Guest Contribution: “The End of Zero Interest Rates”

Today, we present a guest post written by Jeffrey Frankel, Harpel Professor at Harvard’s Kennedy School of Government, and formerly a member of the White House Council of Economic Advisers. A shorter version appeared at Project Syndicate. 

What a difference a year or two makes!  In 2021, interest rates were close to zero in the US and the UK,  and slightly negative in the eurozone and Japan.  They were expected to remain low indefinitely.  Remarkably, as recently as January 2022, investors thought that the probability the interest rate would rise above 4.0 % within 5 years was only 12% in the US, 4 % for the euro-zone, and 7 % for the UK [p.45].  Those were short-term nominal interest rates.  Correcting for expected inflation, real interest rates were substantially negative and expected to remain so.

Even after most central banks started raising interest rates – the Fed in March 2022 – it was in reaction to higher inflation.  So, real interest rates were still substantially negative, as recently as a year ago.  Furthermore, short-term interest rates rose more than long-term rates did that year:  The yield curve inverted in October 2022, signaling that financial markets thought the central banks would soon be bringing short-term rates back down.  This expectation was in turn attributed to confident forecasts of coming US and global recessions, which in 2022 — and up to a month or two ago — were the conventional wisdom.

  1. The reversal

Much less time than five years has passed since January 2022, and yet the short-term nominal interest rate hass risen to 5.25 percentage points in the US. Real interest rates are now positive in the US and many other countries. Forecasts of future interest rates are also now well above zero.  (One reason for the change in outlook is that, suddenly, it turns out that the US might avoid a recession after all.)

If it were only investors who had swung in a short time from a zero-rate outlook to positive rates, that could be shrugged off.  Financial markets are notoriously fickle.  But a more fundamental paradigm reversal may be coming among academic economists as well, and that is something that happens less often.

  1. The zero equilibrium interest rate

By 2021, many monetary economists had decided that the equilibrium (or neutral or “natural”) real interest rate had fallen to zero or even below.  The neutral interest rate has been defined as the interest rate that safe bonds would pay if aggregate demand were equal to potential output and unemployment were equal to its natural rate. (Central banks generally try to keep the actual interest rate close to the natural interest.)  This outlook was thought to be more or less permanent, with the exception of occasional cyclical fluctuations, e.g., transitory increases in the interest rate at times when fiscal policy was unusually expansionary or dips in recessions.

This zero-rate perspective was important. With a long-term target for the average inflation rate of 2 %, the implication seemed to be that the equilibrium nominal interest rate would have to be below 2 % on average, which would leave insufficient room to cut interest rates in time of recession.  Unfortunately, nominal interest rates can’t be pushed much below 0, due to the famous Zero Lower Bound, an example of what Keynes called a liquidity trap.

To be sure, it had turned out in Europe and Japan that nominal interest rates could be pushed a bit below zero: as low as negative 0.5%.  But that was the Effective Lower Bound; they could be pushed no lower.  The reason is that, hypothetically, if nominal interest rates were as low as negative 1.0  percent, people would take their money out of the bank and store it in safe deposit boxes so as to avoid the negative rate of return.  The monthly costs of the deposit box would be less than the penalty for holding money in the form of bank deposits or bonds.

If the equilibrium real interest rate was sometimes negative and the effective lower bound on nominal rates was close to zero, we would be in trouble. In this worrisome situation, monetary policy, much of the time, would be too tight to deliver the equilibrium rate of growth in the real economy. (Hence the interest in “Unconventional Monetary Policies.”)  At best, the job of sustaining full employment would have to be handed back to fiscal policy, which was politically fraught. This was the secular stagnation hypothesis, revived and made famous by Larry Summers ten years ago, in 2013, and seconded by others.

  1. Implications for rising debt

When it came to fiscal policy, there was a more encouraging implication of chronically low nominal and real interest rates:  they had fallen below the rate of growth of the economy [“r<g”].  This would render high levels of government debt sustainable.  The government could run a primary budget deficit (that is, a deficit even excluding interest payments), and yet debt could be sustainable, because it would be on a decreasing path relative to GDP.  Indeed, negative real interest rates helped reduce the US debt/GDP ratio between 2020 and 2022.

It was suggested that the single best indicator of debt sustainability was not the current debt/GDP ratio, but rather the current interest bill / GDP, i.e., debt/GDP times the interest rate.  Because interest rates were so low, governments could continue to borrow.  An unkind interpretation of this view would be that it underestimated the probability of a future sharp rise in interest rates, which would suddenly render the debt unsustainable, as many an Emerging Market country has learned.

Now that interest rates have risen above the rate of economic growth, the US debt is suddenly a problem again. Debt/GDP is expected to resume its upward path, from here on out.  (The same, globally.)  This was one of the reasons that Fitch Ratings on August 1 downgraded US debt from its longstanding AAA credit rating.

