Our inequality research at Bocconi
GRASS was organized this year at Bocconi. We presented another version of our inequality research. In this version we include lump sum indexation (and still keep the minimum pension benefits). This version of the model has also the implicit link between contributions and future pension payouts, so when the agents choose their labor supply, they do internalize fully the future effects of the current choices.
We find that increasing longevity is the big force behind changes in wealth and consumption inequality. The increase in consumption inequality in the baseline scenario of a defined benefit pension system amounts to over 10% of the Gini coefficient on consumption inequality. Also wealth inequality increases with increasing longevity in the DB system (agents with low consumption profiles expect to receive relatively generous pension benefits, hence see little reasons to save, whereas individuals with high consumption profiles, expecting consumption tax surge to cover the deficit in the pension system, make precautionary savings). The pension system reform introduces two major changes in agents' choices. First, they expect lower pension benefits and lower consumption taxes with the implementation of the reform. Second, they start seeing the link between labor supply and future pension benefits, which implies decline in labor taxation as well. These incentives result in higher labor supply and higher voluntary private savings. Such raises the consumption inequality and reduces the wealth inequality. Growth in consumption inequality is lower than that observed due to longevity.
On this setup we experiment with two instruments that are typically considered as means to reducing inequality, especially old-age inequality. These are minimum pension benefits and lump sum indexation. The effects of lump sum indexation are negligible, as this instrument is too weak to affect choices of agents and distribution of consumption among the retirees. Minimum pension benefits are very effective in reducing old-age and overall consumption inequality. They also contribute to higher wealth inequality, as they eliminate part of the precautionary motive. For a policymaker concerned with consumption equality rather than wealth equality these effects are noticeable. The reduction of inequality comes with positive overall welfare effects, but at the expense of high fiscal cost. Analyzing the channels through which these changes occur we find that this instrument could be particularly effective towards overcoming the inequality of opportunity.
The strong response of wealth inequality to changes in longevity stems in our model from the fact that agents are rational and have perfect foresight: expecting longer lives (and in some scenarios, also higher tax rates), they react with increased savings. The ability to save is heterogeneous along endowments as well as preferences, hence the initial inequality in wealth is amplified. Agents increase savings by altering their inter-temporal choices, which attenuates the initial inequality of consumption. If and to what extent people react to the longer life expectancy remains an area of active research and while the magnitude of response by some subcohorts may appear as relatively high, there is compelling evidence that over the past two decades we have witnessed an increase in wealth inequality and increase in life expectancy in most industrialized countries.
The result that minimum pension benefit operates mostly along the endowments margin implies that the behavioral response to guaranteed income in the old age stems from little or no response in terms of labor supply. Partly, it is a consequence of the assumption about the shape of the utility function. With log-linear intra-temporal preferences, income and substitution effects tend to cancel out, which attenuates any changes in individual labor supply. Whether such assumption about the shape of the utility function is plausible remains an area of empirical research. Gruber (2000) argues in favor of substantial labor supply elasticity to the value of benefits, but these estimates were put in question in subsequent studies (e.g. Campolieti, 2004). Estimates from other countries — e.g. Austria (Mullen and Staubli, 2016) — cannot be compared directly to our setup. Such estimates analyze a sudden, unexpected policy change and measure the effect for the directly affected cohorts. In our setup, the minimum pension benefits are negligible for the older cohorts, because these remain still in the DB system, with relatively high pension benefits. Hence, effectively the instrument operates with delay (even if introduced unexpectedly). An alternative to log-linear preferences is to assume GHH preferences, which relate labor supply directly to changes of wage rate. However, in a recent study, we demonstrate that in the context of OLG the shape of the utility function has much less of a role in determining the labor supply reaction, as the key element is making agents account for pension benefits in the labor supply decisions, which our setup does.