We explore the ability of several univariate models to predict inflation in a number of countries and at several forecasting horizons. We place special attention on forecasts coming from a family of ten seasonal models that we call the Driftless Extended Seasonal ARIMA (DESARIMA) family. Using out-of-sample Root Mean Squared Prediction Errors (RMSPE) we compare the forecasting accuracy of the DESARIMA models with that of traditional univariate time-series benchmarks available in the literature. Our results show that DESARIMA-based forecasts display lower RMSPE at short horizons for every single country, except one. We obtain mixed results at longer horizons. Roughly speaking, in half of the countries, DESARIMA-based forecasts outperform the benchmarks at long horizons. Remarkably, the forecasting accuracy of our DESARIMA models is surprisingly high in stable inflation countries, for which the RMSPE is barely higher than 100 basis points when the prediction is made 24-and even 36-months ahead. Resumen En este trabajo se explora la capacidad de varios modelos univariados para predecir la inflación de un conjunto de países a varios horizontes predictivos. Se pone especial atención en predicciones provenientes de una familia de diez modelos estacionales que es denominada Driftless Extended Seasonal ARIMA (DESARIMA). Mediante el cálculo de la raíz cuadrada del error cuadrático medio (RECM) fuera de muestra, se compara la capacidad predictiva de los modelos DESARIMA con la de modelos referenciales de series de tiempo tradicionalmente utilizados en la literatura. Los resultados indican que las predicciones basadas en modelos DESARIMA muestran una menor RECM para horizontes cortos en todos los países considerados, excepto en uno. Se obtienen resultados mixtos para horizontes mayores. Aproximadamente en la mitad de los países las predicciones basadas en modelos DESARIMA superan a los modelos referenciales en horizontes largos. Se destaca que la precisión predictiva de los modelos DESARIMA es sorprendentemente alta en países con inflación estable, para los cuales la RECM es algo mayor que 100 puntos base para predicciones realizadas a 24 y hasta 36 meses adelante. * We thank the comments of seminar participants at the Central Bank of Chile. Any errors or omissions are responsibility of the authors. The views and ideas expressed in this paper do not necessarily represent those of the Central Bank of Chile or its authorities.

In this paper we analyse the utility of international measures of inflation in predicting local ones. To that end, we consider a set of 31 OECD economies for which monthly inflation data are available. Three main conclusions emerge. First, there is an important share of countries for which relatively robust evidence of predictability is found for both core and headline inflation. Second, the share of countries for which there is evidence of robust predictability is about the same for core and headline inflation, although gains in root-mean-squared prediction error are higher for headline inflation. Third, while the evidence indicates that an international inflation factor may be a useful predictor for several countries, it also indicates that, for many à We thank the comments and suggestions of Claudio Raddatz and two anonymous referees. We also

a los participantes del seminario en Cambridge y a los dos árbitros anónimos de Estudios Públicos por sus comentarios. También agradecemos a Consuelo Edwards por su trabajo de edición y a Romina Oses por su trabajo de referencias bibliográficas. Las opiniones expresadas en este artículo no representan necesariamente las del Banco Central de Chile o de sus autoridades. Cualquier error u omisión es responsabilidad exclusiva nuestra.

The use of different time-series models to generate forecasts is fairly usual in the forecasting literature in general, and in the inflation forecast literature in particular. When the predicted variable is stationary, the use of processes with unit roots may seem counterintuitive. Nevertheless, in this paper we demonstrate that forecasting a stationary variable with driftless unit-root-based forecasts generates bounded Mean Squared Prediction Errors errors at every single horizon. We also show via simulations that persistent stationary processes may be better predicted by unit-root-based forecasts than by forecasts coming from a model that is correctly specified but that is subject to a higher degree of parameter uncertainty. Finally we provide an empirical illustration in the context of CPI inflation forecasts for three industrialized countries.

, and an anonymous referee for their kind help and comments. We also thank the comments of seminar participants at Central Bank of Chile. Any errors or omissions are responsibility of the authors. The views and ideas expressed in this paper do not necessarily represent those of the Central Bank of Chile or its authorities.