The economic principle that explains why a stimulus makes good sense.

Few people have heard of the “paradox of thrift,” a very old economic theory that was revived, and dusted off, and contextualized, by John Maynard Keynes and has lately reappeared as a shuttlecock in the battle between Left and Right over stimulus spending and other responses to the economic crisis. Liberal economists have invoked the paradox to support the argument that large-scale government spending is the key to economic recovery.

The premise of the paradox of thrift is that when people save in a way that removes money from the economy (rather than redirecting it to productive investment), consumption falls without an offsetting increase in investment. The fall in consumption and resulting fall in production may reduce incomes to a level at which total savings decline, even though everyone is trying to save more. That is the paradox and it illustrates what is called the “fallacy of composition”: what is good for the individual may be bad for society as a whole.


The opposite of waste and prodigality, thrift provides protection against adversity. It enables one, if one’s income falls, to reallocate savings to consumption in order to preserve one’s standard of living, or at least avoid overindebtedness and eventual ruin. But in good times, when people should save, they like to spend; it is in bad times that they like to save. People today are overindebted and worried about further economic downturns, and so are spending less of their income; the rate of personal savings has tripled in the past three years. Yet thrift in bad times can be—the paradox of thrift shows—self-defeating.

To illustrate, suppose that average income is $100,000 and everyone increases the percentage of his income that he saves from 5 percent to 6 percent—a 20 percent increase. Aggregate demand for goods and services will fall because people will be consuming less if they are saving more, and therefore production and therefore incomes will fall. Suppose incomes fall by 25 percent, bringing average income down to $75,000. Then the dollar amount that each person saves will fall, because 6 percent of $75,000 is less than 5 percent of $100,000 (the income before the increase in the savings rate). Savings will be only $4,500 (the result of the higher savings rate applied to a lower income), versus $5,000 (the result of the lower savings rate applied to a higher income).

Stated this summarily, the paradox of thrift can be criticized for ignoring the role that saving can play in increasing consumption by increasing investment. If increased savings are transformed into lendable funds, and then lent, interest rates will fall because there is an increased supply of money for lending. The lower rates will stimulate borrowing by both consumers and producers. Much of the money borrowed by consumers will be spent on goods and services rather than saved, and so consumption spending may increase even though people are saving more. And lower interest rates at which producers can borrow will reduce their costs (and hence their prices, which will stimulate consumption) and also encourage them to make capital investments, which will increase employment and therefore incomes

This happy picture assumes, however, that increased savings are channeled into loans to consumers and producers, or other forms of investment that stimulate consumption or production. They needn’t be, and, if they’re not, the increase in savings will not push down interest rates to consumers and entrepreneurs and thus stimulate consumption and capital investment. And then the paradox of thrift is likely to be observed—total savings will fall.

The reason the paradox was discovered early in Western economic history is that until relatively recently savings were more likely to take the form of hoarded wealth (stocks of gold, for example) than of lendable funds. Lending was inhibited by usury laws, the rudimentary state of financial institutions, and difficulties in collecting debts. So increased saving could reduce consumption and hence income and hence savings, in accordance with the paradox.

What Keynes recognized is that conditions that limit the conversion of savings into lendable funds can arise in a modern economy and give rise to the paradox of thrift. All that’s required is a profound financial crisis such as the nation and the world experienced in 2008, against a background of private overindebtedness.

The bursting of the housing bubble, which brought down the banking industry because banks were so heavily invested in financing residential real estate, had three effects that relate to the paradox of thrift, each amplified by the accompanying implosion of the stock market bubble. First, the decline in house and stock values reduced household wealth without reducing people’s debt burdens, because debt is a fixed expense rather than a percentage of the value of the assets that secure it. (That’s why so many mortgages are “under water”: the unpaid balance of the mortgage exceeds the market value of the property that secures it because the value has fallen but not the debt.) So people felt poorer and therefore more vulnerable to economic adversity, and they reacted by reducing their consumption, thus saving more.

Second, the financial crisis reduced the ability and willingness of banks to lend money. Lending is highly risky when prospective borrowers are already overindebted. The banks’ solvency was impaired, moreover, and this increased the likelihood that risky lending could bankrupt them and also made the banking regulators insist on extremely conservative bank lending practices. And with people feeling overindebted and wanting to build up their savings rather than go deeper into debt, the demand for loans declined.

Third, with consumption and credit transactions stagnant or falling, production shrank, triggering a downward spiral of sales, output, employment, and prices that made people all the more reluctant either to spend on consumption items or make risky investments with their savings.

In these circumstances, increased savings were unlikely to be channeled into lendable funds unless the federal government was the borrower and the risk of default therefore negligible. With banks reluctant to lend and consumers and businesses averse to taking risks in a depressed and uncertain economic environment, money deposited in banks tended either to stay in the banks, as excess cash reserves, or to be used to buy Treasury securities—which is to say, to be lent to the U.S. Treasury. As a result, though people were saving more of their income, private investment plummeted in 2009 and was actually negative when depreciation is subtracted. Yet the amount of currency increased, partly because of foreign demand, but partly also because Americans were placing more of their savings in safe-deposit boxes or even in home safes. Currency that is shut away in a bank vault or a safe is an inert form of savings; it contributes nothing to economic activity.

