Nick Flamang, UC Berkeley Graduate Student, UC Berkeley Economics
Abstract: An influential literature has argued that estimates of the life-cycle profile of consumption are biased by "shopping effects" whereby consumers pay lower prices when their incomes fall during retirement. Leveraging a large panel data set, I construct a novel measure of consumption that decomposes expenditure into savings effects through lower barcode-level prices and bulk purchases, quality effects from substituting towards cheaper products, and quantity changes. Estimating event studies around retirement, I find that paying lower prices for identical products and savings from bulk purchases explain only 13% of the expenditure drop at the onset of retirement and successively smaller shares further into retirement, with the bulk of the expenditure drop representing declines in the quality of the consumption bundle and actual quantities. In addition, the shopping behavior at retirement looks strikingly like that during unemployment spells, which represent unanticipated and potentially permanent shocks to income. I enrich these results by estimating life-cycle profiles of household-level prices and the quality of consumption. The life-cycle profile of prices is mildly hump-shaped with little evidence of large declines in prices paid late in life. Meanwhile, the quality of consumption closely follows the life-cycle profile of income. These results suggest that the post-retirement expenditure drop is a true consumption drop and that non-durable expenditures are a good proxy for consumption more generally.