Printer Friendly

The U.S. Embargo Act of 1807: its impact on New England money, banking, and economic activity.



The period 1793 through 1807 was one of unusual prosperity in the United States, paced by the growth of American carrying and re-export trade. (1) The Embargo Act of 1807, however, supposedly halted this period of growth and prosperity. The Act was an attempt by a new nation to persuade Great Britain and France to refrain from seizing neutral American ships by prohibiting the trading of certain goods between the United States and other countries. The provisions of the Act forbade U.S. ships to sail with cargoes to foreign ports. The Act was in effect from December 1807, through March 1809.

Historians generally assert that the embargo had at least three important consequences of the U.S. economy. The embargo allegedly initiated a severe economic depression, led to a signficant increase in the level of smuggling between the United States and other countries, and was instrumental in bringing about an expansion of domestic manufacturing that started the United States on a path to self-sufficiency.

With respect to the depression effects of the embargo, McMaster [1903, 415] described the economic losses of the period as follows:

The newspapers were full of insolvent-debtor notices. All over the country, the court-house doors, the tavern doors, the post-offices, the cross-road posts, were covered with advertisements of sheriffs sales. In the cities, the jails were not large enough to hold the debtors. At New York during 1809, thirteen hundred men were imprisoned for no other crime than being ruined by the embargo. A traveler who saw the city in this day of distress assures us that it looked like a town ravaged by pestilence. The counting-houses were shut or advertised to let. The coffee-houses were almost empty. The streets along the water-side were almost deserted. The ships were dismantled; their decks were cleared, their hatches were battened down.

Illegal trade arose as people searched for other sources of income. Malone and Rauch [1960, 118] described the situation in the following manner:

The ancient art of smuggling was revived, particularly along the land frontier with Canada where drovers of wagon trains shot down federal agents.

And concerning the stimulative effect of the embargo upon the rise in domestic manufacturing, North [1961, 56] stated:

The closing off of the import trade was effective in promoting the rise of domestic manufactures, and capital which had been devoted to shipping and foreign commerce was partially absorbed in a rapid growth of industry. Before 1808 only fifteen cotton mills had been built in the United States: one in 1791, one in 1795, two more in 1803 and 1808. By the end of 1809 eighty-seven additional mills had been constructed, and capacity had been increased from eight thousand spindles in 1808 to thirty-one thousand at the end of 1809 and an estimated eighty thousand by 1811. Other branches of manufacturing also made substantial gains through 1814.

Similar comments concerning the embargo are abundant in most economic history texts and in numerous articles devoted to the subject. While dramatic, such discussion is based on casual empiricism at best, and little in the way of rigorous economic analysis has ever been applied to discern the actual effects of the embargo upon the U.S. economy.

As usual with such historically remote events, a paucity of data on such variables as real income, output, and employment constitutes the major obstacle to empirical analysis. Until recently, economic historians have attempted to make inferences about the effects of the embargo upon the economy by relying primarily upon data reflecting the volume of import, export, and re-export trade, domestic and imported price indices, and governmental duties. (2) In a recent article, for example, Frankel [1982] argues that an examination of British shipping statistics rather than those of the U.S. indicates that 1) smuggling increased during the embargo but not enough to offset the direct effects of the embargo, and 2) the resultant effect on prices in Britain implies that the British economy suffered more from the embargo than the U.S. economy. Frankel's [1982, 307] claim that "...there is historical evidence that Americans were remarkably successful in 1808 at switching into the production of manufactured goods when they were cut off from their usual source of supply..." is still based entirely upon price statistics for a small number of commodities and samples of comments made by individuals writing on the subject at the time.

An example of the available data on trade is given in Table I, which presents figures from an 1816 publication on the dollar value of exports from selected areas of the United States for the years 1803 through 1810, incllusive. The values include both re-exports and the export of domestically produced goods, and the apparent decline in the level of foreign trade during the embargo period is staggering. Regardless of the obvious direct impact of the embargo upon official foreign trade statistics, few inferences can be made from such statistics about the effect of the embargo upon other important macroeconomic variables. In addition, little information is found on illegal or unreported trade.

This paper introduces a unique financial data set for this important period in U.S. history and uses it to analyze the effect of the embargo upon the domestic economy. The data are restricted to values for New England only, specifically the state of Massachusetts and the District of Maine. (3) However, New England was the most financially sophisticated region of the United States at this time and, as is evident from Table I, accounted for as much as 28 percent of all official U.S. export trade during the years 1803 through 1810. Excellent data for this time period are available on price indices of imported and domestically produced goods, on financial stock prices and dividends, and on both state and federal bond prices. Extensive balance sheet banking data for the commercial banks of the State of Massachusetts and the District of Maine are also available. (4) An examination of these data, (never previously utilized for this purpose) motivated by a theoretical model of money, banking, and financial markets, can contribute to the long-standing discussion of the effects of the Embargo Act of 1807 upon the economy of the United States.


The theoretical model underlying our analysis of the economic impact of the Embargo Act of 1807 is based upon a theory of money and loanable funds markets operating within the context of a real business cycle model. Current real business cycle models that include a financial or "banking" sector view the sector as performing no essential function other than providing financail services and intermediation. In King and Plosser's [1984] model, for example, observed correlations between money, prices, and real output result from an endogenous response of the banking system (that produces only transactions services) to business cycles driven by real disturbances. Williamson [1987] extends this type of model of include financial intermediation that functions endogenously in the manner of the banking sector in King and Plosser's model. In Williamson's model, however, financial intermediation and inside, or bank created, money "matter" in that a credit supply effect contributes to fluctuations in real output. As with other real business cycle models, williamson's model generates comovements among time series that are more consistent with observed phenomena than the theoretical models driven by monetary disturbances.

