I don't say this because the pound is undervalued. Brexit (if it actually happens) will probably reduce the UK's long-term growth rate. That justifies a weaker pound. So too does the UK's big current account deficit.
Instead, there's another problem. Sterling is cyclical; it rises in good economic times. And these good times are often global ones, not just UK ones. There's a strong correlation between sterling's trade-weighted index and US industrial production - of 0.66 for annual changes since 1996. For example, sterling fell when US output fell in 2001-02, 2008-09 and in 2015-16. And sterling rose when the US grew well in the late 1990s, 2004, 2007 and in 2011-12.
In fact, if we control for US output growth, there is no significant correlation between UK economic growth and sterling: it's US growth that matters for the pound, not UK growth.
There's a simple reason for this. Sterling is a risky asset. Global investors' demand for risky assets tends to increase in good economic times and fall in bad ones. And good and bad times are defined more by what happens to the US economy than anything else.
This poses an obvious danger. If the US economy strengthens, as most economists expect it to, so might sterling. That would mean losses on our foreign currency holdings.
For many balanced portfolios, this wouldn't be much of a problem. The same increased appetite for risk that would cause losses on foreign currency would also push up share prices, causing investors with moderate FX exposure but decent equity holdings to make a net profit. Only someone with big FX exposure and low equity holdings would lose.
Now, you might object here that economists might well be wrong to expect the US economy to strengthen. True. But this highlights a perhaps underappreciated problem - that in one respect there's a sharp trade-off now between risk and return.
A portfolio that protects you from financial crises and bear markets would hold lots of foreign currency and government bonds, two assets that do well in bad times. But such assets might well lose you money in good times. This is an especial danger when bond yields are very low, because at lower yields bonds tend to have a higher duration - a greater sensitivity of prices to a given change in yield.
The same is true for gold. Insofar as this is a safe haven, it would do well in a crisis. But it could also do badly in good times if interest rates rise. This is because higher yields on cash or bonds increase the opportunity cost of holding gold, which would depress its price; it's no accident that gold's big price gains in the early 2000s came at a time of falling yields.
This gives us a dilemma. It's easy to build portfolios that would protect us from a crisis: they would contain lots of gold, foreign currency and government bonds. But such portfolios would do badly in good times, as they would be clobbered by rising yields and increased appetite for risk. This dilemma is magnified by the fact that the probability of a crisis is unquantifiable: it's a matter of Knightian uncertainty.
This dilemma is an acute one for those professional fund managers who must decide whether to go 'risk on' and chase returns or 'risk off' and pursue safety. We retail investors, however, might be in a happier position. We can simply hold a balanced portfolio of equities and safe assets. Yes, such portfolios offer only mediocre returns, because some portion of it will usually do badly - the exceptions being when loosenings of monetary policy raise all assets (perhaps temporarily). But as the late Herbert Simon said, it is sometimes better to satisfice than optimise. It is certainly easier.
