For almost six decades the US labour share hardly moved. Measured as wages’ share of net national income at factor cost and adjusted for the business cycle using my own estimates, it averaged 78.4% between 1963 and 2019, and it never left a band from 75.9% to 81.4%.

Then it broke. In the second quarter of 2026 the adjusted labour share stood at 72.0%, almost 4 percentage points below the lowest reading in the 57 years before the pandemic.

The adjusted profit share has gone the other way, from 12.4% at the end of 2019 to 18.8%, which is above anything recorded between 1963 and 2019. Corporate profits after tax are up 76% since the fourth quarter of 2019, while nominal GDP is up 48%.

Some readers might be surprised that I am writing about a mostly forgotten Polish Marxist economist, Michał Kalecki (1899-1970).

But his insights about profits are very important for understanding what is happening in the US economy right now. We just need to understand why they suddenly matter, after decades in which they did not.

Today’s (few) post-Keynesians already reach for Kalecki and his famous profit equation: government deficits create profits, and the US has been running deficits that would have been unthinkable in peacetime only a decade ago. Others conclude that the labour market is broken, or that profits are simply eating wages.

I think both miss the point. In my view the explanation is the combination of a series of policy shocks – fiscal, tariffs and war – and a change in the Federal Reserve’s monetary policy rule.

And the irony is that the Fed has made Kalecki and post-Keynesian economics relevant again, while it could make both irrelevant the day it decides to do its job again.

To show this I return to a framework I introduced on this blog more than a decade ago: the IS/LM+ model, now with Kalecki added.

An identity is not a theory

Kalecki’s profit equation follows directly from the national accounts.

He first wrote it down in Polish in 1933, developed it in “A Theory of Profits” in the Economic Journal in 1942, and set it out most clearly in “The Determinants of Profits”, chapter 3 of Theory of Economic Dynamics (1954).

In a closed economy, profits after tax equal investment, plus capitalists’ consumption, plus the government deficit, minus the saving of wage earners:

Here Π is gross profits, T_π taxes on profits, I investment, C_π capitalists’ consumption, G government spending, T tax revenue and S_w the saving of wage earners.

The equation is always true – it is a definition, not a description of economic behaviour. And that is exactly the problem.

An identity holds after the fact whatever happens, so it cannot by itself tell us what causes what.

Kalecki read it from right to left: capitalists’ spending decisions and the deficit are the cause, and profits are the result. Workers spend what they earn, capitalists earn what they spend.

But nothing in the identity says that the right-hand side moves first. If the government runs a larger deficit, profits can rise. Or investment can fall, or households can save more, so that profits do not move at all.

Which of the terms gives way depends on what happens to nominal spending, and in a modern market economy that is decided by the central bank.

This is the key point. Kalecki’s identity always holds. But it is monetary policy that closes the model, and therefore it is also monetary policy that determines the causality.

Kalecki wrote in the 1930s, with plenty of idle capacity at the onset of the Great Depression and a gold standard in most economies.

In that world his reading made a lot of sense. In a world where the central bank targets nominal spending, it makes none. So the real question is under which monetary policy rule Kalecki’s equation becomes a theory of profits.

IS/LM+ with Kalecki inside

In the traditional IS/LM model the money supply is simply assumed to be constant. That is a monetary policy rule too, but it is not a rule any modern central bank follows.

So in September 2012, in “The fiscal cliff and the Bernanke-Evans rule in a simple static IS/LM model”, I replaced the constant money supply with a monetary policy reaction function.

A few months later, in “Daniel Lin is teaching macro! Let’s introduce his students to the IS/LM+ model”, I gave the extended model its name and replaced the Fed’s unemployment rule with a rule for nominal GDP.

The plus is the monetary policy rule. And that one extra equation changes everything: the slope of the LM curve no longer depends only on the interest rate elasticity of money demand, but on how aggressively the central bank reacts.

The same framework can carry Kalecki.

I keep everything as Keynesian (and Kaleckian) as possible: fixed prices, a fixed nominal wage, workers who spend most of what they earn, capitalists who consume a fixed amount plus a share of their profits, and investment that responds to both the interest rate and profits.

