The search for a perfect portfolio

What is a reasonable expectation for outperformance in equity portfolios? When planners are doing their modelling with clients, the first part of the modelling is predicting how much we think active managers can out-perform.

The expectation of what you can get from active management is a huge question. And unfortunately the mathematics of active management — or the theory — is not really very useful. It basically asserts that if a manager has real talent, they can continue to drive their alpha up just by continuously increasing their risk level.

Carry this to the extreme, and you get to the point where you’d hire one asset manager to buy one stock for you, and then you’d hire a group of asset managers to buy one stock each, which would be their maximum alpha generating portfolio.

I don’t care what the mathematics says — I don’t believe that. I believe that the efficient frontier for active management is quadratic — that is, at some point it actually falls back on itself. The more you push managers out, the more you actually start to get a decline.

In most relatively efficient large cap equity markets, if a manager is able to construct an active portfolio that returns something in the neighbourhood of 150 basis points gross of fees, it would be doing pretty well. It would be in the top third of the asset management universe. It varies from market to market, of course, but you would be in that range. Bonds tend to be a little bit less.

Returning to my argument, let’s assume the risk premium for stocks to bonds is really only 300 basis points. If I increase my stock holdings by 10 per cent, in a passive space that’s only good for 30 basis points. If I can pick up those 30 basis points using active managers, and not take on the incremental risk caused by holding 10 per cent more stocks, this would be a good deal. And 30 basis points over a long enough time compounds out to be fairly reasonable money.

I take the view that recalibrating expectations for what active management can achieve will result in better investor behaviour. You have to be prepared to ride out managers who you really believe in when they’re going through a bad run. And they will go through bad runs — it’s predictable.

In practice, do people who decide on a global portfolio versus a mix of regional funds really get better results, or is it just a theoretically better way?

First of all, the history on this issue is limited, so I make no claims about the significance of outcomes at all. But it does look like some of the global portfolios have worked a little bit better.

However, most are also riskier portfolios. So, if you want to risk-adjust the results, they don’t look significantly better.

But I think the long-term opportunity for that approach is pretty significant. However, the number of asset managers who actually have true global research platforms on which to base those products is still relatively small.

The path dependent problem is the clash of demographics. How would you advise planners around the globe to think about the three factors that build into that equation: public provision of retirement income versus mandated private provision versus voluntary private provision? How are those big public policy trends playing into, as you put it, the shift that we have to make from the accumulation to the de-accumulation of wealth?

Right now, I would say the public policy dialogue on this is not very realistic. I don’t really see most countries doing the kinds of things that are likely going to mean those obligations will be met. So I think that leaves the end retiree in a position where, no matter what promises are being made to them by governments, they need to think about self-reliance.

I’m relatively indifferent, frankly, about the potential of mandated versus voluntary contributions, as long as the number is big enough to actually accumulate significant wealth.

I think the demographic problem has other interesting aspects for long-term investing. For example, it is fundamental that economic growth is a function of workforce growth plus productivity growth.

So for markets like Japan, where we can predict declines in the size of the workforce, for those economies to grow they have to have very significant increases in productivity, or they’re going to have to modify long-standing biases against immigration.

Countries have historically solved this problem by allowing in new workforce members. But so far for countries like Japan, this has not been part of the solution.

You have pointed out the problems with mean variance portfolio construction. But what is the alternative? What should people be doing other than creating portfolios by the traditional mean variance optimisation methods?

You need to shift out of a traditional definition of risk from a mean variance viewpoint, and think about risk as an asset/liability problem — that is, asset liability matching. It gets more complicated and it requires new toolkits to be given to advisers.

In that world, the behaviour of assets doesn’t need to be looked at solely in terms of mean and variance. You can look at other characteristics, such as optionality, and symmetry of distributions. Large financial institutions have historically used those kinds of tools. So there are answers out there.

I don’t believe that we will ever get a pure optimisation approach to many of these things.

Optimisers are incredibly precise. But the result is that they always buy the biggest error in your forecast — we can’t forecast the behaviour of assets on a forward-looking basis with sufficient accuracy to take the output of an optimiser with anything other than a grain of salt.

At the end of the day, these tools can be useful because they provide insight and understanding of the dynamics of your problem.

But you can’t really get away from exercising judgement — any more than other professionals, like a physician or an attorney, can avoid exercising judgement in their domain.

Randall Lert is the US-based chief portfolio strategist at the Russell Investment Group .