Determining Returns to Storage: Does Data Aggregation Matter?

Journal of Agricultural and Applied Economics, Dec 2007 by Klumpp, Joni M, Brorsen, B Wade, Anderson, Kim B

Figure 3 shows the frequency of wheat sales by month at each elevator and for the USDA data set. Results with both disaggregate data and aggregate data indicate that grain producers in Oklahoma hold stocks throughout the marketing year. This is inconsistent with Oklahoma State University Extension recommendations that advise producers to use mechanical marketing strategies, such as selling at harvest in the south and selling in lots of one-third on June 20, September 15, and November 15 (Anderson and Brorsen 2004). The results show net prices are highest at harvest (Figure 2), indicating that producers should sell at harvest regardless of location. However, net returns for September and November are close to harvest returns at the central and northern elevators, showing some support for the one-third/one-third/onethird marketing strategy.8 Prices peak around late November and early January (Figures 1 and 2), so storing past these months would be uneconomical for producers. Figure 3 shows that most wheat sales occur before prices start declining in early February but that some wheat sales do take place during the more uneconomical time period of February to May.9 One possible reason for holding stocks longer than it is economical is myopic loss aversion (Anderson and Brorsen 2005). Producers may store grain longer than is economical because they do not want to accept loss. But producers may also store grain for fear that prices will increase. In either case, psychological biasness would result in producers holding losing positions too long (Anderson and Brorsen 2005; Locke and Mann). Government programs are another possible explanation for producers' holding stocks after prices have peaked. However, little wheat was under federally subsidized loans in Oklahoma during the study period.

Conclusions

The storage-at-a-loss anomaly has long puzzled researchers. While most previous studies have offered the convenience yield explanation for "stocks held at a loss," some recent studies are questioning the convenience yield explanation and argue that the storage-at-a-loss puzzle is an illusion caused by aggregation of spatial data. Storage theory posits the optimal storage time increases with distance, so producers closer to the market sell first because they have lower returns to storage. However, while location differences among producers may explain why some producers store longer than others, data aggregation could mask spatial differences leading to the conclusion of "storage at a loss."

This article contributes to this growing debate by showing that data aggregation is not the only explanation for the storage-at-a-loss anomaly since we find no difference between returns estimated with disaggregate data and aggregated data from the USDA. Thus, results are inconsistent with the recent argument advanced by Brennan, Wright, and Williams and Williams and Wright that storage at a loss is due to data aggregation. Results with both aggregate and disaggregate data also indicated that grain producers in Oklahoma hold grain stocks throughout the whole marketing year. A possible reason for limited responsiveness to returns to storage is myopic loss aversion.