Picture a capital request form, the kind with boxes. Someone wants a million dollars for a project that will return $150,000 a year for twelve years. Near the bottom of the form are two boxes. One says "Payback (years)", and in pen someone has written 6.7. The other says "Discount rate", and the number in it, 15 percent, is not handwriting. It was printed with the form.
Those two boxes hold two different ways of judging the same project. The first is payback in years: how long until the money comes back. It is simple arithmetic, and it never asks what money costs. The second is value at today's rate: every future dollar is discounted back to the present at some rate, the cost is subtracted, and what is left is the project's net present value. That answer depends on the rate and on almost nothing else you could change at the table.
My argument is that the discount rate is a policy decision disguised as a parameter. The same two projects can swap places as the rate moves, so the number in the template decides which kind of project wins, quick or slow. Most firms set that number well above their cost of capital and rarely change it, so it often records an old decision more than today's market. This is not a forecast of interest rates and not investment advice. It does not say a high rate is wrong, and it does not say a formula should replace judgment. It says the rate is a choice, and somebody made it.
Payback counts years and never asks for a rate.
The value depends on a number printed before anyone proposed the project.
Two boxes on one form
Tuesday's dispatch on the Federal Reserve's dot plot, "The dots are a survey, not a plan.", ended by leaving today's question: what a rate does to a project's arithmetic. On September 16 the Federal Open Market Committee raised its target range for the federal funds rate by a quarter point, to 3-3/4 to 4 percent, by a vote of 12 to 0.7 This essay is not about that move. It is about the other rate, the one inside the firm, which may or may not follow it.
Most firms use both boxes. In February 1999, John Graham and Campbell Harvey surveyed chief financial officers and got 392 answers.1 About three-quarters said they always or almost always used net present value (74.9 percent) or its cousin, the internal rate of return (75.6 percent), which is the rate at which a project's net present value would be exactly zero. Payback came next, at 56.7 percent. The authors found that surprising, because textbooks had complained about payback for decades: it ignores the time value of money and every cash flow after the cutoff, and the cutoff is usually arbitrary. Two decades later the pattern held. In the 2022 Duke CFO survey, Graham reported, 77 percent of large firms always or almost always used NPV, and 64 percent of large firms and 66 percent of small ones used payback.3
So the two rules usually sit side by side, and most of the time they agree. The interesting cases are the ones where they do not, and the rate is what makes them disagree.
Two projects, one table
Take two hypothetical projects, each costing $1 million today. Project A is quick: $360,000 a year for four years, $1.44 million in all. Project B is slow: $150,000 a year for twelve years, $1.8 million in all. A pays back in 2.8 years and B in 6.7. Payback prefers A, and it prefers A at any rate you like, because it never looks at the rate.
Value at the rate tells a different story. At 4 percent, A is worth about $307,000 more than it costs and B about $408,000, so B wins. At 8 percent, A is worth about $192,000 and B about $130,000, so A wins. Somewhere in between the ranking flips. For these two projects it flips at about 6.25 percent. Below that rate, the slow project creates more value. Above it, the quick one does. At 10 percent B is barely worth doing, and by 12 percent it is underwater.
Nothing about either project changed in that table. The cash flows are the same in every row. The only thing that moved was a number that neither project's sponsor wrote, and it decided which of them looked smart.
Why the slow project feels it more
The mechanism is old and simple. Discounting shrinks a future dollar by the same factor once for every year it has to wait, so a change in the rate compounds over the wait. Raise the rate from 4 to 5 percent and a dollar due in two years loses about 1.9 percent of its present value. A dollar due in ten years loses about 9.1 percent, and one due in twenty years about 17.4 percent. A project whose money arrives late is, in effect, a long bet on the rate.
That is why a single point matters so much more to B than to A. Moving from 6 to 7 percent cuts the present value of A's cash by about 2.2 percent and B's by about 5.3 percent. The same nudge falls more than twice as hard on the slow project, because more of its money is far away.
John Maynard Keynes built his theory of investment on this arithmetic in 1936. He defined the marginal efficiency of capital as the discount rate that makes the present value of an asset's expected returns equal to its price, and he noted that Irving Fisher had already given the same idea a name, the rate of return over cost: what a finance class now calls the internal rate of return.6 Investment goes ahead when that rate beats the rate of interest. At the end of the same chapter Keynes put the point more memorably: "It is by reason of the existence of durable equipment that the economic future is linked to the present." The longer the equipment lasts, the tighter the link, and the more the rate decides.
The rate in the template is not the market's
If firms discounted at a rate that tracked the market, this would be a story about the Fed after all. The evidence says they mostly do not.
In the fall of 1990, James Poterba and Lawrence Summers mailed a questionnaire to the Fortune 1,000 and got 228 responses. Converting the answers to real terms, they found an average real hurdle rate of 12.2 percent.2 For comparison, since the mid-1920s corporate bonds had returned less than 2 percent a year after inflation and stocks about 7 percent. The most common hurdle, reported by a fifth of the firms, was 10 percent.
