In this note, we review two recent papers: “Natural and Neutral Real Interest Rates: Past and Future” by Maurice Obstfel and “Money growth and the post-pandemic inflation surge: Updating the evidence” by Claudio Borio (BIS), Boris Hofmann (BIS) and Egon Zakrajsek (Boston Fed). The first paper is a review of the contributions on R-bar (the real rate of interest that persists in a flexible-price equilibrium) and R* (the real policy rate that keeps prices stable). Obstfeld expects the recent rise in real rates to be short-lived and predicts a continuation of the trends of the last 30 years (for the record: we think that R* might have increased recently and that the risks are to the upside). The second paper updates previous results on the link between money growth and consumer prices; the result is that the link survives the inclusion of recent data. Having said that, money growth continues to be a useful indicator only in periods of high inflation (which can be detected only ex-post), making it a poor instrument for policy purposes.
Paper #1: Obstfeld (2023)
A recent paper (Natural and Neutral Real Interest Rates: Past and Future) by Maurice Obstfeld (Berkeley/Peterson Institute) provides an overview of the literature on neutral /(vs)natural rates. The paper provides a history of the global decline in real rates since the 1990s and discusses the main contributing factors (demographics, excess savings from emerging economies, safe asset demand, rising market power). Obstfeld predicts the past (downward) trends in real rates to continue going forward and foresees the ELB as a relevant concern for central banks.
(Note: we have recently reviewed the R* literature here and here. A PPT of the literature review is here).
What the paper does
Many papers focus on estimating long-run equilibrium interest rates (r-bar), though these need not coincide with the neutral interest rate (r-star) that is relevant for monetary policy. The natural rate of interest (r-bar) is the real rate of interest that persists in a flexible-price equilibrium. In contrast, the neutral rate (r-star) is the real policy rate that keeps prices stable. These two objects can coincide in some models but need not be the same. Obstfeld argues that many of the empirical frameworks used to guide monetary policy are conceptually flawed because they focus on the wrong object of interest (r-bar instead of r-star). In addition, he argues monetary policy should also consider factors other than r* such as the current account balance, financial conditions, and imperfect policy credibility.
Main results
A variety of methods to estimate the neutral/natural rate exist, though all have their own types of shortcomings. Some methods estimate long-run trends in real rates using nonstructural time series models, while others use a bond pricing model to extract information about expected long-run real rates. Another way is to use a calibrated structural model and solve for the flexible-price equilibrium. All of these methods estimate r-bar as a proxy for r-star. Hybrid approaches exist as well, and may yield estimates more closely interpretable as r-star. While all methods yield results pointing to a downward trend in real rates, the point estimates can differ substantially (see Figure 1). Further, the estimates often come with wide error bands (not shown) and are sensitive to models- specification.
Figure 1. Natural rate estimates from different methodologies, from Obstfeld. ‘HLW’ is Holston-Laubach-Williams
There are common trends in global interest rates, but these are likely driven by a range of factors with varying importance over time. The paper distinguishes three periods of the global real rate decline: early 1990s-2000, 2000-2008, and 2008-2018. Phase 1 experienced a rise in saving from baby boomers in advanced economies, lower investment, and rising corporate market power. Advanced economies generally ran current account surpluses. Phase 2 was characterized by a growing US current deficit, partly driven by a buildup in foreign exchange reserves by emerging market central banks, but also by rapid income growth in China and some oil-exporting countries leading to strong savings growth. Risk tolerance in financial markets was high. Phase 3 starts following the GFC and sees stronger risk aversion and hence safe asset demand. The Euro area debt crisis and higher political uncertainty reinforced this trend throughout the 2010s. The ELB was a consequence of these trends but may itself have contributed to growing uncertainty. Lower growth post-GFC may have reduced investment demand. Trends in foreign exchange reserves stabilized. Throughout all three phases, global ageing trends and a slowdown in productivity were key contributors to falling real rates.
Looking ahead, Obstfeld expects the recent rise in real rates to be short-lived and predicts a continuation of the trends of the last 30 years. Investment can be expected to remain weak as long as GDP growth is not projected to pick up significantly, although green investment and military spending may be exceptions. Demographic trends will continue and likely be a key driver. Uncertainty remains high. This suggests the ELB remains a concern for central bankers. On the other hand, low interest rates give fiscal policy more space.
