A spectral approach to stock market performance

Abstract: We pose the estimation and predictability of stock market performance. Three cases are taken: US, Japan, Germany, the monthly index of the value of realized investment in stocks, prices plus the value of dividend payments (OECD data). Once deflated and trend removed, harmonic analysis is applied. The series are taken with and without the periods with evidence of exogenous shocks. The series are erratic and the random walk hypothesis is reasonably falsified. The estimation reveals relevant hidden periodicities, which approximate stock value movements. From July 2008 onwards, it is successfully analyzed whether the subsequent fall in share value would have been predictable. Again, the data are irregular and scattered, but the sum of the first five harmonics in relevance anticipates the fall in stock market values that followed.