Previous Archives:

• 2024

• 2023 and Prior

Active/Ongoing Threads

• December 2024: Hidden taxes in U.S. society that add friction costs with dubious value-add.

• May 2024: Ongoing thread of outbound foreign direct investment (FDI) by Chinese companies in EVs, battery, clean energy and other advanced industries

October 2025

• October 1: Errors & omissions (E&O) in China’s Balance of Payments accounting.

September 2025

• September 29: China still has a long way to go to become fully developed according to its own standards.

• September 27: Confusing EV brands with firms. Related: Latest update on overall China NEV sector profitability.

  

• September 26: Allocating Huawei’s access to elite STEM human capital to chips.

• September 24: “Tiger Parent” CCP and “Tiger Child” Alibaba.

• September 10: David Fishman checks in on the “ghost city” narrative in Ordos in his piece “Ordos City - The World's Most Famous ‘Ghost City’ (That Isn’t)”.

• September 9: “Sitting on the couch with three generations of this one Chinese family, you could catch a glimpse of each phase of the development of China’s modern economy and get some idea of its trajectory in the coming decade.” Reference to old piece from 2018 responding to the question “Why is China's GDP growth rate slowing down?”

• September 8: LFP vs. NMC in commercial NEVs.

• September 5: Solar PV is an extremely commoditized product and cyclicality is a natural part of the industry.

• September 2: Emerging 2025 themes in China’s NEV sector.

• September 1: “Hot take: Solar ‘scaling laws’ will ultimately prove more impactful on humanity than the AI version.”

August 2025

• August 31: “Cat city”

• August 31: China’s version of Forrest Gump / iShowSpeed provides wholesome content for U.S.-China geopolitical relations.

• August 26: “There will be a ‘[fully domestic Chinese advanced node foundry] moment’ in 2026.”

• August 26: Beijing’s top priority is not “beating the West” but domestic development.

• August 25: China’s transition away from the real estate sector has entered the “late innings” stage.

• August 24: Productivity is what ultimately drives per-capita economic growth and increases in living standards over the long run. The key issue with the Pettis/Setser accounting identities framework is lack of acknowledgement of productivity effects.

• August 24: DeepSeek’s deepening collaboration with chip designers.

• August 23: Industrial policy is good/bad depending on the competency of the people and institutions running it.

• August 22: In 2025, China will officially become a “high-income” economy, as defined by the World Bank.

• August 22: Surging token consumption in China, driven by the rise of generative AI.

• August 21: China is now entering the “late innings” of the nearly decade-long transition away from the property sector as the economy’s key principal growth driver.

• August 20: The role of infrastructure (large and small) is the long-term intrinsic value it delivers to society.

• August 20: The rise of renewables has disrupted the traditional economics-based concept of “baseload” supply.

• August 18: China’s retirement age is unusually low.

• August 16: Inconsistencies in recent FT article about alleged delays in DeepSeek’s next model: “DeepSeek’s next AI model delayed by attempt to use Chinese chips: Difficulties of training the start-up’s latest system with Huawei’s semiconductors highlight dependence on Nvidia”

• August 16: To prove HSR represented large capital misallocation in China, critics need to provide a reasonable alternative solution to transporting the roughly 3.6 billion annual passengers per year with superior unit economics. Follow-up response (August 21).

• August 16: The relationship between credit and GDP differs by asset type. Housing and infrastructure are long-lived assets that can be more highly leveraged in China compared to manufacturing.

• August 15: Priors about “scaling laws” may need updating. The locus of competition of AI has shifted from training to inference. Historical analogy of AI with Internet search engines and browsers.

• August 12: Is the market always better at allocating capital compared to the state? I explain how it varies sector-to-sector and how institutional capacity plays a role. Also related response to Greg Ip thread.

• August 12: Some implications of Beijing’s apparent ambivalence towards the nVidia H20.

• August 11: Contextualizing Tamil Nadu’s impressive recent per capita GDP growth.

• August 9: Re-invigoration of biking culture in China and relentless expansion of dedicated biking lanes can be traced to the invention and proliferation of dockless bike-share systems starting around a decade ago.

• August 2: More confusion on Chinese government debt: a critique of Shih/Elkobi paper that does not separate out principal repayments from interest in its credit analysis. This was a follow-up to earlier critique from 2023.

July 2025

• July 30: Addressing confusion over the function of debt in the Chinese economy.

• July 27: Difference between how the U.S. and China are provisioning greenfield power.

• July 27: China unique differentiator is having both (i) high tech and advanced manufacturing and (ii) a vast blue-collar labor force … under a single unified political/economic entity.

• July 25: SpaceX has a large lead [around a decade] over China in reusable rockets.

• July 23: Affordable solar photovoltaics has the same effect as discovering tech that allows you to upgrade a desert tile into an energy-producing one in the classic computer game Civilization.

• July 22: China’s economic model relies on external markets just as much as external markets “rely” on China for supply. Unwinding imposes costs on both sides of the trade.

• July 21: “Foreigners — especially Americans — do not fully appreciate China's predilection for large, capital-intensive infrastructure projects because most do not know what it was like to live in a place starved of God-given natural endowments.”

• July 21: How rising share of wage income from stock-based compensation skews traditional balance of payments interpretation.

• July 16: Taiwan Taxicab Chronicles: Mr. Yu, former CNC operator.

• July 14: Using technology and automation to maintain all the “shiny new infrastructure” China has built in recent decades.

• July 11: Beijing is now mandating that heavy industry start using renewable energy by including sectors like polysilicon, steel and solar in the RPS (Renewable Portfolio Standard) program.

  • Responses to a largely non sequitur attack by Seaver Wang in response to this thread.

• July 11: How rare earths restrictions in 2010 against Japan may have contributed to its (in retrospect) misguided focus on hydrogen and fuel cells.

• July 11: “We’ve been chasing the fusion dream for decades when it was right here in front of us all along.”

• July 10: I am not a big fan of “isms” as it oversimplifies complex systems, ignores idiosyncratic endogenous factors including historical exigencies, and does not account for execution.

• July 10: Commentary on recent developments in 600 km/hour maglev.

• July 3: China’s undercounting of consumption.

June 2025

• June 30: Disagree that there is a cultural/genetic explanation for “invention”.

• June 21: Commentary on BYD and the extended payables issue in China’s auto sector.

• June 19: Challenging certain narratives about Tesla’s role in the development of China’s EV industry.

• June 16: After three years, there still isn’t credible evidence (an accumulating asset pile of trillions in capital outflows since 2021) backing claims that China is manipulating / under-reporting its current account balance.

• June 14: Ignoring FDI income in the analysis of global trade and investment flows and questions of “fairness” and “balance”.

• June 3: “If you pull long enough at every thread, you’ll eventually find your way to some explanation that is related to human capital development.”

May 2025

• May 31: Finding workarounds to export restrictions is an order-of-magnitude easier than advanced chip lithography.

• May 29: There are more than “2 to 3” profitable Chinese EV companies.

• May 25: HSR and air travel are complementary modes of transportation.

• May 16: Chinese coal usage has peaked.

• May 14: On Patrick McGee’s hyperbolic claim that “Apple is investing $55 billion in China”.

• May 13: Beijing does care about its economic relationship with the U.S. and generally believes in the “win-win” benefits of trade and resulting economic specialization.

• May 3: Huawei has a monopsony in elite tech/STEM talent in China.

April 2025

• April 27: The impact of China losing Apple iPhone export processing business.

• April 26: The profits that Apple generates selling products and services to Chinese households is not captured in the trade deficit.

• April 24: “Will the U.S. be able to secure rare earths processing capacity before China develops self-sufficiency in advanced node chips?”

• April 24: The relationship between the United States and the European Union is splintering.

• April 22: We are in a full-blown trade war; gloves are coming off.

• April 22: Disproportionate focus on the goods trade is hampering U.S. trade strategy.

• April 22: The Chinese car market is mainly driven by a productivity-driven virtuous cycle of rising incomes and declining real costs of NEVs, not “reliance on foreign demand”.

• April 21: Steelmanned arguments for the Trump administration’s prosecution of the tariff war are often full of gaping holes.

• April 21: Chinese “culture” can explain why Chinese society has a tendency to converge back into a unified state after periods of dysfunction but it does not make the rise of a strong Chinese state “inevitable” either.

• April 20: Explaining the recent divergence between GDP growth and tax receipts.

• April 20: Jiang’s visit to Hyundai Heavy’s Ulsan shipyard in November 1995 came at a pivotal moment in the debate of the role that SOEs should play in the economy.

• April 19: Huawei has deep expertise in optical and leveraged it in the development of its latest AI server.

• April 18: “Net exports” does a particularly poor job of measuring “reliance on foreign demand” for China because of statistical discrepancies related to “factoryless manufacturing” and because it ignores FDI income.

• April 17: China and the U.S. are at roughly equal levels of comprehensive income inequality (factoring in redistribution and wealth effects), but trajectories are very different.

• April 15: Beijing’s bifurcated view on capital (strategic state capital vs. private sector market-oriented capital).

• April 14: On counterfeiting and China.

• April 13: “Disproportionate tariff focus is counter-productive because it sucks the oxygen away from fixing core domestic problems.”

• April 11: What does economic adjustment look like for China if bilateral trade with the United States completely disappears?

• April 9: China State Council releases white paper laying out its vision for the new global economic order.

• April 8: Being ignorant about the difference between Balance of Trade and Balance of Payments puts U.S. strategy at a disadvantage.

• April 8: Tariff evasion and smuggling is going to increase.

• April 7: Game theorying the Trade War from Vietnam’s perspective.

• April 6: A “thread within a thread” laying out context for how the underlying driver of the export industry has changed over the last two decades (from foreign-funded export processors to domestic firms that have expanded overseas).

• April 6: Layman’s interpretation of the People Daily’s response to the U.S. “reciprocal” tariffs.

• April 6: Menswear Guy on apparel manufacturing. China is SOTA for both low- and high-end clothing production.

• April 5: Potential tariff impact on footwear and apparel manufacturers (Vietnam and China).

• April 4: Vietnam’s future is still bright.

• April 4: U.S. trade deficit does not include offsetting profits by U.S. MNCs.

• April 3: China is Vietnam’s largest final export destination for electronics (HS85).

• April 3: Measuring U.S. demand by global revenue market share overstates how much underlying manufacturing it drives.

• April 3 (Part 1 | Part 2): How much does Beijing care about U.S. tariffs?

• April 2: Random calculation behind the “reciprocal” tariffs.

• April 1: Relative competitiveness of Vietnam and India in the “electrical machinery and electronics” (HS85)

• April 1: Aspirational ceiling of Chinese youth has risen to look beyond the U.S. as a frontier model.

March 2025

• March 25: China is focused on China, not you.

• March 23: Parallels between Beijing’s embrace of containerized shipping and open source technology.

• March 23: Vietnam is now exporting electronics at a higher per capita rate than China.

• March 21: Key difference between China and India in attracting manufacturing FDI is that China created a much more attractive environment to generate returns on manufacturing FDI.

• March 21: IMF released BPM7, which highlights “factoryless manufacturing”.

• March 21: “Abundance” or “excess capacity”?

• March 20: Kai Fu Lee estimates that DeepSeek’s annual operating costs are around $140 million per year and also discusses why open source at the foundation model level is going to win.

• March 19: Piezoelectric walkways generate electricity very inefficiently compared to solar.

• March 17: Thoughts on various predictions of China’s EUV development timeline.

• March 16: Does the Chinese state stymie or spur innovation? It can’t be both.

• March 15: All investment is eventually consumed in GDP accounting.

• March 13: 5G was about industrial automation more than it was about consumer wireless broadband.

• March 11: China is flipping traditional macroeconomics on its head with its focus on elasticity of supply vs. elasticity of demand.

• March 9: Tesla cars manufactured in Shanghai are higher-quality than ones manufactured in Fremont.

• March 8: Mao’s legacy will ultimately be tied to the Communist Party of China, not what people like Chen Yun thought shortly after his death, or even what people think about him today.

• March 7: China and economic autarky.

• March 6: Li Qiang brings up 内卷 at Two Sessions.

• March 5: Balance of trade does not do a good job of capturing the increasing “intangibles” global trade. That trade is captured elsewhere on the Balance of Payments ledger.

• March 5: Chinese households are sitting on an enormous pile of cash/deposits and this is a “coiled spring”.

• March 5: There is significantly more fiscal impulse in the 2025 plan. China is clearly signaling shift to “kinder, gentler” economic policy.

• March 4: Closely watching Tesla’s positioning in China.

• March 3: China is treating “compute” as a public good.

• March 3: The “Great Leap Forward” in chicken and egg production.

• March 3: China does not really care that much about U.S. tariffs this time around.

• March 3: Analytical reasons behind China’s enormous concrete consumption (vs. U.S.)

February 2025

• February 27: Does Beijing actually set persistently high GDP targets?

• February 27: Breakthrough invention vs. incremental innovation.

• February 26: Chinese distant-water fishing and Argentine short-fin squid.

• February 26: AI tools will increase both the ability to find signal as well as the noise.

• February 26: “Velvet glove” needs to be backed up by an “iron fist” for the geopolitical strategy to work.

• February 25: The challenge with re-industrialization is reality that the U.S. will need to re-learn how to walk before it can run again.

• February 24: Taiwan is stuck between a rock and a hard place.

• February 24: Rapid decline in LIDAR component price makes me question whether Tesla pursued the right ADAS strategy based on vision-only.

• February 23 (Day 1 | Day 2 | Day 3 | Day 4 | Day 5 | Bonus Day 6): It’s open source week at DeepSeek.

• February 23 (Part 1 | Part 2): Some updates and rebuttals on the Apple iPhone and BoP debate.

• February 19: China’s shift from a real estate-driven economy to one driven by the “new three”.

• February 17: Different drivers that contribute to an economy’s overall capital efficiency: production efficiency vs. distribution efficiency.

• February 15: Twitter Spaces with on AI in China.

• February 14: On tech transfer, TSMC and Intel.

• February 13: The software developer role will change in the coming years due to AI, just like how human calculators were replaced by electronic ones.

• February 13: Focus vs. diversification was a key difference between prominent Chinese companies and Japanese and Korean ones.

• February 10: Framing exports as one-sided reliance can be misleading.

• February 7: The rising U.S. trade deficit (based on physical flow of goods) is a mirror to the profitability of U.S. MNCs that generate large offshore profits.

• February 6: The Twitter financial turnaround and “doing more with less”, the “scarcity mindset”.

• February 4: If there is one overarching theme in China’s private sector, it is cost and operational efficiency, not extravagant waste.

• February 4: China retaliates to U.S. tariff announcement with an expanded bag.

• February 1: OpenAI is like Windows (closed, proprietary). DeepSeek and other open-source contributors are like Linux (open, shared). Perhaps this is less a “Sputnik” moment (1957) and more a “Netscape” moment (1994).

January 2025

• January 30: Seven years ago, export controls were a strategic mistake (with tactical blunders). Today, it would be a tactical mistake to roll them back. But strategically, we need to re-focus elsewhere (i.e. solving domestic problems).

• January 27: Economist Tyler Cowen on “the mistakes of Michael Pettis”.

• January 26: DeepSeek’s deep and multi-disciplinary technical depth.

• January 25: Electric “intelligent” vehicles and what really makes NEVs disruptive.

• January 25: Pettis is wrong about Germany too.

• January 24: (Part 1 | Part 2 | Part 3 | Part 4): Ongoing series of posts on DeepSeek

  • The rise of Deepseek says more about China’s broader tech/AI ecosystem than the company itself. Open vs. closed. Training vs. inference (Deepseek’s likely adoption of Huawei Ascend 910C chips for inference).

  • DeepSeek demonstrated serious technical chops, using PTX (similar to Assembly) to improve the efficiency of its nVidia H800 chips (which have limitations compared to the H100 due to export controls).

  • Observations on response from Dario Amodei (Anthropic) on DeepSeek and export controls.

• January 24: Destroying trust is no way to build a strong manufacturing base.

• January 23: Update on Chinese auto exports.

• January 22 (Part 1 | Part 2 | Part 3): OpenAI / Stargate consortium announces goal of investing $500 billion in AI datacenter. Two key points:

  • Financing “compute” is different from financing long-lived infrastructure assets.

  • With this scale of resource allocation and potential direct/indirect taxpayer-funded involvement, we need to think carefully about alignment of interests and incentives.

• January 20: The way the U.S. has handled TikTok is just so all-around dumb and reflects poorly on “state capacity”.

• January 20: “The success of industrial policy is largely dependent on state capacity”

• January 17: Reprise on the idea of how “large countries operate on a different set of economic laws than small ones” in the context of China’s dominance of global manufacturing.

• January 17: The rising global competition for Chinese FDI.

• January 17: Updated 2024 Chinese export figures in four renewables categories and related commentary.

• January 14: The interplay between labor, capital and consumer surplus.

• January 13 (Part 1 | Part 2 | Part 3 | Part 4 | Part 5): TikTok refugees flood into 小红书 (“RED”, “Rednote”, “Little Red Book”)

• January 13: China’s goods trade surplus in 2024 will exceed $1 trillion. What are we to make of this number?

• January 13: Green hydrogen takes renewables beyond the grid.

• January 12: The lack of Taiwanese voices in discussions that are about Taiwan.

• January 12: Update on Chinese ICE and NEV vehicle exports.

• January 11: The size and scope of foreign multinational operations in China.

• January 10: Apple/Foxconn, “producer power” in the context of the U.S.-China and India-ASEAN rivalries.

• January 10: How finance is ultimately tied to real world institutional capacity and sector expertise: why “American finance and money” cannot play a significant role in international metro, rail and renewables projects.

• January 9: Continuously underestimating China’s technology and economic development.

• January 9: The long-term strategic effects of export controls.

• January 9: Vietnam is attracting the same amount of manufacturing FDI as India despite a population 14x smaller.

• January 8: China’s household share of GDP has re-balanced to normal levels by global standards, but this remains unacknowledged.

• January 8: China continuously exceeds outside expectations on its technology and industrial efforts, this time in biotech:

• January 7: China’s net international investment position and estimating investment income on the Balance of Payments.

• January 6: I agreed with much/most of this excellent interview by The Wire with Barry Naughton on “the State of the Xi Jinping Economy” (archive). This thread goes through some of the pieces where I did not just nod along.

• January 5: Some more recent indicators that China’s residential property market is completing its transition into just anothe mundane, boring sector:

  • 5th National Economic Census: Property sector transitioning from greenfield development to management (QQ: archive)

  • Residential electricity consumption is skyrocketing, driven in part by residents moving into modern new apartments with air conditioning (from ):

• January 5: Debunking more HSR nonsense: speed round.

• January 5: The Xiaomi SU7 is a direct competitor to Tesla’s Model Y in China.

• January 4: Focusing on 26 HSR “ghost stations” instead of the 1,000+ that are being used — the “Costco” trade-off between cost efficiencies and overbuilding.

• January 4: Wuhu vs. Hefei and what it says about local government competition.

• January 4: China sees failed capital-intensive projects as a challenge.

• January 4: China’s manufacturing has a decent chance of weathering the “age of robots” (archive).

• January 2: On GM-China and technology transfers.

• January 2: An ode to Taiwan alleyways.

• January 1: Updated Q3 2024 China balance of payments bridge.

• January 1: Key China NEV sector themes for 2025.

I enjoy analyzing economic datasets spanning long periods because economic development — especially for developing countries — is a long-term endeavor that unfolds over multiple decades. Foreign direct investment (FDI) is just one facet of economic development and I have found that it is often misunderstood because some aspects of it are not well-understood.

Years ago, I had spent time really trying to understand the accounting behind balance of payments (BoP) and trade flows. In 2019, I explored the question “Why does the US have such a high trade deficit?” by immersing myself into U.S. BoP data on the current account side. This background may provide helpful context for understanding the FDI component, which sits on the capital account side. In recent years, I have also relied heavily on the work that Brad Setser does on BoP to “stay current” on the trends; I highly recommend following him.

This post aims to educate both myself and readers about the accounting mechanics underpinning FDI flows, different archetypes of companies that impact both inbound and outbound capital investment, and how various dynamics intersect with observable trends in the news.

Subscribe now

Accounting mechanics of capital investment in a controlled capital account

The first two categories of capital investment are fairly straightforward:

• Non-resident inflows (blue) represent registered new capital from either foreign companies that are already operating in China or new ones that are looking to invest in China for the first time.

• Resident outflows (red) are mainland Chinese companies that are investing outside of China.

These two categories are relatively easy to understand because they are directly linked to real foreign exchange operations that are easily trackable. China operates a controlled capital account. In practical terms, this means that whenever a foreign company seeks to exchange foreign currency for local renminbi, or when a mainland company wants to perform the reverse exchange, they must submit paperwork to their banks specifying the purpose of the exchange1.

The red and blue lines capture the transactions that are specified as “capital investment” in the foreign exchange conversion operations.

Non-resident reinvested earnings is more complicated

“Non-resident reinvested earnings” is more complicated because it involves some accounting estimation. I have found this component to be the source of much misunderstanding about Chinese FDI.

• Onshore reinvested earnings is equal to profits (or losses) generated at the onshore subsidiary minus dividends that are repatriated outside of China back to the parent2.

As above with capital investment, the repatriated dividends involve currency conversion and will thus involve paperwork with the bank that is handling the foreign exchange. In this case, in its foreign exchange application the company specifies that this foreign exchange operation is to take profits as opposed to liquidation of capital investment.

The profit and loss (“P&L”) component is the more non-intuitive and problematic aspect. Anyone experienced in reading financial statements will appreciate how calculations behind accounting profits and losses are often subject to accounting rules and judgments that may not always be intuitive.

For instance, when a company liquidates an asset or ceases operations, a final calculation is necessary to reconcile or balance3 the difference between the sale or liquidation proceeds and the latest accounting value or net worth on the balance sheet. These figures can often be very different, especially if it has been running an unsuccessful loss-making operation in a business with significant capital investment. This can lead to large realized “one-time” losses. Conversely, if you built a successful operation and decided to sell the business for a large profit, it can realize a very large “one-time” accounting profit when you sell the business.

Even after excluding one-time “extraordinary” profits and losses, another complicating factor with “non-resident reinvested earnings” arises from potential timing mismatches. When a company generates profit, it faces two choices:

• Return the profit to the capital holders as a return on capital via dividends.

• Reinvest the profits back into the business, presumably to foster growth at an attractive rate of return.

These two choices can be driven by different decision-making dynamics. Some companies with currently profitable operations but dim future growth prospects may opt to extract profits from the operation (via repatriated dividends) rather than incrementally investing in it. In this scenario, they would not significantly contribute to the yellow “reinvested earnings” line. Others may see strong growth prospects for their products or services and choose to reinvest all profits back into the onshore operation. This would contribute to “reinvested earnings” FDI.

Additionally, some companies may perceive growth prospects as so high that they seek to invest beyond what the onshore operation can self-fund. In such cases, they would rely on new capital investment from the offshore parent company, which would undergo the same foreign currency process described earlier. The incremental amount would then be counted as new FDI (blue line).

Conversely, companies that are loss-making have no earnings to reinvest. If they do take distributions from the business, it would be drawn from historical reinvested earnings accumulated during profitable periods when they opted to retain profits within the onshore operation (e.g., as working capital). In such instances, reinvested earnings would be negative. This underscores how mismatched timing also influences reinvested earnings data. A company might generate accounting profits in Year 1, but these profits require time to convert to cash that can be repatriated, only becoming available as dividends in Year 2.

Furthermore, it’s important to note that just because foreign exchange operations are documented, it does not always ensure accuracy. Companies can misclassify foreign exchange operations between “dividends” and “capital,” either inadvertently due to negligence or intentionally to disguise their transactions — tax avoidance being a significant motive. Some FDI is registered to foreign-domiciled companies may effectively be Chinese4 that are again either inadvertently or intentionally miscategorizing the flows in their foreign exchange applications.

Additional complications with “reinvested earnings”

There are additional complications involved in the actual calculation of “reinvested earnings”. As Homin Lee notes here, “in 2013, China stopped providing data on the portion of FDI inflows attributable to reinvested earnings.” The methodology underpinning Robin’s “reinvested earnings” flow data is derived from the difference between two different sets of flow data from two different organizations (SAFE vs. Ministry of Commerce or “MOFCOM”). The MOFCOM data shows only new and incremental registered FDI flows that correspond to “non-resident inflows (actually utilized)” in Robin’s dataset while the difference between SAFE and MOFCOM correspond to the “non-resident reinvested earnings”:

There are some differences in what is included in the definition of FDI between the SAFE and MOFCOM datasets5, although as Nicholas Lardy notes here in this report, these differences are “now likely quite small”. This follow-up from Robin also shows high correlation between the data that underpins the SAFE and MOFCOM datasets6.

Understanding these mundane mechanics of accounting are important as we try to parse through and interpret the data.

Four archetypes of private sector companies

Four main archetypes of private sector companies contribute to or impact FDI in China7. They operate in a variety of sectors. Understanding these archetypes is crucial for grasping the diverse factors behind inbound and outbound capital investment in the country.

1. Foreign-funded export-processing companies: These companies primarily target developed markets, focusing on manufacturing goods for export, typically for more labor-intensive operations.

2. Foreign MNCs: Multinational corporations invest in China with the aim of penetrating its domestic market, seeking opportunities to sell their products and services to Chinese consumers.

3. Private real estate developers: In recent years, private real estate developers have raised large sums of capital overseas to support their expansion onshore.

4. Domestic firms: This archetype encompasses private sector manufacturing firms based in mainland China8. These companies serve both domestic consumers and foreign ones via exports.

1. Foreign-funded export-processing companies

Many export-processing companies relocated their factories to China in the 1990s to capitalize on the mainland’s abundant labor force. Hong Kong and Taiwanese exporters, in particular, represent a significant portion of this category. Leveraging existing connections to developed markets and expertise in managing labor-intensive manufacturing operations, these firms focused on sectors such as toys, shoes, clothing, household appliances, and consumer electronics — goods requiring substantial labor input.

Terms like “offshoring” or the “China Shock” in the 1990s and early 2000s often referred to this phenomenon. Notable examples include Belle (shoes), VTech (phones, toys), Allan (small kitchen appliances), and numerous small, often family-run factories producing non-branded goods.

Many foreign-funded exporters struggled to penetrate China’s domestic market because their products did not meet stringent price points of price-conscious Chinese consumers, along with challenges in navigating local distribution networks.

Despite these obstacles, export-processing companies experienced rapid growth through the mid-2000s, considered their “golden age.” However, the global financial crisis of 2007-08 marked the end of this prosperous era.

Rather than disappearing entirely, export-processing companies adapted to the changing landscape. Many invested in “semi-automation” and relocated factories further inland to access lower labor costs. While these adjustments helped initially, relentless increases in Chinese labor (and other) costs and limitations in automation9 within these industries posed ongoing challenges. Additionally, mainland firms became increasingly competitive for both labor and export markets10. The evolution of foreign-funded export processing firms can be observed in the blue bars of Gerard DiPippo's insightful chart:

In 2020-21, there was a noticeable COVID-induced bump in export activity, but these effects have since faded. With the ongoing trade war and continuously rising labor costs in China, there is an increasing trend of foreign-owned exporters relocating labor-intensive processing operations to regions like Southeast Asia and India.

Many are simply shutting down or transitioning into real estate ventures, as the next-generation family members are often unwilling or incapable of managing these struggling, often “gritty” operations.

2. Foreign MNCs

Notably, there is minimal representation by American companies in the export-processing group. Instead, this category primarily comprised larger multinational corporations (MNCs) seeking to tap into China’s vast domestic market.

A prime example is the automobile joint venture (JV) model, where foreign MNCs like GM, Toyota, or Volkswagen partner with local manufacturers such as SAIC, Dongfeng, and/or FAW to distribute their cars11 in the Chinese market. In this arrangement, the foreign MNC contributes technology, designs, branding, and consultancy expertise, while the local partner manages day-to-day operations, including hiring the local workforce, supply chain logistics, and manufacturing operations. Typically, these JVs operate on a 50/50 basis, with joint decision-making and profit-sharing.

Compared to foreign-funded export-processing companies, these ventures are much more capital-intensive, involving large-scale assembly lines with automation and robotics, complex supply chains, and substantial inventories. Whether driven by policy objectives such as “market access for technology” (市场换技术) or purely economic considerations, it was often a prerequisite for MNCs to establish manufacturing operations in China to effectively cater to Chinese consumers. Unlike export-processing companies, these ventures required Chinese partners with a deep understanding of local operations, consumer preferences, and distribution channels. Most of these ventures exclusively targeted the Chinese market and did not consider using China as an export base, often due to restrictions outlined in their JV agreements.

3. Real estate and financial services

At its peak, the property sector and financial services industry accounted for almost a third of FDI12. But since the “housing is for living, not speculation” was announced in 2016 (signaling the peak of the real estate market), foreign investment (including Chinese private developers that have been raising money overseas) in the real estate and financial services industry has been declining and only accounted for 11% of FDI in 2022.

4. Domestic firms

In the backdrop are domestic private sector manufacturers and OEMs, competing with both foreign-funded export-processing companies and foreign MNCs. In Robin’s original chart, they are represented within the red bars of outbound Chinese investment. Gerard’s chart depicts them in the green “Private and Other” line item.

