Zipf's law, agglomeration, and the location of Washington wineries

In studies of the geographical concentration of manufacturing employment in U.S. states, researchers find the wine industry to be very concentrated.  At the state level, we would expect clustering or agglomeration to be largely due to climate, soils and other agricultural factors. In this paper, we take a more micro approach, by examining the location of wineries in Washington State, where the wine industry has grown rapidly in recent decades. Does geographical concentration continue to exist in a more homogeneous natural environment for winemaking? Might other kinds of agglomeration economies exist in the wine industry? After a brief discussion of the general literature on the geographical concentration of firms, we show the distribution of Washington wineries for various levels of spatial aggregation, e.g., zip codes, cities, and counties. We then estimate power function regressions of the logarithm of the number of wineries in a location on the logarithm of the rank (in terms of number of wineries) of that location. Regression coefficients for the zip code and city levels of aggregation are very close to -1, which suggests that Zipf’s Law applies to Washington wineries at these levels of aggregation. It appears not to hold at the county level. As evidence of Zipf's Law is likely the result of agglomeration economies in an industry, we then speculate on reasons why this is the case, and discuss hypotheses about the determinants of winery location and agglomeration that we might address in future research. These include, among others, the relative pull of consumer vs. vineyard location, age of the winery, size of the winery, labor pooling and human capital spillovers.