During the period under examination, 1803 through 1812, the United States operated under a bimetallic standard with convertibility between specie, "currency" (which consisted of banknotes issued by state commercial banks and by the First Bank of the United States) and bank deposits (both interest bearing and non-interest bearing.) (5) As the United States operated under a regime of fixed exchange rates, specie flows between the United States and other countries rendered the money supply endogenous to the United States.

The nominal money supply is defined as specie in the hands of the public ([SP.sub.p]) plus bank money held by the public ([M.sub.b]). Bank money held by the public is the sum of banknotes issued by commercial banks and in the hands of the public ([C.sub.p]) plus demand deposits at commercial banks (DD):

[M.sup.s] = [SP.sub.p] + [M.sub.b] = [SP.sub.p] + ([C.sub.p] + DD) (1)

Commercial Banks operated under a fractional-reserve system within which the level of banknote and deposit issuance depended, in part, upon the level of bank reserves. (6) For the commercial banks of this period, reserves consisted of specie in vault ([SP.sub.v]), bank notes in vault issued by other commercial banks ([NOB.sub.v]), and an accounting entry, specie due from other commercial banks (7) (DFB):

R = [SP.sub.v] + [NOB.sub.v] + DFB (2)

The volume of bank money in circulation, [M.sub.b], is a function of the reserve to bank money ration ([M/M.sub.b]), the circulating specie to bank money ratio ([SP.sub.p./M.sub.p]), and the value of "high powered money" (H), which is the sum of bank reserves and specie in the hands of the public:

[M.sub.b] = [1 / ([R/M.sub.b _ SP.sub.p./M.sub.p])].H, (H = R + [SP.sub.p]) (3)

All other things equal, an increase in the nominal interest rate will result in an increase in total bank reserves as the public economizes on banknote and specie balances (non-interest bearing assets) while increasing holdings of interest bearing assets (including deposits and bank stock), a decrease in excess reserves as banks purchase more earning assets, and a corresponding increase in the money supply. (8)

During the period in question, 1803 through 1812, the demand for loanable funds in New England was primarily a derived demand on the part of businesses that borrowed in order to finance bills of trade, inventories, business expansion, and so on. Most loans were short-term, although they were often renewed and the proceeds used for long-term purposes. Commercial banks hence operated, in part, as financial intermediates channeling private savings to business borrowers.

On analysis proceeds under the premise that early American banking activity of this type can generally be considered endogenous, that is, determined mainly by demand considerations. As previously noted, an endogenous response of the banking system to real disturbances, such as the embargo, is an essential feature of current real business cycle models that include a banking and financial sector. There existed no central banking authority, such as the Federal Reserve System, that attempted to affect banking and financial variables through such policy measures as altering monetary aggregates through open market operations, ro altering legal reserve requirements or discount rates. The First Bank of the United States did exist at this time, but other research by the authors indicates that aggregate banking activity in New England was hardly affected by the presence of the First Bank. (9)

Following the predictions of a real business cycle model, the passage and enforcement of the Embargo Act would initially cause a decrease in the level of foreign trade (as indicated in Table I) and a corresponding decrease in the current level of real output and income. (10) As a result, the demand function for loanable funds shifts leftward and the market interest rate falls as the total dollar amount of bonds traded decreases. The fall in real income causes the demand function for real money balances, as wel as the supply of savings on the part of the public, to decrease.

As compared with the pre-embargo equilibrium, during the embargo period the market interest rate will fall from the pre-embargo level to some lower rate. The nominal money supply decreases as a result of a decrease in the note and deposit issuance of commercial banks. Note and deposit issuance declines as the demand for bank loans by business declines, the absolute level of bank reserves declines, the reserve to bank money ratio, [R/M.sub.b], increases, and the specie to bank money ratio, [SP.sub.p./M.sub.b], increases. The demand for real balances decreases, and the real money supply is predicted to decline along with the nominal stock of money.

The level of bank reserves decreases as the incentive to exchange specie for earning assets falls with the decline in interest rates. The reserve to bank money ratio, [R/M.sub.b], increases as banks hold higher levels of reserves per dollar of banknotes and deposits outstanding in the face of lower interest rates and greater uncertainty. Likewise, the public shifts from using bank money to specie as interest rates fall and uncertainty increases.

The theory hence generates unambiguous testable hypotheses concerning the effect of the U.S. Embargo Act upon key New england money, banking, and financial variables. In addition to the predictions described, there are implications with respect to two alternative hypotheses that will be tested. If the embargo were ineffective, that is, had no significant impact upon the level of international trade, national, output, and real income, then the predicted changes in the variables of interest ought not to have occurred. The possibility arises that the embargo was only partially effective, depending upon various circumstances. For example, it has been mentioned that the embargo was possibly circumvented partially via illegal trading and shipping through Maine and across the Canadian border. Were this actually the case, the decline in economic activity in general, and monetary and banking activity in particular, ought to be less evident in Maine as compared with Massachusetts and the rest of New England. And, to the extent that the embargo would have a greater impact upon the economy of a port city as compared with an interior city, the decline in economic activity again ought to be less evident in the interior.


