Whatever the size of the equity risk premium in the future, clearly a portfolio which is not 100% in equities will see potential returns drop.

What you have also got to realise though is that a portfolio which is not 100% in equities will see potential risk drop too. Hence the need to assess the amount of risk you want to take on in order to get the returns you want in the time frame you want them. At this point things get a bit more complicated.

The expected returns on a portfolio are simply the weighted average of the expected returns of its constituent investments. So, a portfolio which is 100% in equities and has an average annual real rate of return of 16%; and a portfolio which is 100% in T-bills and has an average annual real rate of return of 6%; if combined (50% equities, 50% T-bills), will have an average annual real rate of return of (16+6) / 2 = 11%.

So far so good, but a portfolio’s risk profile doesn’t behave in the same way. We introduced the idea of the risk/return payoff with a graph that looked like this ...

... but the relationship between risk and return is not a linear one. In fact the relationship between risk and return looks more like this:

It is hard to explain exactly why this is without getting into some pretty complicated mathematics which frankly you don't need to know about. Intuitively though, it makes sense. You can't get infinite returns but you can lose all your money; so the line must start to flatten out as you take on more risk.

The question of whether the potential for higher returns is worth the additional risk is the nub of the asset allocation question.