THE ONSET OF CONVERTIBILITY SUSPENSION AND THE BULLION REPORT

The French Wars, 22 years of almost uninterrupted warfare between 1793 and 1815, encompassed Europe's political reshuffling following the French Revolution and Britain's struggle against Revolutionary and later on Napoleonic France. France's drive to export revolutionary ideals soon drove the country to war with most of Europe. The execution of King Louis XVI in January of 1793 sealed the first coalition against France.Footnote ⁶ It would take a total of seven differing coalitions between the Continental European powers and Britain to finally defeat Napoleon at Waterloo on 18 June 1815.

The wars exerted extreme political and financial pressures on all belligerents. The Reign of Terror in France, beginning in early 1793, produced a sharp outflow of capital, with silver and gold leaving France for Britain and providing increased liquidity for the British banking system (Neal Reference Neal1991; Sargent and Velde Reference Sargent and Velde1995). When the assignats collapsed in France in 1795, money for ordinary payments was in short supply and those with claims or credit in Britain drew on them to fill the gap. The ensuing return flow of capital to France, by lowering the stock of precious metal, put deflationary pressure on Britain (Kindleberger Reference Kindleberger1984).

Against this backdrop, French military actions in February 1797 triggered bank runs in Britain with depositors hoping to convert notes into gold coin. The consequences were felt throughout the country, putting numerous (country) banks out of business (Feavearyear Reference Feavearyear1963). Eventually, the run reached the ultimate provider of specie and, thus, de facto center of the British payment system, the Bank of England. With gold reserves threatened, the Bank was given permission to cease payment of its notes in gold on the 27^th of February 1797 and suspension ensued.

With suspension, local bank notes were now convertible only into Bank of England notes, shifting Britain's monetary system from a commodity standard to a fiat currency with flexible exchange rates. Once the constraints imposed by the gold standard were abandoned, it became possible to accommodate the external drain of specie caused by Britain's expenditures on the Continent, and to grant the means of payment necessary for internal transactions, while absorbing increasing public debt issues (Bordo and White Reference Bordo and White1991). Britain now had a payment system in which the money supply was controlled by the Bank and the Treasury.

Conceived as an emergency measure, the suspension was regularized in November 1797 and extended several times. In December 1803, the Bank Restriction Act finally fixed its end at six months after the ratification of a definitive peace treaty. Over the 22 years of the French Wars, resumption of the gold standard was often a remote and uncertain prospect. Peace treaties proved to be inconclusive such as those of Amiens (in April 1802) and Paris (in June 1814). Even after the peace treaty of 1815, resumption was postponed several more times.

The years following 1797 saw the Second, Third, and Fourth Coalition Wars against France. France prevailed in military terms. Yet, Napoleon's attempt to undermine British trade by initiating the Continental Blockade (21 November 1806) remained unsuccessful especially given the defeat of the French fleet at Trafalgar (October 1805). This British victory also averted the danger of a French invasion. Hostilities intensified further when Napoleon initiated the Peninsular War to punish the Portuguese for not enacting the Continental System against the British. French and Spanish troops invaded Portugal in November 1807. Napoleon turned on his former ally by attacking Spain in February 1808. The year 1809 proved disastrous in military terms for the British, on all war fronts. Over the first quarter of the year, the French occupied northern Portugal and forced the British to evacuate Spain. In July, the Fifth Coalition of Austrian- and British-led forces was defeated at the battle of Wagram. By the end of the year, the French beat the remaining forces in the south of Spain; the area being taken in the following winter, leading to French control in most of the Iberian Peninsula. By 1810, French victory seemed possible.

During these years, the paper pound's exchange rates into gold and foreign currencies started depreciating. Figure 2 depicts both the agio and the pound's exchange rate on Hamburg. Before the suspension of the gold standard (shaded area) and even during the first years of the latter, both metrics fluctuated in relatively narrow bands: de- or appreciations of the paper pound hardly ever exceeded 10 percent. Beginning in 1809, the agio and the exchange rate experienced marked declines, peaking at 50 and 30 percent, respectively. Ex-post constructed price indices, shown in Figure 3, exhibited the same pattern. Both series increased substantially with the suspension and particularly after 1809 and declined—albeit more gradually—after the resumption of the gold standard in 1821.

Figure 2

AGIO, 1750–1850; EXCHANGE RATE HAMBURG: SCHILLING PER POUND STERLING, TWO MONTHS' USANCE, 1750–1825. FOR BOTH INDICES JAN 1797=100

Notes: The shaded area marks the suspension of the gold standard from 1797 to May 1821. Sources: Neal 1990; Boyer-Xambeu et al. 1994.

