America’s most recent seismic economic challenge, the financial crisis of 2007-9 and its long aftermath, was a challenge to economics as well. Like the worldwide depression of the 1930s and the pervasively high and persistent inflation of the 1970s and early 1980s, the financial crisis was in part a consequence of happenstance and idiosyncratic events. Also like those two formative episodes, however, the crisis was the product of incomplete understanding and consequently flawed thinking within the discipline of economics, and especially macroeconomics. There is ample ground to believe that if macroeconomics in the decades leading up to these events had followed a different path, the policies pursued would have been different too. If so, the financial crisis would have been less severe, and its broader economic repercussions less damaging.

Much of the public discussion in the United States of what the financial crisis said (and says) about economics has focused on two issues. One was the failure of nearly all economists to foresee what happened. The second was the inadequacy of the federal government’s fiscal response. Neither matter, however, is particularly informative about the state of macroeconomic thinking. Despite the fascination of a generation and a half ago with the theory of “positive economics” that held out predictive ability as the sole test of a would-be science’s veracity and usefulness, in fact most economists are not in the forecasting business. Moreover, it is simply not true that predictive ability is our only test of scientific validity. Almost all scientists today, for example, accept some version of the theory of speciation as first posed by Darwin and Wallace. But because of the randomness of the posited mutations, the theory is inherently non-predictive. This hardly discredits evolutionary biology.

The inadequacy of the government’s fiscal response was a useful learning experience, but not one that was informative about the state of economic thinking. At the outset of the Obama Administration, in early 2009, the President and the Congress agreed on a $787 billion stimulus package combining tax cuts and spending increases. Most economists understood at the time that the size of the program, around 5 percent of one year’s national income but stretched out over roughly three years, was not enough to offset the largest business decline since the 1930s. Most also understood that the program’s composition—tax cuts broadly spread across the population rather than directed toward people most likely to spend the money quickly, spending programs directed broadly across the government rather than targeted toward rapid use, and transfers to state and local governments that permitted them merely to fill existing budget holes if they chose—was far from optimal for spurring economic activity. But democracy doesn’t always deliver what the technical experts recommend, and the package that became law was presumably the best on which Congress and the President could agree. And if it wasn’t, the failing was a matter of politics, not economics.

Assessing those policy failures in the crisis that bear on the state of economics as an intellectual discipline (again, especially macroeconomics)—and, even more so, drawing inferences about what those failures imply for the future direction of the field—instead requires a more focused approach. Several elements of pre-crisis macroeconomic thinking stand out in this regard.

One is the pervasive disregard of credit markets in macroeconomics today, and in parallel the lack of attention to the differences among the various assets that individuals and firms hold and trade in the financial markets. Although the subject was central to the economics of an earlier era, markets for privately issued debt instruments sit uncomfortably within today’s standard macroeconomic methodology of the “representative agent,” which for convenience assumes that all of the economy’s private agents are identical. If two firms or two households are identical in all respects, there is no reason one would borrow from, or lend to, the other. (Why would one person want to borrow funds at 6 percent while someone else—with the same wealth and income and spending, and the same preferences and prospects—chooses to lend at 6 percent?) But of course such credit transactions, in which any given household or firm may be a borrower or a lender, or even both simultaneously, are actually the norm, not the exception. Familiar ways of departing from the representative-agent restriction in today’s macroeconomics include distinguishing young households from old ones (the young borrow from the old), and distinguishing risk-tolerant households from risk-averse ones (those with a greater tolerance for risk borrow from the more cautious). But these highly stylized departures fail to capture most of what happens in actual credit markets.

One reason this difficulty of accommodating credit markets within the representative-agent technology has attracted so little concern is simply that the technology itself is so convenient for purposes of the mathematical modeling on which most of macroeconomics today relies. Here, therefore, is one more piece of evidence supporting those who maintain that economists’ theories too often come down in the wrong place in resolving the trade-off between comprehensiveness and realism on one side and mathematical tractability on the other.

A more conceptual root of the problem, however, is the still-lingering legacy of a tradition, in some quarters of the discipline, of treating all non-money assets as perfect substitutes. Most versions of Milton Friedman’s monetarism, for example, omitted assets other than money (no stocks, no bonds, no savings certificates), and therefore omitted privately issued liabilities like mortgages and ordinary loans altogether. To be sure, economists working within this line of thought understood that households and firms have both assets and liabilities. But the assumption, supposedly grounded in empirical evidence, was that the quantity of assets held by the public other than money did not matter for purposes of macroeconomic questions, nor did differences among those non-money assets, nor did either the quantity or the character of whatever private liabilities were issued.

The origins of the crisis, and the way in which it unfolded and exerted its perverse effect on nonfinancial economic activity, showed that assumption to be a bad one. The lack of attention to private credit markets in general, and the further assumption that if such instruments exist they are all perfect substitutes, rendered macroeconomic thinking unequipped to anticipate, or to address once they arose, many of the phenomena at the root of what happened.