The government sets the budget balance. In deviations from the starting point, measured in dollars, with n as nominal spending (nominal GDP) and b as the deficit, the IS curve becomes:

where a is capitalists’ autonomous spending, r is the interest rate, β measures how strongly spending responds to the interest rate, and the Kalecki multiplier μK is larger the more of their profits capitalists spend again and the less workers save. That is Kalecki’s own mechanism.

The full model, including the AS/AD+ extension, is set out in the appendix.

The monetary side is the LM+ curve from the 2013 post. Money demand is m = n − αr, where m is the money supply and α is the interest rate sensitivity of money demand. The central bank follows the rule m = n* − λ(n − n*), where n* is its target for nominal spending and λ measures how strongly it reacts to deviations from that target:

Solve the two, and the effect of a larger deficit on profits, Π, is:

where ω is the wage share of income, so 1 − ω is the profit share.

Everything turns on λ. If the central bank is fully committed to its nominal GDP target (λ very large), the LM+ curve is vertical, and a deficit only raises the interest rate.

Investment is crowded out one for one, and profits do not move. That is the so-called Sumner Critique, applied to Kalecki.

If instead the central bank pegs the interest rate and lets the money supply follow demand (λ = −1), the LM+ curve is horizontal, and the full Kalecki multiplier comes through.

If workers spend what they earn, every dollar of deficit becomes at least a dollar of profits.

Said in another way, Kalecki’s profit equation is a general theory of profits in exactly the same sense as Keynes’ General Theory is a general theory of employment.

As I argued in January 2014 in “No ‘General Theory’ should ignore the monetary policy rule”, Keynes quietly assumed a gold standard-like regime with a fixed money base.

Kalecki quietly assumes a liquidity trap, or at least a central bank that behaves as if it is in one. In IS/LM+ terms that is the same as an interest rate peg: the interest rate does not move when spending rises. Under another rule, the results disappear.

Why the labour share did not move for decades

The distribution of income between wages and profits only shifts when prices move unexpectedly relative to nominal wages.

Over time nominal wages follow nominal spending. What wage earners cannot protect themselves against is a jump in the price level that nobody had priced into their contracts.

From the early 1990s to 2019 the Fed made sure such jumps did not happen. It never announced a nominal GDP target, but it behaved as if it had one. Between 2010 and 2019 US nominal GDP never deviated by more than about 1% from a 4% growth path.

The Fed reacted immediately when nominal spending growth picked up. I have estimated a simple reaction function for the Fed, regressing the change in the fed funds rate over four quarters on nominal GDP growth. From 1995 to 2019 the fed funds rate rose by 0.40 percentage points over a year for every percentage point of extra nominal GDP growth.

Under that rule, fiscal policy did what IS/LM+ says it should do. From fiscal year 2014 to 2019 the structural federal deficit, which I calculate by adjusting the federal balance for the output gap, rose from 1.5% to 4.8% of GDP, and the profit share fell.

In a regression on annual data for 1990-2019, a structural deficit 1 percentage point of GDP larger went with private investment 0.3 percentage points lower and household saving 0.25 percentage points higher, controlling for the output gap. Kalecki’s equation held every year, but it was the other terms that gave way.

There is one exception, and it proves the rule. When the fed funds rate hit zero in December 2008, the Fed behaved as if it was stuck. It was not.

The Fed has one instrument, the money base, and the interest rate is just one way of describing how it is used. At zero that description stops working, but the money base can always be expanded. For a while the Fed acted as if nothing more could be done. In IS/LM+ terms that is a de facto interest rate peg.

And Kalecki showed up right on cue. The structural deficit rose by 3.6% of GDP in fiscal year 2009, and the year after, the cyclically adjusted profit share rose 2.7 percentage points, the largest increase in any fiscal year from 2010 to 2025.

Then the Fed found its feet again with QE2 in November 2010, and with QE3 and the Evans rule in 2012, nominal GDP was back on a steady 4% path. The link disappeared. The structural deficit was cut by almost 5% of GDP from fiscal year 2010 to 2014, the fiscal cliff included, and the profit share did not fall. The Fed offset it.

So there was nothing mysterious about the stable labour share. It was the result of a monetary policy rule that kept nominal spending on a stable path, so that the shocks hitting the economy did not show up as surprises in the price level.

