Illustration by Rob Dobi
This article appears in the April 2023 issue of The American Prospect magazine. Subscribe here.
Economics is commonly described as the science of scarcity.
We have limited resources, and we have to use them wisely. So there are trade-offs. In the old textbook example, if we want to spend more on guns, we have to spend less on butter. And so, not surprisingly, the question of how much we have to spend, now and in the future, is critical.
There are many things wrong with this seemingly impeccable and simple logic.
The first is that if we are not using our resources fully, then we can have more guns and more butter at the same time. Sometimes, we have people who would like to work but we don’t have jobs to give to them; other times, as the former head of the Federal Reserve Ben Bernanke once claimed, we have a “saving glut,” with firms and households saving so much the money has nowhere productive to go. In those instances, we don’t have to make any trade-offs. The country, and the world, has frequently been in such a situation. In the Great Depression, 1 out of 4 workers were unemployed; in the Great Recession, more than 1 out of 10. At the height of the pandemic, 1 out of 7.
The second problem with the guns-or-butter logic is that the market economy is often inefficient. Resources are wasted when they are not used as productively or as wisely as they could be. Of course, ensuring that resources are used well is supposed to be a core virtue of the market economy, as ruthless competition ensures that firms produce what consumers want at the lowest cost possible. But no one living in 21st-century America should believe that such a myth describes the economy today, marked as it is by mega-monopolies and oligopolies.
Can it possibly be the case that the most efficient use of our limited research resources should be directed toward making an ever-better advertising machine (the business model underlying Facebook and Google) aimed at better exploiting consumers through discriminatory pricing and targeted and often misleading advertising? Would an efficient 21st-century “market machine” be unable to deliver safe baby formula? That’s a simple enough product to get right, and yet last year, the country faced a massive shortage. And why did the market creep so slowly toward cost-saving renewable energy?
When there are inefficiencies of this kind, the economy can produce more guns and more butter by reducing these distortions. The economy is rife with such “market failures.” Public policy needs to be directed at reducing their magnitude.
Another major weakness in today’s economy results from lack of sufficient public investments. The obvious example is infrastructure: If we don’t invest enough in roads, harbors, and airports, it will cost private firms much more than it should to get their goods to market. Because there has been such severe underfunding in these areas, the returns on these public investments today are far higher than those on the average private investment.
There is a remarkable record of poor forecasts in critical moments, like the 2008 financial crisis.
And what matters for economic performance is social returns. When there are market distortions, such as when firms spend money to enhance their market power, they can create large private returns but low or even negative social returns. Arguably, investments in building a better advertising machine to target consumers more precisely may have a negative social return, even if Google and Facebook wind up being among the wealthiest firms on the planet.
Public policies redirecting overall investment toward more socially productive uses increases the size of the economic pie. But the economic returns from public investments in health and education and basic research and technology are even higher than in hard infrastructure, so the scope for increasing the size of the pie is even larger.
There is a third problem in the simplistic trade-off analysis, which centers on how those trade-offs are calculated. To do this, economists use models. Models are simplifications of reality. They attempt to capture statistically what will happen if we spend more on infrastructure or raise taxes. Underneath the arcane mathematics, though, there are always simplifying assumptions. There is nothing wrong with simplification. The problem is that if the models make the wrong simplification, they will give the wrong answer. And often, the simplification determines the answer. If one assumes that the economy is efficient, then of course one can’t get more guns without giving up butter. But why would one make that assumption in the first place, one might well ask, when it is obviously wrong? It’s hard to answer that question without impugning motives.
MOST OF THE MODELS THAT ECONOMISTS currently use ignore the role of market power in today’s economy. And in the gizzards of the models are a variety of other assumptions that affect how the consequences of any policy are calculated, including the macroeconomic consequences that determine the size of the pie and the nature of the trade-offs. The estimated responses to any policy change are claimed to be the most reliable estimates of what will happen, based on past data, using the “best” models and best statistical techniques. Typically, these estimates are not robust—with large variations in the estimates depending on how they are done and the sample period over which they are done. The sample period is in fact critical: The current situation may be markedly different from the one in which the studies were conducted. Applying those results to today leads to faulty conclusions.
For instance, if most of the time, in the historical sample, the economy was near full employment, as it was during the late 1990s and the beginning of this century, an increase in government spending would not lead to an increase in GDP. How could it? But both in 2008 and in 2020, government spending had a big effect, with GDP increasing a multiple of the amount of government spending, as standard Keynesian economics had predicted. During these periods, there were underutilized resources, and the underutilization would have been far worse in the absence of government action. The increased aggregate demand resulting from government spending led to a better utilization of resources.
Often, simple reasoning can beat out seemingly complex and sophisticated econometric modeling. In 2017, then-President Donald Trump proposed, and Congress adopted, a massive cut in the corporate income tax. The claim was made, supposedly based on models, that it would massively stimulate investment. It did not. It simply stimulated increases in share buybacks and dividends, funneling money to investors. It was, in effect, a big gift to rich corporations and their shareholders.
I had predicted that investment would not increase by much. Why? The corporate profits tax is a tax on pure profits, on the excess of returns over all the costs of production: labor, the goods that go into production, and capital. Firms make such pure profits, for instance, when they have market power. Some firms have a little market power. But in our economy marked by ever-increasing market power, many have a lot of market power—and thus large profits. The 2017 tax bill allowed firms not only to deduct the cost of their plants and equipment, but even to deduct some of the interest they paid. A basic result of standard economics is that a tax on pure profits does not discourage either investment or employment—and, by the same token, a lowering of such a tax doesn’t encourage investment or employment.
