July 2006 SGE Monthly Luncheon
Review of Remarks by Roger Betancourt
Professor, University of Maryland
The Economics of Retailing and Distribution
Rapporteur: John Sabelhaus*
Why should economists ignore a sector of the economy that is roughly the size as health care, and second only to manufacturing as a share of GDP? The sector being overlooked is retailing and distribution, and at the July SGE luncheon Professor Roger Betancourt from the University of Maryland explained how his new book, The Economics of Retailing and Distribution, fills that gap. The book embodies many years of research by Professor Betancourt and his students on the subject, and brings important economic insights to a part of the economy that most of us take for granted.
The first thing to note about retailing and distribution is the overall size of the sector. Measured by value-added, retailing and wholesale comprised 15.5 percent of U.S. GDP in 1996. The average over 74 countries was 13.5 percent from 1950 to 1983. Also, these numbers are lower bounds, because certain industries (like insurance and banking) obviously market their services, but it is impossible to separate the marketing from the basic service. Also, as discussed further below, the value-added approach to measuring output in the retail sector does not measure all of the value being generated, because that value reflects an input to household production functions that is not being counted.
The key concept when thinking about retailing is “distribution services.” That term encompasses five broad categories of things that retailers do for consumers: provide accessibility of location, assortment of goods, assurance of product delivery, information, and ambiance. Some of these, like ambiance, are hard for economists to deal with in formal models—Professor Betancourt was quick to acknowledge that his book was inspired by collaboration with researchers in marketing. But these five all share an important feature: distribution services are outputs of the retail sector production function, and simultaneously inputs to the household sector production function.
One implication of the “distribution services” approach has to do with measuring productivity. Much of what the retail sector produces does not show up in standard productivity measures. How do we value the fact that a trip to WalMart today might encompass what would have been trips to several retailers in earlier decades? Clearly, the household sector has gotten an efficiency gain, but we are not measuring that directly in GDP. Like other improvements in quality of life (say improvements in air or water quality) many improvements in distribution services will never show up in standard output measures.
A second implication of explicitly analyzing distribution services has to do with market structure. Professor Betancourt discussed interesting research by his students that plays an important role in the book. For example, one can use customer survey data to explain the following phenomenon: when a large retailer (Home Depot) moved into a particular region, the mid-size hardware stores were driven out because the large retailer was more efficient, but the small retailers survived because they had something of value to offer. Those of us who are both homeowners and economists can quickly see why: we might travel to Home Depot to make purchases that do not require much information from store employees, but then turn around and go to the local hardware store when we have complicated technical questions about how something really works (ever try to find a knowledgeable person at a large retailer on a Saturday at noon?). Not only does this make sense, but professor Betancourt and his students have shown that the numbers add up—the gap between prices paid at the large and small retailers can be explained by distribution services, as measured by customer surveys.
In order to formulate predictive models for the retail sector, it is essential to keep in mind one of the key economic principles in Professor Betancourt’s book: almost all distribution services and retail items are gross complements, and thus there are enormous potential economies of scale. This principle is evident when looking at trends in retail production, with more and more shifting of final sales to large one-stop retailers and shopping malls. However, the importance of considering all five components of distribution services comes into play when thinking about where retail might be headed: although a one-stop retailer like WalMart provides assortment and assurance of product delivery, it is often less accessible, one often gets less information about certain types of products, and the ambiance could be considered (at least by some) to be lacking.
Professor Betancourt’s final set of thoughts about the role of retailing are related to vertical integration, and get to some very basic issues in the industrial organization literature. Vertical integration in the production/retail chain has the potential to improve efficiency for the usual reasons (for example, contracts within firms are simpler to establish and enforce than contracts between firms) but there is also some risk that too much concentration within the retail sector can affect factor and output markets in fundamental ways. One interesting example is in franchising: firms are able to extract rents from franchise owners that go beyond the franchise fees by requiring certain levels of initial investment. That shifts risk from the firm to the franchisee, and that cost should be considered (for example) when looking at franchise rates of return.
* The views expressed in this review are those of the author and should not be interpreted as those of the Congressional Budget Office.