Imrecon

Many years ago, for my maths degree, I read Thomas Kuhn's ‘The Structure of Scientific Revolutions'.  It came back to me this summer when I read Andrew Smithers' recent book, ‘Wall Street Revalued - Imperfect Markets and Inept Central Bankers'.  Smithers writes about a Kuhnian paradigm shift in financial economics, on the Efficient Market Hypothesis.  As I'll argue, at stake is up to 2% on the cost of equity assessment for regulated businesses.

Paradigm shifts in the market

Thomas Kuhn described how science moves on from one view of the world, or paradigm, to the next.  A paradigm has its roots in a brilliant insight which establishes in the mind-set of the scientific community, eventually accepted as self-evidently true.  It becomes part of the community's world-view, structuring the way scientists think about the world.  The sun orbits around the earth, light is a wave rather than a particle, the universe is Newtonian. 

Andrew Smithers puts the Efficient Market Hypothesis in a Kuhnian context.  The paradigm that has prevailed in the market has been similarly all-encompassing, structuring the way economists think about how markets work.  That markets efficiently price assets was a powerful idea that explained a great deal and made intuitive sense.  Investors are rational and actively profit-seeking and mispriced assets would be pounced on quickly by arbitrageurs, so the scope for anyone to beat the market would be strictly limited.

Like a scientific paradigm, the brilliance of the original insight and its predictive power gave economists confidence to extend the theoretical edifice.  But, according to Smithers, the paradigm started to run into trouble in the late 1970s as anomalies started to show up.  For example, by 1979, Robert Shiller observed 'excess volatility' in stock returns, more volatility than is consistent with market efficiency (what Smithers calls variance compression). 

As Kuhn describes in science, rather than challenge the paradigm, the evidence is at first doubted or dismissed and, as the evidence becomes less refutable, the problem becomes a crisis in the community and new heretical ways of thinking emerge.  Painfully, the old paradigm dies.  The earth orbits the sun, light is a particle, the universe follows the laws of general relativity.

But what replaces market efficiency?

In my view, the new paradigm lies in the field of behavioural economics, e.g. the thinking behind Robert Shiller's book 'Irrational Exuberance'.  They give us compelling insights into how market participants really work and it is events such as the credit crunch that make it easier to accept that markets do not always price risk efficiently. 

As Mr Smithers says "few economists still believe the financial markets approach perfect efficiency". 

While market participants can construct highly diversified portfolios of investments, the other side of the market, the investors themselves, are highly undiversified.  They are all human beings, imperfectly rational, with common psychological traits.  Furthermore, professional participants are operating within a social and cultural context across the world that is getting more strongly integrated as markets, organisations and people's networks become more global.  The behaviourists point out that individual judgement and valuation errors are liable to be highly correlated, becoming fundamental market errors.  As Robert Shiller states in his 1989 ‘Market Volatility', "Prices change in substantial measure because the investing public en masse capriciously changes its mind".  In other words, markets do not price efficiently.

Why does this matter for the regulated sectors?

In the hidden mechanics of the equity risk premium, there lies up to 2% that depends on an assumption of market efficiency.  That 2% could die with the paradigm, with big implications for regulatory cost of capital assessments.

The following chart shows the pattern of returns on the UK stock market over the last 109 years (using the Dimson, Marsh & Staunton database).  From the start of 1900 to the start of 2009, annualised returns were some 5.1% in real terms[1].

Yet, five years ago, Ofwat assumed an underlying cost of equity of 7.7% for the relatively safe water sector and Ofgem assumed 7.5% for the similarly safe electricity distribution businesses.  These assumptions are now being revisited.  The supposed rationale for these numbers being rather larger than 5.1% is that investors should actually expect returns in line with the arithmetic average of historical annual returns rather than their annualised, geometric average. 

To illustrate this awkward statistical concept, suppose an asset halves in value or doubles in value each year.  If over a long period the value has halved and doubled an equal number of times (i.e. is now worth the same as it was at the start), an investor may judge that a £1 investment in that asset has an equal chance of being worth £2 or 50p after a year.  So, looking forward, the averaged expected value is £1.25, an expected return of 25% per annum.  After a number of years, the investment would have an equal chance of gaining overall or losing overall but, if the returns are independent of each other, the scale of the potential gains will always be greater than the potential losses.  There would be a statistically expected return of 25% per annum.

If, on the other hand, the underlying or fundamental value of the asset is stable and the oscillations in value arise from changes in the market's perception of that value, an investor could not sustainably expect to make a positive return at all.

As NERA points out in its June 2008 report for Water UK[2], "the arithmetic versus geometric controversy is basically about market efficiency and how one believes the stock market functions".  How much of the volatility reflects changes in fundamental value and how much changes in perception?

A perfectly efficient market means the volatility is about fundamental value, and implies stock returns should follow a random walk.  Projecting that volatility forward will mean the scale of potential gains above the historical annualised figure will always be greater than the potential losses (as in our halving/doubling example).  For the UK market, the arithmetic average has been some 7.0%, nearly 2% higher than the 5.1% annualised figure.  So 7% would look like a good assumption for the future.

But that assumption unravels as the paradigm shifts. 

There is, of course, more to it than this high level description, but my prediction is that the days of 7% or more for regulated utilities' cost of equity are numbered.  Revolution is in the air.

Ian Rowson
8 September 2009

[1]         An annualised 5.1% would draw a straight line between the 1900 and 2009 points on the chart.  I personally prefer the line of best fit, which is close to 6%.

[2]         This quote is NERA's quote from a source that appears not to be fully identified.