Modeling the Public Welfare System: Part 2
Attia I. Sweillam
Helman I. Stern
DOI: 10.2190/XK1E-T0FT-RXNF-NFC5
Abstract
A previous paper of the authors presented a pure markov chain model to forecast New York State public welfare cases. The model is comprised of three major components: transferred cases, closed cases, and opened cases. It was found that the opening and closing portions of the model contributed the majority of the forecasting error. This report compares the previous approach to a modified markov model with opening and closings determined by measures of policy change (version 1) or by linear regression on socio-economic and demographic factors (version 2). In this model, intercase transfer rates are conditional upon openings and closings. A validation procedure selected the modified model (version 1) to forecast total caseloads. The forecasts indicated mean absolute per cent errors for individual welfare categories and total caseloads of: 1.9 and 0.65 (1972), 1.3 and 1.0 (1973), respectively. The errors are well below the five per cent level required for departmental budget preparations. The model may be implemented with internally generated data.This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.