The endogeneity of the natural rate of growth – an empirical study for Latin‐American countries

Abstract

The aim of this paper is to analyse the sensitivity of the natural rate of growth to the actual rate of growth for a sample of 11 Latin‐American countries, assuming the natural rate to be determined endogenously by changes in the actual rate of growth. The natural rates of growth are estimated in a system of SUR estimations over the period 1986–2003. In order to determine whether they react endogenously to changes in the actual rate of growth, a dummy variable for boom periods is added to the system of regressions. The results confirm the hypothesis about the endogeneity of the natural rate of growth.

Keywords:

• natural rate of growth
• actual rate of growth
• endogeneity
• Latin America
• seemingly unrelated regressions

JEL classifications:

• O40
• E10
• C23

Acknowledgements

This paper was presented at the 10th Workshop of the Research Network ‘Alternative Macroeconomic Policies’ in Berlin. The author would like to thank Ulrich Fritsche, Eckhard Hein, Thorsten Drautzburg, the participants of the above‐mentioned conference and two anonymous referees for very helpful remarks. Remaining errors are, of course, mine.

Notes

1. For a summary of Harrod’s growth theory, see for instance Hein (2004).

2. For a summary and critique of Solow’s growth model, see Hacche (1979) and Hein (2004).

3. Mankiv, Romer and Weil (1992) also examine the divergence of growth rates between countries. However, they defend the Solow model in many aspects while at the same time arguing that factors such as human capital formation and technological progress should not be neglected in the explanation of economic growth.

4. Note that an endogenous growth theory was only ‘new’ to neoclassical growth theory, since various heterodox growth models embodied the notion of endogenous growth either as an equilibrium outcome of the model itself or through modelling technical change already in the 1950s and 1960s, see for example Kaldor (1957, 1966) and Robinson (1962).

5. For a more extensive summary and critique of ‘endogenous’ growth theories, see Hein (2004).

6. León‐Ledesma (2000) stresses that increasing returns can also be found in the services sector and might explain regional differences in growth rates through labour mobility and the high labour intensiveness of work in the services sector. This sector thus constitutes another ‘engine of growth’ in the sense of Verdoorn’s Law.

7. A more extensive model which integrates also non‐technical productive factors can be found in León‐Ledesma (2002).

8. For an extensive discussion of the causes of the endogeneity of increased labour productivity and technological change as a result of higher output growth and investment of entrepreneurs see for example Cornwall (1977) and the references therein. Cornwall and Cornwall (2001) stress the endogenous evolution of technology as well as institutions.

9. Note, however, that it is not assumed that supply adjusts passively to demand in a manner of ‘Say’s law in reverse’, but rather it is recognised that its response may not necessarily be one of equal proportion and thus has to be modelled explicitly, see for example Dutt (2006), Setterfield (2006) and Felipe and McCombie (2006).

10. Note also that if the natural rate of growth shifts in the same direction as the actual rate, a convergence of the warranted rate of growth to the natural rate is made more difficult. If the warranted rate exceeds the natural rate of growth, the capital stock grows faster than labour productivity. In order to adjust to the natural rate, the warranted rate would have to decline, but since a warranted rate above the natural rate implies conditions of depression, the natural rate is also likely to fall together with the actual rate (León‐Ledesma and Thirlwall 2002).

11. Results for the estimation of equation (1) are available from the author on request.

12. In addition to equation (3), further dummies were constructed using a Hodrick–Prescott (HP) filter and a 5‐year moving‐average on the actual rate of growth. The dummy was assigned the value of one for periods where average actual rate of growth was higher than the HP filter, i.e. the moving‐average. The results confirmed the findings of equation (3) and are available on request.

13. All equations were also estimated using simple OLS regressions and estimating all equations for each country separately. However, the estimations for the coefficients using SUR regressions were considerably more efficient and showed better results, so that only those will be discussed in this paper. Results of the OLS regressions are available from the author on request.

14. For a summary of economic and political developments in the Latin‐American countries, especially during the 1980s, see Kraemer (1995).