Global Inflation in Forecasting - A model comparison

Abstract: An essential input to monetary policymaking is inflation forecasting, and one important factor regarding inflation forecasting debated by researchers is the use of a global component. This thesis builds upon Ciccarelli and Mojon (2005, 2010) as well as Gillitzer and McCarthy (2019), who investigate the performance of a global component when forecasting domestic inflation. This paper examines the inflation forecasting performances of Global Vector Autoregressive models by comparing it to the Atkeson and Ohanian (2001) model augmented with global inflation and the benchmark Autoregressive model. This paper appraises the robustness of the forecasting models via comparisons of the root mean squared error with the purpose to determine if global inflation can be helpful to forecast domestic inflation. This is performed by using quarterly data for 22 Organisation for Economic Co-operation and Development countries and the Euro area over a period from 1980 to 2016. We apply two different in-sample estimation windows; a 15-year and a 10-year estimation window with the purpose to evaluate if, or how, the result changes depending on the window size. The findings from this paper indicate the importance of the global factor and how it seems to be sensitive to both the window size and the forecasting horizon. Overall, this paper finds that none of the models’ performances systematically outperforms the other models in forecasting inflation four to eight quarters ahead across countries and samples. The result also supports the central conclusion by Gillitzer and McCarthy, that global inflation helps forecast domestic inflation.