Tracking the hog and pig inventory in real-time using nowcasting

U.S. hog and pig inventory is a principal agricultural indicator published by the U.S. Department of Agriculture on a quarterly basis. During disruptions in the pork value chain, such as the initial outbreak of porcine epidemic diarrhea virus and the COVID-19 pandemic, there is a need for more frequent reporting of inventory levels. These more frequent estimates, if accurate, would improve production and marketing decisions, aid in policy assessments, and help reduce price volatility. This study proposes a mixed-frequency dynamic factor model to nowcast inventory estimates published in Quarterly Hogs and Pigs reports from 1993 to 2024, using more frequently published hog production and price data. In each round of nowcasting between two consecutive Quarterly Hogs and Pigs reports, we obtain ten nowcasting predictions of hog and pig inventory with progressively updated data. Our results show that reasonably accurate predictions can be obtained as early as two months before the release of a Quarterly Hogs and Pigs report, and they improve further as new information becomes available. The dynamic factor model outperforms several alternative methods and yields superior forecasts compared to analysts’ pre-report expectations during extreme events. Government agencies and market participants could utilize our nowcasting model framework to publish inventory estimates more frequently and at relatively low cost.