A rising tide is supposed to lift all boats. But anyone who has watched a booming economy leave large swaths of its population behind knows the metaphor has limits. When GDP expands, who actually gets richer? And what determines whether the benefits spread or concentrate at the top?
A new study in The Journal of International Trade & Economic Development takes on that question by looking at 43 countries over a quarter-century. The central finding: economic growth tends to widen income inequality on its own, but two specific institutional features can flip that script. Stronger employment protection laws and greater gender parity in higher education both weaken, and in some cases reverse, the inequality-increasing pull of growth.
The question behind the numbers
Researchers have long argued over what drives the gap between rich and poor. Early thinking, going back to Simon Kuznets in the 1950s, suggested that inequality rises in the early stages of development and falls as economies mature. More recent work, including Thomas Piketty’s, has pointed to capital accumulation and policy choices as forces that can keep inequality climbing indefinitely.
Most empirical studies have tested these ideas one variable at a time, asking whether unions reduce inequality, or whether corruption worsens it, or whether trade openness helps the poor. Bernadette Louise Halili and Carlos Rodriguez Gonzalez, both at the University of the Basque Country in Spain, argue that this piecemeal approach misses something important: institutions and growth interact. The effect of economic expansion on inequality might depend entirely on whether a country has the rules, protections, and access policies in place to distribute the gains.
Their hypothesis, drawn from theoretical work by Daron Acemoglu, James Robinson, and Piketty, is that growth alone does not trickle down. It trickles only when institutions channel it that way.
Building the dataset
The authors assembled panel data spanning 1995 to 2019 for 43 countries, a mix of wealthy OECD members and developing economies drawn from the expanded “Varieties of Capitalism” literature. That second group matters because most prior work on institutional determinants of inequality has been confined to rich countries. By bringing in places like Brazil, Indonesia, Turkey, and Egypt, the authors could test whether their broader findings generalize across different stages of development, though their analysis of labor market institutions remained restricted to OECD countries due to data availability.
Their outcome measure was the Gini coefficient for post-tax, post-transfer household disposable income, a standard gauge of inequality after governments have collected taxes and distributed benefits. Higher numbers mean more inequality.
For the institutional side, the authors examined three categories. Labor market institutions were captured through employment protection legislation (how hard it is to fire workers, how restricted temporary contracts are) and collective bargaining coverage (the share of workers covered by union-negotiated agreements). Gender-based institutions were measured through the wage gap between men and women and through gender parity in tertiary education enrollment. Governance-based institutions were tracked through World Bank indicators for rule of law and control of corruption.
They also controlled for financial market size, trade openness, government debt, and the persistence of inequality from one year to the next.
The analytical approach
To test whether institutions change how growth affects inequality, the authors built interaction terms, statistical constructs that let the effect of one variable (GDP growth) depend on the level of another (say, employment protection). They ran the analysis three different ways, using techniques designed to handle the quirks of panel data: correlated random effects, feasible generalized least squares, and systems-generalized method of moments. Using multiple methods is a way of checking that results are not artifacts of any single statistical choice.
They lagged their explanatory variables by a year to reduce the risk that inequality was driving the supposed causes rather than the other way around.
What the analysis revealed
Across all three statistical approaches, two interactions stood out as consistently significant.
First, employment protection legislation. In countries with weak protections, GDP growth was associated with rising inequality. As protection levels increased, that relationship weakened. Above a certain threshold, more growth was actually linked to lower inequality. The authors calculated that when employment protection scores ranged from 3 to 4 on the OECD’s scale, each unit of GDP growth corresponded to meaningful reductions in the Gini coefficient. At the low end of the scale, growth was pushing inequality up.
Second, gender parity in tertiary education. Countries where women’s enrollment in higher education matched or exceeded men’s saw growth translate into lower inequality. In countries where women lagged, growth widened the gap. The inequality-reducing effect of growth kicked in only once parity crossed roughly 1.0, meaning equal access, and strengthened as women’s enrollment exceeded men’s.
Other institutional variables showed results in the expected direction but with less statistical consistency. Collective bargaining coverage, rule of law, and control of corruption all appeared to soften the inequality-increasing effect of growth in some models but not all. The gender wage gap, when interacted with growth, tended to amplify inequality, though the effect did not reach statistical significance.
The authors also found that inequality is highly persistent. The Gini coefficient from one year was the strongest predictor of the next year’s figure, with coefficients close to 1. Inequality, in other words, tends to stick.
What it means in practice
The authors argue that their findings point to specific policy levers rather than vague calls for “better institutions.” Labor market reforms that strengthen job security, limit reliance on precarious contracts, and balance the power between employers and workers appear to change how growth is distributed. So do policies that expand women’s access to higher education, including financial aid, scholarships, and programs that address enrollment gaps.
They are careful to note that these reforms will look different in different places. Cultural norms, existing legal frameworks, and public support vary enormously across the 43 countries in their sample. A policy that works in Sweden may need substantial modification in Indonesia or Argentina.
Caveats worth noting
The authors flag several limits. Their study relies on observational data, not experiments, so the relationships they identify are associations rather than proven causes, even with the one-year lag structure they use. Panel data across dozens of countries can obscure what happens inside any single economy, where unique political dynamics might produce different patterns.
Their focus on formal institutions also leaves out informal ones, the unwritten rules, social norms, and non-market arrangements that vary across capitalist systems. The authors suggest that future research could incorporate these through configurational analyses of how bundles of institutions work together, rather than examining them one at a time.
And the persistence finding deserves attention on its own. If inequality is as sticky as the data suggest, with this year’s Gini almost fully determining next year’s, then changing the trajectory requires sustained institutional effort, not one-off reforms. Growth, on its own, will not do the work.
