An econometric forecasting model for a segment of the construction market

The purpose of this study was to develop an econometric forecasting model for a segment of the construction market. The same method used in developing this model should be applicable to other segments of the construction market. The three principal objectives of this model were: 1) to use existing available data, 2) to predict future construction volume of a segment of the construction market with “acceptable accuracy” and 3) to require minimum servicing. The model was developed by formulating a list of possible economic variables that would logically explain the variations in the one dependent variable which was the value of construction contracts for one quarter for office buildings in the United States. Historical data for both the dependent and the independent variables was assembled This historical data was studied to be able to specify the interrelationships, between the economic variables in a multiple regression analysis. At this stage of development of the model, some preliminary multiple regression equations which contained all of the likely independent variables were examined. There were some independent variables that logically should lead the model by a certain time interval. The interest rate at the time the project is being financed is an example, because the financing would be arranged some time prior to the award of the construction contract. The lead relationship for the selected independent variables was determined. The model (or models) being investigated at this time were preliminary, so the lead relationship must be verified in the final model selected. Reliable forecasts were required for each independent variable used in the model for every quarter for the horizon of the model. For those selected independent variables that led the model by, say, six quarters, no forecasts were required. because this unique time relationship permitted the use of historical data for that variable in conjunction with forecasts of the other independent variables. Forecasts of selected economic variables which are a part of the output of a large-scale macroeconometric model were used as input to the model developed in addition to other independent variables. This unique feature of this study allowed the complex interrelationships of the many factors of the United States economy to be captured and introduced into the model developed. The “best” model was tentatively selected using both judgement and statistical calculations. An evaluation of the independent variables was required to determine the reliability of forecasts and select the proper independent variables to be incorporated in the model. Statistical calculations are required to determine that the independent variables are significant in the model; that there is no systematic lack of fit; and that the model explains adequately the variations in the dependent variable. As stated earlier, the lead-time relationship of the variables in the final model must be verified. The econometric forecasting model developed in this study provides a forecast of the value of construction contracts for office buildings in the United States for each quarter for six quarters into the future in both current and constant 1958 dollars. Existing available data was used. These forecasts on an annual, basis are considerably more accurate than any method currently available. The model developed can be updated and serviced, by one person in a few hours any time a forecast is desired. The· objectives of this study have been accomplished.