Preface: Is the “answer” to China’s economic conditions expanded central government debt and fiscal expansion? Well, that depends on how we diagnose the nature of the situation and the challenges at hand. In this essay, I leverage Li Daokui’s recent interventions to reflect on my own unfolding analytical frame and diagnostics of China’s economic conditions and suggest that the “answer” isn’t necessarily just a question of “more” but a question of “faster.”
Li Daokui is a prominent Chinese public intellectual-economist. His recent comments (original Chinese here) on China’s economy sparked something of a frisson of excitement, again mainly amongst those who’ve been critical of China’s economic model, seeing in Li’s comments some kind of validation of their own complaints. Given his prominence, it’s undoubted that his commentary would and indeed, should, spark some discussion. This short note is a response to his comments, and in doing so, seeks to clarify and extend something of my own diagnosis and discussion of China’s economy and how we can analyse it.
Li argues that China’s most pressing macroeconomic challenge is not a so-called K-shaped divergence. Rather, for him, it is a sustained period of what he calls economic coldness over the past three years. This coldness is evidenced by a broad unemployment measure (not the official statistics) that incorporate so-called “discouraged workers,” which elevates the rate to a little over 10% (from the official 5.1-5.3% range); the historically rare decline in fixed asset investment growth evident in the last quarter; and the financial blockages embodied by indebted local governments. The remedies, on these views, are for the central government to expand debt more decisively and refinance high-cost local government liabilities, convert unsold housing into other uses, and provide support for livelihood projects. Given that central government debt to GDP ratios remain relatively now, Li argues that addressing local government debt would raise economic activity overall — or to put it another way, lead to economic warming.
This diagnostics and remediation plan is coherent on its own terms. However, I would suggest that it remains somewhat “coarse.” The remainder of this essay explores what I mean by that.
To begin with, Li’s analysis is anchored primarily in aggregate flow and stock measures – GDP shares, investment in aggregate, debt ratios and such like. It therefore tends to treat the coldness as a function of insufficient total demand that can be addressed by expanding the net stock of public purchasing power. This is a fairly conventional fiscal expansionary approach. Yet, before we wonder whether such an approach is meritorious, we need to take a step back and ask a prior question: in what sense is the economy “cold”? By examining what “cold” could actually mean, we can better understand the extent to which Li’s wider diagnostics and prognosis is likely to be most effective.
Let’s start with some basics. We know that real disposable incomes in China continue to rise at roughly 4-4.5% per year. This has been the case for many years now, including the last three — the period of Li’s “coldness.” We also know what services consumption has been a growing proportion of household spending, contributing the lion’s share to overall consumption growth, which totalled 2.8% for H1 2026. Retail growth in H1 2026 was recorded at 1.7% with consumption of services at closer to 5.4%. Industrial profits, after several years of relative subdued stability, accelerated markedly in the first half of 2026, rising over 18%. Overall GDP growth was 4.7% for the year to June 2026.
These core data point not to an economy that is contracting or suffering deep demand deficiency. Output growth continues apace, and investment and consumption continue to drive overall GDP expansion. Net exports contributed about 18% to H1 GDP growth, with the remainder accounted for by domestic demand (investment and consumption). Indeed, around 75-80% of manufactured output is consumed domestically and PMI indexes have been expansionary for the most part. Manufacturing capacity utilisation hovers between 72% and 75% for the first half year (73.3% on average). Inflation overall (CPI) is modest, though factor input costs have fluctuated somewhat particularly with the supply disruptions occasioned by the war against Iran impacting the flow of energy and chemicals from the Persian Gulf. All that said and done, the Chinese economy is not suffering from contraction.
So, what is the coldness that Li describes? He specifically speaks of contraction in volume-intensive investment, soft private fixed asset formation and elevated broad labour under-utilisation. At the same time, these indicators co-exist with positive income, output and profit growth. The reconciliation of these two lies in the question of velocity.
