Exchange rate and inflation: a neo-structuralist approach for Brazilian manufacturing sectors (2010–2019)

Abstract

The objective of this article is twofold. The first goal is to comprehend the effect of changes in the exchange rate on prices using a neo-structuralist approach. Our theoretical model showed that the greater the share of imported inputs in costs, the more powerful the pass-through. Furthermore, considering the structuralist notion of neutral inflation, the model also indicated that the exchange rate pass-through of sectors with a small (great) share of salaries in costs is less (more) sensitive to salary changes. Also, the model was extended to endogenize the exchange rate pass-through on prices in relation to the effects of a regime of real exchange rate oriented to economic development on the markup rate and the productive structure. The second goal of the article is to provide time-series evidence on the exchange rate pass-through in prices across different manufacturing sectors of the Brazilian economy over the 2010–2019 period. Empirical results delivered evidence that pass-through is partial and varies across sectors. Moreover, the empirical findings confirmed our theoretical results insofar as sectoral differences of pass-through are associated with the markup rate, the degree of outward orientation, the competition between firms, and the share of imported inputs of each sector.

Keywords:

• Neo-structuralist approach
• inflation
• exchange rate pass-through
• Brazil

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Such a vision, concentrated on the positive effects of RER’s policies oriented for economic development via its effects on productive structure, capital accumulation, and technological progress (which occurs by redirecting resources from traditional (largely non-tradable) sectors to modern (tradable) sectors characterized by increasing returns) may be roughly summarized in the vision of new developmentalism (see Bresser-Pereira 2010 55, 2016, 2019, 2020 and Bresser-Pereira et al. 2015 55 to obtain a detailed and extensive discussion on this topic).

2 Even though a competitive RER results in a contractionary effect on output growth in the short run, insofar as the workers employed in non-manufacturing activities (traditional sectors with lower labor productivity) are absorbed by manufacturing activities (modern sectors with higher labor productivity), the greater labor productivity is absorbed by higher real wages in the long run (Ros 2015). The adoption of an export-led strategy based on a competitive RER requires that workers accept lower real wages, in the short run, in exchange for possible higher real wages in the long run; it is a tradeoff with distributive effects in the present and possible gains in future (Guzman et al. 2018).

3 To the best of our knowledge, only Pimentel et al. (2020) have studied the determinants of manufacturing prices within the existing literature. However, their goal was to measure cost pass-through into manufacturing prices within the Brazilian economy over the period between 1996 and 2014.

4 It is implicit in our argument that the share of labor costs in relation to the total costs is a measure of distributive conflict. This can be justified on the ground of two arguments. First, this variable represents the different technical conditions of different sectors. More labor-intense sectors (low capital:labor ratio) have a greater share of costs associated with labor. Thus, in line with Equation (13), these sectors have stronger exchange rate pass-through on prices, all other things being equal. They are more sensitive to changes in costs. The second argument is that this variable changes over the business cycle. In moments of a faster pace of economic growth, the unemployment rate is low and wage demands are stronger, which leads to a high share of labor costs in relation to total costs. In this context, firms are more sensitive to changes in costs due to a more competitive RER and therefore the exchange rate pass-through on prices tends to be stronger (also, firms pass the enhanced costs on to prices without losing market position). Conversely, in moments of a slower pace of economic growth, the unemployment rate is higher and thus wage demands are weaker, which leads to a low share of labor costs in relation to total costs. In this context, firms are less sensitive to changes in costs due to a more competitive RER, so the exchange rate pass-through on prices tends to be weaker (also, firms avoid passing the enhanced costs on via prices because there is the risk of losing market position).

5 See Table A1 in appendix.

6 Using quarterly data, Campa and Goldberg (2002) introduced three lags of the exchange rate to capture the sluggish price adjustment to exchange rate over a year.

7 We estimated a second specification of Equation (14)(14) $${\mathrm{p}}_{\mathrm{t},\mathrm{s}}=\mathrm{\alpha }+{\mathrm{b}}_1{\mathrm{x}}_{\mathrm{t}}+{\mathrm{\beta }}_{\mathrm{j}}{\mathrm{e}}_{\mathrm{t}-\mathrm{j}}+{\mathrm{\epsilon }}_{\mathrm{t}}$$(14) , introducing the first lag of inflation to capture inertial inflation, demand growth (demand pressures), and oil price (as a measure of imported costs) in addition to the exchange rate. Contemporaneous and lagged values of these variables were considered, using three lags of each variable as instruments. The regressions are not presented in this study. They are available upon demand.

8 If the lag length that minimizes the information criterion was insufficient to eliminate residual autocorrelation, we performed additional specifications with further lags (until the Lagrange multiplier test does not reject the null hypothesis of no residual autocorrelation). This is a problem when the econometrician is concerned with parameters estimated by ordinary least squares as the tests t and F are no longer valid due to biased error variance. As we are interested in the IRF and FEVD of the VAR estimates, we opted for the more parsimonious model (according to the information criterion) since the addition of further lags (used to eliminate residual correlation) increase the error variance, as one further lag means the addition of one parameter of cross-correlation for each endogenous variable/equation. Thus, the results with further lags produced large confidence intervals in the IRF, confirming that the more parsimonious model is a better fit. The results were similar to some extent, indicating the robustness of the results.

9 Assuming that demand can be higher than supply in the short run but equals supply in the long run.

10 The variables are presented in Table A2 in the Appendix.

11 The tests, as well as the descriptive statistics of variables, are presented in the Supplementary Material.

12 The full output of GMM estimates is presented in the Supplementary Material.

13 The full output of VAR estimates is presented in the supplementary material.

14 Articulating an argument that considers Brazilian deindustrialization and its many consequences over productive structure (market power, outward orientation, imports, and capacity to compete with foreign firms) and their inflationary consequences exceeds the scope of this study and constitutes an interesting contribution to future studies.

15 The association between markup rate and pass-through was performed by a graphical analysis. This strategy is the only one possible, as no monthly data exist to allow the use of econometric methods. The markup variable was constructed using the annual data from the IBGE Annual Industry Survey (PIA) over the period 2010 to 2017. The computation methodology follows Nucci and Pozzolo (2001) and is calculated using the following procedure: Markup Rate = (Value of Sales + ΔInventories – Payroll – Cost of Materials) / (Value of Sales+ ΔInventories). All variables came from IBGE Annual Industry Survey (PIA).

16 Sectors with extreme values were removed from correlograms (black line). The trend line of the correlogram with all sectors is represented by the blue line.

17 The export coefficient came from National Conference Industry and represents the share in a percentage of revenues associated with exports. The GMM estimates refer to the equation performed with four degrees of freedom.

18 The import penetration is the share of national consumption that is imported. The higher the sectoral import penetration, the stronger the competition that national manufacturing sectors face against foreign firms.

19 This is plausible assuming that foreign firms have higher productivity growth than national firms.

Additional information

Notes on contributors

Hugo Carcanholo Iasco Pereira

Hugo Carcanholo Iasco Pereira is a professor at Federal University of Paraná, Curitiba, Brazil.

Fabrício José Missio

Fabrício José Missio is a professor at CEDEPLAR, Federal University of Minas Gerais, Belo Horizonte, Brazil. The author acknowledges CNPq and Minas Gerais Research Funding Foundation (FAPEMIG–Universal Demand Program–APQ-01964-18). The authors would also like to thank the professors Frederico Gonzaga Jayme, Jr., Gilberto Libânio, Martin Rapetti, and Nelson Marconi for their comments and suggestions as well as the participants of the 14th Brazilian Keynesian Association Conference. Any errors are our own.