![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
ABSTRACT
The article uses the KLEMS database to estimate equations for the rate of private, non-residential capital accumulation in manufacturing, whole tradables, and non-tradables in Mexico during the period 1992-1994. It shows the rate of capital accumulation is positively correlated with the profit rate and its components – the profit share, the output/capital ratio and the relative price of capital goods, negatively correlated in the latter case – in the three sectors, but with considerably larger profitability effects in manufacturing and tradables. It also shows that capital accumulation is positively correlated with the real exchange rate in manufacturing and tradables, and negatively (and less strongly) correlated in non-tradables. A real depreciation increases the relative prices and profit rates of the former sectors, shifting accumulation toward them and away from non-tradables, with a positive overall effect on capital accumulation. The estimations – based on the error-correction ARDL bounds testing approach, and supplemented by OLS equations using the lagged values of the explanatory variables – imply that the flat trajectory of capital accumulation in Mexico is explained in part by a downward trend in the profit rate, which resulted from a fall in the output/capital ratio that more than offset an increase in the profit share.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. www.inegi.org.mx/est/contenidos/proyectos/cn/ptf/default.aspx. Site accessed in March 2016. In this context, KLEMS refers to categories of inputs used in production and for which the database contains information, namely, capital (K), labor, energy, materials, and services purchased.
2. Magud and Sosa (Citation2010) survey the literature on the so-called Dutch Disease, and show that a majority of studies find a negative impact of real currency appreciations on the relative size of the tradables sector. Gluzmann, Levy-Yeyati, and Sturzenegger (Citation2012) is a rare exception in which the RER affects growth without involving an expansion of tradables. In their analysis, a depreciated currency stimulates growth by its positive effect on profit margins, which redistributes income from workers to capitalists and in that way raises saving and investment.
3. In the panel-data studies that dominate this literature, the degree of depreciation is typically measured in relation to an equilibrium level of the RER, with the latter determined by the Balassa–Samuelson effect (as in the original work by Rodrik Citation2008), or instead by a set of macroeconomic equilibrium conditions a la Nurkse (see Berg and Miao Citation2010 for an empirical comparison of the two approaches).
4. Forestry, fishing and quarrying.
5. The output/capital ratio in any sector i can be calculated as ui=(si,y/si,k)u, where u is the aggregate output/capital ratio, and si,y and si,k are the sector i’s shares in output and capital stock, respectively. During the period under study the output/capital ratio for the whole Mexican economy remained practically constant, at a level a little below 0.4.
6. Since we are interested in estimating long-run effects, we do not attempt to adjust the output/capital ratio for changes in capacity utilization, which we assume reflect mainly short-term variations in demand.
7. As is evident from the series, the fall in the output/capital ratio in manufacturing is a persistent, medium-term phenomenon, which must reflect the characteristics of technical change and capital accumulation – and their effect on the relative pace of capital deepening and labor productivity growth – taking place in the sector rather than short-term fluctuations in aggregate demand and capacity utilization. An analysis of this issue, however, is beyond the scope of the present article and is left for future work.
8. There are, we believe, three reasons for the large differences in measured profit rates between sectors and their persistence. The first has to do with differences in market structure: the average profit rates of the non-tradable sector are very large partly because the largest and more capital intensive activities in this sector operate, unlike the tradable sectors which are subject to international competition, under highly imperfect competition, often characterized by oligopolies and quasi-monopolies (telecommunications, transport and others). Second, there is also the fact that we should not expect equalization of average profit rates, even under competitive conditions, but rather equalization of marginal profit rates; and the more imperfect is competition (as happens in many non-tradable activities) the higher is likely to be the average in relation to the marginal profit rate (to the extent that the price elasticity of demand is lower the more imperfect are competitive conditions). Third, as discussed below in the text, the measured profit share (and thus profits in the numerator of the profit rate) includes self-employment income which is particularly large in the less capital intensive and more competitive activities of the non-tradables sector. By contrast, self-employment income is much less important in manufacturing, the largest segment of the tradables sector. This biases upwards the measurement of the profit rate in non-tradables relatively to tradables (particularly manufacturing).
9. Specifically, the ENE (Encuesta Nacional de Empleo) and ENOE (Encuesta Nacional de Ocupación y Empleo) national surveys.
10. See the unit-root test results in the appendix . For a previous application of the bounds testing methodology to the estimation of accumulation equations, see Van Treeck (Citation2008) which studies the influence of profitability on capital accumulation in a sample of developed countries.
11. The null hypotheses are σ=0 (t test) and σ=d1=…=dk=0 (F test). For the F test we use the small-sample critical values calculated by Narayan (Citation2005), while for the t test – for which no small-sample values are available – we use the asymptotic values reported by Pesaran, Shin, and Smith (Citation2001).
12. The negative t-test results for manufacturing and tradables in the equations with a complete block of first-differenced variables, which seem anomalous given the rest of the evidence, are explained by the statistical noise created by the presence of several highly non-significant coefficients on the lags of the first-differenced variables and which, following the suggestion by Pesaran, Shin, and Smith (Citation2001), are not removed from the equation until after carrying out the F and t bounds tests. As noted in the text, in the final equation, once those non-significant lags are removed, the t test also supports the existence of a long-run relationship in manufacturing and tradables.
13. Of course, the smaller values of the estimated coefficients may reflect not only the correction of an endogeneity bias, but also a lesser ability of the simple OLS regression to capture long-run effects when compared with the ARDL approach.
14. As a further robustness exercise, we re-estimated the equations for the shorter period 1997–2014 using alternatively the unadjusted profit rate and the profit rate adjusted for self-employment income in non-tradables (on the latter, see the last paragraph of section 3 above). The results showed no major difference in the estimated coefficient on the unadjusted and the adjusted profit rate, which in both cases had a value of about 0.04 in the OLS equations (and slightly lower in the ARDL ones); for comparison, the estimated coefficient on the profit rate in equations for the same shorter period was 0.44 in manufacturing and 0.33 in tradables. Since these results are based on a sample of only 18 annual observations, we do not report them in the tables, but they do increase our confidence in the reported results, and in particular in the much smaller effect that the profit rate appears to have on accumulation in non-tradables compared with manufacturing and tradables.
15. For completeness, the ARDL equation for non-tradables is shown in the table, but the very large (in absolute value) speed-of-adjustment coefficient suggests caution.
16. Excluding the production growth rate worsens some statistical properties of the ARDL equations. The bounds tests, in particular, do support the existence of a long-run relationship but only after simplifying the lag structure of the first-differenced variables.
17. Results not shown but available upon request. Recall that to obtain statistically significant results in the equations of , the profit-rate components were lagged one more year than was the production growth rate.
18. To pass the diagnostic tests, the estimated equations include outlier year dummies for the crisis years 1995 and 2009, and 2013. If the dummies are omitted, the estimated RER coefficient becomes smaller; without the 2009 dummy, for example, the coefficient in manufacturing falls by one half.
19. See Rapetti (Citation2016) for an analysis of the RER’s growth channels, and Ibarra (Citation2016) for a review of the empirical literature, with a focus on the Mexican case.
20. The reasoning assumes that labor is mobile and thus that the wage rate is equalized across sectors.
21. It may be possible to detect significant inter-sectoral effects exploiting the more disaggregated data available from KLEMS, perhaps following a panel estimation methodology. This possibility is left for future work.