John Cotter at the University of Dublin and colleagues have complied measures of the effectiveness of international diversification in major assets and have found a "precipitous" decline in such effectiveness since the 1990s. The ability to reduce risk by diversifying across countries, they estimate, has halved since the 1990s. Because of this, the riskiness of apparently diversified portfolios has increased.
This is another aspect of globalisation. In recent years Asian economies and oil exporters have been running huge current account surpluses which they have reinvested around the world. This has generated higher correlations between asset prices in different countries simply because those assets have risen and fallen as demand for them from countries with surpluses have risen and fallen. This has long been true of equities and bonds, but it's also becoming true of property. For example, Chinese investors have been buying property not just in London and New York but also in Canada. This has contributed to property prices becoming more correlated, which means property investors can't so easily spread risk by investing in different cities.
Another factor behind rising correlations has been simply that investors have followed similar strategies. For example, investors in high-yield bonds have generated co-movements between bonds in emerging markets and in southern Europe.
On top of these, some common risks can generate co-movements. For example, fluctuations in liquidity risk (as measured by the TED spread, the gap between interbank rates and Treasury bill yields) can cause all high-yielding assets to rise and fall together as tighter liquidity causes investors to close positions while ampler liquidity causes them to increase such positions. And the risk of tighter US monetary policy can also hurt assets around the world.
All this poses the question: what might reverse these trends and so make international diversification a better way to spread risk?
One possibility would be a retreat from globalisation in goods markets. If a country becomes less well integrated into the world economy its economy should become less sensitive to fluctuations in global growth and more sensitive to local shocks. This might increase the chances of national stock markets moving in different directions.
It would, however, require a big decline in goods market globalisation to achieve this. And such a decline would probably be hugely damaging to the global economy and stock markets.
Another possibility would simply be that if investors realise that international diversification doesn't work so well they'll stop doing it and retreat into their local markets. And as such markets become more driven by local investors, they should be more likely to move independently of each other.
There are, however, big obstacles to this. One is that current account surpluses must be invested overseas: this is true by identity. This alone would generate co-movements across countries. It is only if global current account imbalances decline that everyone will be able to become a local investor.
And even then they might not want to. Many global investors want to invest overseas because their domestic markets are either illiquid or prone to political upheaval: if you fear that a change of government would cause your wealth to be confiscated, you want to put it into other countries where it is beyond your government's reach.
All this suggests that international diversification will remain a poor way to spread risk.
This doesn't mean you should confine your investments to the UK. Investing overseas helps protect us from a fall in sterling - which is especially useful as such falls often occur in bad times for the economy in general. And even if markets are correlated, it's possible that overseas ones will outperform the UK.
It does mean, though, that spreading your investments around the world isn't enough to reduce risk very much.
So, what can we do to reduce risk?
You might think the obvious solution is simply to invest in different assets. Even here, though, we have less ability to cut risk than we had a few years ago. Correlations between UK equities and gilts have risen recently and are now positive. They might remain so. If bond and stock markets are driven more by expectations about monetary policy or inflation and less by swings in risk aversion the two might stay positively correlated. This would reduce our ability to spread risk.
Herein lies a case for cash. The great thing about cash is that it protects us from correlation risk. Bonds and equities, or property and equities, can fall at the same time, as can stock or bond markets around the world. But your losses on cash are limited to the gap between inflation and the nominal interest rate. This gives you downside protection.
Perhaps, though, investors have got here before us. One feature of recent years has been their willingness to hold cash despite nugatory interest rates. This willingness might be due in part to a recognition that one cannot so easily reduce risk merely by diversifying.
