Abstract
Using a comprehensive data set of 22 industries in 86 countries over the period 1980–2012, we empirically identify the effect of inequality on industry-level value added growth. We show that an unequal income distribution increases the growth rates of physical-capital-intensive industries and reduces the growth rates of human-capital-intensive industries by lowering human capital and raising physical capital accumulation. Our study suggests that the empirical difficulty to identify a monotonic relationship between inequality and aggregate growth reflects differences in the relative importance of human and physical capital in a country’s production structure.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
See Cingano (2014) for an extensive literature review on the relationship between inequality and growth.
See also Galor (2011) for an extensive survey on the relation between inequality and economic development.
As argued by Ciccone and Papaioannou (2009), the cross-country cross-industry approach has also been proven useful in research areas where data availability is scarce and where variables of interest are highly correlated, which is often the case in cross-country growth regressions because countries that are similar in one dimension (e.g., GDP per capita) are often also similar with respect to other characteristics, such as financial development or the accumulated stock of human and physical capital (see also Table 1, Panel B).
In this test, we find that the statistical significance of the initial value added share is lower than the corresponding coefficient on the initial logarithm. Thus, in our sample, the initial logarithm of value added seems to be a better proxy for the concept of growth convergence than the initial share of value added.
Many countries with high levels of inequality, such as Nicaragua and South Africa, do not report Gini coefficients prior to 1990. A shorter window around 1980 would, consequently, reduce the variation in our measure of inequality substantially. The inclusion of these observations to our data set is justified by the high persistence of Gini coefficients over time (see Christopoulos and McAdam 2017).
It is a usual correction (e.g., Halter et al. 2014) to account for the fact that expenditure/consumption-based Gini coefficients are systematically smaller than Gini coefficients calculated based on income data. 6.6 is the mean difference between Gini coefficients if both types of construction methods are available for the same country and year (Deininger and Squire 1996).
The data are not available for 1980.
The UK is the only country for which we obtained industry-level data on the real capital stock. Yet, the data are not available prior to 1995.
The 75th percentile of the distribution of Gini indices is equal to 51 and the 25th percentile is equal to 36.3. The 25th percentile of the distribution of HCI is equal to 0.55; the 75th percentile is equal to 0.78. Using these values, we calculate the economic magnitude for the human capital channel as follows: \(-\,0.366*(0.78-0.55)*(36.3-51)\approx 1.2\).
The 75th percentile of the distribution of Gini indices is equal to 51 and the 25th percentile is equal to 36.3. The 25th percentile of the distribution of PCI is equal to 0.86; the 75th percentile is equal to 2.23. Using these values, we calculate the economic magnitude for the physical capital channel as follows: \(0.0401*(2.23-0.86)*(51-36.3)\approx 0.8\).
We use the capital stock of machinery and non-transport equipment, rather than the total capital stock, as the latter includes residential real estate, which is arguably unrelated to the supply of physical capital in industries’ production processes and, hence, to physical capital driven growth. The data on the capital stock of machinery and non-transport equipment come from the capital details in the 9.0 version of the Penn World Table.
All of the channel variables have a low time variation, which is important because Gini coefficients also display a pronounced autocorrelation. That is why inequality is highly correlated with the stock of human and physical capital, as shown below, but not, for instance, with the share of investment in GDP, a more volatile proxy for physical capital.
The 75th percentile of the distribution of Gini indices is equal to 51 and the 25th percentile is equal to 36.3. As we take the logarithm of the dependent variable, we obtain the economic effect for HC as follows: \((51-36.3)*100*(-\,0.00807)\). Equivalently, the corresponding effect for PC is calculated as follows: \((51-36.3)*100*0.03\).
Instead, we test whether countries in which inequality is driven by low income shares of the middle class, compared to countries in which inequality is driven by low income shares of the poorest part of the population, behave differently in terms of industry growth.
We hence drop the years of 2010–2012, as otherwise we would have five 5-year and one 8-year period. This could be problematic because the magnitude of the convergence coefficient varies with the length of the time period.
The number of countries included in these regressions is larger than in previous estimations, as some countries only report their Gini coefficients starting in the 2000s and, hence, not in the 15-year window around 1980.
In this sub-section, whenever the initial value of the variables is not available, we use the nearest value of this variable in the same period in order to maximize the number of observations.
Column (1) refrains from country-time fixed effects, as they would absorb the effect of GINI.
Relative to column (2), the coefficient corresponding to initital industry size decreases significantly. Since the effect of the average size of each industry in a given country is controlled for with the country-industry fixed effects, the coefficient of this variable captures the growth effects of abnormal industry size (see Braun and Larrain 2005), and can thus not be compared to the one obtained in column (2).
A further reason for using US data is that human and physical intensities at the industry level are unavailable for most countries in our sample.
For the reasons outlined in Sect. 2, we drop the top 1% of value added shares since they take extreme values (larger than 42%), thus distorting our estimates.
