This article was sent me by Andrew Smithers on 1st May 2026. You can read about Andrew here.
Andrew Smithers
Introduction.
The world economy is currently fragile due to its imbalances, which include excessive share prices and, for several key countries, including France, Japan, the UK and the US, an unsustainable level of fiscal deficits. These are likely to be resolved by a financial crisis, whose trigger depends on two things:
- Which way AI affects the world economy.
- The policy response.
A large decline in real share prices and a major recession are highly probable. Their triggers, depth and timing will become more predictable, as we learn whether AI’s technology is predominantly either “hard-baked clay” or “putty-putty” or some combination of the two. Different policy responses, to whichever form AI takes, will be needed to maintain low and stable levels of inflation and unemployment, while resolving the economy’s imbalances.
New technology is termed hard-baked clay when it is embedded in tangible equipment and its benefits can only be realised by additions to the stock of capital; it is known as putty-putty when it can be used to improve the productivity of the existing stock of capital.[1] In either form, AI’s arrival will enhance the trend growth rate of the world’s economy. Short-term growth depends, however, on the level of current demand; output will fall short of its potential if unemployment rises above the non-accelerating inflationary rate of unemployment (NAIRU).
Putty-Putty.
If AI is predominantly putty-putty, employment is initially likely to weaken relative to output capacity, as companies can maintain output with fewer employees. Unless governments or central banks boost demand, companies will sack workers and output will fall as unemployment rises, so that profit margins and share prices will decline. To avoid a recession, the policy response would normally be some combination of fiscal and monetary stimuli. But today’s fragile economy is abnormal, because the usual tools for stabilising it are either unavailable or may not work. Reducing short-term interest rates will not support share prices if the decline is offset by a steepening yield curve as the cut encourages further rises in expectations for inflation, and supporting demand by increasing fiscal deficits could precipitate the bond market collapse, which the current level of the deficits already threatens to produce.
Hard-baked Clay.
If AI is mostly hard-baked clay, the level of tangible investment will rise as the expected return on it increases. Companies finance about two thirds of this with equity and one third with debt. The need for both will rise; companies will spend less on buybacks and takeovers and the stock market will fall.
Inflation will rise unless savings rise to match the increased investment. In aggregate, households or governments will have to save more to provide the debt finance needed. Over time, the additional savings needed to finance the rise in tangible investment will follow from the economy’s accelerated growth, but new tangible investment takes time to boost output, while immediately increasing demand. We will need some combination of smaller budget deficits and more household savings, either in response to higher interest rates or a financial crisis.
Both.
If the immediate increase in potential supply from putty-putty advances in AI technology are accompanied by an immediate rise in demand, due to increased tangible investment driven by new hard-baked clay technology, the trend rate of growth is enhanced, while no changes in interest rates or fiscal deficits are needed to maintain demand at the previous level of the NAIRU. But this is not sufficient to avoid a rise in inflation as the NAIRU is likely to fall due to an increase in frictional unemployment.
In equilibrium evenly balanced rises in AI’s two forms will accelerate growth with unchanged fiscal and monetary policies, but the immediate impact will be disruptive. A major shift in employment will be needed from sources of output benefiting from putty-putty change, first to the production of tangible plant and equipment and then to their operation. This will not be instantaneous, and will involve much disruption, which will only be resolved after a considerable time lag. When faster growth requires more tangible investment, a switch from the output of consumer goods and services to the output of investment goods is also needed, and must be matched by changes in financial flows, which policy adjustments must facilitate.
Corporate Savings and Cash Flow.
An important complication arises from the difference between changes in corporate savings, as shown in national accounts, and in the cash flow. Around two thirds of the increased investment by companies will need to be financed by equity, which will mainly be provided by reductions in buybacks and debt-financed takeovers. This will not show up as a rise in savings in national accounts, in which they are categorised as capital rather than income transfers. The reduction in buybacks will, however, lead to a falling stock market and a consequential rise in the required level of savings.
Household Finances.
