Though their risk management systems can never be infallible, the chief financial officers and treasurers at banks, brokerage houses, money management firms and insurers do understand - some from firsthand experience - how and why their portfolios might blow up.
By Rich Blake
February 2001
Institutional Investor Magazine
The industry calls their method of evaluating financial asset volatility “value at risk.”
In the past five years or so, a spirited group of academics and economists has considered how to apply VAR analysis to nonfinancial companies. Last month National Economic Research Associates, a division of insurance giant Marsh & McLennan, introduced a tool that it hopes will get the job done. NERA dubbed its offering C-far, for cash flow at risk.
Harvard University economics professor Jeremy Stein and independent consultant Stephen Usher spearheaded the project to create C-far. The model, they believe, will become a widely accepted standard for understanding and managing cash flow risk at nonfinancials, just as VAR functions in the financial sector.
“For nonfinancial firms, accurately assessing exposure to risk is not straightforward,” says Usher. “There are too many complicated risks with little or no supporting data. Nonfinancial executives have been forced to rely on rule-of-thumb assessments to factor risk into earnings estimates and strategic planning.”
John Faraci, CFO at International Paper Co., believes that the new tool will be quite useful. “Historically, the forest products industry has experienced significant volatility in earnings and cash flow because of the wide fluctuations in product prices through a cycle. Increased leverage from mergers and acquisitions and greater emphasis on controlling capital spending have highlighted the risks of sudden, rapid downturns in cash flow. C-far would help us to quantify that volatility.”
Essentially, C-far is a computer-simulated method to analyze the probability of fluctuations in a company’s cash flow over a specific time horizon, usually the coming quarter or year. From the probabilities, users generate a variety of potential worst-case scenarios.
Harvard’s Stein likens his procedure to an insurer’s actuarial table. “Insurance companies examine life expectancy by dividing the population into groups and collecting data - age, weight, occupation, whether you’re a smoker. We divide companies into groups and look at certain criteria - market cap, earnings, stock price volatility and industry risks - to see what the risk exposure is. C-far tries to answer the question ‘What’s the worst thing that can happen to me in the next quarter or year?’”
The practice of analyzing value at risk in the financial services industry goes back to the 1970s. VAR analysis involves sprawling and expensive computer networks, such as J.P. Morgan’s RiskMetrics. They can take real-time pricing data and simulate thousands of possible outcomes, producing a ballpark estimate of how much money the institution could lose on a really bad day.
Using VAR, a bank or brokerage can identify every one of its assets - loans, trading positions, derivative-based hedges. Then the institution can quantify and aggregate across its entire portfolio the risk exposure to such variables as interest rates, credit risk and foreign exchange risk for each of the assets. The VAR model for financial institutions can look to a robust set of historical data on price movements of its traded assets. “Imagine trying to apply this same approach to a nonfinancial firm,” says Stein.
During the mid-1990s Usher, then a NERA vice president, attended meetings with Marsh & McLennan risk consultants. He noticed that the same question kept coming up: Is there an analogy to VAR for nonfinancial companies? The consultants were trying to model nonfinancial companies’ risks, operational, strategic and financial. They took a bottom-up approach, trying to identify potential sources of risk from the nuts and bolts of a company’s business, but they weren’t getting very far.
In 1999 Usher teamed up with Stein, and within a year the top-down C-far approach was born.
C-far starts with the logical premise that risk reveals itself in nonfinancial companies as deviations from targeted cash flow.
“We decided to look at the culmination of all of the minutiae, the item of ultimate interest: cash flow,” says Usher. Instead of studying historical cash flow patterns - where, for most companies, data sets were too small to make analysis meaningful - the NERA economists developed comparative groups against which to benchmark companies.
Semiconductor companies are not necessarily lumped with other semiconductor companies. Rather, each one is placed into one of 81 boxes based on the four dimensions of cash flow risk noted by Stein. Each of the four dimensions is carved into three subsets: top third, middle third and bottom third. So, for example, one peer box might contain companies ranked along these lines: market capitalization (bottom third), earnings (middle third), industry-specific risk (top third) and stock price volatility (bottom third).
C-far has a number of practical applications. To take one example, it can be helpful to a company in its efforts to determine an optimal mix of debt and equity. Textbook finance theory teaches that the more volatile a company’s cash flows, the less debt it can safely carry. Although this concept is well understood, it can be difficult to put into practice, because cash flow volatility is such an unknown. The C-far model makes it possible to measure that volatility, and in that way it allows a more quantitative analysis of optimal capital structure.
And that’s just the sort of thing that a CFO likes to have in his toolbox.
