Comparison of the Performance of Macroeconomic Finance Models for Financial Planning (MFM-FP) and ARIMA-Common Size in Forecasting ROE of Real Estate Developers in the Stock Exchange of Thailand

Abstract

An econometric model was created for strategic planning and ROE forecasting of real estate development companies in Thailand. The efficiency of the MFM-FP model was compared against an ARIMA-Common Size time series model by creating a simultaneous equation model using 2TSL, for analysis and forecasting. Assumptions on the proportion of the financial statement structure in each quarter were tested for differences. If proportional structure is not different, it can be used for forecasting together with ARIMA. MAPE was also used to assess the accuracy of the forecast results. The built MFM-FP model consisted of 15 identity equations, 22 behavioral equations, and 1 condition equation, with a total of 38 equations. For the ARIMA-Common Size model, the main assumption that the proportion of the structure of the financial statements in each quarter was not different, was accepted, with statistical significance at 0.05. Therefore, the data could be used for forecasting with the ARIMA model (3,1,2). For all variables assessed during the estimation period and the ex-post period, the MFM-FP model was more accurate than the ARIMA-Common Size model. It is therefore summarized that the MFM-FP model performed better than the ARIMA-Common Size model.