These include companies like Huawei (tech/telecom), BYD (batteries/EVs/contract manufacturing), Longi (solar PV), and CATL (batteries). Launched mostly in the last three decades, they have been steadily catching up to both foreign-funded export-processing companies and foreign MNCs. The green line on Gerard’s chart reflects their progress, surpassing foreign-funded exporters around 2017-18. They have also outpaced foreign MNCs in sectors like automobiles, particularly in electric vehicles, in recent years.

What FDI trends tell us about the Chinese economy

With this crucial background in mind, let’s delve into the more intriguing part of the discussion: what do these numbers tell us?

Last October, the FT published a chart illustrating how FDI in China has reached a two-decade low. Notably, this chart was based on the SAFE FDI dataset, which includes “non-resident reinvested earnings”. I post it here alongside Robin’s chart, where the black line on the right corresponds to the dark blue monthly trends on the left:

As I pointed out in a Twitter/X thread previously, one issue with the FT chart its aggregation of new incremental FDI with the more complex “reinvested earnings” component. Robin’s chart is valuable because it disaggregates FDI into two distinct components, providing deeper insight into underlying dynamics, especially considered in the context of FDI accounting mechanics and the various company archetypes impacting capital investment decisions

Non-resident inflows

Non-resident inflows primarily comprise FDI from the two foreign-funded archetypes: export-processing companies and MNCs. This figure has remained relatively stable, steadily increasing since the early 2000s until experiencing a spike in 2021-22. While the topline numbers have been very steady, the contribution from industries has fluctuated over time13. Last year marked the first time it had declined and it will be interesting to see whether this is the beginning of a new trend.

These inflows are fairly reliable because they can be cross-checked and validated with external data from large publicly listed foreign MNCs14.

Non-resident reinvested earnings

Non-resident reinvested earnings exhibit greater volatility compared to inflows as they depend on both profitability and incremental reinvestment decisions.

Following the Global Financial Crisis (GFC), export-processing companies transitioned into being “cash cows” in contrast to growing companies requiring significant capacity expansion in the 1990s and 2000s. Consequently, a larger portion of their FDI stemmed from decisions to reinvest profits generated onshore rather than from incremental new investments. These reinvestment decisions fluctuated with the shifts in their export markets.

In contrast, foreign MNCs experienced significant growth in their China domestic sales throughout the 2010s. This growth is evident in the 10-Ks and annual reports across various industries, including smartphones (Apple), cars (GM, Toyota), planes (Boeing, Airbus), brands (Nike, Hermes), fast-food restaurants (McDonald’s), and coffee shops (Starbucks). This growth led to both increased profits and reinvestment of those profits back into their onshore operations.

Although it is challenging to precisely delineate the proportions of profits versus dividends and reinvested earnings from this dataset, we can infer trends that are becoming increasingly apparent.

However, things started to shift in 2022. Foreign MNCs are now experiencing lower profitability and are scaling back reinvestment of these reduced earnings into their businesses. Why?

Domestic Chinese firms are catching up to Foreign MNCs in certain sectors

The primary reason appears to stem from catch-up by Chinese firms. As mentioned earlier, by the mid-2010s, they had already reached parity with foreign-funded exporters. By the late 2010s, domestic Chinese firms had more or less caught up to foreign MNCs and companies in areas such as cars, smartphones, and even coffee, within the Chinese domestic market.

This trend is particularly evident in the transition from internal combustion engine (ICE) vehicles to electric vehicles (EVs). The emergence of Chinese new energy vehicle (NEV) brands has led to a sharp decline in ICE sales15, traditionally dominated by foreign MNCs through their joint ventures (JVs). For instance, as recently as 2017, GM’s equity earnings in China were contributing the majority of its global net income. However, China sales have plummeted since 2018. The realization that their products were no longer competitive came abruptly and recently.

Some automakers continued to anticipate significant market expansion until the onset of the global pandemic. Over the last five years, not only have these growth plans been shelved, but many foreign automakers — especially the smaller, weaker ones — have been compelled to shutter their China operations.

Japanese automaker Mitsubishi presents an illustrative case study here. For decades, it was a minor exporter of its branded cars to China but in 2012 it invested in a formal JV with GAC (Guangzhou Automobile Group16) to manufacture locally. Sales steadily increased from ~45k in 2013 to a peak of ~158k vehicles in 2019 before suddenly halving in 2020.

While company executives could attribute this to the pandemic, deciding to continue to invest new models until 2022, it became increasingly clear that this was a secular (not cyclical) shift in Chinese consumer preferences17. In October 2023, the business announced that it had terminated its China operations, transferring its 50% ownership to GAC and incurring a JPY24.3 billion ($162 million) one-time loss related to the restructuring. GAC would re-purpose the existing factory for the new electric vehicles that it has developed under the Aion brand. Sales of its Aion brand (launched in 2019) were now running at ~480k per year.

The realized losses incurred from the restructuring of Mitsubishi’s China operations would be recorded as a realized loss and count as negative reinvested earnings in Robin’s chart above. The shift to electric vehicles and the rapid evolution of Chinese car-buying habits have also affected numerous other foreign automotive JVs, particularly smaller brands similar to Mitsubishi. Consequently, there has been a significant decline in reinvested earnings (yellow line) since 2021, as profits have turned into losses, prompting foreign MNCs to shut down operations and withdraw any accumulated reinvested earnings from previous periods, if any.

As we can see from the non-resident inflows (blue line), not all foreign companies are struggling quite as much as the foreign auto JVs in China:

• Hermes recently announced quarterly sales rising 17% on “strong China demand”, a possible indicator that “mass affluent” Chinese consumers may finally be opening their pocketbooks again.

• Sales to China made up a record 49% of Dutch SME provider ASML’s revenue in Q1 despite18 export controls.

That said, even non-resident inflows experienced a decline in 2023, potentially signaling more challenges ahead. Even companies that were previously thriving in China are now encountering challenges. After years of growth Apple and Tesla have recently witnessed declining sales in China, while Starbucks recently reported decreasing same-store sales. Much of this can be attributed to the rise of Chinese competition.

Export-processing companies are moving manufacturing operations offshore at an accelerating rate

Another significant factor contributing to the decline in FDI, including both reinvested earnings and new inflows, is the trend of exporters relocating their manufacturing operations offshore. This shift is driven by rising labor costs and the imposition of bilateral Section 301 tariffs by the United States in 2018. Consequently, both foreign-funded and mainland exporters are moving their operations out of China to regions with lower labor costs, such as Southeast Asia and India.

Export processing companies, particularly those funded by Hong Kong owners, have increasingly been relocating their operations to Southeast Asia. One example is VTech, a Hong Kong-based exporter that acquired a factory in Malaysia in 2018 from Japanese consumer electronics brand Pioneer to expand its audio equipment and contract manufacturing business. Additionally, foreign brands such as Apple, through their contract manufacturers, have been transitioning manufacturing operations to Southeast Asia and other regions like India.

Since the Global Financial Crisis (GFC), foreign exporters have faced mounting challenges in China, primarily due to escalating labor costs. Samsung was early to recognize this and and initiated relocation of its manufacturing operations out of China, mainly to Vietnam, long before the imposition of the 2018 tariffs. Before the tariffs, export processing companies often deliberated whether to relocate their operations further inland or offshore. However, following the imposition of tariffs, these decisions have become increasingly clear.

It’s crucial to emphasize that the impact thus far has primarily affected export processing industries. China’s growing domestic demand is increasingly bolstering onshore manufacturing, even in sectors requiring higher labor outputs. Moreover, policymakers appear to be making concerted efforts to retain a critical mass of most manufacturing industries, aiming for a “complete supply chain”. Mainland exporters continue to manufacture in China for the domestic market while supplementing with offshore manufacturing targeted at export markets. In contrast, foreign-funded export-processing companies that did not cater to the mainland market are either relocating entirely or ceasing operations.

Real estate developers went bust

While underlying real estate construction economy had plateaued by the mid-2010s, property developers continued to sell apartments on an increasingly forward basis through aggressive pre-sales programs. This allowed them to continue booking financial profits and accumulating retained earnings. However, with the introduction of “Three Red Lines” in 2020, which essentially cut off financing and liquidity, this approach came to an end.

Evergrande, probably the most aggressive large developer, began revealing financial difficulties in 2021, announcing a massive write-down of assets on its balance sheet of ¥374 billion, representing over 100% of its shareholders’ equity19. Evergrande had also raised significant funds overseas, most of which was repatriated onshore to support domestic property development activities and would be counted as FDI even though it is a Chinese company.

Other developers, many of whom had also raised overseas debt, were soon to follow. These large write-downs have also likely contributed to the sharp decline in the “reinvested earnings” component of FDI in since 2021.

Macroeconomic factors impacting treasury management

Gerard DiPippo also highlights another important factor that has an impact on the estimated “reinvested earnings” figure. Companies that operate in multiple jurisdictions will have bank accounts and hold cash and liquid assets in multiple currencies. This exposes them to foreign exchange rate risk and interest rate differences. In this interesting chart, he notes the correlation between “reinvested earnings” flows and “portfolio inflows debt” flows (which are heavily impacted by interest rate differentials):

The theory is that when interest rates are higher in China, companies will choose to keep their cash in Chinese banks in the form of renminbi to take advantage of those higher rates. This would reduce the amount that they distribute in the form of repatriated dividends and thus increase the “reinvested earnings” estimation. Since this is just treasury management of liquid assets, this decision does not actually represent a true capital investment (in the form of plant, machinery, inventories, etc.) decision.

In my experience20, large foreign multinationals like Apple, GM or Starbucks do not play the “interest rate arbitrage” game and typically repatriate most of their retained earnings in China subject to local investment needs (e.g. expansion funds and working capital needs) and local regulations21. In contrast, the treasury management departments at Hong Kong-owned firms appear to be more prone to make directional bets on the USD-CNY rate which can result in more volatile swings in the flow of liquid capital that “pollutes” the “reinvested earnings” measure.

Overall I do agree with Gerard that macroeconomic conditions play a key role here and need to be taken into account as well.

Bottom line

Multiple trends and effects have contributed to declining FDI over the past three years. Due to the complicated mechanics described above, it is hard to know with full precision, but there are some noticeable drivers:

• A primary driver is the gradual catch-up of Chinese players, which is now increasingly “crowding out” investment by foreign MNCs and companies. In certain sectors, this is causing foreign MNCs to slow the rate of expansion or in some cases (like Mitsubishi) completely pull out.

• The downturn in the property sector has also caused large write-downs on previous FDI and associated accumulated retained earnings. This has contributed to the sharp drop in the “non-resident retained earnings” component of FDI.

• Trade tariffs are also exerting a significant impact, particularly on foreign-funded exporters. This is hastening the redirection of real capital investment flows to regions such as Southeast Asia, India, and Mexico. Chinese firms are also increasing their investments abroad, which shows up in outbound FDI.

• Lastly, macroeconomic conditions also play a role in the treasury operations of companies, impacting domicile and currency that they choose to hold their cash and liquid assets. Due to the way that reinvested earnings is calculated, this can also have a big impact on the “flow” data.

In 2022, after decades of predominantly net inflows (the black line in Robin’s chart), this figure turned negative, indicating that Chinese outbound investment had surpassed inbound FDI. This trend persisted in 2023, and I anticipate it to continue or even accelerate as many remaining exporters continue to shift their mainland China operations offshore. Additionally, Chinese manufacturers and brands are increasingly investing in local manufacturing in markets worldwide.

After decades during which China was a net importer of FDI, it is likely that in the coming years, China will shift to being a net exporter of FDI.

Subscribe now

6

Source: Homin Lee via Robin Brooks.
”Reinvested earnings from 2010 to 2014 explain 80% of this ‘flow’.”

19

Source: Evergrande 2022 Annual Report
Evergrande had been “profitable” all the way through 2020, booking profits from pre-sales developments that would not be delivered until years in the future and using the deposits collected to finance existing projects through synthetic financing arrangements (and likely illegal, if not formally than in spirit). These profits are accumulated on the balance sheet in the form of retained earnings. When the assets are written down, this represents a large operating loss that will impact “reinvested earnings” component of FDI.

In the grand scheme of things, this is only a quarter’s worth of numbers and as such there is a limit to how meaningful they are. I tend to focus on the longer-term and am using the release more as a framework to synthesize some of the higher-level macroeconomic trends that I have been discussing — mainly on Twitter/X — over the past year:

1. Transition from property to advanced manufacturing

2. China’s trading relationship with the world

3. Labor dynamics and structural change driven by demographics

In this essay, I will tackle #1: the ongoing transition of China’s economy from property to advanced manufacturing. My views are different from mainstream views that GDP growth and productivity are in permanent, secular decline — instead, that the Chinese economy is transitioning to a new era in which GDP growth gets a healthy boost from the structural shift into advanced manufacturing that more than offsets demographic decline in its working-age population. This goes back to the age-old question of “Will China become old before it becomes rich?”.

To understand my rationale requires stepping out very far beyond the Q1 numbers, putting them into the context of (i) three decades of growth and multiple structural economic shifts, (ii) the GDP arithmetic and mechanics of different categories of investment, and (iii) the experiences of comparable economies (during rapid urbanization) like South Korea and Taiwan. Hopefully with this context, we can better understand how the Q1 numbers fit in but even more importantly, what future GDP and other economic releases may tell us about the long-term direction of the Chinese economy.

Before getting into it, I first want to briefly comment on the value of examining primary vs. secondary sources which is relevant here as well.

Subscribe now

Primary vs. secondary sources

I have seen a spate of commentary about the Q1 data on Twitter/X that usually links to news articles from third-party news sites instead of the original press release. As a general rule, I prefer to go to directly to the primary source, which in this case is the original NBS press release.

Every article you read about Q1 GDP that is not this NBS press release is ultimately based on the data contained in this press release. Newsrooms and media outlets will naturally take this primary source data and modify it in some way — otherwise what is their raison d’être? Even news-centric media that pride themselves on journalistic impartiality need to “add value” and this will naturally modifies the primary source data even if truly impartial. Of course there are media outlets with stated or clear goals to align certain narratives. Choosing as a starting point to re-post skewed takes is typically an endorsement of that take.

Of course, my raison d’être here is to offer my own commentary and takes and I will certainly selectively focus on certain sections of the report1 but I try my best to start with primary sources. In this case, it is relatively easy because NBS offers an English-language version of the release. What sometimes makes Chinese economic analysis more challenging is that many primary sources may not be offered in English, requiring one to review primary sources in Chinese.

I also understand why people seem to avoid posting the NBS link. First, Chinese government website design is stuck in early Web 1.0 Internet era. Typically when you try to link to these government websites on Twitter/X, it does not offer up a nice “social preview” image that you will get from bleeding-edge-in-comparison websites like Reuters. It just shows up as a partial link — often with random letters that look like spam that nobody will engage with:

Second, and more significantly, there is a tendency (somewhat understandably) for many people outside China to simply not trust anything coming from China. Linking to a Chinese government website could be seen as endorsing the article or the views contained in that article. And of course the bean counters at NBS in charge of putting together this primary source will offer their own spin and take, just like newsrooms and media outlets around the world. Indeed, just look at the title and preamble of the Q1 report:

Clearly, the bean counters want you to know that the national economy “made a good start” to the year. In my head, I imagine them as expectant puppies asking for a nice pat on the back for their efforts. And I am always amused by these long-winded preambles that seem to be a prerequisite of every government release extolling the virtues of the Communist Party of China, the Central Committee and Comrade Xi Jinping. While entertaining (at least to me), one can and should ignore the ornate writing style and focus instead on the real substance: the data.

Stripping away the noise, these press releases contain much useful data. These press releases also tend to follow the same format from release to release which allows one to look at the same trends over time. In this way, these press releases remind me of public companies and their earnings announcements. There is a lot of ritualistic fluff that you can ignore — real analysts will know where to look having seen multiple versions of the press release over the years.

There are of course some who do not even think the data itself is reliable (except it seems, when it suits a particular narrative). In this case, whether you rely on secondary sources or primary sources does not make a difference. The secondary sources are still based on the same data that is viewed as unreliable. While we always must be open to the idea that the data is manipulated, if we assume it is all wrong then there is no reason to do any analysis at all.

The point here is that analysts should always rely or at least consult primary sources. This is true whether it is Chinese data or any other data.

(1) Transition from property to advanced manufacturing

Both industrial production and fixed asset investment (FAI)2 data in Q1 confirmed the broad trend that we have been witnessing for some time now in the economy’s transition from property as a key growth driver to advanced or (as it is referred to in the press release) “high-tech” manufacturing.

This is one of the major Chinese macro themes that I have been discussing at length over the past year. One of the elements of the popular and seemingly mainstream “overinvestment” narrative is that Chinese policymakers are merely shifting investment from one flavor (property) to another (manufacturing). For advocates of this narrative, there is no meaningful difference between the two — it still represents an “imbalance” that needs fixing.

I strongly disagree with this narrative. There are fundamental differences in the nature of the underlying assets being formed and the level of maturity within the asset class’ “Pareto cycle” that warrant a deeper level of analysis to really understand impacts to aggregate GDP.

Using the “Pareto efficiency” framework to analyze economic transitions

Investing in housing and “old” infrastructure at the tail end of their pareto-optimal stages is fundamentally different from investing in advanced manufacturing at the front end of their pareto-improvement stages.

Industrial production is typically oriented towards consumption GDP while FAI represents investment GDP. At a high-level, industrial production can be viewed as the “R” and FAI as the “I” in the Return on Investment (RoI) calculation. From an “R” and “I” perspective, property and infrastructure assets behave very differently from industrial assets.

I will warn in advance that there is arithmetic and some modeling involved. But I think it is worth absorbing to appreciate how these differences will manifest in historical GDP as well as future trends.

Not all capital stock is created equal: housing and infrastructure

Property and infrastructure share several common characteristics. They are (i) capital-intensive, featuring large upfront investments that (ii) typically generate low percentage returns on the upfront invested capital but (iii) over a very long useful life (50-100 years3).

With greenfield urban housing, the GDP impact is initially dominated by gross capital formation (“I”). When residents move into their newly built apartment, the incremental GDP impact (“R”), represented by actual or imputed rent, is modest — often representing low-single-digit percentage4 of the upfront capital investment. But these assets have very long useful lives (50 years or more) so if the asset is utilized (lived in), it eventually pays for itself (and more) in a long tail of future GDP.

Here is an illustrative example of what the GDP impact of housing might be:

In the illustrative example above, a $100,000 investment5 in newbuild construction starts to yield a small return (4%) after a three-year construction period. Over fifty years, the long tail of GDP consumption grows steadily. By year 50 (when it might need to be replaced), it is delivering a 19.5% real return annually on the original investment (driven by increases in real household income) and has generated >5x accumulated real returns on the original investment. This turned out to be a nice payback if one could be patient.

Infrastructure also tends to have the same characteristics; if anything it features even longer useful lives and even lower initial returns on equity. But similar to housing, if properly utilized, it will deliver a positive GDP return to society over many decades of use. This is especially true the more household incomes rise — this increases the real GDP return as these assets are “consumed” over time. All things equal, the value of shelter and infrastructure is higher for people as their incomes rise.

Not all capital stock is created equal: factory equipment

In contrast, industrial assets have much different characteristics. While there is a property or physical plant component to industrial investment, the majority of industrial assets are comprised of factory equipment.

This factory equipment tends to have shorter useful lives (3-12 years6) and much higher percentage returns on investment so that businesses can recover the value of their capital investments within that shorter useful life period. An illustrative model of GDP impact industrial equipment might look like this:

As you can see, the first notable difference is that the initial investment (the “I”) can happen quickly — with machines manufactured (or imported), delivered, installed and rapidly put to work to deliver incremental GDP return or growth (the “R”). This means less capital that is tied up in non-productive construction-in-progress and working capital. Second, the initial percentage returns on investment are much higher than housing and infrastructure — in this case, 20% per year.

But because the asset only has a useful life of 12 years, the lifetime GDP return is only 2.0x, lower than the housing example above. At that point, the asset will need to be replaced due to obsolescence or simply being worn out. One of the key takeaways here is that “lower RoI” housing and infrastructure is not necessarily worse than “higher RoI” business investment because of the differences in useful life.

A (much) longer-term view of the property transition

Using these illutrative frameworks for thinking about “R” and “I” in GDP terms, one can see how the economy has been making this transition over multiple years with the Q1 data simply another datapoint in a long series of datapoints confirming this trend.

“Three Red Lines” (3RL) in August 2020 was the beginning of the end for property and “old” infrastructure as leading drivers of the economy. But the transition started well before7; most people just didn’t notice.

The rise of newbuild housing and infrastructure were the principal components of the multi-decade “urbanization” uber-macro economic policy that really kicked off shortly after the post-reform period. It started to accelerate in the early 2000s as legislative reforms were enacted that enabled and formalized property ownership, a prerequisite to home ownership by households. And it really got a boost after the Global Financial Crisis in 2007-08 when authorities approved a massive stimulus to rapidly absorb millions of jobs (mostly migrant laborers) that were abruptly shed in the coastal export processing regions.

From the 2000s to the early 2010s, investment in property and infrastructure drove gross capital formation to hitherto unseen levels, high even by “East Asian development model” standards. But there was a notable change around 2013-14 when property investment as a percentage of GDP peaked. After plateauing for several years, triggered by 3RL in 2020, it began a steady decline. One of the best illustrations of this multi-decade cycle is illustrated in floor space completions8:

While it is a lagging indicator compared to starts and new sales, floorspace completions better correlates with the amount of economic resources — mainly blue-collar migrant labor — dedicated to the construction industry. This long-term chart illustrates the multi-decade rise, plateau and now decline9 in the amount of economic resources dedicated to residential housing, which accounts for 70-80% of total floorspace built. Other related metrics like construction employment and real estate employment also corroborate these long-term trends10.

Property and infrastructure suppressed GDP growth since 2010.

Among others, major SOE reforms in the 1990s (抓大放小 or “grasping the big, letting go of the small”) and China’s admission to the WTO in December 2001 set the stage for the heady days of double-digit annual GDP growth in the mid-2000s.

Much of this rapid growth was fueled by increasing investment in the business sector, and in particular the export processing industries in the coastal regions. In the charts below we can see how “business investment” (yellow line) rose from ~14% (of GDP) in the late 1990s to a peak of ~22% around 2010.

The Global Financial Crisis hit the export processing sector hard. Virtually overnight, millions of factory workers were unemployed as thousands of factories shut in the Pearl and Yangtze River Deltas. In crisis mode, authorities rapidly approved the largest-ever stimulus plan in Chinese history, authorizing ¥4 trillion of funds in November 2008. These funds flowed through local governments into the property and infrastructure construction.

We can see this in the blue and orange lines above, representing housing and infrastructure investment. From a combined ~15% in the mid-2000s, housing and infrastructure investment nearly doubled to ~28% of GDP by 2016.

These stimulus funds supported GDP growth in the post-GFC period. Without these funds, all of the laid-off workers would not have been able to find work and GDP would have dropped significantly. Instead, by mid-2009, GDP growth had recovered back to nearly 10% again.

While I agree that the 2008-09 Chinese economic stimulus plan supported growth and prevented mass unemployment in the immediate aftermath of the GFC, continued investment in housing and infrastructure ended up lowering aggregate GDP growth from the average rate of ~10% from 2000 to 2011 to ~6% from 2012 to 2023.

This relates back to the mechanics of housing and infrastructure investment described above. The low RoI nature (offset by very long useful lives) of housing and infrastructure investment meant that GDP growth necessarily had to slow down once the investment in these asset classes plateaued around 2012-13. This is just an arithmetic reality when you essentially “crowd out” business investment that regularly generates >20% RoI with housing and infrastructure investment that initially yields returns in the low single-digits.

The offset to lower GDP (a flow concept) was rapid accumulation in capital stock that sits on the national balance sheet (a stock concept). In effect, while GDP slowed down, Chinese people were benefiting from rapidly improving quality and amounts of living space paired with improved public infrastructure like modern highways, metro systems, high-speed rail and airports. If you really think about it, households were the ultimate beneficiaries of all of this housing and infrastructure spending.

Today’s shift from property and infrastructure investment to manufacturing boosts GDP growth

If the shift from high-RoI business investment to low-RoI housing and infrastructure investment in the 2010s suppressed GDP growth, a reversal of that shift should have an opposite effect i.e. boost growth. Surprisingly robust GDP growth numbers, particularly over the last two quarters, may be indicators of this trend.

Every resource (again, we are mainly talking about labor) that shifts from mature “Pareto-optimal” sectors like residential housing and LGFV-funded local infrastructure to “Pareto-improvement” sectors like electric vehicles or clean energy infrastructure represents a boost in RoI.

As the net impact of these resource shifts accumulate, the higher GDP return will build up and should have a boosting effect on GDP growth. Instead of investments in residential housing that only yield 3 to 4% incremental housing consumption the following years, each investment in a new Gigafactory could yield 80%+ returns11 on GDP investment.

I call these advanced manufacturing industries “Productivity Kings” because they leverage R&D and high-value capex to drive production efficiencies that are not only captured by their employees (wage and personnel growth) and capital holders (increases in net profit) but also even more significantly, large consumer surplus that all of society benefits from. That electric vehicles are now just as affordable (relative to average incomes) for Chinese people as they are for rich countries shows how much impact companies like BYD have had on the economy and not just for their shareholders.

Getting the hard part out of the way

Like China, South Korea and Taiwan also underwent rapid urbanization from the 1960s through the 1990s.

The two “East Asian Tigers” also relied fairly heavily on investments during their rapid urbanization phases. South Korea made large business investments (e.g. heavy industries like chemicals, steel and shipbuilding) and Taiwan made large public infrastructure investments to support their small and medium family-run entrepreneurial businesses.

However, China pushed the investment-centric economic development model even farther, investing heavily in housing, public infrastructure and business all at the same time.

Now we know that South Korea and Taiwan were probably the two most successful economic development stories in the 20th century, as I describe in one of my favorite early writings in 2015 “What are the most fascinating stories of nations that went from poor to rich?”

Does China stretching the investment-centric development model take it one step too far? I do not think anyone can say for certain because it will depend entirely on how well these built-up housing and infrastructure assets sitting on the balance sheet perform over the next 50-100 years of their useful lives. Nobody has a crystal ball that can tell you how those housing blocks in Changsha will be utilized between now and the 2070s.

However, what I can say with a high degree of confidence today is that China has better housing and infrastructure than South Korea and Taiwan did when they were at similar stages of economic development. All of that housing and infrastructure investment enable the average Chinese person to enjoy higher quality modern housing and better infrastructure than the average Taiwanese person in the 1980s or Korean person in the 1990s.

I can attest to this, having spent an aggregate of three years living in Taiwan over the last decade. Taiwanese authorities decided to eschew large and fancy housing as a matter of policy under the 1990s after Martial Law ended. The net result is that today, many parts of Taipei are still fairly “shabby” (at least on the exterior) and most homes are quite small. I biked around Taipei during the 2021 lockdown and these photos are quite representative of the typical neighborhoods.

In a sense, one can actually view heavy investments in housing and infrastructure as fundamentally aimed to benefit households. They represent consumption choices. Much of that consumption is deferred (because of the long useful life) but it does not change how it is fundamentally a consumption-oriented policy choice.

Beijing’s multi-decade urbanization policy and its large emphasis on low RoI housing and infrastructure investment represented sacrificing near and medium-term GDP growth for the future consumption as these assets rise in value with household incomes decades in the future benefiting multiple generations. In effect, for those who tend to focus heavily on GDP, it got “the hard part out of the way” and now the economy can focus on putting more resources to work in more “exciting” higher-RoI activities like advanced manufacturing and high-tech services that are critical to fully developing such a large economy.

10

Source: Rogoff & Yang (2020) “Peak China Housing” Figure 6. Contribution of Real Estate and Construction to Urban Employment.

“Given that the stock market hasn’t been especially rewarding to the volume-over-profits strategy undertaken by big Chinese manufacturers, what policy levers does Beijing have to sustain and encourage the existing approach?”

Many people may have noticed that despite the impressive growth of Chinese manufacturers in sectors like electric vehicles, the market capitalizations of these companies are dwarfed by Tesla. This seeming paradox lies at the heart of the the question posed by Joe.

In 2020, I shared an observation that China cares much more about GDP than market capitalization. I was making this observation in the context of Alibaba1 but would soon broaden the observation to encapsulate many more situations. In sharp contrast to Americans, Beijing just does not seem to care that much about equity market valuations but do seem to very much care about domestic growth and economic development.

This is quite an enigma because many (if not most) of us grew up equating the stock market with the health of the underlying economy. Indeed, renowned capitalist Warren Buffett, who is by some barometers more celebrated in China than America, is known for the “Buffett Indicator” that explicitly ties stock market capitalization to GDP. In his annual letters, he has repeatedly talked about his confidence in the stock market on the basis of the last two centuries of continuous economic development and growth in America.

Those who know me know I am a huge fan of Warren Buffett and Berkshire Hathaway. One area where I have evolved my opinion with respect to China over the last decade is that key elements of his “value investing” style just would not work, particularly as Berkshire Hathaway has accumulated an ever-increasing capital base2.

I have been pondering various iterations of this question for much of the last decade. While it might seem like a simple question, a proper answer is necessarily complex and requires a multi-faceted approach involving not just economics but other disciplines including accounting and philosophy.

The particular approach I will take here today to analyze this question is to break this down into more finite, digestible components and consider how Chinese policy impacts each of these components.

