Robert Shiller and Eugene Fama were two of three academics who won the Nobel Prize for economics for different accounts of asset price predictability: the former regarding its bounded long term predictability and the latter regarding its short term unpredictability.
It should be clear that providing the factors that generate return are fairly stable, over the long term, asset price valuations will be constrained by the limitations of that universe: money supply growth (which defines nominal GDP growth and nominal demand for assets) and real economic growth (which defines the supply of and the real return on assets) and its components (labour, capital and total factor productivity). In this sense high market valuations tend to reflect lower potential returns and vice versa. Over the short term, demand (both economic and financial market demand) and expectations of demand and supply can change quickly and significantly making short term assessment of price movements difficult.
If we have structural imbalances in the system the predictability of asset price returns becomes even fuzzier even over longer horizons, although the truism, that returns are bounded by the limitations of the universe remains true. At the moment it is extremely difficult to assess just which paradigm is going to come out of the current dysfunctional economic crisis we are in. One thing should be clear though, high structural economic and financial system risks, high uncertainty and high prices are not good for future returns.
I have always sided with the price dependency crowd, that is that group of investment professionals and economists who do not believe in “efficient markets and modern financial economics which bases the portfolio paradigm on general equilibrium models”. General equilibrium models assume price independency which would mean it is impossible to predict even the boundaries of long term returns and therefore extend short term unpredictability of asset prices to longer term horizons – hence the “Monte Carlo” simulations of the distribution of future random independent price movements.
Indeed, in developing my own integrated asset and liability models (way back in the early 1990s) I incorporated bounded return assumptions and this forms the basis of my risk and return modelling even today – that is as market valuations and economic cycles become more advanced future returns over and above low risk assets decline and vice versa. This is something which is lacking in MPT based Monte Carlo modelling.