FORECASTING WOMEN, INFANTS, AND CHILDREN CASELOADS: A COMPARISON OF VECTOR AUTOREGRESSION AND AUTOREGRESSIVE INTEGRATED MOVING AVERAGE APPROACHES VICTORIA LAZARIU, CHENGXUAN YU and CRAIG GUNDERSEN ∗ Under the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), each state receives a fixed federal grant for the operation of WIC in the upcoming federal fiscal year. Accurate forecasting is vital because states have to bear the expenses of any underestimation of WIC expenditures. Using monthly data from 1997 through 2005, this paper examined the performance of two competing models, autoregressive integrated moving average (ARIMA) and vector autoregression (VAR), in forecasting New York WIC caseloads for women, infants, and children. VAR model predicted over $120,000 less per month in forecast errors in comparison to the ARIMA model. (JEL H7, C5) I. INTRODUCTION Forecasting the number of participants is an integral part of the oversight of every public assistance program in the United States. This is particularly true for programs where there is only a fixed amount of money available for the program. These programs differ from “entitle- ment programs” where the amount of money available is not capped. One nonentitlement pro- gram in the United States, and the topic of this paper, is the Special Supplemental Nutri- tion Program for Women, Infants, and Children (WIC). Under WIC, each state receives a federal grant from the U.S. Department of Agriculture ∗ Previous versions of this paper were presented at the 2005 Student Poster Day of School of Public Health, University at Albany; the USDA Northeastern Regional Office meeting; the New York State Department of Health SPEED Rounds; and the Albany Chapter meeting of the American Statistical Association. The authors wish to thank participants at those venues for their comments and Mary Lou Woelfel for her comments. Craig Gundersen gratefully acknowledges financial support for this project from the New York State Department of Health. Lazariu : Research Scientist, New York State Department of Health, Albany, NY 12204. Phone 1-518-402-7109, Fax 1-518-408-0254, E-mail vgl01@health.state.ny.us Yu : Research Scientist, New York State Department of Health, Albany, NY 12204. Phone 1-518-402-7109, Fax 1-518-408-0254, E-mail yxc04@health.state.ny.us Gundersen : Associate Professor, Department of Agricul- tural and Consumer Economics, University of Illinois, Urbana, IL, 61801-3671. Phone 1-217-333-2857, Fax 1-217-333-5538, E-mail cggunder@illinois.edu (USDA) for the operation of WIC for the upcoming federal fiscal year. Because of the limitation of funds, states may not be able to serve all people eligible to receive WIC services. Therefore, states have to estimate the caseload levels that can be supported with available fund- ing. If a state exceeds this fixed amount of money, the state may bear the fiscal burden of any additional expenditure. If a state does not spend 97% of the grant received, the state has to return funds to USDA which may impact the next year’s appropriation level. As a con- sequence, accurate forecasting is vital in WIC. There are, of course, many different ways one can forecast caseloads in WIC or in other pro- grams. The most straightforward method is to assume that future caseloads are only depen- dent on past and present caseloads. In consid- erations of other assistance programs, however, other factors, including participation in similar programs and macroeconomic conditions also ABBREVIATIONS ARIMA: Autoregressive Integrated Moving Average KPSS: Kwiatkowski, Phillips, Schmidt, Shin MAE: Mean Absolute Error MAPE: Mean Absolute Percentage Error ME: Mean Error SNAP: Supplemental Nutrition Assistance Program USDA: U.S. Department of Agriculture VAR: Vector Autoregression WIC: Women, Infants, and Children 46 Contemporary Economic Policy (ISSN 1465-7287) Vol. 29, No. 1, January 2011, 46–55 Online Early publication April 1, 2010 doi:10.1111/j.1465-7287.2010.00203.x  2010 Western Economic Association International