Stephen Opata – “Inflation forecasting performance of a developing economy central bank- The Case of Ghana” – PhD Seminar

Abstract:

This study used historical inflation forecast data of the Bank of Ghana and quarterly inflation outcomes to assess forecast performance through the use of Mincer-Zarnowitz regressions and the incorporation of step indicator saturation (SIS) methodology. This paper also compared the forecast performance of the Bank of Ghana against other forecasts, notably a benchmark forecast using the random walk model proposed by Atkeson and Ohanian (2001) and the IMF WEO forecasts. We concluded that the central bank’s forecast outperformed the other forecasts followed by the random walk forecast judging from the ratio of the root mean squared forecast errors. The Diebold Marino test further supported the conclusion that the central bank’s one-step-ahead forecast was superior with a statistically significant difference in the accuracy of the one-step-ahead central bank forecast compared to the random walk forecast. Using the concept of encompassing, we concluded that the central bank’s forecasts reflected all information embedded in the random walk forecast but the latter forecast did not reflect all information embedded in the central bank’s one-step-ahead forecast.. The Bank of Ghana’s one-quarter-ahead inflation forecast was found to be efficient with or without the incorporation of SIS variables, however a stronger efficiency was exhibited when SIS variables were incorporated in the forecast. The stronger efficiency exhibited by the forecast that incorporated the SIS variables points to the importance in addressing outliers and structural breaks when evaluating inflation forecasts especially in developing economies such as Ghana. There was evidence that the Bank of Ghana’s forecasting performance as measured by the inflation forecast error improved with time.