2021: the Fed suspended the rule

In 2021 the Fed stopped behaving as if it had a rule. Nominal GDP grew 17.5% year on year in the second quarter of 2021 and stayed above 10% for the next three quarters. The fed funds rate stayed at zero. Had the Fed reacted as it did from 1995 to 2019, the policy rate would already have been around 2.4-3.1 percentage points higher in the third and fourth quarters of 2021, when nominal GDP was above its pre-pandemic level. The first hike came in March 2022.

In IS/LM+ terms the Fed went from a nearly vertical LM+ curve to a horizontal one. It was a de facto interest rate peg, and it came in fiscal year 2021, when the structural deficit was 11.6% of GDP.

This is the one situation where Kalecki is right. With the interest rate fixed, the deficit was not offset by higher rates, lower investment or higher saving. It went straight into nominal spending, and from there into profits. The cyclically adjusted profit share rose 1.9 percentage points in fiscal year 2021, a year after the largest structural fiscal easing since the Second World War. Take a look at the chart below: 2021 and 2010 are the two years that stand out, and in both the Fed did not offset fiscal policy.

When the Fed returned, it returned with a different rule. Estimated on data since 2020, the same reaction function shows the fed funds rate rising by only 0.17 percentage points for every percentage point of extra nominal GDP growth, less than half the pre-2020 response, and the reaction comes around five quarters late rather than immediately.

And it is a growth rule, not a level rule. Nothing has been done to bring nominal spending back to its old path, so the 2021 jump has become permanent. Nominal GDP is now 15% above the 4% path from the fourth quarter of 2019.

Tariffs, war and an accommodating Fed

The 2021 deficit explains the first jump in profits, but not why the profit share has kept rising. The structural federal deficit has narrowed from 7.7% of GDP in the third quarter of 2024 to 6.1% in the second quarter of 2026, and still profit growth has accelerated from around 4% in 2024 to 14% in the first half of 2026. So something other than fiscal policy is at work.

Two caveats.

First, 6.1% of GDP is still a very large deficit for an economy with no slack, and in Kalecki’s accounts it is the level of the deficit, not only the change, that adds to profits.

Second, some of the narrowing reflects higher tariff revenue, so part of the fiscal tightening is the supply shock itself showing up in the budget.

Neither changes the conclusion. A deficit that is high but shrinking cannot explain why profit growth has accelerated.

The answer is a series of negative supply shocks. The effective US tariff rate rose from around 2% in early 2025 to between 6.5% and 8.8% over the following year. Energy prices jumped again in the second quarter of 2026 with the war in Iran.

Take a look at what happened to nominal GDP in that quarter: nominal growth rose to 6.3% year on year, but real growth slowed to 2.2% while GDP deflator inflation rose to 4.0%. Faster inflation and slower real growth is the signature of a supply shock.

To analyse this I need the AS/AD version of the framework from the 2014 post, with one change. Instead of completely fixed prices, firms pass part of any change in nominal spending and any cost shock on to prices, while nominal wages adjust slowly.

Call it AS/AD+. With sticky nominal wages and productivity held constant, the profit share rises whenever the price level rises faster than wages, and the labour share falls by the same amount.

Now consider a negative supply shock under two monetary policy rules. Under an NGDP target, nominal spending stays put.

Prices rise and output falls, so the cost of the shock is shared between lower real wages and lower activity.

If the central bank instead accommodates the shock by letting nominal spending rise, output is protected, but prices rise further.

With nominal wages lagging, the real wage and the labour share fall by more, and profits rise by more.

That is what the Fed has done. It cut the fed funds rate from 5.33% in mid-2024 to 3.63% while the GDP deflator accelerated and nominal GDP was already far above its old path, and only in September 2026 did it raise rates again, by a quarter of a percentage point. In the language of IS/LM+, λ has at times been below −1: the Fed has eased into a nominal spending boom rather than leaning against it.

And because the Fed now runs a growth rule rather than a level rule, every accommodated shock lifts the price level for good.

The labour share recovers only if wages catch up before the next shock hits. With an energy shock in 2022 after Putin’s invasion of Ukraine, tariffs in 2025 and another energy shock in 2026, wages never got the chance.

Each shock has ratcheted the labour share lower.

