PodBrowser
Impact Theory

Planned Episode 8/18/2026

Tuesday, 18 August 2026 · 4 min read · Listen to the episode ↗

In July 2026, China recorded a historic contraction in total social financing and new RMB loans, with outstanding loan growth hitting consecutive record lows worse than both 2024 and 2025, driven by bank de-risking after LGFV loan restructurings slashed yields from roughly 7 to 9 percent down to 2 percent.

China's banking sector recorded a historic contraction in total social financing and new RMB loans for July 2026. Although July is a seasonal low point on the Chinese lending calendar, the speakers argue this amplifies rather than excuses the severity. The outstanding stock of RMB loans has been hitting record low growth rates consecutively for several years, and conditions in 2026 are described as worse than both 2024 and 2025, confirming an accelerated downturn the channel had been forecasting since the prior summer.

Bad loans have accumulated across Chinese banks not only from real estate but also from Local Government Financing Vehicles, corporate structures created by local governments to borrow commercially and circumvent CCP rules prohibiting direct government lending. Banks were earning 7 to 9 percent on informal LGFV loans before the CCP pushed to restructure them at approximately 2 percent yields. Banks were reluctant to accept the swap given that local governments had originally pressured them into making those loans. In response, Chinese banks have de-risked their balance sheets, pulling back from household and corporate lending and accelerating purchases of government bonds as a flight to safety. This drove long-term government bond yields down sharply, and in 2024 the PBOC warned banks to stop buying government bonds ahead of anticipated stimulus-driven bond supply. Banks ignored the warning and were correct to do so, as PBOC stimulus did not materialize as predicted and bond supply did not move yields.

The speakers invoke Milton Friedman's interest rate fallacy and cite Steve Keen to argue that low interest rates reflect economic weakness rather than stimulus, and that when debt created as new money is repaid it ceases to exist under double-entry accounting, removing liquidity from the system. Chinese aggregate financing to the real economy has continued rising, but the source is government bond issuance rather than bank lending. Increased government borrowing since the September 2024 bazooka stimulus has coincided with economic growth going down, not up. The Chinese stimulus mini-cycle is described as self-reinforcing: stimulus fails, government does more, stimulus fails again, government does even more, piling government bonds with no growth effect. A short-run pickup in Chinese lending in early 2025 was attributed to the September 2024 stimulus and characterized as a fluke.

The speakers draw a parallel to Japan's post-1989 housing bubble, where consumer and business psychology shifted toward debt repayment and extreme caution, stalling the economy for decades and creating zombie corporations. China has been trying to disguise the severity of its downturn to maintain investor confidence and prevent domestic psychological pullback in spending. China's exports remain the one part of the economy still performing, and local governments facing growth quotas will push businesses toward export markets as domestic consumption weakens. The speakers predict China flooding global markets with cheap goods will force the US to raise tariffs further, hollow out the European middle class, and risk making Europe an economic vassal state of China. A further prediction holds that China, Japan, and Europe will reduce purchases of US debt simultaneously at a time when the US needs to continue large deficit spending, pushing the US government to lower interest rates and debase the dollar. If both the US and China race downward simultaneously, the speakers characterize the outcome as a potential global depression.

US consumer spending in July 2026 dropped dramatically, and the speakers are watching August data to determine whether a rebound occurs or a downward trend consistent with depression economics is taking hold. Depression economics are described as psychology-driven, where saving instead of spending collapses the velocity of money, destroys jobs, reduces liquidity, and creates a self-reinforcing spiral of the kind Japan could not break for decades. Jeff Snyder argues that expectations are the most powerful force in economics and that knock-on effects in the real economy follow directly from what people believe will happen. The speaker ties this to a current shift in American psychology toward anti-wealth sentiment, arguing that if workers and investors come to believe that billionaires are stealing from them and corporations exploit employees, participation in the economy will decline and growth will stall. Supporting evidence cited includes real wages failing to keep pace with inflation, jobs at an all-time low, and people leaving the workforce in significant numbers.

The speaker's current positioning reflects acknowledged uncertainty about timing even where he believes his directional read is correct. He is limiting exposure to AI investments without eliminating it entirely and increasing exposure to short-term US debt while avoiding full concentration there, explicitly flagging the risk that holding short-term US debt for three years could mean absorbing three years of inflation that returns may not offset. He attributes his hedging approach to Warren Buffett's principle that diversification protects against ignorance and notes that running a company as his primary focus means he must account for his own limitations as a part-time market participant. The Treasury does not control rates on anything beyond short-term bonds because longer-duration bond rates are set at auction by banks rather than by government directive, and rising yields are complicating the traditional safe-haven role of bonds. AI stocks and housing are at all-time highs while crypto sits at a relative low.

This summary was generated from the episode transcript and can contain mistakes.