Kevin Warsh recently coauthored a piece in the WSJ suggesting that lower interest rates actually penalize business investment, contrary to common knowledge and mainstream theory. The specific argument itself is wrong, at least by itself. Even if lower interest rates did somehow encourage buybacks, that’s cash ultimately returned and theoretically invested where it’s most needed. Now this might not increase business investment, but that’s another argument entirely. Brad DeLong isn’t happy and Larry Summers agrees, though is open to any logic about why this might be the case. 

I’ve vaguely thought about this before, and think it’s worthwhile to think about the assumptions under which lower short term rates might reduce current investment. Even accepting this is most likely not the case, the process of figuring out how something might happen at least provides a set of empirical projects researchers might pursue in determining this question. To the extent these arguments are reasonable they also provide a useful counterpoint to conventional wisdom.

Note the point here isn’t to argue in favor of lower or higher rates – indeed answering this question has little bearing on that topic, without further assumption about the quality of the business investment itself. We’re also not trying to tell ad hoc stories – for example how, perhaps, low interest rates might signal uncertainty about the economy, reducing investment. Rather, the interesting if impertinent question is the existence of any obvious theme or set of incentives that create a world where interest rates are positively correlated with investment.

I have four suggestions.

True versus observed investment

The first point I would make is regardless of true investment, lower interest rates probably tend to positively bias the observed investment. This doesn’t move the ultimate question in one direction or the other, but raises questions about any data used to justify a particular position. My intuition behind this is simple – though possibly incorrect. Consider a firm that borrows from a bank to invest in what turns out to be a bad project. In theory managers learn about a loss and recognize it immediately. But in practice it’s hard to know when the economic fundamentals of an investment fail to match their invested value – that is fail to generate a positive return net of a capital charge (in this case the interest rate).

An obvious example to this effect might be Japanese banks that can delay recognition of economic loss simply because they can return the cost of capital to investors – who expected the project to work out – by rolling over inexpensive debt. Eventually, of course, all accounts must be settled and someone won’t be happy. You might think of this as an illusory double-counting mechanism that is particularly prevalent when the cost of doing nothing is low.

This requires nuance about the constitution of ‘business investment’. We must in some sense think about the true ex post value of the investment – otherwise it wouldn’t matter whether business investment meant cash spent on a cup of coffee or building a new tower. Of course this isn’t something we can know for sure in real time, but a simple effect of investment cycles that we should note when thinking about interest rates. The important point here is this won’t get out of hand in a world with high interest rates.

Agency dilemmas and a growing gap between volatility to manager and volatility to investor

Another story might involve agency problems between owners and managers. Investments are made in expected value – with some compensation for risk. Investors can diversify firm or project risk, whereas managers cannot. There is an enormous literature discussing and debating manager risk-aversion and its effect on investment though there is an important wedge when individual firms can commit to a limited number of projects, unable to diversify risk and thereby failing to invest in positive expected value projects that investors like.

Low interest rates affect this wedge. If the marginal project that becomes profitable at a slightly lower rate is one with high variance, individual firms may not be compelled to invest – despite every economic argument that the opportunity cost of holding money is now higher. This is particularly true when projects are not traded and limited to a small number of firms. (Do you know about various potential investments a Wisconsin waste management company might make to decrease the number of routes it has to make every morning and, if you did, would you have the capacity to purchase this investment and use it to the same effect?)

It might be – and I’m only theorizing potential assumptions, not claiming any to be practically true – that marginally profitable projects are increasingly risky for the individual firms that compete. But this wouldn’t be an outlandish claim, either. The discounted value of a project grows dramatically faster as interest rates fall. Small errors in judgement have larger feedback into economic calculations, amplifying the gap between investments that are made and should be made. There’s evidence to back this story. Where interest rate have fallen to zero, corporate executives still demand a 10% hurdle rate. Indeed lower interest rates change the opportunity cost for investors, not managers. This shouldn’t matter, but maybe agency dilemmas are significant.

This matters only because price changes manifest in both an income and substitution effect. If the income effect of lower interest rates, a smaller supply of investable capital, works as predicted at low interest rates, but the substitution effect is increasingly challenged by other factors, this change might reduce business investment.

Attenuation of relative debt preference

Also note the general agreement that the tax code privileges debt. There’s the explicit tax shield, along with various other interest deductions that bias towards lending more and borrowing more (with an indeterminate effect on price). Some of these are restricted to interest payments, excluding settling the principal. Lower interest rates would tend to reduce the debt bias. That is they might increase overall debt, but reduce the divergence vis-a-vis equity in a world where both are treated equally.

High leverage industries tend to be more capital-intensive, with large outlays for business investment. Airlines are a good example. If the relative bias changes with interest rates, and investors privilege industries that are typically equity-financed – again, keep in mind the distinction between change in bias and change in actual quantities – business investment might decrease.

