**The Pillars of Finance: The Misalignment of Finance Theory and Investment Practice**. By Guy Fraser-Sampson. *Palgrave Macmillan, 2014; 260 pages*

*Guy Fraser-Sampson’s new book argues Modern Portfolio Theory fails to capture the real nature of risk. Better methods have long been available, but MPT is dogma to academia and industry.*

In the early 20^{th} century Frank Knight, Ludwig von Mises, and John Maynard Keynes shaped the concept of risk. These economists depicted risk as a quality of assets that cannot be measured, but it nonetheless has to be addressed by managers and academics alike.

Their ideas were influential until the post-war period, when they were supplanted by the work of Harry Markowitz, a graduate student at the University of Chicago. In 1952, Markowitz changed how finance professors and professionals view risk with his article, “Portfolio Selection” in *The Journal of Finance*.

Markowitz’s view of risk as volatility forms the foundation of Modern Portfolio Theory (MPT). Its working definition of risk, expressed as *E-V*, the “expected returns – variance of returns rule,” occupies the commanding heights of both theory and practice in academia and the industry. So influential were these ideas that Markowitz was awarded the Nobel Prize in Economics in 1990.

**Quantifying the heart**

Markowitz’s concern was that the discounting methods with which Knight, von Mises, and Keynes were familiar did not address risk quantitatively. If it cannot be calculated, it cannot be risk, he thought. In the paper, he “illustrate(s) geometrically relations between beliefs and choice.”

Deriving objective equations from fixed probability beliefs, “in part subjective,” Markowitz showed that risk not only can be quantified, but can be done so easily and precisely. At the time, no one thought this strange. He theorized that risk is best understood as the difference between expected return and actual return, the volatility of returns as measured by variance.

Markowitz’s parameters are the mean (the expected return) and the standard deviation, so actual returns are distributed normally. In MPT, risk is identical to variance. “If the term…‘risk’ (were replaced) by ‘variance of return,’ little change of apparent meaning would result,” Markowitz wrote.

**At odds with himself**

Though his parameters are those of the normal distribution, Markowitz wrote, “The presumption that the law of large numbers applies to a portfolio of securities cannot be accepted.” He did not explore the implications of this statement for his theory.

There are, on occasion, attempts to challenge the risk as volatility concept – such as Benoit Mandelbrot’s *The (mis)Behavior of Markets* (2004) – but none have successfully breached the battements the world of finance has erected around MPT.

In a new book, *The Pillars of Finance*, Guy Fraser-Sampson leads a direct assault against formidable Fort Markowitz. Fraser-Sampson, a Visiting Lecturer at the Cass Business School, City University London, is determined to lay siege to MPT, though a flanking maneuver may have been the wiser approach.

**Subjective risk**

To Fraser-Sampson, risk is a subjective measure defined by individual investors according to their needs. “What an investor really needs to know,” Fraser-Sampson writes, “is…the risk (an asset) may fail to meet the investor’s objective of discharging future liabilities.” Fraser-Sampson argues describing risk in terms of the normal distribution is meaningless. He stresses risk’s qualitative character over its quantitative nature and defies MPT’s proposition that one measure of risk is appropriate to all investors.

Fraser-Sampson goes further. He proposes a return to the basics: Time Value of Money, Net Present Value, and Internal Rate of Return. He points out the calculation of periodic returns, used to calculate standard deviation, ignores cash flows, the basis of asset valuation. IRR is a better method of calculating equity returns. It captures cash flows and takes into account the time value of money. Bond returns are IRRs.

He makes the observation that risk measures vary by asset type. Equity risk is measured differently than bond risk. Many asset classes have no useful risk measures. How can assets with different risk measures be combined in portfolio with a single value for portfolio risk? Why use the *E-V* rule at all?

The answer is, variance is easy to calculate and the normal distribution is easy to understand. But that does not mean variance is an accurate or even useful measure of risk.

**The efficient portfolio**

Debt and equity assets are commonly combined in a single investment portfolio. Diversification results. As Markowitz put it, “the *E-V* rule leads to efficient portfolios almost all of which are diversified.”

Equity risk is measured by variance of returns, or standard deviation, and in terms of beta (β), the regression of an asset’s returns against the market’s returns. Bond risk is measured differently, usually in reference to default and liquidity risk premia beyond the basic risk-free rate.

**A Crash every several billion billion years**

In practice, periodic returns present inconsistencies. Equity returns are calculated using changes in stock price (capital gains) and on the basis of any equity income. Bond returns (or yields) are the spreads above the risk-free rate. Like equity prices, bond yields change minute-by-minute throughout the day. So, too, do bond prices. But there is no calculating bond risk as a standard deviation. And which risk-free rate is the right one to use?

β is also problematic. What is the appropriate market to measure against the returns for Microsoft? The Dow? NASDAQ 100? S&P 500? S&P 100? The Russell Indices? Microsoft has a different β for each. There is no β for debt.

Despite such problems, MPT asserts it is sensible to express risk as “variance of return.” As Fraser-Sampson points out, this measure doesn’t work in a world where moves of 20 standard deviations (as in 1987) are not as uncommon as probability theory suggests. How uncommon is that? Mandelbrot wrote, “The 1987 stock market crash was, according to (normal) models, something that could happen only once in several billion billion years.”

So much for theory.