The standard models used by corporate interests to sell the tax cut assumed that the tax was equivalent to a tax on capital, simply forgetting the fact that expenditures on capital were tax-deductible. (It was, I suspect, not an innocent mistake.) If it were a tax on capital, it would have discouraged capital expenditures. One can easily calculate by how much a tax on capital might discourage investment, and voilà, one has an estimate of how much lowering the corporate income tax will encourage investment. The magic is in the assumptions, which are hard to discover.
JOHN LOCHER/AP PHOTO
Models assume a level of unemployment below which inflation starts to increase. But it cannot be reliably estimated.
A coherent model of the entire economy recognizes that such a corporate tax will lower the value of firms’ equity—and lowering the tax will correspondingly increase the value of equity. If there is a pool of savings to be allocated between holding equity (reflecting the value of after-tax pure profits) and productive capital, then an increase in the value of equity will crowd out real capital accumulation. At least in the medium term, lowering the corporate income tax may actually result in less investment, and reduced GDP.
Another critical assumption that goes into the standard modern macro econometric model concerns full employment. This is usually taken to be the level of unemployment below which inflation starts to increase, a number that is referred to as the natural unemployment rate, or NAIRU (non-accelerating inflation rate of unemployment). The idea is simple: If labor markets get too tight, wages start to increase, increasing the rate of inflation.
The problem is that the NAIRU cannot be reliably estimated, as the debate in the aftermath of the pandemic illustrates. Before the pandemic, we had very low levels of unemployment with very little inflation. Some thought that the pandemic had induced a permanent change in the labor market; for example, Larry Summers believed that undoing the inflation (which he wrongly attributed to excess aggregate demand, but was clearly the result of a series of pandemic-induced supply-side shortages and demand shifts) would require a high level of unemployment for a long period of time. Others thought the pandemic, with its unprecedented levels of job separations (particularly stark in the U.S.), had temporarily shifted the relevant curves, but that eventually matters would normalize. It might take a while; we know, for instance, that quit rates are much higher in the early years of a new job. With a much larger fraction of workers getting new jobs, aggregate quit rates would be expected to be higher. In fact, there is mounting evidence of a normalization of labor markets in just a few years as the pandemic winds down.
A third example of a macro model assumption involves estimating the impact of public investments. I’ve already pointed out that public investments yield very high returns, and even if taxing corporations resulted in less private investment (which it does not), diverting resources from private to public investment would increase the size of the national pie. But because public investment may actually increase the returns to private investment, such investments may crowd in private investment. Typically, such longer-term effects are not included in the budgetary analyses. While there may be uncertainty about the value of these effects, we can say, with some certainty, that they are significant. Assuming them away, as much of the budgetary modeling does, is wrong, and prejudices the policy analysis.
THERE ARE A HOST of other examples. Models build in our views of how the economy and society function. We know that there are differences in these views, and that predominant views change over time, so we shouldn’t be surprised that models incorporating different views would yield different results. Tragically for our country, the models that have been prevalent for the last quarter-century embed a particular set of views that are increasingly out of touch with the realities of today’s economy.
I’ve mentioned one aspect—the assumption of a competitive economy. More broadly, the “neoclassical economy” presumes profit-maximizing firms interacting with utility-maximizing individuals in perfectly competitive and efficient markets. But we know that neither firms nor households behave according to that model, and that markets are far from perfect.
These deviations can be of first-order importance. To take but one example: In the perfect-market model, there is what Arthur Okun called “The Big Tradeoff.” One can only have more equality at the cost of poorer economic performance. But increasingly, experts recognize that in our economy marked by high levels of imperfections, including rent-seeking from firms with market power, equality and economic performance can be complementary. We pay not only a high price for inequality in terms of social and political divides, but even more narrowly in terms of economic performance. Even establishment institutions like the Organisation for Economic Co-operation and Development and the International Monetary Fund view it this way. Yet, this perspective is still not incorporated into standard macro budget models.
To be fair, the models used in the U.S. are not as bad as they could be. A standard model of the right—dating back to Herbert Hoover and before—entails “expansionary austerity.” This view says that cutting spending, even in a recession, is expansionary, not contractionary! The magic is worked by what Paul Krugman has called the “confidence fairy.” Somehow, the cutbacks inspire such confidence that investors rush in and, in a self-fulfilling prophecy, this not only undoes the effect of the cutbacks but propels growth.
A main problem with this “theory” is that it runs counter to virtually all experience. Hoover’s cutbacks didn’t propel the economy into a new boom, but into an ever-deeper Depression. As did the IMF’s cutbacks in East Asia, Greece, Spain, Portugal, and Ireland. Investors understood the underlying economics better than the “modelers.” They understood that contractionary policies, like raising interest rates and budget cuts, are … well … contractionary. They understood that when the economy goes into a downturn, sales decrease and bankruptcies increase—and returns to capital decrease. Austerity leads to less investment. Households, worried about the future, husband their resources, so it can even lead to curbing consumption. The knock-on effects of austerity deepen the downturn. Common sense, once again, triumphs over the model.
There is a remarkable record of poor forecasts in critical moments, like the 2008 financial crisis and the euro crisis, by central banks and the international financial institutions. All of them were based on bad modeling. If the flawed modeling were just an academic exercise, that would be one thing. But policies are based on these models. Educations were interrupted and lives were broken by austerity. Millions lost jobs, homes, and livelihoods.
Flawed models have made us face false choices. It’s time to formulate new ones that accurately reflect the world in which we live. Only then can we make informed decisions that will lead to a healthy and robust economy for all citizens.