• “Stock market” a.k.a. Market cap = Profits (to equity) x Market multiple

• Profits = Volume x Profits per unit

• Profits per unit is a function of revenue and operating expenses, in particular fixed overhead and variable costs

• Profits to equity = Profits less return on debt (interest expenses) and taxes

• Market multiple = A function of many factors, including at times (it seems) magic

Subscribe now

The positive effects of volume on profit

In the traditional manufacturing model, profits are a function of the residual after fixed overhead (including capex via depreciation, factory overhead and R&D) and variable costs. Fixed overhead unit cost are ultimately a function of volume: the more volume, the larger unit base on which a company can amortize fixed costs.

For fixed overhead, volume is your friend.

Meanwhile, variable costs will be line items like energy and materials inputs, third-party component costs and retail/distribution3. In the past, with labor-intensive manufacturing, labor was a variable input but for advanced manufacturing it has shrunk.

All things equal, scale can drive profit growth. If a company has fixed overhead of $1,000 and sells 10 widgets, its fixed overhead per widget is $100. If it sells 100 widgets, fixed overhead per widget declines tenfold and all things equal, profit per widget rises commensurately.

“Scale matters” in manufacturing and the natural instinct for Chinese manufacturing companies is the rapid pursuit of scale to establish a defensible market position.

One company’s variable cost is another’s fixed overhead

In a complex manufacturing supply chain, costs that are variable to the purchasing department of a manufacturer represent the revenue of an upstream supplier. For example, BYD runs a ~20% gross margin which factors in both fixed overhead and variable costs. It purchases large amounts of aluminum from upstream aluminum suppliers (e.g. Chalco) that would be counted as variable costs on its income statement.

If we now turn our attention to Chalco’s income statement, we will find that it also has its own fixed and variable overhead structure, which we can break down similarly to how we have done above. We can continue to recursively make our way farther and farther upstream in the supply chain involving Tertiary (mainly services) to Secondary (mainly manufacturing) to Primary (mainly extractive) sectors.

Ultimately, we will discover that the majority of “operating costs” in society are simply labor costs. There is a question of how labor cost is categorized: either as expenditures (consumed within a year) or counted as gross capital formation and consumed over time. On top of this we also have separate categories for corporate income and government expenditures. In effect, this is a reconstruction the income approach to calculated gross domestic production (GDP), shown here on the right-hand side:

The trade-off between current profit and future growth

When a manufacturer generates incremental profit — through any combination of various approachs volume growth, production efficiencies, more leverage over suppliers, etc. — an important decision is whether to return these profits to capital holders (dividends or stock buybacks) or reinvest back into the business (retained earnings) presumably with the rational expectation of some kind of incremental return on investment.

Increasingly as manufacturing has become more automated in China, this means more reinvestment takes the form of line items like R&D and marketing (brand-building) instead of traditional capitalized expenditures like factory equipment. The accounting affect is delayed gratification: current-period operating profits are sacrificed for growth in future profits.

Once again turning to BYD as an example, in 2023 it increased its number of research personnel by over 50%. This increased R&D investment represents a massive reinvestment of earnings back into future growth — improved car models, less expensive batteries, expansion in manufacturing capacity.

The ideas captured in the “overcapacity is a feature, not a bug” phrase are also relevant here. Rapid expansion in manufacturing capacity requiring large amounts of expenditures can depress financial profits, even if most of those expenditures are capitalized as factory assets. Depreciation of this factory equipment is often mismatched with the actual underlying usage of the equipment over its entire useful life. This can result in depressed earnings, especially as factories are still in ramp-up mode.

While all of this should be theoretically reflected in the market multiple (in the form of higher growth expectations on profits to equity owners), as I will discuss later on, this is not the only factor that determines valuations in the public markets.

Labor vs. capital

In society, a balance must be struck in how economic gains are split between labor and capital (i.e. profits on capital investment). On the one hand, it is zero-sum because $1 of economic gain can only be split it between labor and return on capital. It is non-zero-sum in how this split might ultimately impact long-term growth of those economic gains.

Compared to Western capitalist nations, in China the general policy orientation is to favor labor over capital, especially privately owned equity capital (vs. state-owned equity such as state-owned enterprises). You can see this in how so many policies favor labor, especially lower-income labor, for example:

• Foreign exchange policies that encourage weak currency, which makes Chinese labor cheaper compared to foreign labor on a productivity-adjusted basis

• Policies to encourage a relatively labor-intensive manufacturing model as the basis of industrialization

• Heavy stimulus programs during the Global Financial Crisis (GFC) in reaction to the sudden loss of millions of jobs in the export processing sector

Prioritization of labor encompasses more than just rising wages and incomes for the gainfully employed, but spillover knowledge benefits that contribute to building up the hard-to-measure but significant human capital asset on the national balance sheet.

The question of balance is important. Do you favor labor over capital so much that private entrepreneurs become less motivated? This is a huge, extremely complex and nuanced question that one can argue is more philosophical than economics.

But we can look at this empirically by watching the behavior of private entrepreneurs as well: China has notably cracked down on some areas of the private sector (like Internet and Fintech) while encouraging others (like advanced manufacturing). Its policies have had a notable impact on growth (or lack thereof) in these respective areas.

Debt vs. equity financing

One of major differences between China and the United States is how long-term investments in the economy are financed. In China, the dominant financing approach for capital-intensive industries is debt, mainly funneled through the state-controlled banking system:

• Out of Total Social Financing representing ~294% of GDP in 2023, only 3% was accounted for by equity financing. Even though this figure does not include all forms of equity4 it still shows how much more important debt is than equity financing in China.

• The dominant household asset class in China is equity held in real estate instead of the stock market or other liquid assets.

The stock market is an equity concept. The implication here with the primacy of debt over equity is that the performance of the domestic stock market is much less indicative of the underlying health of the overall economy5. For these reasons, the “Buffett Indicator” is largely irrelevant for China.

State-owned enterprises (SOEs) also tend to play a disproportionate role in industry sectors that are by nature capital-intensive. As a result, they are also by far the largest users of debt. Tight state control over the state banking system (household and private sector business deposits) allow the banking sector to provide financing at a relatively low cost, especially compared to developing nations. For capital-intensive industries, this is an advantage.

Chinese manufacturing companies tend to be more private sector oriented and less capital-intensive business models compared to industries SOEs tend to operate in. For example, BYD’s is mainly capitalized with equity and does not feature significant long-term borrowings despite investing increasingly large amounts into expanding manufacturing capacity.

Mr. Market and market multiples

To borrow from another Warren Buffett favorite, “Mr. Market” was an allegory created by his mentor Benjamin Graham to characterize a stock market that at times seemed irrational and temperamental. One key lesson of this allegory was to recognize groupthink in the markets and “to be greedy when others are fearful and to be fearful when others are greedy”.

Much of Mr. Market’s temperament can be captured in a variety of market multiples that investment banking analysts develop a lifelong passion for calculating in increasingly arcane formulations:

• Price to earnings

• Price-to-book value

• EV-to-unleveraged-FCF, EV/EBITDA, EV/Revenue

• And my favorite: Wework’s “community adjusted” EV/EBITDA where adjustments include adding back opex (both real and imagined) 6

All things equal, high multiples mean that Mr. Market is happy and exuberant while low multiples indicate a sadder, more despondent version.

Of course, temperament is not the only factor. Interest rates (opportunity cost of capital), risk perception, economic growth (and how it translates into growth in profits to shareholders), tax policy (how operating profits are shared between the government and business) etc. can all play a role in driving market multiples.

Except for perhaps a brief period before the GFC (when PetroChina A-shares drove an implied market capitalization of over a $1 trillion), Chinese market multiples have tended to be much lower than U.S. equity market multiples. And this gap has only been increasing over the last decade.

Even a company like BYD which has been growing faster than Tesla tends to have a lower market multiple. Or compare a company like Alibaba that is trading at veritable early Buffett “cigar butt” levels to Amazon.

While the “Buffett Indicator” might help with market timing and provide some measure of how “cheap” or “expensive” the market may be, Warren Buffett himself pays attention more to underlying earnings and even Berkshire Hathaway’s book-value per share (BVPS) than the more volatile market capitalization of Berkshire Hathaway which is subject to the changing emotions of Mr. Market.

For the first four-plus decades of of his annual letters7 to shareholders of the former textile mill, he exhorted investors to pay attention to Berkshire Hathaway’s BVPS in favor of its stock price — not because BVPS was a more accurate formulation of the more abstract and difficult-to-precisely-measure “intrinsic value” concept, but because it was a better indication of how intrinsic value was trending.

This actually illustrates an important and related concept — that the “mood swings” of Mr. Market should not have a meaningful effect on the underlying operations of the business, better captured by measures like annual operating earnings (but even then subject to the limitations of accounting described above). In short, stock market valuation at a given point in time could very well be driven by factors completely divorced from the underlying performance of an economy, especially one that (per above) relies much more on debt financing anyway.

The “Dungeon Master”

With respect to private sector market forces, Beijing tends to see its role as coordinators of an elaborate “game” that is meant to foster industry dynamics that drive desired market behaviors. The metaphor I sometimes use is as the Dungeon Master role in Dungeons & Dragons8.

These “desired market behaviors” tend to overwhelmingly revolve around this focused effort to maximize economic development and growth over multiple decades. Beijing has been very consistent about the goal to become “fully developed” by the middle of the 21st century9.

To date, I would say that Chinese policymakers have been relatively successful using the approaches and principles described above to drive economic growth:

• Priority on labor over capital / wage growth over capital income growth. Prioritizing labor is a key pillar of China’s demand-side support strategy. Growth in household income drives growth in domestic demand (whether in the form of household gross capital formation or expenditures).

• Setting up rules to foster the create competitive industry dynamics and motivate economic actors to reinvest earnings back into growth.

• Periodic crackdowns to disrupt what is perceived to be rent-seeking behavior, particularly from private sector players that have accumulated large amounts of equity capital (vs. small family businesses):

  • Anti-competitive behavior (e.g. Alibaba e-commerce dominance in the late 2010s)

  • Regulatory arbitrage (moral hazards inherent in Ant Financial’s risk-sharing arrangement with SOE banks)

  • Societal effects (for-profit education driving “standing on tiptoes” approach to childhood education)

• Supply-side support to encourage dynamic, entrepreneurial participation from private sector players like in the clean energy transition to drive rapid industry through scale and scale-related production efficiencies. China has relied on supply-side strategies to support economic for decades despite repeated exhortations by outsiders to implement OECD-style income transfers.

• Encouraging industry consolidation (vs. long drawn-out bankruptcies) once sectors have reached maturity although there are often conflicting motivations between Beijing and local governments.

A consistent theme is Beijing’s paranoia to rent-seeking behavior by capitalists (especially those who have accumulated large amounts of capital). It is sensitive to the potential stakeholder misalignment when capitalists — who are primarily aligned with one stakeholder class (fiduciary duty to equity owners).

It would prefer that rent-seeking behavior be handled by the party instead, whose objective (at least in theory) is to distribute these rents back to “The People” — although naturally in practice it never turns out this way; Yuen Yuen Ang has written multiple volumes about the prevalence of Chinese-style corruption and its corrosive economic effects.

So to bring it back to Joe’s question, the answer on whether Chinese policymakers can continue these policies going forward very much revolves around public goods and rent-seeking: is it better to be handled by the government or by private sector capitalists? What should be abundantly clear is that Beijing is definitive on this question: the party will maintain a tight grip, if not outright monopoly, on rent-seeking.

Will this dampen the “animal spirits” of entrepreneurs and private capital that play such a critical role in focus industries like the clean energy? So far it does not seem to have slowed them down much at all — quite the opposite in fact. This is despite financial returns on capital not being particularly high, often quite volatile / difficult to predict, and Mr. Market often assigning market multiples that are at times indicative of extreme levels of depression.

This remains a puzzle and paradox and something that we can only observe over long periods of time to truly judge whether this was ultimately the right approach for the “Dungeon Masters” in Beijing to take.

O beautiful for spacious skies,
For amber waves of grain,
For purple mountain majesties
Above the fruited plain!
America! America!
God shed His grace on thee
And crown thy good with brotherhood
From sea to shining sea!

Having spent my entire life on the East coast, in my early 20s I finally had a chance to truly understand what those words meant.

Following the footsteps of my father when he first arrived in North America in the 1960s, in June 2001 two buddies and I set off on a classic post-college graduation roadtrip. We started in San Francisco and drove south to Nevada, Arizona and New Mexico and then all the way to Houston, Texas. We then drove north through Texas, Oklahoma and Colorado until we reached the Rockies. After driving up through Wyoming, we veered west through Idaho and Nevada and back to San Francisco.

Similar to song composer Katharine Lee Bates more than a century earlier, I was awestruck by the natural beauty and splendor of the American West.

Interestingly, in the song composer Katharine Lee Bates may have unwittingly embedded lessons about certain structural conditions of this country by identifying key areas of natural economic advantage:

• “Amber waves of grain” and the “fruited plain” — bountiful land that has fed the nation for centuries and turned the country into an agricultural powerhouse

• “Purpled mountains majesty” — enormous reserves of shale oil and gas (not to mention minerals) sitting beneath the Rocky Mountains

Buried deep in the poem’s fourth stanza, Bates also references gleaming “alabaster cities”. This was a reference to the growth of cities like Chicago1 and represented the rise of man-made industry and manufacturing in the late 19th century. In the 20th century, these natural and man-made advantages combined to create immense absolute advantage, resulting in the most powerful nation the world has ever seen, peaking at one-third of global GDP at the end of World War II.

Following the war, the rest of the world has been in catch-up mode. First, with the reconstruction of war-torn Europe and Japan. Then with the rise of former colonies in Asia, Latin America and Africa in the post-colonial era. Finally, over the past three to four decades, the rapid rise of continent-sized economies in China and India.

To better understand these long-term trends, it has become increasingly important to understand comparative advantage, how it has changed over time and how it contributes to changes in absolute advantage over the long run.

While references to “amber waves of grain” and “purple mountains majesty” are meant to imbue in us a sense of enormous gratitude for the natural resource endowments of this country, in the long run what will matter more is that less well-known reference to “alabaster cities”. In school, we never quite got to the fourth verse of “America the Beautiful” so this is something I had to look this up:

O beautiful for patriot dream
That sees beyond the years
Thine alabaster cities gleam
Undimmed by human tears!
America! America!
God shed His grace on thee
And crown thy good with brotherhood
From sea to shining sea!

The potential downside of such bountiful natural resource endowments is how it can create comparative disadvantages in manufacturing. In this essay, I will discuss two examples of how bountiful land and resources have contributed to the relative decline in global competitiveness of our manufacturing industries.

Subscribe now

Corn-based ethanol and the “Paradox of Plenty”

I have been a huge baseball fan ever since I got my first Nolan Ryan baseball card2 sometime in the third grade. Perhaps inspired by Field of Dreams, one of my childhood wishes was to one day take a summer roadtrip across the country, stopping at small towns and cities in the Midwest like Peoria and Cedar Rapids to watch recently drafted high school and college baseball players start their professional careers3. Like in the movie, I imagined baseball fields surrounded by endless rows of corn extending deep into the horizon.

While “amber waves of grain” tends to be associated with wheat, corn4 is actually America’s largest agricultural product, accounting for about one-third of agricultural output. Corn is an incredibly versatile vegetable/grain and can be used in natural form as corn-on-the-cob5 to a multitude of uses including cornstarch, syrup, cooking oil and many others. It can also be converted into ethanol and used as a fuel to supplement gasoline.

Ethanol was first used in 1826 to power an engine. It fueled Ford’s original Model T in the early 1900s and was used during World War II with the war effort and allies needing all the gasoline that could be produced. However, the modern ethanol industry did not really take off until the 1970s and was driven by two key factors:

• The formation of the OPEC cartel leading to rapid increases in the price of oil

• Environmental concerns leading to the elimination of leaded gasoline, which was a key additive to increase octane6

High oil prices drove industry to explore alternative fuels like ethanol. Federal and state subsidies encouraged the use of higher-octane ethanol to boost octane levels in gasoline. Indeed, taxation and industrial policy have played a key role in fostering the development of the corn-based ethanol industry over the years. For example, one of the objectives of the “Minnesota Model” for ethanol production was to boost profits and create jobs in the community by adding value to agriculture products.

For a long time, ethanol was an interesting case study of how the country’s natural advantage in arable land translated into economic benefits and advantage. In the early 2000s, when we were still worried about energy independence, ethanol was touted as a mitigant to reliance on imported oil from the Middle East. The Energy Policy Act of 2005 increased targets of blending ethanol use with gasoline and the Energy Independence and Security Act of 2007 raised targets to 36 billion gallons by 2022.

But rapid improvements in solar photovoltaic (PV) and battery technology and costs in recent years — especially since the late 2000s — have not only negated this advantage but possibly turned it into a crutch in our difficulties developing domestic clean energy manufacturing. At the heart of the problem is the enormous amount of land required to generate corn-based ethanol.

To compare solar to ethanol, we must examine corn-based ethanol production at a more fundamental level as the conversion of solar energy into useful chemical energy or other forms like electricity:

• With solar, sunlight is absorbed by the PV cells, causing electrons to move, creating electricity7. This electricity must then be used right away by the grid or or put into longer-term storage, typically in the form of chemical energy (batteries or conversion to a fuel like green hydrogen) or gravitational (pumped storage).

• With corn-based ethanol, sunlight is absorbed by corn through photosynthesis and stored in various sugars (chemical energy). The corn is harvested and processed, typically in a dry mill. The processed corn flour is then fermented, which converts sugars into more complex hydrocarbons like ethanol8.

As you can see, the ethanol production process is significantly more complex and also involves energy inputs at each stage to create the final output. Energy Return on Investment (ERoI) measures the ratio of energy inputs vs. energy outputs. Corn-based ethanol is at 1.2, meaning for every unit of energy you put in9, you get 1.2 units of energy back. Solar PV is 8.10 The other downside to “clean” corn-based ethanol is that the energy inputs are typically petroleum-based fossil fuels.

Further, the transition to electric vehicles only further exacerbates these already extraordinary differences in efficiency. Because there is so much energy lost to heat during the combustion process, traditional ICE vehicles are far less efficient at converting stored energy into useful work (movement) than electric vehicles:

This means the differences in land use are yet another order-of-magnitude greater than ERoI. According to this report from Clean Wisconsin, “net energy production per acre is 100-125x greater for solar PV than for corn-based ethanol”11.

Countries like China suffering from relative land scarcity could not pursue corn-based ethanol — they needed all the arable land to produce food. Instead, it was forced by these natural endowment disadvantages to come up with more creative solutions to solve the problem of solar energy capture.

The “Paradox of Plenty” is a term typically used with poor or developing countries that are blessed with natural resources but puzzlingly have lower development outcomes. Obviously, with the highest per capita GDP among large countries, the U.S. does not exactly fit this profile. But there are some indications that over-abundance of land may have contributed to the United States — once the pioneering leader — falling behind in solar PV and other clean energy industries.

For example, given massive differences in ERoI and land efficiency, one might think the obvious thing to do is completely shift federal and state support away from corn-based ethanol to solar, battery and other related clean energy manufacturing. However, there are strong corn-based farming constituencies in a number of important swing states like Iowa, Minnesota and Wisconsin that are highly motivated to maintain ongoing federal and state-level support for corn-based ethanol12. This slows down the transition to clean energy.

In other words, our natural advantages in land led to the establishment of the corn-based ethanol processing industry, which is well-positioned politically to reinforce this advantage. Because comparative advantage is relative, this necessarily widens our comparative disadvantages in manufacturing needed to produce solar PV and other clean energy technologies competitively.

The fracking boom and “Dutch Disease”

The Economist coined the term “Dutch Disease” to describe how the Dutch economy changed after the discovery of natural gas fields in the northeastern part of the country in 1959. It describes how significant natural resource windfalls can shift the comparative advantages of an economy, making other parts of the economy less competitive. Much of this impact was brought about by a rising currency, the Dutch Guilder, which increased the cost of Dutch labor relative to the the rest of the world, which (all things being equal) decreases its competitiveness. This had a negative impact on the country’s manufacturing sector.

The U.S. has also recently had its own resources boom. As I describe here in this 2018 essay, the advent of new “hydraulic fracturing” techniques (“fracking”13) in the Barnett Shale by American wildcatters unlocked immense shale oil and gas reserves, pushing what had once been the world’s largest buyout (TXU) into bankruptcy and contributing to a rare black eye for Warren Buffett and Berkshire Hathaway.

More significantly, it turned the United States from a major oil importer14 to the largest oil producer on the planet. We became a net oil exporter in late 2019 and have not looked back:

This has been one of the most under-appreciated economic stories of the last two decades. Following the Global Financial Crisis (GFC), the huge trade deficit shrank as Americans underwent a deep and painful recession to cut back on consumption. However, as we progressed through the 2010s, this necessary change in consumption habits was in part alleviated by rising onshore oil production fueled by fracking.

This cushioned the full impact of the GFC and helped America recover faster than Europe, allowing Americans to resume its old consumption habits without significantly increasing the trade deficit, as illustrated by the green area charts below from 2010 to 2020.

As you can also see, the trade deficit ballooned following the global pandemic as households shifted spending on domestic-focused services to foreign-produced (predominantly China) consumer goods. Normally when a country’s trade deficit rises for an extended period of time, its currency will shrink. However, the opposite happened with the U.S. dollar. It strengthened:

These are classic signs of “Dutch Disease”.

The fracking boom attracted rising investment into the oil and gas sector, and dominated growth in non-residential fixed investment after the GFC until recently. Petroleum engineers and wildcatters commanded high salaries, allowing the energy sector to attract a disproportionate share of “hard/deep tech” science, engineering and vocational talent, pulling human capital away from traditional manufacturing sectors. Meanwhile, a rising dollar made those traditional manufacturing sectors even less competitive in global markets.

In other words, like corn-based ethanol, fracking has also shifted our comparative advantages away from general manufacturing. We can see this when we “peel the onion” on the trade deficit and focus on manufactured goods, which after a sharp recovery following the GFC, have been in a consistent downward spiral (blue line below). Fracking masked the decline in the topline numbers for a period of time until it became noticeable with the pandemic-induced demand spike in manufactured goods15 .

Meanwhile, major exports like travel and education were cut off overnight and the cratering of oil prices, which would have helped in the past, were now a slight negative with the country having flipped into a modest oil exporter. As a result, the overall trade deficit spiked to all-time high levels before moderating with the re-opening of the country in 2022.

Comparative disadvantage and the long decline of American manufacturing competitiveness

Fracking-induced “Dutch Disease” is only the latest in what has been a long decline in American manufacturing dating back to the 1970s, as we can see when we go farther back in Brad’s chart above. The post-war rise of Japan, particularly its automobile sector, pushed what had been a surplus in manufactured goods to a deficit.

The machine tools industry was the “canary in the coalmine”, collapsing in the 1980s. The 1985 Plaza Accord forced rapid appreciation of the Japanese Yen and temporarily reversed the decline. But the rise of the “East Asian Tigers” and later, China (which joined the WTO in December 2001), continued to exert pressure on the trade deficit. As a result, we lost critical mass in our supply chains and manufacturing human capital.

In the past, our immense absolute advantages were enough to keep our comparatively disadvantaged manufacturing sector globally competitive. In the post-war period, American multinationals had the most advanced technology and intact manufacturing capital base by a wide margin, which allowed the nation to sustain a quality life that was significantly above the rest of the world, not to mention supporting European allies with the Marshall Plan.

Over the decades, as others developed and our absolute advantages shrank, the effects of comparative disadvantage in our manufacturing sector became increasingly exposed to global competition. Margin of safety provided by overwhelming technological and capital advantage eroded until it was no longer enough to overcome structural disadvantages like high labor wages, an inefficient healthcare system, and higher regulatory burdens. That offshoring manufacturing to East Asia and IT/BPO services to India and the Philippines proved to be so attractive in the 90s and 2000s reflected these trade realities.

To stem further decline of our manufacturing sector, we need to think more in terms of comparative advantages. This means that we need to change fundamental aspects of our economy to shift comparative advantages away from natural-resources intensive sectors like energy back to manufacturing. If we want manufacturing to truly be globally competitive, we need to make real structural changes, the kind that hurt. If they do not hurt, it probably just means they weren’t enough.

The CHIPS Act and IRA are a good start, but much more is required to combat both the natural advantages of resource endowments we sing about in “America the Beautiful” as well as active efforts by other countries that are actively working to shift their own comparative advantages towards manufacturing.

This is not going to be easy and will take decades of persistent effort to put into effect, because part of this requires training a new generation of American manufacturing workers. But if we want to retain our economic lead over the world, we will need to figure out a way to bend the comparative advantage spectrum in factor of manufacturing.

8

Department of Energy: Ethanol Production

This description simplifies what is actually a much more complicated process, as shown in the diagram below from Nebraska Corn Processing LLC:

11

Based on multiple analyses of land use:

Here, in a recent report whose title “Overcapacity at the Gate” pays homage to the leveraged buyout classic1, research firm Rhodium writes:

China’s National People’s Congress concluded in March 2024 with an explicit focus on industrial policy favoring high-technology industries, and very little fiscal support for household consumption. This policy mix will compound the growing imbalance between domestic supply and demand. Systemic bias toward supporting producers rather than households or consumers allows Chinese firms to ramp up production despite low margins, without the fear of bankruptcy that constrains firms in market economies.

In this piece, I want to focus on “overcapacity” in the auto industry. First we need to discuss exactly what “overcapacity” means.

“Overcapacity is a Feature, Not a Bug”

If you follow me on Twitter/X, you may have noticed that I have been somewhat of a broken record2 on this phrase because many do not seem to fully understand what “overcapacity” means in a China context and to Chinese policymakers3.

Traditionally, the term “overcapacity” was usually used when referring to mature industries. Rhodium describes the term here:

The simplest and most widely accepted definition of overcapacity is when factories’ production capacity is under-utilized. While temporary overcapacity can be harmless and a normal part of market cycles, it becomes a problem when it is sustained through government intervention. Structural overcapacity happens when companies maintain or grow their unused capacity without worrying about making a profit (or a loss), often due to a lack of economic pressure to operate efficiently, like a hard budget constraint.

Chinese policymakers often consider “overcapacity”4 in the context of nascent, developing industries like electric vehicles.

In his book 置身事内：中国政府与经济发展 (English translation entitled “How China Works: An Introduction to China’s State-led Economic Development”)5 Lan Xiaohuan, a professor at the School of Economics at Fudan University, discusses how “overcapacity” can help a new industry get started from scratch and why it is not perceived as a risk in the same way others do6. Pekingnology provides an excellent translation of the passages here.

1. First, “overcapacity” is less of a risk at the early stages of any industry, especially for developing countries where adoption curves are highly predictable, providing comfort that demand will grow into the new capacity.

2. Second, there are learning spillovers that can lead to the accumulation of human capital.

3. Third, done right “overcapacity” creates competitive conditions that can ultimately drive production efficiencies.

There is an element of creative destruction that is reminiscent of the ethos that has driven Silicon Valley to become one of the most productive regions on this planet.

On the flipside, Chinese policymakers themselves are quite sensitive to overcapacity7 in mature industries. Following a buildup of industrial capacity following the 2008 GFC8, in 2016, Chinese policymakers enacted a series of policies to reduce excess capacity in the steel and coal industries9 that proved to be effective10.

Here we can see how China’s industrial capacity utilization rate (black line) rebounded from lows in late 2015.

The ICE-to-EV transition and China’s evolving role as a car exporter

Of particular concern to Western automakers have been rising exports of Chinese automobiles. These have risen significantly in recent years to 5 million in 2023, making China the largest vehicle exporter in the world, surpassing Japan. Some conflate rising exports with “overcapacity” and “weak demand”.

Auto manufacturing is a geopolitically sensitive industry — anybody who was around in the 1980s remembers the backlash against Japanese automobiles11. Four decades later we are going through a similar cycle with Chinese cars.

There are a number of factors behind rising Chinese exports but the most impactful is transport electrification. Rapid rise in NEV adoption in China has been driven by the combination of feature improvement and price reduction that make them very competitive with ICE vehicles.

However, not all car exports are created equal and there is more to this story than just transport electrification. This requires digging deeper into the numbers.

First, we need to focus on the passenger car segment. Out of the 4.9 million vehicles exported in 2023, 769k were commercial vehicles like buses, tractor trailers and light to heavy trucks. Many of these commercial vehicles are sold into markets like Russia/ CIS, the Middle East, and Southeast Asia and are less of a concern to incumbent automakers.

This leaves approximately 4.1 million passenger cars. Out of this number, we can further segment this out by ICE (3.0 million) and NEV (1.1 million) car exports. Each of these categories is driven by a different set of dynamics.

Passenger ICE vehicle exports

ICE vehicle exports have been boosted from rapid consumer adoption of NEVs. Within the ICE market, there is a bifurcation between Chinese brands and foreign brands manufactured through joint ventures.

As domestic market share has dwindled, Chinese ICE brands like Chery are finding success diverting some of the newly freed up capacity towards export markets, primarily in developing countries.

However, foreign JVs were not originally structured with the intention of exporting and have been bearing the brunt of declining capacity utilization. Many foreign automakers that do not have competitive NEV offerings are opting to exit the China market. A recent example was Mitsubishi which gave up its share of its JV to its partner GAC Aion last October.

