If interest rates are low, that means capital is cheap, right? And if capital is cheap, that means more edgy, entrepreneurial projects get funded, right? In an era of low interest rates, shouldn't we see a lot of experimentation in creating businesses with high long-term potential but uncertain short-term return? Chris Dillow asks these questions in a specific case (about which I know little and care less). But as a general proposition, I think the chain of reasoning above is less reliable than you might think. For an explanation, and a suggestive empirical result, read on.

Economics is founded on the notion of opportunity costs. But opportunity costs are frequently overlooked when discussing the "cost of capital". Suppose real interest rates are at 0%, or even negative. Does that mean capital is "cheap" for an entrepreneur with a project expected to earn a positive real return? At first blush, one might shout "Yes!" After all, anyone with money in the bank would be better off investing in the project than earning interest that fails even to keep up with inflation.

But this reasoning is flawed, because earning quoted interest is not an investor's only alternative to the proposed project. Suppose there exist many alternative investments that, for a similar level of risk, offer twice the real return of the proposed project. Then the real financing cost faced by entrepreneurs pushing the project is not a quoted interest rate, but the rate of return offered to investors by the alternatives. If the project cannot match those returns, it will not be financed.

In an idealized world, with no stickiness in prices or information, no non-market intervention in interest rates, and objective appraisals of project risk and return, the scenario described could not occur. The "quoted interest rate" for projects at the relevant risk level would quickly rise to approach the returns of the most profitable available investments, and interest rate benchmarks would accurately approximate the cost of capital.

But we don't live in that idealized world. Interest rate metrics can and do vary in ways that don't obviously track the expected returns of investable projects in the economy. The risk and expected return of projects cannot be accurately measured, and market participants must rely on a variety of strategies from hyperrational modeling to recent-past extrapolation to make investment decisions. I'd suggest that recent past extrapolation is a common approach.

Suppose an exogenous shock reduces benchmark interest rates beneath equilibrium levels. Financial assets should revalue very quickly in response to interest rate changes. But suppose that, because of stickiness, momentum or other effects, an asset price boom of some duration, rather than an instantaneous price change, results. How does this affect the cost of capital of a small entrepreneur with a speculative project?

Benchmark interest rates are low, so the headline cost of capital is cheap. But if investors, extrapolating from recent experience, expect high returns at low risk from asset appreciation, our entrepreneur has to compete with those expected returns. Her real "hurdle rate" is defined not by the headline interest rate, but by asset-boom inflated expectations.

For some entrepreneurs, this distinction between asset markets and available financing would be fictional, as it ought to be in theory. An entrepreneur within a large firm, for example, could take advantage of an asset market boom by persuading the firm to issue bonds, commercial paper, or shares to finance her project, achieving financing costs at or even beneath levels what headline interest rates would suggest.

But for many entrepreneurs, rasing capital by issuing securities in the broad market is not an option. Their projects must instead rely on bank financing or "angel investors", who would require higher returns when broad market expectations are high. If I'm right, during asset price booms this category of entrepreneurs should face an unusually high spread between quoted benchmark interest rates and the rates of return demanded by banks or angels. This is a testable proposition.

Unfortunately, I'm unaware of good data on the average returns required of small entrepreneurs by banks and angel investors. But until the late 1980s, there was a published interest rate reflecting the borrowing cost of businesses that rely on bank financing, the US Prime Rate. Presumably, the cost of bank loans for small entrepreneurs included the (observable) prime rate plus some (unobservable) spread. Though not conclusive, it would be suggestive if the spread of the Prime Rate over a low credit-risk benchmark tends to increase when asset markets boom. Naively, one would expect credit speads and asset prices to be negatively related, as higher credit spreads mean higher financing costs, and usually a risky business environment. So a positive association between a high spread for bank-financed loans and asset prices would be both surprising, and consistent with the hypothesis that asset price booms increase financing costs for firms unable to sell securities into the boom.

Regressing the spread between the US Federal Funds Rate and the Prime Rate against monthly percentage changes in the Dow Jones Industrial Average shows a significant positive relationship, with a coefficient of 0.04 (p < 0.001). In other words, a 1% monthly gain in the DJIA was associated with a 4 basis point increase in the spread, consistent with the opportunity cost of capital hypothesis. 34 years of monthly data were regressed form September 1955 though August, 1989. R2 of the regression is small, at 0.04, as would be expected since DJIA returns are much more volatile than the Prime Rate/Fed Funds spread. (The data is truncated in 1989 because, starting in the early nineties, the Prime Rate was altered to a near fixed 3% spread above Federal Funds. It now has little relevance as a specific measure of the cost of bank loans to business. DJIA was chosen as a proxy for asset prices simply because it is the most famous measure, and therefore intuitively likely to influence capital market expectations. I've not tried a similar test against other potential asset market price or borrowing cost spread measures.)

This was a butt-simple, univariate regression on data taken from FRED and Yahoo, and is very preliminary. Correlation ain't causation, and there could be a variety of other factors accounting for the observed relationship. It'd be nice to come up with a more complete model of the Prime/FF spread, and see whether it seems consistent with the high-opportunity-cost-of-capital in an asset boom hypothesis. But this very simple test provides at least a little evidence that asset booms increase the cost of capital to bank-dependent small entrepreneurs relative to what benchmark interest rates would suggest.

