John Maynard Keynes Essay, Research Paper
The Critics of Keynesianism: A Survey “He who knows only his own side of the case, knows little of that.”–John Stuart MillJohn Maynard Keynes’ General Theory marks a turning point in intellectual history. In less than a decade since its publication, the numerous converts to Keynesianism attained dominance in both the academic and political realms. Their hold on these positions has, since the 1970’s, weakened under the combined force of theoretical criticisms and practical failures. This theoretical criticism is not a coherent edifice; rather, this criticism consists of disparate strands of dissent from both the general principles and the specific applications of Keynesianism. This paper, then, will survey these criticisms of Keynesianism, describing them, classifying them, and tentatively evaluating them.One of the important barriers to a thorough study of anti-Keynesian doctrines is that Keynesianism is a broad tendency of thought rather than a rigid set of theorems. Still, there are certainly basic assumptions that all species of Keynesianism share. By stating them explicitly, it will be possible to see what portions of the Keynesian system the major schools of critics reject.Two fundamental postulates underlie Keynesian theories of all types:1. Unemployment is caused by insufficient aggregate demand.2. The proper means to eliminate unemployment is for the government to increase aggregate demand through discretionary monetary and fiscal policies.This definition is admittedly simple, but will prove its worth by both distinguishing Keynesians from non-Keynesians, and by allowing a clear way of classifying the many schools of Keynesianism’s critics. Roughly speaking, there are those who criticise the first postulate, and those who criticise the second. The first group includes rational expectations, sectoral shifts, real business cycles, the more extreme forms of monetarism, Austrians, and, perhaps inadvertantly, Keynesians who concentrate on microfoundations. The second group adds moderate monetarists, public choicers, and advocates of free banking.The critics of the first postulate can be split into two main factions. The first approach, which focuses on the microeconomic assumptions of Keynesianism, spearheaded the modern wave of criticism but no longer part of cutting-edge discussion; the second and more popular approach emphasizes the influence of real variables such as supply shocks, sectoral shifts, and search.The early critics observed that Keynesians had not fully explained the microfoundations of their models, and that this sin of omission led to dangerous long-run consequences. Unemployment, Friedman and others explained, is not caused by insufficient aggregate demand per se; it is caused by excessive wage rates. Increasing aggregate demand only effects employment if, due to nominal wage rigidity, the real wage falls relative to prices. A Phillips’ curve, then, does not describe a set of long-run equilibrium positions. It works only so long as the market does not anticipate what is going on. The widely varying pattern of inflation and employment combinations both between countries and within countries in different historical periods testifies to the power of this insight.Other modern critics of Keynesianism doubt that money illusion is possible or relevant; at least, it matters much less than Keynesians seem to think. Think of it this way: all economists, Keynesians included, agree that some positive amount of unemployment is the natural and inevitable result of freedom of contract, due to quits, dismissals, search and other frictions. Moreover, all economists agree that a change in real factors can affect the rates at which workers find and lose jobs; hence, there will always be some fluctuations in employment. Given this, is it possible that real factors have caused employment fluctations rather than aggregate demand fluctations? There are many variations on this theme. Robert Barro simply doubts that a “strong” correlation exists between real and nominal variables. David Lilien argues that sectoral shifts can account for about half of recent cyclical unemployment normally thought to be caused by fluctuations in aggregate demand. The radical divergence between this approach and Keynes’ should be obvious.The second group of critics, whether or not they agree with the first group, have a decidedly different orientation. They argue that, even granted the Keynesian view that unemployment is caused by insufficient aggregate demand, that there are better means to cure unemployment than active fiscal and monetary policies. For example, they argue that active intervention makes the economy unstable, since policy can change with the winds of opinions of politicians and the central bank. No one can know policy in advance, so they may make actions which, though reasonable given their ignorance of future policy, are foolish in light of the policy that actually happens. Such critics advocate rules to strictly delimit options the of monetary and fiscal authorities, such as a balanced budget amendment, or a constititionally fixed