A LITERATURE SURVEY

The present article is related to several previous studies on the changes in Britain's financial markets starting in the late seventeenth century and on the institutional changes of that period. Writing long before North and Weingast, John Ramsey McCulloch describes the determinants of interest rates in Britain as follows: “In the beginning of the funding system, the term fund meant that the taxes or funds (were) appropriated to the discharge of the principal and interest of loans …. Owing perhaps, to the scarcity of disposable capital, but far more owing to the supposed insecurity of the Revolutionary establishment (our own emphasis), the rate of interest paid by the government in the early part of the funding system was comparatively high. But, as the country became richer, and the confidence of the public in the stability of the government was increased, ministers were enabled to take measures for reducing the interest, first in 1716 and then in 1749.”¹²

Moving to more recent studies, one related line of research has focused on the evolution of private (nongovernment) interest rates in Britain. Gregory Clark, for example, examines private (real) rents on agricultural land.¹³

In addition to these attempts to provide evidence based on British interest rates, several studies cast some doubt on the relative importance of institutional changes in the protection of property rights, in comparison with other institutional changes in seventeenth- and eighteenth-century Britain. John Brewer, for example, attributes Britain's economic and military success to the emergence of a strong government, which he views as the most important transformation that took place at the time.¹⁷

Beyond Britain, the present study is also closely related to studies of the relation between the cost of capital, institutional changes, and political events more generally. Robert Barro and J. F. Wright document the relationship between long-term interest rates and wars for the eighteenth and nineteenth centuries.²⁴

METHODOLOGY AND DATA

Methodology

Our investigation of the evolution of Britain's cost of government debt during the century that followed the Glorious Revolution begins around 1690, and makes use of several newly constructed and reconstructed interest rate and government debt series, which are described in the next subsection. Methodologically, as noted in the introduction, a central feature of the present article is the treatment of Britain as an international borrower, whose cost of capital is calculated relative to that of the Province of Holland. This approach does not necessarily require that Holland be viewed as a completely risk free borrower; it is sufficient to acknowledge that Holland was an established borrower by the end of the seventeenth century, whereas Britain a relatively new one. This view is certainly consistent with descriptions of the Dutch Republic (sixteenth through eighteenth centuries) by Jan de Vries and Ad van der Woude as the “first modern economy,” whose society, economic efficiency, urbanization, and educated workforce were admired by contemporary and later observers.³⁰

de Vries and van der Woude, First Modern Economy.

The fact that Britain and Holland fought the same wars (on the same side) in the first half of the eighteenth century makes the comparison of the Dutch and British borrowing costs particularly appropriate.

To study the dynamics of the determination of interest rate on government debt we use a VAR (vector auto-regression) analysis describing the relation between interest rates in Britain and the Province of Holland in more detail:

In this specification, interest rates in London are determined by a weighted average of lagged interest rates in London and Amsterdam (denoted by i) and by a vector of additional explanatory variables (denoted by X).

Modern uncovered interest rate parity specifications relate the interest rate differentials to expected depreciation and to country risk. In practice, for nearly 90 years, between 1698 and 1785, the British pound-Dutch schilling (Banco) exchange rate was virtually constant, fluctuating slightly within a very narrow band of 34 to 37 schilling per pound (European State Finance Database). It is therefore reasonable to assume that the expected depreciation was zero, so that changes in relative interest rates should reflect primarily changes in the risk premium.

The control variables we use in the estimation of the determinants of the risk premium on British government debt are in line with the modern literature on emerging market risk premia.³¹

See Mauro, Sussman, and Yafeh, Emerging Markets.

These include two variables: the first is the trade surplus, which serves as a measure of the flow of revenues that can be used to repay (foreign) debt (therefore its coefficient is expected to be negative); exports may also be viewed as a proxy for a country's output (the concept of GDP did not exist at the time).³²

Exports in absolute figures or exports per capita can be used instead of the trade surplus without changing the empirical results.

The second is the fiscal deficit, which proxies for “leverage” or default risk, (therefore its coefficient is expected to be positive).³³

Another macroeconomic variable that we examine is inflation, which turns out to be insignificant, perhaps because this variable did not vary much during the period.

Data Issues

In support of their argument that there exists a relation between interest rates and institutional reforms, North and Weingast provide evidence on a decline in interest rates on British government debt following the Glorious Revolution.³⁴

North and Weingast, “Constitutions.”

This evidence is based on an incomplete interest rate time series, with observations for several years in the late seventeenth and early eighteenth centuries. They also offer evidence on an increase in the volume of British government debt around the same time period, supplemented by a general description of the development of the London Stock Exchange in the eighteenth century.³⁵

Part of the rise in the volume of trade on the London Stock Exchange during the period was related to the South Sea Bubble. There is no doubt that the bubble had ramifications that did not lead to financial development that could contribute to economic growth, and, in fact, it led to popular resentment that hindered the growth of British equity markets for over a century. In particular, the Bubble Act of 1720 limited the use of joint stock companies (Neal, Rise of Financial Capitalism, p. 62). For further details on the South Sea “Scheme,” see, for example, Roseveare, Financial Revolution, chapter 6. Recently, Carlos, Neal, and Wandschneider, “Origins,” suggest that a possibly positive effect of this episode was an improved microstructure of trading on the London Stock Exchange and greater liquidity of securities markets.

Data constraints have made the design of a more precise test of the determinants of the cost of government debt in the seventeenth and eighteenth centuries virtually impossible. High frequency time series of government borrowing rates or government bonds yields are available only starting in 1753, when British Consol prices become available for the first time (although annuity prices are available earlier). Attempts to bridge this gap have generally followed three different routes. The first is based on examinations of interest rates on private loans, and attempts to infer from them whether or not interest rates declined starting in the late seventeenth century.³⁶

See, for example, Clark, “Political Foundations”; Quinn, “Glorious Revolution's Effect”; and other studies mentioned in the literature review.

These studies essentially test the conjecture that the (assumed) reduction in the government's cost of capital led to a decline in the private cost of capital, but do not examine directly if, indeed, interest rates on government bonds experienced a significant change starting in the late seventeenth century. Moreover, this conjecture is based on the assumption that British capital markets were integrated, an assumption that contradicts Moshe Buchinsky and Ben Polak, who show that British capital markets were segmented until the late eighteenth century.³⁷

Buchinsky and Polak, “Emergence.”

The second route is based on substitute financial assets from which inference is made about government borrowing rates. Thus, Wells and Wills use the Bank of England shares as a proxy for government bonds, an approach which requires that the Bank of England stock be correlated with British government bonds.³⁸

Wells and Wills, “Revolution.”

This, however, need not be the case because, at the time, the Bank of England was a private lending institution that made out loans to the government. With the exception of a very severe threat to the existence of the London capital market, the Bank's profit could increase with the rate of interest charged to the government, so that Bank of England stock prices may well have been inversely related to government bond prices. In practice, for the period 1731 to 1753, we find that prices of 3 percent annuities and Bank of England stock prices are not co-integrated at all, and the R-squared of the first differences equation is quite low (0.18). During the Seven Years War (1756–1763), the correlation between the yield on government bonds and the total return on shares of the Bank of England is close to zero (–0.07).

The third approach is based on using the fragmentary available data on the cost of government debt at issue. This series (first compiled by P. G. M. Dickson using British Parliamentary Papers, 1890–1891 and 1898) has been used more recently by Clark and Stasavage.³⁹

P. G. M. Dickson, Financial Revolution; Clark, “Debt”; and Stasavage, Public Debt.

The advantage of this series is that it enables direct measurement of the cost of debt faced by the British government. The main disadvantage, however, is the incomplete nature of this series. Note also that an important feature of borrowing in the eighteenth century was that once a “permanent” debt level was established at a given interest rate, 3 percent, additional borrowing was made by issuing similar bonds with the same coupon. Variations in the risk premium were typically not reflected in new bonds carrying different interest rates, but rather in discounts on the purchasing price of new issues.⁴⁰

According to McCulloch, Dictionary, this form of borrowing became a feature of British debt management during the reign of George II (starting 1727). It continued into the twentieth century.

Therefore, the fact that the coupon on British securities declined, for example, with the introduction of 3 percent coal duties in 1731, does not necessarily imply that from then on the British government borrowed at these rates.⁴¹

See North and Weingast “Constitutions and Commitment,” pp. 823-824. The discussion on p. 823 suggests that with the issue of 3 percent bonds, Britain achieved a permanent decline in government borrowing rates.

In order to draw firm conclusions on the actual cost of debt, primary yields should be adjusted to reflect the discount at which the bonds were sold.

Rather than follow one of the existing routes in the literature, we construct a novel fiscal measure of Britain's cost of capital, the ratio of debt service payments to total government debt. This variable is constructed by dividing total government debt service expenditures (from Brian Mitchell) by a newly constructed total debt series, which includes both long-term and short-term debt obligations.⁴²

Mitchell, British Historical Statistics.

The long-term debt series is based on the British Parliamentary Papers, where all issues of long-term debt are listed, and the evolution of the stock of debt is tracked by recording redemptions and refinancing of older debt (this is one of the primary sources used by Dickson).⁴³

British Parliamentary Papers, History. Because Mitchell, British Historical Statistics, provides data for fiscal years whereas the information in the British Parliamentary Papers is listed per calendar year, there may be minor dating discrepancies. See a comment on this issue in Dickson, Financial Revolution, p. 525.

The calculation of short-term debt is more involved; short-term notes were issued by the Exchequer (“Tallies” or IOUs) and by the Army and Navy, and were not fully accounted for in the British Parliamentary Papers.

⁴⁴

British Parliamentary Papers, History.

Mitchell does not provide data on short-term debt, and Dickson offers only a rough estimate of short-term debt in the late seventeenth century, which he himself describes as “orders of magnitude rather than … strict accountancy.”⁴⁵

Mitchell, British Historical Statistics; and Dickson, Financial Revolution, p. 525.

We therefore impute the short-term debt for the period 1694 to 1703 as follows: Denote short-term borrowing by B^S; long-term borrowing by B^L; the stock of short-term debt by D^S and the government deficit by spending, G, minus taxes T (G-T). Time subscripts are denoted by t. Using the government's budget constraint, short-term borrowing is defined as:B_t^S ≡ (G_t - T_t) - B_t^L. The stock of government short-term debt is therefore the accumulated sum of short-term borrowing (redemptions appear as negative borrowing):

. Starting in 1704, we use the short-term debt series provided by Dickson which includes the Exchequer “Tallies” and IOUs (called “Fictitious Loans”).

⁴⁶

Dickson, Financial Revolution.

The plausibility of the imputed short-term debt series for the pre-1704 period is verified as follows: We extend the imputed short-term debt series to 1711 and compare it with Dickson's short-term debt series for the period 1704–1711, when Dickson's series is complete, and prior to the refinancing of government short-term debt through the South Sea Company; the two series are very similar (see Appendix Table 1).⁴⁷

Dickson, Financial Revolution.

We complement this fiscal measure of Britain's cost of debt by constructing, using the British Parliamentary Papers, a yield at issue series of actual interest rates on newly issued debt, which is conceptually similar to that of Dickson but takes into account the discount at which bonds may have been sold.⁴⁸

British Parliamentary Papers, History; and Dickson, Financial Revolution.

In order to overcome the discontinuity of this series, in the 21 years with no borrowing we impute the preceding yield at issue.

Finally, to verify the robustness of the results we report, we calculate yet another proxy for the government's cost of debt—the coupon interest rate on long-term bonds, which includes both “funded” and “unfunded” long-term debt. (The term “unfunded” means that no specific tax revenues were earmarked for repaying this debt).

All of the measures we use have shortcomings: The fiscal proxy for the cost of debt, the ratio of debt service to total debt, measures the average (not marginal) cost of debt. In the first years following the Glorious Revolution this average is highly influenced by the short duration of the debt, and therefore the series may exhibit more volatility than in subsequent years when marginal debt issues affect the average much less. On the other hand, this volatility conveys important information to the extent that it reflects the inability of the government to borrow at reasonable long-term rates in those years. Another possible shortcoming is that the debt service series for the initial years may also include debt redemptions, causing an upward bias in the series. In practice, there were no substantial redemptions before the second decade of the eighteenth century, so this is unlikely to constitute a significant problem.

The yield at issue series captures the marginal cost of long-term borrowing faced by the government, but because the government did not borrow in every year, this series has missing observations for 21 years. In addition, the series does not take into account the cost of short-term debt.

The interest rate on long-term debt also ignores the substantial short-term debt in the beginning of the sample period; after 1712, when much of the short-term debt was refinanced, the series becomes very similar to the fiscal series.

The three series, together with annuity prices between 1731 and 1753, followed by daily market Consol yields are presented in Figure 1a.⁴⁹

Annuity prices are from Sinclair, History, appendix 2, pp. 28–33. Consol yields are from European State Finance Database. In addition to the shortcomings listed above, none of the measures takes into account the fact that lenders to the British government in the early years were often the Bank of England or the East India Company, which were granted monopolies, partly in exchange for providing loans. Ignoring this form of repayment may lead to an under-estimate of the cost of British government debt; see Broz and Grossman, “Paying for Privilege.”

This latter series, and the yield at issue series, exhibit more volatility, which is somewhat smoothed in the fiscal series and the yield on long-term debt. However, for the period we are mainly concerned with—the end of the seventeenth century and the early decades of the eighteenth century—the fiscal series exhibits considerable volatility. In Appendix Figure 1 we verify that the picture that emerges from Figure 1a is consistent with available fragmentary data on monthly market yields on a variety of government-issued securities in the early eighteenth century.⁵⁰

From Castaing, Course.

All yields begin the century at high levels, increase around 1710 and then steadily decline starting around 1712, the period when short-term government debt was first refinanced through the South Sea Company.

We now turn to variables other than measures of the British government's cost of debt which are used in this study. Data on Britain's population, total government expenditure, and exports are all drawn from Mitchell.⁵¹

Mitchell, British Historical Statistics.

Data on borrowing rates of the Province of Holland, the largest and wealthiest in the Netherlands, are not readily available.⁵²

Although precise data are not available, in the eighteenth century the Province of Holland accounted for about 60–70 percent of total Dutch debt, and its population, around 800,000, constituted about 40 percent of the total population in the Netherlands (based on private communication with Wantje Fritschy).

We construct a fiscal measure of the Dutch cost of debt similar to that of Britain, using the ratio of debt service to debt from the European State Finance Database. Throughout the century, the Dutch cost of debt is very stable at a level slightly above 4 percent (somewhat higher during the War of the Spanish Succession, see Appendix Figure 2), making it a convenient benchmark for comparison with British borrowing rates.⁵³

In comparing British to Dutch interest rates, differences in monetary standards between the two countries may be an issue. The use of a fiscal proxy for interest rates (nominal interest payments divided by nominal debt) mitigates this concern, because interest rates are not expressed in terms of a currency unit. Nevertheless, the Dutch Gulden, which was based on a silver standard, was stable from 1681—each Gulden contained 9.61 grams of silver. Following the Glorious Revolution, the British currency, which at the time, was also based on a silver standard, was clipped and large quantities of silver were exported. In 1696 the British currency was stabilized, and in 1717 the value of the gold guinea in terms of silver was reduced by Newton from 21.5 to 20 shillings, thus putting Britain on a de-facto gold standard.

We also use the European State Finance Database to construct similar fiscal measures of the cost of debt for Denmark. Data on the population of Holland are from de Vries and van der Woude.⁵⁴

de Vries and van der Woude, First Modern Economy.

EMPIRICAL ANALYSIS AND RESULTS

Before turning to a detailed discussion of Britain's cost of capital, we briefly discuss the liquidity of government debt using The Course of the Exchange. Surprisingly, although this source provides continuous quotes on equity prices and exchange rates, it rarely quotes prices or yields of long-term government bonds. Table 1 shows the fraction of long-term government bonds quoted relative to the total number of outstanding bonds in public hands: the market for these bonds was extremely illiquid, almost nonexistent.⁵⁵

WARS, REFORMS, AND BRITAIN'S COST OF CAPITAL IN THE EIGHTEENTH CENTURY

Figure 1b displays the interest rate differential (“spread” in modern parlance) between British government debt and debt issued by the Province of Holland using the fiscal cost of capital series. Early in the eighteenth century, when Britain was involved in the War of the Spanish Succession, the spread was very high and volatile. It began to decline in 1709, and especially in 1711–1712 when short-term debt was converted into capital of the South Sea Company, but the spread remained positive until around 1730.⁵⁷

We now turn to a more detailed discussion of the decline in British interest rates (and spreads) between 1711 and 1730. As noted, part of the decline during this period coincided with the use of the South Sea Company as a mechanism to refinance the expensive government debt issued following the Glorious Revolution (conversion to cheaper government bonds was apparently impossible). The first debt refinancing in 1711—when creditors were to be incorporated as the South Sea Company—involved 9.2 million pounds of short-term debt in arrears.⁵⁸

FIGURE 2 THE COMPOSITION OF BRITISH GOVERNMENT DEBT, 1691–1750

The second refinancing scheme of 1719 involved about 22 million pounds of debt; as a result, annual interest payments were reduced by 415,000 pounds.⁵⁹

In sum, the institutional changes of the seventeenth century clearly did not suffice to guarantee a permanently low cost of capital for the British government. A comparison with the pre–Glorious Revolution borrowing rates would have been interesting, but it is made difficult by the fragmentary nature of the available data. Nevertheless, Roseveare cites interest rates of about 7 or 8 percent for the 1670s, and in some cases interest rates of 6 percent were observed as well, figures which are lower than those for the early decades of the eighteenth century (Figure 1a).⁶⁰

In order to shed further light on Britain's financial development, Figure 3 portrays government debt per capita in pounds for Britain and the Province of Holland during the eighteenth century (see the European State Finance Database for the exchange rate). It is evident that the government of Holland could borrow relatively more than her British counterpart without paying higher interest rates: despite the large increase in British government borrowing, as late as 1749, on the eve of issuing the first Consols, debt per capita in Holland was almost three times as large as in Britain.⁶¹

FIGURE 3 GOVERNMENT DEBT PER CAPITA: BRITAIN VS. THE PROVINCE OF HOLLAND, 1698–1790 (pounds Sterling)

FIGURE 4 GOVERNMENT DEBT SERVICE TO EXPENDITURES: BRITAIN AND THE PROVINCE OF HOLLAND, 1692–1795

Moving to a more formal treatment of the relation between interest rates in Britain and Holland, Table 2 displays the results of the VAR analysis, which suggest that the cost of debt of the government of Holland affected the cost of debt of the British government. During the first half of the eighteenth century the effect of interest rates in Holland on the cost of British debt was particularly pronounced. By contrast, Dutch interest rates prior to 1750 were not very sensitive to British rates.⁶³

Taken together, the empirical evidence we present suggests that the fruits of the institutional reforms of the seventeenth century did not affect the London financial market immediately; over half a century was needed for these changes to be reflected in the market. This conclusion is consistent with that of Nicholas Crafts who describes the early decades of the eighteenth century as a period in which “there were institutional weaknesses related to banking, finance and company legislation which must have had some inhibiting effects both on savers and on business investment.”⁶⁵

ADDITIONAL EVIDENCE FROM AROUND THE WORLD

Moving from eighteenth-century Britain to Continental Europe, we briefly discuss the fragmentary available evidence on the cost of borrowing of several other European countries, see Figure 5. Data on Austrian external borrowing rates (in Amsterdam and elsewhere), drawn from Dickson and the European State Finance Database, suggest that this absolute monarchy borrowed at rates that were similar to those of Holland and far lower than the British cost of debt early in the eighteenth century.⁶⁷

FIGURE 5 GOVERNMENT BORROWING RATES IN EUROPE, 1700–1780

We now turn to the nineteenth-century experience of two belligerent regimes, Tsarist Russia and the Ottoman Empire. For these regimes, that did not advocate the rule of law, the absence of secure property rights was not much of a constraint when it came to foreign borrowing—both countries were among the largest borrowers in the 1870–1914 period.⁷⁵

Moving away from Europe, the conclusion that institutional changes in Britain were not immediately translated into a lower cost of government debt is in line with our earlier findings.⁷⁶