For the past 20 or 25 years Keynesian demand management has been pretty much in eclipse among policymakers, as has Keynesian analysis among professional economists. The fundamental reason is a practical one: demand management as presently understood has proven unable to effectively control inflation. Criticisms by neoclassicists(1) have also played a part, but they did not win much credence until after practical considerations had pretty well emasculated the Keynesian strategy.
Various Keynesian(2) schools of thought have attempted to meet the neoclassical criticisms, but the debate is still unresolved: neither side has yet convinced the other. One school, which professes to be Keynesian and has come to be known as New Keynesianis Frank, 1986; Mankiw and Romer, 1991; Symposium on Keynesian Economics Today, 1993] has attempted to reconcile Keynesian and neoclassical analysis. However, they seem to have accepted so many of the assumptions on which the neoclassical position is based that I am not sure which camp they belong in. I therefore identify them as a third camp, neither neoclassical nor Keynesian.
I believe that the major neoclassical hypotheses in three crucial areas, on which their conclusions heavily depend, are open to challenge: the nature of the equilibrium that is relevant for the discussion of the matters at issue, the nature of the unemployment observable in the real world, and the implications of the present inflationary environment.
Finally, in academic circles one of the most telling neoclassical criticisms of Keynesian analysis has been the claim that it has not provided a convincing microeconomic foundation for believing that nominal prices and wages are sticky; since neoclassicists assert that flexible prices would ensure the prompt achievement of full employment, they see this criticism as very damaging to the Keynesian case for prolonged underemployment equilibria. The importance of sticky prices and wages for the Keynesian case has been disputed by some Keynesians, however, and in any case I believe a plausible microeconomic base for belief in stickiness can be demonstrated.
Neoclassical analysis relies heavily on the tacit or overt assumption that something like the purely-competitive long-run static-equilibrium position is quickly and relatively painlessly attained--in the case of some rational-expectations models, apparently almost instantly [Hoover, 1984]. Curiously, though monetarists and other neoclassicists usually pride themselves on the empirical verification of their propositions, they have made no such effort in this case. Until someone, monetarist or other, produces a definitive empirical study of the matter,(3) therefore, we must stick with the contrary conclusions of a long list of distinguished economists of many schools. Also, with few and unimportant exceptions, neoclassical analysis ignores the traditional Marshallian concept of disparate short-run, medium-term, and long-run equilibria, each of which offers perspectives and insights not discernible in the others.
The long-run equilibrium of the model of pure competition may be identified with the situation that would obtain if the equilibrating forces could be left to operate for an indefinitely long time without new disturbances to the economy. In that situation the labor market would have two important characteristics: first, each worker would be indifferent between a little more real income and a little more leisure; second, the long-run supply curve of labor would have a positive slope throughout. This would ensure that there was no involuntary unemployment. However, since literally zero unemployment is inconceivable in the real world, it is customary to identify some minimum level of frictional unemployment with the closest practicable approach to full employment. In the 1950s and 1960s, for example, 3 or 4 percent unemployment was commonly deemed a realistic goal.
If something like the purely competitive long-run equilibrium were really the correct concept, there would be little for neoclassicists and Keynesians to argue about. There is nothing in any version of Keynesian theories of underemployment equilibria to deny that automatic equilibrating forces might eventually bring about a full-employment long- run equilibrium. However, there is no such meeting of minds here. Keynesian analysis is distinctly Marshallian; the relevant timeframe is that in which real-world policy decisions can be designed and carried out-- presumably a relatively short-run period, perhaps extending at most to the medium term, or a series of short-run to medium-term periods. We needn't wait patiently for a utopian future to unfold at its own leisurely pace, because there are things we can do immediately that will speed things up enormously; or things that we could do, if only we could effectively control inflation independently of other economic goals.
In the Marshallian short run, or even the medium term, neither of the two basic conditions of long-run labor-market equilibrium need be met. Obviously, even in the best of times the irreducible minimum number of those involuntarily unemployed would not achieve indifference between work and leisure at the margin. Also, it has long been recognized in the economic literature that the slope of the short-run labor-supply curve need not be positive throughout its length. If a major disturbance suddenly and sharply reduced employment and incomes, for example, newly unemployed workers might be willing to work more instead of less than they were before, and at a lower wage, and additional family members might join the labor force, to maintain their real incomes or just to keep bread on the table. These effects would be reinforced for home-owners and other property owners if, as would be likely, the price of real estate (and perhaps some other assets) fell sharply at the same time, reducing their wealth as well as their incomes.