Figure 3

PRICE INDICES, 1800=100, 1750–1850

Notes: The shaded area marks the suspension of the gold standard from 1797 to May 1821. Sources: Broadberry et al. 2012; Clark 2014.

These changes in the agio and in prices prompted the government to appoint an investigating committee, which published its work in 1810 as the Bullion Report. According to the report, rising prices and falling exchange rates had a common source in the Bank of England issuing too many notes.Footnote ⁷ Notes were issued to buy the government debt, which in turn was used to finance the Napoleonic Wars. This would have been impossible had the Bank of England retained its obligation to convert its liabilities into bullion.

The Bank of England had indeed expanded its note issue, but the peak was reached only when the bursting of the South American bubble in late 1810 caused widespread panic and business failures (Feavearyear Reference Feavearyear1963; O'Brien Reference O'Brien2010). Amid spreading commercial distress, country banks started to contract their note issue drastically. The liquidity shortage was broadly acknowledged and the Treasury attempted to overcome it by issuing Exchequer Bills. Thus, at least part of its note issue could be accounted for by its normal function to sustain the domestic payment system, when other forms of credit were disappearing (O'Brien Reference O'Brien2010; Clapham Reference Clapham1944; Flinn Reference Flinn1961).

The marked devaluation of the pound after 1809 coincided rather with the military misfortunes of the British than with the evolutions in the Bank's note issue. Moreover, with the intensification of hostilities around 1809 a turning point would also be attained in terms of public finance.

FISCAL PRESSURE AND INTERACTIONS BETWEEN THE BANK AND THE TREASURY

Institutional characteristics enabled Britain to mobilize war finance at an impressive scale and laid the basis for her military and commercial supremacy over most of Continental Europe (Brewer Reference Brewer1988). Parliament gained broad powers over public expenditures and taxation through the Glorious Revolution; loans to government and the taxes earmarked to service them necessitated parliamentary assent (Sussman and Yafeh Reference Sussman and Yafeh2006). Furthermore, the government created a separate sinking fund to cover the planned reimbursements in case revenues earmarked for specific loans proved insufficient (Cox Reference Cox2015). Partisan politics further rendered outright default politically unacceptable (Stasavage Reference Stasavage2007).

The Bank of England acted as an additional private constraint on the government. It was responsible for assuring the continuity of promised reimbursements. As such, it could have refused to grant new loans whenever arrears on existing ones occurred. Channeling all public credit through the Bank therefore implied that it would become difficult to utilize funds of current loans whenever previous obligations were not honored (Macaulay Reference Macaulay1914; North and Weingast Reference North and Weingast1989). Thus, by the time the French Wars had to be financed, contemporaries dismissed the risk of strategic default, making large scale debt accumulation possible (Acworth Reference Acworth1925; Neal Reference Neal1990; O'Brien Reference O'Brien2010).

Nonetheless, by 1809, prolonged and intensified warfare had taken toll on public finances. Following the usual practice, expenditures during the first years of the war were paid for by issuing long-term government debt (Barro Reference Barro1987). As a result the national debt had doubled during the first six years of unsuccessful warfare against Revolutionary France and large loans against bonds (funded debt) continued to be issued in every year of the war.Footnote ⁸ Only in 1809, considerable loans were floated for England, Ireland, and Portugal (Grellier Reference Grellier1812).

As much as debt finance was an essential part of war funding, it was backed by a strong increase in taxation, with tax receipts accounting for 58 percent of the extra revenues necessary to fund the wars (O'Brien Reference O'Brien1988).Footnote ⁹ Yet, by 1809, the government had exhausted its possibilities for taxation. The introduction of the income tax in 1799 had only been possible given the highly exceptional circumstances. By February 1799, the price of consols had declined by 26 percent, to the lowest levels of the French Wars, reflecting grave doubts about the stability of public securities. The strength of opposition to the so-called “repugnant tax” was visible in its instant revocation upon the Peace of Amiens in 1802 and in 1816, shortly after the victory at Waterloo (O'Brien Reference O'Brien1988).

For excise taxes, a broad social consensus argued that it was indecent to impose duties on the necessities or the poor; so mostly goods consumed by the affluent were further taxed. Between 1793 and 1815, two-thirds of the extra taxation required to finance the French Wars came from taxes levied on the incomes and consumption patterns of the wealthy andFootnote ¹⁰ further increases in tax rates had become self-defeating (O'Brien Reference O'Brien1988, Reference O'Brien2007).

