Abstract
In this article the covariate quantile autoregression approach was used to test whether consumption is a constant unit root process, as predicted by the Permanent Income Hypothesis (PIH). The evidence suggests that at low quantiles of the conditional quantile function of consumption the persistence of shocks are lower than that predicted by the PIH. This asymmetry is consistent with credit constraints and/or buffer-stock savings.
Acknowledgements
I am grateful to Antonio Fialho Galvao for his comments and to CNPq for financial support. The usual disclaimer applies.
Notes
1I focus on explanations that, potentially, affect the consumption distribution heterogeneously. However, many other arguments can be found in the literature, for instance, the hyperbolic discounting models (Laibson, Citation1997) and the nonseparability of the utility function over time and between the measure of consumption, leisure and other goods. For instance, Alessie and Lusardi (Citation1997) showed that when there is habit formation, consumption becomes a weighted average of past consumption and permanent income.
2Data were obtained from the National Income and Product Accounts (NIPA) data set. All variables are seasonally adjusted.
3I used the specification of the SIC based on the Laplace distribution, which is implemented by an l 1 type of regression.
4A more detailed description of the steps is given in Galvao (Citation2009).