On the moo-ve: testing for spatial agglomeration economies in the U.S. dairy industry
To examine the effects of county-specific demographic, environmental, and market factors as well as to test for the influence of spatial agglomeration economies on the geographic distribution of the U.S. dairy industry, a spatially explicit, county-level model of the dairy production sector was developed. Quantities of milk marketed through the Federal Milk Marketing Order during the month of May for counties in 45 states during 1997 and 2002 were specified as a function of natural endowments, business climate, production resource availability, milk price, and market access. The model was estimated according to spatial autoregressive (spatially lagged dependent variable) and spatial Durbin (lagged dependent and independent variables) specifications accounting for the censored nature of the dependent variable and heteroskedastic errors. Based on RMSE, the spatial error model was selected to make out of sample predictions for 2004. The change in milk marketings between 1997 and 2002 was regressed on the 1997 independent variables using non-Tobit versions of the same models with limited success.
Results indicated a small but statistically significant presence of spatial agglomeration effects in the dairy industry in both 1997 and 2002 and revealed changes in the degrees of influence of several variables between the two periods examined. Population and the wages of agricultural workers became significant in 2002, while the elasticities of feed availability diminished, consistent with an increase in western-style dairy production. Interestingly, the spatial parameter decreased from 0.052 in 1997 to 0.028 in 2002 suggesting spatial agglomeration economies had a diminishing role in determining the amount of milk marketed in a county.
School:Kansas State University
School Location:USA - Kansas
Source Type:Master's Thesis
Keywords:dairy spatial agglomeration location decision economics agricultural 0503
Date of Publication:01/01/2007