The Permanent Income Hypothesis for Normal Humans
Milton Friedman's 1957 insight that people spend based on what they expect to earn over a lifetime, not what they earned last month, reshaped how economists think about consumption, saving, and fiscal policy.
The Puzzle
In 1936, John Maynard Keynes proposed one of the simplest and most powerful ideas in economics: people spend a stable fraction of their income. Earn more, spend more. Earn less, spend less. This relationship — the “consumption function” — was the engine of the Keynesian revolution. It gave governments a lever. If consumer spending fell during a recession, the government could put money in people’s pockets through tax cuts or transfer payments, and they would spend most of it, generating a multiplier effect that would pull the economy back to health.
The idea was intuitively appealing, and early data seemed to support it. Cross-sectional surveys — snapshots comparing rich and poor households at a single point in time — showed that higher-income families consumed more but saved a larger fraction of their income. This was consistent with Keynes’s “fundamental psychological law” that the marginal propensity to consume declines as income rises.
But there was a problem. When economists looked at long-run time series data — the historical record of income and consumption over decades — they found something different. Simon Kuznets, in work that would eventually contribute to his Nobel Prize, showed that the average saving rate in the United States had remained roughly constant over many decades, even as incomes had risen enormously. If Keynes was right that richer people save a higher fraction of their income, then the national saving rate should have risen steadily over time. It had not.
This was the consumption puzzle, and it demanded an explanation.
Friedman’s Answer
Milton Friedman provided one in his 1957 book “A Theory of the Consumption Function.” His insight was deceptively simple: people do not base their spending on this month’s paycheck. They base it on what they expect to earn, on average, over their lifetime. Friedman called this expected long-run average “permanent income.”
The distinction matters because actual income in any given period is the sum of two components. Permanent income is the stable, predictable part — the income you expect to receive year after year, based on your education, skills, career trajectory, and wealth. Transitory income is everything else — the bonus you did not expect, the overtime pay that will not last, the inheritance from a distant relative, or on the downside, the month you were sick and earned less than usual.
Friedman’s hypothesis was that consumption depends almost entirely on permanent income. When your permanent income rises — you get a promotion, you complete a professional degree, you inherit a steady-income asset — you increase your spending proportionally. But when your income rises temporarily — you win a small lottery, you get a one-time tax rebate, you work extra shifts during the holiday season — you save most of the windfall rather than spending it. You know the extra money will not last, so you smooth it over time.
This single idea resolved the consumption puzzle. Cross-sectional data showed high-income people saving more because some of their high income was transitory — lucky breaks, windfall years — and people save transitory income. Low-income people appeared to save less because some of their low income was also transitory — bad years, temporary setbacks — and they maintained spending by drawing on savings. The long-run time series showed a constant saving rate because, over decades, transitory fluctuations wash out and what remains is permanent income, against which people consume at a stable rate.
How People Actually Smooth
The mechanics of consumption smoothing are worth spelling out, because they reveal how radical Friedman’s vision of the consumer really was.
Consider a 25-year-old who has just graduated from medical school. She is entering her residency, earning perhaps $60,000 a year. But she expects to earn $250,000 or more within a decade. According to the permanent income hypothesis, her consumption should already be significantly higher than $60,000 — she should borrow against her expected future income to maintain a standard of living consistent with her lifetime earnings. And indeed, this is what we observe: young doctors carry significant debt and spend beyond their current income.
Now consider a 60-year-old executive who has just been told his division will be eliminated in two years. His current salary is $200,000, but his permanent income has just dropped sharply. The hypothesis predicts he will immediately cut spending, even though his paycheck has not changed yet. He is forward-looking, and his consumption tracks his revised expectations, not his current bank balance.
This forward-looking behavior is the heart of the theory. Consumers in Friedman’s framework are not passive responders to current conditions. They are planners — imperfect and uncertain, perhaps, but fundamentally oriented toward the future. They use savings, borrowing, and assets to buffer their consumption against the ups and downs of actual income.