rate of money growth. This criticism is typical of both monetarists and rational expectationists.Other critics, especially with a public choice slant, doubt whether the government actually would choose to pursue the “optimal” policy even if it knew exactly how to achieve it. They think government actors are self-interested, not angelic servants of the public good. Put otherwise, they view government actions as endogenously determined by the motives of the officials and the incentives of the system, rather than exogenously determined by the wise advice of Keynesian economists. Some possible motives that would deter the quest for “optimal” policies are: the desire to win elections by subsidizing favored groups when expansionary fiscal policy is not needed, or to obtain seigniorage by economically unjustified expansionary monetary policy.A final notable band of theorists of the second group think that central banking is an ineffective way to maintain monetary equilibrium. Instead, they endorse “free banking,” a system whose mildly unconventional characteristics will be described later. Free bankers argue that unregulated banks would expand or contract their liabilities by varying their reserve ratios in response to changes in the demand to hold money – and that they would do so automatically in response to profit-and-loss indicators. In their view, this compares favorably with a central bank, which cannot easily discover how to adjust the money supply in response to changes in demand to hold or currency-deposit ratios. Since both the workings of free banking (which has, incidentally, existed historically, and is not merely idle hypothesizing) and the reasons why its proponents think their system would improve upon central banking are relatively unknown, this essay will devote extra space to their critique.Monetarism’s and Rational Expectations’ Microeconomic CritiqueMonetarists and rational expectationists have, it is fair to say, completely destroyed naive Keynesianism with their attacks on its microfoundations. In modern discussions Keynesians as well as anti-Keynesians agree that if wages were perfectly flexible, involuntary unemployment would be impossible. And, even if wages are not perfectly flexible in the short-run, almost everyone agrees that, first, wages are flexible in the long-run, and, second, that this implies that a permanent trade-off between inflation and unemployment is impossible.It is true that all participants to the macroeconomic debate now accept this microfoundational critique of Keynesianism. Still, the power of this attack is clear from the fact that Keynesian theories mitigate, and anti-Keynesian theories emphasize, the practical importance of this fact. Let us, therefore, investigate the classic statements of this critique, then examine the work that builds on it.The microfoundational criticism was forcefully expressed in 1962 by Murray Rothbard, just as the American government for the first time invited leading academic Keynesians to prescribe proper policy for the nation. Rothbard bluntly stated, “Keynesian and neo-Keynesian ‘compensatory fiscal policy’ advocates that government deflate during an ‘inflationary’ period and inflate+to combat a depression. It is clear that government inflation can relieve unemployment and unsold stocks only if the process dupes the owners into accepting lower real prices or wages. This ‘money illusion’ relies on the owners’ being too ignorant to realize when their real incomes have declined – a slender basis on which to ground a cure.”1Rothbard’s thesis was ignored by the academic community at the time, but other critics, notably Milton Friedman and Robert Lucas, successfully advanced similar objections after America experienced rising unemployment and rising inflation simultaneously – a phenomenon inconsistent with Keynesianism in its naive form. Writing in 1968, Friedman criticized the Keynesian view that increased aggregate demand increases employment: “But it describes only the initial effects. Because selling prices of products typically respond to an unanticipated rise in nominal demand faster than prices of the factors of production, real wages received have gone down – though real wages anticipated by employees went up, since employees implicitly evaluated the wages offered at the earlier price level.”2 Robert Barro is a leading advocate of replacing the Keynesian model with an alternate, “market-clearing” model. After defining the model, he summarizes its implications: “The theory predicts that changes in the monetary base are neutral. In particular, a one-time shift in the quantity of base money leads to proportional changes in nominal variables, but to no changes in the real variables.”3 Barro then makes a novel move. Economic theory, he says, must explain the effect of nominal variables on real variables only to the extent that the effect exists. Surely no one would disagree with this. Barro then proceeds to examine the historical evidence of relationships between nominal and real variables.Barro’s econometric methods have drawn fire from critics. We shall focus only on his conclusions. In the United