Typical neoclassical and New Keynesian textbooks [Dornbusch et al., 1993; Baumol et al., 1994; Mankiw, 1994; and Lipsey et al., 1994, for the New Keynesians, and Parkin, 1994, for the neoclassicists] identify "normal" unemployment as the level that would obtain at "full employment", or as the average rate around which the economy fluctuates, and in some cases they also identify it as the non-accelerating-inflation rate of unemployment (NAIRU). However they define "normal" unemployment, they all allow both actual unemployment and actual output to fluctuate around their "full employment" levels. This raises questions about how full their "full employment" really is, and what interpretation is to be put on their "normal" unemployment. Are they identifying "full employment" with a long-run trend line of calculated potential outputs over historical time? That is quite different from the locus of the long-run-equilibrium points (as defined earlier in this section) over historical time, and their "normal" unemployment is quite different from the irreducible minimum of frictional unemployment associated with real-world approximations of those equilibrium points. Long-run equilibrium implies that the resources of the economy are truly fully employed, but a calculated potential output does not--especially when based on data for a period in which some calculations of the "normal" rate of unemployment (or the NAIRU) have approached or entered the double-digit range, as in much of the last 20 years.(4)
Neoclassicists cite the real-balance or wealth effect as a fail-safe device that will rescue the economy from an underemployment equilibrium even if all else fails. Frank [1986, 8] thinks this effect is rather weak at best, and might take decades to be fully effective. He is surely right. In fact, the proposition is highly debatable in its entirety. The idea is that holders of money would become progressively better off as prices fell, with no offsetting losses to anyone else, and would eventually begin to spend. Nowadays, however, money typically consists of claims on some financial intermediary or other, no matter what definition of it one prefers, and its counterpart is the debts of their borrowers; any advantages for the one group are pretty well offset by disadvantages for the other, so the net effect is approximately zero. Certainly, the long period of deflation this scenario envisages is not consistent with the neoclassical assumption of instant equilibrium. Instead, it suggests the possibility of serious defaults by some ultimate borrowers, which might well bankrupt the money-issuing intermediaries and destroy the supposed advantage to their creditors, so *all three* parties would lose.
Turning now from theoretical postulates to practical experience, economic history demonstrates not only that the equilibrating forces postulated in the purely competitive model, or whatever is their real- world equivalent, are pretty sluggish in correcting the recessionary phase of the normal business cycle, but also that periods of depression and unemployment may sometimes be quite substantial and prolonged. The Great Depression of the 1930s was until now the prime example. Automatic restorative forces did show themselves to some extent, in that the decline bottomed out about 1933 in most countries, but they proved unable to cope with the world-wide combination of adverse circumstances that developed between 1926 (in Britain, where an overvalued pound brought years of depressed output) and 1931 (in Continental Europe, where the collapse of the Kreditanstalt set off a series of disturbances that included driving sterling off the gold standard), with the North American contribution coming in between (in 1929, when a worsening economic slowdown triggered the stock market crash and not the other way around, contrary to the conventional wisdom). It was only about two years after the start of World War II in 1939, and the massive injection of new spending this entailed in the belligerent countries and their suppliers, that full employment was achieved in most countries. These are not mere hypotheses, but historically recorded facts, which are not open to doubt. The legal dictum *ipse dixit* applies: the thing itself speaks. Clearly, long-run equilibrium is not quickly or easily attained in the real world.
Now, however, the even longer period from the mid-1970s to date seems to make the point still more strongly: some twenty years of neoclassical economic strategies have failed to achieve anything approaching the measures of full employment that were regularly achieved in most major countries in the first 25 or 30 years after the war. Instead of the promised renewal of prosperity they have brought spreading poverty and despair--and they have not even achieved their first objective, which was the elimination of inflation. At least two economists [Peterson, 1994, 10 and *passim*; McLeod, 1995, 10f] have identified this period as a new *depression* that has already lasted twice as long as The Great Depression of the 1930s and is still counting.
