A value of beta (systematic risk measure) that is constructed artificially when market-generated betas are not available (or cannot be estimated). To figure out beta in an artificial way, analysts usually start with a beta coefficient for the company or unit, and then adjust that beta for the differences between a specific project and the company or unit. However, if the betas for the company or unit are unknown, an analyst can usually choose a market-traded company’s beta to which financial characteristics of the project are comfortably comparable. Betas can be artificially expressed as the product of an industry portfolio correlation coefficient and a company-specific relative volatility coefficient:

Volatility of market returns- and a correlation coefficient for the industry- may be derived directly from market data. If the company being analyzed is not traded, return volatility may be estimated using as a proxy the standard deviation of changes in capitalized net operating profit after tax (NOPAT) or earnings before interest and taxes (EBIT). To get rid of any possible seasonality in operating results (monthly basis data), the percentage change in capitalized NOPAT or EBIT over the same period last year can be regressed against respective annual market returns.

A constructed beta is especially instrumental for non-traded companies and illiquid stocks where the beta is artificially depressed by a low market correlation due to illiquidity (or very low liquidity).