There is a huge literature on the equity risk premium that probably could sustain its own journal, trying to document it or explain it. But somehow, people keep finding new ways to look at the equity risk premium, so it sometimes worth checking out what they last came up with.
Georges Prat looks at the equity risk premium at various horizons and studies how and why they evolve differently. The study highlights that there is a time-varying term structure of equity risk premia, and that it depends on interest rates (expectedly) and a hidden state variable that the author attributes to psychological factors. Now, it is easy to blame changes in tastes for anything one cannot explain, but this is hardly convincing, here or elsewhere. The study uses the S&P 500 index and Treasury bonds and calculates premia at one and ten year horizons. If taste shocks make that risk tolerance of some people changes, they may get completely out of particular maturities. Looking at big aggregates is then not appropriate to measure how risk-averse they are. For example, if I find that long term risk is getting too high for me, for example because I am approaching retirement, I will get out of the blue chip stocks I was holding and into ten-year government bonds. Blue chips will then be priced by a different demand. There is thus a composition effect, that is, the risk tolerance of those holding these stocks is different, but it is because these are different people. That is not psychological, this is demographic.