Search
Search type

Methods@Manchester

Time Series Analysis

Time series analysis covers a wide range of statistical and econometric techniques designed to capture the patterns observed over time in one or more data series.

Examples

Examples include monthly patterns in average recorded temperature (seasonality), the number of airline passengers since the 1960s (trend) and patterns of growth and recession in national output (business cycles).

The methods of time series analysis are inherently dynamic, so that, say, UK inflation in October 2010 is viewed in terms of how inflation and other macroeconomic variables have moved over preceding months and years. As such, time series analysis can be viewed as a form of regression modelling, but with the focus on dynamics and recurring temporal patterns.

Forecasting

Time series analysis is fundamental to forecasting in macroeconomics and finance, and it plays a key role in macroeconomic policy decisions.

For example, the Bank of England’s most recent (August 2010) forecasts for annual consumer price inflation, shown in the shaded area of the plot below, are based on time series models.

This plot is quite sophisticated, showing not only the Bank’s “central” forecast, but also ranges around this, with deeper shades of red indicating higher probabilities. The Bank’s monetary policy decisions are, in turn, based on assessments of the likely future trajectory for inflation and the economy more generally. (The plot has been downloaded from the Bank of England web site).

Experts at Manchester

Additional information 

There are many introductory texts on time series analysis, although these require some knowledge of statistics. Some suggestions:

  • Enders, Walter: Applied Econometric Time Series, 2nd ed., Wiley 2004
  • Harris, Richard and Sollis, Richard: Applied Time Series Modelling and Forecasting, Wiley 2003.