Profit share and capital accumulation in Mexican manufacturing

ABSTRACT

Mexican manufacturing shows a long-run divergence between a rising profit share of income and a flat rate of capital accumulation. Following the Cambridge equation, the paper studies the determinants of capital accumulation and finds the divergence to be partly explained by a reduction in the output/capital ratio, which opened a gap between a rising profit share and falling profit rate. But besides this classical, Marxian mechanism, two additional factors were at play: first, nonlinear profitability effects, which reveal that, at the high profitability rates observed in Mexico, profitability was not a binding constraint on accumulation; and second, foreign profitability effects, which show that a rise in US profits rates affected negatively accumulation in Mexico. The results come from estimated equations for the capital accumulation rate and investment share of profits in an annual panel of 17 manufacturing industries during the period 1992–2019.

KEYWORDS:

• Capital accumulation
• profit share
• profit rate
• nonlinear profitability effects
• foreign profitability effects
• profits–investment puzzle
• Cambridge accumulation equation

JEL CLASSIFICATION:

• C23
• E22
• E25
• O14
• O47

Acknowledgement

The author would like to thank two anonymous reviewers for their very insightful comments, which significantly improved the paper. Naturally, the usual caveat applies.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. Or from 56 to 68%, if we deduct from gross operating surplus an estimate of the labour income of the self-employed. Since the adjustment, which uses data on the composition of the labour force (self-employed versus wage-earner), cannot be readily made at a disaggregated industry level, in the paper we will measure profits by the unadjusted operating surplus.

2. To borrow the term used in this context by the growth-diagnostics literature (Hausmann, Rodrik, and Velasco 2007; Ibarra 2011).

3. Staying within a classical approach, we will assume that causality runs from profits to accumulation. In the econometric analysis, the use of lagged profitability variables will support this interpretation. In a more general framework, accumulation could affect profits, for example because of Kaleckian demand effects or if faster accumulation triggers a process of Marx-biased technical change. A formal, econometric testing of the possible two-way causality is left for future work.

4. Likewise, a rise in the investment share of profits, sometimes despite a fall in the profit rate, can be observed in acceleration episodes in countries like Brazil and the US in the 1960s and early 1970s (Kliman and Williams 2015; Mateo 2018), and China during the 1980s and 1990s (Felipe, Laviña, and Fan 2008).

5. More specifically, if the variables were not lagged there would be some reduction in the economic and statistical significance of the profit rate or its components, while the RER and M2 would no longer be statistically significant.

6. The maximum number of lags is restricted to two to reduce the number of instruments and hence the potential downward bias in the standard errors of the estimated coefficients (see Windmeijer 2005).

7. For example, the ratio between the real exchange rate coefficients in columns (2.1) and (2.4) is equal to 0.57 = 0.048/0.084.

8. In line with this view, if we removed the latter variables from the estimations, the RER coefficient would rise, for example from 0.048 to 0.072 in Equation (2.1) of Table 2.

9. Not shown to save space, the lack of significance of the output growth rate in column (2.4) remains if we eliminate the profit rate from the estimated equation or if we use the lagged rather than current output growth rate. Similarly, the output-growth coefficient is not affected if we remove the profit rate from column (2.1). The results suggest that the lack of significance of the output growth rate in the investment-share equation cannot be explained by a possible correlation between the rates of profit and output growth.

10. An unexpected result, whose intuition is difficult to grasp, is the negative coefficient on the capacity/capital ratio in the investment-share equations.

11. The ratio between the estimated coefficients, equal to 0.48 = 0.11/0.23, is not far from the mean profit rate of 0.61, when C&E is excluded.

12. Regarding other explanatory variables, we see (a) a smaller interest rate coefficient in the equations that include the crisis dummy (which suggests that the negative effect of the interest rate on investment was abnormally large in the midst of the crisis, when the interest rate shot up), and (b) a further weakening of the accelerator effect in the equations that include either year fixed effects or the crisis dummy, which again suggests that the mentioned effect was abnormally large during the crisis, in the midst of a severe output contraction. This reinforces the puzzling observation of a weak accelerator effect, particularly in the investment share equations, as already mentioned in the text.

13. The share of inward foreign direct investment (FDI) in domestic fixed investment was also considered as an indicator of integration, but results were never close to statistical significance. Perhaps a better indicator would be the FDI stock rather than flow, but industry stock data are not readily available.

14. Although the industrial surveys do not specify country of origin, the great majority of Mexican imports come from the US. According to the World Input – Output Tables (WOIT) database, in the year 2000 the US accounted for 67% of intermediate imports in the Mexican manufacturing sector. Unfortunately, the industry classification in WIOT does not exactly match that in Mexican KLEMS, so we cannot use the WIOT data to calculate the import ratios included in the following estimations.