Meanwhile, ICE factories are being converted into NEV factories. Following the shutdown of the joint venture, GAC Aion has converted former ICE manufacturing facilities into NEV production, even expanding production capacity by 50%. It takes time to build a “greenfield” NEV factory and starting with an existing “brownfield” facility can shorten lead times. Expect more consolidation of ICE factories in the coming years, which will result in the absorption of excess capacity.

This excess ICE capacity does not fit the definition of “structural overcapacity” defined earlier in the Rhodium report. It was added mainly by foreign joint ventures where there was a profit expectation and shareholder “pressure to operate efficiently”. The investment decisions were made by economic actors subject to Kornai’s “hard-budget constraint”.

When these plants were built, very few expected the speed at which they would be made obsolete by NEVs. Things do not always turn out exactly as one expects at the time investments are made; this happens in many industry sectors and is a normal part of doing business. The key here is that this excess ICE capacity has a limited lifespan and will be absorbed as quickly as automakers can setup new NEV lines.

Passenger NEV exports

For many, the most concerning aspect of rising Chinese exports are NEVs — as they represent a direct threat to incumbent automakers. People see the decimation of the ICE industry in China and very easily imagine the same thing happening in their own sectors12.

China’s NEV exports generally fall into two camps. First, there are exports of higher-value foreign brands that are produced in China and exported abroad, mainly to Europe13. This includes Tesla, MG14, Volkswagen, BMW, Mercedes-Benz and Renault. Second, some Chinese brands like BYD and Geely are just starting to explore export markets.

It is notable here that the foreign brands are relying on China as an export base more than Chinese brands. For example, last year, Tesla exported 36% of its China production. In contrast, only 8% of BYD cars were exported15.

It indicates how Chinese car companies are more focused on the domestic market for their NEV offerings. Unlike the export-oriented industrialization boom of two decades ago, Chinese domestic demand is supplying the energy to drive its advanced industries. Compare this to the auto manufacturing industries in Germany, Japan and South Korea, where the majority of domestic production is exported.

It also means that exports by foreign brands are naturally displaced as they open up their own local manufacturing facilities — for example, NEV exports to Europe declined in the first two months of the year, likely due to the ramping up of Gigafactory Berlin which began operation in 2022 and has been gradually ramping up production.

Some also conflate rising Chinese NEV exports as indications of “weak domestic demand” and “overcapacity”. However, export demand is real demand — Chinese NEVs are selling overseas because of their competitiveness; they are typically sold at much higher prices16 offshore compared to onshore.

Chinese automakers are also aggressively investing in manufacturing facilities abroad. Again in sharp contrast with the export-oriented industrialization boom from the 2000s, Chinese automakers are focusing on developing markets instead of trying to fight rising protectionism in developed economies. BYD is building local manufacturing facilities in Thailand and Brazil. Chery just announced a plant in Vietnam. As these manufacturing facilities ramp up, NEV exports from China will be replaced by locally manufactured vehicles.

Claims of automaking “overcapacity” not supported by the numbers

The hard data does not support the idea that there of significant overcapacity in auto manufacturing.

To the extent there is excess capacity in China’s broad industrial sector, it tends to be in categories that have been hit by the decline in housing construction like cement and glass. As you can see (black line below), auto manufacturing has actually been ticking upwards after bottoming out in the middle of 2022 during Zero-COVID.

Claims that China has upwards of 48.7 million17 in annual automaking capacity likely includes significant obsolete capacity in their calculation.

Other measures of overcapacity also corroborate the capacity utilization figures. The turnover of non-current assets (mostly factory assets) for auto manufacturing (yellow line below) has, if anything, been trending upwards over the past two years.

Inventories of Chinese automakers are also at normal levels18. There aren’t tons of cars piling up in dealer showrooms. Indeed, the transition to NEVs is also changing the way cars are sold. One under-appreciated aspect of the transition to electric vehicles has been the use of more direct-to-consumer sales approaches, famously pioneered by Tesla as it sought to disrupt the Big Three oligopoly in the United States.

Xiaomi, which made a name for itself in selling smartphones directly to consumers, recently demonstrated the power of this approach with the launch of the Porsche-esque SU7, racking up 88,898 reservations on the first day of its release for deliveries starting in April and May.

This distribution approach mitigates the amount of inventory required by linking demand directly to production. Moreover, advances in manufacturing19, simpler designs and deepening of the supply chain mean shorter manufacturing lead time and lowers the risk of building up excess capacity in the system. Capacity utilization at NEV automakers in China is generally much higher than traditional ICE manufacturers20.

With the industry moving rapidly from ICE vehicles to NEVs — over half of domestic sales are expected to be NEVs this year — the auto industry’s capacity utilization and working capital requirements are poised to continue improving over time.

China’s car market is right now hands-down the most brutally competitive market in the world with even Tesla’s Q1 sales wilting from the competition. Consumer subsidies for NEVs were phased out at the end of 2022 and after a brief dip, NEV sales have started accelerating once again.

NEVs are taking over because they have long surpassed ICE vehicles in total cost of ownership and the brutal competition continues to wring out impressive gains in production efficiencies throughout the supply chain. Lower prices unlock new sources of demand and drive further market growth.

Indeed, not only are Chinese NEV makers not suffering from overcapacity, they face real constraints of not being able to open up new factories and production capacity fast enough. To merely reach Taiwan levels of vehicle penetration (~350 per 1,000)21 over the next decade, China would have to increase automaking capacity by 3-4 million per year, every single year22. This is roughly the equivalent of South Korea’s entire domestic production capacity and Japan’s domestic sales.

Given the huge consumer surplus and quality of life benefits from Chinese households switching over to NEVs, from a long-term perspective it is probably far more accurate to say the industry is woefully under-supplied.

Subscribe now

13

Bloomberg (December 2023)

18

Bloomberg (April 2024) “China’s Overcapacity Is More About Solar and Batteries Than EVs”

20

Bloomberg (April 2024) “China’s Overcapacity Is More About Solar and Batteries Than EVs”

Some consensus is forming on the pros and cons of the U.S. dollar’s status as the world’s reserve reserve currency. Here, Brad Setser who provides some of the most insightful and matter-of-fact commentary and analysis on the sometimes-controversial topic, outlines this dyadic phenomenon:

One of the benefits is having what is essentially a a no-limit credit card that permits the United States to borrow at low rates. In 2019, I wrote about this as part of a piece explaining why the U.S. had such a high trade deficit. But having the world’s “Most Amazing Credit Card” comes with drawbacks, including the ease with which policymakers can stimulate their way out of economic crises like the recent global pandemic.

More significantly, as Brad alludes, this exorbitant privilege also comes with less control over bilateral exchange rates, which has contributed to a long-deteriorating comparative advantage in domestic manufacturing.

Mercantilism and the “East Asian development model” playbook

For many decades, Japan and the “East Asian Tigers” have used this weakness to develop their domestic manufacturing sectors. Japanese industrial policymakers “pioneered” the use of high trade and current account surpluses to turbocharge their manufacturing sectors through the early 80s.

This led to rising trade tensions that eventually spilled out into the “first” trade war between the United States and Japan. Fortunately (for the United States), we had geopolitical leverage over Japan and put it to use, forcing Japan to allow its currency to appreciate in 1985 Plaza Accords.

The “East Asian Tigers” of Taiwan and South Korea followed in Japan’s footsteps. Taiwan has long (and continues to) run one of the most mercantilist economies on the planet with current account surpluses north of 15%. South Korea started generating substantial large trade surpluses after the Asian Financial Crisis. Both relied on exports — underpinned by demand from the United States, in particular — to drive the development of their advanced industries.

Then China came along and continued this now-seemingly standard maneuver of the “East Asian development model” playbook to drive its early industrialization. Of course, the difference here is China’s scale: its population is more than seven times the combined populations of Japan, South Korea and Taiwan.

China’s current account surplus rose to over 10% of GDP on the eve of the Global Financial Crisis (“GFC”). The GFC itself was in part triggered by rising financial imbalances caused by rapidly rising borrowing of U.S. treasury bond and agency debt, which suppressed interest rates with the now well-known downstream effects with subprime securities.

This triggered painful global adjustments with a sharp economic downturn in the developed world and export factories shutting down overnight in China, leading to a large shift in economic resources to its housing and infrastructure sectors.

For the next decade through the end of the 2010s, there was a respite in the growth of the U.S. trade deficit, as you can see with the green shaded area in another one of Brad’s amazing charts:

However, the underlying issue had not really gone away. Moreover, beneath the seemingly calm waters of a stable trade deficit churned one of the most under-appreciated macro factors of the last two decades. The United States was turning from one of the world’s largest oil importers into an exporter:

Declining trade deficits with the oil-exporting countries of the world meant that the deficit in manufactured goods continued to rise during this period, even as China turned inwards to focus more on its domestic economy, contributing to further deterioration of onshore manufacturing. In the most recent iteration of “Dutch Disease”, a beleagured American manufacturing sector had to deal with even greater support for the appreciation of the U.S. dollar.

This came to a head with the Global Pandemic, when we discovered just how much we relied on the supply of manufactured goods from the Asia-Pacific region, especially China. Imports of manufactured goods grew while “exports” of travel and education (two of the largest imports by China) came to a standstill. The trade and balance of payments deficit once again ballooned.

Further exacerbating trade tensions has been the rise of Chinese advanced manufacturers. While everyone was distracted by the pandemic and China’s emerging property woes, the clean energy triumvirate of electric vehicles, batteries and renewables were incubating within the comfortable confines of protective Chinese industrial policy. Now, as we have seen exemplified by rapidly rising vehicle exports, Chinese advanced manufacturers have emerged as ferocious global competitors.

Whither the almighty dollar

Compared to Japan in the 1980s, the United States has minimal leverage with China today. At that time, the U.S. was Japan’s largest car market. Chinese automakers are de facto banned and have minimal presence in the country. China’s domestic markets in most manufactured goods categories are much larger than the United States or even the combined markets of the G-7 in some cases. Notably, the United States has its largest military forces stationed in Japan. Further, frosty geopolitical and trade tensions have made it difficult to even have conversations.

Since it is clearly out of the question to negotiate or cajole China into moving away from a weak RMB policy, what options does that leave us? Other countries that have more control over their currencies can weaken them. That is what has happened with the Japanese Yen in recent years. But our hands are mostly tied because of this exorbitant privilege.

I do not have a good answer here except to say that now seems as good a time as any to consider the pros and cons of keeping this status, with manufacturing competitiveness in the front of our minds.

Previous Archives: 2023 and Prior

Active/Ongoing Threads

• December 2024: Hidden taxes in U.S. society that add friction costs with dubious value-add.

• May 2024: Me versus the AI bots from Dubai.

• May 2024: Ongoing thread of outbound foreign direct investment (FDI) by Chinese companies in EVs, battery, clean energy and other advanced industries

• May 2023: What the heck are Japanese automakers thinking on the ICE-to-EV transition?

• January 2023: Significant differences in the cost of construction/capex is a persistent issue that saps the global competitiveness of the U.S. economy.

December 2024

• December 28: Abundance mindset to human capital development vs. zero-sum “brain drain” approach.

• December 27: Heavy investment in housing and infrastructure investment suppressed growth rates in the 2010s but lower growth rates should not necessarily be equated with bad policy. The trade-off for higher capital intensity was achieving modern living standards earlier and enabling more natural transition to a modern white-collar workforce with the next generation.

• December 26: The economics and financial viability of the next massive Chinese hydroelectric project on the Yarlung Tsangpo River in Tibet.

• December 26: Huawei is leading China’s advanced chip efforts in the background. If China is able to catchup in advanced chip fabrication, historians will look back at the December 2018 arrest of Meng Wanzhou as a tactical and strategic blunder in the tech war.

• December 26: Immigration is good. Immigration fraud is bad. How to fix the H1B program: focus on the IT consultancies.

• December 25: It took time to develop the human capital and technology to enable the rise of a new type of Chinese factory.

• December 24: “Consumption” (a.k.a. “household expenditures”) defined for GDP purposes is not consumption that regular people are familiar with and we should not conflate the two.

• December 23: Designing renewables-centric energy infrastructure and embracing a green future where renewables drive primary energy capture and creation and everything else is some form of storage.

• December 22: The unit economics of Costco soup dumplings.

• December 21: It makes little sense to compare fossil fuels to renewables on the basis of primary energy consumption measured in exajoules.

• December 20: On the reliability of Chinese data sources.

• December 19: Triple whammy of seasonally weak hydro, solar and wind fluctuations puts “China coal-fired generation peaked in July 2023” at risk and a broader discussion of how renewables are changing the dynamics of power grid design with renewables as the primary energy source and all other sources conceptually viewed as some form of a “battery”.

• December 18: It’s not an either-or choice between HSR and air travel. It is a choice between HSR + air travel and conventional rail + air travel.

• December 18: Reversing the “China as Japan in the 1990s” to think about it what Japan at 1990s level of development might have looked like at China scale, particularly in the context of the current U.S.-China trade war and debates around a Plaza-like agreement around capital controls.

• December 18: Hyperbolic journalism.

• December 17: Are trade surplus countries really more vulnerable in a trade war? Are we measuring “trade surplus” using the right metrics?

• December 16: Ideas vs. execution.

• December 14: “Temporal wage arbitrage” of long-lived infrastructure assets like high-speed rail.

• December 14 (bluesky): ~22 billion passengers have taken China’s high-speed rail network since 2009 with follow-up reaction thread on “temporal wage arbitrage”.

• December 14: The relationship between debt and GDP. This is a follow-up to this piece from 2020.

• December 13: Comparative vs. competitive advantage.

• December 12: (Part 1 | Part 2): China’s seafood industry has been driven by aquaculture, not marine capture.

• December 12: Unless autonomous driving changes individual driving preferences in the next decade, Nio CEO William Li is significantly under-forecasting the size of China’s auto market.

• December 12: We do not need to settle the China balance of payments debate today. But if more time passes and evidence of massive, accumulating hidden capital outflows does not surface, the debate will settle itself.

• December 12: What the sudden wind-down and liquidation of Jiyue says about automotive “venture capital with Chinese characteristics”.

• December 11 (bluesky): At the core of the trade/tech war is the reality that the United States and China are becoming less economically compatible. There is nothing wrong with that; we just need to manage the breakup peacefully.

• December 11: Cross-border fund flows are what determine FX movements. Trade flows are just a subset of cross-border fund flows.

• December 10 (Part 1 | Part 2): The success of Chinese batterymakers was not just about “unfair” local factors. The Northvolt bankruptcy illustrates how they have also been able to accumulate significant technology and human capital. For Europe’s car brands to hold onto global market share, they will also need to accumulate more local technology and human capital and build out local supply chains, including battery cell production.

• December 10: Frontloading capital-intensive economic model made sense in China if you consider the demographics of its workforce: building as much physical long-term capital stock as possible before the blue-collar workforce started to decline precipitously in the 2020s.

• December 10: The issue with Michael Pettis’ predictions a decade-plus ago is not whether China eventually had to transition to a less capital-intensive economic model, but being upwards of three decades off on timing. The core framing issue was believing that accounting identities (e.g. “investment as a % of GDP”) are the driver of economic development policy / strategy. Accounting identities are the output and sit downstream of fundamental drivers like demographics.

• December 9: Intel is a cautionary tale for the Western technology sector vis-a-vis rising Chinese competitors.

• December 9 (twitter/x): GM just took a $5 billion writedown on its China operations. But over the years it has generated tens of billions of profits on less than $1 billion of initial registered capital investment in the 1990s. The lesson here is not that GM should not have entered China, but in what it decided to do with those windfall profits.

• December 9: Economies are dynamic, not static: addressing logical inconsistencies from Michael Pettis in framing the trade-in voucher program for home appliances and automobiles and stimulative effects on the economy that can be better visualized using the circulation model.

• December 8: Autos are another example of where cross-border trade data tracked by customs tells an incomplete story. If we look at auto companies by brand nationality, China is still a net importer of foreign brands.

• December 8: We need to conceptualize foreign direct investment (FDI) not in abstract dollar terms but for the necessary human capital, technology and real capital that sits behind it as a critical component of risk mitigation. We can see this in the outsize role that ethnic Chinese diaspora business communities played in investment into mainland China in the early decades of reform and opening up.

• December 6 (bluesky): ASEAN is at the frontlines of a major shift in global manufacturing in the coming decades. China will likely play an outsize role

• December 6: U.S. industrial policy is mainly about reacting to China (external focus) rather than domestic prosperity whereas China's industrial policy is mainly about domestic prosperity (internal focus) than the U.S.

• December 6: Apple drives more value to U.S. stakeholders than Chinese ones. Messing with this relationship will hurt the U.S. more than China.

• December 5 (bluesky): Billy moneyballs Gallium.

• December 4: 90% of gallium is produced as a byproduct of aluminum production. The economics of greenfield gallium mines are just not going to work.

• December 4: Some correlation between engineers in the workforce and manufacturing as a share of GDP.

• December 3: Critique of Gao Shanwei’s recent note (archive) about the missing “47 million” unemployed.

• December 3: Building off David Fishman’s excellent thread on the detailed construction timetable for Hualong One nuclear reactor in Fujian to discuss why China’s scale and focus/specialization combine to make it such a formidable competitor across industry, manufacturing and tech.

• December 2: The difference between “flow” (GDP / production) and “stock” (national balance sheet / wealth) is why even if countries like China catch up in productivity, it will still need decades to fully catchup on wealth.

• December 2: U.S. manufacturing relative decline predated NAFTA and China by decades.

• December 1: Illustrative example of misapplication of trade statistics. The “trade surplus” is not equivalent to fund flows.

November 2024

• November 30: Who ultimately wins in the next round of the trade and tech war between China and the U.S.?

• November 30 (bluesky): The recent “Chinese honeypot AI” meme reveals deeper is a “Masters of the Universe” level delusions of grandeur in the Silicon Valley AI community.

• November 29 (bluesky): People under-appreciate just how much of a “9/11” or “Pearl Harbor” moment the arrest of Huawei CFO (and daughter of the founder) Meng Wanzhou in December 2018 was for Huawei, the Chinese tech sector, Beijing and Chinese people.

• November 29: The RMB is persistently weak because of Beijing’s capital controls on inbound financial flows. This thread also discusses some views on long-term development of key Balance of Payments line items.

• November 28: Michael Pettis has been writing a version of the same article for well over a decade on the relationship between GDP, income, consumption, investment and trade surplus. It is useful to examine what went awry last time as an indicator of how it might be wrong in the future — it comes down to the base assumptions behind the model.

• November 26: (Part 1 | Part 2): Clearing up inaccuracies in the Chinese EV sector. Part 1 on the aggregate financial health of Chinese carmakers and Part 2 on properly calculating capacity utilization.

• November 26 (bluesky): We have lost the plot when the solution to “re-industrializing the economy for national security purposes” is reframed around the goal of trying to “rebalance China”. Reaction to presumptive incoming Secretary of Treasury Bessent’s remarks on tariffs and US dollar appreciation.

• November 24: Compilation of key points from analysis on China’s NEV sector economics.

• November 23: Twitter Spaces on “Which car companies will emerge as winners and losers out of the coming consolidation?” and a summary of some key points that came out of the discussion.

• November 23 (bluesky): Chart crime when it comes to measuring the “middle class”. Reaction to Economist article “The surprising stagnation of the Asian middle class” (archive).

• November 22: When it comes to property rights, perhaps China is not the outlier here. It might just be the United States that is the outlier and the one that needs fixing.

• November 21 (bluesky): Thoughts on China’s recent $2 billion sovereign bond issuance in Saudi Arabia by the Ministry of Finance. There was significant excitement from “de-dollarization” advocates about how transformative the transaction was given listing locale and pricing. However, the reasons are likely more mundane: this is just one step in a series of steps to integrate China’s trade and investment links to the world. In this case, the sovereign bond issuance opens up more fundraising avenues for Chinese corporates that are seeking to invest overseas in the coming decades.

• November 21 (twitter/x): Good opportunity to update my HSR charts in reaction to this recent article from WSJ: “China Is Building 30,000 Miles of High-Speed Rail—That It Might Not Need”. Also here.

• November 20 (twitter/x): October youth unemployment data update.

• November 20: How “factoryless manufacturing” led to overstated current account surpluses for decades and some implications on China’s trade relationship with the world.

• November 20 (Part 1 | Part 2): China is transitioning from a blue- to white-collar workforce in the coming decades. First “exclusive” quasi-thread on Bluesky (albeit covering a very familiar topic).

• November 19 (bluesky): Technology transfer was always about technology for profits.

• November 19: While high-speed rail in the United States is too expensive right now, instead of giving up a more constructive approach is to diagnose root causes on why and work on fixes. That’s how societies move forward.

• November 18: How kernels of objective truth like trade surplus and consumption as a % of GDP data are conflated to present a distorted reality.

• November 17: China’s urban infrastructure is better than Taiwan’s at the same stage of development in part because China has simply invested significantly more into housing and public urban infrastructure.

• November 15: October 2024 “flash” numbers and analysis, including an update of the “New LKQ Index”.

• November 14: Conflating China’s very large manufactured goods surplus with holistic trade. The bridge from manufactured goods to the current account balance:

• November 13: Response to Zoho founder Sridhar Vembu’s excellent post on how technology progress concentrates wealth and the importance of how society’s think about this phenomenon. Proud Twitter/X moment: earned a follow from Sridhar, who is one of my favorite thinkers on the platform.

• November 12: The practical and operational implications of “moving manufacturing”.

• November 11: The positive externalities of bike-sharing and the youths who are biking from Zhengzhou to Kaifeng for soup dumplings.

• November 10 (bluesky): Billy moneyballs Chinese structural reform

• November 10: Manufacturing is being automated even at the micro level: A CNC lathe with milling functionality and aluminum bar feeder/magazine can run unassisted and churn out custom parts at extremely low unit prices.

• November 9: The rise of Spain after two decades of painful structural adjustment. “Spain: open economy, unstoppable run?” h/t Benjamin Wolf.

• November 9: Who wins and loses with another round of increased tariffs on China by the U.S.? Focusing on the Apple iPhone trade.

• November 7: Reactions to an interview with Robert Lighthizer, who will likely play a significant role in the incoming Trump adminsitration: “Bob Lighthizer on Changing the Balance in U.S.-China Trade” from The Wire China.

• November 6: Helpful framework for describing different viewpoints within the incoming Trump administration with respect to overall policy to China from Tanner Greer: “The Battle to Shape Trump’s China Policy”

• November 5: A deeper examination of Chinese subsidies (credit, tax and R&D) in reaction to Bloomberg article “China Can’t Cut EV Subsidies It Isn’t Paying” (archive)

• November 4: China’s trade and current account surplus were historically overstated due to the “factoryless manufacturing adjustment. Adjusting historical figures significantly changes the narrative of just how much China’s economy relied on trade surpluses to drive growth since peak “mercantilism” during the Global Financial Crisis.

• November 3: “Industry 4.0” and the rise of Chinese advanced manufacturing.

• November 1: There is increasing access to high-frequency data in China. Also, thoughts on how “having to pick a side” in data-oriented analysis can pollute the analysis.

October 2024

• October 31: Direct rebuttal to the hypothesis that China’s current account surplus has been massively understated since 2022. Applied retroactively, the “factoryless manufacturing” adjustments meaningfully change the perception of China’s trade balances with the rest of the world.

• October 30: “Factoryless manufacturing” has resulted in an overstated trade surplus equivalent to approximately 1% of GDP since at least 2012. The 2021-22 changes in methodology that switched from using China Customs data that tracks physical flows to data tracking direct fund flows addresses these distortions in the Balance of Payments data tracked by SAFE.

• October 28: More likely that the Chinese trade surplus is overstated than understated. This thread pulls together significant work to understand the distortions caused by the measurement of imports and exports using physical flows vs. fund flows.

• October 25: Updated view on household wealth in China and splitting the household balance sheet into “in-use” and financial assets. In doing so, it is clear that there is a tremendous amount of household liquidity sitting on the sidelines in the form of banking deposits.

• October 24: Comparing Japan’s failed effort to develop a commercial airliner vs. China: it wasn’t technical capabilities but scale and market size that was the biggest constraint.

• October 23: National banks have re-taken the lead from local and regional banks, which dominated growth in incremental lending after the Global Financial Crisis, driven by the local nature of property and traditional infrastructure development.

• October 23: Applying the circulation model to recent package of reforms designed to release local government payment arrears back into the local economy.

• October 21: Addressing the fiction that the NEV and battery sectors are losing money in China; a response to the “White House warns China using overproduction for global dominance”.

• October 20: Delving further into the concept of “circulation” and pairing it with concepts of “Pareto efficiency” to visualize how policy intervention can be used to drive sectoral transition: in this case from the property sector to advanced manufacturing.

• October 19: On long-term human capital development.

• October 18: This compendium is supposed to be a list of my threads but this thread is mostly about pseudoerasmas’ take on the “potted history of institutions in macro-development thinking, at the time of the AJR intervention”.

• October 18: Using the recent NEV trade-in subsidies program to explain “circulation” and analyze the policy impacts.

• October 18: “Live tweet” of reactions and observations from an interesting podcast from Odd Lots featuring and Richard Koo on China’s recent stimulus.

• October 17: Distinctions between “demand-side” and “supply-side” stimulus is unnecessarily ideological and confusing. What matters is how policy intervention impacts productivity. Economies should be viewed holistically as dynamically interconnected circulatory systems:

• October 13: Line-of-sight now into the full-year 2024 domestic Chinese NEV delivery numbers. It is running ahead of forecasts.

• October 13: Thoughts on re-capitalization of the “mega banks” in China which have been sharpened via back-and-forth with and his recent post “A Depression with Chinese Characteristics”.

• October 13: Michael Pettis and the spectacular collision of finance, economics and China.

• October 12: “Supply-side” and “demand-side” policy is a false dichotomy as supply and demand are ultimately: Producers are also consumers.

• October 11: Implicit assumption of homogeneity in capital stock in traditional calculations of TFP is a significant issue that was quite specific to China in the post-GFC period.

• October 9 (Part 1 | Part 2 | Part 3 | Part 4 | Part 5): Referring to a series of detailed analysis by Brad Setser on why he thinks China’s current account might be significantly overstated. In the first part, I discuss the implications of a significantly understated current account surplus, namely the offsetting effect of “hidden” capital flight. In the second part, I use Apple to illustrate why the divergence in trade flow calculations under the traditional customs vs. BoP method and explain why I think the BoP method is actually more accurate for China. In the third part, I examine Apple Japan to confirm certain model assumptions and put a range on the estimated discrepancy of between $23 and $65 billion. In the fourth part, I reference a recent report from CF40 to expand on the technical differences between the physical trade flows captured by Customs data and the funds flows captured by SAFE and explain why for “factoryless” manufacturing, funds flows are more accurate. In Part 5, corroborating data from Alex Etra on China’s customs valuations allows us to get a much more precise measurement of the amount of overstatement.

• October 8: The rise of the next cohort of pureplay Chinese NEV makers.

• October 4: Contrary to popular belief, Chinese battery manufacturers are quite profitable.

• October 3: First Twitter Spaces (along with ) on “Socioeconomic inequality and ‘Common Prosperity’”

• October 2: Run (don’t walk) from reasoning based on GDP accounting identities.

• October 1: Acknowledging the risk of ridicule, disagreeing with Elon Musk here about the role that combustion will continue to play in vehicles in the coming decades.

September 2024

• September 30: Achieving scale and industry verticalization are the key factors that enable Chinese advanced industrial companies to be so competitive. Reactions to David Fickling’s deep dive into polysilicon (archive), the key base material used in solar photovoltaics.

• September 30: Framing the “epic rally” in Chinese capital markets in the context of the “wealth effect” on the real economy.

• September 29: Conventional wisdom is that Beijing sets “unrealistically high targets” and then uses (credit-driven) “non-productive” investment to meet them.

  A common corollary is that China’s headline GDP is generally inflated or “lower quality” than comparable economies. But have we considered that the opposite may be true?

• September 29: Long-term electricity production and consumption drivers and trends in China.

• September 27: Structural reforms in China are underpinned by demographic change from a workforce and population dominated by blue-collar workers to one dominated by college-educated white-collar ones.

• September 27: More detailed thoughts on how to read the recent announcement of Chinese stimulus, including both fiscal and monetary components.

• September 26: August youth unemployment data increased to 18.8%. Some more context and analysis around these numbers. Update: September figure declined to 17.6% (and nobody seemed to care anymore).

• September 25: Quantifying how the “wealth effect” has driven relatively weak demand during China's sectoral transition away from real estate and what the Chinese household and economy may look like coming out of the property transition through 2030.

• September 25: When analyzing the impact of policy, it is important to go beyond top-level analysis and consider specific impacted groups. Here, I analyze a couple of recently announced stimulus items from the perspective of upper and bottom quintile groups that they target.

• September 23 (Part 1 | Part 2 | Part 3): Illustrative archetypes of different socioeconomic levels in China, split by quintile. Part 2 breaks the upper quintile into more discrete segments. Part 3 introduces the “Common Prosperity Ratio”.

• September 22: Understanding socioeconomic inequality in China is key to understanding the political economy of the "Common Prosperity" era. I analyze socioeconomic and Gini coefficients using a more comprehensive measure of income that incorporates wages, capital income and redistribution. The results help contextualize both short- and long-term economic prospects for China.

• September 20: Breakdown of China’s income and wealth by socioeconomic level over time and the impacts of high socioeconomic inequality on demand.

