Intertemporal Choice and Consumption Mobility

Abstract

The theory of intertemporal consumption choice makes sharp predictions about the evolution of the entire distribution of household consumption, not just about its conditional mean. In a first step, we study the empirical transition matrix of consumption using a panel drawn from the Bank of Italy Survey of Household Income and Wealth. In a second step, we simulate the transition matrix of the consumption distribution using parameters for the income process estimated on the same dataset. Comparison between the actual and the simulated transition matrix for consumption is favorable to the permanent income hypothesis once we allow for measurement error in consumption and a moderate degree of excess sensitivity to income shocks. The theory of consumption insurance is strongly rejected.

The Public Pay Gap in Britain: Small Differences that (Don't?) Matter

“…22 The second comment relates our paper to the recent literature on the use of copulas in models of income dynamics and sheds additional light on the status of the the τ 's as measures of relative income mobility. An alternative way of approaching the likelihood of individual income sequences consists in decomposing the joint density ℓ (y t , y t−1 , y t−2 ) appearing in the numerator of (6) as the product of marginal densities of log income y t , denoted f t (y t ), f t−1 (y t−1 ) and f t−2 (y t−2 ) and a copula density of income ranks…”

“…Whereas all these studies measure inequality using data on individual earnings, a third approach consists of looking at consumption inequality, which, under the Permanent Income/Life Cycle Hypothesis conveys information on long-run income inequality (e.g. Blundell and Preston, 1997; Blundell, Pistaferri and Preston, 2004; Jappelli and Pistaferri, 2004). None of these, however, examine lifetime differences between job sectors.…”

“…The use of copulas thus allows decomposition of the likelihood of an income trajectory into the product of cross-section densities at all time periods and the likelihood of the trajectory of relative 22 We choose to parameterize the autocovariances of income disturbances (the τ 's-see equation (9)) rather than the AR coefficients (ρ's) because the former are the moments that we end up matching in the estimation (see Appendix B). In fact, the τ 's are easier to handle as they come up "naturally" in the likelihood function (10) and also because the fact that τ (n) it is a correlation matrix imposes a number of natural restrictions on its elements (for instance, because τi,t,t−1 is a correlation coefficient, it has to lie within the interval [−1; 1]).…”

Fiscal Policy and MPC Heterogeneity

“…17 Above we noted a large concentration of responses in the MPC = 0 and MPC = 1 categories. Figures 3 and 4 plot the fraction of respondents 17 Using a completely different approach based on matching actual and predicted consumption transition probabilities, Jappelli and Pistaferri (2010) estimate an MPC of 0.4 for the low educated and 0.10 for the high educated, confirming a negative relation between measures of permanent income (measured here as education) and the MPC. reporting MPC values at the two extremes of 0 and 1, by cash-on-hand percentiles.…”

“…Although the latter assumption may seem ad hoc, it is supported by empirical evidence from Italy (Jappelli and Pistaferri 2010), and it is isomorphic to a model in which the poor have higher discount rates than the rich (ex ante heterogeneity). 30 Using a completely different approach based on matching actual and predicted consumption transition probabilities, Jappelli and Pistaferri (2010) estimated that the fraction of rule-of-thumb consumers is 40 percent among the low educated (which is a group that includes disproportionally low cash-on-hand consumers) and 10 percent among the high 29 We thank Fabrizio Perri for sharing with us the Matlab and Stata codes generating these results. Details are available in Appendix B.…”