Essays on inflation forecasts and Bayesian estimation of central bank policy preferences

This thesis consists of three essays which address important topics on inflation forecasts and central bank policy preferences in Ghana and South Africa. The second Chapter evaluates the Bank of Ghana’s (BoG) inflation forecasts. First, it investigates whether the BoG’s inflation forecast is efficient. It also assesses how its inflation forecast performs when compared to other benchmark forecasts. Finally, this Chapter assesses the effectiveness of an intervention technique, the Step Indicator Saturation (SIS) method in explaining forecast performance when there are structural shifts in the data. The third Chapter investigates whether New Keynesian Philips Curve (NKPC)-based forecasts that incorporates a time-varying-trend (TVT) improves on forecasts when compared to institutional and other empirical-based benchmark forecasts. The Chapter also investigates whether the incorporation of indicator variables improves on the performance of NKPC forecasts. The fourth Chapter investigates whether African Inflation Targeting (AFIT) central banks are committed to price stability as mandated. The Chapter also examines whether there are alternative policy objectives that distract from the commitment of AFITs to price stability. Finally, I address the question of whether there are significant differences in the level of commitment to inflation stabilisation among AFITs, Latin American Inflation targeters (LAIT) and advanced small open economy inflation targeters (ASOE). Inflation Forecasts Evaluation - The Case of Ghana. The main contribution of Chapter 2 is its use of a combination of Mincer-Zarnowitz regressions and step indicator saturation (SIS) variables to evaluate inflation forecasts of a developing economy central bank. Using a Mincer-Zarnowitz regression, I conclude that the one-quarter ahead BoG inflation forecast with SIS variables provides the strongest evidence in support of forecast efficiency. The BoG’s one-quarter ahead inflation forecast is efficient with or without the incorporation of SIS variables, however, a stronger efficiency is exhibited when SIS variables are incorporated in the forecast. The BoG’s two-quarters ahead inflation forecast is inefficient even with the inclusion SIS variables. Inflation Forecasting using the New Keynesian Philips Curve with a Time-Varying Trend and Structural Breaks. Chapter 3 evaluates the performance of a time-varying trend New Keynesian Phillips Curve (TVT-NKPC) inflation forecast versus other time series-based benchmark forecasts in a developing economy. This Chapter’s contribution is its focus on a developing economy as previous research have focused on developed economies and the use of saturation variables to analyse the effect of structural shifts in the data on inflation forecasts. For the immediate forecast horizon (one quarter ahead), the Atkeson and Ohanian (2001) (AO) GDP deflator inflation forecast is the most accurate forecast, when compared to the random walk forecast and the four variants of the TVT-NKPC forecasts. In the medium to long-term forecast horizon, even though the random walk (RW) forecast consistently outperforms the AO and TVT-NKPC inflation forecasts, it is not statistically significantly different from the TVT-NKPC inflation forecast. Bayesian Estimation of Policy Preferences of African Inflation Targeters. This Chapter is the first known research on African Inflation Targeting (AFIT) Central Banks’ policy preferences. It also compares results with those of Latin American (LAIT) and some Advanced Small Open-Economies (ASOE) inflation targeting central banks. The results confirmed that Ghana and South Africa are committed to their price stability mandates with the estimated price stability weights for both countries well within those recorded in Latin America and other ASOE. Output stabilisation is the next policy preference for Ghana, the policy weight is the highest among inflation targeting (IT) central banks. Aside inflation stabilisation, interest rate smoothing is the next important policy consideration for South Africa. As expected of IT central banks, Ghana and South Africa placed the least weight on exchange rate stability.