Typically, neoclassicists either assert or implicitly assume that most of today's unemployment is voluntary--if offered a job in their fields, at the going wage or slightly below it, most of the jobless would refuse. The implication is that the real-world rates of unemployment observed in the major industrialized countries nowadays--roughly double the rates regularly achieved in the 1950s and 1960s--are never far from the long-run- equilibrium rate.
The experience of the 1930s is relevant here. There is no question about the facts, merely about their explanation: the existence of substantial and prolonged involuntary unemployment and underemployment on a world-wide scale for a decade or more is incontrovertible. The labor market did not clear in the short run, because neither of the conditions that ensure a full-employment long-run equilibrium in the auction markets of the model of pure competition was satisfied in the real world of monopolistic or imperfect competition and administered prices. Nominal wages fell, slowly but persistently, yet the supply of labor increased instead of decreasing. Each increase in the labor force added to the downward tendency of wages, in a vicious circle. New hirings virtually ceased, or were far outnumbered by new layoffs. Declining wages and net layoffs further reduced consumer spending power and added to the deflationary spiral. To all intents and purposes the marginal wage (both nominal and real) for society as a whole fell to zero.
It would therefore appear that the labor market would not have cleared even if all wages had proven infinitely flexible downwards; the only way the surplus could have been quickly eliminated would have been to allow the excess workers and their dependents to starve to death. The fact that this did not happen, and that a hesitant recovery eventually began, was clearly not due to wage flexibility. On the contrary, evidently what saved the day was the persistent strength of consumer demand, albeit at a sadly deficient level, supported by (a) the stubborn resistance of wages in the face of downward pressures and (b) humanitarian aid to the destitute as the initially rudimentary emergency relief and welfare systems were gradually improved. Between them these two phenomena sustained the economy until the growing need for some capital renewal began to inject a modest positive impetus to the spending stream.
Despite the improved social-security networks put in place in most countries in the i950s and 1960s (which, by the way, are now under attack everywhere), there are indications that conditions in the labor market for the last decade or two differ only in degree from those of the 1930s; the bulk of today's unemployment is involuntary, and slope of the short-run labor-supply curve is once more negative. Here again it should be possible to mount empirical studies that would settle both matters.
Some voluntary unemployment certainly does exist, at least temporarily, in the sense that anyone who finds himself or herself unemployed at any given time may use accumulated past savings, unemployment-insurance compensation, other social security safety-net receipts, or borrowings from friends or relatives for support while taking time to systematically explore whatever employment opportunities may currently be open. However, this does not sound like the voluntary unemployment the neoclassicists describe; far from being at or just below going wage levels, these alternatives are much inferior to a steady job, and most of them are merely temporary expedients.
Many economists find it hard to accept the neoclassical assertions; e.g., Cornwall and Maclean [1984, 18], and Tobin [1992, 398f]. Official statistics report that many people who accept part-time employment would prefer full-time jobs. Many young people accept work in the fast-food industry and in other low-paying fields, although they are clearly qualified for greater responsibilities. These people are not technically unemployed, but they are surely underemployed, and they are certainly not rejecting better jobs at the going rate.
Hasan and de Broucker [1982] found that 21 percent of the weeks of Canadian unemployment in the late 1970s was due to long-term instances (spells of over 6 months), which were concentrated in only 4.9 percent of all spells, and that about half the spells of unemployment ended in withdrawal from the labor force rather than in finding new jobs. Additionally, there was scattered evidence that the incidence of long- term unemployment was greater than these figures indicated; multiple spells of unemployment contributed anything from a quarter to a third of unemployment, so long-term spells accounted for over half the total. On the basis of these findings, Cornwall and Maclean [1984, 63] concluded that Canadian labor markets in those years were characterized by hidden and highly concentrated long-term unemployment.
There is considerable recent anecdotal evidence of involuntary unemployment. Occasional newspaper accounts nowadays report new plants opening (perhaps transferred from relatively high-wage to relatively low- wage locations), and invariably the line-up of applicants for the new positions is several times as large as the number of advertized vacancies, even though the wages offered are by no means exceptional for the region.(5)
Persistent high levels of involuntary unemployment for over 20 years strongly suggest that the slope of the short-run supply curve of labor has been negative, and some observers have cited the persistent shift of women (particularly married women with children) into the labor force over the last couple of decades as persuasive evidence of that phenomenon [Peterson, 87]. Repeated bouts of tight money combined with downsizing in public and private employment offer clear evidence of the sort of repeated new "shocks" that could prolong a basically-short-run phenomenon for decades.
