Forecasting real exchange rate trends using age structure data – the case of Sweden

Abstract

Theory predicts that life cycle saving and consumption behaviour could cause real exchange rate variations as the age structure varies. Time series regressions show that the Swedish demographic structure has significant explanatory power on the real exchange rate during 1960 to 2002. A model using age shares as regressors is used for medium-term out-of-sample forecasts, which perform well both compared to naïve forecasts and forecasts based on an autoregressive model.

Acknowledgements

We are grateful to Hans Dillén, Thomas Lindh and seminar participants at the Institute for Futures Studies, Sveriges Riksbank, Uppsala University and the 21st International Symposium on Forecasting in Atlanta for valuable comments on this paper. Financial support from Jan Wallander's and Tom Hedelius’ foundation and Sparbankernas forskningsstiftelse is gratefully acknowledged

Notes

See for instance Jorion and Sweeney (1996) and Wu and Chen (2001).

For different ways of representing the demographic structure in the empirical literature, see for instance Fair and Dominguez (1991), Higgins (1998), and Lindh and Malmberg (1999).

Children have been excluded from the equation to avoid perfect collinearity since the age shares sum to one.

The dummy variables take the value one for eight periods following the event.

Apart from aggregate saving and the current account – which were mentioned above – demography has been shown to affect a number of other variables, such as growth, inflation, real interest rates and budget balance. See for instance McMillan and Baesel (1990), Bloom and Sachs (1998) and Lindh and Malmberg (1999).

All variables are judged stationary using the Johansen (1988) likelihood ratio test as a panel unit root test, as proposed by Taylor and Sarno (1998) and Österholm (2004). This does not, however, exclude the possibility that a regression is spurious as shown by DeBoef and Granato (1997).