In the years leading up to the crisis, many economists understood that the United States was building too many new houses (more than two million per year during much of 2004, 2005, and 2006). But few paid attention to the credit market conditions that enabled this extraordinary surge of home construction. Increasingly lax underwriting standards facilitated the sale of many of those houses to purchasers with only fragile prospects of servicing the debt they took on to finance them. In parallel, the lenders, some of them highly leveraged, took on significant exposure to risk in the event of delinquency or default by the borrowers. And the proliferation of new “derivative” instruments further compounded these problems by magnifying the risk and attracting yet additional classes of investors to assume it. It is no accident that these disturbing phenomena, as they built up, attracted attention primarily among finance economists and practitioners, not macroeconomists.

A second reason for macroeconomists’ failure either to anticipate the crisis or initially to appreciate its severity once it occurred was the increasingly standard application of rigorous notions of optimization by individual households and firms—as if every economic decision is somehow spit out of a supercomputer having far more information and processing ability than any human but none of the biases, preference for shortcuts, or other foibles. This extreme vision of economic decision making also usually includes the separate but intellectually related assumption that business people and ordinary individuals assess their economic environment according to so-called “rational” expectations. The idea that individuals and firms act to further their self-interest dates to the very beginnings of modern economics. Unlike the mercantilists who preceded him, Adam Smith assumed that people mostly understand what is in their self-interest when they act as producers of goods and services. (In contrast, Smith made no such assumption about people’s behavior as consumers; he was openly contemptuous of many consumers’ misguided choices.) The introduction of utilitarianism in the nineteenth century facilitated extending pursuit of self-interest to outright optimization, according to which people not only act to further their self-interest but do so to the maximum extent possible.

Along the way, everyone recognized that expectations of future economic conditions and events matter for many economic choices. In the early twentieth century, Keynes emphasized the role of expectations, but he had little to say about how they are formed. Following the work of Richard Muth and Robert Lucas in the later decades of the century, the conventional working hypothesis among macroeconomists became that of “rational” expectations—meaning that individuals, on average, form their views of the future as if they completely understand whatever process will actually produce the future outcomes in question, and factor their knowledge of that process into their thinking. Standard macroeconomic analysis has also accompanied this line of analysis with the further assumption that individuals, on average, do not make the mistake of acting as if a process that cannot go on forever will do so—so that, for example, no one makes the mistake of buying a house in the belief that house prices will continue indefinitely to rise by 15 percent every year just because that has now happened for several years in a row.

Taken on its own terms, this set of assumptions is not necessarily a bad way to discipline macroeconomic analysis. In the absence of either optimization or some well-specified alternative to it, a dauntingly broad range of representations of economic behavior is potentially consistent with pursuit of self-interest. In the absence of either true-process-consistent expectations (that’s what “rational expectations” means in practice) or some well-specified alternative to them, any representation of beliefs about the future is potentially admissible. Such undisciplined thinking risks rendering macroeconomic hypotheses untestable against observed experience.

But assumptions can unhelpfully restrict thinking as well, and the recent crisis stands as a case in point. The question is how rigorously to apply the presumption that people always act optimally, or that on average they form their beliefs about the future “rationally,” or that they know when some process cannot go on forever and therefore act accordingly. Surely no one interprets the optimization to mean that nobody ever does anything foolish. But to what extent can foolish behavior predominate in one market or another? Similarly, no one interprets “rational” expectations to imply that no price, in any market, is ever at an incorrect level (meaning, again, a level other than the equilibrium implied by the actual process that determines outcomes). But what kinds of departures are admissible?

Most economists accept that the market can misprice the shares of any one company, or any one borrower’s bonds. Can the market misprice the equity of an entire class of companies? Or the debt of an entire category of borrowers? Can the stock market as a whole establish a “wrong” price? Can the bond market? Can prices in some market—for houses, say—rise to a level explainable only if the buyers believe the price increase will continue indefinitely (even when of course it can’t)? Even if macroeconomists did not take the discipline’s standard assumptions to their logical extreme, it is fair to say that during the build-up to the financial crisis too many of them were unwilling to give serious weight to the prospect that the average price being paid for houses was too high, or that the interest rate being charged on a wide class of mortgage debt was systematically too low, or that the price of the securities backed by these instruments was therefore too high, or that large numbers of institutions that invested in these securities were bearing risk well outside their safe range of tolerance.

The mindset shaped by these assumptions also often blocked serious consideration of what would have been corrective policies. The most glaring example was in the regulation of mortgage lending. Beginning well in advance of the crisis, not only some private individuals but various government agencies urged a tightening of mortgage lending standards, or restrictions on highly leveraged financial institutions’ investment in mortgage-backed securities, or both. Such warnings were largely ignored, however, and the proposed correctives blocked or rejected, on the ground that government regulation of credit markets populated by rational investors was unnecessary: Investors would rationally judge the value of the instruments they bought.