If the Embargo Act had a significant impact upon economic activity in the United States, this impact should be reflected in actual changes in the New England macroeconomic variables that conform with the predicted changes of the theoretical model. To test for this conformity, historical values of the money, banking, and financial variables of the theoretical model, or appropriate proxy variables, were developed and examined. The dollar denominated variables examined include both real and nominal bank loans, or "discounted notes," from the asset side of the individual bank balance sheets, which comprised a major component of the loanable funds market of the period. Also included is the bank-issued component of the nominal money supply, [M.sub.b], which consisted of banknotes in circulation plus the deposit liabilities of commercial banks. Deflating the nominal bank money supply by the price level gives the bank-issued component of the real money supply, [M.sub.b]/P. In order to test the predictions of the theory with respect to changes in these variables, pooled time-series cross-section models were estimated for the years 1803 through 1812, inclusive, using the individual bank as the observation. (11)

To determine the variation of market interest rates during the period, current yields on bank and other financial stocks (measured by changes in market prices) were calculated and examined, along with a weekly reported yield to maturity for U.S. Treasury bonds traded in the Boston market. To measure the effect of the Embargo Act upon these variables, Box-Jenkins intervention analyses were conducted on six proxies of bank stock interest yields developed for 426 weeks between 1803 through 1811, inclusive, and on U.S. Treasury bond yields for 468 weeks during the same period.

The Pooled Time-Series Cross-Section Models

To isolate the effect of the embargo upon the money, banking, and financial variables in question, an attempt must be made to control for as many influential, but non-embargo related, variables as possible. The absence of any yearly data on real income for the period constitutes the most serious deficiency. The lack of such data precludes many direct measures of the impact of the embargo, in addition to disabling attempts to specify explicit structural saving, investment, and money demand functions. Likewise, any change in real money balance holdings on the part of the public cannot be broken down into those attributable to changes in real income and those attributable to changes in velocity. Nonetheless, changes in those variables that can be observed and that conform to the predictions of the theory presented above ought to support inferences concerning the remaining unobserved variables of interest. Two observable variables that, a priori, should affect the levels of both real and nominal bank loans and banknote plus deposit issuance of banks are the population of the bank's market and the asset-size of the bank.

The population of the market in which the bank is operating should be directly correlated with the amount of banking activity undertaken as measured by both real and nominal bank loans outstanding and both real and nominal banknote plus deposit liabilities outstanding. All other things being equal (such as income per capita), a bank operating in a larger city would report larger dollar values of loans, banknotes, and deposits outstanding than a bank operating in a small town. To control for different populations of markets, and changes in population over time, nominals and real loans and banknote plus deposit liabilities are divided by the population of the market served by the bank (n) to state the variables in per capita terms. (12)

All other things being equal, the dollar value of loans, banknotes, and deposits outstanding per bank will be a function of the size of the bank's assets, which is a function of the paid-in capital account. (While total assets and liabilities of a given bank could vary greatly from period to period depending upon economic conditions, the paid-in capital account was generally constant.) Larger banks, in terms of total assets, will obviously report higher values for individual asset and liability items thatn smaller banks. This will be due, in part, to the number of customers using the bank, which itself will be a function of the population the bank is serving. It is less obvious, however, that the per capita level of loans, banknotes, and desposits outstanding per bank will be a function of the asset size of the bank. Yet from an a priori standpoint, this ought to be the case. In general, a wealthier city will exhibit a higher dollar level of economic activity, per capita, than a poorer city, including a higher level of banking activity per capita. Banks serving wealthier markets should report higher levels of paid-in capital than banks serving poorer markets, controlling for population served. In addition, the value of paid-in capital per bank fluctuated little if at all over the period in question, while the values of the other asset and liability variables in the model exhibited wide swings. The paid-in capital (deflated) of the individual bank (SIZE) is hence included as a proxy variable for the wealth of the region served by the bank.

At this time in U.S. history, the United States had a much higher ratio of foreign-trade generated income to total national income than at any time since. All other things equal, banks in seaport cities should have been more affected by a foreign trade embargo than interior banks, controlling for wealth and population of the market served. A dummy variable accounting for the difference between interior and seaport banks (INTERIOR) is hence included as an explantory variable in the pooled model.

To test the phypothesis long held among historians that the embargo may have been defeated partially by an increase in the volume of trade through Maine and across the Canadian border, a dummy variable distinguishing Maine banks from Massachusetts banks (MAINE) is included as an explanatory variable. If an increase in trade occurred through Maine, the predicted decline in the levels of the monetary and banking variables examined should be mitigated for Maine banks when compared with Massachusetts banks. Other things being equal, a higher real volume of trade through a region would require a higher level of banking activity in general as well as a higher level of real and nominal loans, banknotes, and deposits in particular.