Household savings tend to rotate around a stable level, which provide for a reasonable standard of living in retirement and as a precaution against the sudden needs that arise from illness or unemployment, and for short-term bouts of spending on holidays and Christmas. The household sector’s financial assets consist of cash, bonds and equities. Only bonds and equities are suitable for pension saving, as returns on cash are low and liquidity is not needed once adequate provision for it has been made, but only cash is suitable for other types of savings. In equilibrium households hold the amount of cash they need to match their requirements for liquidity, and the remainder is available to finance their consumption in retirement (their pension assets) and will sensibly be invested in bonds or equities. ”[2]
If household liquidity and the perceived needs for it are unchanged, households will invest their other financial assets in equities and bonds. Their relative prices will vary with swings in households’ hopes and fears for profits and inflation but are moderated by the offsetting responses of corporate leverage to changes in interest rates and profit margins. Changes in liquidity or its perceived need will, however, alter the total value of bonds and equities that households wish to own.
Interest Cover.
Companies, in aggregate, seek to maintain a stable level of interest cover, which is the ratio between their profits and interest payments.[3] They also prefer to borrow long-dated bonds rather than short-term debt, as this reduces their exposure to fluctuations in inflation. If bond yields fall or profit margins rise, they will tend to increase their leverage. When the economy is growing with low and stable levels of unemployment and inflation, it is stabilised by central banks who raise interest rates to offset inflation and lower them to stimulate growth. Provided that the yield curve is stable, changes in short-term rates will be matched by changes in bond yields, but the yield curve changes with inflation and becomes unstable if inflationary expectations alter.
When inflationary expectations are well anchored, and fears of unemployment are low, households will generally be happy to hold stable ratios of cash, bonds and equities. While temporary fluctuations in optimism about profits will encourage households to hold more equities, this will also tend to push up demand and be offset by changes in short-term interest rates. The offsetting response by companies tend to stabilise the preferred ratios of bonds and equities in household portfolios.
Household Liquidity.
Cash differs from bonds and equities in that its return is low and its nominal value does not fluctuate. It is therefore the sensible form to hold as a precaution against unemployment, but not for retirement savings. Provided expectations for inflation and worries about unemployment are contained, households’ portfolio preferences (their preferred ratios of cash, bonds and equities) tend to be stable relative to national income.[4] The two key causes of instability are changes in inflation and its expectations and in the liquidity of households or their perceived need for liquidity.
Household liquidity is altered by governments’ funding policies and the perceived need for liquidity is mainly affected by changes in fear, notably that of unemployment, and tends to rise in bear markets. As there is no parallel tendency for fear to fall in bull markets, it explains why equities fall more sharply in bear markets than they rise in bull markets.[5]
The Policy Challenge.
Successful policy, defined as adequate growth combined with low and stable levels of unemployment and inflation, will be difficult to achieve in the face of the unknown impact of AI and the current economic imbalances. I expect it to fail, because there are more sources of disequilibria than there are policy instruments available to balance them. Their current number is therefore inadequate, according to the cybernetics law called the principle of requisite variety. This states that “Anything which aims to be a ‘regulator’ of a system needs to have at least as much variety as that system.”[6] Its use in economics was first proposed by Jan Tinbergen and is thus known as the Tinbergen rule.
I noted in the introductory paragraph the imbalances of excess share prices and the unsustainable level of fiscal deficits. The policy instruments currently available to governments and central banks are changes in interest rates and fiscal deficits. Keynes showed that reducing interest rates fails to stimulate demand in liquidity traps. Unless supported by increased fiscal deficits, cuts in short-term interest rates have persistently failed to support demand since 2000, which is why national debt in several countries has reached unsustainable levels. It appears, therefore, that we have suffered this century from a structural liquidity trap and cannot therefore expect demand to rise in response to declining short-term interest rates. It also seems that national debt ratios can no longer be increased without causing a financial crisis and it may be already necessary to reduce them.
Policy Responses to AI.
