Student Media Cell,
When you attend a session by a globally renowned professor with a stellar profile, having several globally acclaimed papers to his credit, and a complex-sounding topic such as “Analysing Intergenerational Mobility: Measures, Curves, Welfare,” you doubt the session would be relevant to you. However, noting that the audience consisted of students, professors, noted researchers, people from corporate houses, several people joining through webcast, you knew the topic addressed was a pressing one.
The speaker, Professor James Foster is a leading economist who sets the agenda for world policy. He is an Oliver T. Carr Professor of International Affairs and Professor of Economics, at the George Washington University. Professor Foster’s renowned research and policy work focusses on international economic development, inequality, poverty, economic theory and policy. His co-authored paper from 1984 in the prestigious journal Econometrica is the basis for all worldwide measures of poverty and is known as the Foster-Greer-Thorbecke class of poverty measures. His 2011 paper, co-authored with Sabina Alkire at Oxford University, provides the definitive basis for measuring multi-dimensional poverty.
In today’s presentation, Professor Foster presented a work that may become equally important in the public policy literature by seeking to look at how generations today are doing in comparison with its parents. While this can sound complex, Professor Foster, set this up with a very simple, but provocative question: “are we better off than our parents?” The audience responded with a resounding “Yes”. Enthusiastically smiling for photos, he explained that “Intergenerational Mobility” is focused on trying to answer this question using data to give a simple and consistent measure that has a range of interesting and policy relevant properties.
Here we detail a few comments and conceptual issues that he explained very simply.
“Intergenerational mobility has many unrelated approaches and its conceptual foundations are not well grounded.” Thus, it is neither possible to rank one generation as having systematically done better than another, nor do existing measures allow consistent ranking of sub-groups of population within a population. Thus, it is important and policy relevant that any new measure satisfy the following characteristics:
(i) An axiomatic foundation of intergenerational mobility to establish a consistent measure
(ii) Provide alternative measures for finding upward mobility and downward mobility that capture improvement and deteriorations in mobility across generation
(iii) Link mobility measurement to stochastic dominance and other tools from distributional analysis to help unpack the measure
(iv) Finally, it should link intergenerational mobility to measures of welfare
As exhaustive as it sounds, Professor covered it all in the 1-hour session and his talk began and ended on time.
After setting the context, Professor Foster presented and criticized three means of intergeneration mobility that are widely used in the literature:
The first was, measuring the Intergenerational Elasticity of Income (IGI). This refers to the percentage change in a child’s income when the parent’s income rises by 1 %; the IGI coefficient is obtained from the regression of the natural logarithm of the child’s income to the natural logarithm of the parent’s income. Citing an example, based on a US study, a 10% increase in the income of parents, typically leads to a 4-6% increase in the child’s income. And the shortcomings of this is that the income dynamics average out and thus, is insensitive to many changes such as when half the generation does better and another half does work. Further, it is not decomposable in a useful manner to comparing mobility in sub-groups of a generation.
The second method is exploiting Transition Metrics. As intuitive as it sounds, this is grouping people up, measuring probabilities of transition from one group to another, allowing capture of the entire distribution of movements. The shortcoming is that the thresholds set are arbitrary, meaning only movements across the threshold count as mobility and this may censor off few important results.
The third measure he introduced was the Distance Based Mobility. This measures the vector change or the flux in any set of dimensions (example, total changes in income). While this doesn’t need arbitrary thresholds, its major shortcoming is that this does not consider who gained/lost income as the starting point is not defined.
Pausing to take questions, he clarified that we must be able to compare the income of parents to children over time and space, else we would be comparing apples to oranges. Thus, there needs to be some standardization to take account of inflation, and other aspects that make it difficult to immediately compare Rs. value of income for parents with the Rs. value of income for their children.
The key unit of observation is the pair of incomes of parents (x) and the incomes of their children (y) and thus, the vector (x, y). For a population this can be extended to creating a panel p = (xi, yi), where i =1, 2, …, n and there are n households in parent and child generation Thus, the measure of mobility M(p) can be defined as a mapping from the set of all mobility pairs i.e. the domain of p to real number line that generates a measure of how mobile the population has been across the generation.
So how do we do it? It may be done by:
• Averaging out absolute movement or gap between parent and child
• Taking logarithm of value of parent and child, subtracting and averaging it out (logarithm is used to emphasize movement at the lower end)
• Transform each income by a parameter v and find average movement in transformed variable
M(p) must exhibit a range of reasonable properties and be consistent. Some of these properties are the Anonymity Axiom (where the names of families doesn’t matter as long as their nature is the same), Replication Invariance Axiom (where the measure is not affected if the families were cloned-explains why measurement for China wouldn’t be significantly different because it has more people), Decomposability Axiom (where decomposing into sub-groups and analyzing would help linking local initiatives to a national level) and the Expansion Axiom which is built on the idea that any change in one of the variables (parent’s or child’s status variables), increases mobility (movement or difference). This last Axiom is what is trending, with upward mobility meaning the kid is doing better than the parent and downward mobility meaning vice versa. Elaborating on the empirical nature of the oriented mobility model which brought together upward and downward mobility, he also went beyond and explained the practical difficulties in building the “right” model as the challenge lies in choosing the status variable that is to be compared between generations and getting the right data.
Appreciating questions generously, he answered a question if intergenerational mobility can help calculate intergenerational sustainability, saying that it was an interesting scope to study, with the status variables encompassing resources. Another interesting question was that inheritance of wealth and debt was not captured in this and he did agree that it could result that the rich might be better off in this model. With other questions revolving around the broad idea of using a different status variable, Professor Arnab Mukherji added that in the Indian context, data was sparse, where it was easier to get the difference between levels of parents and children than the absolute number of each of them.
Explaining a complex topic in simple terms, it was definitely a very interesting session with Professor Foster getting the audience to think how good a companion indicator Intergenerational Mobility is to income inequality. As a concept, intergenerational mobility may well have the same impact on public policy, as Professor Fosters 1984 paper did. After all, public policy is often concerned with addressing intergenerational persistence of poverty and this measure certainly provides us with a toolkit to measure intergenerational transfers. This session was great way for the Centre for Public Policy, IIM Bangalore to bring together leading scholars to discuss issues of relevance to public policy.
Post the discussion, we got a chance to catch-up with Professor Foster for a quick cup of coffee. Answering all questions patiently, he shared his experiences with dealing with resistance to change in convincing stakeholders about the necessity of the Multi-Dimensional Poverty Index (MPI). Adding his touch of humor and explaining the intricacies in policy implementation, Professor James Foster was a pleasure to interact with.