Life cycle income and consumption patterns in transition

There is vast literature examining how households’ income and consumption changes over the life cycle. These studies, however, are usually restricted to developed economies, such as the US, Japan or the UK. The general objective of this paper is to add to this literature by investigating life cycle profiles and income mobility in a transition economy, facing rapid structural economic and social changes, such as Poland. There are two more specific objectives of this article. The first one is to study the evolution of households’ distribution of income and consumption over the life cycle, focusing on two first moments, i.e. the mean and variance of the logarithm of income and consumption. The life cycle profiles are analyzed separately for more and less educated households, which allows us to determine how the choices made are affected by the level of education. The second part of this study focuses on the mobility of Polish households between income quintiles using transition matrices. The ultimate aim is to compare the obtained results with similar studies for developed countries (in particular the US, see Diaz-Gimenez et al., 2007) and thus identify the crucial factors driving income mobility in Poland, and in transition economies more generally. A useful byproduct of the results obtained in this study is a set of parameters that can be used to calibrate macroeconomic models, in particular general equilibrium models with households experiencing uninsurable income shocks, including models estimated with the method of simulated moments (see Cagetti, 2003). Cagetti, M., 2003. Wealth Accumulation over the Life-Cycle and Precautionary Savings, Journal of Business and Economic Statistics 21, 339-353. Diaz-Gimenez, J., Glover, A., Rios-Rull, J.V., 2011. Facts on the Distributions of Earnings, Income, and Wealth in the United States: 2007 Update, Federal Reserve Bank of Minneapolis Quarterly Review, 34, 2–31. To construct age profiles of mean and variance of households’ income and consumption, I estimate a partially linear model using the two-step estimator proposed by Speckman (1988). In this approach the dependent variable, i.e. a particular moment of consumption or income distribution (for instance the variance of logarithm of income), evaluated for a given year, age of household’s head and cohort group, is explained by two components. The first one is parametric (linear) and consists of cohort and year dummies. The other part is a nonparametric relationship linking the dependent variable to age. The whole specification is estimated in a two-step procedure, being a combination of the ordinary least squares and a standard kernel smoothing estimator. This method was first adopted for a life cycle analysis by Fernandez-Villaverde and Krueger (2006) and was also used i.a. by Yang (2009). The study is based on the Polish Household Budget Surveys (HBS) covering the period of 2000-2010. HBS is conducted every year by the Polish Central Statistical Office on a sample of around 37 thousand of Polish households. It collects households’ monthly income and spending data, as well as a number of other socio-economic characteristics. To control for family size, I apply the square root equivalence scale, used i.a. by the OECD. Since, as shown e.g. by Coulter Cowell and Jenkins (1992) or Banks and Johnson (1994), the results of analyses using microdata are sensitive to the choice of an equivalence scale, I also estimate the life cycle profiles using alternative family decomposition measures. The equivalence scale effect turns out to be significant, but does not change the qualitative conclusions. HBS is a cross-section rather than a panel study and so it does not track households over time. Therefore, to estimate the transition matrix of households income I rely on the Social Diagnostics data from five waves between 2000 and 2009, using spectral decomposition techniques to convert multi-year transitions into one of an annual frequency. For comparative purposes, both matrices with and without controlling for family size and age effects were calculated. Banks, J., Johnson, P., 1994. Equivalence Scale Relativities Revisited, The Economic Journal 104, 883-90. Coulter, F. A. E., Cowell, F., Jenkins, S., 1992. Equivalence scale relativities and the extent of inequality and poverty, The Economic Journal I02, I067-82. Fernandez-Villaverde, J., Krueger, D., 2006. Consumption over the life cycle: Some facts from consumer expenditure survey data. Review of Economics and Statistics 89, 552–565. Speckman, P., 1988. Kernel smoothing in partial linear models. Journal of the Royal Statistical Society B 50, 413–436. Yang, F., 2009. Consumption over the Life Cycle: How Different is Housing? Review of Economic Dynamics 12, 423-443. In line with the previous literature, the life cycle profiles of households’ mean income and consumption in Poland exhibit significant humps. However, in contrast to the US economy, where consumption increases with age at a relatively stable rate (Fernandez-Villaverde and Krueger, 2006), in Poland we observe rapid growth of consumption (and also income) during early years of the life cycle (the age of household head between 18-30), while in a later period (up to the age of 55) this growth is significantly slower. Interestingly, this effect is present only for relatively educated households (with household head having at least secondary education). The savings rate of Polish households is smallest at the beginning of the life cycle, grows up to the age of 30 and then remains quite stable until the retirement age. However, in contrast to the life cycle hypothesis and data from other countries, the savings rate levels off or even increases after retirement. This might reflect the need to insure against sickness or some random events that are difficult to predict in a fast changing environment. As first pointed out by Deaton and Paxton (1994), income and consumption inequality reveals an increasing trend during the life cycle. As regards the rate of growth of inequality, there may exist various patterns. For instance, in the US the approximately linear trend is observed for the inequality of consumption (Deaton and Paxton, 1994 and Storesletten et. at., 2004), while in Japan the variance of logarithm of income begins to increase sharply at the age of 48 (Abe and Yamada, 2009). According to my results, yet another effect can be observed in Poland. The inequality of income and consumption grows rapidly in the early stage of household life (age between 18 and 30), which for household heads aged 18-25 years is mostly driven by the decision to postpone employment and gain additional skills by studying. Interestingly, for households with age 30-55 the inequality of income remains stable and the inequality of consumption even slightly decreases. A comparison of the estimated transition matrices for Poland to those for the US (the latter taken from Diaz-Gimenez et. al., 2007) reveals that Polish households appear to be relatively more mobile between income quintiles. However, the simulations also show that this difference is mainly driven by different shapes of the income distribution. In particular, the distribution of income in the US is characterized by greater dispersion. Abe, N., Yamada, T., 2009. Nonlinear Income Variance Profile and Consumption Inequality over the Life Cycle, Journal of the Japanese and International Economies 23, 344-366. Deaton, A., C. Paxson, 1994. Intertepmoral choice and inequality, Journal of Political Economy 102, 437–467. Diaz-Gimenez, J., Glover, A., Rios-Rull, J.V., 2011. Facts on the Distributions of Earnings, Income, and Wealth in the United States: 2007 Update, Federal Reserve Bank of Minneapolis Quarterly Review, 34, 2–31. Fernandez-Villaverde, J., Krueger, D., 2006. Consumption over the life cycle: Some facts from consumer expenditure survey data. Review of Economics and Statistics 89, 552–565. Storesletten, K., Telmer, C.I., Yaron, A., 2004. Consumption and risk sharing over the life cycle. Journal of Monetary Economics 51, 609-633.