Do empirical tests show that CAPM work s in practice?

No matter how much we diversify our investments, it's impossible to get rid of all the risk. As investors, we deserve a rate of return that compensates us for taking on risk. The capital asset pricing model (CAPM) helps us to calculate investment risk and what return on investment we should expect. Here we look at the formula behind the model, the evidence for and against the accuracy of CAPM, and what CAPM means to the average investor.

Birth of a ModelThe capital asset pricing model was the work of financial economist (and, later, Nobel laureate in economics) William Sharpe, set out in his 1970 book "Portfolio Theory And Capital Markets." His model starts with the idea that individual investment contains two types of risk:

1. Systematic Risk - These are market risks that cannot be diversified away. Interest rates, recessions and wars are examples of systematic risks.

    

2. Unsystematic Risk - Also known as "specific risk," this risk is specific to individual stocks and can be diversified away as the investor increases the number of stocks in his or her portfolio. In more technical terms, it represents the component of a stock's return that is not correlated with general market moves.

Modern portfolio theory shows that specific risk can be removed through diversification. The trouble is that diversification still doesn't solve the problem of systematic risk; even a portfolio of all the shares in the stock market can't eliminate that risk. Therefore, when calculating a deserved return, systematic risk is what plagues investors most.

Roots of CAPM are in the Markowitz (mean variance) portfolio theory in the 1950’s.

CAPM is credited to Sharpe, Lintner and Mossin in the beginning of 1960’s, and is the

first model to quantify risk and the reward for bearing it.

CAPM marks the birth of the (modern) asset pricing theory. It is widely used in applications

(estimation the cost of capital, portfolio performance evaluation, etc).

However, the empirical record of the model is (very) poor.

Empirical testing relies usually to this excess return form by focusing on three implications:

(1) The intercept is zero

(2) Beta completely captures the cross-sectional variation of expected returns.

(3) The market risk premium λ = E[zm] is positive.