Kingdom, he says, there is a negative relationship between wage inflation and unemployment for 1862-1913; no relationship from 1923-39; and a positive relationship for 1947-84. For the United States, there is a negative relationship for 1890-1913, with no relationship over 1923-39 or 1947-84. Extending the comparison to seventy-eight nations, he concludes that there is no significant relationship between real GNP growth rates and inflation, currency growth, or M1 growth. Barro notes that it is possible to explain even such relationships as do exist by unique historical shocks to the real economy. He concludes that, “At least in the long run, it is untrue that more inflation leads to a lower unemployment rate, or that to have low inflation a country must accept a high unemployment rate.”4Barro then turns to the evidence for short-run relationships between real and nominal variables. He here limits his analysis to the United States; first he discusses the period of 1890-1940, then the period from 1947-1984. In the earlier era, Barro finds no evidence that changes in the monetary base have real effects, but thinks that there is good evidence for a relationship between real output and changes in the ratio of M1 to the monetary base – specifically, banking panics and changes in reserve requirements. For post-World War II, Barro says that anticipated monetary expansion had no real effect, while a 1% increase in unanticipated money normally increases real GNP by 1% over a one or two year period.Here is Barro’s calm conclusion to this chapter: “it is the nonneutral effect of the monetary base – to the extent that it exists – that conflicts with our theory. Although this result deserves some weight in our thinking, it is probable that the weight has usually been too large.”5 Here we see the important features of the rational expectations’ critique of Keynesianism. First, although the the market-clearing theory fails to perfectly explain the data of history, Keynesianism has severe empirical problems of its own. Whatever their claims to be more interested in “facts,” than their opponents, Keynesians have made strong assertions without sufficient evidence. Second, Barro does not claim that nominal variables never effect real variables. He argues that the closeness of their causal relationship has been over-rated.The Sectoral Shifts HypothesisMost macroeconomic models treat the economy as a single market. Then, if unemployment rises, we can usually assume that it is caused by insufficient aggregate demand. But some economists, notably David Lilien, have noted that the economy can also be treated as composed of several markets or sectors, and that more unemployment could be caused by shifts between sectors with total demand unchanged. Lilien then shows that if the percentage of persons who lose their jobs increases temporarily – as might happen due to a large sectoral shift – then the natural rate of unemployment rises. This change is not a deviation from the natural rate that might be easily corrected by demand adjustments. Instead a sectoral shift represents a change in the natural rate itself. If wages were perfectly flexible, Lilien’s problem would not arise. But even if wages are not flexible, traditional Keynesian policies would not really solve the problem. As Lilien states, “Such policies may have been successful in delaying or smoothing the cyclical pattern of unemployment, but since inadequate demand was not the source of unemployment, aggregate demand policies were not an appropriate cure.”6Lilien’s statistical support for his sectoral shifts hypothesis has been challenged – for example, by Katz and Abraham.7 Their dispute is too complicated for this paper. What seems important about sectoral shifts, whether or not it has in fact been important in modern times, is that if it were true, a jump in unemployment would not necessarily indicate aggregate disequilibrium. It might be a short-run change in the natural rate in disguise. So a fall in employment is not ipso facto grounds for expansionary monetary or fiscal policy. One must also show that the job losses are not mainly the result of sectoral shifts.Explanations of Wage-RigidityA. Implicit ContractsThe success of the microfoundational criticisms of Friedman, Barro, and associates can be seen in the wealth of attempted responses. No longer do the conversants debate whether wage rigidity is necessary for Keynesian theories and policies to work; now they discuss why wage rigidity exists. Since everone agrees on the major premise, that wage rigidity is necessary for Keynesianism to be true, the debate focuses on the minor premise, of how wage rigidity is possible.Probably the first major attempt to explain wage rigidity was “implicit contracts” as advanced by theorists such as Costas Azariadis.8 This theory begins with plausible assumptions: workers are risk-averse, and therefore prefer a certain and stable series of payments to one that varies with their short-run, fluctuating productivity; employers, in contrast, are risk-neutral. Hence, we have a clear situation where both parties may gain by shifting the burden of