25 examples of research with impact: The science of fairness – inside inequality

Socio-economic inequalities are often described as a defining challenge of our time. Ever since the turn of the century, there has been a growing in interest among academics, the public, and policy makers alike to understand the potential negative consequences of excessive concentration of economic resources. The field of social inequalities has become important in both economics and sociology. Developments in the possibilities to gather and analyse data on a large scale, in areas such as income, wealth and living conditions has been of significant benefit for the field.

“Understanding and addressing socio-economic inequality is not only an academic exercise, it is now considered central to building inclusive, democratic societies,” explains Philippe Van Kerm, Professor of Social Inequality and Social Policy at the University of Luxembourg, and previously at the Luxembourg Institute of Socio-Economic Research (LISER).

Van Kerm & team study the dynamics and determinants of various forms of socio-economic inequalities, with a focus on high-income countries.

Methods for a clearer look at impact of specific dynamics

Van Kerm considers the main contributions of him and his team’s research to be the development of statistical methods that enable the study of distributions of outcomes, as well as the dynamics of these outcomes.

“Most tools in data scientists’ toolboxes are designed to assess impacts ‘on the mean’ of outcomes of interest. But studying inequality obviously requires tools that to go beyond examining “averages” towards examining broader “distributive” impacts – we need to assess effects ‘at the bottom’, ‘in the middle’ and ‘at the top’ and understand how these combine to shape inequality.”

As an example, Van Kerm and team worked on developing and refining “influence function regression methods” – methods easy to implement which have found widespread use. The methods have recently provided insight into not just whether inheritances increase wealth, but whether and how much inheritances increase inequality in wealth.

They have also helped to understand not just whether foreign workers increase average earnings but whether they tend to increase overall earnings inequality.

Examining trends over time to identify what works

Once it has been established how to measure inequalities, the next step is to examine trends over time in order to make comparisons across societies.

“The goal is to try and identify what are key micro-determinants – arising from individual decisions and circumstances – as well as macro-determinants, which arise from global trends such as technological change, demographic trends, social and economic policies, etc. This is key knowledge to think of potential interventions.”

Van Kerm’s main future research goals are identifying “what works” in terms of reducing poverty, mitigating inequality and fostering prosperity, which he will approach by rigorously assessing the implications of interventions and policy decisions and evaluating the impacts of major socio-economic shocks.

But – getting a clear picture of social inequality and its many drivers is nowhere near simple, and Van Kerm explains many challenges lie ahead. For example, despite all the research already conducted, a key question remains: What are the precise drivers of long-term levels and trends in inequality in contemporary societies? The challenge in answering this question stems from the fact that it has multiple layers and it is in constant evolution.

Unravelling the effects of policies and economic shocks

Another challenge is identifying the causal effects of policies or economic shocks on inequality and the distribution of welfare in a broader way.

“Obviously, one cannot resort to the classic experimental designs since potential determinants of social inequalities cannot be randomised. Furthermore, we are looking at factors which are broad – they rarely hit well-defined populations in isolation – and slow-moving – their effects build up over time. Natural experiments may sometimes come to the rescue and artificial intelligence-based tools may potentially help in the future for simulation-based approaches, but whether they will be able to provide definite, universal answers some of the big questions about inequality drivers remains uncertain.”

How the field will evolve: Data, data, data

Van Kerm explains that large-scale administrative are increasingly being made available for research, equipping them with valuable data that enables the tracking of trajectories across key phases of people’s lives: early childhood, school, work. This gives researchers the chance to study the dynamics of individual welfare over time.

“Broader access to large matched administrative records and surveys will challenge how we assess welfare in a long-run perspective; how we assess long-run inequality and how we assess policy interventions. This will feed both more theoretical and empirical work.”

Advancements in technical tools are also revolutionising the possibilities of inequality research:

“Machine learning methods are now used in the field – for example to predict socio-economic outcomes from past events, circumstances or life courses; or to handle missing data issues, or help in matching data from disjoint sources. Machine learning and artificial intelligence tools will certainly be leveraged further in research.

“I can see them being used to augment information available in existing data – by helping to match sources, by generating or imputing missing information, by helping simulating counterfactual societies under alternative scenarios.”