In the early 1690s the English crown was borrowing at rates that could reach 14 percent. By the 1750s the British state was issuing perpetual annuities at 3 percent. Few price series in economic history have carried as much interpretive weight. For Douglass North and Barry Weingast, the decline was the signature of credible commitment. After the Glorious Revolution, Parliament controlled taxation, and that control made the sovereign a trustworthy debtor.

Yet the same kind of evidence has supported the opposite conclusion. Gregory Clark examined returns on land and private rent charges and found no structural break around 1688. Nathan Sussman and Yishay Yafeh showed that British sovereign yields stayed high and volatile for roughly three decades after the settlement. They converged toward Dutch levels only from the 1720s onward. One political event and one family of price series produced two incompatible readings.

The disagreement is not a failure of quantitative history. It shows how the evidence has to be handled. Interest rates are composite prices. They bundle compensation for time, inflation, default, illiquidity and the cost of enforcing a claim. Read naively, a single rate tells us almost nothing. Decomposed carefully, and compared across borrowers who share some shocks but not others, rates become one of the few continuous, comparable and high-frequency measures of institutional quality available for the pre-statistical past.

Start with an accounting identity. A nominal yield equals the sum of a real risk-free rate, expected inflation, a default premium, a liquidity premium and, for longer maturities, a term premium. The identity is trivial. Identifying its parts is not. Historians observe the left-hand side, sometimes at monthly or even daily frequency. Every component on the right must be inferred from auxiliary data, from theory, or from carefully chosen comparisons.

Expected inflation is the first component to strip out, through the Fisher relation. For much of the period before 1914, metallic standards kept long-run price levels fairly stable, so researchers are tempted to treat nominal and real rates as interchangeable. That shortcut fails during debasements, paper-money episodes and wartime suspensions such as Britain's Restriction Period of 1797 to 1821. In those episodes, ex post inflation from series like Phelps Brown and Hopkins is only a noisy proxy for what lenders actually expected.

The default premium is usually isolated by differencing. Suppose two instruments share a currency, a maturity and a trading venue but have different issuers. The spread between them then approximates the market's view of relative default risk. This is why gold-standard sovereign bonds traded in London are so valuable to cliometricians. A common denomination and a common market neutralise most of the inflation and liquidity terms, so the spread is a cleaner measure of perceived creditworthiness.

Institutional factors enter in less tidy ways, and usury ceilings are the clearest example. England capped interest at 5 percent from 1714. Peter Temin and Hans-Joachim Voth have shown that the cap bound private lending through much of the eighteenth century. A rate observed at the ceiling is therefore a censored observation, not a market-clearing price. The true scarcity of credit shows up instead in rationing, collateral requirements and lending volumes, which only quantity data can recover.

Liquidity and enforcement costs complete the decomposition. A perpetual annuity traded on an organised exchange will yield less than an otherwise identical private mortgage, even when default risk is the same. If these terms are ignored, a falling government yield can be misread. It might reflect a deepening secondary market rather than a more trustworthy sovereign. Both are institutional changes, but they carry very different implications for theories of property rights and constitutional constraint.

Takeaway

An interest rate is a sum, not a signal. Its institutional information lives in whichever component you can credibly isolate, and that component usually appears as a difference between rates rather than in the level of any one of them.

The logic of inference is simple. A lender demands compensation for expected loss, which equals the probability of default multiplied by the loss given default. Institutions that make repudiation politically costly lower the first term. Institutions that make enforcement cheap, such as courts, registries and bankruptcy procedures, lower the second. So when borrowers face similar capital markets but live under different institutions, the differences in what they pay should carry information about those institutions.

Early modern Europe supplies a natural cross-section. Stephan Epstein and David Stasavage documented that city-states governed by representative assemblies, including Genoa, Venice and the province of Holland, borrowed far more cheaply than territorial monarchies. Holland's provincial debt yielded around 4 percent by the mid-seventeenth century, while the French crown routinely paid considerably more. Stasavage refined the claim further. What mattered was not simply the existence of an assembly but whether creditors held power within it.