As I have indicated before, China’s income velocity of money is at historically low levels of around 0.41. The overall money stock in the economic system is large and growing, but is turning over relatively slowly. Evidence? Household deposits are rising, corporate cash is accumulating and payment cycles are stretched. The same underlying income and profit generation produces weaker measured expenditure and investment. Rising real incomes can co-exist with subdued retail momentum if a large share of the income is held idle. Strong industrial profits growth can co-exist with softening fixed asset investment if retained earnings are not being mobilised to expand new capacity at the same rate as before. As a result, the aggregate picture can look cold in the investment and broad labour-market metrics, even while income and profit flows remain positive.
A perspective focused around questions of structural change and capital rotation, combined with attention to monetary circulation, suggests that the binding constraint is less a scarcity of money capital per se than it is one of subdued rates of money turnover. This is a question of flow. The old growth drivers of real estate and the dense web of related intermediate demand, coupled with large-scale urban and transport infrastructure, have slowed dramatically since 2020 (on which see my previous essay). This is the consequence of policy-induced sector deleveraging. Yet, this contraction was not – and has not been – coupled with a collapse in either employment or real wages. According to official statistics, between 2019 and 2026, unemployment has hovered between 5.1% and 5.3% nationwide. This prima facie suggests that the contraction in real estate was offset to a large extent by new subsystems that emerged, accelerated and expanded. That real wages also grew during this period suggests that system adjustment to real estate contraction did not lead to labour market absorption via compressed wages; rather, the reverse was the case.
High productivity manufacturing in new energy vehicles, batteries, renewable energy systems and technologies, electronics and increasingly advanced machinery and equipment, including robotics, together with associated upstream supply chain reorganisation, have absorbed labour and generated income even as the old volume-intensive real estate and related activities contracted.
When we look at profit growth and investment by sector, we see that there is strong growth in certain sectors, pointing to ongoing restructuring of the economic system’s production coefficients. As I have argued previously, these shifts improve thermoeconomic efficiency and reconfigure intersectoral proportions in ways that standard one-commodity or aggregate models capture poorly. Pasinetti-style structural dynamics make the required re-proportioning of demand and capacity across vertically integrated subsystems visible; Li’s framework largely does not. I would suggest that contrary to Li, replacement growth drivers have already emerged. Whether they are operating at full pace, or at the capacity potential as yet, is a separate question. In other words, things are far less “cold” than Li would imply.
If there is a retardant, it’s the fact that we have yet to see the accelerated obsolescence of “zombie” firms — that is, low-productivity enterprises that have been kept afloat by soft credit or local protection. It should nonetheless be noted that a large proportion, about 40% of so-called “zombie” firm are a legacy of real estate deleveraging. The national government is aware of the issue, and has initiated measures to accelerate the euthanising of such enterprises. However, while there has been ongoing talk of a need to facilitate orderly obsolescence, the current balancing of gradual obsolescence against the expansion of new growth activities is, in principle, a sensible approach. Keeping these zombie firms in existence is not pure deadweight loss: they continue to support employment and therefore consumption demand. Accelerating their exit without adequate mechanisms for worker transition risks raising under-utilisation and weakening demand precisely when velocity is already low. Proper plans for reskilling, job matching, and income support during transition are needed if China is to successfully accelerate the processes of creative destruction, without generating new circulation frictions by way of rising unemployment.
Mainstream analysts often speak of China’s high savings rate, blaming it for what they call “weak” consumption. This has the causal effect around the wrong way. High savings is a result of high real incomes growth, a residual, which in turn is a function of investment expansion. Rising real incomes does not lead to a commensurate yuan-for-yuan increase in consumption expenditure, because the income elasticity of consumption is less than one. As such, while the cost of living remains relatively low, in conditions of rising real incomes the residual — savings — continues to rise. Households save what they can, not what they must.
