ABSTRACT
In this chapter we direct the spotlight away from forecasting and focus attention on time series analysis. The objectives of a time series analysis are to identify and estimate patterns of behaviour, to characterise the historical development of time series, and to compare and contrast related series. The present study examines the annual number of marriages in Greece over the 22 year period 1968 to 1990. Data is available by geographical region (official Greek publications). We examine the regional breakdown later in the chapter. To begin the study we take a look at the aggregate figures for Greece as a whole, shown in Figure 6.1. Leap years are indicated. Several prominent features of the data are easily identified: a strong four year cyclical pattern with a low coinciding with leap years, and a dramatic change in the underlying trend pattern after 1979. Peering a little more closely there is a suggestion that 1972 and 1988 are unusually low even for leap years. Also, the period 1980-1983/4 has the appearance of a transitory phase. The Greek population expanded from 8.5 million in 1968 to 10 million in 1990, so the decline in marriages cannot be the result of declining pop ulation. A more likely cause is cost. Weddings in Greece are expensive occasions (civil weddings became possible only after 1981 and still account for just a tiny proportion of the total), and the Greek economy experienced a serious deterioration during the 1980s.