• September 18: HSR ridership in China over the first eight months of 2024 has increased by more than the entire Shinkansen network during its peak year.

• September 17: The modern “middle” class in China is actually the upper quintile of society. The median Chinese household is a family that moved from the village to the city around a dozen years ago. Understanding this socioeconomic strata is key to understanding much of its policy direction today.

• September 16: Asia Society “China's Post-Plenum Tech Gambit: Chips, Shifts, and the Race to the Next Frontier”

• September 16: Technology and software are important, but not everything in the world of EVs today — reactions to Gartner’s latest 2024 “Digital Automaker Index” (archive) rankings.

• September 16: China is raising its retirement age. Productivity, not demographics, will be the key driver of China’s economy and its chances at reaching full modernization over the next two to three decades.

• September 15: A recent study on systemic corruption in China: “How Corruption Fuels Inequality in China” by Branko Milanovic.

• September 15: We can watch Chinese industrial policy in real-time through the effort to develop solid-state batteries and the role of a newly formed cross-industry organization, the China All-Solid-State Battery Collaborative Innovation Platform (CASIP).

• September 14 (Part 1 | Part 2): Review and commentary on August 2024 monthly GDP release and some more “macro” thoughts on China.

• September 13: The VC and PE investing environment in China is bad. The data and analytical skills embedded in this chart from this FT article are even worse. Follow-up note (March 2025): DeepSeek was founded in 2023.

• September 10: Higher-end vehicles like the Aito M9 are a key 2024 China NEV sector theme, along with NEV makers like Geely directly entering battery manufacturing.

• September 8: Tesla’s current slowdown in China has nothing to do with the “China Cycle” narrative espoused by .

• September 6: New World Bank paper (“How Redistributive Is Fiscal Policy in China? New Evidence on the Distributional Impacts of Taxes and Spending”) that illustrates how China uses “supply-side” policies to execute against its socioeconomic rebalancing policy goals.

• September 4: The battery sector is also profitable, but profitability is disproportionately concentrated with the largest players.

• September 4: BYD’s auto business will surpass Tesla’s by revenue and on a gross profit basis is already ~75% larger. BYD is able to achieve higher margins through its proprietary battery business.

• September 1 (Part 1 | Part 2 | Part 3): Contrary to popular belief, China’s NEV sector is profitable even as subsidies reduce over time. Compared to every other EV sector, it is the only one that is even close to making money.

August 2024

• August 31: Plaza Accord hurt Japan’s economy, but not in the way most people think.

• August 29: In the debate about LGFVs, there is too much focus on transferring value from the state to the private sector and not enough focus on how changing management and/or transferring ownership can unlock productivity.

• August 28: The youth unemployment rate jumps from 13.2% to 17.1% in July. Is it time to get hysterical?

• August 27: A collection of thoughts on the renminbi, “hot money” flows, capital flight and its future policy direction.

• August 26: Re-visiting the Chinese ICE “overcapacity” debate and discussing the concept of gross vs. effective capacity by incorporating natural depreciation and obsolescence rates of factory equipment.

• August 25: NEV growth is driven by supply-side factors in China while ICE decline is driven by demand destruction.

• August 25: Domestic consumers are the main absorber of increased manufacturing output in China. This is “funded” by the productivity gains that drive increased manufacturing output.

• August 24: Thought exercise to illustrate why market cap isn’t everything: A five-step plan to turn China Railway into a trillion-dollar market cap company and for me to collect my $30 billion banker fee.

• August 24: The Chinese banking system was in much worse shape in the 1990s and early 2000s due to the fundamental inefficiency of SOEs before large structural reforms.

• August 24: It’s time to revive the Li Keqiang Index. We can keep electricity and credit but railway cargo volumes should be replaced by retail sales of goods and services.

• August 24: Framework of different ownership and labor models and applications to specific companies and industries in China.

• August 23: China and the United States are both self-sufficient economies.

• August 20: China Railway’s July 2024 performance — the first month of the summer travel season.

• August 19: Indigenization is the only reliable way for a continent-scale economy like China to become a high-income country: the case study of tunnel-boring machines from Li Zexin.

• August 18: There isn’t just one approach to raising household incomes. Beijing’s core approach is targeting full, gainful employment which necessitates the use of supply-side measures. Supporting wages supports downstream demand (consumption and savings). Fiscal income transfers commonly used in OECD countries are not viewed by Beijing as a policy tool to stimulate ddemand, but as an income redistribution tool.

• August 18: Unit economics of high-capex, commodity businesses like solar and why I believe the United States should strategically build onshore solar PV manufacturing capacity.

• August 17: China is not in a “classic deleveraging trap”. Understanding credit in China requires delving into the underlying assets that collateralize the debt.

• August 16: PBOC governor Pan Gongsheng explains why he thinks systemic risk in the financial risk has decreased. This is an update and follow-up to remarks from last November.

• August 16: Conflating financial efficiency with productivity.

• August 14: Total social financing (TSF) is slowing in China: implications on capital formation and GDP growth.

• August 13: Beijing uses a variety of policies to stimulate demand (vouchers, trade-in programs, etc.). But direct fiscal transfers to households are more aimed at income redistribution vs. non-market demand stimulation. The most important policy to stimulate demand is targeting full gainful employment.

• August 10: I have thought a lot over the years about why the U.S. has had such difficulties building a modern high-speed rail network.

• August 9: On the problems of calculating overcapacity in rapidly growing industries when using forecast “planned capacity”. Capacity expansion is required to meeting rowing demand, and planned capacity is more indicative of industry growth expectations and lead times to build out new capacity.

• August 9: On Japan’s industrial and technological stagnation since the 1990s

• August 9: Various thoughts in reaction to this Economist article “China’s manufacturers are going broke” including Schumpeter’s creative destruction and contextualizing this chart from Adam Wolfe:

• August 8 (Part 1 | Part 2): Thoughts on the recent Foreign Policy article “China’s Real Economic Crisis: Why Beijing Won’t Give Up on a Failing Model” and accompanying thread by Brad Setser. I generally agree that RoW needs to make adjustments to China’s industrial policy but I disagree with the premise that China’s economic model is “failing”, which suggests that it is unsustainable. It is precisely because within China the model is seen as working that the United States and other countries should expect these policies to continue.

• August 3: Policy guidelines to boost consumption focus on supply-side measures.

July 2024

• July 31: The ongoing shift from labor-intensive to capital-intensive manufacturing and the move up the economic value chain in China, spurred by this FT article “Chinese low-tech manufacturers hanging on by ‘their fingernails’”.

• July 30 (Part 1 | Part 2 | Part 3 | Part 4): Net Domestic Product (NDP) is a more precise measure (vs. GDP) of actual production of goods and services in any given period. I present a framework for how various production factors contribute to production as measured by GDP/NDP.

• July 26: The Chinese economy may finally be at an inflection point on the balance between consumption and investment. This is driven by the demographic transition of its labor force.

• July 25: Consumption is undercounted in China and as an economy that is still in the middle stages of development, it should not be compared to fully developed OECD countries anyway.

• July 23: Framing different types of equity categories along disruption/risk vs. capital requirement vectors and how “State-led VC” has developed to address the difficult task of disruptive innovation-driven industries like clean energy that also have high capital requirements.

• July 23: Reasons why even though unit labor costs are much cheaper in Southeast Asia, finished solar panels end up being significantly more expensive. How industry verticalization and scale drive down friction costs in solar PV manufacturing in China.

• July 21: Pekingnology released a translated full text of the long-form resolution from the Third Plenum. This thread is an ongoing “live stream” of commentary as I go through the document from top to bottom.

• July 20: “Retail Sales” is not a good proxy for consumption because it ignores services. Critique of Rhodium’s report “No Quick Fixes: China’s Long-Term Consumption Growth” and long-term consumption forecast.

• July 19: Response to Noah Smith’s interesting post about whether China has developed better industrial policy for an advanced economy. I actually don’t agree with the fundamental framing because I do not think it is possible to lift a system that one country uses divorced from its underlying conditions and apply to a different country with a whole different set of conditions. I discuss two examples of major differences (China’s phase of development and its lack of land). What I do think is important is understanding Chinese industrial policy in the context of its unique conditions.

• July 19: From Bloomberg: “Trump Welcomes China to Build Cars in US in Departure From Biden”. But Chinese automakers would not invest substantial sums of money in the U.S. market without a reset in U.S.-China relations and a landmark reciprocal access agreement.

• July 18: Short-term thinking putting profits over long-term strategy: While most of the rest of the world is shifting to EVs, incumbent U.S. carmakers are doubling down on ICE.

• July 17: Discussion on the impact elevated investment in low-RoI (but ultra long lived) housing and infrastructure had on GDP (and TFP) over the last two decades and how the shift to high-RoI (but shorter useful life) manufacturing assets and services might impact both GDP and TFP.

• July 17: We should stop thinking about FDI in abstract financial and money terms and in real-world production terms. I explain why the U.S. is not well-positioned to grow manufacturing FDI in India. Instead, India needs to rely on countries with recent and relevant experience in targeted sectors. That so many roads lead to China makes it tricky given the current geopolitics.

• July 17: Comparing manufacturing value-add to consumption is not all that useful in assessing trade balances.

• July 15: Domestic sales of ICE vs. NEVs in China are driven by both demand- and supply-side factors. Understanding contrasting dynamics better informs long-term forecasting of aggregate domestic auto sales.

• July 14: Quick update on China Railway’s passenger rail and freight results through May/June.

• July 14: The ICE-to-EV transition is re-organizing the value chain and creating new points of competitive differentation. Chinese auto manufacturers like Chang’an are the closest to turnkey contract manufacturing in the car industry. And technology companies like Huawei are vying to become the Google Android for aspiring EV makers. This is lowering industry barriers to entry and leading to a proliferation in brands.

• July 13: The retail value of domestic car sales has recovered to 2017 levels. Rising net exports boosts economic value-add contribution beyond this. Improvements in underlying car quality and lower future fuel costs also means higher quality of GDP. The increase in domestic brand share also means more value-add from activities like R&D and marketing is captured onshore. The combination of all of these factors means that China’s car industry has played a key role in in helping the economy transition away from real estate.

• July 12: Chinese “gigafactories” (a.k.a. “advanced manufacturing”) have very low direct labor input and are more capital efficient (both working capital and fixed assets).

• July 9: There are even bigger projects in the pipeline. Three Gorges Energy, the same company that manages the Three Gorges Dam, confirmed plans for a ~16 GW integrated solar/wind/coal/battery project outside Ordos, Inner Mongolia. This project is part of a larger 455 GW plan that is targeted to be completed by 2030.

• July 6: Last week a 3.5 GW solar project in Xinjiang switched on. In this thread, I go through the energy economics of a solar PV farm vs. farmland to grow corn and what this might mean for the evolution of geopolitics and balance of power.

• July 6: Deeper analysis of the unit economics of the EV business in the context of the EU adding incremental tariffs (+17.4%) on BYD imports.

• July 4: Indonesia is adding tariffs on “footwear to clothing and ceramics to protect its domestic industries”. How this fits into the tectonic shifts in trade relationships between China and developing vs. developed nations.

• July 3: Migrant workers are the “beating heart” of the Chinese economy. How the generational relationship between migrant workers and their increasingly college-educated children sits at the core of China’s social compact. How structural change in China is underpinned by this demographic transition.

• July 2: China can substitute its entire kerosene demand with $165B of investment in large plants that can convert wind power into aviation fuel.

June 2024

• June 30: The polytunnel and other productivity-enhancing investments in China’s oft-overlooked agricultural sector.

• June 26: Analyzing the steel and concrete requirements for onshore wind turbines.

• June 25: China and the Costco metaphor.

• June 24: Response to thread by Michael Pettis on local vs. central tax reform where I dispute that there have been implicit transfer from households to businesses and governments since the GFC. Instead, I argue the opposite. Since the GFC, government policy has favored households, particularly lower-income ones, over businesses. I discuss how getting the “system framework” right is critical in answering the question about local vs. central tax reform.

• June 24: The international money connection between (i) Mexican drug traffickers (ii) U.S. drug users and (iii) Chinese HNW seeking to move assets overseas and the role that Chinese money-laundering organizations (MLOs) play.

• June 22: More accurately calculating Chinese EV sector subsidies.

• June 17: China “Retail Sales” is only a subset of broad-based consumption. It does not capture consumption-oriented services. If we examine broader-based measures, consumption is much closer to pre-crisis trend than suggested by “Retail Sales”.

• June 13 (Part 1 | Part 2 | Part 3): Analyzing the EV transition, specifically BloombergNEF’s EV transition compared to my more aggressive forecasts (Part 1) and then downstream impact on Chinese petroleum consumption (Part 2). Part 3 examines how the pace of the EV transition impacts assessments of excess capacity in upstream industries like batteries.

• June 12 (Part 1 | Part 2): The European Commission releases its long-awaited tariffs. It will apply additional individualized tariffs to BEV exporters (ranging from 17% to 38%) from China. Some analysis on how it might affect the prospects of a car like the BYD Atto3 (Yuan Plus in China) in Europe. We should expect a short negotiating window which, if successful would yield some sort of memorandum of understanding, followed by a longer negotiation over definitive terms.

• June 11: India will need to work with China if it pursues an labor-intensive export-oriented manufacturing strategy. India will need to pursue such a strategy to drive sustainatable, balanced economic growth. Reactions to two op-eds from Mihir Sharma (Bloomberg) and Raghuran Rajan (FT) on the importance of manufacturing to India’s economic development.

• June 11: EV and solar companies are all “feeling the 卷” these days but in distinct ways.

• June 10: “Overcapacity” or “involution” (卷)? What did Geely boss Li Shufu really mean in recent comments about EV industry competition? The analogy between the Chinese EV industry today and VC-backed rideshare industry in the 2010s. “Overcapacity” is quite different from “involution”.

• June 7: Land subsidies mitigate natural disadvantages; they do not provide a significant competitive advantage for Chinese automakers trying to compete globally.

• June 6: Sources of competitive differentiation have changed in the auto industry with the transition to “new energy vehicles”. How this ties into comparative advantages and why China was able to build decisive advantages in this space.

• June 4: Upfront subsidies used to get a nascent industry to scale should be amortized over future demand generated on a non-subsidized basis. If applied to the EV industry, the upfront subsidies ($173 billion high-end estimate) from 2009 to 2022 amortized across hundreds of millions in future NEV sales will be relatively de minimis on a per-vehicle basis.

• June 3: Fordism in the 21st Century — the number of years it takes the average Chinese manufacturing worker to afford a basic entry-level plug-in hybrid has fallen from 120 to less than 10 since 2008.

• June 1: Clean energy deployments are not only growing faster in China (everybody knows this by now) than the G7, but in the rest of the world as well.

• June 1: Chinese SOEs come in a variety of forms. Some of them look like Korean chaebols.

May 2024

• May 30: Cars still look the same on the outside, but their insides are changing rapidly. How “hybrids” are really just incrementally improved ICE designs while PHEVs and BEVs represent new “species” of vehicles.

• May 29: Step 1 of the property correction, involving the transition of economic resources away from property to other sectors, is largely complete. Step 2, involving the restructuring of accumulated malinvestment on the national balance sheet, is beginning in earnest. There are now numbers coming out that help quantify the total scope and scale of this restructuring effort.

• May 28: Comparative advantage at the country-level vs. sector-level. Clearing up some confusing aspects of Michael Pettis’ op-ed in FT (archive) “No, trade surpluses aren’t caused by comparative advantage”. Related follow-up thread with Brad Setser.

• May 26: Comparative advantage “stock” and “flow” and STEM labor markets.

• May 24: China’s solar numbers are about to get crazier.

• May 23: Solar and its similarities to commodity chips, “overcapacity” in growing vs. mature sectors, and what Chinese solar industry insiders are really asking for from Beijing.

• May 22: Morocco’s Al-Boraq HSR line — the first high-speed segment on the African continent.

• May 22: You need to pay more attention to Donghua Jinlong. No, seriously, you need to pay more attention to Donghua Jinlong.

• May 21: Technology transfer has now reversed with EV platform development technical knowledge and know-how developed in China and flowing back to German automakers like Audi.

• May 20: Summary of Sun Ninghui’s lecture to the Politboro Standing Committee (archive) on the history of computing and how recent AI developments fit into the long arc of computing history.

• May 20: The fossil fuel to renewables transition and how path dependencies have made the U.S. — at its core an oil and gas superpower — waver on its climate commitments.

• May 18: The irony of Trade Wars are Class Wars is that Beijing’s proactive policy support of the lower class in the post-GFC era is what is causing global spillovers in financial markets and trade.

• May 17: China announces a ¥300 billion fund to support local governments to convert existing excess buildings into affordable housing.

• May 16: Land subsidies, capital investment and winning the “hearts and minds” of non-aligned, developing nations.

• May 15: Throwing some chum out into the water. Why doesn’t Apple just acquire or partner with a Chinese automaker? Follows up this 2018 post on Apple and Tesla.

• May 14: Response to NY Times article “China Is Raising Bullet Train Fares as Debts and Costs Balloon” that makes quite a leap interpreting a pilot program for flexible pricing strategy as a signal for deeper economic troubles with local government finances and China Railway, rail network operator.

• May 13: Clarifying the excess capacity debate for Chinese ICE exports. Responding to Brad Setser’s thread.

• May 11: Auto industry evolution in the ICE-to-EV transition and how it impacts everything from accounting to changes in the critical success factors for competitive car companies. Analysis of three key components: manufacturing, developmand and marketing/distribution.

• May 10: A response to Tanner Greer’s excellent essay on 新一轮科技革命和产业变革 “New Round of Techno-Scientific Revolution and Industrial Transformation” with some critiques on framing but more importantly augmenting it with a specific focus on the clean energy revolution.

• May 9: China’s transition away from the property sector has been smoother than many expected. What does this say about the “Japanification” comparison? Observations from Richard Koo’s recent interview in Yicai and a follow-up to a thread from last summer on a talk he gave on this topic.

• May 9: German automakers are “holding service” on the ICE-to-EV transition.

• May 8: The implications of a global economy that is transitioning from “finite” to “non-finite” goods and services. Re-post of a piece I wrote in 2018 “What will happen to the United States if China's technology continues to advance?” on Quora.

• May 7: China’s complicated relationship with Myanmar.

• May 6: Xiong’an is the culmination of four decades of urbanization in China.

• May 3: Thread on declining FDI into China that accompanies a recent post (“Controlled capital investment”) that tries to breaks down the mechanics of the misunderstood “reinvested earnings” component. Follow-up thread looking at Apple’s segment data and explaining how it flows through FDI and balance-of-payment line items.

• May 2: How road names reflect Taiwan’s core ambition as a “developmental state”. Also the latest episode of Californians: Taipei.

• May 1: Migrant workers are the “beating heart” of the Chinese economy. The latest numbers from the annual migrant worker survey and how I interpret them.

April 2024

• April 30: Reminiscing about fantasy baseball and how it impacted everything from math to microeconomics to Excel modeling.

• April 30: The Third Plenum is officially on for July. It came about a year later than usual so much speculation about the content and agenda for the meeting. I expect substantive forward-looking reforms tied to the the impact of rapid technological changes, particularly in clearn energy, and Chinese “full acceptance” of a bifurcated globe and policies designed to adapt to it.

• April 29: The direct-to-consumer (DTC) distribution model for EVs decreases the cash conversion cycle and lowers working capital requirements. This is one of the many disruptive changes to the traditional car business model that are making it hard for incumbent ICE manufacturers to adapt.

• April 27: How asset rotation is decoupling credit formation from GDP growth in China. This is a a response to an excellent set of long-term credit and other charts from Shanghai Macro Strategist and a follow-up to a very important chart shared in October 2023 showing how credit was flowing from property to industrials.

• April 26: China and ASEAN are getting closer, driven by powerful and natural economic forces.

• April 25: BYD is now using electromagnetic linear motors in its high-end Yangwang cars replacing traditional hydraulic-based suspensions. This is an example of technology spillover, leveraging skills and IP that was developed in adjacent industries.

• April 24: The origins of TSMC from a radical new concept in Morris Chang’s head to one of the most geopolitically significant companies on this planet in less than four decades.

• April 23: TSMC’s Arizona fab and observations on TSMC’s “intensive, military-style work environment”.

• April 23: Global and China EV forecasts from IEA. I am forecasting significantly higher NEV sales and production than the IEA forecasts.

• April 22: All about Asian conglomerates from Japanese keiretsu to Korean chaebols to the rise of Indian conglomerates on the world stage like Tata Group today.

• April 22: Are Taiwanese contract manufacturers the big losers in evolving Apple and broader consumer electronics manufacturing supply chain?

• April 22: Texas controls its own power grid and it is not a coincidence that is leading the country in clean energy deployment.

• April 21: Applying Ferguson’s Law to China.

• April 20: The emissions payback period on the production of solar PV.

• April 19: China’s “mature node” chipmaking strategy. This was a continuation of an earlier thread from from last year where Lu Feng described this strategy.

• April 18: I did my first podcast with Steve Hsu at Manifold. Discussed my background and how it has shaped my perspectives on the topic of the Chinese economy. We talked about many of the topics I have been covering:

  • China’s evolving relationship with the West as it has moved up the economic ladder

  • Its policy emphasis on human capital development

  • The importance of advanced manufacturing

  • My “deep dive” into high-speed rail

  • Thoughts on the future evolution of the economy

• April 17 (Part 1): China’s Q1 GDP numbers came out yesterday. I wrote about the numbers, but more in a broader context of how they fit into several important themes I have been talking about for the past year. Here I focus in on the structural transition away from property to advanced manufacturing and other higher-RoI activities.

• April 15: In response to a question posed by Joe Weisenthal about China’s propensity to favor volume over profits in manufacturing, I wrote a long-form blog post and thread discussing this seeming paradox.

• April 13: Toyota’s bet on solid-state battery technology as a lynchpin to its EV strategy is a sign that it is not taking the transition to EVs seriously.

• April 10: Vietnam announces that it will prioritize the high-speed rail route to China for its North-South express railway project. This would be the second leg of an envisioned regional rail network in Southeast Asia that would enable high-speed passenger rail travel from Beijing to Singapore. This elevates Nanning (南宁), capital of Guangxi Province, in importance as a regional gateway city into Vietnam, the 3rd-largest member of ASEAN.

• April 9: Two reactions to Xu Gao’s recent speech about low consumption in China, translated here by East is Read. My two main reactions:

  • Household demand (expenditures plus gross capital formation) matters more than consumption

  • Corporate profits (via dividends) are not the only way to distribute economic gains to households. Chinese SOE’s play a major role with “public goods” and are distinct in the industries they operate in (highly regulated/strategic, capital-intensive and domestic industries prone to rent-seeking).

• April 9: Second-order effects of the shift from ICE to EVs on where and how they are produced.

• April 8: A follow-up on “Gigapresses” from the previous week and recent Noah Smith tweet — diecasting is a mature technology where most of the world’s foremost expertise is now in China. It’s used to produce an EV chassis, which is not close to being the most critical component in the product (batteries or the drive train), so insinuating that China is “stealing this IP” distracts from the much more critical point that so much of the industrial machinery sector has moved to China and how behind America is falling in complex mass manufacturing.

• April 7: As with cars, “excess capacity” or “overcapacity” in solar and battery manufacturing must be contextualized.

• April 6: Chinese advanced manufacturing are “digital-native” and leading the new “intelligent manufacturing” paradigm, also known as “Industry 4.0”.

• April 4: We know that the United States needs to consume less and invest more. What’s not clear is how we can make this happen. The path forward for China is much more clear at this point — deeper economic integration with the developing world.

• April 1: Indonesia should develop its own national auto brand. It can leverage changes to the auto supply chain (fragmentation and specialization) and the transition to electric vehicles.

• April 1 (Twitter | Substack): Exorbitant privilege of the U.S. dollar as the world’s reserve currency and its impacts on American manufacturing competitiveness.

• April 1: The recency of China’s development gives it an advantage over the G-7 to trade and invest in developing countries, especially in ASEAN.

March 2024

• March 30-31 (Part 1 | Part 2): Review of 置身事内 (“Chinese Government and Economic Development” h/t Rui Ma) wrapped around a critique of misapplied models (e.g. Janos Kornai’s “hard-budget constraints” vs. “soft-budget" constraints) on the Chinese bureaucracy. Using a more realistic model of the Chinese bureacracy and understanding of its weaknesses shows how the property problem is one of distribution instead of aggregate malinvestment.

• March 29 (Part 1 | Part 2): Huawei financials and compensation at Huawei and BYD. Huawei and BYD as the next evolution of East Asian tech/industrial conglomerates.

• March 28: To reach high-income levels of development, China will have to rely primarily on its domestic market to drive advanced industries. Its foreign expansion strategy for advanced industries will also rely more on developing economies than the G-7. In this way, the “Second China Shock” is the opposite of the first one.

• March 28: Tesla and some of its impacts on the Chinese EV industry: “Gigapresses”. Despite the hoopla over these largest diecasting machines, Tesla’s contribution was more on product design, technology and distribution.

• March 27: First podcast with Steve Hsu at Manifold

• March 26: Focusing in on Chinese NEV and ICE passenger vehicle exports. Breaking down the vehicle export numbers to separate out these exports from trucks, buses and smaller ICE vehicles that do not really touch developed markets.

  

• March 25: Response to Michael Pettis’ thread with counterpoints on why the property to advanced manufacturing transition represents a leveling up of the economy.

  

• March 23: Analyzing China’s Belt & Road Initiative (BRI) from the perspective of labor and human capital flows.

• March 21 (Part 1 | Part 2): Some interesting recent data from January-February on the Chinese EV industry. EV exports to the EU declined 20% in the first two months of 2024 increasing 36% to RCEP countries. What this says about the evolution of trade flows. From SCMP article “China EV exports, slammed by West’s trade barriers, find greener pastures with free-trade partners” (archive). Chinese EV penetration about to cross 50%, much earlier than most expected.

• March 18 (Part 1 | Part 2 | Part 3): High-quality electric vehicles that — due to rapidly declining input and production costs — are cheaper than equivalent (often less fully-featured) ICE vehicles on an upfront basis, and much cheaper on an TCO (total cost of ownership) basis, represent huge productivity gains that result in both higher quantity and quality of GDP. Response to this article (archive) by Thomas Klitgaard entitled “Can Electric Cars Power China’s Growth?”.

• March 18: More evidence from the first two months of the year of the property to industrial/manufacturing shift and how it may potentially create a tailwind for GDP growth. Retail also ticked up but is still well under pre-pandemic trends. Reactions to nice data-heavy thread from Patrick Zweifel.

• March 18: Capital-intensive vertical farming and clean energy sit at the foundation of an epochal shift.

• March 16: The economics of government-funded industrial subsidies to jumpstart a nascent industry by solving economic problems that private capital cannot.

• March 16: Widening gulf between politicized economic statistics in India and on-the-ground realities. Ashoka Mody responds (archive) to the announcement of the “elimination of extreme poverty”.

• March 14: Analysis of new central government funded “trade-in” stimulus program for businesses and households as reported in this SCMP article (archive) and original release from the State Council.

• March 14: Introducing a new mental model that compares sectors by their underlying “productivity” and “labor intensity” characteristics:

  

• March 12-13 (Part 1 | Part 2): Trying to figure out the true program development cost of the C919 at COMAC. Previous estimates had been in the $72 billion range. More likely in the $4-7B range (as a high-end estimate).

• March 12: Reactions to FT Article “Will China be forced to stimulate?” (archive) on 2024 GDP, focusing on the knowns (policy tools) and unknowns (macro and exogenous factors).

• March 12: Technical discussion on the GDP mechanics of sectoral transition from property to industrial investment. How “GDP Return” is different from vs. financial return measurements like RoE and RoIC.

• March 12: Correction of a misleading chart that shows the decline in the number of available Chinese data series by the National Bureau of Statistics. It turns out that a third-party data provider (CEIC) was removing access to data series for other reasons. Some data series have been indeed taken down but much milder than what a widely circulated FT chart (archive) indicated.

• March 11: India is the best bet to replace China as the labor-intensive manufacturing partner for the United States but there are some challenges. It’s main competition is ASEAN and while the United States can provide demand, FDI will likely have to come from allies like Taiwan, Japan and Korea.

• March 11: Ongoing debate between Robin Brooks, Mike Bird and Noah Smith about whether the U.S. has “meaningfully decoupled” itself from the China manufacturing supply chain. Recent data shows that it has not. I also raise concerns that U.S. “decoupling” efforts may accelerate China strengthening its relationship with the developing economies, potentially pushing the U.S. out farther to the periphery.

• March 10-11 (Part 1 | Part 2): Introducing the “Pareto efficiency” framework for analyzing Chinese economic development. Why the “China is over-investing” narrative is overly simplified. Applying the model to Chinese historical and ongoing economic development during the rapid urbanization era with export-oriented manufacturing, domestic-oriented housing and infrastructure investment and today’s clean energy transition.

  

• March 9: Why electric vehicles and batteries go hand in hand. Clean energy generation needs to be evaluated as a system, which can include a “dizzying array” of combinations of solar panels, wind turbines, green hydrogen production, transformers, inverters, UHV lines, batteries, pumped storage and many more.

• March 9: How many hours does it take for the average wage-earning worker to buy a car? 1909 + 1925 Model T vs. 2007 Ford Mondeo (in China) vs. 2024 BYD Seagull.

• March 8: Silicon Valley “deep tech” vs. massive integrated clean energy projects in China.

