According to AK growth models, permanent changes in investment rates have permanent effects on a country’s rate of economic growth. Jones (Quarterly Journal of Economics, 1995, 110, 495-525) finds strong evidence against this prediction studying the time series properties of GDP growth rates and investment output ratios in fifteen OECD countries for the period 1950-1988. In this paper, we test the same hypothesis in four OECD countries using a longer span of data (1870-2002 for Canada, the UK and the US and 1885-2002 for Japan). Moreover, instead of using classic approaches, which are based on stationary I (0) or unit roots I (1) processes, we use methodologies based on fractional integration. After examining the order of integration of GDP growth rates and non-residential investment rations for these countries, we do not find much evidence against the “growth effects” prediction of AK models. In fact, we only find clear evidence against this theory for the UK case.