risk through a system of insurance. The result: wages become rigid because market participants desire it. Some economists misinterpreted Azariadis’ conclusions. They observed that implicit contracts make wages rigid. Next, they made the plausible jump that while this is reasonable from the point of view of those making the contracts, it allows the familiar Keynesian results of possible involuntary unemployment and a wide role for discretionary monetary and fiscal policies. However, this view is wrong. Azariadis never made the above conclusion, and in a later article, Akerlof and Miyazaki showed explicitly that it is specious. Put simply, if workers are risk-averse to drops in wages, they are surely even more risk-averse to losing their job altogether. If faced with a choice, risk-averse workers would clearly prefer “implicit insurance” against job loss to a “policy” that covered them against the lesser hazard of a pay cut. Akerlof and Miyazaki state this thesis clearly: “Because unemployment in many Keynesian macro-models is caused by rigid wages, by a parallel argument it has been held that the smoothing of wages caused by implicit contracts results in non-Walrasian fluctuations in employment. The present paper questions this last claim on unemployment; for it is demonstrated here, if workers can (implicitly) make contracts with the firm, they can also readily insure against employment variations (i.e. layoffs) and, as a result, implicit contracts even with sticky wages will lead to full employment, in most instances, rather than to unemployment,”9 (emphasis added).So we find a curious conclusion. Implicit contracts, rather than being the underlying defect that prevents the efficient operation of the labor market, turns out to be another way for individuals to quietly better their condition through voluntary exchange.B. Efficiency WagesEven though implicit contracts theory cannot explain involuntary unemployment, Keynesians see that its basic approach is persuasive to economists since it relies on traditional assumptions. Implicit contracts theory assumes that firms maximize profits and, more broadly, that economic institutions are not brute facts but instead the rational solution of market actors to problems. Efficiency wage theory is another explanation for involuntary unemployment that shares this basic approach. Put simply, it argues that firms may prefer to pay above market-clearing wages rather than face the consequences of cutting them. And since firms’ sole desire, by assumption, is to maximize profits, efficiency wage theory tries to explain why paying a lower price for labor may, on the net, be an unwise move. The most plausible case of this is in undeveloped countries where cutting laborers’ incomes may leave them unable to secure their bare sustenance – leading to poor work performance.10But what matters this for industrialized nations where everyone earns well above a subsistence income? Here, advocates of efficiency wage theory point out other plausible cases where employers might gain by paying above the market wage. First, they might do this to reduce shirking. Workers paid more than their opportunity cost would work more diligently lest they find themselves replaced by another worker eager for a high wage. Second, they might do so to solve an adverse selection problem: namely, workers willing to work for unusually low wages may be less able than those who insist on holding out for normal wages.11In its pure form, efficiency wage theory can explain some involuntary unemployment but cannot explain why nominal changes should make any difference. Yellen agrees in principle, but argues that if one weds efficiency wage theory with the assumption that employers set wages according to “rules-of-thumb” – that they only optimize perfectly in the long run – Keynesian results follow. Be that as it may, “rule-of-thumb” behavior could yield Keynesian results all by itself. So efficiency wage theory seems somewhat skew to the whole issue of why nominal changes have real effects – although it can explain why involuntary unemployment might exist even in the long run.Efficiency wage theory shares a major defect with Keynesian theory generally: it does not boldly state its basic assumptions upfront.12 The critical underpinning of efficiency wage theory is imperfect information of employers. If employers could perfectly monitor workers, efficiency wage behavior reduces to compensating wage differentials. Jobs that demand more effort would pay more, those that require less effort, that openly condone “shirking” as part of the benefit package, would pay less. But when employees can potentially fool their boss about how much effort they put into their job, the bosses may rationally choose to give their employees more money so that they will fear job loss and cheat less. Here is a plausible analogy that shows why an employer might pay efficiency wages: “It is instructive to contrast the labor market and certain product markets. Firms that rent bicycles, cars, or video cassettes wish to deter misuse and damage but are not usually thought to create rents with their pricing. Instead