Where the money went

In the second quarter of 2026 nominal GDP was 15.1% above the 4% path from the fourth quarter of 2019, or 14.0 log points. Of those, 11.5 log points reflect a higher price level than a 2% path would have given, and 2.5 reflect higher real GDP.

The output gap, measured against the Congressional Budget Office’s (CBO’s) estimate of potential output, is essentially where it was in 2019.

The adjusted labour share has a statistical break in the third quarter of 2021 (Bai-Perron, p = 0.02).

That is the quarter in which the Fed kept rates at zero while nominal GDP grew at a double-digit pace.

Corporate profits follow nominal spending rather than the deficit.

Since 2010 the deviation of profits from their pre-pandemic trend has a correlation of 0.91 with the deviation of nominal GDP from its 4% path, and profits move with almost twice the amplitude.

Households have absorbed the squeeze by saving less. Personal saving fell from 5.0% of GDP in the second quarter of 2024 to 3.2% two years later, as real hourly wages stopped growing.

In Kalecki’s accounts that shows up as a contribution to profits, but it is not an independent impulse.

It is the same redistribution recorded twice: once as lower real wages, and once as lower saving by wage earners who try to keep consuming as their purchasing power erodes.

These are simple estimates on short samples, especially after 2020, and the difference between the two estimated reaction functions is only borderline significant (p = 0.07). So I would not put much weight on any single coefficient. But the pattern is consistent with the model.

The Fed made Kalecki relevant

For decades post-Keynesian economists have told us that deficits create profits, and for decades the data refused to cooperate, because the Fed offset fiscal policy shocks. Then in 2021 the Fed de facto pegged the interest rate, and suddenly Kalecki’s equation worked exactly as advertised.

So the post-Keynesians are right about the last five years, but for completely the wrong reason.

Kalecki did not discover a law of capitalism. He described how profits behave when the central bank lets nominal spending float with fiscal policy, and that is a choice the Fed made, not a feature of the US economy.

The same goes for those who conclude that the labour market is broken or that profits are eating wages. That is nonsense. Nothing structural happened to the US labour market in 2021.

What happened was that the Fed stopped guaranteeing a stable nominal anchor. Wage earners had signed contracts on the assumption that it would, and they have paid for that with a falling share of national income ever since.

In my view it is pretty damn ironic that the institution most post-Keynesians regard as irrelevant to income distribution is the one that has made their favourite equation hold. And it can make that equation irrelevant again.

Under an NGDP level target, the deficit would not show up in profits, a supply shock would be shared between prices and output rather than dumped on wage earners, and the 2021 overshoot would have been reversed rather than locked in. An inflation target that looks through supply shocks would deliver the first two. Only a level target delivers the third.

In January 2014 I ended my post on Keynes’ General Theory with “It is the monetary policy regime, stupid!”. Twelve years later the same is true of Kalecki. It is not the deficit that determines profits. It is the Fed.

References

Kalecki, M. (1933), Próba teorii koniunktury (An Essay on the Theory of the Business Cycle), Warsaw: Instytut Badania Koniunktur Gospodarczych i Cen.

Kalecki, M. (1939), Essays in the Theory of Economic Fluctuations, London: George Allen & Unwin.

Kalecki, M. (1942), “A Theory of Profits”, Economic Journal, 52(206/207), pp. 258-267.

Kalecki, M. (1954), Theory of Economic Dynamics: An Essay on Cyclical and Long-Run Changes in Capitalist Economy, London: George Allen & Unwin. Chapter 3, “The Determinants of Profits”.

Kalecki, M. (1971), Selected Essays on the Dynamics of the Capitalist Economy 1933-1970, Cambridge: Cambridge University Press. Includes “The Determinants of Profits”.

Christensen, L. (2012), “The fiscal cliff and the Bernanke-Evans rule in a simple static IS/LM model”, The Market Monetarist, September 2012.

Christensen, L. (2013), “Daniel Lin is teaching macro! Let’s introduce his students to the IS/LM+ model”, The Market Monetarist, January 2013.

Christensen, L. (2014), “No ‘General Theory’ should ignore the monetary policy rule”, The Market Monetarist, January 2014.

Appendix: the full model

The model is deliberately as Keynesian as possible. The only thing that is not Keynesian is that there is a central bank with a monetary policy rule. The economy is closed, and all variables are nominal.