The Life-Cycle Hypothesis
Friedman was not the only economist working on this problem. Franco Modigliani, working with Richard Brumberg in the early 1950s and later with Albert Ando, developed a closely related idea called the life-cycle hypothesis. Where Friedman emphasized the distinction between permanent and transitory income, Modigliani emphasized the arc of a lifetime: people earn little when young, more during their working years, and nothing (from labor) after retirement. Rational consumers would borrow when young, save during their peak earning years, and draw down savings in retirement, keeping consumption relatively stable throughout.
The two theories are more complementary than competing. Both predict consumption smoothing. Both predict that temporary income changes will have small effects on spending. Both challenge the simple Keynesian consumption function. The main difference is in emphasis: Friedman focused on the stochastic nature of income fluctuations, while Modigliani focused on the predictable demographic pattern. Together, they reshaped the economics of consumption.
Modigliani received the Nobel Prize in 1985, in part for this work. Friedman had already received his in 1976, though the committee cited his contributions to monetary theory and consumption analysis jointly.
The Fiscal Policy Bombshell
The permanent income hypothesis had an explosive implication for fiscal policy, and Friedman was fully aware of it.
If consumers spend based on permanent income, then a temporary tax cut or a one-time rebate check will have little effect on spending. People will recognize the windfall as transitory, save most of it, and continue consuming at their usual rate. The Keynesian multiplier — the whole mechanism by which deficit spending was supposed to stimulate the economy — would be dramatically smaller than Keynesian models predicted.
This was not a minor quibble. It struck at the heart of the policy framework that had governed Western economies since the 1940s. If temporary fiscal stimulus does not significantly boost spending, then governments cannot reliably manage the business cycle through demand-side policy. The case for fiscal activism weakens enormously.
Friedman pressed this point repeatedly. When the U.S. government implemented a temporary income tax surcharge in 1968 to cool an overheating economy, Friedman predicted it would have little effect on consumption. He was largely right — consumers mostly maintained their spending and reduced saving, treating the surcharge as transitory. The episode became a textbook illustration of the permanent income hypothesis in action.
The same logic applies in reverse. The one-time tax rebates that have become a staple of U.S. fiscal policy — in 2001, 2008, 2020, and 2021 — have always generated debate about their effectiveness. If the permanent income hypothesis holds, most of the rebate should be saved rather than spent. Empirical studies have generally found that consumers do spend a significant fraction of rebates — perhaps 20 to 40 percent within the first quarter — but considerably less than the Keynesian model would predict. The permanent income hypothesis does not perfectly describe behavior, but it captures something real about how people respond to windfalls.
When the Theory Breaks
No theory in economics is universally true, and the permanent income hypothesis is no exception. Its most important failures reveal as much about the economy as its successes.
Liquidity constraints are the most obvious problem. Friedman’s consumer is free to borrow and lend at reasonable rates, smoothing consumption across good times and bad. But many households — particularly young people, low-income families, and those with poor credit histories — cannot borrow on reasonable terms or at all. A household that is already at its credit limit cannot smooth consumption in response to a negative income shock; it has no choice but to cut spending immediately. For these households, current income determines current consumption, almost exactly as Keynes described.
The empirical importance of liquidity constraints is substantial. Research by Christopher Carroll, Jonathan Parker, and others has shown that a large fraction of the population — estimates range from 25 to 50 percent — behaves in ways more consistent with the Keynesian consumption function than the permanent income hypothesis. These “hand-to-mouth” consumers spend most of any income increase, whether permanent or transitory, because they have little savings and limited access to credit.
Uncertainty and precautionary saving also complicate the picture. Friedman’s framework assumes that consumers can form reliable expectations about their permanent income. But when the future is deeply uncertain — during a pandemic, a financial crisis, or a period of structural economic change — people may save more than the basic theory predicts, not because they are smoothing against transitory shocks but because they are building a buffer against an unknowable future. This precautionary motive, formalized by Christopher Carroll and others, generates behavior that looks quite different from the smooth optimization of the permanent income hypothesis.