• March 8: How China and the United States approach de facto bans of companies.

• March 8: AFRE (Aggregate Financing of the Real Economy”. Review of detailed primer and historical review of China’s financial system.

• March 7: Chinese homes are back to 2014-17 levels of affordability. This suggests that China is closer to the end of the property sector adjustment process.

• March 7 (Part 1 | Part 2): The GDP mechanics of long-lived housing and infrastructure investment and how it contributed to slowing GDP growth, only to be offset by accelerating real asset gowth on the “national balance sheet”. What does GDP measure and not measure?

• March 6: Bytedance will not force-sell TikTok. It may sell the US operations (or maybe just let it get banned).

• March 5: On the importance of updating priors.

• March 5: When Chinese policymakers talk about “domestic demand” (archive) they are talking about both consumption and investment, which are accounting-driven definitions. Policymakers do not focus on them. Instead, they view things from a sectoral perspective. “Supply-side reform” is a common phrase that refers to the coordination in the shift of economic resources. Thinking about it this way helps us better the new slogan “New Productive Forces”.

• March 4: Medium- to long-term dynamics in the global auto industry. Some predictions on how national automakers will shake out.

• March 4: Why property prices are structurally higher in China than the United States.

  • Land scarcity vs. land abundance

  • Construction & materials costs

  • Land use rights vs. ongoing property taxes

• March 3: Large countries (like China, India and the United States) operate on a different set of economic laws than small ones. The rational basis for greater autarky in larger economies that is driven by the principle of specialization.

• March 2: History of Chinese-American immigrants, including some family history.

  

• March 1: Applying Bayesian thinking to applied sciences and technology development. “But I cannot help but wonder whether the long-term ramifications of export controls are to give China, leveraging its enormous STEM pool and human capital, the space to leapfrog in yet another advanced high-tech industry built on applied sciences.”

February 2024

• February 29: Response to the Biden Administrations press release (archive) on “National Security Risks to the U.S. Auto Industry”, including the sidelining of Tesla, not even mentioning the word “electric”, de facto U.S. ban on Chinese EVs, and potential long-term risks related to this strategy.

• February 29: The different productivity characteristics of various industries. Some drive (manufacturing and technology) productivity growth in the broad economy. Others (labor-intensive services) are consumers of technology and are better at creating jobs.

• February 28: Survey of the global auto market as a baseline to evaluate how different regions and countries will evolve in the EV transition. Background for why Chinese EVs are focused so heavily on developing markets.

  

• February 27: How inbound FDI (into China) and outbound FDI (from China) are related. Outbound: Chinese industrial FDI is replacing infrastructure FDI (Belt & Road. Increasing FDI to support offshoring of labor-intensive Chinese manufacturing. Inbound: Rising competition from domestic manufacturers is replacing foreign FDI, particularly in the auto industry from the EV transition.

• February 26: Time considerations in travel mode comparison between high-speed rail vs. airplane. “Breakeven (km) = HSR Time Premium (hrs) / Air Velocity Advantage (km/hr)”

• February 26: Discussion about how comparative advantages can change over time. Comparison of comparative advantages between the United States and China.

• February 26: China is building a “backup” network of Free Trade Agreements (FTAs).

• February 23: The benefit of technological progress and productivity gains does not always show up in direct employment numbers. It can also show up in production efficiences that lower the real costs of goods and services in the broad economy.

• February 23: GDP is a flow concept and the national balance sheet is a stock concept. We need to pay attention to both when evaluating major resource shifts between “consumption-oriented” and “investment-oriented” production.

• February 22: The biggest obstacle to U.S. manufacturing revival is a lack of quantity and depth of manufacturing skill and know-how. Cites Tim Cook’s 60 Minutes interview and talks about how this came to be.

• February 21: One of the biggest geopolitical shifts over the next two decades is China transforming from the world’s largest energy importer to an energy exporter, driven by the clean energy transition.

• February 20: While China has made strides in aviation, I agree with Airbus' CEO that it won't significantly disrupt global aviation for decades. Discussing the impact of long product cycles, China’s relatively smaller scale advantage and the national security factor.

  • Note: subsequent information on how much COMAC has actually invested in C919 development changed my views here

• February 16: Deflation tells you something different about China than from what it told you about Japan in the 90s.

• February 14: The domestic market accounts for the vast majority of Chinese EV growth.

• February 13: HSR is not a “terrible idea”. Critique of issues I had with Casey’s blog post on why HSR is obsolete.

• February 9: Labor-intensive and low-capex manufacturing vs. high-tech and high-capex manufacturing.

• February 5: Different ways to re-capitalize Chinese banks — capital injections, net interest margin, inflate away and move impaired assets off the books.

• February 5: What are the embedded losses in the Chinese banking system? Key factors: aggregate level of malinvestment, equity cushion / NPL recovery, and banking system liquidity / time horizon.

• February 4: China has been operating a two-speed labor economy.

• February 2: Comments on IMF’s medium-term outlook for the Chinese economy and why I have a more optimistic view of 4-6% growth through 2030.

• February 1: Taiwan during its labor-intensive manufacturing era with case study of Mattel / Barbie in the 60s to the 80s.

January 2024

• January 31: Compilation of ideas on why developing countries are different from deveoped ones.

• January 31: Scale in infrastructure can drive innovation and lower costs by changing the unit economics of fixed cost R&D. How “Going Big” played out with HSR in China.

• January 30 (Part 1 | Part 2): Issues with the TFP calculation for China.

  • Calculating capital increase using fixed vs. variable depreciation (rising asset duration)

  • Problematic GDP data from Penn World Tables (PWT)

  • “All roads lead to Harry X. Wu”

  • Housing is not related to “productive capacity”

  • TFP trendlines based on Taiwan/Korea experience

• January 28: My personal experience with Cantonese

• January 27: Is it better to invent or commercialize? The story of Lithium Iron Phosphate (LFP).

• January 26: The massive amount of investment China is putting into solar and all of the medium- to long-term impacts.

• January 26: On China’s “persistent” trade surpluses. What drives China’s manufacturing surplus (comparative advantage) and overall trade surplus (FX policy)

• January 25: Thoughts on Rhodium Groups’s low-end estimate of 1.5% GDP growth in 2023 (archive).

• January 24: TFP estimate from Lowy Institute’s 2022 report “Revising down the rise of China” (archive). I argue that declining TFP was partially the result of sectoral shift into “low RoI” housing and infrastructure investment. Sectoral shift back into “high RoI” industrial and high-value services should provide a TFP tailwind.

• January 23 (Part 1 | Part 2): The generational shift from blue-collar to white-collar labor force.

  

• January 21: Supply-side subsidies can benefit consumers.

• January 19: Observations on The Economist’s take on per capita GDP adjusted for hours worked (archive).

• January 19: Air travel vs. high-speed rail adjusted for traveler comfort.

• January 18: The seeds of U.S. industrial/manufacturing decline were sewn in the 70s and the decline of the machine tools industry was one of the first troubling signs. What happened to the US machine tools industry?

• January 18: China Railway full-year 2023 operating results and commentary.

  

• January 17: China’s youth unemployment rate is back after being suspended for a few months to adjust the survey methodology. Additional follow-up analysis on it in April.

• January 14: China’s NEV market dynamics.

• January 12: The causal link between debt creation and GDP growth is weak, at best.

• January 11: Massive upwards revision in renewables driven by China growth.

• January 10: Evolution of China’s manufacturing sector from a comparative advantage based on low-cost but productive labor to one based on technology and embedded know-how.

  • Product type

  • Target markets

  • Labor dynamics

  • Geopolitical environment

• January 8 (Part 1 | Part 2 | Part 3): China vs. Malaysia and Thailand

• January 2: Housing is consumption.

• Project Finance with Kweichow Characteristics

• LGFVs: the Good, the Bad, and the Mierda, Part I

• LGFVs: the Good, the Bad, and the Mierda, Part II

• LGFVs with Korean Characteristics

The previous essay took a detour to examine the case study of Korean chaebols from the 1990s and how a currency and liquidity crisis (the Asian Financial Crisis, or “AFC”) provided them impetus to force major corporate governance reforms on the conglomerates that ultimately led to a “leveling up” of the economy over the following two decades.

In a similar way, LGFVs today are currently in a slow-motion version of the AFC, one that was — unlike the exogeneous currency crisis — largely self-induced by the impact of the financial regulator’s tightening on the real estate sector in August 2020. Looking at the Chinese economy holistically, in a previous essay I highlighted how LGFV assets tend to sit at the “bottom of the barrel”, meaning that the private sector and national state-owned enterprises (“SOEs”) are relatively healthy with strong balance sheets, current malaise notwithstanding.

In this essay, we will discuss the path forward for the local government sector, first by defining the problems that it faces and then discussing potential solution approaches. Let’s start by first talking about how the role of local governments itself will likely evolve in the coming decades.

Subscribe now

The “Mayor Economy” and the evolution of the role of local governments over the next decade

For the last two-plus decades, local governments have played a critical role as the gatekeeper in the development of land, part of what has been arguably the most significant development trend for the Chinese economy: urbanization. As Keyu Jin describes the “Mayor Economy” in her recently published book The China Playbook:

“Pioneering local government officials have rushed at a frenetic pace to expand their local economies, transforming former fishing villages and farmland to technology hubs and industrial centers. These officials vie with each other for primacy in everything from economic growth to foreign investment, from the number of industrial, trade and horticultural exhibitions to the size of cultural events like concerts and film festival.”

We know from past precedents that rapid urbanization eventually ends. But how do we know how much urbanization is left for China? One way is to look at the experiences of South Korea and Taiwan, as they are probably the two “cleanest” and most comparable historical examples of this for China.

China’s rapid urbanization phase started about three to four decades after South Korea and Taiwan. Both South Korea and Taiwan urbanized rapidly, growing near-linearly at an average of 1.7% and 1.4%, respectively, in the roughly two decades after they reached 36%. After reaching betwen 75-80% urbanization rates, urbanization growth rates slowed considerably, averaging <0.5% per year in the subsequent two decades.

This would suggest that China has about another decade left of rural-to-urban migration until it reaches around 75%. This means another 150-200 million people will still be moving into urban areas at a pace of around 15-20 million people per year which is moderately slower than the pace of the last three decades1. 2021 and 2022 were impacted by the pandemic and the “Three Red Lines” liquidity tightening in August 2020. Based on historical precedents, we can expect a moderate decline in the pace of urbanization for the next decade — and there probably will be some pent-up demand that manifests over the next couple years as regulators loosen up after almost three years of tightening.

Infrastructure is the other large multi-decade trend that has involved land-finance and local governments. Similar to urbanization, there is still at least another decade left of historically elevated construction and corresponding gross capital formation. For example, the current plan for the high-speed rail network is to increase from the current 43,000 kilometers to 70,000 kilometers by 20352. In recent years, the pace of investment in urban metros and regional rail systems has been even more intense than HSR.

As we enter the 2030s, the pace of land-centric construction activity will fall off and the role of the local government will have to evolve. In mature economies, local governments typically focus mainly on local activities like schools, hospitals, safety and city services. Land-related activities (e.g. zoning) will continue to be a responsibility, just much smaller.

I expect local governments in China to also naturally evolve in this direction. This would inevitably change the composition of local government finances and the eventual reduction in the scope of LGFVs as well as long-discussed changes in the hukou system. We can speculate about the pace and timing of such a transition but these end goals are fairly clear.

Solvency vs. Liquidity

Timing is one thing, but can China afford to make this transition? In the first essay of the series, I threw out two critical questions that needed to be answered out of this exercise:

• Is this a solvency or liquidity issue?

• Who pays for the social costs?

The tongue-in-cheek response to the first question is a resounding “yes”. If you examine at the asset or project level, the majority of LGFV assets appear to be financially impaired, which would suggest a solvency issue. The degree of impairment matters — when your equity cushion is 38%3, there is a significant difference between assets being impaired 15% and 50%. But on this question of degree, I do not presently have a good answer for you because I do not have access to the underlying data needed to make this assessment. The only way to precisely determine this answer is to perform an exhaustive set of “fresh look” assessments on trillions of dollars worth of assets.

Zooming out and examining it from a national perspective, it is very clear that this is not a solvency issue. The national SOEs are large, well-run with market values that generally exceed the recorded book values of their assets ($30 trillion). Even in an extreme scenario4 where local government assets of $23 trillion5 were impaired by 50%, between the national and local governments, you still have sufficient asset coverage on $9 trillion in LGFV debt.

Ultimately, the crux of the matter revolves around the second question concerning the losses — it is not a question of whether the losses can or will be covered but how. The central government has the resources and local governments need financial support — this has been the way the administrative system has functioned since the 1990s. Resolution hinges on the dynamics and cooperation between the central government and local authorities.

One (admittedly imperfect) analogy to conceptualize the Chinese bureacracy is as vast and decentralized conglomerate where Corporate (the central government) manages far-flung branch offices (local governments). Borrowing from one of my all-time favorite shows, The Office, there is an adversarial relationship between corporate and the branch offices with corporate playing the role of the responsible adult (David Wallace) and the branch offices playing the roles of the entreprenuerial, less-dicisplined children (Michael Scott).

“How?” is a much more complicated question and requires looking into the problems at an individual level. Each asset, particularly those dealing with financial problems, needs to be assessed or audited. Coming out of this analysis should be a plan on how to deal with the assets, if necessary. And by looking for patterns across thousands of these assets, policymakers and reforms will be able to figure out what went wrong and how to address these in the next round of local government reforms.

“Fresh Look” assessment on $23 trillion worth of assets

Much of the commentary I have seen on local government (or broad Chinese economic) issues exists at too high a level to be useful in formulating practical solutions. What does it actually mean in practice to “unshackle the economy and stimulate consumption”? That type of high-level commentary is not helpful to policymakers that need to figure out detailed, practical solutions to very specific real-world issues.

Instead, issues need to be identified at a deeper level than just “there’s too much investment and debt”. It needs to recognize that there have been a variety of assets created with a range of outcomes from the good to the mierda. Policymakers need the right frameworks to guide the restructuring and reform process and ultimately tailored approaches with assets at the individual project level.

Different asset types require differentiated restructuring solutions. Dealing with real estate assets that are stuck in work-in-progress construction requires a different approach from real estate assets that are intended to be held on the balance sheet. Dealing with an infrastructure asset like an under-performing highway is very different from managing an eco-tourism development or amusement park.

In a previous essay in this series, we discussed the use of “Economic RoI” to determine one measure of asset and project performance. There we started from a high level and successively applied the framework at deeper levels. At the individual project level, the framework would be applied by evaluating its individual performance in a “fresh look” analysis and then diagnosing the reasons for project success or failure.

In private equity, these “fresh look” assessments are a common way of managing a portfolio of assets and similar, modified approach can be used for this massive portfolio of LGFV assets. The key difference is that these projects need to be evaluated based on both financial and social factors vs. private equity where the primary stakeholder is the equity owner. In the context of economic development, the stakeholders include the entire spectrum of society.

This is what a summary of such “fresh look” analysis might look like at the project level:

The above is the summary of a project assessment (illustrative only) of the Guiyang-Guangzhou HSR project (“GuiGuang HSR”). A detailed assessment would look more like this World Bank implementation report from 2017. The GuiGuang HSR project happened to be one of the more successful HSR projects. A less-successful project in another sector (e.g. a failed real estate development) would look different and generate a different set of recommendations.

Multiply this assessment across tens of thousands of projects6 and one can start accumulating a nice catalog of case studies of all the problems that need to be fixed. This is the type of work that is currently ongoing, undertaken by Chinese creditor banks, auditors and teams from the policymaking bodies. I expect that many of the issues that arise are typical of those that we find in conglomerates around the world.

LGFVs are conglomerates and deal with typical conglomerate issues

Imagine, if you will, that all of the LGFVs were grouped together under one organization as a large conglomerate. To a certain degree this reflects reality:

• LGFVs are all affiliates of one another by virtue of being state entities

• LGFVs draw on implicit guarantees from the state to raise financing

• LGFVs often lend and borrow from each other and co-invest together on projects and operating ventures

If one were to combine the balance sheets of all the LGFVs, this is what it might look like:

On the righthand side, I have annotated the balance sheet with descriptions of the major asset categories that sit within LGFVs. In the “LGFV conglomerate” think of each asset category as a focused division. Each division will have to formulate its own operational improvement strategy.

Restructuring and reform

In the last essay, I described some of the longitudinal problems that Korean chaebols faced in the 1990s and how LGFVs face similar problems today. The experience of the chaebols can serve as a guidepost for LGFV reform. Capital efficiency can be improved by:

• Re-structuring: Re-organizing the existing assets that have been created to be better managed

• Reform: Improving the capital and resource allocation decision-making process going forward

Re-structuring and reform are related. For example, re-organizing assets into more discrete categories will also create new market players that can be utilized to improve capital allocation going forward. Financial sector reform, particularly the use of newer types of specialized financing mechanisms like REIT-like structures, will help in the re-structuring process by providing fresh capital to unlock liquidity.

Restructuring: the over-arching need for greater focus

History (e.g. Korean chaebols in the 1990s) has shown that as conglomerates get larger, their economic advantages are eroded away by the burden of managing an increasingly diverse array of assets. A key goal for reform will be introduce greater focus around differentiated business models.

The business model groupings above provide guidelines, with different operational improvement approaches for different types of assets:

• Financial intermediary / investment holdings — Similar to a bank or financial institution, this grouping represents assets primarily listed under “Financial and Other” on the aggregate LGFV balance sheet above that are net off against associated liabilities on the debt side7. Ideally LGFVs should not be in the business of acting like unregulated banks: lending and borrowing money, holding a portfolio of investments and assets using public money. This activity can be done more effectively by the formal banking system or dedicated investment funds. Regulations need to be tightened up to clarify this type of financing activity at LGFVs (more on this below under “Reform: working capital”).

• Structured asset-backed warehousing — This division represents the role that local governments play as a gatekeeper on land development. While there is still another decade of rapid urbanization, this responsibility should eventually wind down. As China’s economy matures, the role of the local government as a formal gatekeeper to land will shrink significantly as other priorities take precedence. In the meantime there may be some interesting financial sector reforms that can improve the way this division finances its activities (more on this below under “Reform: financial sector”).

• Real estate asset management — LGFVs hold large portfolios of real estate assets “on the books”. These include many properties in new planning districts, government buildings, tourist sites, etc. In my view, the local government should not be in the business of real estate asset management and these assets represent trapped capital for many cash-strapped local governments. These assets should be liquidated over time and transferred to private sector entities like real estate management firms that specialize in operating and managing real estate assets. Select properties, like government building and schools, may be retained.

• Infrastructure — LGFVs hold large portfolios of infrastructure assets such as highways, joint ventures in rail assets and airports. In my view, local governments should not be in the direct infrastructure management business. These assets could be consolidated under larger, dedicated infrastucture management companies, with a mix of state-owned and private ownership. Introducing private ownership could help with liquidity and management but also requires putting clear rules in place for how natural monopoly infrastructure assets like tollroads and bridges are operated.

• Other operating assets — LGFVs also own a potpourri of other operating assets. Local officials partnered with local entrepreneurs to build a variety of operating businesses spanning from construction companies to amusement parks to tourist sites. Going along with the theme of focus, there are opportunities to consolidate operating assets under a larger, dedicated entity. For example, a dedicated operator could acquire a number of amusement parks that were built to create a Walt Disney Parks and Resorts equivalent that is dedicated to operating amusement and theme parks. A local government that contributes its local amusement park into the operation could receive equity in the larger entity. The key is to allow competent and focused operators that have the right combination of skills and incentives to run these businesses at larger scale, more efficiently. Increased clarity and separation between the board of directors and management would reduce risk of misaligned incentives and blurred lines between local government and private sector. A new generation of focused enterprises could arise out of this diversified blend of assets.

These approaches will help deal with the $23 trillion or so worth of assets that have been created in LGFVs to date. Greater focus will also improve the capital and resources allocation process going forward. Creation of more focused enterprises that specialize in specific asset categories are also natural partners for local governments on future projects.

Restructuring: Disciplining through financial support

As mentioned earlier, nobody — not even the central government — really quite knows exactly how large of an impairment LGFVs will have to collectively take on the $23 trillion in assets that have been created to date. This will not be fully known until auditors have had a chance to comb through a large representative sample of the assets.

While the size of the impairment is meaningful, arguably equally important are the reasons for the impairments. There are “good” reasons and there are “bad” reasons.

As discussed in Project Finance with Kweichow Characteristics, some of the impairments may be due to infrastructure projects that had specific poverty alleviation mandates. These are examples of “good reasons” — the local governments were fulfilling national policy mandates that were handed down to them by the central government policymakers and the projects should understandably not be evaluated purely on its financial metrics. To the extent the projects were executed well (cost-efficient, no/low corruption, achieved social goals) but are still struggling financially, the right thing to do would be to get continued financial support.

There are also “bad reasons”. If a project was underwritten and executed poorly, or riddled with corruption, those responsible should be held accountable. Provinces and municipalities where there was more bad execution might just need to clean up the mess themselves and not expect significant support from the central or provincial governments, respectively.

The reality is that every project will have its own individual assessment and its performance will be explained by multiple factors, ranging from “good” to “bad” to plain luck and chance. Each asset should have its own tailored plan. Good assets might become consolidators in the industry, with their managers rewarded with more responsibilities, including scaling up by acquiring other assets or being put in charge (with funding support) of dealing with troubled assets. Bad assets might need to be restructured and dealt with in various ways.

Reform: Privatization and transparency

With the prospect of significant re-organization of LGFV assets, improvements in the asset transfer mechanisms like auctions would help accelerate this process. Enlisting the support of the private sector and its capital increases the urgency to implement transparency.

There have been improvements in the asset sale process in recent years but greater transparency is required when it comes to public assets that are potentially being privatized. For example, there were rumors around the the sale of certain operating rights to manage the famous Leshan Giant Buddha in Sichuan for ¥1.7 billion. As David Fishman (Crossing the River) describes here, in this case it looks like it was merely the transfer of rights from one LGFV to another, but if there is to be widescale privatization involving private capital, transparency is of paramount importance.

In the last cycle of significant SOE privatization in the 1990s, things were not transparent at all. Many public assets were stripped from the state by insiders. In retrospect, the massive gains from introducing private capital far outweighed the losses from proto-crony capitalism — leading to significant net gains in overall capital efficiency. But the China of today is very different from three decades ago and the low-hanging fruits of market reform have already been picked. It gets harder from here.

Reform: Incorporating “lessons learnt”

Optimizing and re-organizing $23 trillion worth of assets is a big task. But trillions more worth of new assets will be created in the coming years and it is just as important that the process of underwriting new projects be improved going forward.

In private equity, one of the most important outcomes of “fresh look” analyses is figuring out what went wrong and using these lessons to improve the investment and portfolio management process. In a similar way, the aggregate output of thousands of “fresh look” case studies will be evaluated and analyzed to inform policymakers on the next round of reforms aimed at improving capital allocation processes going forward.

Much of the responsibility for this evaluation work will be with the National Development and Reform Commission (NDRC), an umbrella organization that works with other governmental bodies and is responsible for much of China’s long-term economic planning. One of its main responsibilities is drafting the Five-Year plans and no doubt many of the lessons learned from the LGFVs will be incorporated into the strategic planning cycle. Currently, China is in the third year of the 14th Five-Year Plan spanning (2021-2025). Research should be starting next year for the 15th Five-Year Plan (2026-2030) but this does not preclude reforms from being implemented within the current plan.

Reform: Corporate governance

As shown in the Korean chaebol experience, re-organizing assets held in diversified conglomerates into more focused operating entities should be paired with improvements to corporate governance affecting capital allocation decision-making in all of the major asset categories that LGFVs play in.

Management of LGFV assets will benefit from greater clarification and separation in the roles and responsibilities of the board of directors (representing the investors) and management responsible for day-to-day operation. Today, the lines are too often blurred between local officials and the folks in charge of managing the assets sitting on LGFV balance sheets. These blurred lines result in greater opportunities for corruption, misaligned incentives and lower operational performance and capital efficiency.

Reform: Working capital

When most people think of capital, they think of buildings or factory machinery. However, another significant and poorly understood component of capital are working capital items like receivables and inventory. One way to think about working capital is the amount of time a business takes to receive cash from the time they start working on a product or service that they are selling, also known as the cash conversion cycle8.

China has unusually long working capital cycles, which means it has more capital tied up in working capital than other economies. The average payment delay is almost twice as long as Japan’s and also higher than Australia, Hong Kong and Taiwan. SOEs are especially notorious for stretching out payments, which increases working capital requirements for upstream suppliers. While payment delays do not show up in the income statement, they do affect cash flows and businesses ultimately run on cash inflows and outflows, not profit accruals.

Referring back to the “aggegate LGFV balance sheet”, you can see the working capital intensity in the size of the Receivables category, which at ¥18 trillion at the end of 2020 was over 60% of the entire Infrastructure fixed assets category (¥29 trillion). The “Inventories” category is comprised mainly of real estate, some of which is owned “on the balance sheet” but another large portion of which is essentially long-dated working capital tied up in work-in-progress construction projects that have yet to be completed or sold.

Elevated working capital levels reduce the capital efficiency of the economy as they increase the amount of total capital required in an economy and reduce the velocity of money and corresponding economic activity. Significant reforms in this area would improve the capital efficiency throughout the economy and particularly with LGFVs. J.P. Morgan estimates that ¥2.3 trillion ($335 billion) tied up in working capital can be released with improvements in the cash conversion cycle and used for more productive purposes.

Reform: Financial sector — loosening (short-term)

A recurring theme in this essay series is the idea that the Chinese financial system is by mature economy standards still quite basic, conservative and unsophisticated. This has given rise to informal lending and investing that has undoubtedly created risk in the financial system through increased credit intermediation.

This is part of the cycle of regulatory tightening and loosening where entrepreneurial forces are tacitly encouraged to experiment outside the formal system for a period of time before regulators step in. In the case of LGFVs, the key date of tightening was August 2020 when financial regulators stepped in with the “Three Red Lines” guidelines that limited indebtedness and curtailed access to liquidity for mny LGFVs. If this cycle follows past financial sector reform cycles, financial regulators should step in with a slate of new regulations to usher in a new era of growth and experimentation.

China’s post-pandemic recovery has been relatively tepid because capital is simply not flowing as smoothly through the system as it has in the past. The formal banking system, while basic and conservative, fit well with two of the main growth engines of the Chinese economy — residential real estate and infrastructure. It was efficient at re-deploying cash deposited in the banks by savers into loans for projects, a large portion of which went back into the pockets of wage-earners, which would then flow back into the economy and eventually find its way back into cash deposits, starting the cycle all over again.

But with liquidity tightening and general lack of confidence in the economy, this once-smooth cycle has slowed down. Households, which had been driving new urban household formation, have been tentative to make such large financial commitments. While household incomes have continued to rise, instead of saving it in the form of residential real estate, they have been saving it in the form of cash and liquid reserves. Banks are flush with cash but there is stagnant demand for loans — the balance sheets of LGFVs and private sector developers are stretched and the private sector is still tentative. Significant capital is stuck in the system and not being put to work. This has dampened momentum in the post-pandemic recovery.

In the short to medium-term, getting out of this liquidity trap is to a large extent about restoring confidence in households and private sector businesses. The lack of confidence is caused by lack of visibility in the future outlook and greater uncertainty. Part of restoring confidence will be simply loosening as part of the natural tightening and loosening cycle described above.

But this needs to be done in moderation. Urban household formation peaked in the 2015-19 period and completions of floorspace in the 2012-17 period. Infrastructure is also likely at or approaching its peak intensity. While regulators may have tightened too early or let it linger too long, they also need to be mindful that a return to 2010s decade would not be prudent in light of where China is in the back-end of its multi-decade urbanization super-cycle. A “goldilocks” approach with respect to the tightening/loosening cycle is warranted here.

Reforms: Financial sector — more sophisticated financing tools, role of private capital (long-term)

In the long-term, financial sector reform are important to enable the economy to reach the next level of capital efficiency. Part and parcel with greater focus and a move away from diversified conglomerate approach of LGFVs are more sophisticated financing tools that align with the asset categories they are financing.

For example, China has been experimenting with REIT-like structures as a way to diversify financing options for real estate and infrastructure projects away from plain-vanilla bank lending and bond markets. Since June 2021, approximately ¥90 billion was raised for over two dozen projects — a drop in the bucket compared to the total pool of real estate and infrastructure assets but China’s securities regulator recently released guidelines to expand both the size and scope of this financing method.

REITs would strengthen the alignment between investors, management and the assets they hold, a more dynamic mechanism to enforce capital discipline and improve the resource allocation process. They would also bring new sources of liquidity into the sector by diversifying its sources of capital to private investment funds and household investors.

Another area of financial sector reform would be to formalize many of the informal financing activities that have been undertaken by LGFVs. This includes investing, lending and borrowing operations between LGFVs mentioned earlier but also the role that LGFVs play in warehousing work-in-progress construction projects on their balance sheets before liquidating by selling to the household end buyers (in the case of residential real estate projects). There are more specialized and sophisticated structured asset financing vehicles that can be used to finance this process more efficiently.

Ultimately, the primary goal of introducing more sophisticated financial tools is to better align investors with management and the assets that they manage. Private capital and capital markets can introduce capital discipline more dynamically and responsively than the “brute force” disciplining approach undertaken under China’s adversarial administrative system and conservative, unsophisticated banks. By diversifying sources of financing, it also mitigates future risk of liquidity traps.