As a result, the Treasury had to rely heavily on the Bank for war finance. Short-term borrowing against Exchequer or Navy bills (unfunded debt) increased by 22 percent in 1809 (Mitchell Reference Mitchell1988). Unfunded debt had the great advantage of not necessitating parliamentary assent and of not coming with earmarked taxes.Footnote ¹¹ It was, thus, often used to finance unforeseen expenditures (O'Brien Reference O'Brien2007). As both manager and holder of public debt, the Bank of England became essential for financing the war effort.

The Bank could hardly refuse requests for credit from the Treasury in times of war, especially since the monopoly of issue it enjoyed in the wider London area was subject to periodic renewal. With the Bank charter renegotiable at will, bargaining over renewals were often reached along with war loans, as was the case in 1781 and 1800 in the midst of the American and French Wars (Broz and Grossman Reference Broz and Grossman2004).

The Bank's interactions with the Treasury came with certain risks. During the Restriction Period, more than half of the Bank's assets were held in public securities; in the aftermath of the war, this rose to almost 70 percent. Holding substantial amounts of sovereign short-term debt on its balance sheet, the Bank's financial solidity depended on the Treasury's will to honor its running obligations.

War finance was, nonetheless, a very lucrative business for the Bank and its shareholders. Figure 4 displays the evolution of the Bank's stock price over the 25 years from January 1796 to December 1821. The monthly price index is set to 100 in January 1797, the month before the suspension, and deflated by the pound's value in gold. The stock price almost tripled between the suspension of convertibility (solid left line) and its peak in November 1817. Especially after the beginning of the Third Coalition War against France in May 1803 (dashed line) the stock price increased markedly.Footnote ¹²

Figure 4

STOCK PRICE BANK OF ENGLAND, DEFLATED BY AGIO, JANUARY 1797=100, JANUARY 1796 TO DECEMBER 1821

Notes: The solid right and left lines mark the suspension and the decision to end the latter in February 1797 and May 1819, respectively. The dashed line depicts the beginning of renewed hostilities between Britain and France in May 1803. Sources: Neal 1990.

The Bank's stock price reflected changes in its expected cash flows to share-holders. As the latter hinged largely on the management of public credit, times of increasing debt—such as war times—were profitable for the Bank: beyond the interest the Treasury paid on the majority of loans with the Bank, the latter also charged management fees for handling the public debt. Billing £450 per million of public debt, the Bank's income increased automatically when war debts soared.Footnote ¹³

In June 1815 with the news of the victory at Waterloo, the Bank's income and stock price decreased; this was also the case with the decision to resume the gold standard in May 1819 (solid right line in Figure 4). Yet, these did not fall back to levels usual before the suspension of payments. Despite revised management fees and falling discounts, the sheer size of the public debt assured a comfortable level of revenue to the Bank. Parliament's refusal to maintain the income tax certainly helped that cause as well, since it implied that borrowing continued when war-related expenses ceased.Footnote ¹⁴

THE LONG PATH TO RESUMPTION

Having won the war against France, fiscal pressure delayed resumption of the gold standard. The Bank Restriction Act had initially determined that specie payments would be resumed six months after the end of the war. But between the end of the Napoleonic Wars in June 1815 and the resumption of the gold standard, the legal limit was extended several times, carrying the restriction to the 5th of July 1819 with resumption occurring slowly and only fully in 1821.Footnote ¹⁵

In the immediate post-war period, Bank directors insisted that an effective resumption was only possible if the government reimbursed a sub-stantial portion of debt to due to the Bank.Footnote ¹⁶ The argument was acknowledged in the parliamentary report preparing the resumption of the gold standard: “The effect […] of the extent of the advances to Government upon the situation of the Bank, when preparing for a resumption of cash payments, is evidently to cramp its operations, by placing a large proportion of its issues beyond its control.”Footnote ¹⁷ Along with the definite resumption of specie payments, Peel's Act, enacted 2 July 1819, a separate act imposed the repayment of £ 10 million of government short-term debt and forbade the Bank to lend to government for more than three months without express parliamentary approval.Footnote ¹⁸

These legislative actions determined the modalities of resumption. The suspension of cash payments was to be maintained until 31 January 1820. Following that date, the Bank would gradually resume payments, by exchanging its notes against ever higher amounts of specie between 1 February 1820 and 1 May 1823.Footnote ¹⁹ Discretion was left to the Bank of England to accelerate resumption, granted that any increase in the pound's gold content had to be irreversible. At the same time, the Bank of England was to be repaid gradually the £ 10 million of government short-term debt.

