Global real rates, 1311–2018

Paul Schmelzing

Paul Schmelzing is an academic visitor to the Bank of England, currently based at Yale University. In this guest post, he summarises his research on the differential between real interest rates and real growth rates over the past seven centuries…

There is a lively academic and policy debate about whether a build-up of excess savings in advanced economies has created a drag on long-term interest rates. In a recent paper, I provide new context to these discussions. I construct a long-run advanced economy (DM) public real interest rate series geographically covering 78% of DM GDP since the 14th century. Using this series, I argue that current interest rate trends cannot be rationalized in a “secular stagnation” framework that has been “manifest for two decades”. Rather, historical data illustrates that advanced economy real rates have steadily declined for more than five centuries, despite important reversal periods. This post draws from long-run economic history to provide additional insights concerning the current interest rate environment.

Public rates, nominal and real

Chart 1 displays the new 707-year series in nominal and real terms compared to two existing series. The two existing series include: the “lowest-yielding” semi-centennial trend in Sidney Homer’s and Richard Sylla’s classic book (green series), and data from a recent compilation by Huang, Chilosi and Sapoznik (2019) featuring municipal annuities between the 14th–18th centuries (red series, in which I arithmetically weigh all their nominal data points).

The yellow and blue series display the new “global” data in nominal and real terms: these have to be constructed from a wide array of scattered archival and printed sources, and should be thought of as a DM public long-term debt series that incorporates both consolidated and unconsolidated voluntary lending across the full risk spectrum. In this sense, they are comparable to the “global” series presented by King and Low (2014), with the key differences being that I measure ex post (rather than ex ante) inflation, and GDP-weigh my sample throughout.

Chart 1: GDP-weighted “global” public real rates and previous series, 1311–2018

The data are less suited to illustrate shorter-term, cyclical-level fluctuations: except for some geographies, early modern cyclical dynamics remain simplistic given my need for interpolations. However, robustness increases at the decadal level, and I argue that the compilation goes substantially beyond existing aggregated series, thus serving to illustrate more granular long-run dynamics. For instance, we can observe the apparent existence of two secular reversal periods, during which public capital markets broke from their general downward trend and entered lasting periods of higher real rates, ca.1320–1480, and ca. 1520–1650, highlighted in Chart 2. The first episode coincides with the Hundred Years War, the European “Bullion Famine”, and raging Condottiere warfare in Northern Italy. The second covers the “Triple Default of 1557–1559”, the long global wave of bank and merchant defaults, and accelerating population growth from the second half of the 16th century.

Chart 2: Public R-Reversals, advanced economies, ca.1320–1480s and ca.1520s–1650s

Safe(r) real rates: evidence from private long-term debt

Aggregated series such as those in Charts 1 and 2 incorporate premia over the “safe” rate, which reflects default, inflation, or debasement expectations. Is it possible to obtain a series closer to the “safe” rate? One way to tackle this is to construct an (ex post) default-free series of leading political powers with strong commitment mechanisms over time, an approach discussed elsewhere. Chart 3 on the other hand displays German private long-term mortgage contracts (today’s Pfandbriefe) since the 13th century, in nominal and real terms, sourced from a variety of municipal archives ranging from Frankfurt over Mainz to Cologne, plus printed sources. It has been argued that such contracts represent a useful approximation of a “risk free” financial asset over time, since they were secured by real estate, fully transferable, and exempted from any usury laws.

Such rates trade notably below public rates during 1300–ca.1650, though one has to remember that they are not of course entirely riskless (there being no truly riskless asset). I similarly observe that such private, relatively safe rates have been falling since the late Middle Ages – more steadily and less aggressively (by -0.6 basis points p.a. in real terms) than public series, but similarly unimpressed by major historical institutional and regime changes.

Chart 3: German nominal and real mortgage rates, and “private R”-G

A derivative measure, “R-G”, has attracted interest in the context of debt sustainability debates recently. R-G is the difference between the real long-term interest rate (R here is not referring to the general return on wealth debated in the context of inequality trends) and real GDP growth. I reconstruct a new, long-term R-G by combining the new “R” with existing GDP data. Though early modern German GDP estimates are less robust than similar ones for Britain, Spain, or Italy, there is a common trend in each country R-G series. Namely, the presently depressed R-G environment as discussed by Blanchard or Barrett (2018) has been long in the making, and the general decline in the spread has not previously been permanently reversed by exogenous shocks (note the “Kipper and Wipper” plunge of 1619–1621) beyond the short-term. Secularly, advanced economies have in this sense continuously improved debt sustainability: by roughly 1 basis points (bps) per annum in the case of “private R”-G, with the spread itself averaging just under 2.5% over the past 500 years (Chart 3).