A comment to the paper
The paper provides a good criticism of existing methods and hints at relevant areas for improvement in state-of-the-art models. However, it doesn’t offer attempts of its own to provide better estimates. While this may well be beyond the scope of the paper, it may also be a reflection of the fact that the current state of the literature appears dire from a practical point of view: Model estimates can yield significantly different yet imprecisely estimated results for (proxies of) r-star, and there is simply no easy answer on how to improve state-of-the-art methods to achieve reliable policy guidance from such models in the near future. Indeed, there is an important conceptual distinction between estimates of r-bar and r-star. While the two measures can coincide, they need not be the same. An implicit assumption in many empirical applications that aim to measure r-bar using long-run averages is that nominal rigidities and temporary shocks vanish at long horizons but it is not certain that all economic state variables reach their steady state over those horizons.
Policy implications
According to Obstfeld, real rates can be expected to continue their downward trend as multiple drivers of the past 30 years are likely to persist, especially demographic changes and the high demand for safe assets. The ELB therefore remains a concern for policymakers in the medium- to long-term. Fiscal policy and unconventional monetary policy tools may have to become the key tools of stabilization policy.
(For the record: we are not fully convinced by Obstfeld’s view and we invite the reader to take a look at our R* PPT here. In our view, R* might have moved up recently and the risks are to the upside)
Paper #2: Borio et al. (2024)
A new article (Money growth and the post-pandemic inflation surge: Updating the evidence) by Claudio Borio (BIS), Boris Hofmann (BIS) and Egon Zakrajsek (Boston Fed) updates previous results on the link between money growth and inflation (a year ago, we reviewed the original article here) using longer samples. The updated evidence provides nuance to the previous findings. On one hand, it strengthens the robustness of the findings for the period 2021–22, when the inflation surge took place. The cross-country link survives when controlling not only for the post-pandemic rebound in economic activity, but also for the fiscal response to the pandemic. On the other hand, the new analysis casts doubt on the persistence of the link once inflation declines. Though money growth can carry useful information for policy, using that information reliably is still a question mark.
What the paper does
The authors define two inflation regimes (high vs low) instead of focusing on the level of inflation directly. The small movements in the price level during the low-inflation regime are predominantly driven by relative price changes across sectors. In contrast, price level movements in high-inflation regimes become synchronized across sectors and the probability of a wage-price spiral increases. Based on this, the authors study the correlation between money growth and consumers’ price inflation
Main results
Higher money growth is associated with stronger inflation rates during the Covid episode. The authors compare excess money growth (money growth minus real GDP growth) in 2020 with the average inflation rate in 2021/2022 for a set of advanced and emerging economies. Figure 2 shows the main result. For the full sample, the relationship is close to one-to-one and statistically significant (panel A). This result is partly driven by the close relation between money growth and inflation in countries in the high-inflation regime. Excluding Argentina and Turkey from the analysis weakens the coefficient to 0.34 (panel C). The same holds for inflation forecast errors (panels B and D): Countries with larger money growth experienced stronger inflation surprises.
Figure 2: Main result from Borio et al. (2024): Excess money growth vs inflation rates/forecast errors
The results confirm previous evidence on the state-dependence of the money-inflation link. The literature generally finds that money growth is associated nearly one-to-one with inflation when high-inflation countries are included but the relationship weakens for countries and periods with lower inflation rates. The authors also confirm that the results are robust to controlling for the size of fiscal stimulus and for the strength of the inflation rates in 2022-2023.
A critique of the paper
The paper provides a good overview of the empirical facts on money growth and inflation during the Covid period but suffers from the usual limitations. It covers a wide range of countries and provides both raw correlations as well as evidence with some key controls for fiscal stimuli and the effects of economic recovery. The authors are transparent in their focus on associations with no claim to establish a causal link between money growth and inflation. Indeed, the limitations of MV=PQ are well known: (i) the observation that money growth today helps to predict inflation tomorrow does not, in and of itself, imply causality (which in fact can run the opposite way), (ii) in some cases (i.e. demand-driven, as in 2020, when companies drew heavily on their credit lines) the increase in M is largely endogenous, (iii) if the causality works only in high inflation regimes, MV=PQ is almost impossible to use in real-time because the regime switch becomes clear only ex-post, (iv) if any, monetary aggregates are correlated with price increases only at low frequency, and (v) a positive correlation between money growth and price inflation does not imply the same during disinflation periods (or deflation as some have argued in the last year), as there is a degree of price rigidity.
Policy implications
Monetary aggregates can be useful predictors of inflation rates but only in high-inflation regimes. Policymakers may therefore want to pay attention to the growth rate of money as a possible way to improve inflation forecasts. However, the success of the forecast will ultimately depend on the correct assessment of the inflation regime, based on which the slope of the relation between money growth and inflation changes strongly. The key point is that estimating the probability of a regime switch requires a model itself, making the use of monetary aggregates a poor instruments for policy pursposes.