As argued in Sect. 2, this is likely to be the case because the number of industries providing data on value added varies significantly across countries. For instance, whereas the UK provides data for all manufacturing industries, Gambia only provides data on 14 industries. Consequently, the initial share of value added for each industry in Gambia is consistently higher than in the UK, thereby potentially biasing the results.
References
Bartelsman, E. J., & Gray, W. (1996). The NBER productivity manufacturing database.
Baumol, W. J. (1986). Productivity growth, convergence, and welfare: What the long-run data show. American Economic Review, 76(5), 1072–1085.
Berg, A., Ostry, J. D., Tsangarides, C. G., & Yakhshilikov, Y. (2018). Redistribution, inequality, and growth: New evidence. Journal of Economic Growth, 23(3), 259–305.
Bourguignon, F. (1981). Pareto superiority of unegalitarian equilibria in Stiglitz’ model of wealth distribution with convex saving function. Econometrica, 49(6), 1469–1475.
Braun, M., & Larrain, B. (2005). Finance and the business cycle: International, inter-industry evidence. The Journal of Finance, 60(3), 1097–1128.
Brueckner, M., & Lederman, D. (2018). Inequality and economic growth: The role of initial income. Journal of Economic Growth, 23(3), 241–366.
Christopoulos, D., & McAdam, P. (2017). On the persistence of cross-country inequality measures. Journal of Money, Credit and Banking, 49(1), 255–266.
Ciccone, A., & Papaioannou, E. (2009). Human capital, the structure of production, and growth. Review of Economics and Statistics, 91(1), 66–82.
Cingano, F. (2014). Trends in income inequality and its impact on economic growth. OECD Social, Employment and Migration Working Papers No 163.
Cörvers, F. (1997). The impact of human capital on labour productivity in manufacturing sectors of the European Union. Applied Economics, 29(8), 975–987.
Deininger, K., & Squire, L. (1996). A new data set measuring income inequality. World Bank Economic Review, 10(3), 565–591.
Deininger, K., & Squire, L. (1998). New ways of looking at old issues: Inequality and growth. Journal of Development Economics, 57(2), 259–287.
Durlauf, S. N. (1996). A theory of persistent income inequality. Journal of Economic Growth, 1(1), 75–93.
Dynan, K. E., Skinner, J., & Zeldes, S. P. (2004). Do the rich save more? Journal of Political Economy, 112(2), 397–444.
Galor, O. (2011). Inequality, human capital formation, and the process of development. Handbook of the Economics of Education, 4, 441–493.
Galor, O., & Moav, O. (2004). From physical to human capital accumulation: Inequality and the process of development. Review of Economic Studies, 71(4), 1001–1026.
Galor, O., & Zeira, J. (1993). Income distribution and macroeconomics. Review of Economic Studies, 60(1), 35–52.
Halter, D., Oechslin, M., & Zweimüller, J. (2014). Inequality and growth: The neglected time dimension. Journal of Economic Growth, 19(1), 81–104.
Iacovone, L., & Zavacka, V. (2009). Banking crises and exports: Lessons from the past. The World Bank.
Kaldor, N. (1957). A model of economic growth. Economic Journal, 67(268), 591–624.
Kuznets, S. (1955). Economic growth and income inequality. American Economic Review, 45(1), 1–28.
Mayer, S. E. (2001). How did the increase in economic inequality between 1970 and 1990 affect children’s educational attainment? American Journal of Sociology, 107(1), 1–32.
Orazem, P., & Tesfatsion, L. (1997). Macrodynamic implications of income-transfer policies for human capital investment and school effort. Journal of Economic Growth, 2(3), 305–329.
Rajan, R. R., & Zingales, L. (1998). Financial dependence and growth. American Economic Review, 88(3), 559–586.
Acknowledgements
We thank Michael Berlemann, Luis Catão, Valeriya Dinger, David Dornekott, Francisco Ferreira, Oded Galor, Stefan Homburg, Robert Inklaar, Aart Kraay, Vlad Marincas, Gian Maria Milesi-Ferretti, Alexander A. Popov, Hans-Werner Sinn, Sven Steinkamp, Frank Westermann, Joachim Wilde, two anonymous referees and an associate editor, as well as conference participants at the 11th Workshop for Macroeconomics (ifo Dresden and University of Hamburg), at the University of Osnabrück, at the University of Groningen, at the 71st European Meeting of the Econometric Society, at the Development Economics and Policy Conference at the University of Göttingen and at the 2017 Annual Meeting of the German Economic Association for helpful comments. All remaining errors are our own.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Appendix
Rights and permissions
About this article
Cite this article
Erman, L., te Kaat, D.M. Inequality and growth: industry-level evidence. J Econ Growth 24, 283–308 (2019). https://doi.org/10.1007/s10887-019-09169-z
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10887-019-09169-z