Finally, a dummy variable (EMBARGO) separating the embargo period from the non-embargo period is included to measure the change in the dependent variables, bank loans per capita ([LF.sub.b/n]), and bank money issued per capita ([M.sub.b/n]), which can be attributed to the (unobserved) decrease in real income during the embargo. The dependent variables are specified in both nominal and real terms, generating four equations to be estimated. The pooled regression equations thus take the form:

[([LF.sub.b/n]) = [[alpha].sub.0] + [[alpha].sub.1][(SIZE)] + [[alpha].sub.2](INTERIOR)

+ [[alpha].sub.3](MAINE) + [[alpha].sub.4](EMBARGO) + [] (4)

[([M.sub.b/n])] = [[beta].sub.0] + [[beta].sub.1][(SIZE)] + [[beta].sub.2](INTERIOR)

+ [[beta].sub.3](MAINE) + [[beta].sub.4](EMBARGO) + [] (5)

where i = observations on a given bank, i = 1,..., 18 t = observations for a given year, t = 1803,..., 1812 and u, v are error terms.

Generalized least squares estimation is appropriate since the dependent variable values, and hence the error terms, for the cross-section banking units are serially correlated over time and heteroscedastic within each given year. (13)

The Box-Jenkins Intervention Models

To determine the effect of the embargo on New England market interest rates, seven time-series were developed. Six of the series serve as a proxy for the Boston market interest rate, and consist of 426 weekly high prices of banking and financial stocks traded on the Boston market during the period. (14) The remaining series consists of 468 weekly yield-to-maturity quotations on U.S. Treasury 6 percent bonds of 1798 traded in the Boston market during the period. Univariate time-series models for the seven series were specified and estimated, including an intervention component (a binary dummy variable) that would measure differences in the behavior of the series during the embargo period.

A general univariate autoregressive integrated moving-average (ARIMA) model can be written as:

[[phi](B)W.sub.t] = [[theta](B)a.sub.t] (6)

where [W.sub.t] = [(1-B).sup.d.Z.sub.t], B is the backward shift operator, and [Z.sub.t] is the variable in question, such as the index of bank stock returns. The polynomials in B are defined as:

[phi](B) = 1 - [[phi].sub.1]B -...- [[phi].sub.p.B.sup.p] [theta](B) = 1 - [[theta].sub.1]B -...- [[theta].sub.q.B.sup.q] (7)

and [A.sub.t] is the usual random shock or white noise series. As usual, p, d, and q represent the autoregressive, difference, and moving-average polynomial orders, respectively. This model can be written in level form as:

[Z.sub.t = {[theta](B)/[[phi](B)[(1-B).sup.d][} . [A.sub.t]

To include the intervention component, we write:

[Z.sub.t] = [[Omega](B)I.sub.T/.sub.t] + {[[phi](B)/[[theta](B)(1-B).sup.d][} . [A.sub.t]

where [I.sub.T/.sub.t] is the intervention variable, that begins in period T, and [Omega](B) is the general functional form of the intervention component and its coefficient, [Omega].


Bank Loans, Bank Money, and Bank Reserves

Table II presents mean per capita dollar loans per bank and mean per capita note plus deposit liabilities per bank for the years 1803 through 1812 (excluding 1811) in both nominal and real terms. The movements in these variables over time are consistent with the predictions of the theoretical model. In both nominal and real terms, mean bank loans per capita increased steadily between 1803 through 1806 inclusive. Loans, both nominal and real, monotonically decreased between 1807 and 1809, inclusive, and then recovered after the embargo until the War of 1812. Nominal loans per capita fell an average of $5.16 per bank during the embargo period, which was over 5 percent of a mean value of nominal loans per capita per bank of $99.06 for the entire period 1803 through 1812. During the embargo, real loans per capita were $4.20 lower per bank, on average, which was just under 5 percent of a mean value of real loans per capita per bank of $90.11 for the entire period 1803 through 1812. (15)

Both nominal and real bank-money supplies increased from 1804 through 1806, inclusive, monotonically decreased between 1807 and 1809, inclusive, and then monotonically increased each year thereafter. During the embargo years, nominal banknote plus deposit liabilities per capita fell an average of $10.28 per bank, which was over eighteen percent of a mean value of nominal banknote plus deposit liabilities per capita per bank of $56.50 for the entire period 1803 through 1812. During the embargo, real banknote plus deposit liabilities per capita were $8.97 lower per bank, on average, which was over 17 percent of a mean value of real banknote plus deposit liabilities per capita per bank of $51.36 for the years 1803 through 1812.

These four measures of banking activity comprise the dependent variables of the pooled regression models discussed above. Table III presents the results of these pooled time-series cross-section regressions. As indicated in Table III, the levels of all four dependent variables were significantly lower during the embargo period than at other times, which is a finding consistent with the predicted effects of the embargo from the theoretical model. However, coefficients of the Maine bank dummy variable and the interior bank dummy variable are insignificant in the pooled model estimated for the entire period. Further analysis indicated that multicollinearity between the MAINE and EMBARGO variables significantly inflated the variances of the coefficients on the MAINE regressor. To isolate the changes in the dependent variables for Maine banks versus Massachusetts banks, the pooled data set was divided into two components, one containing observations from the embargo period and the remaining containing observations from the pre-embargo and post-embargo years. (16) Table IV presents the results of the regressions on the divided sample using real values of the dependent variables. Results indicate that the real banknote plus deposit liabilities as well as the real value of bank loans for Maine banks were significantly higher during the embargo period, but this difference is not evidence in the pre-embargo and post-embargo years. During the embargo, real loans per capita per bank averaged $46.41 higher for Maine banks than for Massachusetts banks, which amounted to more than 51 percent of a mean value of loans per capita per bank of $90.11 for all of New England during the entire period 1803 through 1812. During the embargo period, real banknote plus deposit liabilities per capita per bank averaged $19.26 higher for Maine banks than for Massachusetts banks, which amounted to more than 37 percent of a mean value of banknotes plus deposits per capita per bank of $51.36 for all of New England during the entire period. The mean increase in both variables for Maine banks during the embargo is greater in magnitude than the mean decline in the variables for New England as a whole. These results are consistent with the hypothesis that an increased real volume of trade occurred through Maine and Canada during the embargo.

