From American Affairs Journal, Spring 2024 / Volume VIII, Number 1:
....MUCH MOREWell-managed economies grow at a decent pace while keeping unemployment and inflation at low and stable levels. By these criteria, all major developed countries have been run incompetently for the past two decades. They have experienced stagnation of output and incomes, the worst recessions since the Great Depression, and, more recently, a surge in inflation. Support for liberal democracy has been seriously endangered by this failure of economic policy.
In a previous article in this journal, “Monetary Policy, Tax Policy, and Investment,”1 I explained that, while economic policy has been appallingly bad, it is wrong to blame either central bankers or governments. Official policy must be rationally based on accepted theory and, as today’s consensus theory is wrong, the economy cannot be well managed until this theory is replaced. It would have been unfair to blame eighteenth-century doctors for bad medical practice, and it is unfair to blame economic policymakers today for their bad management. We should not blame the ignorant for their ignorance. Economists in central banks and treasury departments believe what they were taught when young; and students are still being taught similar nonsense when they study macroeconomics at universities and finance at business schools. Academic economists today are making a similar mistake to that made by British generals in the 1790s. In the words of James Thomas Flexner, “Washington learned the lessons of the American war . . . because he had no conventional lessons to unlearn. The British and the Hessians, on the other hand, suffered the confusion common to acknowledged experts when their expertise ceases to function.”2
The basic errors of the current consensus were introduced when the discipline was being developed after World War II. Macroeconomics barely existed as a discipline before the Depression. It was assumed, with little debate or analysis, that output rose steadily, with its fluctuations being unfortunate but mild, as demand responded to changes in short-term interest rates and could therefore be boosted by central banks if it were weak. The Depression exposed the inadequacy of existing economic theory to deal with a dramatic collapse in employment and output. We were fortunate that John Maynard Keynes was able to show that massive unemployment could be avoided by using government budget deficits to boost demand. Keynes observed that there are circumstances, which he called liquidity traps, when cuts in interest rates could not always increase demand. When monetary policy is incapable of reducing high levels of unemployment, fiscal stimulus is needed, which involves increasing government expenditure or reducing tax revenue.
Despite the great opposition that heterodox views always generate, events proved too powerful for change to be resisted, and the assumed expertise entrenched in academia crumbled. As old ideas were no longer held to be gospel, economics was in turmoil after the war. It had become an exciting subject and great intellectual efforts were made to produce a macroeconomic theory which incorporated Keynes’s ideas and included a way to manage the economy through the combined use of both monetary and fiscal policy. The result would eventually become the consensus theory, which is currently taught to undergraduate and graduate students. Following much hard work, the result is a mathematically coherent model of the economy, whose conclusions follow logically from a small number of assumptions. Its only fault is that these conclusions are wrong.
From Untestable to Incorrect
In any model of how the world operates, be it about physics, medicine, or economics, if the assumptions are wrong, it is highly likely that the conclusions which follow from them will also be wrong. This is what happened with the neoclassical synthesis. Its mistaken assumptions have become, over the past seventy years, so deeply embedded in academia that they are exceptionally hard to change. It is not only difficult for academics to unlearn what they learned when they were students; they must accept the bitter pill that their teaching and the papers on which their prestige rests are fundamentally invalid.
Economists are often accused of excessive reliance on math due to “physics envy.” As Paul Krugman wrote in 2009, “The economics profession went astray because economists, as a group, mistook beauty, cloaked in impressive-looking mathematics for truth.”3 This is a criticism, however, to which I only partly subscribe. The failure of consensus economics is not primarily caused by an excessive reliance on algebra, but through basing sound mathematical reasoning on mistaken assumptions. The glaring errors of current consensus economics come from a failure to insist that hypotheses must be capable of being tested against data and proved to be robust when so tested.
Nothing in science is ever proved; knowledge is encapsulated in models, and the one currently assumed to be correct is simply the best we currently have, and we arrive at the decision about which model is best by debate. The debate involves three main processes: one consists in testing the model’s conclusions to ensure that they follow logically from the assumptions, and the others are to test both the assumptions and the conclusions which follow from them to see that they are consistent with data evidence. Algebra is an excellent tool for checking that a model is self-consistent and is, therefore, a valuable tool for economists, though Krugman is correct that mathematical rigor, which shows the self-consistency of a model’s assumptions and conclusions, is sometimes assumed to validate the assumptions. Mathematics is a useful tool for checking models, but it produces dangerous nonsense when the consistency of the model is assumed to justify the assumptions. As Ricardo Caballero and others have pointed out, this is a far from rare mistake among economists.4
Such errors are often the sad result of bad luck rather than bad judgment. When the consensus theory was being developed, there were little data available which could be used to check the model. It could not be based on evidence and was, by necessity, aprioristic—founded on assumptions whose validity cannot be tested. Aprioristic models, insofar as they are untestable, cannot have the status of being scientific.
