Cross‐country evidence on the determinants of inclusive growth episodes

• DOI: http://doi.org/10.1111/rode.12605
• Author: João Tovar Jalles,Luiz Mello
• Date: 01 November 2019
• Published date: 01 November 2019

1818

|

wileyonlinelibrary.com/journal/rode Rev Dev Econ. 2019;23:1818–1839.

© 2019 John Wiley & Sons Ltd

DOI: 10.1111/rode.12605

REGULAR ARTICLE

Cross‐country evidence on the determinants of

inclusive growth episodes

João TovarJalles1*

|

Luizde Mello2

*João Jalles is also affiliated to UECE—Research Unit on Complexity and Economics, University of Lisbon, and ISEG/

UTL—Technical University of Lisbon, Portugal

1Nova School of Business and

Economics,Universidade Nova de Lisboa,

Lisbon, Portugal

2OECD, Economics Department, 2 rue

Andre Pascal, 75775 CEDEX 16, Paris,

France

Correspondence

João Tovar Jalles, REM/UECE. R. Miguel

Lupi 20, 1249‐078 Lisbon, Portugal.

Email: joaojalles@gmail.com

Abstract

Widening income disparities and slow productivity growth

in most advanced and several emerging‐market economies

have rekindled interest in the empirical analysis of the de-

terminants of inclusive growth, defined in this paper as epi-

sodes of increases in GDP per capita without a concomitant

deterioration in the distribution of household disposable

income. The empirical analysis is based on a chronology

of inclusive growth episodes between 1980 and 2013 for

a sample of 78 countries. Logit and multinomial probit

estimations show that human capital accumulation, the re-

distributive potential of tax‐benefit systems, increases in

multifactor productivity and labor force participation, as

well as trade openness and a range of institutional factors,

including political system durability and electoral regimes,

are important determinants of the probability of occurrence

of inclusive growth. This empirical evidence contributes to

the policy debate about how countries can deal with effi-

ciency–equity tradeoffs.

KEYWORDS

growth, income distribution, multinomial probit, relogit

JEL CLASSIFICATION

O47; O15; D31

|

1819

JALLES And dE MELLO

1

|

INTRODUCTION

A combination of slow productivity growth and rising income inequality in most advanced and several

emerging‐market economies has rekindled interest in the empirical links between economic growth

and income distribution. In particular, although the specific channels through which growth and in-

come distribution are related are complex and often difficult to disentangle empirically, there is in-

creasing, albeit weak, cross‐country evidence that income inequality undermines growth1

and policy

initiatives that are growth friendly often do affect social groups differently2

.

Of particular interest in this strand of literature is the experience of countries that have managed

to sustain spells of uninterrupted output growth without a concomitant deterioration in the interper-

sonal distribution of income. In fact, these episodes, which we characterize as inclusive growth, are

not infrequent. Nor are they circumscribed to specific regions or time periods. For example, on the

basis of the data available from the World Bank's World Development Indicators (WDI), there are

268 episodes of increases in GDP per capita without an associated deterioration in the distribution

of household disposable income in the sample of countries for which information is available for the

period 1980 to 20133

.

Based on this chronology, we estimated pooled logit and multinomial probit models to gauge

empirically the determinants of inclusive growth episodes. We started by looking at a variety of

output growth and income distribution covariates, including indicators of the population's educa-

tional attainment, the size and redistributive potential of tax‐benefit systems, as well as indicators

of demand and economic structure, including trade openness, inflation and unemployment. We also

looked at various measures of financial deepening, infrastructure and institutional characteristics

that are known to influence growth and income distribution. We tested for the robustness of our

empirical findings by looking at alternative estimators and different sets of growth and income

distribution covariates.

We depart from the literature by focusing on a chronology of episodes linking growth to income

distribution, rather than estimating (jointly or independently) growth and income distribution equa-

tions and testing for specific channels of causality. In this regard, our approach is akin to that of Anand,

Mishra, and Peiris (2013), who also identify episodes of inclusive growth on the basis of changes in

GDP per capita and the distribution of income for a selected group of countries. We nevertheless use

different estimation techniques and look at additional macroeconomic and institutional variables that

can shed light on the likelihood of an inclusive growth episode.

Our main findings are that the probability of an inclusive growth episode rises with increases

in human capital accumulation, redistribution through the tax‐benefit system, labor force partici-

pation and multifactor productivity, as well as trade openness, and it falls with a rise in inflation,

output volatility, and unemployment. Also, we show that institutions matter, and inclusive growth

episodes are more likely in countries with durable political systems, regular parliamentary elec-

tions and electoral regimes based on proportional representation. Moreover, although the empir-

ical findings are somewhat weaker, financial deepening, as reflected in an increase in different

measures of the stock of credit in the economy, reduces the probability of an inclusive growth

episode, probably because of its association with a higher probability of occurrence of banking

and financial crises, which are in turn detrimental to growth and improvements in the distribution

of income.

The remainder of the paper is organized as follows. Section 22 reviews the literature. Section 33

outlines the empirical methodology and Section 44 describes the data. Section 55 presents and dis-

cusses the main results. The last section concludes.