Income mobility is usually considered a good thing. It implies higher social welfare as the ability of individuals to move up and down the income ladder mitigates the impacts of poor income distribution. But it is also true that when income jumps up and down unexpectedly, life becomes riskier and planning, difficult. This is why making a general link between the mobility we observe in the data and welfare is not straightforward.
A common approach used to show high mobility is a low correlation of present and past incomes is captured, for instance, by the Hart index (cov lnyt, lnyt-1). If we assume, as is often done, that an individual’s income is comprised of a transitory component (short-term blips up or down in a self-employed person’s income that we can smooth, or even measurement error), and a permanent component where each income shock is persistent (say, an income loss after an involuntary job change (an AR (1) process with autoregressive coefficient, ρ), then the Hart index can be broken into three parts.
The first part is comprised of measurement error or transitory shocks, the second part corresponds to income risk, and a third part is a “convergence term” that captures movement back to the average income for an individual of his/her level of human capital ( 1- ρ ).
Each of these components may affect welfare. If we assume that individuals can draw on savings or a short-term loan from friends or family to smooth a transitory shock, neither these shocks nor obviously measurement error are likely to have important welfare effects. We can consider income risk to be “bad mobility” and convergence to be “good” mobility. Whether a measured rise in mobility is good or bad depends on the relative welfare impact of the latter two effects.
In a working paper entitled “Income Risk, Income Mobility, and Welfare,” co-authored with Tom Krebs, and Pravin Krishna, we use data from Mexico to estimate the magnitudes of these distinct elements. Across 1 year, the vast majority of measured mobility is due to measurement error or transitory shocks, and roughly 1% is due each to income risk and convergence, respectively. This means that a large component of measured mobility is really of limited interest from a welfare point of view. The other two elements matter more. Removing all measured income risk raises welfare (as measured by lifetime consumption) by 12%. Eliminating the convergence effect reduces welfare by roughly 8%. Both experiments reduce mobility, both are of important magnitudes, and they work in opposite directions.
Clearly, other welfare functions and mobility measures would generate different findings. But these results, using very standard measures, suggest several important lessons. First, given the large impact of measurement error on measured income mobility, care should be taken in comparing mobility across countries or surveys. Second, it isn’t clear that an increase in measured mobility necessarily improves welfare. If it arises purely from an increase in random shocks—income risk—more mobility leads to a reduction in welfare. Third, the relevant measure of good income mobility in this case is convergence of incomes.
* The post has also been published in Jobs and Development blog.
Join the Conversation