Productivity Growth and Economic Performance

Empirical economic laws, viewed as statistical relationships between economic variables that are robust with respect to different data sets and time periods, have often fascinated economists. This may be because they suggest that economics is a science, with a methodology analogous to physics, although without the benefit of controlled experiments. However, these laws have also given rise to lively debates concerning both their empirical validity and their economic interpretation — consider, for example, the original Phillips curve. ‘Verdoorn’s Law’ is no exception to this.

One of the difficulties in long-term planning is to estimate the future level of labour productivity. Unless this is known, one does not know the relation between output and employment.

‘Fattori che regolano to sviluppo della produttività del lavoro’ (henceforth simply referred to as ‘Fattori’) is an article which, although much quoted, has been, perhaps, seldom read. The fact that it was originally written in Italian certainly did not help it to achieve widespread recognition.¹ Indeed, over time, the increasing fame of this paper stands in marked contrast to the lack of knowledge generally available about the author’s personal life.

It is somewhat ironical that without Kaldor’s celebrated inaugural lecture of 1966 the widespread use of the term ‘Verdoorn’s Law’ (Verdoorn, 1949) to describe the relationship between productivity and output growth may never have come to pass. Moreover, Verdoorn himself made no further major contribution after his seminal 1949 paper to the extensive literature that has developed concerning the law. Indeed, the main impetus for the subsequent revival of interest in the law ironically may be traced back to Rowthorn’s (1975a) critique of Kaldor’s specification of the law. Verdoorn’s (1980) only other notable article was to reinterpret the law within a neoclassical framework and simultaneously to distance himself from it.^2,3

Verdoorn’s Law postulates the existence of a significant positive relationship between the growth rate of labour productivity and that of output, at least in manufacturing. The relationship is generally interpreted to be of a technological nature, reflecting the existence of both static and (mainly) dynamic economies of scale and thus the presence of increasing returns to scale.

In his inaugural Cambridge lecture, Kaldor (1966) refers to what he terms Verdoorn’s Law — the statistical relationship between the rate of growth of labour productivity and the rate of growth of output — as evidence of the pervasive existence in industrial economies of static and dynamic economies of scale. Since this contribution, it has often been suggested that attempts at estimating the law (including, of course, Kaldor’s own one) suffer from serious specification problems. As is well expressed by McCombie and Thirlwall (1994, p. 167), ‘the debate over the Verdoorn Law would make a good textbook example of the problems that can beset statistical inference!’ As can be seen from the surveys in Bairam (1987a) and in McCombie and Thirlwall (1994, ch. 2), problems with estimating the law are related to three major issues.

More than fifty years on, Verdoorn’s Law remains significant because it enters as the key ‘stylised fact’ in the Dixon-Thirlwall (1975) model, a model that has come to be regarded as the ‘standard’ model of cumulative causation (see McCombie, 2002, this volume). Indeed, it is precisely because of Verdoorn’s Law that growth in this model is cumulative. This is because without it there would be no tendency for an initial growth advantage in the model to (positively) feed back into labour productivity growth, which is a necessary prerequisite in the model if an initial growth advantage is to be retained (Dixon and Thirlwall, 1975, pp. 205–6). However, the emphasis in this chapter will not be on the vital role that Verdoorn’s Law plays within the Dixon-Thirlwall (1975) model, for this is both obvious and well-known. Rather, we first concentrate on relaxing the model’s implicit assumption that workers are passive to the implications of firms’ pricing decisions for their real wages. To achieve this relaxation we draw on the NAIRU literature,² a literature in which workers only accept the implications of firms’ pricing decisions for their real wages if they are consistent with their own aims in the wage bargaining process.³

The economic growth debate of the 1980s and 1990s has opened the door to a greater range of ideas about why and how growth rates differ across countries and regions. The main achievement from a theoretical viewpoint is the construction of models that allow growth rates to be positive in the steady state without the help of any exogenous variable. Growth can be positive in the long run and depends on the investment decisions of individual economic agents. This may seem obvious to the amateur economic growth practitioner. However, this intuitive idea clashes with the complications and constraints that both real data and mathematical models impose for the theoretician of economic growth. How can we explain the continuous growth of output without the generation of an explosion in per capita income that cannot be observed in the data? What factors lie behind this possibility? ‘New’ growth theory has indeed contributed to specifying growth models in which both questions are addressed. Technical progress, either disembodied or embodied in capital goods, has been placed at the centre of the analysis. National specific non-exogenous factors may now explain why some countries have been more successful than others. Even more, a world of diverging per capita incomes has been more plausibly explained by this more recent analysis.

The issues that I discuss in this chapter, and that appear in the title, might be viewed at being first as independent of one another. In particular, the first one (output measurement) may seem to be quite untheoretical; and the second one might be expected to contribute little to a school of thought, namely the Kaldorian one, which has already, over many decades, stressed the importance of the demand side of the economy. However, I hope to show that coupling some observations drawn from both parts may help in developing a viewpoint about the Verdoorn effect which, although somewhat unconventional, might be useful in interpreting some actual productivity trends that have been observed in the advanced industrial economies.

This chapter tells a story which has an apparently unfortunate ending. The story concerns a basic, although rather neglected, aspect of Verdoorn’s Law — namely its compatibility with steady-state growth. The story appears to be unfortunate, since the attempts to marry Verdoorn’s Law to the analysis of steady-state growth produce inconsistencies and weaknesses. This fact seems to prevent Verdoorn’s Law from playing a satisfactory role in any long-run analysis. However, as this chapter will briefly show, Verdoorn’s Law can still have a useful role in long-run analysis if its role in economic development is explained and made endogenous.