A1. Income, spending and Kalecki’s identity

Nominal income Y is split into wages W and gross profits Π, with a wage share ω that is fixed in the short run because nominal wages and prices are sticky:

Taxes are a share t of income, T = tY, and the government sets the deficit B = G − T. Workers save a share s_w of their disposable income. Capitalists consume a fixed amount plus a share q of their profits after tax, and investment depends on the interest rate and on profits after tax:

Here C_w is workers’ consumption, C_π capitalists’ consumption with fixed part C̄_π, I investment with autonomous part Ī, r the interest rate, β_I the interest rate sensitivity of investment and γ the share of profits after tax that is reinvested.

The national accounts identity Y = C_w + C_π + I + G can be rewritten as Kalecki’s profit equation, which holds for any values of the parameters:

where T_π is taxes on profits and S_w the saving of wage earners.

A2. The IS curve with Kalecki inside

Inserting the behavioural equations into the identity and solving for Y gives the IS curve:

μK is the Kalecki multiplier. It is larger the more of their profits capitalists spend on consumption and investment (q + γ), and the less workers save (s_w). The model requires q + γ < 1, so that capitalists spend less than a dollar out of each extra dollar of profits after tax.

In deviations from the starting point, measured in dollars, with n for nominal spending, a for capitalists’ autonomous spending, b for the deficit and β ≡ μK β_I, this is the IS curve used in the text:

Because the wage share is fixed in the short run, profits change by a share 1 − ω of any change in nominal spending, ΔΠ = (1 − ω)Δn. In logs, profits and nominal spending move one for one.

A3. The LM+ curve

Money demand is m = n − αr, where m is the money supply and α is the interest rate sensitivity of money demand. The central bank follows the rule from the IS/LM+ posts, with the money supply reacting to the deviation of nominal spending from the target n*, and λ measuring how strongly it reacts:

Setting money supply equal to money demand gives the LM+ curve:

The traditional IS/LM model is the special case λ = 0, a constant money supply. λ → ∞ is a strict NGDP target and gives a vertical LM+ curve. λ = −1 is an interest rate peg and gives a horizontal LM+ curve. λ < −1 means that the central bank eases when nominal spending rises, and the LM+ curve slopes downwards.

A4. Fiscal policy and profits under different rules

Solving IS and LM+ together, with n* normalised to zero, gives:

Under a strict NGDP target the effect is zero. Higher deficits are offset one for one by lower investment through the interest rate. Workers’ saving depends only on their income, so it does not move, and Kalecki’s identity holds with investment giving way. Under an interest rate peg the full Kalecki multiplier comes through. If workers do not save, profits after tax then rise by 1/(1 − q − γ) per dollar of deficit, which is at least one. With λ < −1 the effect is larger still, as long as the denominator stays positive, which requires −1 − α/β < λ < −1 for α, β > 0. Below that range the model has no stable solution.

Two further features of the rule matter for the US after 2020. First, timing. If the central bank reacts with a lag of k periods, r_t = φ n_{t−k}, where φ is the strength of the reaction, the economy behaves as under an interest rate peg for k periods, and a fiscal shock passes fully into profits before the rule bites.

Second, growth versus level. Under a growth rule the central bank only stabilises the growth rate of nominal spending, so the level shift from a fiscal or supply shock is never reversed. Under a level rule, nominal spending is brought back to the old path.

A5. AS/AD+ and the labour share

To analyse supply shocks, prices are allowed to move. In this section, n denotes the log deviation of nominal spending, rather than the dollar deviation used above. Firms pass a share κ of any change in nominal spending and all of a cost shock ε on to prices, while nominal wages w̄ stay fixed in the short run. With p the price level and y real output, both in logs:

The labour share is the real wage relative to productivity a, so with sticky nominal wages and productivity held constant, it falls one for one with the price level in log terms:

If the central bank accommodates a cost shock by letting nominal spending rise, n = θε, where θ measures how much of the shock it accommodates, then:

Under an NGDP target (θ = 0) the labour share and output both fall by ε in log terms. With accommodation (θ > 0) output falls by less, but the labour share falls by more, and the profit share rises by more. The labour share only recovers as nominal wages catch up with prices. Under a growth rule the price level never returns, so if a new shock hits before wages have caught up, the labour share is ratcheted lower with each shock.