Historical versus “neutral” rates

Willem Buiter sensibly pointed out that even the narrower “safe R” approximations I use (be it the default-free public R sample, or the private R series) may not be consistent with the “neutral rate” concept often employed in the context of the secular stagnation debate. Indeed, since the measurement of “neutral” rates even for recent years is not entirely straightforward, its construction for early modern markets presents additional challenges. However, since I isolate tradable, transferable, and repression-free instruments, with a meaningful bias on relatively financialised municipalities, a smoothed version of the “safe public R” or the “global R” series should allow us to reasonably approximate a steady state condition. Jordà, Singh and Taylor’s exercise along these lines on the basis of my data appears to suggest exactly that: “neutral” real rates have just as well declined since the Renaissance, and on this basis, too, developments since the 1980s equally mark a “return to historical trend”.

Chart 4 now illustrates these points by grouping all real rate data according to three broad risk profiles – the “safe” private series, an ex-post default-free public series tracing the leading economy over time, and a “risky” public real rate series, which includes historical default events. A finer dissemination is certainly desirable, but once more these exercises keep suggesting that simply ignoring long-run real interest rate trends by invoking “risk premia” or “neutral rate” dynamics is too simplistic, and the idea that real interest rates have historically trended around a “normal”, “stable” level correlated to factors such as GDP growth, remains highly doubtful.

Chart 4: R trends across three risk classes, advanced economies, 1311–2018

Liquidity and market integration back then

Similarly, GDP-weighting and aggregating separate municipal-level and country-level observations should not obscure the fact that financial market integration was far from perfect in the medieval and early modern periods. This is despite evidence of significant progress during the 15th century towards financial market integration, and despite the fact that the debt contracts considered here were overwhelmingly denominated in international gold currency, the Florentine Florin, or its German derivative, the Rheingulden.

It is often assumed that serious market depth for public debt in advanced economies was only achieved in 17th century Holland and 18th century Britain. High levels of pre-1700 interest rates are against that backdrop often assumed to reflect in part prohibitive levels of illiquidity. However, actual evidence is more nuanced. No doubt, various early modern geographies fall short of even remotely comparable financial market depth, particularly in peripheral European towns and regions. But on the other hand, political bodies in Northern Italy, the Holy Roman Empire, France, and Flanders often showed real per capita levels of debt stocks and annual funded debt issuance exceeding those of 18th century England or 17th century Holland. In general, it appears rather underestimated how “financialised” the pre-1700 era already was. To that end, Chart 5 calculates real per capita funded debt stocks in a range of “advanced economies” between 1313–1500, benchmarking against the 1752 English real consolidated debt stock per capita. Full sources to these calculations are available here. We note that the majority of the geographies included in our sample at least reach a 65–80% debt level of our benchmark, and often comfortably more. A handful of geographies fall below such a level – but even if we decided to omit such issuers entirely and imposed a strict 80% threshold, our aggregate trends would only be little changed.

Chart 5: Real funded debt stock per capita outstanding, selected locations versus 1752 British levels, 1313–1500

Where next?

With recourse to new archival and printed data, I show that real interest rates trended downwards, typically in a range of 0.5-2bps p.a., since at least the days of Charles VIII’s triggering of the Italian Wars in the 1490s and across geographies, monetary and financial regimes. Such a statement obscures much historical and empirical nuance, but the new evidence is able to qualify “secular stagnation” propositions: the trend fall since the 1980s appears much less of an aberration once we factor in deep history. Liquidity and other risk premia components are of substantial importance in the early modern period, but they cannot fully explain the downward trend fall over time. I suggest that there must remain additional long-run drivers working on advanced economy public and private markets across the DM risk spectrum for half a millennium now. Significant short-term fiscal or monetary responses, or fundamental changes to the international financial regime (such as the transition to fiat money), have often caused sharp initial volatility (and have historically by no means been “costless”), but they have not thus far permanently upended these long-run drivers.

If you want to get in touch, please email us at or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.

5 thoughts on “Global real rates, 1311–2018

  1. Can you please explain why 1780-1790s had this sharp drop in real interest rates? I can understand the situation in 1920s and later, but this part is a puzzle.

  2. A wonderful piece which has changed by view on long-tern real rate trends. A few typos need correcting: the title of Figure 3 “private R”-G should be “private R-G” and I think “weigh” should be “weight” in a few places. But a magnificent piece of work.

  3. Is this trend (and some of its reversals) not related to effectively how ‘money like’ such neutral debt instruments are? Or in a softer version of that possible theory, is more of the variability not possibly explained by the changing nature of ‘safe’ debt instruments? Thank you for this important contribution.

  4. I would love to understand how the author can omit China and India from the definition of advanced economies over the observed time frame, considering that these economies were substantially larger at the beginning and towards the end of the period. Also, how would their inclusion change the results?

  5. Andrew Sheng – something to do with the formation of the USA? And the French?

    Apart from that, I’m looking at the chart and wondering about network effects and the consumer surplus.

Comments are closed.