This paper uses a rich weekly price database from the largest supermarket in Uruguay to analyze the relationship among prices, micro-level sales, and business cycle conditions. On average, 7% of products were on sale each month, and we show a positive and statistically significant but small relationship between sales and unemployment at the aggregate level, controlling for product and time effects. For each product category, there is a high and positive statistically significant association for these variables. Also, for some categories like food products, alcoholic drinks, beverages, and fruits and vegetables, we found a large, negative, and statistically significant relationship between the probability of being on sale and employment. The mechanism we tested is the following: As the economy worsens and unemployment rises (and/or employment falls), the supermarket uses sales to adjust prices while keeping reference prices fixed. Our results indicate that sales do not alleviate much of the price rigidity in economic recessions. We also studied sales behavior through a principal component analysis in the time domain. This analysis revealed that the first two principal components explained more than half of each sector’s variance in price dynamics. These components are significantly correlated with variations in unemployment and employment. Additionally, we found that movement in the principal components across sectors systematically Granger-cause employment changes.
Read full abstract