English Français |
An-Chi Tung, Institute of Economics, Academia Sinica
Kevin Yu-Ching Hsieh, School of History and Sociology, Anhui Normal University
A commonly asked question is: Are Millennials a Lost Generation? More deeply, we ask: Does it matter when one is born regarding lifetime welfare? We conduct an Age-Period-Cohort (APC) analysis of consumption, as consumption is the most commonly used welfare indicator. Our task is to separate impacts of age, time and cohort. We begin with the conventional approach by Deaton and Paxson (1997), which sets a constraint on time effect to solve the identification problem of perfect linear dependence among age, period and cohort. Yet this constraint is harmful when data series is long. Therefore we resort to the Intrinsic Estimator (IE) approach, which yields a unique solution to the model. We refine the estimation by comparing over subperiods, introducing extra explanatory variables, and experimenting with individual-based consumption data. We use Taiwan’s household survey data of 1976–2016. We choose Taiwan because it has gone through rapid economic changes, as well as huge-scale social and demographic development, hence the results carry rich policy implications to societies at all development stages Preliminary results suggest that younger household heads face less favorable time or cohort effects, which calls for appropriate policies. Further, methodologically, IE approach is better when the society is non-stagnant.
Keywords: Econometrics, Decomposition analysis/methods, Economic analysis, Age structure