Verdoorn’s law and productivity dynamics: An empirical investigation into the demand and supply approaches

ABSTRACT

According to Verdoorn’s law, productivity growth is endogenous to output growth, due to the existence of increasing returns to scale, broadly defined. Such an idea is at the root of both the endogenous growth theory and the Kaldorian approach. While in Kaldor’s view, a country’s growth is demand-driven, in the endogenous growth theory, growth is determined by the growth of the factors of production and hence growth is supply-constrained. This article empirically tests both assumptions for Verdoorn’s law by using a dynamic panel of manufacturing industries for seventy countries at different stages of development for the years between 1963 and 2009. In order to distinguish between these approaches, two different specifications are estimated where the growth of output and the supply of factors of production are instrumentalized by system generalized method of moments (GMM)estimators. The results show that, if it is assumed that the growth rates of countries are demand-driven, a faster growth of output increases productivity growth due to the existence of increasing returns. Alternatively, if it is assumed that output growth is driven by the growth of the supply of the factors of production, it is not possible to conclude that productivity growth is induced by output growth.

KEYWORDS:

• Demand-led growth
• increasing returns
• manufacturing
• productivity growth
• Verdoorn’s law

JEL CLASSIFICATIONS:

• E12
• O33
• O47

Notes

¹Some authors consider that this relationship could result if increases in productivity growth reduce the growth of relative prices, which in turn might increase the growth rate of demand for particular industries (Salter, 1966). Nevertheless, according to Kaldor, in this view, productivity is mainly to be explained by autonomous technical progress, but if this is true, how can we explain large differences in productivity growth across different countries at the same level of development in similar industries for such long periods? Hence, Kaldor argues that this relation can only have one significant direction of causality: from output to productivity growth.

²Kaldor (1966) estimated Verdoorn’s equation using data for the advanced countries over the period 1953/54 to 1963/64, and found a coefficient of 0.484 for the elasticity of manufacturing productivity to output, which is very similar to the value found by Verdoorn.

³Kaldor (1975) suggested, alternatively, to regress employment, rather than productivity, growth on output growth. As productivity growth is defined as the growth of output minus employment growth, there is an element of spurious correlation engendered by having output growth on both sides of the regression of the Verdoorn law. Nevertheless, regressing employment on output growth does not affect the value of the derived Verdoorn coefficient, which is the same whichever of these two specifications is estimated.

⁴See McCombie (2002, pp. 95–96) for a detailed discussion about simultaneous equation bias in both estimations.

⁵Under the assumption of constant growth of the capital–output ratio for the long term, Verdoorn’s law can be reduced to Equation (1). However, it is not assumed here because short-term variations in the growth of the capital–output ratio can affect the value of Verdoorn’s coefficient.

⁶McCombie (2002) presents a detailed discussion about the problems and advantages of this theoretical approach.

⁷There are many explanations for the existence of increasing returns to scale. Krugman (1991, 1998), for example, emphasizes the effects of external economies of scale. According to him, the geographical clustering of activities can result in localization economies. Clustering facilitates research and innovation in an industry, as well as the exchange of ideas and knowledge between firms. Furthermore, there is also the importance of Marshallian sources of external economies, such as market-size effects (backward and forward linkages), a thick local labor market, especially for specialized skills, and information spillovers.

⁸Many studies, such as Angeriz et al. (2008, 2009), estimate the Verdoorn coefficient regressing total factor productivity growth on output growth. The advantage of this approach is that there is no need to assume that the capital–output ratio is exogenous to output growth to obtain the Verdoorn coefficient, such as assumed in this work. Nevertheless, to obtain the total factor productivity it is necessary to assume a value a priori for a, and (1 – a), which, following the neoclassical tradition, is usually obtained as the share of wages and profits, respectively, in total income.

⁹Harris and Liu (1999), for example, use annual time-series data for GDP for a number of individual countries, but the data (and hence the estimates) are subject to the serious measurement errors discussed in the text. Most studies of the Verdoorn law use long-term average growth rates and very few use annual or quarterly data.

¹⁰Hypothesis tests comparing these results were constructed based on the estimated coefficient and standard errors presented in Table 1. The t-statistic comparing capital goods and natural resources for Equations (19) to (22) are, respectively, 0.91, 0.79, 0.82, and 0.79, indicating that the null hypothesis that they present different Verdoorn coefficients cannot be rejected at the 10 percent level.

¹¹The t-statistic comparing high-tech and low-tech products for Equations (19) to (22) are, respectively, 0.21, 0.06, 0.13, and 0.07.

¹²The authors found higher increasing returns for all subsectors than for manufacturing, with the only exception being Textiles.

Additional information

Funding

This work was supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior [Grant Number BEX 9442/11-1].

Notes on contributors

Guilherme R. Magacho

Guilherme R. Magacho is PhD in Land Economy, University of Cambridge.

John S. L. McCombie

John S. L. McCombie is a Fellow in Economics, Downing College and director of the Cambridge Centre for Economic and Public Policy, Department of Land Economy, University of Cambridge.