Profits are an enterprise’s savings. That these savings aren’t mobilised and remain idle for long periods of time is one of the reasons for the apparent “cold.” Aged receivables and systematically extended payment terms extract working capital from suppliers, especially smaller firms. This has been a persistent feature of Chinese supply chains, with reports dating back to at least 2014, in which larger firms hold cash and delay payments to others. Debt recovery of unpaid bills can take a long time, disincentivising many firms from doing so. New regulations were enacted in early 2025 that aim to compel larger firms to settle trade invoices within 30 days, and at most 60 days. See here for details on the regulatory provisions. Supply chain financing remains an undeveloped arena, as my book China, Trust and Digital Supply Chains discusses, and is ripe for development.
Local governments, constrained by vertical fiscal imbalance (heavy expenditure responsibilities relative to limited own revenues), frequently delay payments, turning liquidity mismatches into further freezes in the production chain. Tightening payment discipline while developing genuine supply-chain finance instruments that mobilise dormant balances into short-duration fixed-income vehicles can accelerate cash flow without expanding the overall monetary stock. Directions in monetary policy and institutional innovation announced recently by PBOC Chair Pan Gongsheng, which I discussed previously, would suggest that one of the key targets of financial products innovation is to tackle this issue. Progressive vertical fiscal rebalancing that makes central support more reliable for required local payments — while conditioning it on verified clearance of arrears — addresses the same problem at the intergovernmental level.
An increase in velocity of circulation would have a marked impact on overall GDP. As Velocity V = nominal GDP / M2, at roughly 0.41, a rise to 0.45 (holding the money stock fixed) implies nominal GDP higher by a factor of 0.45 / 0.41 ≈ 1.098, or nearly +10%. Given that M2 is already in the mid-to-high 300-trillion-yuan range, these are not trivial increments; they are macro-significant. Simply raising velocity from 0.41 to 0.45, holding the money stock constant, would lift measured nominal GDP by roughly 10 per cent. Even a more modest recovery — say from 0.41 to 0.43 — would still lift measured nominal GDP by about 5%. This is no small order, however, and we shouldn’t expect some instantaneous transformation in the velocity of circulation.
There is a further inconsistency in much of the western commentary surrounding Li’s argument. Over the past twelve to twenty-four months, many mainstream observers criticised “involution” — excessive competition, overcapacity and price wars — as evidence of too much investment and capacity expansion. Yet the same voices now praise the diagnosis of an investment slowdown that must be remedied by renewed fixed-asset growth. This is difficult to reconcile. Part of the recent deceleration in overall fixed-asset investment was precisely a result of an attempt to put the brakes on capacity expansion in sectors already experiencing intense involution. A wholesale call to reverse that slowdown, without distinguishing between efficiency-enhancing, higher-productivity investment and a return to volume-driven overcapacity, risks re-igniting the very problems those critics previously condemned.
On the question of savings, standard explanations revolve around claims of either precautionary savings or balance-sheet impairment resulting from falling house prices. These are not particularly persuasive as primary drivers once the data are examined closely. I actually discuss the supposed mystery of Chinese household savings in Thermoeconomics in a Time of Monsters, so won’t go into things in too much detail here. Suffice to note that real disposable incomes continue to rise, with rural incomes growing faster than urban ones and lower-tier cities often outpacing first-tier cities from a lower base. Wage and salary income remains the dominant component of household receipts; property income is a small share (~8%). China’s social expenditures as a share of GDP have risen substantially and now approach levels seen in other large upper-middle-income economies. While residual gaps in benefit depth and portability remain, though in the main they have closed as hukou restrictions have faded away, the claim that a uniquely shallow safety net is forcing high savings is difficult to sustain as the dominant explanation.
In the first half of 2026, rural per capita disposable income rose 6.4% nominally and 5.5% in real terms; urban income grew 4.4% nominally and 3.4% in real terms. The same differential has been visible for several years. That is, rural growth has consistently outpaced urban growth, and the urban–rural income ratio has continued to narrow. Within the urban system, absolute levels remain highest in the first-tier cities, but percentage growth (and especially growth in real purchasing power after housing and living costs) is frequently stronger in second-, third- and fourth-tier cities that start from a lower base. Cost-of-living adjusted disposable income can already be competitive in many of those places.
