A Multivariate Time Series Analysis of Fertility, Adult Mortality, Nuptiality, and Real Wages in Sweden 1751–1850: A Comparison of Two Different Approaches
The relationships between fertility, adult mortality, nuptiality, and real wages in Sweden from 1751 to 1850 are studied using multivariate time series analysis methods. This approach allows an empirical determination of the relationships between the four time series, a considerable advantage where existing theory provides insufficient guidance.
Both the vector autoregressive moving average (VARMA) model and the vector autoregressive (VAR) model are used and the two approaches are compared with regard to parsimony, interpretation, and post-sample forecasting performance. We also compare the VARMA and the VAR approaches with regard to their sensitivity to outliers.
Multivariate time series model; VARMA model; VAR model; historical demography; outliers.