A trio of Federal Trade Commission economists (Christopher Taylor, Paul Zimmerman, & Nicholas Kreisle) have revisited Justine Hastings’ 2004 AER analysis of the ARCO/ Thrifty vertical merger in their paper, “Vertical Relationships and Competition in the Retail Gasoline Market: Comment.” (HT: Danny Sokol). Hastings’ analysis is viewed as particularly important because it is one of the few empirical results that suggests that vertical integration is associated with lower retail prices and that regulations restricting vertical integration might improve welfare.
The FTC economists challenge both the underlying empirical results using a similar differences in differences methodology and also take a closer look at the welfare implications of Hastings’ theoretical model. Their conclusion:
According to Hastings’ analysis of the W-L data, the ARCO/Thrifty acquisition increased prices at nearby competing stations by $0.05 per gallon, on average. Our analysis of the OPIS data provides a very different estimated price effect of this transaction, suggesting that if there was any price effect, it was an order of magnitude smaller. In a number of instances we cannot reject the hypothesis of no effect at standard levels of significance. While we used a different data set, we have found no reason that would explain this discrepancy. Regardless of which data set more accurately depicts the transaction’s actual impact on prices, our theoretical analysis reveals some difficulties in inferring welfare effects from price changes when consumers attach value to particular brands of gasoline.
This sort of merger retrospective work is very important for formulating competition policy and regulation that is well grounded in both theory and evidence. As the FTC authors notes, however, their is still a significant shortage of this type of work. For those interested in the state of evidence on vertical integration, Francine Lafontaine and Margaret Slade also recently published a survey on the topic in the Journal of Economic Literature (available here).