This matters for velocity and residual mobilisation. Households at lower absolute income levels typically have a higher marginal propensity to consume. Regions that are already recording faster income growth therefore offer a higher “bang for the buck” when additional incentives or infrastructure that support spending and local services are introduced. Creating more visible “buzz” — better local services offerings, cultural and experience amenities, improved digital and physical connectivity, targeted support for local entrepreneurship and supply-chain linkages — can accelerate the translation of rising incomes into circulating expenditure rather than further accumulation of idle balances.
It also aligns with the structural shift already under way. As high-productivity manufacturing and related services expand beyond the traditional coastal first-tier hubs, the income gains are more widely distributed. Channeling complementary demand-side and services-side support into the places where those income gains are arriving most rapidly is a spatially efficient way to raise the overall turnover of money without simply flooding already high-income, lower-MPC locations.
Engel’s Law helps fine-tune the picture. This speaks to the compositional change in consumption patterns with discretionary expenditure growing as a proportion of total spend, as household incomes rise. As a substantial segment of households reach relatively comfortable levels of material provision, incremental income is increasingly directed toward experiential and quality-of-life services — domestic tourism, education, cultural activities, and seniors’ travel and learning. The evidence on services growth confirms this emerging pattern. These categories are labour-intensive, geographically dispersible, and — importantly for our purposes here — high-velocity. Payments are usually instant, made when services are procured. It’s a “real time” economy. Because income elasticity of expenditure is higher among lower- and middle-income households, and because incomes are rising faster in rural areas and many lower-tier cities, targeted support for services capacity and amenities in precisely those regions converts rising incomes into circulating expenditure more efficiently than broad fiscal volume directed at already high-income locations.
A recent PIIE assessment is also relevant to understanding the dynamics of household savings. According to this report, social expenditures as a share of GDP in China have more than doubled since 2010 and now sit at roughly 9.1%. This is approaching the levels of other large upper-middle-income economies such as Mexico and Turkey. Coverage of basic pensions and health insurance is near-universal in formal terms, and many education and health services carry substantial subsidies. The residual gaps (depth of benefits, rural–urban differences and portability for flexible workers) still matter, but the simple claim that a “shallow or limited” safety net is forcing high savings is weaker than it is often presented. Elevated deposit growth can occur for many reasons other than pure precautionary motives — preference for liquidity in an environment of uncertain returns, limited attractive risk assets, general thrift or simply the mechanical residual from rising incomes when desired consumption growth lags.
Wealth effects from property-price declines are real for higher-income households with significant housing exposure, yet these households typically have lower marginal propensities to consume; for the majority, income is wage-driven (58%) and daily necessities remain readily available and inexpensive. Property-income flows are modest (~8%), and the households whose net wealth has been most impaired tend to be higher-income ones whose marginal propensity to consume is already lower. Balance-sheet repair and sentiment effects exist, particularly among those with leveraged property or multiple units, but they do not dominate the aggregate consumption story for the median household. The rapid recovery in certain services categories is consistent with COVID-era restraint fading rather than a permanent wealth-driven collapse.
Taking all of this into consideration, fiscal expansion retains a role, particularly where it reduces genuine liquidity mismatches or strengthens the social foundations that support confidence. On this, I can only concur with Li’s proposition that the central government plays an even more active role in restructuring the balance sheets and fiscal imbalances vis-a-vis local governments.
Beyond what is already planned in terms of public infrastructure investment, however, the higher-leverage task is to raise the turnover of the existing stock of money capital. Structural re-proportioning is under way; the residual after rising real incomes is substantial; and the frictions that keep that residual idle are identifiable and addressable. The appearance of coldness is in significant measure a velocity problem. Focusing policy on circulation — payment discipline, supply-chain finance, services development matched to actual income elasticities, carefully managed obsolescence of low-productivity capacity accompanied by transition support, and spatially intelligent incentives — offers a more precise route to restoring momentum than treating the problem primarily as a shortfall in the net stock of public demand, or reversing an investment deceleration that was itself partly a response to involution.
For mine, we are dealing with the ongoing challenges of dynamic balance; the issue is less about “more” but more about “faster.”
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.