If you've read this far, thank you.

Update History:
  • 11-Oct-2006, 4:00 a.m. EET: Removed an unnecessary "indeed".
Steve Randy Waldman — Tuesday October 10, 2006 at 4:07pm permalink
Gabriel M. (mail) (www):
Very interesting!
It's great to see these sort of issues come up.
10.10.2006 5:24pm
Aaron Krowne (mail) (www):
I think the core insight above is a really good one, and count me in the category of "small entrepreneurs" having trouble finding funding in an asset boom (something always seemed fishy about that to me).

However, I'm not sure your experiment proves what you want it to prove. One major reason is that the credit spread you're comparing with the return on the Dow can increase *either* if the funds rate falls or the prime rate rises. We might expect that if the funds rate falling is responsible for one such event, then this will indirectly result in liquidity pouring into the stock market, thus boosting the Dow (at least, uh, before 1995).

The second problem with this methodology is that the companies in the Dow are a long shot from "small entrepreneurs". These companies are the "illuminati" of capital and can access it pretty much whenever, especially in an asset boom (and especially when the underlying drivers are intentionally inflationary monetary policy).

To really be convinced the theory has been validated, I would have to see data more directly pertinant to small business financing success. Perhaps the SBA has something. Or perhaps a venture capital returns index of some sort?

There is another nuance that might come into play in your underlying theory: as you hint at, start-up financing is nuanced; there are (broadly) loans, equity, and convertibles. Looking at just the first two "pure" categories, we might expect loan financing to behave differently than equity because in loan financing your upside returns are limited by the terms of the loan. Yes, issuance itself might be somewhat effected by estimates of business success, but it is certain your point about sacrificing higher returns from other similar projects would not apply.

E.g., a bank would see no distinction between financing Project A with 40% estimated rate of return using a 10% bank loan, or Project B with 20% estimated rate of return using a 10% bank loan, where Project A and B have the same likelihood of success. Either way the bank makes 10%.

So one might expect to see the *very* small scale business lending category to improve purely by virtue of lower interest rates, but the mid-range becoming hollowed-out, as only large-scale, proven projects can get equity funding.

On the other hand, as far as society-(or bank-)wide lending strategy, one might expect to see a shift from this relatively-unattractive small-scale lending to the juicier-seeming large-scale, equity-like lending in an asset boom.

And what do you know, in this asset boom, we've seen banks become bedfellows of home builders, going as far as equity partnerships in joint ventures. Hmmm...

So, there do seem to be some nuances, but broadly, I'm sure your point still holds.
10.11.2006 5:26am
Steve Randy Waldman (mail) (www):
Aaron,

Yeah, there are definitely lots of problems with my "test". It was just something I could do to get some inkling that a relationship I expect might hold.

Re the first problem you point to, you identify an implicit assumption in my test that may well not hold. Basically, I presume that Prime is always defined as Federal Funds plus a spread, that it adjusts very quickly to changes in Federal Funds, and that the spread adjusts independently. The post-1991 Prime works like this, in the sense that the spread is fixed, and the Prime adjusts instantaneously whle the spread is independent. But if, back in the market-driven prime days, Prime reacted with a lag to changes in Federal Funds, so that the spread widened for some period of time on a fall in FF, your alternative explanation is more parsimonious than mine. It's a testable proposition, regress ΔFF against Δprime and expect a coefficient of approximately 1 if my assumption holds, something less than one if your alternative explanation holds. When I have more time, I'd like to do some more testing, and this would be a good one.

The use of the Dow as something far from small entrepreneurs was intentional, in the sense that the Dow was supposed to represent the "asset market alternative" that small entrepreneurs implicitly compete against for financing in a low-interest-rate-asset-boom environment. If there were an "SBA financing index", I think I'd use that to replace the Prime Rate in my regression rather than DJIA.

Your point about banks being unconcerned about differences in return when the return is above a certain level is very subtle, and I think accurate. It's another "stickiness". Banks don't really set financing rates for projects wholly independently, based on unique analyses of risk/return characteristics. They bundle projects into categories, and set different rates for different bundles, within conventional limits. Banks never do lend at "Prime + 30%", even though there might be some projects for which that's be rational. (Those projects have to seek equity financing.) So I think you're right that the effects I'm talking about would be more relevant for marginal projects, projects amenable to bank financing at the highest conventional rates. Projects that are easy winners and good credits get the same, small spreads regardless of the returns realizable by the businesses. In a no-risk theoretical world, these good projects are "underleveraged", and equity holders would want to the firm to borrow and to fund dividends or stock buybacks until the bank's required rate approached the expected return of the ventures. But we don't live in that world, and real ventures (perhaps with the exception of those run by private equity firms) are nervous about stuff like not meeting projections and going bankrupt. So in the real world, you are right that spreads wouldn't be sensitive for the creditworthy projects, which is another reason to dislike Prime Rate as a measure of variable borrowing costs faced by entrepreneurs.

Anyway, thank you (as usual!) for very thoughtful and thought-provoking comments.
10.11.2006 8:31pm
Note: You don't have to register to comment! Just use the pull-down menu to comment as "Guest".