they rely on deposits. At least in part this is because denying future access to those who misuse products is likely to be difficult. On the other hand, landlords often price apartments below the market to insure loyalty on the part of their tenants rather than relying on a huge security deposit.”13Efficiency wage theory is, therefore, reducible to two aspects: compensating differentials and imperfect monitoring. No economists think that there is anything wrong with compensating differentials. As for imperfect monitoring, critics of efficiency wage theory note that it could be solved by auctioning off jobs to workers in exchange for security bonds. Workers would place a suitable bond with their employer. If caught shirking, they would be fired and lose their bond; if not caught shirking (either because they don’t shirk or because no one notices) they would get their bond back at the end of the period. (In a multi-period setting, presumably, the employer would keep the bond until the employee chooses to quit.) Advocates of efficiency wages concede that this would work if employees were willing, but suspect that legal restrictions and damage to worker morale kill the labor market’s cure in the cradle. This is a complicated issue, but it is clear that the market has many tools to whittle away at the power of efficiency wages.A close cousin of efficiency wage theory based on rational behavior is efficiency wage theory based on “sociological” considerations. Here, it is fair to say, “sociological” is a euphemism for behavior that economists would call “irrational.” Thus, they may posit “partial gift exchange” behavior where employers pay workers more than the minimum necessary to secure their services, and the employees respond by applying more effort than their job description demands. Or workers may retaliate against their employer if they feel that their wages are unfair. The standards for “fairness” would be set by prevailing customs.14This species of efficiency wage theory is more open to criticism that its profit- maximizing relative, since it is hard to imagine that the situations that it describes could be stable. Since the individuals aren’t acting optimally, they create profit-opportunities for anyone who is willing to act optimally. One could draw an analogy between sociological theories of employment and sociological theories of racial discrimination. Neither can show why sub-optimal behavior does not beget the seeds of its own destruction by creating blatant profit opportunities for anyone willing to break with the suboptimizing herd.C. The Insider-Outsider ApproachThe last major explanation of wage rigidity that we will examine is in many ways a throwback to an older theory of unemployment that blamed current employees’ use of violence and sabotage to deter potential competition from the unemployed. Mises, clearly sympathetic to this theory, writes: “The labor unions are practically free to prevent by force anybody defying their orders concerning wage rates and other labor conditions. They are free to inflict with impunity bodily evils upon strikebreakers and upon entrepreneurs and mandataries of entrepreneurs who employ strikebreakers.”15A updated version of this is known as the “insider-outsider” approach.16 Here, the insiders are current workers – possibly but not necessarily unionized. The outsiders are unemployed workers who would be willing to do the same job as those currently employed for the same or slightly lower wages. There is thus a clear conflict of interest between these two groups. Earlier theorists such as Mises focused on the use of violence and sabotage to deter “strikebreakers.” But the insider-outsider theorists implicitly note that this is a crude and needlessly bloody method for the insiders to secure their ends. It is also possible for them to refuse to share their firm-specific training and knowledge with new workers – thereby reducing the marginal product of the outsiders below what it otherwise might be. Or, attacking the problem from the other direction, the insiders might be unfriendly toward outsiders or otherwise artificially worsen working conditions, thereby raising outsiders’ reservation wages to such a point that they prefer to be idle. The success of these tactics depends strongly upon the employers’ imperfect information of the insiders’ conduct: else, the employers would probably retaliate.This theory has some virtues that other explanations lack. Some versions can explain why short-run money illusion may imply significant real effects. If insufficient aggregate demands leads the pool of outsiders to swell, the insiders’ exploitation of their position can explain why wage rates do not quickly adjust to solve the problem. Conversely, it explains why an increase in aggregate demand could solve the problem by “transforming” outsiders into insiders – at least in Lindbeck and Snower’s model where an outsider only needs one period to graduate into an insider.Still, aggregate demand management seems like a backwards way to solve or at least substantially mitigate the insider-outsider problem. The individual firms might