Keynesian demand management in the real world has been emasculated by inflation, not by the intellectual problems some academic theoreticians have found with the underlying model. Policymakers drew back from full- employment policies between the late 1960s and the mid-1970s, because of the practical problems they encountered in trying to maintain them in the face of inflationary pressures; thereafter they replaced them with repressive neoclassical anti-inflation strategies.
From the end of World War II in 1945 to the early 1970s the steadily evolving techniques of demand management gave the world as a whole, and most nations individually, the strongest and most sustained period of economic growth and prosperity ever recorded before or since. However, it turned out that, on balance, very little net stimulus in the form of deliberate deficit financing by governments was necessary in most of the major industrialized countries; generally speaking, much of the government borrowing that did occur in this period was for the perfectly orthodox purpose of financing capital expenditures on projects that were necessary parts of the economic infrastructure, including postwar reconstruction. In the early years, especially in former belligerent countries, the task was made much easier by the fact that the public held substantial liquid savings left over from wartime financing, co existing with pent-up demand for consumer goods that had long been very scarce or unavailable; producers could sell those goods as fast as they could turn them out, so a gentle hand on the demand-management tiller was all that was needed most of the time, to reassure the public that there would be no postwar depression this time. Nevertheless, demand management's developing skills deserve full credit for maintaining sustained prosperity.
One small cloud was visible on the horizon from the very first: low levels of "creeping inflation" persisted even in those countries that had suffered no war damage and that followed what were generally considered responsible domestic economic policies. There is no denying that inflation is a serious problem, for even at relatively low rates it distorts the role of the price mechanism in directing production and consumption in a money-and-market economy [McLeod, 1991], but for a long time creeping inflation remained relatively stable and did not seem to be a serious threat to postwar prosperity.
It had always been clear that using monetary and fiscal policies too aggressively to combat unemployment would make inflation worse. However, it soon become equally clear that using them to combat inflation made unemployment worse, and in fact they proved far more effective in creating unemployment than in curing inflation. Practical policymakers therefore had to resign themselves to seeking the least painful balance between two equally intolerable evils. Unfortunately, by the late 1960s inflation was clearly accelerating everywhere, and had reached levels that the authorities in country after country found they could no longer tolerate;(6) they increasingly eased off on the use of monetary and fiscal policies to support high levels of output and employment.
A worrying feature of the postwar period, which was at the root of the recurring crises in international financial relationships, was the fact that price trends varied materially among the major nations. The discrepancies became more serious as time passed and inflation accelerated; by 1973 they brought the Bretton Woods system of normally fixed exchange rates to an end, and replaced it by a regime of fluctuating rates. Inflation soared everywhere thereafter--it soon reached 32 percent per annum in Japan, 24 percent in Britain, 11 percent in the U.S.A. and Canada, even 7 percent in prudent Germany--because balance-of-payments discipline was greatly relaxed as a consequence of the new exchange-rate regime. Lacking any better advice from professing Keynesians(7), or from anyone else, most national authorities soon adopted monetarist, quasi- monetarist, or neoclassical remedies. The neoclassical era had begun, as sanctioned by the 1976 G-7 summit; in most countries monetary and fiscal policies were thereafter addressed to the anti-inflation fight with increasing severity [McLeod, 1995, 146-151]. However, inflation has proven stubbornly persistent everywhere; it has only recently been reduced to rates that various national authorities consider tolerable, and it threatens to accelerate again if economic growth reaches quite moderate levels. A plausible explanation of its persistence and its tendency to accelerate in good times is suggested at the end of the next section.
Neoclassicists accept continuing inflation as a fact of life for the foreseeable future, and build their policy proposals around continued reliance on repressive monetary and fiscal policies to combat it, but their prescriptions have signally failed to restore prosperity in 20 years of trying. A revival of Keynesian demand management seems to be the best hope for again achieving a reasonable approximation of full employment, but that is impossible unless monetary and fiscal policies can freed be from inflation-control duties; like human agents, they can not serve two masters. We must find a new way of controlling inflation--one that does not make unemployment worse, and that is compatible with the free- enterprise money-and-market system--before those techniques can again be used to support output and employment.
A number of alternatives to monetary and fiscal restraint for controlling inflation have been proposed in the past decade or two. They include incentive-based incomes policies (TIPs and MAPs), extensively examined by Colander et al. [1986]; several other types of incomes policies; and a proposal by Cornwall [1994] explicitly designed to revive Keynesianism and thought to be applicable in full to the U.S.A., Britain, and Canada and in part to the other OECD countries. Most incomes policies apply primarily or solely to wages and salaries, which leaves returns to the ownership of productive resources unfairly free of restraint, though remedies for this deficiency appear to be possible [McLeod, 1994, 182 and 208-10]. My own short-run proposals [McLeod, 1994, Ch. 12; 1995, 156] apply equivalent restraint to both types of factor returns. The prices of goods and services would not be directly controlled, and the profit motive would not be interfered with, but stable factor costs would be reflected in stable consumer prices. Longer-range remedies are also proposed for some of the inflationary biases now to be found in the economic system.
Failing widespread agreement by the economics profession on anything better, a new look at arbitrary price controls might be in order, despite their well-known faults. In view of the evidence that oligopolistic firms and industries can ratchet up their selling prices even in depressed times and in the absence of significant wage pressures, presented at the end of the next section of the paper, some form of price control should at the very least be applied to them. A carefully designed permanent or quasi- permanent program, including continuing or periodic *ad hoc* correctives for the price anomalies that would inevitably occur, would surely be less damaging than the stark choice between uncontrolled inflation and perpetual depression that neoclassicism offers us. The elements of such a system are suggested by Salant in his description [1986] of how U.S. price controls in World War II affected oligopolistic firms and industries, though Salant does not attempt to construct such a system. His essential points are two: (1) setting a price cap below the firm's ideal price (where MR = MC] resulted in increased output at a lower selling price, and (2) when controls were lifted in 1946 these prices did indeed rise, but not to the level they would have attained had there been no controls. A system of permanent controls designed around these findings should damp inflation sufficiently to permit substantial monetary and fiscal stimulus to the spending stream.
It appears that the widespread belief that Keynesian underemployment equilibria require sticky nominal wages and prices may be wide of the mark. Keynes [1936, Ch. 19] argued persuasively that fully flexible wages and prices would not ensure full employment, but he did not claim that rigid wages and prices were necessary to achieve that goal; he merely said that he thought a stable general level of money-wages is the most advisable policy. Some modern Keynesians have reached substantially the same conclusions [Davidson, 1992, 453; Tobin, 1993, 46].
However that may be, a plausible explanation of sticky wages and prices can indeed be constructed on the basis of standard textbook representations of the administered prices of firms operating under conditions of conventional imperfect- or monopolistic-competition theory. It would appear that the onus is on the neoclassicists to show that the model of pure competition offers a better explanation of real-world price phenomena.
The appropriate microeconomic paradigm is not an auction market, as neoclassical theory implies, but an adaptation of the workings of organized commodity and securities exchanges, in which trading occurs in standard board lots, would-be sellers set specific asking prices for their offerings, and would-be buyers set specific bid prices: no trading takes place until some buyer accepts some seller's asking price, or some seller accepts some buyer's bid price [McLeod, 1994, 101-104]. If market conditions prove less favorable than expected, sellers may have to reduce their asking prices sooner or later, but if market conditions prove more favorable than expected they will feel free to raise them promptly. In general, then, wages and prices tend to be sticky downwards but flexible upwards.
In typical labor markets an individual would-be worker faces relatively few potential buyers for his or her particular skills conveniently accessible within reasonable commuting distance, and there are significant costs in time, trouble, and money in seeking out additional possibilities; he or she normally has little choice but to accept the best job offer to be found with a reasonable search. It is the buyers (especially the major employers) who have the greatest market power, and who take the initiative by posting their bid prices. Also, there are additional advantages for the employer in a posted-wage policy; it permits an orderly array of the wage