Brian T. Ratchford

According to the "official" BLS index of labor productivity for retail food stores in the United States, the productivity performance of this industry has been dismal. By 1995 this index was about 13 percent below its peak 1971 value. The performance looked even worse until the BLS recently revised its output index upward by amounts ranging from approximately one percent in 1980 to approximately 10 percent in 1992-95. The revision for 1992-1995 was due to employment of a different methodology for calculating price deflators (BLS Handbook of Methods, Ch. 11). Thus the official government index indicates that labor productivity at retail food stores has declined over a long period. Because food retailing is one of the largest sectors of retail trade in the U.S., and because it is patronized by virtually everyone, productivity gains in this sector that lead to lower prices would lead would benefit virtually everyone in the U.S. economy, particularly the very poor who spend a relatively high share of their income on food. Consequently the issue of productivity growth in this sector, or lack of it, is a potentially important public policy issue.

Because the measured decline in labor productivity in food retailing after 1971 took place at the same time as the adoption of scanners, widely documented to be a major cost-saving technological improvement, the decline is puzzling. The explanation investigated in this paper is that changes in the mix of services offered by food stores that are not reflected in the output measure employed in the official index account for the measured decline in productivity. The output measure used by the BLS is a weighted sum of deflated sales across a number of categories. There is no adjustment to either the sales measure or the deflators for changes in service offerings that have taken place in this industry.

To test our explanation, we have constructed indexes of breadth of assortment and of the number of different services offered (deli, service meat, service bakery, etc.) by food retailers in the U.S., defined as firms classified into SIC 54, over the period 1959-95. Breadth of services and number of different services are two factors that have been documented by other studies to both have value to consumers and increase retailer costs. Both indexes show market increases during the time period of our study.

To quantify the impact of changes in breadth of assortment and different services on the productivity of food retailers, we estimate cost functions relating the total costs of retailers in SIC 54 to physical output (the BLS Index of output), breadth of assortment, services, price of capital used in SIC 54, price of labor used in SIC 54, and trend terms that capture shifts in the cost function. For any reasonable level of economies of scale we find that the costs of food retailing are declining if breadth and services are held constant. This indicates that, contrary to the evidence provided by the BLS Index of labor productivity, productivity has grown in this sector. In comparison with estimates for the entire private non-farm economy in the United States, these estimates indicate that total factor productivity growth in food retailing was comparable that of the entire private non-farm economy in 1959-79, exceeded that for this entire sector in 1980-89, and was lower than growth for this entire sector in 1990-95.

A decomposition of the BLS Index of labor productivity into its components indicated that the BLS Index of labor productivity would have grown substantially after 1969, instead of declining, if the effects of increases in services and breadth of assortment were subtracted out. Thus the decline in measured productivity change for the BLS index can be attributed largely to the increase in costs resulting from increased service offerings.

Thus our evidence indicates that productivity in food retailing did grow in a way that one might expect given the adoption of the scanning technology during the period of our study. While it may be a reasonable measure of the ratio of physical output to labor input, the BLS Index was unable to capture productivity growth in food retailing because it did not account for the added provision of services documented to be of value to consumers. If they are to accurately measure the performance of a given retail or service sector, and therefore provide useful information for policy makers, the productivity measures provided by the government must be able to account for changes in services provided as well as physical output. For managers, our results indicate that official productivity measures do not necessarily accurately reflect he performance of their industry. Attention must be paid to exactly what the official indexes measure, and to whether these measures accurately capture the conditions of a given industry. The results of our study suggest that official productivity indexes for any retail or service industry that has experienced major changes in the nature and composition of services offered per dollar sold should be scrutinized carefully before being used as an indicator of industry performance.