themselves subdue the problem by adopting a seniority-based pay scale, such as the one described in Salop and Salop’s “Self-Selection and Turnover in the Labor Market.”17 In the Salops’ article, firms solve the problem of variable employee tenure by setting a pay scale dependent upon seniority to discourage turnover. Employers with an insider-outsider problem might set such a pay scale so that insiders feel less threatened by outsiders who have newly joined the firms – since they would no longer be in such direct competition with one another.Second, rather than taking positive action to mitigate the insider-outsider problem via demand-management or industrial policies, the government could just take the negative step of repealing laws that promote the market power of the insiders. What is needed here is not a draconian police-state crackdown on labor unions, but a return to unabridged legal freedom of contract in the labor market for both employers and employees. While this would not solve the insider-outsider problem entirely, it would probably get rid of its extreme manifestations.The Argument Against Active Policy from Uncertainty and LagsLet us assume that Keynesian theory is perfectly correct. At first it would seem that Keynesian practice would follow immediately. The government should use its monetary and fiscal policies to insure the “full-employment” quantity of aggregate demand, expanding if there is a short-fall, contracting if there an excess. And of course, since the private sector is free to alter its spending patterns at any moment, the government needs the flexibility to counterbalance any disturbance. Hence, the monetary and fiscal authorities should have free reign so that nothing prevents them from correcting the defects that they alone have the ability and will to fix.But a second look here reveals that Keynesian policies do not necessarily follow even if Keynesian theory were conclusively proved. They must also demonstrate a second series of propositions about the ability and willingness of the government to act as the theory prescribes.Critics of Keynesianism note this and react appropriately. They have challenged the ability of monetary (and fiscal) authorities to actually know when and how much they should adjust aggregate demand. In fact, one could doubt that the government will get the direction correct. On top of this, other critics question the willingness of the government to behave as the Keynesian theory prescribes. Unless we assume that government officials, unlike other human beings, joyfully serve their fellow men and never take advantage of their position, we face a clear agent-principal problem if authorities hold wide powers unchecked by clear and binding rules. So even if the “first-best” situation without agent-principal difficulties set no prior restraints on the government, it is futile to act as if such a state of affairs is really possible.Let us begin with the argument that the government discretion is harmful because they do not really know when to do what. Milton Friedman is the most famous and persuasive critic of Keynesianism on these grounds. He has two main arguments: first, that there are “long and variable lags” between the identification of a problem and the effects of the designed remedy; second, that activist policy often itself becomes a source of instability since policy itself becomes a variable that the market must guess at.In his classic A Program for Monetary Stability, Friedman summarized his empirical findings on long and variable lags: “on the average of 18 cycles, peaks in the rate of change in the stock of money tend to precede peaks in general business by about 16 months and troughs in the rate of change in the stock of money to precede troughs in general business by about 12 months+ For individual cycles, the recorded lead has varied between 6 and 29 months at peaks and between 4 and 22 months at troughs.”18If this were the case, it is difficult to see how any policy maker could know which direction to adjust his policy, much less the precise magnitude needed. Due to long lags that preclude accurate forecasting, he cannot know what the state of the economy will be when his policy takes effect. And due to variable lags, even if he knew all future states of the economy (in the absence of his adjustments, that is) he could not know which future state of the economy he should design his policies to correct. Many Keynesians, Friedman notes, advocate “leaning against the wind.” By this they mean, in some sense, that the monetary (and fiscal) authorities should try to balance out the private sector’s excesses rather than passively hope that it adjusts on its own. Friedman tests the Fed’s success by checking whether the rate of money growth has truly been lower during expansions and higher during contractions. He admits that this method of grading the Fed’s performance is open to criticism, but decides to go ahead and see what turns up. He finds that Fed has – for the periods surveyed – been unsuccessful. “By this criterion, ” he explains, “for eight peacetime reference cycles from March 1919 to April 1958+actual