If AI is predominantly putty-putty, policy will need to stimulate short-term demand which, in today’s abnormal conditions, neither monetary nor fiscal stimuli can be relied upon to achieve. Cuts in short-term interest rates may be offset by a steepening yield curve and hikes in fiscal deficits will increase the existing risk of a financial crisis. The use of either fiscal or monetary policy threatens to raise inflationary expectations which, due to the Iran war, are already far from quiescent.
If AI has a significant element of hard-baked clay, the rise in savings needed to finance the rise in tangible investment, (which will precede any increase in the output and income that it will produce), will reduce corporate buybacks and debt-financed takeovers, so that share prices will decline. At the same time savings will rise and demand will weaken.
Tax Restructuring – a Third Policy Instrument.
Policy based on classical, pre-Keynesian economics, assumed that growth, combined with low and stable levels of unemployment and inflation, could be achieved by changes in short-term interest rates. Keynes showed that in liquidity traps, variations in fiscal policy were also needed. In France, Japan, the UK and the US national debt ratios have reached levels at which fiscal stimuli are no longer an available policy tool. A third policy tool is thus now needed which will stimulate demand without raising fiscal deficits. This could be provided by tax restructuring, which leaves the fiscal deficit unchanged, but stimulates demand by reducing the net revenue from corporation tax and raising that from other forms of taxation. The net impact of these changes is to boost investment even more than it cuts consumption.
Three Provisos.
I am expecting policy to fail, but expectations are not forecasts. The least unlikely ways in which we could maintain growth and avoid high and volatile levels of inflation and unemployment are:
- The introduction of a third policy instrument in the form of tax restructuring. I think this is unlikely as the Stock Market Model (SMM) of economics on which it is based differs significantly from the Consensus Model (CM) on which current policy is based.
- It is possible that the current level of share prices can be maintained for many years ahead, by relying on short-term debt rather than bond issuance to finance the prevailing fiscal deficits, without raising inflation.
- The US stock market ceases to be extremely overvalued, through a rise in real corporate net worth, rather than a fall in real or nominal share prices. The putty-putty element in AI would need to be so strong that the rise in NDP from the existing capital stock would be sufficient to raise the real value of that stock by around a factor of around two. [7] As the produced capital/NDP ratio is mean reverting around 4 times, this would require a rise in NDP of about 50%.
All of these are possible, but none seem to me to be likely.
Conclusions.
- We are likely to experience another financial crisis within the next few years.
- The monetary and fiscal policy tools used to boost demand after the last financial crisis in 2008 are, today, likely to either fail or precipitate a crisis.
- While we cannot predict the timing, or even the certainty of the next crisis, it is highly probable. The fall in the stock market and the consequent recession are likely to be much more severe than last time.
- It is often claimed that “markets do not like uncertainty”, but today uncertainty provides their key support.
- The triggers for the next recession are likely to be falling profit margins or rising bond yields.
- Rising bond yields are likely if governments seek to fund current deficits or because expectations for higher inflation rise.
- Inflationary expectations are likely to rise if governments rely on short-term debt in preference to funding deficits, which will trigger a financial crisis.
Andrew Smithers
[1] Robert Solow termed these ‘embodied’ and ‘disembodied’[1] though the terms most used today are ‘hard-baked clay’ and ‘putty-putty’. Substitution and Fixed Proportions in The Theory of Capital by Robert M. Solow (1962) Review of Economic Studies Vol. 29 No. 3.
[2] How Share Prices Fluctuate by Andrew Smithers (2024) World Economics Vol. 25 No. 4.
[3] See Figure 1 The Economics of The Stock Market by Andrew Smithers (2022) Oxford University Press.
[4] NDP, which is GDP net of depreciation.
[5] See Figure 6 How Share Prices Fluctuate op. cit.
[6] The Unaccountability Machine by Dan Davies (2024) Profile Books.
[7] In equilibrium the stock market value of quoted companies equals their net worth as set out in A General Equilibeium Approach to Monetary Theory by James Tobin (1969) Journal of Money, Credit, and Banking. Vol. 1 No. 1. The Federal Reserve’s Z1 Table B. 103 shows this ratio at the end of 2025 as 2.47.
