in six or more years were the 1900s and the 1980s.
The results suggest that some of the impact of changing wages, prices, and productivity on unemployment comes long after those variables change, although there is a short-run impact as well. Table 3.4 indicates the proportion of the ultimate effect of changing money wages, prices, and productivity that is observed immediately, after one year, two years, etc. Starting with money-wage changes, the findings suggest that in a year when money wages rise by 1 percent, unemployment rises by 0.16 percent as a consequence. Four years later, this year’s 1 percent wage increase will have an impact on unemployment in that year of 0.64 percentage point. Thus only about 25 percent of the effect of the wage increase is what might be termed “short-run” or immediate, with the remainder coming over time. Note the full (maximum) impact is reached after four years—after that the lagged wage-unemployment relation weakens somewhat.
With productivity change, however, more of the impact comes immediately. A 1 percent productivity increase per worker this year will lower unemployment by 0.48 percentage points. The maximum cumulative impact is reached after three years, 0.77 percentage points. Thus over 60 percent of the productivity-change impact is felt in the year that it occurs. With price changes, the findings are similar. The immediate impact of a 1 percent increase in prices is a fall in the unemployment rate by 0.48 percentage points; the maximum long-term impact, after six years, is 0.67 percentage points, suggesting that about 70 percent of the impact comes immediately.
With all three components of the adjusted real wage, the impact of a change is mostly (85 percent or more of the maximum observed impact) seen within two years. The long-term effect as measured by the maximum regression coefficient runs from a .64 percentage point impact for a 1 percent change in money wages to a .77 percentage point movement for a 1 percent change in labor productivity. The three variables are approximately equal in potency, with the long-term impact somewhat greater than indicated in (4) above, especially for money-wage changes.
ISSUES OF CAUSALITY
Just demonstrating that a statistical relationship exists between two or more variables does not, of course, prove causality. Going back to our simplest expression of the neoclassical/Austrian perspective in (1), it is at least theoretically possible that instead of rising adjusted real wages causing higher unemployment, higher unemployment causes higher adjusted real wages. Yet such a conclusion is implausible. One might dismiss that possibility on the basis of economic theory or logic. It makes sense that higher wages would price workers out of the market causing increased unemployment, but makes no sense that higher unemployment would cause wages to rise (if anything, higher unemployment might induce wage-cutting.)
However, critics of the model above would argue that only the proximate causes of unemployment are examined, not the underlying factors. For example, the unemployment-price change relationship observed in (4) and table 3.3 above ignores the causes of price changes. Similarly, the underlying causes of changes in the money-wage variable are not made explicit. These criticisms have some validity, and as the discussion unfolds in the next several chapters we will look at some of the deeper causes of changing prices and wages.
The most criticism, however, relates to productivity change. Keynesians would argue that productivity change is highly procyclical, responding to fluctuations in aggregate demand. For example, they might make the point that if aggregate demand declines, businesses face a decline in the demand for labor (the demand curve in figure 2.1 shifts downward and to the left). At least initially, businesses do not reduce staff proportionally with reduced production, leading to a decline in output per worker. If true, changes in the adjusted real wage are at least partly determined by Keynesian-style aggregate-demand shifts, making the statistical results cited above far less unambiguously supportive of Austrian or neoclassical perspectives on the determinants of unemployment.
It is true that there is a positive statistical association in the twentieth century between short-term fluctuations in economic activity and changes in the productivity of labor. Running a regression between the percent annual change in labor productivity and the percent growth in real GNP (the best measure of economic activity), we find that for the ninety-year period, about one-third of the annual variation in labor productivity growth is explainable by fluctuations in the level of economic activity.21 Moreover, the observed relationship is highly significant in a statistical sense. Over some subsets of the century, the proportion of productivity variation explained by changing real GNP is slightly higher, although never as much as half.
We could call productivity change induced by cyclical shifts in aggregate demand “Keynesian productivity change.” Yet a large majority (about two-thirds) of the observed variation in labor productivity is not of this nature. Thus, the claim that the productivity variable’s behavior is largely determined by shifts in aggregate demand seems questionable.
Moreover, in any given year, the growth in real GNP reflects not only cyclical forces (such as changing aggregate demand), but also the longterm growth in real output, influenced by such things as the formation of productive inputs, especially capital, and the resultant increase in the capital-labor ratio. This is the type of productivity advance talked about by Adam Smith in The Wealth of Nations, and can be termed “Smithian productivity change.”22 Moreover, even with respect to cyclical fluctuations, it may be true that spurts and lapses in technological progress and innovation may themselves cause business cycles and also explain variations in productivity growth over time: this was the argument of Joseph Schumpeter, and consequently, we might speak of “Schumpeterian productivity change.”23 In short, the cyclical component of real-output change may be caused by multiple factors.
One final test of the role of productivity change was performed. Keynesians have often argued that shifts in human behavior regarding one of the key components of aggregate demand have been a decisive factor in explaining cyclical activity. For example, the Great Depression has been explained in large part by downward shifts in the investment and/or consumption components of aggregate demand.24 Out of any given income in 1930, people spent less than what historical experience would have predicted, thus triggering a depression (helped by a “multiplier effect”). In the lingo of economists, “autonomous consumption fell.” Similar demand shifts have allegedly explained both the onset of the World War II boom (increase in autonomous government spending) and the lack of a depression at the end of the war (increased autonomous consumption and investment).25
The relevant question here is, to what extent have shifts in one of the key components of aggregate demand impacted on labor productivity, the adjusted real wage, and thus unemployment? The proposition that fluctuations in labor productivity are partly induced by Keynesian-style shifts in one of the components of aggregate demand is subject to empirical analysis. We performed a two-stage regression procedure. First, we identified shifts in autonomous consumption or investment spending that potentially could cause a shift in aggregate demand. This involved, in the case of consumption, estimating a consumption function where consumption is related to disposable income.26 Similar investment and government-spending functions were estimated. The discrepancy between actual spending (say, for consumption) and predicted spending is an indication of the extent that spending in a given year deviated from the normal long-run trend consistent with that income or output level. It provides a means of measuring changes in autonomous consumption (or other components of spending). Second, we related the year-to-year changes in autonomous spending to observed productivity change using a simple bivariate regression. In every case (consumption spending, gross private domestic investment spending, government purchases of goods and services, and net exports), we found no statistically significant relationship, even at the 10 percent level using a one-tailed test. Indeed, a majority of the results had a negative sign, not the relationship
