Collaborative Research: International Prices, Costs and Markups

National Bureau Of Economic Research Inc, Cambridge MA

Investigators

Abstract

International Prices, Costs and Markups SBE/SES 0820468: Gourinchas The questions of how markets function and the limits to arbitrage are central for many areas in economics. In the context of international economics, these questions take the form of determining whether markets separated by country borders are more or less segmented than regional markets within a country. The empirical literature that attempts to answer this question faces two important challenges. First, empirical tests require that the goods compared are identical. Second, without data on costs or profits, it is difficult to ascertain whether evidence of price segmentation indicates varying mark-ups or differences in manufacturers' or retailers' costs across different locations. Both limitations are addressed in this project. The PIs use a new confidential dataset with detailed product-level (UPC or barcode) information from one of the largest North American food and drug retailers, with stores located in the United States and Canada. The dataset contains weekly data on 6347 goods sold in 250 US stores (in 19 states) and 75 Canadian stores (in 5 provinces), between 2004:1 and 2007:6 (178 weeks). Since the dataset contains information on prices, quantities sold and cost of goods, the PIs can decompose variations in relative prices across locations into their local relative costs and relative mark-up components. Besides new data, the PIs propose to estimate the economic importance of the border using a new methodology, borrowed from the treatment effects literature, namely a regression discontinuity design. This design exploits the precise location of stores and their distance to the border. Preliminary results indicate significant differences in the behavior of prices within the United States, within Canada, and across cross-border pairs. The median of the absolute price gap is twice as large for cross-border stores as compared to within-US pairs. The estimated border effect over time is found to co-move closely with the US-Canada nominal exchange rate. These finding are important for understanding the sources of fluctuations in real exchange rates, the transmission of monetary, fiscal and technology shocks across countries and the impact of nominal exchange rate movements on relative prices and consequently on the adjustment of external balances.