A parallel implication was that bank depositors with amounts beyond the deposit insurance limit (at the time, $100,000 per account), and purchasers of uninsured bank obligations, would monitor the balance sheets of the institutions to which they lent and would exercise a non-government—and, in the eyes of many, therefore presumably more effective—regulatory function by not lending to any bank that took on excess risk. This form of privately imposed safety-and-soundness regulation would apply even more stringently to “shadow banks” (think Bear Stearns or Lehman Brothers) whose obligations were entirely uninsured, and which did not have access to Federal Reserve lending in the event of distress.

To what extent these beliefs were typical of economists as opposed to political and intellectual conservatives more generally—a category that includes many economists but excludes many others—is impossible to know without detailed opinion surveys or some other form of evidence that does not exist. In the end, however, these beliefs were dramatically proven wrong. As Alan Greenspan stated, after the largest U.S. banks avoided failure only by turning to government bailouts, “Those of us who have looked to the self-interest of lending institutions to protect shareholders’ equity, myself included, are in a state of shocked disbelief.”

A third lacuna in macroeconomic thinking that turned out to be important was the failure to take account of the arrangement that then-soon-to-be Justice Brandeis, just over a century ago, labeled “Other People’s Money.” Economists certainly do appreciate the complications that arise when one person acts on behalf of another. Such “principal-agent” issues are the heart of what the discipline calls “contract theory,” and they figure in other areas too. Macroeconomic thinking, however, has mostly not taken serious account of this aspect of reality, again presumably on the ground that differences between the interests of decision-makers and the interests of those on whose behalf they are acting do not matter for purposes of understanding how an economy as a whole behaves.

The issue turned out to be important in the crisis, however. Many of the individuals who bought the mispriced mortgage-backed securities and unsound bank deposits—and thereby drove the interest rates on these instruments down, hence stimulating ever greater use of them—did so not with their own funds but for institutions who employed them, whose interests were only loosely aligned with their own. In the end, most of the key figures whose actions led to the failure, or near-failure, or failure-but-for-government-intervention, of many of these institutions came out very well personally, often earning tens of millions of dollars in salary and bonus even as the value of their employer’s stock plummeted. At one point during the aftermath of the crisis, Alan Greenspan commented that he did not personally know, or even know of, any executive of a U.S. bank who had either gone bankrupt or lost his or her house in the crisis. (In a related phenomenon, beyond the realm of economics per se, this pattern likewise extends to statutory violations; violations of the law during the crisis by employees of U.S. banks have mostly resulted in penalties imposed on the bank’s shareholders, not the individual perpetrators.)

Finally, the assumption that markets—for goods and services, or for labor, or for financial assets—always clear, so that demand equals supply, has increasingly come to dominate macroeconomic thinking in recent decades. The experience of the 1930s, with long lines of the unemployed vividly disproving the claim that anyone willing to work at the going wage could get a job, has become the distant past. So too has the more temporary (and more specific) experience of the oil shortage in the early 1970s. A generation or two ago, models in which demand fell short of supply, or vice versa, and for more than a fleeting interval before the relevant price adjusted to equalize them, were familiar in macroeconomics.

Today such disequilibrium theorizing is rare. But during the financial crisis the markets-always-clear presumption also proved problematic. In particular, it clouded many macroeconomists’ assessment of proposed central bank intervention. Once the reality of an out-and-out financial crisis became apparent, many long-time students of monetary economics urged central banks to follow the nineteenth century Englishman Walter Bagehot’s famous dictum to lend freely to solvent institutions, at a penalty rate, on good collateral. The Bagehot principle has much to recommend it—if it is clear which institutions are solvent and what collateral is good. Under conditions of market failure like those prevailing during the crisis in the market for mortgage-based derivative instruments, however, neither judgment is straightforward.

The problem in applying the Bagehot rule was more deep-seated than just the operational difficulty of evaluating some credit market instrument, or some lender’s portfolio, under disorderly market conditions. During the crisis, markets were sufficiently dysfunctional that whether some credit was worth 80 or 40, or perhaps nothing at all, depended crucially on what action the central bank itself would take. Would the Federal Reserve intervene? If so, what instruments would it buy? And how much? Which instruments constituted “good collateral” and which institutions were solvent depended on the Federal Reserve’s own actions. Under such conditions the Bagehot rule becomes operationally meaningless, not just as a matter of technicalities of implementation but in its fundamental logic.

The 2007-9 financial crisis challenged macroeconomics, and in key ways macroeconomics failed the test. It is impossible to know what would have happened differently if the ideas and methods with which macroeconomists approached the crisis as it developed and then assessed potential reactions to it once it broke—not just the concrete models they had in their toolkit, but the underlying assumptions that shaped their intuition—had been different. Contemplating the possibilities points the way to what may be the most useful avenues for new research in the field. Without assumptions there is no theory, and without theory there is no analysis. Whether macroeconomists will be willing to find new assumptions on which to base their ideas, leaving behind the ones that straight-jacketed their thinking during the crisis, will largely determine whether the discipline will regain the usefulness it once seemed to have.