In the large literature on the stock return–inflation puzzle, existing works have used constant coefficient linear regression models or change point analysis with abrupt change points. Motivated by the time varying stock return–inflation relationship and the drawbacks of change point analysis, we propose to use the recently emerged locally stationary models to model stock return and inflation. Although the model exhibits non-parametric time varying dependence structure over a long time span, it has local stationarity within each small time interval. Detailed empirical analysis is conducted and comparisons are made between various approaches. We find that the stock return–inflation correlation is negative during early sample periods and turns positive during late sample periods, but the turning time point is different for the total inflation rate and core inflation rate.
|Original language||English (US)|
|Number of pages||20|
|Journal||Journal of the Royal Statistical Society. Series C: Applied Statistics|
|State||Published - Nov 1 2019|
All Science Journal Classification (ASJC) codes
- Statistics and Probability
- Statistics, Probability and Uncertainty