Comparison of the grey model and the box jenkins model in forecasting manpower in the UK construction industry

Accurate forecasts of manpower in the construction industry are very important for the government, educational institutions and individual construction firms in their manpower planning. Although many good forecasting models have been developed, they require a sufficient amount of good-quality data to produce accurate results. Grey systems theory developed by Deng has become popular due to its ability to deal with systems in which some information is unknown. Grey models require only a limited amount of data to estimate the behaviour of unknown systems. This paper compares the grey model with the Box-Jenkins model in the forecasting of manpower in the UK construction industry. A GM (1, 1) grey model is proposed to forecast construction manpower one quarter ahead using manpower data from the Construction Statistics Annual published by the Office for National Statistics covering 72 quarters from 1991 Q1 to 2008 Q4. Within this period, two sets of manpower data were used: the total manpower available for employment (labour supply) and the total number of employees in employment (labour demand). An Excel programme was formulated to execute the forecasts using the grey model. An SPSS programme was used to conduct autoregressive integrated moving average (ARIMA) forecasts (Box-Jenkins model). The minimum mean absolute percentage error (MAPE) forecasted by the grey model for the total manpower and total employees time series was 1.52% and 2.14%, respectively, whereas the MAPE forecasted by the ARIMA model was 1.61% and 2.33%, respectively. Given the small forecasting error, it is concluded that both the GM (1, 1) model and the ARIMA model can accurately forecast manpower in the UK construction sector, but that the GM (1, 1) model performs slightly better than the ARIMA model.