Behavioral anomalies further erode the theory’s universal applicability. People exhibit mental accounting — they treat money differently depending on its source, spending a tax refund more freely than an equivalent amount of regular income. They are present-biased, valuing immediate consumption more than future consumption in ways that standard discounting cannot explain. They anchor on current income in ways that Friedman’s forward-looking consumer would not. The entire field of behavioral economics, pioneered by Richard Thaler, Daniel Kahneman, and others, is in some sense a catalog of deviations from the permanent income hypothesis.
Habit formation poses another challenge. If people develop consumption habits that are costly to break, then spending will be stickier than the permanent income hypothesis predicts. A household whose income drops permanently may continue spending at the old rate for months or years, drawing down savings and accumulating debt, because cutting back is psychologically painful. Conversely, when income rises permanently, spending may adjust slowly upward as new habits form. This “excess smoothness” — consumption that is smoother than even the permanent income hypothesis predicts — has been documented empirically and has spawned its own literature.
The Empirical Record
Testing the permanent income hypothesis has occupied economists for decades, and the results are genuinely mixed.
Robert Hall’s 1978 paper made a sharp prediction: if the hypothesis holds and consumers have rational expectations, then changes in consumption should be unpredictable. Consumption should follow a “random walk” — the best forecast of next period’s consumption is this period’s consumption, because any predictable change would have already been incorporated into spending plans. Hall found partial support for this prediction in aggregate U.S. data, but subsequent research identified systematic deviations. Consumption appears to respond to predictable income changes — paydays, Social Security payments, tax refunds — in ways that should not occur if people are truly forward-looking.
The natural experiments provided by tax rebates have been particularly informative. Studies of the 2001 and 2008 U.S. rebates by David Johnson, Jonathan Parker, and Nicholas Souleles found that households spent 20 to 40 percent of the rebates within a few months — far more than the permanent income hypothesis would predict for a clearly temporary windfall, but far less than the near-100 percent that a naive Keynesian model would suggest. The truth, as so often in economics, lies between the polar cases.
More recently, research using high-frequency transaction data — made possible by financial technology and the availability of anonymized bank records — has confirmed that many households are highly responsive to income timing. People spend more in the days after receiving a paycheck, even though the paycheck was entirely predictable. This is difficult to reconcile with the permanent income hypothesis, which would predict smooth spending across the pay cycle.
A Theory That Changed Everything Anyway
The permanent income hypothesis may not be literally true for all people at all times. But that is not the right standard for evaluating a theory. The relevant question is whether it illuminated something important about economic behavior and changed how we think about policy. By that standard, it was a triumph.
Before Friedman, the default assumption in macroeconomics was that consumption tracked current income. After Friedman, the default shifted to forward-looking, optimizing consumers who smooth over time. This change in baseline assumption — even when modified by liquidity constraints, precautionary motives, and behavioral frictions — fundamentally altered how economists evaluate fiscal policy, design pension systems, think about household saving, and model business cycles.
The hypothesis also forced a productive research agenda. The deviations from the theory — hand-to-mouth consumers, mental accounting, precautionary saving, excess sensitivity to predictable income changes — became research programs in their own right. Much of what we know about household finance and behavioral economics was discovered by economists trying to explain why the permanent income hypothesis did not work as cleanly as Friedman predicted.
In that sense, the theory succeeded even in its failures. It set the terms of the debate. It told economists what the benchmark should be, and every deviation from that benchmark revealed something new about how people actually make financial decisions. Nearly seventy years after its publication, it remains the starting point — the first chapter, not the last — for understanding why people spend what they spend.
Friedman, who never lacked confidence, would probably have said: of course. The theory was not meant to describe every individual. It was meant to describe the tendencies that dominate in the aggregate and over time. And in that, it does its job remarkably well. The bathtub of consumption is smoother than the faucet of income. That much, no one disputes.