Many people find negative interest rates confusing. Why should anyone pay a bank to make a deposit? Why should a bank pay someone to borrow? How can we value an asset with a future cash flow when the interest rate is negative?
Policymakers also wonder whether the effects of negative interest rates on the economy are favorable or unfavorable. Do negative interest rates help central banks achieve price stability by stimulating economic activity? Do negative rates spur banks to make more good loans or to evergreen bad ones? Will borrowers and banks take on too much risk because they can fund investments at a negative rate? Will households reduce their saving rate because the return is so low, or raise it because low returns leave them farther from their wealth target? Will negative rates influence the ability of pension funds, insurance companies and governments to make good on their long-term promises to future retirees?
In this primer, we examine these questions, starting with key facts about negative nominal interest rates. We note how low some central banks have set their policy rates and that there is an effective lower bound (ELB) below which a policy rate reduction would be contractionary. We then turn to a few basic issues, including the impact of negative rates on asset valuation. Finally, we discuss how negative interest rates affect the behavior of key financial institutions, especially banks, but also intermediaries that manage assets for retirement.
Our conclusion: there is little magic about having a slightly negative, as opposed to slightly positive interest rates. Thus, much of the criticism of persistently negative nominal interest rates applies similarly to very low, but positive rates. That said, financial system frictions limit the favorable impact from modestly negative nominal rates, but our experience with them remains limited. Given the likely need for unconventional policy tools to address the next recession, learning more about the benefits and costs of negative nominal interest rates is a high priority.
Stylized Facts. It is important to distinguish between nominal interest rates—which you can find online or in the financial press—and real interest rates, which are adjusted for inflation. For most investment decisions, the real interest rate matters. Yet, without attracting great attention, over the past 25 years real interest rates have frequently been negative. For example, since 1995, the German and U.S. one-year real interest rates on safe government debt have been below zero for 40 percent and 47 percent of the time, respectively (see chart). (We plot the ex post real interest rate. But, given the stability of inflation, ex ante real interest rates computed using inflation expectations have very similar properties.)
One-year real interest rates on government debt, 1995-October 2019
Note: The real rate subtracts past 12-month inflation from the current nominal rate. This implies that expected inflation equals actual inflation. Sources: Bundesbank, OECD and FRED.
While real interest rates have frequently been negative, virtually all the market and media attention has focused on the fact that, in recent years, nominal interest rates fell below zero. To counter stubbornly low growth and inflation, beginning with Denmark’s central bank in 2012, several central banks lowered their policy rate modestly below zero. Denmark and Switzerland have set policy rates as low as -0.75%; the ECB’s policy rate edged down to -0.50% in October; and the Bank of Japan set the policy rate at -0.10%, while targeting the long government bond yield in a range around 0%. Today, short-term money market rates in Denmark, Germany (and much of the euro area), Japan, Sweden, and Switzerland all remain below zero (see chart).
Nominal three-month interbank rates, 2007-October 2019
Source: FRED.
Recently, attention has focused on the large volume of long-term bonds with negative nominal yields (see chart). Most of these are sovereign debt issued in Europe and Japan, but some long-term, high-grade corporate liabilities are also trading at yields below zero. Indeed, according to the IMF, as of October 2019, about one fourth of outstanding fixed-income debt—or $15 trillion—carried a negative yield (see the Global Financial Stability Report, page viii).
Long-term government bond yields, 2007-October 2019
Source: FRED.
The ZLB and the ELB. For many years, nearly all economists and market participants acted as if nominal interest rates could never sink below zero, believing that households and firms would prefer cash (which pays zero) to any instrument offering a lower yield (see the discussion here). However, the notion of a zero lower bound (ZLB) ignores the elevated transactions costs of using cash—including storage, transport and insurance costs—compared to instruments that trade and settle electronically.
Today, we speak of an effective lower bound (ELB), which differs from the ZLB solely because of the costly frictions associated with using cash. No one knows the precise level of the ELB. What we do know is that anything that raises the transactions cost of using cash—such as eliminating the largest-denomination currency notes (as the ECB did with its 500-euro note in 2018)—tends to lower the ELB (for details, see our posts here and here). So, it surely fluctuates, albeit modestly.
