Forecasting Inflation Using Univariate and Multivariate Time Series

Authors

  • Azam Ali Deputy Director, State Bank of Pakistan, Karachi.
  • S.M. Husnain Bokhari Joint Director, S&DW Department, State Bank of Pakistan, Karachi

Keywords:

Forecasting, consumer price index, quasi money, CMR, ARIMA, vector auto-regressive model

Abstract

The purpose of the study is to forecast inflation in Pakistan from January to June 2008. This study set out to redress the deficiency and explicitly use of time series techniques solely for forecasting purposes. The analysis based on the time series available from July 1995 to December 2007. Inflation is a trending series with the possibility that the trend is time varying. It is also plausible that monthly inflation could move around a time varying mean. Forecasting inflation is a difficult but essential task for the successful implementation of monetary policy. Inflation forecasts are central to macroeconomic analysis. There are a number of approaches available for forecasting economic time series. One approach, which includes only the time series being forecast, is known as univariate forecasting. An alternative approach is multivariate time series forecasting. To forecast inflation in Pakistan on monthly basis, we segregated our study in two parts, univariate with ARIMA Model and multivariate with VAR Model. Forecasted inflation for the month of January-08 and February-08 are close to the actual inflation, while in March 2008 there is found significant differences in forecasted and actual values of inflation. The reason of the high rate of actual inflation in March 2008 is the rise in oil prices.

Downloads

Download data is not yet available.

Downloads

Published

2009-06-30

How to Cite

Ali, A., & Bokhari , S. H. (2009). Forecasting Inflation Using Univariate and Multivariate Time Series. JISR Management and Social Sciences & Economics, 7(1), 79–89. Retrieved from https://jisrmsse.szabist.edu.pk/index.php/szabist/article/view/324

Issue

Section

Original Articles