Cash payments were finally resumed at the mint rate on 1 May 1821. It appears that the plan for gradual resumption and debt reimbursement was credible in the eyes of contemporaries. Upon the plan's announcement in May 1819, bank stocks and gold reacted immediately, suffering sizable declines in value; and the pound's exchange rate on Paris appreciated. The London Times lead article of 27 May asserted that resumption would ultimately succeed. Yet, neither cash payments nor effective debt reimbursement had begun: as late as January 1820, the Bank was still pressing the Treasury for repayment (Clapham Reference Clapham1944). The change in market sentiment and its impact on asset prices hinged on the mere anticipation of an important debt reduction that came with the reversion to the old monetary standard.

ECONOMETRIC METHODOLOGY

The Data

The agio, the difference between the pound's exchange rates on the spot and forward markets, captures how credible the promise of resuming the gold standard at its pre-war parity was in the eyes of contemporaries. The public, and the Bullion Committee in particular, used the agio to form an idea regarding the fluctuations in the pound's internal value and to assess inflationary tensions more broadly. Classical value theory held that the equilibrium price of a good was determined by its production costs (Ricardo Reference Ricardo and Sraffa1817). As the production prices of gold and silver were relatively stable over the period, a rise in the sterling price of gold bullion, and hence, an increase in the difference between the mint price and the market price of gold, was perceived as evidence of inflation.

Evolution in the agio reflected the internal conditions of the British gold market well. The external supply of gold was stable during the Restriction Period because there were no major gold or silver discoveries (Barro Reference Barro1987). Melting down and export of British coin were prohibited; the Continental Blockade and British counter measures made smuggling more costly. Finally, there is no reason to assume, that the demand for non-monetary uses of gold changed during the Restriction Period. Thus, price increases for precious metals in times of financial or political turmoil most likely reflected precautionary monetary motives, that is, a hedge against future inflation.

The data set spans the years 1750–1850 and is based on Castaing's and—later on—Wettenhall's Course of the Exchange, published twice a week. As noted earlier, the agio, shown in Figure 1, was essentially stable and close to zero over the period,Footnote ²⁰ except for the Restriction Period, where it rose above 10 percent after 1808 and reached a peak of 45 percent mid-1813.Footnote ²¹ With no significant changes in the supply of gold, these evolutions meant a sizable internal devaluation of the pound: an increase of gold's market value in terms of paper pounds meant that more pounds were required to purchase the same amount of gold.Footnote ²² Table 1 provides descriptive statistics for the agio with the agio being on average substantially higher and more volatile during the suspension than for the rest of the sample.

Table 1

DESCRIPTIVE STATISTICS FOR THE AGIO (MONTHLY DATA)

Sources:

Author's calculations, data see text.

The agio contains information on inflationary tensions. For the available monthly and yearly price indices, correlation coefficients, shown in Table 2, ranged from 0.53 to 0.72. It is noteworthy, that the correlation between the agio and the different price indices is higher for domestic goods and increases substantially for the Restriction Period, ranging from 0.65 to 0.79. This is indicative of the fact that the agio reflected the internal value of the paper pound. An increase in the agio meant a decline in the purchasing power of the paper pound: aggregate prices and the agio varied in the same direction at a time when Britain's payment system rested on the paper pound.

Table 2

CORRELATION COEFFICIENTS BETWEEN THE AGIO AND PRICE INDICES

Sources:

Author's calculations, data see text.

Archival Evidence: The Press and the Bank

This approach does, however, bear the risk of “over-interpretation,” that is, the risk that historical events of minor importance are matched with the determined break dates at any cost. I verify the importance of events matched with detected break dates by considering contemporary sources and archival evidence. In particular, I consult the London Times, for which archives are available over the whole period (Bignon and Flandreau Reference Bignon and Flandreau2014). I also examine the London Gazette, the government's official journal. This was published twice a week (on Tuesdays and Saturdays) and as a Gazette Extraordinary whenever deemed necessary by the nature of the news. The London Gazette contained official government announcements, such as declarations of war and peace, and intelligence from different war fronts. The news came in the form of dispatches, composed by the actors of war themselves (to a lesser extent that was the case for the London Times as well).

These sources also conveyed information regarding transmission times for news. For instance, Madrid capitulated to the French on 4 December 1808. News of the capitulation, however, only reached London on 19 December. I, consequently, pay close attention to the fact that break dates have to coincide with the moment at which information concerning an event reached London, rather than with the timing of the event itself.

