Is the Return on Investment of Norwegian Pension Funds a Result of Luck or Managers’ Skills? Is it Possible to Predict Which Pension Funds that Will Outperform the Market?

Abstract

The purpose of this research is to get insight into the Norwegian pension fund market. We will look further into the performance of pension funds by analyzing the drivers of the return of the funds. The data of study consists of 70 Norwegian pension funds delivered from Pensjon Norge AS with period from 2014-2017. In this thesis, we will study whether the performance of the pension funds is a result of luck or the stock-picking skills of managers by using bootstrap simulations. In addition, we will investigate the persistence of pension fund performance. Firstly, we use the four-factor model of Carhart (1997) to compute residuals, factor loadings and alphas of each pension fund. Further, we bootstrap a new dataset of alphas, and generate t-statistics of alpha, for each fund which is used to compare with the original four-factor alphas and t-statistics of alpha. We also compute the parametric and bootstrapped p-values for each fund to conclude our hypotheses. The main analysis in this study is over a period of three years (2015-2017). Besides, we also test three different subperiods as robustness tests. We operate with two different datasets, one dataset which contains gross numbers while the other dataset is net numbers. Our main findings in this analysis are that there is some degree of stock-picking skills of managers for several pension funds. By evaluating the bootstrapped pvalues, we can reject the null hypothesis for significant p-values and hence, rule the return of some funds as a result of managers’ skills. On the contrary, we are not able to make any conclusions if the bootstrapped p-values are insignificant. In general, there are indications of some persistence in the performance of the pension funds. However, we are not able to predict which pension fund that will outperform the market since the ranking of the funds varies across the percentiles.