Your ERP estimate is still too high

I recently had a conversation with a colleague who had been told that “Credit Suisse recommended an Equity Risk Premium of 7%”.  I’m curious to know whether they truly view that as an appropriate ERP.  If your ERP is 7%, it’s still too high.

The authors of Triumph of the Optimists have joined forces with Credit Suisse to publish the Credit Suisse Global Investment Returns Yearbook 2010 (pdf) which is a brief update of their brilliant research.  You should definitely read the original book.

The updated research shows a very familiar picture to that of the book.  Here are a few important outcomes:

  • Realised excess returns of equities over bonds have been negative for most countries for the last decade.

Clearly, using realised excess returns (or historical ERPs) over a short period as a measure of future ERP is a bad idea.  I’m fairly sure the future ERP is positive.

  • For the World, the US, the UK, Australia, Belgium, Canada, Denmark, France, Germany, Ireland and South Africa (a few countries I chose to look at before I realised the trend is near-universal) have had declining historical ERPs over the last 110 years. Some have had a few bumps in between, but the overwhelming trend has been downward.  The last decade’s poor performance has obviously helped establish this trend, but it was pretty well established for most of these countries even without the last decade.

Using unadjusted historical ERPs over long periods is a dangerous idea because trends in the data make it a poor estimate of future experience.

  • Over the last 110 years, the average realised ERP in the World has been 3.7% (over bonds).

ERPs of 7% or 8% or higher are way too high to be realistic even compared to the historical record which probably overstates future ERPs

  • The historical ERP of South Africa over the last 110 years has been 5.4% over bonds. This includes a 25 year period from 1960 to 1984 where the ERP was approximately (I unapologetically hacked the authors’ pristine numbers to estimate this) 11%.

Even allowing for this exceptionally good period (associated with gold boom years) an ERP for South Africa of 7% or 8% is just too high. We should further be cautious of using a single country’s historical experience given the wide standard errors around these estimates given the variability of the data. South Africa, along with Australia, have the highest and second highest real equity returns per year over the 110 year period. Unless we expect this miracle to continue, we should expect more average growth (mean-reversion in this context is alive and well(.

  • The mean-reversion effect in equity returns is small if it exists at all (this from the 2009 report).

One should not expect to make strong returns in years following poor returns. The time to recover from significant falls, in real terms, can be decades (see Japan) or easily longer than 5 years (several examples).

Overall, the appropriate ERP for South Africa is probably still somewhere between 3% and 5%, with plenty of evidence to support estimates towards the lower end of that range.

If you use a basic prospective ERP estimate tool with the JSE All Share’s current dividend yield of 2.17%, you need to be expected real GDP growth of 7% in future. Any takers?

Published by David Kirk

David Kirk runs Milliman’s actuarial consulting practice in Africa. He is an actuary and is the creator of New Business Margin on Revenue. He specialises in risk and capital management, regulatory change and insurance strategy . He also has extensive experience in embedded value reporting, insurance-related IFRS and share option valuation. He has been involved in significant insurance projects across Sub Saharan Africa and in the Middle East.