While it is rather straightforward that Madrid's capitulation made the news in December 1808, it is less easy to determine which event affected contemporaries' perceptions, when there were several important battle lines. In these cases, I search the Times and the Gazette for keywords in the vicinity of the break date and retain the event yielding the highest number of references. I pay particular attention to the frequency of mentions in the London Times lead articles. In August 1812, for example, intelligence from neither the Russian nor the Spanish fronts made the headlines; rather the American declaration of war occupied most news space.

In addition to these sources, I rely on ledgers and related documents from the Bank of England Archives that contain prices for precious metals, exchange rates, and different titles of public debt.Footnote ²⁴ The ledgers recorded events significant to the Bank, establishing a link between observed price movements and the incidents that caused them. In other words, contemporaries were well aware of the fact that political and military events had a bearing on the evolution of asset prices. The ledger (10A2702) recorded fortnightly averages of the aforementioned price data and was compiled in real time for the years 1705 to 1880. This makes the ledger a reflection of how events were understood as they unfolded. It is, therefore, reassuring that practically all estimated break dates and related events, discussed in the next section, are mentioned in the ledger.

My approach is inspired by Steven B. Webb (Reference Webb1986) in that it allows detection of events that influenced contemporaries' perceptions in real time, by combining ex-ante agnostic break tests with the study of contemporary sources. The Online Appendix provides an example of how this methodology forestalls detecting events that suffer from ex post bias. For example, historiography regards the Battle of Leipzig in October 1813 as Napoleon's decisive defeat, yet contemporaries' perception of the event explains why it did not affect asset prices; the British victory was anticipated and was not expected to alleviate fiscal strain.

Various studies in economic history (Guinnane, Rosen, and Willard Reference Guinnane, Rosen and Willard1996; Brown and Burdekin Reference Brown and Burdekin2000; Flandreau and Oosterlinck Reference Flandreau and Oosterlinck2012) have relied on break point tests. The procedure used here to estimate break dates is based on Jushan Bai and Pierre Perron (Reference Bai and Perron1998, 2003). It has the advantage of allowing for a very wide range of specifications (serial correlation and heteroskedasticity in the errors, lagged dependent variables, trending regressors, as well as different distributions for the errors and the regressors across different segments). The procedure also encompasses different break point tests (sequential and global methods), increasing the robustness of the results.

Consider the following multiple linear regression model with m breaks and m+1 regimes:

(1)

for t = T_j–1 + 1, …, T_j and j=1,…,m+1. y_t is the observed endogenous variable, x_t (p × 1) and z_t (q × 1) are vectors of co-variates and β and δ_j are the corresponding vectors of coefficients. u_t is the disturbance at time t. Given the T observations of (y_t, x_t, z_t ), I estimate the unknown regression coefficients β and δ_j and break dates. For the agio, I will posit that all coefficients are subject to change, obtaining a pure structural change model (p=0) of the following form:

(2)

Note that the variance of u_t does not need to be constant. Breaks in the variance are permitted, provided they occur at the same dates as the breaks in the parameters of the regression.

I define a break in the series as a change in the conditional mean of the series y_t , that is, I specify that the intercept can change (z_t = 1). The focus here is on abrupt structural changes in the mean that reflect the advent of unexpected news affecting Britain's war fate and, hence, public finances. For that reason, I exclude past values of the endogenous y_t . When lagged variables are included, changes in the level of y_t also depend on the auto-regressive dynamics of the series; thus, the change takes effect gradually. Conversely, when specifying x_t = Ø, all the dynamics are contained in the error term and the change is abrupt (Bai and Perron 2003).Footnote ²⁵

I first run the break point procedure on monthly data for the 1795–1823 period. This data set is comparable to the ones used in earlier studies (Neal Reference Neal1990) and contains 340 observations. I then use biweekly data, in order to more precisely detect break dates. Due to data availability issues, I estimate biweekly break dates only for July 1810 to May 1823. The biweekly data set encompasses 1,350 observations.

Increasing the number of observations included in the sample allows detecting more break dates, as the number of breaks depends on the size of segments (the number of observations between two adjacent breaks). A minimal segment size guarantees that each segment contains sufficient observations to estimate parameters precisely. Furthermore, I ignore break dates indicated as imprecisely estimated by a large confidence interval.Footnote ²⁶

In the presence of autocorrelation and heteroskedasticity, as is the case here, each segment should include at least 24 observations. A trimming parameter, ε, links the minimal segment size, h, to the maximal number of breaks, k, allowed: ε = h/T. For a sample of T observations one obtains the maximal number of breaks, k: ε = 0.05 → k = 10; ε = 0.10 → k = 8; ε = 0.15 → k = 5; ε = 0.20 → k = 3; ε = 0.25 → k = 2.