This chapter discusses how times series methods can be used for forecasting purposes. As mentioned in Chapter 1, their main distinguishing feature is that they take a statistical view which leaves limited room for economic analysis. Their appeal stems from the ease with which they allow generation of numerical forecasts for a host of variables. The flipside is that these forecasts do not lend themselves to much if any economic interpretation, which is a major handicap when it comes to disseminating and explaining them. Therefore, times series methods usually play an ancillary role, as an auxiliary tool or a benchmark. They can be useful to produce forecasts that are needed but for which the available resources are limited, such as ad hoc extrapolations in the context of business cycle analysis or forecasts of exogenous variables in a macroeconomic model. They can also serve as a check on forecasts obtained through other methods.
KeywordsCentral Bank Granger Causality ARIMA Model Exponential Smoothing Cyclical Component
Unable to display preview. Download preview PDF.