Skip Navigation

The Costs and Benefits of Retail Activities at Military Bases October 1997

The impact of many policies involving the commissary system depends on the relationship between changes in sales and changes in costs to the Department of Defense and society. Would expanding sales to reservists increase costs? Would raising commissary prices save the federal government money not only because of the higher prices but also because the loss of sales it would produce would decrease costs? What about policies that would reward managers for increasing commissary sales? Because not all costs vary with sales, information about the average cost per dollar of sales does not answer those questions.

The Congressional Budget Office (CBO) analyzed the relationship between unit costs (costs per dollar of goods sold) and sales for 220 commissaries in the United States, using 1995 data provided by the Defense Commissary Agency (DeCA). The cost for each store included all supplies, salaries, and other operating costs that could be tracked directly to individual stores, regardless of whether they were paid for with appropriated funds or with receipts from the 5 percent surcharge. In addition, the per-store figure included an estimate for the costs of the services provided by DeCA headquarters and of the personnel and financial services provided by the Defense Logistics Agency and the Defense Finance and Accounting Service. (DeCA allocated headquarters costs equally among stores under the assumption that larger stores did not require any more headquarters services than smaller ones. The other estimated costs were allocated partly by sales.) The costs of capital were not included.

To capture the relationship between costs and sales, CBO estimated a regression of the form cost/sales = a + b(sales) + c(1/sales) + d(sales²). That functional form allows for the kind of nonlinear relationship between unit costs and sales that is seen in Figure 4 in Chapter 2. It allows increases in sales in small stores to be associated with large declines in unit costs even though increases in sales in large stores have little impact on unit costs. That regression yielded the following statistics:
 

R-square equals 0.93, there are 220 observations, and all of the coefficients are significant at the 1 percent confidence level.

If that regression is used to predict costs based on actual 1995 sales at the 220 commissaries, total estimated costs are $773 million. If sales in each store are reduced by 10 percent (a total reduction of $454 million), the regression predicts total costs of $712 million. Thus, a 10 percent decline in sales is associated with an 8 percent decline in operating costs--or a $1 decrease or increase in sales (distributed among all stores in proportion to their sales) is associated with a 13 cent decrease or increase in operating costs, not including capital costs.

In the case of an increase, not all of the 13 cents in additional costs would have to be paid through appropriations; surcharge receipts would rise by 5 cents. Depending on how much of those receipts would be used for capital costs, the extra appropriations needed would range from 8 cents to 13 cents. (Those figures would be slightly higher, however, if DeCA's assumption that additional sales do not lead to any additional headquarters' costs is incorrect.)

Besides increasing DeCA's need for appropriations, each additional $1 in commissary sales raises forgone state and local taxes by about 5 cents. Thus, from a social perspective, an additional $1 in commissary sales raises subsidy costs by between 13 cents and 18 cents.

---