The first era of globalisation offers sharper tests. Niall Ferguson and Moritz Schularick found that British colonies borrowing in London between 1880 and 1913 paid spreads roughly a percentage point below comparable independent states. They attributed this empire effect to imperial legal and fiscal oversight. Michael Bordo and Hugh Rockoff argued that gold-standard adherence acted as a seal of approval. Later work by Marc Flandreau and Frédéric Zumer suggested that fiscal fundamentals, especially the debt burden relative to revenue, did much of the explanatory work.

Three identification problems recur. First, good institutions tend to come with wealth, peace and deep capital markets, and each of these lowers rates on its own. Second, spreads measure perceived quality, and markets can misprice. Third, and most important for the 1688 debate, sovereign credibility is not the same thing as private property security. North and Weingast reasoned from government yields toward private-sector rights. Clark's evidence from private land returns showed why that step needs separate testing.

Best practice therefore borrows from modern econometrics. Difference-in-differences designs use a comparator market to absorb common shocks, as Sussman and Yafeh did with Dutch yields. Event studies with high-frequency price data test whether spreads moved on the dates of institutional reforms rather than on the dates of plausible confounders like battles, peace treaties or harvest failures. An inference about institutions is strongest when the spread responds to the reform and stays still when the confounders occur.

Takeaway

A lower interest rate certifies that lenders believed something. The historian's job is to establish what they believed, and whether the institution under study was actually the reason they believed it.

Financial development leaves two signatures in price data. The first is a declining level of rates, which reflects capital abundance and lower perceived risk. The second is narrowing dispersion across regions, borrowers and instruments. Dispersion measures market integration: whether capital moves freely enough to arbitrage away differences in its price. The two signatures can move independently, and keeping them apart is essential for any careful interpretation of institutional change.

Lance Davis's 1965 study of American regional interest rates remains the classic demonstration of the second signature. After the Civil War, lending rates in the South and West stood well above those in the Northeast. By the early twentieth century these differentials had narrowed sharply as the commercial paper market and new intermediaries linked regional economies, a pattern John James later extended. Larry Neal applied similar reasoning to the eighteenth century. He showed that share prices in London and Amsterdam moved closely together, which points to an integrated international capital market.

The British long series, compiled by Sidney Homer and Richard Sylla among others, illustrates the first signature. Pelham's conversion between 1749 and 1757 consolidated much of the national debt into 3 percent stock. Goschen's conversion of 1888 cut the coupon to 2¾ percent, with a scheduled fall to 2½. Each step reflects a market that would accept lower compensation for holding British government paper, and the state captured that gain through refinancing.

A third indicator is volatility. Mature markets show lower variance and react less to war news and political shocks. A caution applies here. Falling levels can come from global capital supply, demography or peace dividends rather than from domestic institutions, so levels should be benchmarked against an external market. Convergence can also reverse. The interwar collapse of international lending, followed by widespread default in the 1930s, widened spreads that had been narrowing for decades.

Methodologically, historians now borrow tools from the growth and finance literatures. Sigma-convergence tracks the coefficient of variation in rates across markets over time. Beta-convergence tests whether high-rate markets fell faster than low-rate ones. Cointegration analysis asks whether paired price series share a long-run equilibrium. These price measures should be read alongside quantities such as outstanding debt, the number of instruments and the distribution of lenders, because rationing can hide in prices that appear stable.

Takeaway

Falling levels and shrinking spreads both mark financial development. The spreads are usually the more trustworthy guide to institutional change, because they net out the shocks that move every rate at once.

Interest rates are among the longest and most comparable series economic historians possess, and they do encode institutions. They do so only through differences, though, and only after inflation, liquidity and legal censoring have been accounted for. The Glorious Revolution debate shows what is at stake when those steps are skipped.

The weight of current evidence points to gradual change in Britain rather than a constitutional rupture. Sovereign credibility improved over decades, and the improvement depended on fiscal machinery: excise administration, the Bank of England and a liquid secondary market. Private property security followed its own separate trajectory.

Several research areas look promising. Notarial credit records, of the kind exploited by Hoffman, Postel-Vinay and Rosenthal for Paris, can recover private rates and quantities. High-frequency event studies can sharpen identification. Comparable series from Ottoman, Chinese and South Asian credit markets would test whether European findings generalise beyond the cases that built the theory.