When data started to become available, the consensus model became testable. And when tested, the evidence falsified the model. Consensus theory has therefore moved from being unscientific, because it was untestable, to being simply wrong because it fails when tested. Without data, consensus theory was necessarily aprioristic and had the bad luck to choose its assumptions badly. Even at the time, this did not go unnoticed, as several economists held that serious errors were present in its assumptions. Four of these critics, in particular, stand out in hindsight as having been proved correct.
When Nicholas Kaldor criticized consensus theory for its aprioristic and thus unscientific approach, he wrote that “one must not fall into the error of supposing that assertions about reality can be derived from a priori assumptions. Whether well-behaved homogeneous-and-linear production functions exist or not is a question of fact. They cannot be presumed to exist as a consequence of some basic postulate. . . .” He also remarked on “the intellectual sterility engendered by the methods of Neo-classical Economics.”5
Robin Marris criticised the way in which the new consensus oversimplified by dividing the economy into only three sectors: government, private, and foreign. By failing to make a distinction between the business and household sectors, the neoclassical model assumed that those running companies would behave as if they owned the companies they ran. He argued that it was necessary to divide the private sector into two—a household sector and a business sector—as the decision-makers in the business and household sectors were motivated differently; they had, in economic terms, different utility functions. He therefore began “from the proposition that corporate directors may subject corporate policy decisions to utility functions of their own.”6
Fortunately, the distinction between companies and their owners is so obvious that national accounts separate the data on businesses from those that apply to households. It is therefore easy to show, from the way the two sectors behave, that Marris was correct in his criticism. Nevertheless, the error of consensus theory in conflating business and households into one sector was compounded by another: the model also assumes that shareholders are concerned with the value of their companies’ balance sheet net worth rather than with their share prices. This has led to considerable confusion in economic analysis and policy.
Hyman Minsky argued that the economy could not be kept in balance by any combination of monetary and fiscal policy. These tools could not prevent the bouts of instability engendered by financial speculation and the periodic crises that followed. Minsky therefore claimed that “Modern orthodox economics is not and cannot be a basis for a serious approach to economic policy.”6
Milton Friedman pointed out that, while the consensus model sought to include interest rates, it had no role for money, and claimed that ignoring the relationship between the amount of money in circulation and the size of the economy’s output was another serious error, notably with respect to inflation.8
These four criticisms were ignored when the neoclassical model was developed and, now that data have become available, not only does each of these objections seem totally vindicated but eminent economists have recently added another major criticism. Nobel laureate George Akerlof and MIT professor Ricardo Caballero have argued that we do not live in an economy whose stability can be maintained solely by keeping demand at its equilibrium level. The conclusion of consensus theory—that we live in a world in which stability can be maintained if one single equilibrium condition is achieved—follows logically if the cost of the capital needed to finance new investment varies only with fluctuations in real short-term interest rates, but Akerlof has argued that this assumption oversimplifies Keynes’s work.
All models simplify by cutting out variables that do not affect their validity; oversimplification means cutting out too much, so when applied to any specific example, it is a polite way of saying that the model is wrong. As equity provides the bulk of the capital needed to finance investment and is more expensive than debt, this oversimplification of consensus theory requires that share prices and equity returns fluctuate with short-term real interest rates. Yet, as I show in The Economics of The Stock Market and illustrate in figure 1, they don’t.9 As the chart shows, the returns on equity fluctuate around a stable mean, whereas those for short-term interests rates and bonds wander in unrelated ways. Thus, while the logic of the neoclassical synthesis has been impeccable, its failure is simply due to its assumptions.
Looking back at the postwar history of economic thought, the criticisms of Kaldor, Marris, Minsky, and Friedman should not have been ignored, and this failure can be summarised by saying that we live in an economy which does not have a single equilibrium. Akerlof, paraphrasing Caballero, has called this mistake “one-deviation-at-a-timisim.”10 Economic theory must therefore be revised so that these criticisms are no longer ignored, and economists must accept that we live in an economy in which several different equilibria must be maintained for stability.
Yet it is not enough simply to show that the current consensus is wrong; it must be replaced by a better one, not only for teaching theory but also for policy. We must have a sound theory if we are to avoid the failures of economic management which have been such a sad feature of the twenty-first century. To be an improvement, the new model must be as logically coherent as consensus theory, but differ crucially in that its assumptions must be testable and robust when tested.11
A Better Model
As all four of the older historic criticisms, together with the more recent questioning of its single-equilibrium conclusion, are valid, the model that replaces the current consensus must satisfy all five of these objections. This requires replacing the assumptions of the consensus model with empirically derived postulates, along with the pragmatic addition of policies which avoid rapid growth in the ratio of money to output....
Previous visits at various venues:
- Take a Walk on the Dark Side of Macro Expectations With Andrew Smithers
- Andrew Smithers: "Investment, Productivity, and the Bonus Culture"
- Andrew Smithers: "Savings Glut or Investment Dearth: Rethinking Monetary Policy"
- Andrew Smithers on Labor, Capital and Taxes
- Andrew Smithers: "The Rise of Carry and Macroeconomic Risk"
- "The economic performance of the United States and other major developed economies in the twenty-first century has been appalling"
