Forecasting revenue using time series techniques in South Africa’s commercial transactional banking industry
Malepane, Benny Christian Thapedi
Budgeting and forecasting is a crucial exercise carried by finance teams in organisations to predict their future performance. The use of time series models to forecast financial data has sparked interests from many researchers around the globe. Even though time series studied were carried on financial data, they haven’t been widely applied to total Net Interest Revenue (NIR) from transactional banking. This highlighted a need to apply time series models on this data. This research studies the efficiency of four time series models (AR, ARIMA, VAR and MA) on total NIR for transactional accounts in commercial banking. Macro-economic factors for instance, inflation and prime interest rates are also studied via the cointegration test to identify their influence on model performance using the VAR model. The best performing model is chosen based on statistical metrices, forecasting errors and significance of parameters. The forecasting errors reveal that VAR is the best performing model amongst the four with MA coming second. This highlighted the need to consider inflation and interest rate movements when forecasting NIR on transactional accounts. However, the VAR models did not produce significant parameters, therefore, MA model had to be chosen as the overall best performing model as it produced better forecasting errors and significant parameters than the VAR model. As a result, this study recommends the use of MA models to forecast total NIR for transactional accounts in commercial banking. This study recommends an amendment of certain policies within the banking sector in order to benefit from the results of this dissertation. The polices include budgeting, hiring and remuneration policy. Due to the government dependency on banking forecasts, it was also recommended that government adjusts its budgeting policy to stay consistent with the way banks forecast as they do their forecasts based on the results banks produce.
A research report submitted in partial fulfilment of the requirements for the degree of Master of Management in Finance and Investments to the Faculty of Commerce, Law and Management, University of the Witwatersrand, 2022