There is major precedent for this in China’s recent history. One of the biggest reasons why Chinese economic efficiency boomed in the 2000s was from the introduction of private capital and entepreneurs to formerly state-owned assets that were “let go” in the major SOE reforms in the 1990s. Similarly, private capital played a significant role in improving capital efficiency in Korea in the aftermath of the Asian Financial Crisis. Figuring out how to better harness these forces should be an important theme in the next cycle of reforms.

“Country Roads, Take Me Home (to Guizhou)”

Economic development is driven by a continuous series of reforms. In the coming months and years we are likely to see a steady stream of news and announcements related to local governments, LGFVs, and this large pool of assets they have spawned. The goal of this series of essays was not to predict everything that is to come but to provide perspective and frameworks on how to place changes as they arrive.

While this topic has been mostly about abstract concepts like economic development, it is always important to ground ourselves in the human impacts, often hidden behind all the numbers and charts. Whether American or Chinese, West Virginian or 贵州人, no matter where we are in the world, “going home” is something we can all relate to. This essay series began in Guizhou and I think it is fitting that we end it there as well.

Regardless of what happens with Guizhou province, its LGFVs and the assets they hold, there is no doubt that there has been tremendous change and the long-term impact of the past two decades will continue to echo into the future, often in ways that nobody could have predicted at the outset. But even through such tremendous change, home will always be home.

When I began writing this series of essays, I knew very little of Guizhou and the 39 million people that call it home. While reading and research can barely scratch the surface, I now know a little more about this corner of the world and I hope you do too as well.

Just one last tidbit on Guizhou: I recently read this interesting article from Baiguan about the budding trend of “Village Super Leagues” (村超) where a grassroots effort by locals that are passionate about football has grown into a national phenomenon:

“We invite you to explore captivating stories of individuals who, driven by their unwavering love for the sport, are keeping the football spirit alive in their rural local communities in the mountainous, historically underdeveloped province of Guizhou. Through personal tales of dedication to the sport, this examination of China’s Village Super Leagues presents a new, hopeful side of the nation's relationship with ‘the beautiful game’”

The accounting identity is Maslow’s Hammer of Chinese economic discourse. These days it seems every explanation, policy recommendation, and prognostication of the future revolves around the idea that China’s economy — reduced to an accounting identity — is imbalanced with too much investment and too little consumption.

Alternative, nuanced explanations are not countenanced, or hand-waved away in favor of this re-balancing narrrative. There is minimal questioning of the underlying assumptions or even the validity of the original assumptions that supported the thesis. Updated data and findings that raise questions about the predictive ability of this narrative are dismissed. Data that confirms the narrative is amplified.

These maxims are repeated in derivative conversations, and like the game of telephone diluted where repeaters often misinterpret the highly technical underpinnings of the narrative; when challenged, they make discussion-silencing appeals to authority. This is not objective analysis but behavior typically attached to conspiracy theorists and cults.

All while the fundamental basis of the narrative remains highly flawed, yet unquestioned. But it needs to be questioned to prescribe the right reforms at the right pace and better understand how things are likely to unfold.

Note: in this essay series, to distinguish between technical GDP accounting and colloquial definitions, I will generally refer to the technical definitions as “Consumption (C)”, “Investment (I)” (or “Gross Capital Formation”), and “Government Expenditures (G)” and the colloquial definitions as “consumption”, “investment”, and “government”.

Subscribe now

Lies, Damned Lies, and Statistics

One of the core premises of the Over-Investment Thesis is that consumption is too low and needs to regress sharply to the mean. The key piece of evidence is this chart presented in Part I showing how China’s Consumption (C) as a percentage of GDP is “extremely low”, even when compared to economies that followed high-investment growth models like Japan, South Korea and other “Asian Tigers”:

But when you take a closer look at these numbers and pair it with a deeper understanding of how the accountants and economists formulate GDP, it will become very clear that, for a variety of reasons, China’s numbers are counted very differently from the others on this list, especially the OECD members, and need to be adjusted if they are going to be compared to others.

These adjustments will show how the headline Consumption (C) numbers for China significantly understate the true consumption of the Chinese household compared to the countries to which it is being compared. To understand how, we need to once again jump into the weird world of GDP accounting.

Technical jargon vs. colloquial usage

In Part I, we reviewed how the GDP “sausage” was made and hinted at how words like “consumption” and “investment” could mean different things in different contexts. In a GDP context, Investment (I) and Savings (S) are precisely defined but when the average layperson hears the terms “investment” and “savings”, they usually have something else in mind.

The diagram below shows how non-intuitive the GDP accounting identity categories really are and how they do not really match up well against actual socioeconomic groupings like the “household sector”. Private sector businesses do not even show up directly in the GDP expenditures approach, as the expenditures approach only measures final demand by households — businesses provide intermediate goods and these activities are not considered final demand.

Further complicating this is the overlap1 between socioeconomic groups like the “household” and other groups like the “government” and “private sector businesses” — ultimately the government and private sector businesses are made up by individuals that are also part of “households”.

The expenditures approach and the income approach do not line up neatly; for instance, the relationship between Consumption (C) on the left side and household income on the right side is not a direct 1:1 relationship:

• Consumption (C) itself can be broken down further into components like housing (but only the imputed rent or rental portions), food, energy & transport (but only the consumable components), durable goods, services and many others. All of this is attributable to the “household sector” and paid for with “household income”.

• Investment (I) can be broken down into categories like housing, infrastructure and business investment. Its “housing” is related but ultimately distinct from the “housing” that is part of Consumption (C). The “residential housing” portion is greenfield construction activity that is ultimately driven by purchase decisions made by the “household sector” and also paid for out of “household income”. This shows up a “savings” and residential housing makes up the largest component of household savings in China2.

It’s not just consumption and investment that have very different colloquial definitions from their precise, technical GDP accounting definitions.

When the average person hears “savings” they probably think more of the cash and deposits in their bank, brokerage account and 401-k. In other words, it is a “stock” concept that is analogous with a balance sheet. But in a GDP context, Savings (S) is actually a “flow” concept — analogous to an income or cashflow statement — that is based on total GDP minus Taxes minus Consumption (C) (i.e. “savings of the private sector”) and Taxes minus Government (G) (i.e. “savings of the Government”). Savings (S) of the private sector can be further bifurcated into savings of households vs. private sector enterprises.

In official policy communication, it is typically the colloquial definition of these terms that are being used and not the highly technical GDP accounting identity definition. For example, in this strategic planning document by the Central Committee, the authors are clearly using the colloquial definition which includes housing construction:

The basic role of consumption continues to strengthen. The proportion of final consumption expenditure in GDP has remained above 50% for 11 consecutive years. Traditional consumption such as housing and transportation consumption has grown significantly, the per capita housing construction area of ​​urban residents has steadily increased, and the sales of new cars have ranked first in the world for 13 consecutive years. New forms of consumption and new models are developing rapidly. In 2021, online retail sales of physical goods will account for 24.5% of total retail sales of social consumer goods, and per capita service consumption expenditures will account for 44.2% of per capita consumption expenditures.

(Translated by Google)

And just to make the point very clear, it goes on further to talk about investment and includes categories like healthcare and education in the discussion — which, in the GDP accounting identity would be categorized as Consumption (C).

The key role of investment is better played. The proportion of the country's total capital formation in GDP has been maintained at a reasonable level, providing strong support for optimizing the supply structure and promoting stable economic development. The level of infrastructure construction has been improved in an all-round way, the national comprehensive transportation channel has been accelerated, and a number of major water conservancy facilities have been completed and put into use. The construction of new infrastructure such as 5G has been accelerated, significant achievements have been made in the construction of major scientific and technological projects, and investment in high-tech industries has continued to grow rapidly. Weaknesses in the fields of medical and health, ecological and environmental protection, agriculture and rural areas, and education have been speeded up.

(Translated by Google)

GDP accounting identity adjustments

With this background in place, let’s start going through the adjustments with some real numbers. To align China’s GDP accounting identity figures with OECD members, there are at least three identifiable categories of adjustments:

1. Differences in GDP accounting standards with OECD economies

2. Differences in China’s economic structure and approach and how expenditures that should really be categorized as Consumption (C) are categorized as Government (G)

3. A certain portion of Investment (I) represents capitalized subsidies that would fall under Government (G) or Consumption (C) in OECD countries

(1) China’s conservative GDP accounting and its impact on Consumption (C)

To standardize GDP calculations, there are global accounting standards that countries try to adhere to when tabulating GDP. For various reasons certain countries are behind others in adopting the latest accounting standards and this leads to differences in how certain economic activities are measured.

Despite various claims about how Chinese GDP data is inflated3, it is actually quite conservative from an accounting perspective. China calculates its GDP using a modified version (CSNA-2016) that is formulated in accordance with System of National Account (SNA) 20084. SNA-2008 has been adopted by most OECD countries but within this standard there is significant leeway for accounting judgment.

A well-known difference is China’s conservative approach to calculating imputed rent. Imputed rent is an estimation of the value of housing “consumed” over a certain period — think of it as rent that you pay yourself (as the homeowner) to use the property. It is derived from applying a percentage to the value of owned residential housing stock. With 90% of Chinese households owning their own homes, for all intents and purposes, it is the how the vast majority of the housing consumption portion of GDP is calculated.

Historically, under CSNA-2002, imputed rent was calculated based on the cost basis, or book value, of residential housing, yielding a much lower imputed rent figure than if you were using market rates which are significantly higher. While calculations based on market rates are allowed under CSNA-2016, this particular change does not appear to have yet been implemented5.

This conservative accounting is validated by comparing China’s imputed rent calculation with other countries. In this excellent data series, Adam Wolfe highlights how China’s imputed rent is a much smaller percentage of the economy compared to developed economies. Typically housing would make up a larger, not smaller, share of income for less-developed countries like China, so this is a strong indicator of this accounting difference.

Multiple sources corroborate that China’s GDP figures undercount imputed rent by amounts ranging from 4% to 8% of total GDP, or ¥4.6 trillion to ¥9.2 trillion of 2021 GDP. This is not an insignificant sum that would be tacked onto the GDP estimate and counted as Consumption (C). Besides imputed rent, there are other items from under-counted employment income to medical services to automobiles where China’s conservative GDP accounting generates a lower estimate of GDP, and they are almost all related to consumption factors.

(2) Government (G) spending on behalf of households and consumers

In China, the government provides subsized goods and services to certain segments of the household population. Prime examples of these are the thousands of community canteens (公社大食堂) in Chinese cities where certain groups like the elderly can get subsidized meals within walking distance.

At this community canteen in Shanghai, it’s an all-you-can-eat buffet for ¥28 per person with half-off discounts for senior citizens. This one has been open since 2019 and can accommodate eighty people at a time. In addition to lunch and evening buffets, it also provides à la carte breakfast and dining options that cater to specific communities. For example, fresh food can be prepared and delivered to elderly people who prefer to eat at home for ¥16. While these are run by private entrepreneurs like regular restaurants, their costs are subsidized by the government. The closest OECD analogy might be food stamps.

The subsidized costs of these community canteens are counted under Government (G) but they are benefiting households and should count as Consumption (C). Hongshen Zhu noted in a recent tweet that these subsidies equate to a 6.9% (of GDP) transfer to the Consumption (C) category. The National Bureau of Statistics officially provides for this adjustment in its annual accounts:

In OECD countries, social benefits are typically made via direct payments (e.g. social security) and show up in household income. Re-balancing advocates often comment about China’s “thin social safety net” and recommend that it increase direct transfers to the household to boost the household’s share of income which they contend would lead to more consumption. But these advocates fail to recognize that the government is already providing large transfers to households (7% of GDP), just in a different form.

That China takes a different approach — “social benefits with Chinese characteristics”, if you will — in a manner that is categorized differently under GDP accounting standards, does not take away from the reality that it is straight-down-the-fairway consumption6. It represents real household consumption, and for the purposes of this re-balancing question, should be re-categorized as Consumption (C) from Government (G).

(3) Components of Investment (I) are really embedded subsidies that should be re-classified as Government (G) or Consumption (C)

In Project Finance with Kweichow Characteristics, I discussed the role that poverty alleviation often played in investment considerations, particularly in less-prosperous provinces such as Guizhou. As you can see in the chart below, infrastructure is one of the pillars of China’s national poverty alleviation approach.

There are a number of ways that Consumption (C) and Government (G) might be disguised as Investment (I):

1. Poverty alleviation strategies that are funded through infrastructure projects

2. Subsidized housing programs funded through construction of public housing

3. Local governments funding ongoing expenditures by de facto borrowing from related-party LGFVs

First, infrastructure projects are typically funded through local government financing vehicles (LGFVs). The LGFV would inject equity (typically in the form of land) and borrow money to fund construction and development. Certain infrastructure projects had significant poverty alleviation elements in their underwriting mandates. They were not expected to fully support themselves with incremental revenue generated by the asset and might eventually need additional government support.

An example would be a bridge in heavily mountainous terrain that could cut down travel times from hours to minutes for local residents. While the bridge may not generate enough incremental taxes or tolls from lower-income local residents to fully cover its financing costs, it increases quality of life for local residents, effectively representing a transfer of wealth from the government to these local residents.

It is difficult to calculate exactly how much of a subsidy this amounts to, but one way of thinking about this is that a significant portion of LGFV losses can be attributable to poverty alleviation programs. If these “social losses” had been funded by the central government directly as part of explicit funding schemes — again, more similar to how it is typically done in OECD economies7 — then it would show up in the Government (G) column and not Infrastructure (I).

Second, similar to infrastructure, a portion of residential housing built over the past two decades has been in the form of public housing that is subsidized by the government. From 2010 to 2019, 141.3 million urban housing units were sold of which 29.5 million (~21%) were newly-built public housing units8. Construction of public housing since 2019 has been “slightly more resilient than private housing”. This report suggests that affordable housing subsidies of approximately 1% of GDP were granted in 20129 as a proxy for capitalized Infrastructure (I) related to affordable housing programs. This adjustment is likely conservative.

Third, local governments have been using related-party LGFVs to borrow funds that are being used to fund ongoing expenditures. This is a legacy of a conservative banking system and central government controls that were put in place to limit direct borrowing. I also discussed this point here in Project Finance with Kweichow Characteristics. Under a different system, these capitalized expenses would show up under the Government (G) category as fiscal spending.

This number would be a portion of the “augmented fiscal deficit” metrics that the IMF has used to augment the official view of local government finances with the net borrowing activities through related-party LGFVs. The IMF calculated the augmented deficit at 13.8% of 2021 GDP, most of this driven by LGFV activities (vs. local government fiscal deficits). For the purposes of this exercise, I am assuming 1% of GDP that has been capitalized in Investment (I) that should really be in Government (G) in the adjusted GDP accounting identity. Again this adjustment figure may be conservative.

Summary adjustments: China’s true Consumption (C) is closer to 50% than 38%

Pulling this all together, China’s true Consumption (C) percentage is closer to 50% than 38% and its adjusted Investment (I) percentage is closer to 37% instead of 43%.

Said another way, if China used GDP accounting and approaches — like direct instead of in-kind transfers — more similar to OECD members, this is where its GDP component percentages would be.

Other economists have also researched and corroborated these adjustments. Zhang Jun at Fudan University estimates that adjusting for these accounting differences, China’s true household consumption should be 10-12% higher than the official rates. For 2021, this would imply Consumption (C) as a percentage of GDP between 48% to 50%, similar to where I ended up with my analysis.

With the proper “OECD adjustments” in place, while still low, China’s consumption is not as extreme an outlier from global averages. The need to “re-balance” goes away when you more accurately benchmark China’s consumption numbers to others and eliminates a core pillar of the Over-Investment Thesis — that China is an extreme outlier10.

We can tie these concepts back into the diagram we reviewed earlier by breaking apart the top-level accounting identities into smaller components and attributing these components to socioeconomic groups like the household sector (generally colored in maroon). For example, the diagram depicts 7% out of 16% of Government (G) expenditures are really in-kind social benefits for the household sector that would have been classified under Consumption (C) for most OECD member economies.

This diagram gives a much clearer view on how GDP can be attributed to the household sector.

The Right “Comps Set”

Beyond making these technical accounting adjustments, there is an even bigger question at play here. Implicit in the observation that China’s consumption percentage deviates from the rest of the world is the idea that the global average is the correct model for China to conform to.

But why should China be expected to conform to this particular group of countries, at this particular time? China today is at a very different phase of economic development from most advanced OECD members. Compared to middle-income countries, it is pursuing a differentiated growth model. The assumption that there is a universal standard that every economy must conform to is simply a bad assumption to make.

I find the original list of comparable countries problematic, particularly as it relates to benchmarking Consumption (C) as a percentage of GDP. With a few exceptions, all of the economies on the list are fully urbanized. China is still in the middle of rapid urbanization, having averaged ~20 million new urban residents a year for the last two decades. These new urban residents need homes and greenfield construction of apartment buildings drives Investment (I) at the expense of Consumption (C). The urbanization difference mandates that China’s Investment (I) as a percentage of GDP will be elevated compared to mature economies. The inverse is also true because of the accounting identity: Consumption (C) must be lower.

China works with a very different capital and labor structure from the median economy. China has a massive population and relatively little land and natural resources. In other words, it is over-supplied on labor and under-supplied on capital. The only practical way to address this imbalance is by building which will necessarily result in elevated Investment (I) as a percentage of GDP for a lengthy period of time. Even after almost two decades of heavy investment, China is still under-capitalized compared to other economies at the same level of development11:

Basic developmental challenges that China faces are most similar to the ones other “Asian Tiger” economies like Japan, South Korea and Taiwan faced. Like China, they suffer from a relative dearth of land and natural resources and used a high-investment growth model over many decades to build up their capital stock and grow their economies.

Whittling down the list to these much more comparable “Asian Tigers” makes the comps analysis more meaningful. But we cannot compare their Investment (I) and Consumption (C) rates to “Developed Asia” economies today as they are already fully urbanized and developed. We need to reach back into history and see what their numbers looked like when they were at a similar phase of urbanization as China is today.

China was at 65% urbanization in 2021. Taiwan and Korea were at those points in 1979 and 1985, respectively. Japan reached this point in the early 1960s.

China’s adjusted investment rates are slightly lower than Japan’s in the early 1960s and moderately higher than South Korea in the 1980s and Taiwan in the late 1970s. Taiwan is a bit of an outlier here. Out of the “Asian Tigers”, its economic development relied less on heavy industry and more on small and medium-sized family businesses that developed capital-light export-oriented business models. Moreover, during much of its urbanization drive, Taiwan’s authoritarian government prioritized economic resources on defense and re-taking the mainland, with specific regulations in place to limit the amount of investment that could go into residential housing. As a result, Taiwan’s investment rates lagged the other “Asian Tigers”.

The inverse of high investment rates are low rates of final consumption expenditures. In his analysis of adjustments to Chinese consumption rates, Zhang Dun noted that China’s consumption rates of “60-65%”12 are comparable to the “Asian Tigers” during their high-growth periods:

Japan’s consumption rate was only 64-70% from 1968 to 1997. Hong Kong’s consumption rate in the past 50 years was around 60%. Taiwan was around 70%. These East Asian economies have one thing in common with China: they are all high-growth economies …

… based on these consumption rates, we can speculate on the future growth prospects of the Chinese economy. With consumption of the four “Asian Tigers” at 60%, more than 30% of GDP was used for capital investment. After all, the improvement of productivity ultimately depends on the formation of capital to drive rapid economic growth …

… if 70-80% of our annual national income is used for consumption, there will be fewer investments in technological research and development, capital will not be formed as quickly and labor productivity will not increase rapidly, limiting future economic growth prospects. The experience of the four “Asian Tigers” illustrated these points.

(Excerpts translated by Google)

Compared to its most relevant comparables at the same stages of economic development, China’s investment rates are high but not extraordinarily high13. Far from being a cautionary tale, if anything the astonishing economic success of the “Asian Tigers” offers valuable lessons for relative latecomers like China to borrow from the same playbook and follow similar development principles, including a model based on high savings and investment.

Ultimately, using the GDP accounting identity as Maslow’s Hammer diverts attention from what should be the main goal of any economy — maximizing the long-term growth of an economy and improvements in quality of life for its people. Instead of focusing narrowly on an obscure set of accounting identities, we need to focus on how to maximize long-term growth rates. Tunnel vision focus on the GDP accounting identity obscures a deeper and more complex, but ultimately more accurate way of analyzing the Chinese economy.

This recent Project Syndicate article was another in a string of pieces adopting this narrative:

Household consumption typically accounts for 60% of a country’s GDP. For the 2011-20 period, it was 68% in the United States, 59% in India, and 61% in middle-income countries excluding China, where it was only 37%. While China accounted for 16.7% of global GDP between 2017 and 2021, its share of global household consumption was a mere 11.5% …

Here is another one from the recent FT article “China after the property boom”:

Given the relatively modest role that consumption plays in its economy, the IMF has described China has a “global outlier”. The country’s gross domestic savings as a percentage of GDP is 44 per cent, compared with an average of 22.5 per cent among OECD members. Over the long term, much of this is believed to be precautionary savings, cash put aside for housing, education, healthcare and retirement.

In tracing the origins of China’s “exceptionally high savings and low consumption” rate today, the IMF notes “inadequate social spending” as well as earlier changes such as the one-child policy and the “gradual dismantling of the social safety net” in the 1980s and 1990s.

These articles, and countless others like it, all play on some variation of themes related to this narrative:

• China’s growth model is imbalanced with too much investment and not enough consumption. This was made possible primarily by subsidies to investment and production activities and is the main reason for its “weak household demand”.

• Over-investment leads to an increasing proportion of non-productive projects and large, “systemic” misallocation of capital that will accumulate until investment rates fall significantly.

• China has been able to continue over-investing by taking on increasing amounts of debt. This is not sustainable and when China runs out of debt capacity, it will be “forced to adjust”.

• Chinese policymakers are following this path because they target GDP growth rates that are above sustainable levels, as the party’s political legitimacy depends on high growth. This makes it “politically difficult” to change.

In many discussions about China’s economy today it is hard to avoid this narrative and I often wonder if those that reference it truly understand the narrative’s intellectual foundation and historical origins. This is not just a petty academic debate but one with significant real-world impact. As I describe in my ongoing series on LGFVs, there are major reforms needed that will have long-term impacts on billions of people around the planet. Getting the framing correct and formulating the right policy prescriptives are extremely important.

In this first part of this multi-part essay series, I will explore the historical underpinnings and development of this narrative. Most of this is just review for anyone who has been part of the Chinese economic debate for a reasonable amount of time. I will probably get a few things wrong and welcome feedback and criticism on the characterization of the thesis. This background is important to set a common baseline understanding that will be referenced in follow-on essays in this series.

Subscribe now

Let’s Get Technical

Economics can be a rather technical subject, characterized by abundant use of domain-specific vocabulary. It frequently delves into references from obscure historical periods and encompasses a wide array of academic theories and “isms”. I took my first economics class at a pre-college program in 1995, later matriculated at the same university and eventually graduated with a B.S. in Economics.

However, I would by no means consider myself an economist1. Although I retained some concepts from my college courses, fully delving into the Over-Investment Thesis necessitated conducting more in-depth refreshers on the subject, including going beyond just academics and tying it to the what I had learned in the real world, particularly my experiences living and working in the Greater China region over many years.

Calculating Gross Domestic Product

International economics is a sub-category that deals with, among many other things, the intricacies of measuring economic activity across different economies. Gross Domestic Product (GDP) is one of its more important metrics and sits at the heart of the “Over-investment Thesis”. Understanding this narrative requires a technical understanding of how GDP is calculated/interpreted that is not typically well-understood by the average layperson.

GDP is the measure of economic activity within some defined geographic region, typically a country, over a specified period of time. It is calculated by accountants pulling together various data series and making adjustments/estimates while trying their best to adhere to evolving global GDP accounting standards.

There are multiple ways of calculating GDP. The first is called the “Income Approach”. Under this approach, accountants count all of the labor and capital used in the production of goods and services in a certain period. The second is called the “Expenditures Approach”. Under this approach, accountants tally the sum of “final” goods and services purchased in the economy over a certain period by consumers and the government as well as net import/export balances.

In practice, accountants are not counting every single transaction that takes place. Instead, they rely on existing data series (like tax receipts) and supplement with surveys taken across a representative subset of the population. Economists then apply judgment, making estimates and adjustments to the data. This is why the first GDP numbers you see are estimates that get “firmed up” as data from more series comes in over time.

The less developed an economy, the larger the informal economy and the more economists must rely on adjustments and estimates. GDP can also be a highly political number, which creates its own set of issues. In the United States, the expenditures approach is considered to be the more accurate way of estimating GDP than the income approach2. Ultimately both approaches are used to triangulate to the final numbers.

The GDP Accounting Identity

One of the fundamental formulas describing GDP is from the expenditures approach:

• GDP = Consumption (C) + Government (G) + Investment (I) + Net Exports (NX)

This formula is referred to as the “National Income Identity”3. Consumption (C) (a.k.a. “Household Expenditures”) consists of spending by the private sector and Government (G) comprises expenditures of the various levels. Investment (I), also known as “Gross Capital Formation”, is comprised of fixed assets and working capital items like inventory — final spending that is capitalized and ends up on the official national balance sheet in the form of “Capital Stock”. Net Exports (NX) is Exports less Imports.

A derivative formula creates what is known as the “Saving-Investment balance”4:

• Savings (S) = Investment (I) + Net Exports (NX)

Here, Savings (S) refers to the savings of both the government and household sectors. For economies that have a neutral trade balance (Net Exports = 0), Savings (S) must equal Investment (I).

Note: in this essay series, to distinguish between technical GDP accounting and colloquial “layperson” definitions, I will generally refer to the technical definitions as “Consumption (C)”, “Investment (I)”, and “Government Expenditures (G)” and the colloquial definitions as “consumption”, “investment”, and “government”.

The Over-Investment Thesis

Much credit should be given to Michael Pettis, a finance professor at Peking University’s Guanghua School of Management, as one of the main advocates of this “Over-Investment Thesis”. Through the years he has been remarkable in his consistency and adherence to both writing and promoting this narrative.

I first came across Michael’s writings in the aftermath of the Global Financial Crisis (“GFC”). I enjoyed reading his long e-mail letters where he deftly weaved together economic topics and historical examples with the then-current Chinese economy that was going through significant change.

He has been writing about China’s propensity to over-invest for almost two decades now. In October 2009, he wrote:

Consumption growth in any country is necessarily limited by the growth in household income and wealth, neither of which has grown nearly as rapidly in China as the country’s economy. China’s development model is a steroid-fueled version of the classic export-led model common to many high-saving Asian countries. This model involves systematically subsidizing production and investment, often leading to very inefficient investment.

In his 2013 book The Great Rebalancing, he wrote:

… but with consumption so low, it would mean that China was overly reliant for growth on two sources of demand that were unsustainable and hard to control. Only by shifting to higher domestic consumption could the country reduce its vulnerability and ensure continued rapid growth. This is why in 2005, which household consumption at a shockingly low 40 percent of GDP, Beijing announced its resolve to rebalance the economy towards a greater consumption share.

How GDP accounting identities explained the Chinese economy

The GDP accounting identities lay at the heart of the Over-Investment Thesis and narrative. As Michael explains in this 2011 paper, in the 1980s after China’s reform and opening up, “household consumption comprised 50-52% of China’s GDP”. During the 1990s, both household income and Consumption (C) grew quickly, but “not as quickly as China’s economy”. This resulted in a sharp decline in Consumption (C) to the mid-45s. But in the 2000s, it started to drop to levels that were “unprecedented in history for a large economy in times of peace.” It dropped to 36% on the eve of the Great Financial Crisis and then dropped even further, bottoming out at below 35% in 2010:

Michael almost always explains this historical trend through the lens of the GDP accounting identity. The rise of China’s export-led industrialization growth model in the coastal regions of the Pearl and Yangtze River deltas led to rapidly rising Net Exports (NX), peaking at 8.7% in 2007 “far surpassing the trade surpluses Japan generated in the late 1980s”. This contributed to global trade imbalances that led to the GFC and was ultimately unsustainable. And indeed it was — over the subsequent decade, Net Exports (NX) would decline, bottoming out at 0.8% in 2018.

With such a large part of the economy in decline, how could China keep up its high growth rates? For the better part of the 2000s, rising exports had helped drive double-digit growth rates, topping out at a whopping 14% in 2007. But in a post-GFC trade environment, the Chinese economy was forced to replace sharply declining net exports with domestic investment. You can see in the above chart how starting in 2008, Net Exports (NX) and Investment (I) flipped. This caused Consumption (C) to bottom out at 34.6% in 2010.

“The end of the growth model”: Imbalances and unsustainable growth

It is understandable why trade imbalances were unsustainable. A large trade imbalance means that you are exchanging goods and services with the world in exchange for foreign assets.

In 2016, Warren Buffett explained the dangers of high trade deficits from the American perspective in the allegory of Thriftville vs. Squanderville:

In effect, our country has been behaving like an extraordinarily rich family that possesses an immense farm. In order to consume 4% more than we produce--that's the trade deficit--we have, day by day, been both selling pieces of the farm and increasing the mortgage on what we still own.

