Is secular stagnation over?
I do not think so. Today’s inflation will not last, but I believe that low rates will.
Secular stagnation, a concept put forward by the economist Alvin Hansen in 1938, refers to economies suffering from structurally low private demand, requiring very low interest rates to sustain demand and achieve potential output—indeed requiring interest rates so low that they are lower than the growth rate. This state is summarized by a formulation that has become famous (at least in macroeconomic discussions…): r < g, a safe interest rate lower than the growth rate of the economy. In 2013 Lawrence H. Summers stirred widespread interest by declaring that the advanced economies were in a state of secular stagnation.
The empirical evidence was indeed overwhelming that most economies were in such a state. Real interest rates (i.e., interest rates adjusted for inflation) had steadily declined since the mid-1980s and were much lower than growth rates, indeed typically negative. Accordingly, Summers and I (and many others) explored the fundamental implications of such a condition. I focused on the implications for fiscal policy, and my new book, drawing conclusions for fiscal policy, has just been published by MIT Press.
In the current inflationary context, however, the title of my book, Fiscal Policy under Low Interest Rates, seems like a provocation. Some will say the publication may be a nice history book but that it is already out of date because the world has changed. According to this view, interest rates are now high, not low, and we should now prepare for a world in which r > g on average. Indeed, Summers himself, at the recent meetings of the American Economic Association, opined that "we’ll not return to the era of secular stagnation."
Disagreeing about secular stagnation with the first macroeconomist to identify it rightly as a contemporary phenomenon is something that one should do very carefully. Yet I must.
First, because the issue is a fundamental one: There may be no variable more important for macroeconomic policy than r − g.
Second, because I believe that global secular stagnation was and is driven by deep structural factors that neither COVID nor inflation have done anything to reverse. Once central banks have won the fight against inflation, which they will, we shall most likely return to a macroeconomic environment not dramatically different, at least in this respect, from the one before COVID. That means that safe interest rates—that is, rates for assets with no risk of default—will be low again. As a result, I see a high probability that r < g remains the prevalent regime for some time to come.
Why do I believe rates will remain low?
The decline in safe rates on government bonds (adjusted for expected inflation) has been a steady and global one since the 1980s. The decline may have started earlier in Japan, but essentially all advanced economies shared the condition until the post-COVID inflation of 2021. This trend did not result from particular events, say the global financial crisis or the COVID crisis: while these led to temporarily lower rates, these lower rates were short-lived deviations from an underlying downward trend.
What market rates tell us
This process was indeed interrupted by the rise of inflation starting in 2021, which has forced central banks to increase interest rates to get inflation down to target. But even now, at what may be close to the height of central bank interest rates in their fight against inflation, real interest rates are still surprisingly low (and real rates are what matters for r − g).
In the United States, the 10-year nominal rate on government bonds is currently 3.4 percent, while the 10-year inflation forecast by the Congressional Budget Office (CBO) is 2.4 percent, implying a 10-year real rate of 1.0 percent. At the same time, the 10-year CBO growth forecast is 1.7 percent, implying that, even today, the 10-year forecast for r − g is still negative: −0.7 percent. Similar computations for the eurozone and for Japan yield even more negative values for the 10-year r − g: −1.3 percent for the euro zone, −1.2 percent for Japan.
These are 10-year averages, and thus reflect both the forecasts of higher short rates in the near future and the depressing effects on long rates of quantitative easing on long rates. An alternative is to use data from the overnight index swaps curve for 1-year bonds 10 years out and from 1-year inflation swaps also 10 years out, together with the CBO forecast for 1-year growth 10 years out. These imply an even greater negative value of r − g: −1.2 percent 10 years out from today.
Or one can instead look at what the Fed expects to happen. The December “Summary of Economic Projections” based on forecasts from the Federal Open Market Committee gives a median long-run forecast for the policy rate of 2.5 percent, an inflation forecast of 2 percent, and a GDP growth forecast of 1.8 percent. These imply a real rate of 0.5 percent and a value for r − g of −1.3 percent.
