– From the introduction to Capitalism: Competition, Conflict, Crises –
Chapter 14 derives the crucial linkages between unemployment, wages, profitability, and growth. The key conclusion is that competition under flexible real wages creates a sustained rate of involuntary unemployment. This is in sharp contrast to neoclassical theory in which flex-wage competition necessarily leads to full employment, and to Keynesian and post-Keynesian theory in which competition may or may not give rise to unemployment. Goodwin formalized Marx’s argument that competition creates a persistent pool (Reserve Army) of unemployed labor and set the stage for modern heterodox approaches. A striking implication of both orthodox and heterodox approaches is that workers have no say in their own standard of living: in neoclassical theory the real wage is determined by the full employment condition; in post-Keynesian theory it is determined by productivity and the monopoly markup set by firms; and in Kaldor and Pasinetti’s extension of Harrod, it is determined by productivity and the requirements for full employment. Even in Goodwin’s formalization of Marx the real wage is determined by productivity and the requirement for the normal rate of unemployment (section II). But once it is recognized that labor force and productivity growth may themselves respond to accumulation through increases in labor force participation and/or immigration rates and through accelerated technical change (section III), then there is full room for the effects of labor struggles on the real wage and wage share.
Keynes who based himself on competitive markets specifically cites the role of wage bargains and labor struggles in determining money wages. He conceded that prolonged unemployment would erode real wages, but argued that in periods of high unemployment state intervention was preferable to a slow and socially devastating erosion of the living standards of workers. Keynes’s views are consistent with the classical theory of real competition. In Kalecki’s theory, the labor share in net output is entirely determined by the monopoly markups set by their employers. As previously noted, Kalecki struggled to incorporate some degree of worker agency into his story. These and other views in the post-Keynesian and post-Goodwin traditions are analyzed in considerable detail in the text.
Section III constructs a framework in which labor struggles play a significant role in determining the real wage and normal-capacity accumulation maintains a persistent pool of unemployed labor. Shop-floor conflicts between labor and capital bring about a particular division of the money value-added in each firm. At an aggregate level this translates into a real wage linked to labor productivity through a term reflecting the average bargaining strength of labor. The labor strength term itself rises when unemployment falls below some critical level and falls in the opposite case. This implies that the rate of change of real wages relative to productivity (i.e., the rate of change of the wage share) is a negative function of the unemployment rate. I call this the classical Curve. It is one of the two basic relations in the Goodwin model and can be shown to imply the aggregate log-linear “wage-curve” estimated empirically by Blanchflower and Oswald and many others. The unemployment rate in turn depends on the levels output, productivity, and the labor force. The rate of growth of output was previously shown in chapter 13 to depend on normal net profitability (which drives accumulation) and a driving term reflecting various factors including private, public, and foreign injections of new purchasing power. Productivity and labor force growth are assumed to respond to unit labor costs (the wage share) because a rise in the latter provides a strong incentive for firms to raise productivity and to increase the labor force by importing workers and/or raising the participation rate. The mutual adjustment between output and productivity growth creates a correlation known as Verdoon’s Law.
The classical dynamical system yields a growing economy with a normal rate of unemployment and a stable wage share (β) which reflects the social-historical strength of labor. When the stable wage share is combined with the national income identity that net output per worker (y) equals the sum of the real wage and real profit per worker, the result is the relation y_t = A_t·k_t ^ (1–β), which looks just like an aggregate Cobb–Douglas production function even though it is explicitly derived from a labor-struggle theory of the real wage. Since growth is endogenous to the classical dynamic even a temporary rise in the growth of aggregate demand arising from state deficits, export booms, or from an acceleration in investment spending due to higher animal spirits, will permanently raise the levels of the growth paths of output, employment, productivity, and the real wage without affecting the wage share, the profit rate, or the rate of growth. On the other hand, persistent demand growth at a rate above that induced by net profitability will lead to a persistently higher wage share (and hence to a lower normal profit rate). Yet the growth rate would also be raised because the negative effect of lowered profitability is offset by a rising stimulus until some limits come into play (see the next section). A striking feature of the classical model is that the long-run wage share depends positively on the initial values of the wage share and unemployment rate, and negatively on the initial values of productivity and labor force growth. Hence, local actions that raise the existing wage share or employment rate will raise the long-term wage share, while local actions that raise productivity or labor force growth will lower the long-term wage share. Workers and employers are therefore justified in thinking that local actions do matter even in the long run. However, none of these will affect the equilibrium employment rate.
Section IV examines the further implications of a normal rate of involuntary unemployment. Pumping up aggregate demand can increase employment and output growth, but will not permanently eliminate unemployment because there are internal mechanisms that restore the normal rate. Therefore, it would take an increasing stimulus to maintain an unemployment rate below the normal rate. Even so, inflation is not an automatic outcome (chapter 15). On the other hand, the normal rate of unemployment can itself be lowered if the balance of power shifts against labor. Section V takes up the relation of the classical curve to various types of Phillips curves. Phillips’s original question was about the effect of unemployment on wages. His own answer was posed in terms of the rate of change of money wages, very much in keeping with a Keynesian money-wage perspective. Friedman and Phelps argued that workers’ struggle for a standard of living (i.e., for a real wage, not a money wage), so that the correct “Phillips-type” relation should be in terms of (expected) real wages. The classical argument is that real wage struggle is conducted in relation to the general level of development (productivity), in which case the appropriate “Phillips-type” relation will be in terms of the rate of change of nominal wages relative to inflation and productivity growth. Given a stable classical curve, a stable real wage exists only if productivity growth is roughly constant, and a money wage curve exists only if inflation is also roughly constant.
Section VI presents the empirical evidence. As expected, from 1948 to 2011 the US wage share rose and fell broadly in line the growth rate of nominal output (a proxy for aggregate demand). The unemployment rate roughly doubled over this interval, and the unemployment duration quadrupled. The unemployment intensity, which is their product and as such a much better measure of the pressure on wage changes, rose to ten times its original value. As in the theoretical classical system, the actual wage share and the unemployment intensity trace out a clockwise three-dimensional spiral over time. Most importantly, a scatter diagram of the rate of change of the wage share versus unemployment intensity clearly displays a negative slope. Phillips’s original curve was based on cyclically adjusted data points in order to identify the underlying structural relations. I use the Hodrick–Prescott (HP) filter for the same purpose. The dramatic result is a stable classical curve from 1949 to 1982 which then shifts down in the face of subsequent neoliberal attacks on unionization and labor-support mechanisms. The new lower curve in turn reduces the critical unemployment intensity at which the wage share is stable. Also clearly visible are movements up the curve as the economy is pumped up during the Vietnam and dot.com booms and down the curve as the stimuli peter out. All such movements are fairly slow, as Keynes argued long ago. Finally, the lack of stable real or money wage Phillips curves is easily explained by the fact that inflation rose and productivity growth fell dramatically precisely in the era when Keynesian economics had to retreat from the price Phillips curve. Had Phillips answered his own question in classical rather than Keynesian terms (i.e., through a wage-share relation rather than a nominal-wage one), it might have been possible to avoid the theoretical crisis about the price “Phillips curve” during the Stagflation era of the 1970s and 1980s. Keynesian theory would still have required an explanation of inflation, and even if it had retained a markup theory of inflation based on nominal wage, the shifts in the underlying nominal wage curve would have been entirely comprehensible. Of course, the political attack aimed at weakening labor and raising the profit share might well have won the day in any case.