Because demand for loans was weak and the Federal Reserve was flooding the economy with money in an unsuccessful effort to increase the amount of lending, interest rates were very low. That reduced the income generated by savings, and it might seem that the lower the return to savings, the less people would save. But people anxious about their economic prospects may stick with safe savings even as interest rates fall, and indeed may increase the amount they save if they are dependent on the income from their savings: greater savings may offset the loss of income on one’s savings that is caused by the lower interest rate. And when interest rates on bank deposits or other safe financial assets are very low, cash in a safe or a safe-deposit box (or under one’s mattress) becomes a less unattractive alternative; that is one reason the amount of cash held by consumers and by business firms has increased during the economic crisis.

Another reason is anticipation of deflation. If prices are expected to fall, and thus the purchasing power of money to increase (the same amount of money will buy more goods and services if prices fall), the hoarding of cash is encouraged, because the real value of cash is expected to rise. And, unlike interest, an increase in purchasing power is not taxable. Moreover, cash is safer than the expectation of repayment of a loan (other than a loan to the federal government).

Even without expectations of deflation, uncertainty about the economic outlook tends to freeze spending, including lending, because hoarding is a common, and incidentally a thoroughly rational, response to uncertainty. Building cash reserves increase one’s ability to survive a future economic shock and meanwhile gains time (by avoiding commitments) for the uncertainty to dissipate.

The increase in savings since the beginning of the crisis has not reduced consumption, and therefore production and incomes, so drastically as to cause aggregate savings to decline—which would be required for the paradox of thrift to hold. But it has reduced them enough to retard economic recovery and thus strengthen the economic argument for deficit spending—what is now called “stimulus”—as a means of speeding economic recovery from a recession or depression.

The argument for stimulus begins with the observation that when people are fearful about their economic prospects, they may be willing to lend their savings only to the federal government, either directly by buying a government bond or indirectly by placing their money in a federally insured demand deposit in a bank. The bank, fearful of defaults by borrowers, may turn around and use its deposits to buy Treasury securities rather than to make loans to businesses and individuals—in which case the depositors are really lending their money to the government, with the banks as intermediaries.

What should the federal government do with all the money that is flowing into it directly or indirectly from nervous savers? It can use it to reduce the federal deficit, and this will have a positive effect on economic activity because it will check anticipation by businessmen and consumers of future inflation and tax increases. But the effect is likely to be modest because no one knows how the political process will respond to a lower deficit; the government may simply spend more, and on projects that do not increase people’s incomes.

The alternative is for the government to use the money that it’s borrowing from the private sector to convert inert savings (the savings that consist of direct or indirect loans to the government) into active investment. It can invest the borrowed money in construction and renovation of public buildings; beautification of public spaces; education and the environment; scientific research; preventive measures against natural disasters; expansion of airports and other transportation infrastructure (how about filling the potholes in the Obamas’ neighborhood in Chicago, which is also my neighborhood, for a start?); rebuilding inventories of military supplies depleted by our continuous combat operations; improving the national communications infrastructure; and promoting energy independence by building nuclear power plants, insulating buildings better, and making greater use of windmills, solar panels, and batteries to generate electricity. Such investment projects would increase employment and incomes and reduce economic anxieties, and by doing these things increase consumption and in turn private investment.

The problem is implementation. To be effective, a stimulus program has to be large, timely, well designed, and well executed. The $787 billion stimulus program (actually, it turns out, $862 billion) enacted in February 2009 has turned out to be none of these things. It should have been adopted in the fall of 2008, though obviously it’s not Obama’s fault that it wasn’t. It probably should have been larger. The money should have been spent more quickly and should have been concentrated in industries (such as construction) and areas of the country (such as California and Michigan) in which unemployment was (and remains) high. An experienced executive, rather than the Vice President, should have been placed in charge. And the stimulus program should have concentrated on investing in projects, which would put people to work.

Instead about two-thirds of the stimulus money consists of transfer payments, such as unemployment benefits, other welfare payments, and tax credits. Transfer payments are trickle-down stimulus. The hope is that the recipients will spend the money on goods and services, thus stimulating production. But suppose that instead they save the money and lend it back to the government by depositing it in a bank that buys Treasury securities with its deposits, so that the money passes in a circle from the Treasury to the Treasury. No stimulus there. Or suppose they use the money to buy goods that are sold out of inventory and the seller prefers to draw down his existing inventory rather than restock it with new purchases from his suppliers; then the additional purchases by recipients of stimulus funds do not stimulate new production. Even a stimulus measure favored by most economists—an investment tax credit—is not a sure thing. The tax credit that a business could obtain by making a new investment may not be large enough to overcome uncertainty over whether the investment can be recouped; and then the business will forgo the credit rather than invest.

The stimulus fell victim to delay, pork-barrel politics, congressional incompetence, and bureaucratic inertia—in short to the familiar pathologies of modern American government, together with conservative demagoguery. But it was not a complete failure. The enactment of the program built confidence at the peak of the nation’s economic anxieties, in the winter of 2009, and the injection of so much money into the economy in the ensuing months can’t have failed to have some positive effect on production and therefore employment. Had the unemployment rate reached 11 percent, as might have happened without the stimulus, the increase in fear and uncertainty might have triggered a deflationary spiral.

All this is speculation—and the administration’s trumpeting of the stimulus’s success is propaganda. But the paradox of thrift should make us hesitate to reject out of hand the idea of stimulus as a sensible response to an economic plunge triggered by a financial crisis.

Richard A. Posner is a judge on the U.S. Court of Appeals for the Seventh Circuit and a senior lecturer at the University of Chicago Law School.