Cash Flow at Risk: How Merck Hedged What Really Mattered

In the late 1980s, Merck earned a large share of its income outside the United States, and the strong swings of the dollar in that decade moved its reported results by hundreds of millions. The finance team, led by Judy Lewent, who later became CFO, faced a question that every multinational eventually asks: should we hedge, how much, and why?

The answer they published in 1990 in the Journal of Applied Corporate Finance, with John Kearney, became a reference case in corporate risk management. Its core was statistical. Instead of hedging every exposure or none, Merck simulated the distribution of its future cash flows under thousands of currency scenarios and asked what really needed protecting: the funding of research and development, a multi-year commitment, and the dividend.

From value at risk to cash flow at risk

Banks measure market risk with value at risk: the loss on a portfolio that will be exceeded only with a small probability over a short horizon. Industrial companies care less about the market value of a portfolio tomorrow and more about the cash they will generate this year and next. Cash flow at risk (CFaR) applies the same idea to them: how far below plan can operating cash flow fall, with 95% confidence, over the planning horizon?

1. Map the exposures: which cash flows are earned or spent in which currencies, and how prices and volumes respond to exchange rates.
2. Model the risk factors with their volatilities and correlations, from market data and history.
3. Simulate thousands of paths of the factors and recompute the cash flow on each one.
4. Read the distribution: the median, the 5th percentile and the probability of falling below the level the business must protect.

What a hedge does to the distribution

Consider a company planning 2,000 million of operating cash flow, of which 900 million is earned in foreign currencies whose basket moves by about 12% in a typical year. It needs at least 1,800 million to fund its research programme and its dividend without new borrowing. Merck's choice, buying currency options rather than forwards, keeps the upside and protects the downside, for a premium.

Unhedged, the cash flow at risk is about 223 million, and there is a 7% chance of falling below the 1,800 million the company must protect. With options on 60% of the exposure, the left tail is cut: the 5th percentile rises by almost 40 million, the cash flow at risk falls to about 176 million and the probability of a shortfall halves to 3.5%. The cost is the premium, 15 million, visible as a slightly lower median. The upside is almost untouched.

Why the statistical framing matters

• It defines the objective. The goal is not to remove volatility, which is costly and often impossible, but to protect specific uses of cash that would be expensive to interrupt.
• It sizes the hedge. The simulation shows how much protection each additional unit of cover buys, so the hedge ratio becomes a decision about the price of insurance.
• It handles correlations. Exposures in currencies that move together add up; those that offset each other need no hedge at all. A distribution captures both; a list of exposures does not.
• It speaks the board's language. "There is a 7% chance we cannot fund the plan without borrowing, and 15 million cuts it in half" is a decision. "We have 900 million of exposure" is a fact.