Importantly, if interest rates were to sink temporarily below the ELB, at least some households, businesses, and institutional investors would shift from bank deposits to holding cash, putting a floor under nominal interest rates. Indeed, if a central bank were to push its policy rate below the ELB on a persistent basis, the switch could make the policy contractionary as depositors flee banks into paper bank notes, undermining banks as a key source of credit. That risk likely accounts for why no central bank has yet tried to lower its policy rate even to -1%.
Why do some central banks wish to set interest rates modestly below zero? The answer is that it provides added room for stimulating aggregate demand and securing price stability. Most advanced economy central banks have an inflation target of about 2% (see here). Yet, because of the decline in the equilibrium real interest rate (r*)—a development that central banks accept as an exogenous fact of life—the normal level of the nominal policy rate also has fallen. To be specific, estimates of r* are currently around ½%, which is nearly 2 percentage points below the level in 2007, prior to the financial crisis. As a result, the neutral policy rate is now around 2½%. This makes it far more likely that central banks will run out of conventional monetary policy space the next time the economy turns down. Indeed, in several economies, inflation has languished below the central bank target even in the presence of negative rates.
In practice, most central banks that impose negative rates only impose the lowest (headline) rate on a fraction of bank reserves. That is, they practice tiering to diminish the impact on bank profits. For example, the Bank of Japan pays a positive rate on a base level of reserves, a zero rate on an amount that increases with the total level outstanding, and a negative rate only on the marginal reserve holdings (see here). Similarly, in September 2019, the ECB adopted a two-tier system that exempts part of excess reserve holdings from the negative deposit facility rate (which currently pays -0.50%). A tiered system means the average rate paid on reserves exceeds the marginal rate paid. Another way to interpret a tiered system is that the central bank charges a negative rate on all reserves, but pays a subsidy on some base level of reserves, lowering the average cost. Since competitive pressures likely force banks to pass this subsidy on to clients, the average rate charged on reserves may be a better indicator of the monetary policy stance than the marginal rate that central banks highlight.
To be sure, the ELB is an artifact of the current monetary system. If we were to eliminate cash, as Goodfriend, Rogoff, and others advocate, then central bankers would be free to set whatever interest rate they feel is necessary to meet their inflation and growth objectives. Indeed, one argument for creating a universal central bank digital currency is that it would substitute for cash and permit deeply negative nominal interest rates. However, it is hard to see how even a major jurisdiction could do this on their own. If, for example, the euro area were to eliminate all paper bank notes (something unimaginable to even a casual visitor to Germany), unless Switzerland did the same, one could expect massive shifts into Switzerland’s 1000-franc notes. In other words, the benefits of eliminating cash depend on an extraordinary level of cooperation among leading financial jurisdictions.
Valuation. The formula for evaluating the present value of a future cash flow—such as a bond’s coupon payments or the dividends from a share of stock—works equally well for positive and negative nominal interest rates. Consider, for example, a five-year $100 bond that pays five annual coupons of $5 at year-end and returns the principal of $100 after five years. For a given nominal interest rate i, the formula is as follows:
The black line in the chart shows how the present value of this simple 5%-coupon bond changes with the interest rate (or yield to maturity) (i). Note that there is no discontinuity at or near the level where i equals zero. The red line shows the same pattern for a zero-coupon bond. In principal, the same pattern would arise for a bond with a negative coupon. (However, such a negative-coupon bond might be difficult to administer, since the issuer would need to collect the coupon payment from each bondholder in the first four years, before making the net-of-coupon principal payment after five years.)
Present value of a five-year bond with a 5% annual coupon and with a zero coupon
Source: Authors.
Greater concern arises regarding the pricing of assets—like stocks—using an equilibrium dividend-discount model like the simple one below:
In this model, the price of the asset equals the product of the current dividend (D) times one plus its steady-state growth rate (g), all divided by the difference between the required nominal return on equity (i) and the dividend growth rate (g) (see here for a simple derivation). When the denominator is negative, so that the dividend growth rate exceeds the required return on equity, this model is ill-defined. But, since firms set the dividend growth rate in response to investors’ required return on equities, it would make little sense for firms to pay a growing dividend when the required return on equity is negative. Put differently, the model still works even if i is negative, so long as it is greater than g. And, because we can interpret the model in real (inflation-adjusted) terms, it still works in world of deflation, so long as (i-p) exceeds (g-p) where p is inflation.
