Correction of Measurement Error in Monthly USDA Pig Crop: Generating Alternative Data Series

Journal of Agricultural and Applied Economics, Aug 2008 by Kim, In-Seck, Plain, Ronald L, Bullock, J Bruce, Jei, Sang Young

The imputed pig death loss contained in the reported monthly U.S. Department of Agriculture (USDA) pig crop data over the December 1995-June 2006 period ranged from -4.93% to 12.75%. Clearly, there are substantial measurement errors in the USDA monthly pig crop data. In this paper, we present alternative monthly U.S. pig crop data using the biological production process, which is compatible with prior knowledge of the U.S. hog industry. Alternative pig crop data are applied to a slaughter hog model and tested comparatively to USDA pig crop. Test results reject the validity of USDA pig crop data in favor of the alternative data.

Key Words: biological production process, measurement error, monthly USDA pig crop data, pig death loss

JEL Classifications: Q11, Q13, C12

Most livestock supply analyses have concentrated on the incorporation of farmers' expectation of future prices into the supply function to explain cyclical patterns of livestock prices and quantities. The analyses also examine supply elasticities of slaughter animals, or meat products, in response to price changes. Aadland and Bailey well documented the previous research of these approaches in the U.S. beef cattle industry. In cases of the pork hog market, Chavas investigated the nature of the expectation formation within the U.S. pork market with the use of annual data from 1960 to 1996.

However, unlike annual or quarterly models, monthly changes in slaughter hogs (barrow and gilt) are rarely affected by economic variables, once the size of the breeding herd is determined by the hog producers within the monthly framework. Fluctuations of monthly slaughter hogs are determined by seasonal variations of production variables in the biological production process, rather than by producer expected price changes.

Hence, the challenge of modeling supply response for monthly slaughter hog forecasting is not to specify the nature of monthly changes in producer expectations about slaughter hog prices or pork prices in month t; rather, the challenge is to model the biological production process so as to predict the number of animals slaughtered in month t given the number of pigs farrowed in month (t - 6), which was determined by the number of females bred in (t - 10), which, in turn, was determined by the size of the breeding herd in (t- 11).1

Thus, monthly estimates of the size of the U.S. pig crop are necessary to develop an estimate of monthly slaughter hogs in future months. Before 1995, only quarterly estimates of the size of the pig crop were available. In December 1995, the U.S. Department of Agriculture (USDA) began reporting estimates of sows farrowing and pig crop on a monthly basis.

However, these monthly estimates of farrowing and the size of the pig crop lead to some logical and biological inconsistencies in historical data. For example, comparison of the calculated monthly slaughter hog produced in month t,2 as well as the reported pig crop 6 months before, leads to imputed death loss during the growing/finishing phase of pork production over the December 1995-June 2006 period. During that period, it ranged from -4.93% of reported pigs farrowed to 12.75%.

This observation suggests a need for an alternative pig crop projection system that reflects (is constrained/generated by the use of) widely observed and accepted norms regarding trends and seasonal patterns in conception rates, pigs per litter, and pig and sow death losses during the hog reproduction and growing process.

The objective of this study is to present an alternative monthly U.S. pig crop data series that is compatible with prior knowledge of the U.S. hog industry. A second objective is to test alternative pig crop data comparatively to the USDA pig crop. In the balance of this paper, we discuss potential problems of monthly USDA pig crop data, describe the alternative data generation process, compare alternative pig crop data with reported USDA numbers, develop monthly slaughter hog model, and test the appropriateness of two different data series on the basis of the developed monthly slaughter hog model.

Problems of Monthly USDA Pig Crop Data

Monthly USDA pig crop data are obtained from quarterly surveys of pork producers. Producers are surveyed quarterly regarding the number of sows farrowed and the size of the pig crop during each of the previous 3 months. Data for each of the months are published in the USDA Hogs and Pigs Report. In some quarters, previously reported monthly data regarding farrowings and the pig crop are adjusted on the basis of recently observed slaughter numbers.

The USDA Hogs and Pigs Report production data system treats pig death loss during the growing process as the residual required to balance hog slaughter numbers in month t and the reported pig crop in month (t - 6). However, this simple comparison between observed slaughter in month t and the reported pig crop in month (t - 6) might misrepresent actual pig death loss contained in the USDA pig crop data because the pig crop is used not only for slaughter but also for breeding, and some of the observed slaughter hogs in month t are not raised from the pig crop in month (t - 6).