To put the $2.5 trillion of net foreign ownership in perspective, contrast it with the $12 trillion value of publicly owned U.S. stocks or the equal amount of U.S. residential real estate or what I would estimate as a grand total of $50 trillion in national wealth. Those comparisons show that what's already been transferred abroad is meaningful--in the area, for example, of 5% of our national wealth.

While China did not run the highest trade imbalances in percentage or per capita terms, the sheer scale of its population meant that it would have enormous impact on global trade that would likely result in enormous trade frictions. Different from others like Japan, which could ride the export-oriented industrialization wave to becoming advanced economies, this strategy was simply not possible for China and its then-1.2B+ population5. It had to find a new growth model to continue its economic ascent.

On November 9th, 2008, China unveiled a ¥4 trillion economic stimulus package to offset the impact of the GFC and millions of suddenly-unemployed workers in its coastal export regions. It was the largest such stimulus that had ever been put in place in the country, equivalent to almost 12% of China’s then-GDP. Much of it went to “fast track” domestic infrastructure and construction projects that were already in the planning phase. Millions of migrant workers simply moved from shuttered export factories in Shenzhen to construction sites in places like Wuhan and Guiyang. The GFC and resulting stimulus plan ushered in the global unveiling of China’s new primary growth model: domestic investment.

But according to Michael, there was a big problem. Similar to how trade imbalances were unsustainable in driving growth in the long run, he argued that the surge in domestic investment was also unsustainable. To him, these were fundamentally the same underlying models — and similarly unsustainable. As he wrote in “The end of the growth model” in 2011:

It is increasingly obvious that China has reached both constraints. The global financial crisis has eliminated the ability of the rest of the world to absorb China’s large trade surpluses, and capital misallocation has been a serious problem for much of the decade …

… if capital misallocation is becoming a problem, clearly China must make a major transition to a different growth model—one that allows household wealth and consumption to catch up with the enormous wealth generated in the past three decades so as to become an increasingly important driver of growth.

From unsustainable trade surpluses to unsustainable domestic investment

The case for why high trade surpluses were unsustainable was quite reasonable; in any case, the aftermath of the GFC proved this out. But (at least to me) it was less obvious why elevated domestic investment was an unsustainable growth model sitting there watching this unfold in 2009.

Pettis explains why by bringing up examples from history:

But historical precedent also makes it clear how powerfully addictive that model is. It is hard to find a single example of a country adjusting smoothly and quickly from a period of excessive investment-driven growth—just look at the United States in the 1930s, Brazil and Venezuela in the 1980s, the Soviet Union in the 1970s and 1980s, Japan in the 1990s, and the Asian Tigers after 1997. In ever single case excessive debt led to a sharp growth contraction and a “lost” decade or two.

Up to this point, I had generally nodded along to his explanation of the global economy, particularly the risks and distortions of large global trade imbalances and some of the mechanisms like financial repression suppressing Chinese household income. But this is where the views began to diverge.

It all revolves around consumption

For Pettis, the switchover from growth enabled by unsustainable increases in the trade surplus to domestic investment was merely a symptom of the underlying root cause — under-consumption. Because of the GDP accounting identity, whether it was high Investment (I) or high Net Exports (NX) — or both — the offsetting effect was low Consumption (C).

The root of the problem, in his view, is the idea that the household sector is repressed by “politically entrenched Chinese elites” (defined somewhat vaguely) that “distort the Chinese economy by strangling purchasing power and subsidizing production at the expense of consumption.”6

Michael goes on further to say that China has the most extreme levels of under-consumption in modern history, often sparing no hyperbole in descriptives. In the chart below, we can see how China at 38% is 17% below the global average and only ahead of trading entrepôts like Singapore and Ireland that are not comparable and suffer from their own distortions. China’s “extremely low consumption rate” provides overwhelming evidence of this imbalance and a growth model that is not sustainable unless it reverts to the mean.

Ultimately, the picture that Michael Pettis has consistently painted is of a China that is trapped in this low-consumption, high-investment model, ostensibly because entrenched interests have tied themselves to the model and make it “nearly impossible” to adjust. It has resulted in “massive misallocation of resources” and “spiraling debts”. Like every other “economic miracle” its growth story is coming to an end. His outlook is grim, and yours should be too.

From Mungerism to Maslow’s Hammer

Charlie Munger, Vice-Chairman of Berkshire Hathaway, is well-known for his multi-disciplinary thinking and liked to use the phrase, “To the man with only a hammer, every problem looks like a nail”. Originally, I had thought this was just another Mungerism but turns out it is officially attributable to psychologist Abraham Maslow, of Maslow’s Hierarchy of Needs fame.

Maslow’s Hammer is the idea that to specialists in their field, everything looks the same. For many economists and analysts looking at China, this GDP accounting identity has become that type of hammer. The tool can be useful in some situations. But to view an economy as large and complex as China through this lens can turn into a crutch. Especially if the tool itself is not as robust as it seems once you start poking into its fundamental logic.

The purpose of this essay series is an attempt to explain why we need to move on from Maslow’s Hammer and advance the collective understanding of how the Chinese economy works at a more detailed, granular level with a wider array of mental models and tools. In a follow-up essay, I will show how the core logic of the narrative — that China is under-consuming at an extreme degree — is at best, shaky.

Here’s a small preview: Chinese Consumption (C) is extremely low. But Chinese consumption is not particularly low at all.

Ultimately, understanding the differences between technical Consumption (C) and colloquial consumption is very important to figuring out what the GDP accounting identity numbers are really telling us.

I am still collecting my thoughts on the various reform paths for LGFVs. I thought that I could capture it all in one final series post but it is getting long so I may have to split it into two.

This past week on social media there were a number of interesting conversations on a number of topics. I have pulled out the most interesting ones and summarized below.

musings

Quantifying malinvestment in Chinese real estate (2023.7.19)

In the debate about capital efficiency in the Chinese economy, particularly real estate assets, residential vacancy rates are a key metric to estimate wastage and malinvestment. At the end of 2019, Chinese households were sitting on $33 trillion in residential real estate. This compares with $99 trillion in total capital stock1.

Residential vacancy rates across 28 Chinese cities are approximately 12%. Cross-country comparisons suggest that China may be higher than other countries by up to double2. If we assume that the global standard for residential capital efficiency should be half, this would mean 6% of outstanding Chinese residential housing stock could be considered “bad” GPD.

Combined with around $1 trillion of potential bad real estate assets sitting on the balance sheets of LGFVs, one can argue 3% of GDP equivalent gross capital formation may have been wasted3. From 2010 to 2019, investment as a percentage of GDP averaged 45%. Removing all this potential “bad” investment and replacing it with consumption-oriented expenditures would have lowered it to around 42%.

HSR — the Good, the Bad, and the Mierda (2023.7.17)

In the table below, I separated various Chinese HSR lines into grades based on the level of ridership, which can be estimated based on the number of daily train pairs on a line4. At the top end, lines like the 1,318-km Beijing-Shanghai line rivals the highest-usage lines in the world: Japan’s Tokaido Shinkansen line running from Tokyo to Osaka. At the bottom end are lines like the 1,777-km Lanxin line5.

As with everything else in the Chinese economy, HSR was a mix of good and bad projects. This nuance is often lost with the tendency to over-simplify.

“Savings” means something very different to economists vs. regular people (2023.7.16)

The definition of “Savings” in a GDP accounting context is very different from the definition “savings” that most regular people have in mind. This can cause confusion and misinterpretation in discussions about the Chinese economy and its tendency to “save too much”.

How HSR indirectly reduced GHG emissions (2023.7.15)

There is a non-linear relationship between velocity and energy consumption. This means that passenger travel by HSR consumes significantly more energy than passenger travel by conventional track. And because the current Chinese grid is dominated by coal, HSR replacing motor vehicles has actually raised GHG emissions over the past decade.

So how did HSR still led to overall reduced GHG emissions? You may recall one of Zhao Jian’s arguments was that China should have invested more in its freight network. It took some time, but this is exactly what has happened. As HSR lines grew in popularity, conventional lines were ramped down and eventually converted to freight traffic instead. Bulk freight transport is as much as three times more efficient than trucking. Shippers started shifting their bulk freight loads from ICE-powered lorries to freight trains.

The GHG savings here have more than offset the increase in emissions from shifting to electrified HSR passenger traffic (powered by a coal-dominated network). That said, with the rapid rise of renewables and the likely peaking of coal consumption in the next 2-3 years, the de-carbonization benefits of HSR will really start to compound.

Lack of confidence in households and the private sector (2023.7.13)

Confidence in this context is the level of certainty in the future.

The triple-whammy of zero-COVID, crackdown on technology and for-profit eduction sectors and “Three Red Lines” liquidity tightening has eroded the confidence of households and the private sector in China. While zero-COVID was an exogenous factor, it is now over. The other two are self-induced and from that perspective it is within control of policymakers to remove these overhangs.

This does not mean avoiding potentially difficult reforms — particularly in the local government / real estate sector. Indeed, signaling what these reforms are could bring the clarity that increases the confidence of households and private sector companies.

Cross-country wage comparisons (2023.7.11)

Cross-country comparisons are always challenging because of different ways of calculating metrics and different practices between countries. For example, in China, it is very common for room and board to be includes as part of the compensation package, especially for lower-wage workers. When we compare the wages of manufacturing workers in China to other countries, we need to keep this in mind. Some data series will adjust for this, but others may not.

charts & data

China has been undergoing a multi-decade urbanization boom. Past precedents like South Korea and Taiwan suggest that urbanization is a process that takes around four decades.

In China, there has been a fairly consistent stream of rural villagers moving to the cities in search of jobs and a better life for themselves and their kids, averaging 20 million per year over the last 25 years. For the first time in more than two decades, this rural-to-urban migration feell well below that average. Was this a function of the pandemic or is it a harbinger that China’s urbanization path will deviate?

from the archives

Is China’s state-driven model running out of gas? (2019.7)

This seems like the age-old question that everyone is interested in. More comparisons to other “East Asian Development” economies.

Why is China’s growth rate slowing down? (2018.9)

Explaining China’s growth rates through three archetypes.

On electric vehicles and vertical integration (2018.6)

How important is vertical integration to success in electric vehicles. Interesting to look back on my thoughts from five years ago. I think vertical integration is more important than it was back then and will separate the big EV winners from the rest of the pack.

diversionary

In the post LGFVs with Korean Characteristics, I shared a link to H.O.T.’s “Candy”, the first modern Korean boy band. But if you were to ask my Korean friends who the real K-pop O.G. is, it would be Seo Taiji.

• Project Finance with Kweichow Characteristics

• LGFVs: the Good, the Bad, and the Mierda, Part I

• LGFVs: the Good, the Bad, and the Mierda, Part II

The previous essays were intended to establish foundational understanding of this oft-misunderstood $20+ trillion pool of assets. At the risk of repetition, the key themes to highlight include:

1. The importance of looking at things from an asset-centric perspective.

2. The centrality of land to local government activities.

3. The role that social policies such as poverty alleviation played in creating these assets.

4. The role that adverse selection bias played in asset formation.

5. The adversarial relationship between central and local governments.

Originally, I thought this would be the last post in the LGFV series, but as I contemplated Korean chaebols in the 1990s and its similarities with today’s LGFVs, I thought it would be helpful to take a detour before getting to the final segment discussing potential reform paths for LGFVs.

The Rise of the Chaebols

I began my career working in Asia in the early 2000s. One of the first live projects involved traveling to Seoul over several months working with SK Corporation’s energy division as it considered expanding into the power sector. Very much in-line with the free-market zeitgeist of that era, Korea Electric Power (KEPCO) was considering the privatization of some of its power generation assets. This was my first brush with chaebols.

Chaebols were family-run business groups that formed when the Korean economy was still in its early development stages, many starting as construction companies helping rebuild Korea in the aftermath of the war. Patterned after Japanese zaibatsu, these business groups developed major advantages during this period.

First, they established tight, inter-linked supply chains through both formal and informal links giving them a leg up over smaller domestic competitors. Chaebols had a tendency to vertically integrate the industries they entered. Second, they were able to access capital at a time when capital was scarce, especially debt: chaebols, like LGFVs, used significant debt financing to grow, taking advantage of close relationships with the government and banking sector. Third, they developed strong internal group cultures, often driven top-down by the controlling families, that would come to exemplify the Korean economy: paternalistic, hierarchical and very hard-working1.

With scale, established supply chains and strong group working cultures, they could deploy capital at attractive rates of return compared to the rest of the economy. They formed tight partnerships with the authoritarian Korean government, and played a critical role in General Park Chung Hee’s industrialization drive in the 1970s. In the process, chaebols diversified and became large, politically connected business groups that eventually came to drive the majority of economic activities.

SK Group was a typical Korean chaebol. Founded by Chey Jong-hyon in 1953, it got its start by taking over the assets of a former Japanese-Korean textiles plant, becoming Korea’s first producer of polyester. In the 1970s as part of General Park’s heavy industrialization drive, it expanded upstream into oil refining, following the vertical integration strategy that was common with many chaebols. It also started diversifying into other areas, notably purchasing the famous Walkerhill hotel (and casino). In the 1990s, it moved into telecom, taking over Korea Telecom and launching Korea’s first CDMA cellular network2. In 2012, through SK Telecom, SK Group moved into semiconductors with its takeover of Hynix, forming SK Hynix. Today SK Group is the second-largest Korean chaebol.

As their capital bases expanded, their advantages were eroded by the burden of managing a large, diversified business: less focus and misaligned incentives resulting in declining capital efficiency. It is impossible for a large conglomerate to be great at everything. Today, most successful companies are ones that have developed specific advantages under competitive conditions and emerged as focused operators with scale, company culture and expertise to continue building on these advantages.

As the Korean economy entered the 1990s, cracks started to appear: chaebols were experiencing declining capital efficiency and there were general worries that they wielded too much power, often squeezing or crowding out smaller businesses and trying to lower market competition. Chaebols would bribe politicians in exchange for government contracts or gain preferential treatment3.

That said, on the surface, the economy was strong and people’s spirits were high. The global economy was basking in the post-Cold War glow. Netscape had gone public in 1995, ushering in the Internet era and “New Economy”. Korea was about to join the OECD at the end of 1996, a symbolic culmination to a stunning rise from the war-torn aftermath of the Korean War to a fully developed country within two generations.

In 1996, SM Entertainment — not a traditional diversified chaebol4, but part of a new generation of entrepreneurs from the 1980s — created the first modern Korean boy band called Highfive Of Teenager, or H.O.T., for short. Long before BTS and Blackpink or turn-of-the-millenium groups like Jinusean and Turbo, there was H.O.T. On September 7th, 1996, the catchy single “Candy” was released and it came to epitomize the “shamelessly happy” cheerfulness and optimism of mid-1990s South Korea (as well as, clearly, “peak baggy” in fashion).

The Asian Financial Crisis and Aftermath

On July 2nd, 1997, the Thai baht collapsed as its government ran out of hard currency to maintain its hard peg to the U.S. dollar. This immediately triggered capital flight as investors pulled capital out of emerging Asian economies in cascading fashion. By November, the Korean won devalued by more than half, putting intense pressure on a banking sector that had borrowed heavily in U.S. dollars with short-duration debt5. This cut off liquidity to the over-leveraged chaebols and many of them were suddenly teetering on the brink of bankruptcy.

With Korea running out of hard currency, the IMF stepped in with a rescue plan that forced Korean chaebols to undergo significant reforms. Widely characterized as a “tough medicine” approach, the adjustment period was particularly difficult featuring large-scale job losses and sending Korean households scrambling to dig out their jewelry in a patriotic drive to meet the Treasury’s need for hard assets to pay off foreign creditors.

But looking back, these reforms were also responsible for setting the foundation for the “leveling up” of the South Korean economy we see today. Today, with the benefit of more than two decades of hindsight, the South Korean economy has continued its ascent into becoming one of the world’s most technologically advanced economies. Korean chaebols emerged from the aftermath of the Asian Financial Crisis more focused, stronger operationally, and more capital efficient. Groups such as Samsung Electronics and SK Group were able to make the leap from investing in capital-intensive heavy industrial businesses to investing in technology and R&D-intensive efforts like consumer electronics and semiconductors.

Not only that, they were able to build global brands and eventually catch up and surpass their Japanese counterparts whom they had modeled themselves for many decades. Along with Taiwan, South Korea is one of the most successful examples of economies that have been able to make the leap from “dirt poor” to “fully developed” in three generations.

Parallels and Lessons for LGFVs

Like chaebols, LGFVs are conglomerates that hold a diverse collection of assets, businesses and investments. Today, similar to chaebols in the 1990s, they are facing similar issues of declining capital efficiency, large debt burdens and most recently, liquidity issues.

Average Korean per capita GDP in 19966 was $13,400, which was around 44% of then-prevailing U.S. GDP. Today, China’s per capita GDP is approximately $13,700 which is around 17% of U.S. per capita GDP. On a PPP basis7, the percentages are closer: Korea was 47% of U.S. GDP while China is 29%. As described earlier, Korea had relied on more foreign debt (net off against fewer foreign assets) — which is what eventually led to its currency crisis — while China holds a large net foreign asset position8.

In lining up LGFVs and chaebols, there are several interesting parallels and potential lessons as we think through potential reform pathways.

Enforcing discipline via crisis

Export-oriented chaebols (or divisions of chaebols) had a natural disciplining mechanism in competition in the global markets9. But for domestic markets, there was less of a natural disciplining mechanism — and domestic units of chaebols had this tendency to vertically integrate in and dominate industries they participated in.

In the past, the Korean government was successful in “enforcing discipline” to make sure chaebols “toed the line” on national development policies. One way was to provide rewards to chaebols that were successful at the export game. But domestically it was trickier to maintain that natural competitive dynamic — by the 1990s, chaebols had gotten so big with “crony capitalist” behavior and corruption on the rise.

From that perspective, the Asian Financial Crisis may have ironically been a blessing in disguise for the long-term health of the Korean economy. The currency crisis and resulting debt crisis gave Korean policymakers the leverage they needed to push down significant reforms and force chaebols to adjust.

Today, LGFVs — which are exclusively domestic concerns — face similar issues. Here the central government in its oversight role has responsibility for enforcing discipline. One might even view the “Three Red Lines” tightening in August 2020 as a self-induced crisis for China analogous to the exogenous Asian Financial Crisis for Korea.

Today as a result of the tightening, many LGFVs are struggling with liquidity, which gives the central government more leverage for potential reforms. A critical question is whether or not they take advantage of that leverage. Because it is self-induced, there could be a higher temptation to forego “tough medicine” reforms that would likely be better in the long run, as South Korea has experienced.

Focus and specialization

Chaebols played a dominant role in powering the growth of the Korean economy participating in nearly every sector of the economy. However, this led to the problem of “di-worsification” and declining capital efficiency. The IMF reforms and bailout conditions targeted this by forcing consolidation of business groups around “core competence areas”10. This meant forcing chaebols to consolidate or get rid of under-performing divisions, e.g. Hyundai Motors taking over Kia, dissolution of Samsung Motors, and the sale of Daewoo Motors to GM.

In a similar way, LGFVs need greater focus and specialization. Assets may need to be moved around to create more focused operations with greater scale (sector consolidation). There should be a realignment of the best industry operators in charge of these more focused asset groups. The way in which different asset categories are financed could also use change — infrastructure assets attract a different type of investor and capital base compared to real estate, for example.

Corporate governance

Similar to inter-LGFV financial flows, chaebols were held together by intricate cross-holding structures that allowed the head families to exercise control with only minority ownership. Chaebols often lent to and borrowed from each other as well. Additional IMF reforms targeted corporate governance with the IMF recommending “capital structure improvements”, “elimination of cross-debt guarantees”, “reduction in indirect cross-ownership”, and “enhancement of management transparency and accountability”.

Many LGFVs also feature complicated cross-holding structures, the result of creating co-investing syndicates with other LGFVs on common projects. For example, HSR projects often involved multiple cities and counties along the proposed rail line, with many separate LGFVs involved. To improve corporate governance, in addition to the greater focus and specialization discussed in the previous bullet point, LGFVs may also need to unravel some of these cross-holdings.

Financial reforms

The close relationship between the government, banking system and chaebols was a target for reforms. Banks were critical to the growth of chaebols during their rise but this eventually contributed to moral hazard issues, over-investment and declining capital efficiency. The Korean financial sector was under extreme stress during the financial crisis and its need for capital injections helped to break the “too close” relationship between chaebols and banks.

Many of the working capital issues with LGFVs are the result of local officials of the “mayor economy”11 trying to get around the rigid formal banking sector and turning LGFVs into quasi-financial intermediaries. This has increased the risks posed by creating “hidden” informal debts in the form of inter-entity borrowing and lending, stretched payables and other types of liabilities.

Role of private capital

One interesting thing that emerged out of the Asian Financial Crisis was the rise of private equity’s role in the Korean economy12. The South Korean economy was desperately in need of capital and private equity firms rose to the occasion, attracted to potentially attractive investment opportunities in the aftermath of the crisis.

It is notable here that private capital also played a role during China’s own large-scale SOE reforms in the 1990s. Many assets that were “let go” by the state ended up in hands of the private sector and, motivated by wealth creation, managed the assets better. Industrial sector returns on capital improved through the 2000s and made many of these entrepreneurs very wealthy.

The role of private capital is related to focus, specialization and consolidation themes above. Bringing in capable operators and providing incentives to succeed (wealth creation) is a powerful force that could “level up” industries struggling with capital efficiency. It needs to be handled carefully to avoid asset-stripping and more clear rules need to be setup defining the role of the local government vs. the private sector — but these are the types of reforms that could usher in a multi-decade tailwind just as the 1990s SOE reforms did for China and the IMF reforms for Korea.

Differences between chaebols and LGFVs

It is also important to note the major differences between chaebols and LGFVs:

• LGFVs are purely state sector entities; Korean chaebols were family-run private sector companies — with close linkage to the state.

• LGFVs were restricted to operating in less attractive (from a financial perspective) sectors and subject to adverse selection bias.

• While LGFVs also took a quasi-vertically integrated approach to real estate (e.g. investing in local construction companies, building tourist sites), supply chains are an order-of-magnitude less complex for land development as they are to industrial enterprises.

• Korean chaebols had some crown jewels like Samsung Electronics and Hyundai Motors; the best asset held by provincial governments is probably Kweichow Moutai.

Perhaps it would be more accurate to draw parallels with chaebols if you exclude their better assets: e.g. Samsung Group excluding Electronics and Heavy Industries. In other words, LGFVs are most comparable to the lower-performing subsidiaries or smaller and less well-run chaebols like Halla Group, the “bottom of the barrel” of chaebol assets, if you will13.

However, this distinction does not change the fact that the 90s-era Korean chaebols had taken its model to its limits, reaching pareto efficiency, and needed significant reform just like Chinese LGFVs need major reforms today to be able to advance to the next level.

Hopefully in the next post I will be be able to finally discuss some potential reform pathways for LGFVs.

As you may have noticed on Twitter and Substack, I have been wrapping my head around Chinese local government assets and finances to better understand the nature of the bulk of the country’s investment in recent years and to envision what a reform path going forward might look like in the coming years. I shared my thoughts in a series of related essays:

• Project Finance with Kweichow Characteristics

• LGFVs: the Good, the Bad, and the Mierda, Part I

• LGFVs: the Good, the Bad, and the Mierda, Part II

There are still some related areas I am thinking through and I will hopefully be able to wrap this series up in the coming weeks.

I’ve officially renamed the sub-domain readwriteinvest.substack.com. Links to the old domain should forward to this new one.

Like everyone else it seems, I setup an account on Meta/Threads under “glenn_luk”. Feel free to follow me there! I am not quite sure what to do with it for now but I may start posting there as well. I will go wherever the best discussions are.

musings

Impact of transport electrification on China’s oil consumption (2023.7.9)

The sharp rise in electric vehicle and HSR adoption is going to have a major impact on China’s oil consumption. Motor vehicles (commercial and passenger) account for approximately 41% of China’s oil demand. Over the next two decades, this could fall by up to 90% depending on how quickly the existing ICE fleet is retired.

How likely is this to happen? The fact that this is based primarily on China’s drive to achieve energy independence vs. de-carbonization increases the chance of it happening, in my view.

China’s declining TFP and implications on LT growth (2023.7.8)

Total Factor Productivity (TFP) has declined in China since the Global Financial Crisis with GDP increasingly dominated by large increases in capital investment. While many point to this and see secular, irreversible productivity decline, I note how the changing composition of gross capital formation has also played a role significant role. The contribution of TFP to GDP growth could very well rebound as China gradually winds down a multi-decade long investment-centric housing and infrastructure boom.

Risk of a balance sheet recession in China? (2023.7.5)

Nomura economist Richard Koo discussed the risk of a balance sheet recession in China based on his extensive experience studying the Japanese economy, particularly as it underwent its stagnation period starting in the 1990s. You can watch the video here.

Overall, I agreed with some but not all points. Some short-term stimulus can help get liquidity flowing again but in the longer-term, real work and reform is required both to (i) manage the large pile of assets created at the local government level and (ii) improve the process going forward for continued asset development.

There are some key differences between China today and 1990s Japan that make the comparison difficult. Perhaps the biggest of all is that China still has significant catch-up growth, with its urbanization rates at 65%1 — compared to a Japan that was fully urbanized (80%+) by the 1990s. This concept of catch-up growth also relates to the demographics question: while China might seem like it is closer to demographic decline than Japan in 1990, it still has some additional rural-to-urban migration tailwind left.

I agree with Koo that the geopolitical environment is more challenging. But I also pointed out how China’s economic scale (12x the population of Japan) means that it can pursue a radically different economic strategy.

Finally, I thought Koo made a great point about how China has the benefit of learning from Japan’s experience. Japan was dealing with some of these economic issues for the first time — like the balance sheet recession — and this likely contributed to the undemanding pace of its reforms.

The paradox of renewables and coal (2023.7.3)

There is a building boom for both renewables and coal in China today. For some, this seems like a paradox. But there are good reasons. Intermittent renewable needs backup power and China is looking to coal as one of its backup solutions. China needs to replace hundreds of gigawatts of coal plants built in the 1990s and 2000s with cleaner, more efficient designs that are more suitable for this backup role. New nameplate capacity does not mean actual coal consumption will increase — capacity factors have been declining and are expected to continue declining as coal takes on more of a backup role over time.

Technology enabling local production of food, fish and freshwater (2023.7.3)

New technology advancements are enabling a scalable, sustainable model that integrates vertical farming, aquaculture, conservation techniques, freshwater production and renewable energy that countries that have historically lacked abundant arable land and water supplies can harness.

Technology shifts the economics of food supply in favor of local. The old model of growing food in one place and transporting it thousands of miles is being supplanted by food that is grown locally using the latest vertical farming techniques. There is significant energy wastage built into such a long food supply chain that is analogous to the energy wastage from internal combustion engine.

charts & data

After three years of the pandemic, passenger rail in China is expected to recover to pre-pandemic levels. Recently, China Railways reported 311 million passenger trips in May, which works out to an annualized 3.7 billion riders per year.

from the archives

Will China become an innovator? (2018.12)

Here I discuss China’s ability to innovate in the context of one of my favorite childhood PC games: Civilization. I learned a ton from that game.

Will "shadow banking" in China lead to a financial crisis? (2018.10)

Credit disintermediation “with Chinese characteristics”. Five years later we are talking again about hidden debt and the need for reform.

Could China become another Japan? (2016.1)

China is Japan in the 70s, not the late 80s. Seven years after writing this article, China may now be closer to late 70s or early 80s Japan. Maybe in a decade it will finally be late 80s Japan.

Gold standards of economic development success stories (2015.5)

The fascinating development stories of Taiwan and South Korea.

diversionary

• Project Finance with Kweichow Characteristics

• LGFVs: the Good, the Bad, and the Mierda, Part I

In the previous post, we examined various categories of assets that make up the bulk of the LGFV portfolio. In this essay, we will step out and see where these assets fit and intersect with others in the broader Chinese economy, namely assets held by the central government and the private sector.

The Holy Trinity of Assets

One way to look at the Chinese economy from an asset perspective is to divide it into three distinct pillars:

1. Private Sector

2. Central Government

3. Local Government

Each grouping has its advantages and disadvantages within the context of the Chinese system and over time that has led to each finding its way to certain preferred sectors and types of economic activity.

Private Sector Assets

The private sector emerged over the past four decades out of China’s once-100% planned economy. Much of the development of the private sector has been driven by entrepreneurs trying to find spaces within the Chinese economy where they could pursue either “Blue Ocean” strategies or co-exist or compete successfully with the state.

Today, the private sector has taken the lead in sectors like agriculture, manufacturing, travel and tourism, technology, Internet and real estate. Examples of successful private sector companies include Alibaba, Foxconn, Huawei and BYD.

Private sector companies are generally less capital-intensive than SOEs. Part of this is simply a function of capital being more expensive, typically relying on higher-cost equity to capitalize their businesses instead of relatively inexpensive bank loans. This served as a forcing function for them to operate more efficiently. Another part of this was being explictly crowded out of capital-intensive domestic sectors by the state itself (more on this below).

Market competition weeds out poor performers, rewarding better operators with higher profits that can be re-invested back into the business via retained earnings. Despite its cost of capital disadvantages, higher efficiency and growth rates have led to impressive accumulation in its asset base: private sector industrial companies held aggregate assets of approximately $6 trillion as of the end of 20221.

Off this asset base, the private sector is generally able to generate significantly higher returns on capital compared to SOEs: