Looking at the first three periods, note that the standard deviation of the adjusted real wage variable is about the same from 1960 to 1990 as it was in 1900–1929; the fluctuations in both periods were much lower than from 1930 to 1959, when that variable seemed to bounce around more. Note also that the absolute mean value of the adjusted real wage was highest from 1930 to 1959, and lowest from 1900 to 1929. It is not surprising, then, that unemployment was also highest on average from 1930 to 1959, and lowest, on average, from 1900 to 1929. Also the variability in unemployment was greatest when the standard deviation of the adjusted real wage was greatest.
TABLE 3.1 CHANGES IN THE COMPONENTS OF THE ADJUSTED REAL WAGE, 1900–1989
Although the standard deviation of the adjusted real wage variable is a measure of variation in the adjusted real wage over each time period examined (usually thirty years), short-run fluctuations in that variable are better measured by the standard deviation of annual percent changes. On that score, fluctuations in the adjusted real wage were markedly lower in the 1960–90 period than in the earlier eras.
Not only has the instability in the adjusted real wage declined since the era of the Great Depression, but there has been a pronounced diminution in the fluctuations in the components making up that wage. Productivity, price, and money-wage changes tended to vary far less from year to year in the latter decades than early in the century. Even unadjusted real-wage growth tended to bounce around less in the more recent decades.
The evidence from table 3.1 seems to be consistent with what some economists view as a decline in the downward flexibility of wages over time. According to this view, markets cannot function to eliminate unemployment because of wage rigidity. It is true, for example, that in nine of the first forty years observed (1900-1939), money wages actually fell, whereas in none of the past forty years have money wages risen less than 3 percent.
From the standpoint of real wages, however, there is a different pattern. In ten of the first forty years in the century, they fell. Yet, in six of the past twenty years (1970-89), real wages also have declined, a slightly higher proportion of years. Early in the century, falling real wages were brought about in some years by downward wage adjustments, in other years by upward price adjustments, and in some years by both factors; late in the century, however, decreases in real wages have come about exclusively through upward adjustments in prices.
Put differently, early in the century, private decision-making in the private sector in response to market conditions seemed to initiate downward real-wage adjustments in several years; in modern times, downward real-wage adjustments seem to have resulted from price increases that may be largely initiated by macroeconomic policy. Also, in the early part of the century, downward real-wage adjustments may have come quicker than in recent years. For example, real wages fell some 11 percent between 1979 and 1989—but in no single year did they fall as much as 2.7 percent. Real wages were quite flexible downward—but it took time for the full adjustment to occur. By contrast, on eight different occasions before America’s entry into World War II, real wages fell more than 3 percent in a single year.
It may be, however, that the observed decline in the variability of wages has been exaggerated by biases arising from splicing together sets of wage data for different periods computed using different methodologies. Steven Allen has replicated methods used to calculate wages in the prewar era for postwar data, concluding that money-wage volatility has not fallen—the observed decline is a statistically created illusion.12
Nonetheless, the prevailing wisdom still is that real-wage flexibility has diminished. The reasons for this will be explored in greater detail in coming chapters. We would observe here that two constraints on wage flexibility are found today that did not exist to a major extent in the first decades of this century. First, in the earlier years, relatively few employees worked under contracts. The growth in collective bargaining brought about more wage rigidity. Workers usually work under multiyear contracts; cost-of-living adjustment clauses in some contracts make it difficult to reduce real wages in the short run. Also, the growth in the relative importance of government employment has increased the “wage inflexibility” problem since government compensation has historically been notoriously slow to respond to changing market conditions.
A second problem has been the growth in direct government intervention in labor markets. With the Fair Labor Standards Act of 1938, the federal government imposed minimum-wage requirements that have materially altered the market for relatively unskilled labor. Other legislation (e.g., the 1931 Davis-Bacon Act and similar state laws) even impacted on compensation for skilled labor. More recently, civil-rights legislation and perhaps even the Medicare program have affected the ability of employers to pay market-determined wages. Further substantial effects on employee compensation patterns have come from governmentally mandated fringe benefits, such as the social security program.
The decline in wage flexibility has been interpreted by some as a sign that market adjustments will be inadequate to deal with substantial unemployment, and that governmental macroeconomic intervention is desirable. We are skeptical of that conclusion for at least three reasons. First, if indeed wages are less flexible than previously, the solution to that is to restore wage flexibility by removing obstacles to that flexibility, not to consider self-correcting market adjustments as an inadequate policy tool, particularly given the strong evidence in (4) that unemployment is closely associated with movements in the adjusted real wage. Second, there is some question as to the degree to which relative wage-price flexibility has truly declined, given the nature of some of the data available. Errors in early data collection and presentation may well serve to distort early wage patterns. More recently, the consumer price index has come under a good deal of criticism for overstating the inflation of the 1970s and early 1980s. Third and perhaps most important, it appears that a reduction in volatility in the adjusted real wage and its components has contributed to reduced unemployment volatility. If wages are somewhat less flexible on the down side, they also have tended to increase in a predictable and not very volatile fashion in recent years. There has not been a single increase in real wages in excess of 4 percent in any of the past thirty years, for example, a factor contributing to the relatively few observed surges in unemployment in that era.
Criticisms of the Model
The model presented in (4) above is very unfashionable these days. To most contemporary mainstream economists, it is naively low tech, a simple single equation. Allegedly, far more complex models are capable of dealing with the material of the simple model while also offering more profound insight. The model here is appropriate for 1950 or possibly even 1970, but not 1990 or beyond, since econometric knowledge has expanded in the past generation. Also, it focuses directly on the labor market, whereas most economists of the mid- and late twentieth centuries have tended to explain unemployment indirectly by looking at other markets, in particular money markets. The model does not talk about consumption or investment or the stock of money or interest rates or any of the other things that economists usually bring up in discussing unemployment.
The approach here is even subject to criticism by the one group of economists that in general solidly agrees with the conclusions, the Austrian school. Austrian economics, which has had a healthy resurgence in recent years, tends to be highly skeptical of econometric and mathematical work.13 The Austrians tend to reject the modern scientific perspective that there is no a priori knowledge. Also, they question the aggregation used in constructing data series. For example, there are very serious conceptual problems in the construction of price indices, not to mention practical difficulties. Thus the approach used by us is criticized by one group of economists as being insufficiently empirical and by another group as being excessively empirical and quantitative.
Regarding the mainstream criticism, it is our goal to communicate our ideas to an audience of intelligent laypersons as well as to professional economists, and highly complex econometric models are beyond the understanding of much of our audience. Great economists from Smith to Keynes wrote books that the intelligent layperson