The decline in economic activity was apparently the same for interior banks as for seaport banks during the embargo. Per capita loans and per capita banknote plus deposit liabilities, per bank, both real and nominal, exhibited no significant differences for interior banks versus seaport banks during the embargo as compared with the years preceding and following the embargo. In the pooled models, the coefficients delineating interior banks from seaport banks are never statistically significant.

The reserve positions of banks during the peirod conform closely with the predictions of a conventional theory of fractional reserve banking. The absolute level of bank reserves fell from an average of $188,835 per bank during the non-embargo years to an average level of $78,180 per bank during the years 1807 through 1809. The vault specie holdings of banks fell from an average of $94,622 per bank during the non-embargo years to an average level of $50,507 per bank during the years 1807 through 1809. These declines amounted to over a 34 percent drop in total reserves and over a 46 percent drop in specie holdings, respectively, yet no banks in the sample either failed or suspended specie convertibility during the period.

Interest Rates

The changes in market interest rates as measured by fluctuations in financial asset prices are entirely consistent with the predictions of the theoretical model. The decline in the current yield on bank stock is statistically significant and ranged from a minimum of 1.25 percent for the Union Bank, (from a weekly average of 7.5 percent to 6.22 percent) to a maximum of 2.0 percent for the National Bank (from a weekly average of 7.5 percent to 5.5 percent.) The yield on U.S. Treasury bonds traded in the Boston market fell from a weekly average of 6.2 percent before the embargo to an average of 5.9 percent during the embargo.

Table V presents the results of the time-series intervention analysis of the six financial stock price indices and the U.S. Treasury bond yield described above. The estimation techniques are the same for each series. First, the autocorrelation functions and partial autocorrelation functions of each series were estimated and examined. In each case the series required first order differencing to achieve stationarity. Higher-order differencing was not required as the adopted ARIMA models were tested for stationarity and exhibited no significant drift in mean or variance. In each case the auto- correlation functions of the differenced series suggested the ARIMA processes to be specified. Each model was then estimated in two stages. First, conditional least squares produced initial estimates of the values of the parameters. These estimates were used as starting points for a Gauss-Marquardt nonlinear maximum-likelihood procedure that produced the final estimates of the parameters by backcasting. The residuals were examined to determine the adequacy of the models and the need for further modeling. When the residuals and their auto- correlation revealed no obvious inadequacies in the model, it was adopted. (17)

The ARIMA parameters are significant at the .01 level or better in nineteen of twenty-one cases, and significant at the .05 level in all cases. The EMBARGO parameters are significant at the .01 level in all six cases for the bank stock series, and the values of the parameters are of uniform magnitude. (18) This finding would be expected in an efficient market with arbitrage tending to equate current yields across all stocks rated in a market. The decline in the Treasury bond yield is statistically significant when comparing the mean values for each period (t-value = -4.81, [n.sub.2]=[n.sub.2]=69), but the embargo intervention coefficient, which constitutes a more powerful test, is statistically insignificant in model 7. (19) In each case, plots of interest rate movements produce similar results, viz., current yields begin falling in late 1807 about the beginning of the embargo period, reach a minimum around July and August of 1808, during the middle of the embargo period and rise back to the pre-embargo period level by the middle of 1809.

The realizations, which are the first differences of the financial asset prices, all have long moving averages, suggesting little variation from period to period. (A statistically significant but weak upward trend in the indices was eliminated by the differencing.) A majority of the series exhibit a quarterly disturbance, MA [13], that was correlated with fixed dividend payments on the respective financial stocks. A similar four-month disturbance common to all six stock price indices, MA [17], offers no similar obvious explanation. Again, the absence of additional data on macroeconomic and financial conditions for the period hinders greater in-depth analysis.


This study examines the impact of the Embargo Act of 1807 on New England economic activity during the years the law was in effect. A theory of endogenous money and loanable funds markets operating within the context of a real business cycle model motivate an empirical analysis of changes in monetary, banking, and financial variables for the periods before, during, and after the actual embargo. Using New England bank statistics, the weekly prices of six stocks and the yield on a U.S. Tresury bond traded on the Boston market, the study considers four major hypotheses:

1) The embargo significantly affected the levels of bank loans, the bank-issued component of the money supply, and current financial stock and bond yields, in a manner predicted by the comparative statics of the theoretical framework.

This hypothesis is strongly supported by the empirical results. The decline in the level of foreign trade and the resultant decline in national income were accompanied by a significant decline in bank loans. Both nominal and real loans per capita were lower for all of New England during the embargo period as compared with the years preceding and following the embargo. The bank-issued component of the money supply of New England significantly decreased during the embargo. Both nominal and real banknote plus deposit liabilities per capita ([M.sub.b]) were lower for all New England during the embargo as compared with the years preceding and following the embargo. Current market interest rates were lower during the embargo as compared with the years preceding and following the embargo.

2) These effects were more evident for seaport markets than for interior cities as a result of the direct impact of the embargo upon the level of foreign trade. Contrary to the predictions of the model, we found no evidence to support this hypothesis.