  1. The historical downward trend in interest rates

It is not too hard to see why investors and economists alike had concluded by 2021 that equilibrium interest rates had fallen to near-zero more or less permanently.  US short-term rates had been near-zero for nine years [2009-15 and 2020-21] out of the preceding 13 years, since the Global Financial Crisis of 2008.  Similarly for the Euro Area [below 1 per cent since 2009 and below zero since 2015] and Japan [below 0.5 % since 1996].  This had not happened before, at least not since the Depression of the 1930s, the era when Alvin Hansen originated the concept of secular stagnation.

The trend among major countries in both nominal and real rates had been downward ever since 1992 (and even since the 1980s, but that decade could be explained by temporary macroeconomic factors). More surprisingly, when Rogoff, Rossi, and Schmelzing[1] looked at seven centuries of data on long-term real interest rates, they found a long slow downward trend since the Renaissance, estimated at about 1.2 percentage points per century (between 0.6 and 1.8 percentage points), despite a lot of short-term fluctuations that can mask the trend.  The implication seems to be that equilibrium long-term real interest rates in the 21st century are entering permanently negative territory.  (An alternative interpretation is that the rates have asymptoted to zero.)

The decline in real interest rates was so widely taken as permanent, that these papers see their finding of a 7-century downward trend as a rejection of the proposition that the decline is only a phenomenon of the last four decades, rather than as a rejection of the proposition that the decline was only temporary.  But the latter hypothesis should be taken seriously. Statistically, the econometrician should expect to need a hundred years of data, to find significant evidence that fluctuations in such financial variables are in fact only transitory departures from a long-run equilibrium [2]

What might be the fundamental cause behind declining real interest rates?  A variety of possible reasons have been offered.  Suggestions to explain the last several decades include: slowing productivity growth (attributable to slowing technological progess), various demographic factors (including the post-war baby boom), rising global demand for safe and liquid assets (possibly due to increased risk perceptions), increasing inequality (where the rich save more), a global savings glut (attributable to high-saving East Asia), declining prices of capital goods, and more.  Explanations that might also explain declines in real rates going back several centuries include rising life spans.  (I might add declining transactions costs in financial markets, allowing increased competition among lenders.)


  1. Five percent vs. zero


It is not that eminent economists ruled out the possibility of future increases in interest rates. They recognized that there was a chance that interest rates could go up for a time.  But many thought it improbable in the near term and likely to be transitory in the long run.  Summers in 2018 wrote, ”we are likely to have, by historical standards, very low rates for a very large fraction of time going forward even in good economic times.”  Jason Furman and Summers in 2020 reiterated, “real interest rates are expected to remain negative” [p.6].   Olivier Blanchard, in June 2022, wrote in a very impressive book, “the long decline in safe interest rates stems from deep underlying factors that do not seem likely to reverse anytime soon” (p.28) .


What was said to be improbable has now happened.   The short-term interest rate is above 5 %.  Perhaps all those monetary economists are right, and the interest rate will revert to zero in the future.  But it seems to me that some may have been too quick to extrapolate from the record of 2008-2021.  After all, up until 2008 almost nobody thought that equilibrium interest rates could be zero.  Macroeconomists gave almost no thought to zero interest rates before 2008 (or 1998 in the case of Japan); and they thought of little else after.  If we changed our minds once, we can do it again.  Moreover, it is not easy to explain equilibrium real interest rates at or below zero in theory, so long as firms show any productivity growth and consumers show any impatience.

I don’t know what interest rates will do in the future.  But I don’t think they are going to snap back to zero anytime soon.  If this is right, it is good news for monetary policy: it will be less constrained than previously.  But it is bad news for fiscal policy, which is once again constrained by the prospect of ever-rising debt/GDP ratios.

[1]  The Schmelzing paper is from 2020.  The latest version of the paper that he coauthored with Rogoff and Rossi is dated 2023.

[2]  E.g., for the real exchange rate: Frankel (1986, pp. 55-57; revised in Frankel (1993, p.52).


 This post written by Jeffrey Frankel.

12 thoughts on “Guest Contribution: “The End of Zero Interest Rates”

  1. pgl

    Excellent essay. Permit me to ask for a wee bit of claification on this sentence:

    ‘It was suggested that the single best indicator of debt sustainability was not the current debt/GDP ratio, but rather the current interest bill / GDP, i.e., debt/GDP times the interest rate.’

    I would agree if “the interest rate” means the real interest rate. I get the sense that Dr. Frankel is saying that given his appropriate focus on real rates. I only ask because a previous post by Dr. Chinn plotted this using nominal interest expense. I asked in a comment but my point seems to be not quite understood.