Three decades later Graham, in his 2022 presidential address to the American Finance Association, put numbers on the same gap.3 In the March 2019 wave of the Duke CFO survey, the average hurdle rate was 15.0 percent and the average weighted average cost of capital was 9.0 percent, a buffer of about six points. For at least two decades, he writes, hurdle rates have carried a buffer of that size. They are also sticky: they have moved only about two or three percentage points in 35 years, while market interest rates fell substantially. And the hurdle is not even the last word. In a 2017 wave, only 20 percent of companies said they accept every project whose expected return beats it.
Other researchers find the same thing by other routes. Ravi Jagannathan and coauthors, matching a CFO survey to company data, report discount rates that average twice the firms' cost of financial capital.4 Niels Gormsen and Kilian Huber read the conference calls of roughly 2,500 large firms in 20 countries from 2002 to 2021 and recorded the rates managers said they used. Their finding is that the link between discount rates and the cost of capital is far weaker than standard theory assumes: it takes many years for a change in the cost of capital to show up in the rate a firm uses.5
This is a different gap from the one in March's dispatch, "The Fed held. Your rate may not.", which followed the Fed's target to a borrower's loan rate. That gap is between the central bank and the lender. This one is inside the firm, between what capital costs and what the template says it costs.
The wedge on the form
Now go back to the form. Suppose, as in Graham's averages, the firm's cost of capital is about 9 percent and its printed hurdle is 15. At 9 percent, Project B is worth about $74,000 more than it costs. At 15 percent it falls short by about $187,000. Project A passes at both rates. Nothing in B's forecast decides its fate. The six points between the two numbers do, and they were set before anyone proposed B.
There is a further twist. The higher the rate, the less the distant years count, and the more value at the rate starts to rank projects the way payback does, by how soon the money comes back. In the table above, every rate above 6.25 percent agrees with payback. A high, sticky hurdle quietly turns the blue rule into the gold one. A firm can do the textbook calculation in every proposal and still be running a payback shop.
One more habit makes it stronger. Graham and Harvey found that 58.8 percent of CFOs would always or almost always judge a new project at the company-wide rate, even when its risk differed from the firm's.1 One rate applied to every proposal does not treat them equally. It favors whichever kind of project that rate flatters, and at a high rate that is the quick one.
This is also not the question in No. 10, "The money is already gone.", which was about money already spent and measured from today without discounting. That essay set the rate aside. This one is about the money not yet spent, where the rate is the whole question.
Objections, considered
A high hurdle is deliberate, not a mistake. Often it is. Some firms told Graham that a high hurdle helps them focus on the best projects. Jagannathan and his coauthors connect high rates to firms with plenty of capital but too few qualified managers or staff, which pass up good projects to keep capacity for better ones later. I agree, and the claim here is narrower than "the hurdle is too high". It is that the hurdle is a choice with a tilt toward quick projects. A firm should be able to say why its number is its number, and when anyone last checked it.
If hurdles are sticky, the rate barely matters. This turns the evidence against the thesis: if firms seldom change the number, market rates cannot be reordering anything. But that strengthens the point. If the market does not decide, the template does, and whoever last set the number made the call. Stickiness moves the decision from the market to the committee. It does not remove it.
Payback is crude, and nobody serious relies on it. More than half the CFOs in 1999, and about two-thirds of firms in 2022, said they did. It survives for reasons: it is easy to explain, it speaks to firms short of cash, and Graham suggests that firms which lean on it may find the alternatives hard to put into practice. The right complaint is not that payback is naive. It is that payback is the one rule on the form that cannot see the rate at all.
Rates rose, and companies are still investing. Nothing here is a claim about the economy as a whole. The flip in the table is arithmetic inside one firm's review, and it holds whatever aggregate investment does. Graham himself suggests that sticky hurdles may make corporate investment insensitive to interest rates, which fits the argument rather than contradicting it.
By the time a proposal reaches the committee, the rate on the form has already decided what kind of project will look smart.
Before the next capital review
- Find the rate in your template. Write down who set it, when, and on what basis.
- Put it next to a current estimate of your cost of capital. Name the gap, and decide whether it is a buffer you mean to keep.
- Show payback and value at the rate side by side, and rerun at one point higher and one point lower. If the ranking of the top projects changes, say so in the memo.
- Flag the projects whose value mostly arrives after year five. They are the ones the rate decides. Then put the rate itself on the calendar for review, separate from any project.
Back to the form. The 6.7 in pen is a fact about the project: the money comes back in the seventh year, whatever happens to rates. The 15 percent in print is a fact about the firm, set by someone, at some point, for reasons that may still hold. It belongs in the review as much as the project does, and it is the one number on the form that nobody is asked to defend.