Investors, and even the Fed, can be wrong, however. Indeed, they did not predict the current increase in nominal rates. So, one must dig deeper.
Looking at fundamentals
One must try to both identify the factors behind the earlier decline in rates and forecast their evolutions in the future. Conceptually, one can think of three sets of factors. The first two are the factors affecting saving and the factors affecting investment; together, they determine capital accumulation, and by implication the economy’s marginal product of capital and the distribution of real rates depending on their relative riskiness. The third are the factors that affect the preference for safety among investors, which in turn determines how low the safe rate is relative to more risky rates of return such as stock returns.
It is fair to say that, while the (abundant) literature has identified the likely factors at work, it has not pinned down the relative contributions of these three sets of factors. Some give a main role to stronger saving and weaker investment. Others focus on an increased demand for safe assets. Yet I believe that, for most of the relevant factors, there is no reason to expect a dramatic turnaround from their pre-COVID evolution.
Take saving. I see the two main forces behind high saving as demographics and higher income. To the extent that longer life expectancy—which dominates demographic evolutions in advanced economies—leads to a longer retirement, it induces people to save more in anticipation of their longer retirement. Higher income levels (rather than higher income growth) increase the proportion of the population that is able to save, either for precautionary motives or for life cycle saving purposes. These factors will continue to be relevant in the future.
Summers has argued that the increase in public debt due to the fiscal response to COVID will lead, other things equal, to an increase in r. He is right about the sign of the increase in public debt’s impact on r, but the effect is likely to be quite small. The debt-to-GDP ratio in advanced economies has increased only from 75 percent in 2019 to 82 percent in 2022; under standard assumptions, this implies an increase in r of no more than 15–30 basis points. That would be insufficient to offset the pre-COVID downward trend in safe rates, let alone to close the gap between r and g.
Turn to the demand for safe assets. A rise in this demand was an important factor in leading to a larger discount on the safe rate pre-COVID. In particular, regulation requiring financial institutions to hold a higher proportion of liquid assets clearly played a role. This is unlikely to change. And, if anything, the world looks more uncertain than it was pre-COVID, leading not only to more precautionary saving but also to a higher proportion of wealth to be held in liquid and safe assets (and fewer countries’ assets being considered safe).
This leaves the likely evolution of investment, and this is where I freely admit that there is more uncertainty. The rate of technological progress is notoriously difficult to predict from decade to decade. If progress were to speed up again, the effect on r − g would be ambiguous: it would lead to higher growth, which would lead to an even lower r − g; but it would also trigger higher investment, which in turn would lead to a higher r and thus a higher r − g. Such a technological explosion did not happen in the last 40 years, but it could.
One can think of other reasons why investment might become stronger. Geopolitics suggest that defense spending, a form of investment, may go up. Reshoring and friendshoring, for security or other reasons, may imply both higher investment and possibly lower growth as some of the benefits of trade are lost. The fight against global warming will increase green investment, while at the same time potentially slightly decreasing growth. All these may lead to an increase in r − g and thus reduce the room for fiscal space and the use of fiscal policy. It would however take an unusual investment boom to reverse the sign of r − g and this cannot be the baseline scenario.
In short, one can never be sure. But I see the inflation episode and the higher rates as an interlude. Looking beyond this episode, I see low interest rates and a negative r − g as the most likely scenario for the future. If I am right, this is the time to think about fiscal and monetary policy once the fight against inflation is won.
1. It has been argued that longer life expectancy means a larger proportion of dissavers and thus lower saving. This is incorrect. While it is true that the proportion of dissavers goes up, over their lifetime people need to save more in anticipation of a longer retirement, leading to a higher aggregate saving rate.
2. To be clear and as I argue in the book, such an increase in r, if it came from an increase in desirable investment, would be a good thing, even if it reduced fiscal space.
3. A summary of what I see as the main implications for macroeconomic policy in general, and for fiscal policy in particular, is in my previous blog.
This publication does not include a replication package.