    1. Jeffrey Frankel

      PGL : As you say, real interest payments would be a more accurate predictor of debt sustainability than nominal. (Better yet, (real interest rate – real growth rate)(Debt/GDP). But the nominal interest rate can be observed directly, while the real interst rate (and real growth) require some assumptions or estimation,

  2. pgl

    “More surprisingly, when Rogoff, Rossi, and Schmelzing[1] looked at seven centuries of data on long-term real interest rates, they found a long slow downward trend since the Renaissance, estimated at about 1.2 percentage points per century (between 0.6 and 1.8 percentage points), despite a lot of short-term fluctuations that can mask the trend.”

    Looking back to the Renaissance? Forgive me but this sort of made me think about how CoRev wants to take the Global Warming debate back to the Ice Ages.

  3. pgl

    The header of slide 14 from Furman-Summers:

    The more relevant metric of real debt service expected to stay low relative to the economy
    U.S. Federal Real Interest Payment

    Not to harp on this but this was the graph I was hoping Dr. Chinn would have used in his previous post on this.

  4. pgl

    Slide 9 of Furman-Summers

    ‘Implication 2: Lower interest rates necessitate new measures of a country’s fiscal situation’

    They refer to a forward looking flow measure based on the present value of the government’s primary surplus relative to the current debt. This is NOT new. Several papers over 40 years ago were all over this (Barro and Sargent-Wallace and many more) addressed this. But then it was taken as a given that r would be at least as high as g. But yea that changed over the period following the Great Recession. But as this excellent post notes r > g is back.

  5. baffling

    inflation drove rates up in the past couple of years. quickly. the driver for that inflation does not appear to exist today. inflation is falling. the fed will, mistakenly, keep rates high over the next year. for reasons that are not economic, but face saving-it will be hard for the to admit a reversal soon. inflation will continue to fall. interest rates will not return to zero. but they will stay in the 2-4% range starting in 2024. recession is a greater risk than higher inflation over the next 12 months. that is what my crystal ball says.

  6. pgl

    The link to the 1-Year Real Interest Rate (as reported by FRED) is interesting. This rate reached 8% as of June 2023. While very high, this rate was 9.3% during June 1982. That’s how bad the toxic Reagan-Volckeer macroeconomic mix became.

  7. Anonymous

    Some such as retweets at Zero Hedge are comparing central bank holdings divided by GDP.

    Japan is highest%, ECB, less than Japan, and Fed lesser of the 3.

    The Fed’s reverse repurchase agreements are interesting. Last week they priced at 5.3% annual rate, while the $ volume is off peak, it remains substantial, and pays good to take in cash. Reverse repurchase at high volume are a relatively new phenomenon.

    The central banks seem to presume high real rates are needed for cooling the price trends.


    Central banks have set the rate in attempt to rein in rising price trends. There remain huge stashes of cash, last week reverse repo we’re over $2.2 trillion, that is a lot of cash temporarily out of reserves, and a lot of treasuries that don’t show on the balance sheet

  8. Macroduck

    Given the seven-century decline in real long-term interest rates, explanations for the decline in real rates in the past few decades probably amount to explanations of variability around trend. Similarly, the newly adopted view that real long-term rates may not be permanently negative is, for now, mostly speculation about variation around trend.

    Frankel notes that drivers of the long-term decline in rates probably include extended life-spans and increased competition among lenders. Do either of these, or any other likely explanations for a seven-century downward trend in r*, suggest an underlying long-term zero or negative real lending rate? As long as a grain of barley planted in the ground can be expected to yield many grains of barley over time, won’t it remain incumbent on borrowers to find uses for funds which produce positive yields?

    Does climate change imply a fundamentally different trend in interest rates? Life spans and total factor productivity, and so returns on assets, are all likely to be different when water and useful land are in shorter relative supply.

  9. JohnH

    Wow! After more than 2 decades of paltry returns for folks trying to save to buy a house, save for their kids’ college, or eventually have enough to retire, it looks like savers might actually get a positive inflation-adjusted return on their investments. A positive, real time value of money! What a concept!

    pgl must be devastated…

  10. Vasja

    This post is informative, insightful, and links to some of the best research on natural rates. Thanks a lot!

    And, personally, for me, the dilemma centres around the optimal mortgage rate fixation period. If the natural rate remains persistently distant from zero, opting for the current high 30-year fixed mortgage rate is a prudent choice. Conversely, if the natural rate is expected to revert to zero within the next five years, opting for the 5-year fixed or variable rate would be better.

    I am leaning towards the 5y fixed since I believe that the natural and real rates will again decline. In section 4, Professor Frankel explores potential explanatory factors for the prevailing low rates, such as decelerating productivity growth, demographic influences, and more. None of these factors have undergone significant shifts, so once the inflation subsides I expect the interest rate to decline toward an “unchanged” close-to-zero natural rate. So I wonder why Professor Frankel concludes “I don’t think they are going to snap back to zero anytime soon”? The only significant shift that I see is the end of QE-like policies. If QE-like policies caused the low (natural) rates seen in the last 5y and these are phased out, we could return higher average rates.

Comments are closed.