3) The increased volume of trade through Maine and into Canada, including illegal smuggling, increased monetary and banking activity in Maine as compared with Massachusetts and the rest of New England during the embargo.

This hypothesis is strongly supported by the empirical results. A substantial difference occurred in the effect of the embargo upon banking activity in Maine as compared with Massachusetts. Both nominal and real bank loans per capita and nominal and real banknote plus deposit liabilities per capita were significantly higher for Maine banks than for Massachusetts banks during the embargo. There was no significant difference in these variables between each region during the years preceding and following the embargo.

4) The decline in national product and income as a result of the severe reduction in the level of foreign trade caused a shift toward domestic manufacturing and self-sufficiency in the U.S. economy.

Our study provides little evidence to support or refute this argument. Again, an absence of good yearly estimates of real income and national output for the period under examination precludes a rigorous test of whether the embargo Act significantly affected the trend toward domestic manufacturing and self-sufficiency for the United States. Some inferences none the less can be made from our results. The drop in the level of foreign trade from the pre-embargo peak to 1808, as measured by the official export statistics in Table I, amounted to a 75 percent decline for Massachusetts and an 80 percent decline for the United States as a whole. Assuming no substantial differences between the changes in the yearly banking statistics used in our study and the unobserved values for these variables for the remaining banks in the United States (which likewise assumes an insignificant volume of inter-regional capital flows for the period) the decline in foreign trade greatly exceeds the 5 percent fall in bank loans and the 15 percent drop in the money stock. However, using Goldin and Lewis' [1980, 25] estimate of the ratio of value of exports to national income of 14.4 percent for 1807, the (estimated) percentage drop in (unobserved) aggregate real income between 1807 and 1808 (-10.8 percent) is remarkably close to the percentage drop in mean real bank loans (-10.3 percent) and the drop in mean real bank-issued money (-10.5 percent) for the same year. Comparing the ratio of banking variables to total exports for New England at the beginning and end of the period leaves a similar impression. The dollar value of exports for the United States was 16.4 percent higher in 1810 than in 1803; mean real bank loans 15.5 percent higher in 1810 than in 1803; and nominal bank loans 19.3 percent higher in 1810 than in 1803. In other words, examining the money and loanable funds markets provides little evidence to support North's hypothesis of a major shift of the economy towards domestic manufacturing: The changes in the values of important money and banking variables between 1803 and 1810 are proportional to the exogenous changes in the level of foreign trade. A significant rise in domestic manufacturing, had it occurred, should have resulted in a corresponding increase in domestic money and banking activity. Such an increase is unobserved.

The embargo could have initiated such a shift, or accelerated a previous trend, but the four year period between 1808 and 1812 is simply too short to measure such a shift as North argues occurred by 1811. Rather, changes in the level of activity in the loanable funds market and the money market appear to be proportional to an estimated change in the level of real income that can be attributed entirely to foreign trade considerations. whereas our results do not disprove North's view (a shift from agricultural or handicraft production for domestic markets to manufacturing for domestic markets could have occurred without necessarily affecting the aggregate level of bank loans) neither do these results, or new data, support his argument in any way.

Finally, the data used in the study allow only the tracking of intersection points of curves in the money and loanable funds markets based on yearly values for a priod of nine years. As such, any attempt to identify and estimate the structural relationships of these markets through a simultaneous equation system is not possible. Still, we feel the current study does shed some new light upon an old issue, while at the same time reinforcing the applicability of more recent economic theories and econometric techniques to distant historical events.

(1) See North, [1961, 53].

(2) For an attempt to estimate the role of exports upon the level and growth of real income in the United States at this time, and a discussion of the problems involved, see Goldin and Lewis, [1980]. Other articles of importance concerning the history of the embargo include Adams [1980] and Frankel [1982]. For an earlier discussion of the growth of smuggling during the period, see Heaton [1941].

(3) During the early nineteenth century, New England, consisting of what was later to become the states of Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, and Vermont, was relatively more financially integrated than the remainder of the United States. Unlike the remainder of the United States, for example, convertibility between specie, banknotes, and deposits was generally not suspended by banks in New England over the entire period 1785 through 1837. Our data set for the present study is based, in part, upon yearly balance sheet variables for fifteen banks that operated in the state of Massachusetts, (which represent over 80 percent of the Massachusetts banks in existence during those years), and three banks in the District of Maine, (which represent 50 percent of the banks in existence in the District during the period.) The eighteen banks in the sample represent over 62 percent of all the banks in existence in New England in 1803, and over 41 percent of all banks in existence in 1806. The banks in the sample account for almost 40 percent of the total assets of all New England banks for the period of the study, as measured by the authorized capital of the particular bank.

(4) For a great number of years, economic historians had to generalize from fragments of data about the operations of commercial banks and their attendant contribution to the nation's money supply in the early nineteenth century. In 1965, however, Fenstermaker [1965] provided a considerable amount of aggregated state bank statistical data for the 1819-37 period. These data were aggregated from asset and liability items on individual bank balance sheets located in banking, governmental, and historical archives. The authors have undertaken a project to extend these statistical data back to 1785 and forward through the Civil War, to fill in missing components, and to analyze more thoroughly the growth of early American banking and financial markets and the components of the nation's money stock. This article utilizes the New England component of this data set, as described in Fenstermaker, Filer, and Herren [1984].

