Does Collective action by workers always increase unemployment?

Thinking out-loud

The answer to this question is almost a unanimously agreed upon yes across paradigms even within formalized Marxian models.  In the classic Marx-Goodwin formulation successful collective action by workers only serves to increase the long-run rate of unemployment: some workers maybe able to raise their wages but this will be balanced out in the aggregate by increased unemployment.  However, if as Shaikh (2003) points out the strength of labour endogenously influences the rate of technical change then workers can increase their wages in the aggregate (wage share) without causing an increase in long term rate of unemployment.

Shaikh sums up the case thus:

6. Summary and Conclusions

This paper has attempted to analyze the manner in which alternative macroeconomic frameworks portray the dynamics of the labor market. Two types of dynamics have been of interest, both of which depend upon the mutual interactions between the wage share and the employment rate. In disequilibrium dynamics, the issue is the manner in which these variables respond to imbalances in the labor market, while in growth dynamics the issue is their response to technical change and growth in labor supply growth. We examined the basic neoclassical, Keynesian, Harrodian and Marx-Goodwin models, since each embodies a particular approach to macroeconomics

Dynamics require explicit analysis of stability of various equilibria. But even the existence of a particular stable equilibrium need not imply that the economy will be at or even near that point. The analysis of the neoclassical model in Section 2 demonstrates that if real wages respond to the current excess demand for labor, then the labor market converges to a particular wage at full employment (Figure 1). But if real wages respond to the cumulative excess demand for labor, then the labor market would exhibit endless and possibly large fluctuations in real wages and excess labor demand, around but not at, the equilibrium real wage and full employment (Figure 2). This second type of response is reminiscent of Goodwin’s elegant representation of Marx’s argument about the reserve army of labor, except that in his model the center of gravity is a persistent level of unemployment, not full employment (Section 5). In any case, this type of disequilibrium dynamic remind us that we should be careful to distinguish between equilibrating paths and equilibrium points. At an empirical level, this cautions us not to confuse observed variables with their putative equilibrium levels.

In the case of growth dynamics, a second type of finding emerges. It turns out that in each of the four macroeconomic approaches, the paradigmatic case is one in which the organizational or institutional strength of labor has no influence whatsoever on the path of real wages and on the level of the wage share. In all of the approaches, it is technical factors and labor supply growth which determine the standard of living of workers. The degree of labor strength in the struggle over wages has no effect at all. In the neoclassical case, this is instanced by the ubiquitous Cobb-Douglas production function, in which the labor elasticity parameter directly determines the wage share. Hence the profit-wage ratio is entirely determined by production conditions. In the standard Keynesian case, the corresponding outcome arises from mark-up pricing, in which changes in money wages are said to cause equiproportional price changes. This not only leaves the real wage unchanged, but also implies that it is unchangeable. In the Harrodian framework, unemployment affects the wage share, which in turn affects the warranted rate of growth via the dependence of the savings rate on the wage share, a la Kaldor and Pasinetti. This feedback loop leads the system to stabilize around full employment in the long term. But it also implies that the wage share is completely determined by the rates of technical change and population growth, completely independently of labor strength. Finally, even in Goodwin’s classic formalization of Marx’s theory of the reserve army of labor, “class struggle” over wages has no effect whatsoever on the rate of surplus value. Indeed, greater labor strength would only serve to increase the long-run equilibrium rate of unemployment. This is a particularly unkind cut for a Marxian model.

Two critical questions are raised by the general theoretical finding that wage shares are independent of labor strength. First of all, it is at all empirically plausible? The stability of wage shares is a well-known “stylized fact.” But then so too are differences between wage shares across nations and across levels of development. Are these differences reducible to those arising solely from technical factors and conditions of labor supply?

Alternately, if social forces do indeed influence the wage share, how might such a mechanism operate? The key expression to consider is equation 15, in which the rate of change of the employment ratio depends solely on two critical variables: the rate of accumulation g_K = s(u)R and the rate of mechanization g_k, assuming that the rate of growth of the labor supply g_n is exogenous.

v‘/v = g_K – (g_k + g_n) = s(u)R – (g_k + g_n)                                            (15)

We saw that if the output-capital ratio R and the mechanization rate g_k are exogenously given, then there is only one wage share u = u* consistent with a stable employment rate (i.e. with v‘/v = 0). But this conclusion would not be altered if R and g_k, and indeed even g_n , were to also depend on the wage share.²¹ What is needed, therefore, is some other mode of feedback between the employment rate and one of these variables. A particularly simple one is to suppose that the rate of mechanization depends not only on the wage share (i.e. indirectly on the employment rate through its effect on the relative cost of labor) but also directly on the employment rate (i.e. directly on the relative availability of labor). Rowthorn (1984, pp. 203-205) notes that this is precisely the argument in Marx.²² Then g_k = f(u,v), and

v‘/v = g_K – (g_k + g_n) = s(u)R – [g_k(u,v)+ g_n]                                      (15a)

The results of this apparently minor extension are dramatic. Suppose we consider the extreme case in which the wage share is now entirely determined by “class struggle,” so that u = u₀. Then if v‘/v > 0 initially, the employment rate v will rise, which will raise the mechanization rate g_k(u₀, v), thereby bringing the employment rate back into balance. It follows that the same result would also obtain if we assume that the wage share depends on both “class struggle” and the employment rate. Thus the preceding simple modification completely reverses the general theoretical conclusion that the wage share is independent of labor strength, for now there is plenty of room for the influence of the relative strength of labor.

However, if the mechanism is technical change–the swapping of machinery for living labour–it strikes me that over the short run the Marxist model would have to say that unemployment must initially increase unemployment.  This is consistent with Marx’s observation that capital can always adjust to existing supply shortages through technological / organizational innovation. Surely this process takes time.  As Marx was well aware.   I think Marx’s position was that even in cases where capitalism managed to produce full employment that such a condition would not prevail for long do to the tendency to substitute away from labour.

Moreover, what about the case of successful collective action by workers in the face of not labour market shortages but rather excess supply aka unemployment?  It seems the inescapable answer is that such action would serve to increase the short term rate of unemployment even if over the medium to long term there were not any adverse effects.

This result undoubtedly sharpens the political question for unions during recessions.