1.8 Time series

A time series is created when an observation is monitored in economic statistics at regular intervals (annually, monthly or daily). The time series describes how the observed phenomenon changes over time. It is often presented as a line graph. Time series are used to describe, among other things, economic up and downturns and trends.

Unemployed persons, trend and original series

Source: Statistics Finland, Labour Force Survey

A time series can be broken down into several parts: trend, business cycle, seasonal variation and random variation.

Trend

Underneath a time series there usually lies a trend, which means the long-term change that is taking place. It can describe, for example, increase in the productivity of labour, volume of output or energy consumption.

Business cycle

Especially economic time series show fluctuations of the business cycle over a couple of years, which follow the up and down trends of economy. The business cycle is not always distinguished from the trend in short-term statistics. For instance, in the graphics on unemployed persons presented here the trend also includes fluctuations of the business cycle.

Seasonal variation

Regular variation within one year is called seasonal variation. Many phenomena, for example construction work, vary by season. Holiday periods and moving public holidays also have a visible effect on many statistical series.

Random variation

Most statistical time series also contain random variation. Random variation can have a significant impact in small economies. For instance, random variation is more important in the Finnish economy than in the economy of the USA. One large enterprise can produce variation in the economy of a small nation (e.g. Nokia in Finland).

It is important to separate these factors from each other when analysing time series. Random variation is not usually considered as interesting. The causes for seasonal variation are quite well known, too. However, business cycles and trends are objects of continuous interest.

In order to reveal the trend and business cycle in a time series, seasonal and random variations have to be eliminated from it. The simplest way to do it is to compare a measurement result with the one taken twelve months earlier. For example, the unemployment figure for this year's July is compared with the unemployment figure for last year's July.