Moreover, keep in mind that the required nominal return on corporate liabilities (i) exceeds the return on safe government debt, so that even if the safe government interest rate is modestly below zero, the expected return on equity and on privately-issued bonds is still likely to be positive. For example, since 1995, the Moody’s Baa corporate bond yield has exceeded the 5-year Treasury yield by an average of 3 percent. Similarly, estimates of the long-run risk premium on U.S. equities put it at 5 to 6 percent (see, for example, Jordà et al).
Bank Profitability and Behavior. Banks’ primary business is maturity transformation; their liabilities are shorter term than there assets. This means that whenever the yield curve flattens, the gap between the bank lending and fund rate—the net interest margin (NIM)—falls. All other things unchanged, a shrinking NIM typically drives down profitability and reduces banks’ incentive to lend. Importantly, when nominal interest rates sink below zero, the reluctance of banks to impose negative deposit rates on customers also tends to squeeze NIM. For example, Eggertson et al find that banks in Sweden generally did not pass through the negative policy rates of the Riksbank.
However, bank behavior may change with the persistence of negative nominal short-term rates. A recent Bundesbank survey finds that nearly 80 percent of total nonfinancial corporate sight deposits (and about one-fourth of household sight deposits) now bear negative rates (November 2019 Monatsbericht, page 32). More broadly, Altavilla et al report that well-capitalized banks in the euro area have been able to impose negative rates on corporate deposits, with the pass-through increasing as rates decline. Moreover, euro-area banks that pay negative rates to hold balances at the central bank supply relatively more credit, suggesting that they are not funding-constrained. Finally, pooling data from Europe and Japan, Lopez, Rose and Spiegel conclude that, while negative rates squeeze net interest income, increases in fees and other non-interest income have cushioned overall bank profitability (see also here).
Finally, it is worth recalling the critical role of bank supervision in an environment of persistently low interest rates—both modestly negative and positive. A key problem is that low interest rates help sustain weak, undercapitalized banks that evergreen loans to weak, unprofitable firms (see here). The first line of defense against such zombies is an effective regulatory framework that either recapitalizes or liquidates fragile banks without delay. Absent such a mechanism, Acharya et al argue that zombie credit can lead to a perverse disinflationary effect from low interest rates.
Pensions and Life Insurance Companies. A frequent criticism of central banks with a negative rate policy is that they are undermining the returns on retirement savings. At first glance, this would seem to be true―the lower the interest rate, the higher the saving rate one needs to achieve a given wealth target level at any age, including retirement. However, the challenge in most advanced countries today is that the equilibrium real rate of return on safe assets (r*) has declined for reasons that have little to do with central banks. Factors responsible include slower technological progress, slower (or declining) labor force growth, and increased post-crisis incentives to save (see, for example, here). The result is that retirement planning requires people work and save longer or take greater risk to achieve their longer-term financial goals.
In addition to the difficulties they pose for individual savers, very low real and nominal yields create considerable challenges for those intermediaries that offer generous long-run guarantees. Unrealistically high expectations of returns lead these institutions to take on greater risks in order to meet promises that may no longer be plausible. Such “reaching for yield” compresses the returns on risky assets, and poses risks for financial stability. Consistent with this, in their study of U.S. state and municipal pension funds, Lu et al. show that funds take more risk when risk-free rates are lower. However, promoting risk-taking through interest rate reductions is a feature, rather than a flaw, of conventional monetary policy. Moreover, the finding that lower interest rates boost risk-taking assigns no special role to negative nominal rates.
The Bottom Line. In a world of low inflation and a low equilibrium real interest rate (r*), central banks will need a diverse toolbox when the next recession hits. Some central banks are likely to rely on mildly negative rates to provide additional conventional policy space. Against this background, persistently low interest rates have both benefits and costs that policymakers need to assess.
Importantly, however, critics exaggerate the difference between modestly positive and modestly negative rates. Modestly negative rates do not threaten the basics of asset valuation or other foundations of modern finance. At the same time, both policymakers and practitioners need to be wary of “guarantees” that may be impossible to honor in a world of very low (and even negative) interest rates, as well as of the financial stability risks that may arise.