(5) During the period under examination, banks did not distinguish between interest bearing deposits and non-interest bearing deposits on their reported balance sheets. For the purposes of this study, all such deposits at commercial banks are simply labeled "demand deposits." While general convertibility was not suspended in New England during the period, individual banks at times would suspend specie convertibility when the bank failed or reserve levels became critically low. None of the banks included in the present study failed or suspended convertibility during the years 1803 through 1812. The discounting of banknotes was a general rule, however, outside the area of original issue. The First Bank of the United States did not act as a central bank during these years. Rather, along the lines of the Bank of England, after which it was modeled, the First Bank competed with the existing state charted commercial banks. Not all commercial banks at this time were privately owned. Vermont, for example, chartered a commercial bank that was owned and managed by the state.

(6) In general, the dollar volume of banknotes and deposits created by a bank depended upon the equivalent value of specie (or banknotes of other commercial banks that were redeemable for specie) deposited with the bank plus some multiple (usually double) of the paid-in capital account of the bank. The capital account was assumed to be paid in specie.

(7) Banks often made deposits of specie in other banks or with agents at various commercial centers to facilitate redemption of their notes. For example, Maine banks often deposited specie in Boston banks for local note redemption purposes. These deposits were listed as assets on the depositor bank's balance sheet and were counted as part of the depositor bank's total reserves. It would have been interesting to test specie alone as reserves, but bank reports often lumped specie together with due from banks and notes of other banks, and in fact, all three were used as reserves.

(8) None of the substantive relationships or predictions of the model depends upon the interest elasticity of the nomical money supply being significantly different from zero. However, empirically this finding seems to be the case, which is consistent with the money supply being endogenous under fixed exchange rates.

(9) See Fenstermaker and Filer [1986].

(10) Note that we do not assume an exogenous shift in income. The comparative statics model used examines shifts in the money and loanable funds (bond) markets, and is isomorphic to an IS-LM model. (We merely drop the goods market, under Walras' Law, rather than the bond market, because of the lack of data on real income.) To close this model, we implicitly assume an aggregate demand and aggregate supply model. The Embargo Act constituted an exogenous event that significantly decreased the level of real exports, as documented in Table 1. (There was no prohibition of imports under the conditions of the Act, and while imports did decrease, the decline was significantly less than the decline in exports.) The significant decline in the level of real exports caused the aggregate demand curve to shift to the left. Given an upward sloping aggregate supply curve, the result was a lower equilibrium price level and a lower equilibrium level of real income. While income is unobserved, the empirical movement of the price level conforms with the model and is summarized in Smith and Cole, [1935, 146]. The price level (base = 1802) rises steadily from 100 in February 1803, to 121 in January 1807, falls to a low of 84 in June and July of 1808, and then rises back to 111 by December 1810. The embargo theoretically could cause the aggregate supply function to shift inward as well, by interfering with international specialization. However, the significant drop in the price level indicates that, if present, such a shift of aggregate supply constituted a second order effect compared with the decline in exports and aggregate demand.

(11) For econometric reasons, the years 1803 through 1812 were chosen since the number of banks in the data set is constant for that period By 1814, two additional banks had begun operating in Maine, and an additional ten in Massachusetts (with one older bank ceasng operations). Yearly cross-section regressions were specified and estimated for each of the years 1803 through 1814, inclusive, excluding 1811, for which only aggregate data were found. As the results of these regressions are consistent with those from the pooled model, albeit less powerful statistically, the results are not reported here. Interested readers may obtain these results from the authors upon request.

(12) In Maine, two banks were located in the city of Portland. In Massachusetts, three banks were located in Boston, and two banks were located in the cities of Nantucket and Salem. In each of these cases, to avoid underweighing the observation, the dependent variables were divided by one-half the population of the city in which the bank was operating (and by one-third in the Boston case). For all other banks, there was one bank per city, and for these observations the values of the dependent variables were divided by the population of the relevant city. Actual population figures for all years are generally not available. The authors obtained population figures from numerous sources, including the U.S. Census of Population between 1780 and 1840, and for each city fit a polynomial regression of the form:

CITYPOP = [[tau].sub.0] + [[tau].sub.1] (YEAR) + [[tau].sub.2] [(YEAR).sup.2] + [[tau].sub.3] [(YEAR).sup.3] + u.

The estimated regression equations were used to interpolate the population of cities and towns in years for which actual data were unavailable.

(13) The only variable that changes much over time and effects the time series behavior is EMBARGO. The other variables scale the level of loans of loans or liabilities between banks. The implict assumption we make is that all other disturbances can be included in a well-behaved error term. Again, we are limited by the paucity of macroeconomic data. The estimator is described in judge [1985, 518].

(14) Changes in current market interest rates can be measured by examining, say, actual reported bond discount rates or calculated from actual changes in bond prices. During the embargo period, actual market interest rates were not reported regularly. But weekly financial stock prices (high and low for the week) were reported in the Boston newspapers. (Closing prices would improve the argument, but are unavailable.) After being issued, these bank stocks were traded at market prices and paid dividends at announced yearly percentages of par (100). While called "stocks", these financial instruments more closely resembled consoles and prices were reported as "discount" or "advance" (premium) deviations from par. A "stock" that paid a "dividend" of 7.5 percent actually paid $7.50 per year, regardless of the market price of the stock. If the market price were actually $100, the current yield would be 7.5 percent. If the market price rose to, say, $115, the annual dividend payment remained $7.50, and the current yield fell to 6.5 percent. Announced dividends changed very little over the period for some stocks examined, and not at all for others. Hence, changes in (unobserved) current market interest rates should be highly correlated (inversely) with changes in reported financial stock prices, adjusting for the infrequent changes in dividends.

