One of my A-list investment firm is the Singapore GIC. They publish annual report on their performance, including rolling 20Y performance metrics. I glanced at their 2015 metrics. The preceding 20Y (1995 – 2015) had some major ups and downs including dotcom boom and bust, 1997 Asian financial meltdown, the 2008 GFC, the boom leading up to it and the spectacular recovery afterwards.

If you look at a 5Y horizon you could see oversize annual returns (like 20%/year return) or negative returns…. all smoke and mirrors! I really want to see a long-enough window showing the cumulative effect of all of those short-term ups and downs. Well, 6.1% was the annualized return GIC achieved, which is comparable to a 6.9% return in their “reference portfolio” — 65% global stocks and 35% global bonds. (I think insurance products can deliver 3-4%.) 6.1% is considered high, so it’s probably unrealistic to achieve higher return than 6% over 20Y.

Other findings in the report:

• property – GIC managers see it as an inflation-hedge. I see it as good diversification from stocks and bonds.
• GIC portfolio is believed to be S$350b in total. 34% of that is allocated to US, 25% to Europe but not only stocks and bonds.

A fundamental implicit assumption in my comments above is ..

.. no regime change. In other words, the assets’ growth/devaluations are treated similar to natural phenomena like temperature, rainfall… that follow a consistent trend. There are random factors but the source of those factors are stable.

This is a fundamentally questionable assumption. Regime change is frequent and drastic IMO. If we look at equity return figures in U.S. stocks, the post-WW2 returns are very different from pre-WW2. Also, the equity returns in the 30 years after the war (reconstruction boom) were much higher than the recent 30 years. Yet another factor is technology (and energy) stocks, which occupy a significant position in U.S. equity indices. Did these growth factors exist in the 60’s? Regime change! The evidence in developing markets (Asia, Latam, BRIC…) are even “worse” i.e. too many regime changes. The volatility level is much higher now than N decades ago (N > 2). Therefore, I always feel cheated when a research uses 100 years of return data.

I think all serious economists know the limitations of “models” which ignores regime change, but I feel they can’t throw away most of the data and only use the last 20Y of return data.

We the lay public don’t realize the limitations. We see a long history of data and we feel reassured and emboldened. Surely if a doctor has treated 600 patients over 40 years, then she must be more trust-worthy than someone with experience on 200 patients over 10 years, right?

More and more scientists now recognize climate change is affected by human behavior so rainfall level fluctuations are different from 500 years ago i.e. pre industrialization. There are regime changes in nature too.