Note this doesn’t reduce ex post returns – for example software firms have been able to produce remarkable returns without much invested capital. Indeed investing much in an industry that returns little isn’t always better than investing less in an industry with outstanding potential. This is a point of stipulation in the secular stagnation argument, and is potentially very true.

Permanent patience

Finally, I think we should distinguish between a dramatic fall in short-term rates to encourage investment in a recession to prolonged low rates because of a lower bound. Low rates now, and commitments to lower rates in the future where there isn’t an obvious ‘return to normal’, substantially decrease the urgency of any capital decision on the technical principle of ‘well, what else are you gonna do with your money’. This sort of flies in the face of what a discount rate even means, but your life period loss function for delayed investment will certainly be greater when the penalty from impatient investors is higher. Furthermore, if low interest rates are associated with comfortable liquidity conditions the cost of waiting may not be high.

Distinguish between the average investment (which may be amazing, and does not matter) and the marginal investment (which may be mediocre, and definitely matters). This patience and a slight risk aversion can’t be explained away by simply noting ex post fantastic investments that were risky and did turn out well, as much as noting that the incentive to make the many marginally profitable investments might be lower today for reasons outlined above.

None of this is to say that lower interest rates reduce investment, as much as to propose a pathway through which that argument might be made.

Some people argue that inflation causes high tax rates. In the obvious – and obviously correct – sense that taxes are progressive, resulting in bracket creep. But also in the more insidious sense that asset appreciation through inflation creates a nominal tax burden that must be paid out of a base of no real income. This is a point forwarded by the Tax Foundation, prominent economists, and just about every entry on the search entry “capital gains tax inflation”. These arguments all describe how inflation substantially increased the real tax burden between 1955 and 1990. Unfortunately, none of it is as straightforward as the writers make it out to be.

Let’s make a few assumptions. At the margin all taxes are proportional to value, that is no lump-sum levies. Assume there is no bracket creep, and that marginal government spending is restricted to transfer payments. The role of taxes in this world is, then, limited to assigning the distribution of income among various groups. The argument, then, is tantamount to claiming that the “inflation tax” perverts the tax system to skew post-tax income away from capitalists towards everyone else.

When you think of it this way, inflation doesn’t matter. If I own a security that appreciates $10 through inflation, generating a tax burden of $2 the natural reaction might be that the government is creating a real tax burden out of no nominal income. An “infinite” tax rate. Except inflation affects all future cash flows, including income from labor, which also increases in the state with inflation. If inflation affects most assets evenly over the long-run, not charging a tax on nominal gains would increase the relative distribution of income accruing to capital owners, necessarily at the expense of everyone else.

Another way to think of it is that the government has real liabilities – workers, indexed transfer payments, and material costs will all eventually rise with inflation – and to finance the larger nominal outlay it must also receive a larger nominal income from each group. To the extent taxes are levied as a percent of income this does not change the final distribution between capitalist and worker.

The truth of this simple claim can be noticed by taking the infinite tax argument to its logical conclusion. Suppose we have really high inflation for a while and taxed only real gains – out of concern for the Tax Foundation. If the stock of capital increased from $1 to $10, without any real gains, the capitalist faces no tax burden. Except everyone else – rentiers, workers, lenders – does face the nominal burden, and pay more in taxes. This naturally results in a capital subsidy.

One might argue that really we should tax only real gains on both capital and labor. This, of course, isn’t the argument that everyone cited above is making as they restrict the argument to an “infinite tax on capital” and believe there is something special about the nominal appreciation that makes capital different. Also note this would mean there was an “infinite tax” on labor, making the claim that there is some net inflation tax (i.e. change in distribution due to inflation) very likely false. More importantly – supposing the relative distance of various brackets were indexed – indexing real income on capital and labor is a really challenging strategy.

For one there’s a lot of error in these measurements, creating a political bias to understate inflation figures to juice tax revenue. It also assumes there is some measurable, true inflation ignoring the complex heterogeneity of various liability structures. Of course, ultimately, if we did figure out how to get around these problems and taxed everyone on t – 10 dollar values, the net distribution would change none at all since all income would be modulated by the same factor.

In reality, of course, some of the assumptions that led to this conclusion aren’t true. Inflation doesn’t affect everyone uniformly – at least not over any fixed period of time – and each individual has different preference for changing prices since the goods market may react more sensitively in one industry than the other, hurting some people more.

It might be worthwhile to study these assumptions, and figure out cases where higher inflation changes the distribution away from capitalists purely through the nominal tax rate. But the sort of work you saw from the Tax Foundation, apparently detailed, wonky, and evidence-oriented is misleading. It is either tautologically true saying absolutely nothing (inflation means bigger numbers) to certainly false (that inflation changes the net distribution without further assumption).

This isn’t particularly tricky to see, and is simply a consequence of efficient markets. If present value is the weighted sum of future cash flows, it makes no sense to argue that inflation affects the two sides of the equality any differently.