(15) While new loans created dropped significantly during the embargo years, many of the loans reported on the banks' balance sheets undoubtedly were of longer term and possibly "troubled" or uncollectible. While banknote plus deposit liabilities for a given bank could drop quickly as the public redeemed bank money for specie, the total dollar value of bank loans outstanding ordinarily would decline less rapidly. Indeed, this process is often the first series of events leading to a liquidity crisis for a bank. The ratio of total loans to total banknote plus deposit liabilities per bank ([LF.sub.b]/[M.sub.b]) increases significantly from a pre-embargo and post-embargo mean of 1.74 to an embargo period level of 2.28 (t-value from a dummy variable regression: 3.98, d.f. = 161).

(16) A Chow test was applied to test for the effect of pooling upon the MAINE coefficient. The difference between the unconstrained sum of the squared errors, SSE, (from the separate regressions for the real dependent variables from Table IV) and the constrained SSE (from pooled models using real dependent variables and estimated for the entire period) is significant at the 1 percent level ([F.sub.4,154] = 7.09), indicating that the coefficient for MAINE is different for the embargo versus the pre-embargo and post-embargo periods. An interaction term for MAINE and EMBARGO was included in equations (4) and (5), but the high degree of collinearity between the three dummy variables significantly inflated the variances of the parameter estimates. As the results for the separate regressions with nominal dependent variables are consistent with those reported for real dependent variables, only the real results are presented here.

(17) As is usually the case with Box-Jenkins models, it can rarely be concluded that one has found the "best" model. The "parsimonious" model that whitens the residuals is "adequate." In the present study, because of the importance of identifying the best ARIMA model possible to properly test the effect of the embargo on the series, numerous specifications of the ARIMA models for each series were estimated and tested involving higher order ARMA processes and differencing. No alternatives produced mean square errors as 1 w as the specifications reported (see Box and Jenkins, [1976]).

(18) EMBARGO, which is the intervention parameter estimate for Table V, is defined as [tau] in the intervention component, [Omega](B)l = [[tau]B/(1-B)]l , of equation (9). This linear intervention expression assumes that the effect of the embargo on the asset price series is constant each period (see Box and Tiao, [1975]).

(19) For the six bank stocks examined, the volume of earning assets of the banks involved clearly decrease during the period along with note and deposit issuance. These factors would strengthen the decline in interest returns on the bank stock of the firm. For the U.S. Treasury bonds, however, we track only the quoted weekly yield to maturity and do not examine the level of bond issuance, spending, and taxation of the federal government during the period in question. Changes in these variables could mitigate a drop in the Treasury bond return as compared with the drop in return on regional bank stock in the Boston market. We were able to find little information on the total volume of 1798 six-percent bonds traded in the Boston market during the embargo, but the total net increase in the federal debt in 1798 was only $4.5 million, (which does not control for refunding.) The first federal budget deficit in over a decade occurred in fiscal year 1808-1809. Treasury bond issuance and lower bond prices might account for the relatively small decline in the market return. It should be noted that the Treasury bond rate was rarely considered the "risk-free" rate; uncertainty over the future of the nation ordinarily resulted in the Treasury rate including a significant risk premium.


Adams, Donald R., Jr. "American Neutrality and Prosperity, 1793-1808: A Reconsideration." Journal of Economic History, December 1980, 713-37.

Boston Gazette. Boston, Massachusetts (microfilm), weekly, 1803-1812.

Box, George E. P. and Gwilym M. Jenkins. Time Series Analysis: Forecasts and Control, 2nd ed. San Francisco: Holden Day, 1976.

Box, George E. P. and George C. Tiao. "Intervention Analysis with Applications to Economic and Environmental Problems." Journal of the American Statistical Association, March 1975, 70-79.

Fenstermaker, J. Van. The Development of American Commercial Banking 1782-1837. Kent State University Bureau of Business and Economic Research, Printed Series No. 5, 1965.

Fenstermaker, J. Van, John E. Filer, and Robert S. Herren. "Monetary Statistics of New England, 1785-1837." Journal of Economic History, June 1984, 441-53.

Fenstermaker, J. Van and John E. Filer. "Impact of the First and Second Banks of the United States and the Suffolk System On New England Bank Money, 1791-1837." Journal of Money, Credit, and Banking, February 1986, 28-40.

Frankel, Jeffrey A. "The 1807-1809 Embargo Against Great Britain." Journal of Economic History, June 1982, 291-307.

Goldin, Claudia D. and Frank D. Lewis. "The Role of Exports in American Economic Growth During the Napoleonic Wars, 1793 to 1807." Explorations in Economic History, 1980, 6-25.

Heaton, Herbert. "Non-Importation, 1806-1812." Journal of Economic History, November 1941, 178-98.

Judge, G. G., W. E. Griffiths, R. C. Hill, H. Lutkepohl, and T. C. Lee. The Theory and Practice of Econometrics, 2nd. ed. New York: Wiley & Sons, 1985.

King, Robert G. and Charles I. Plosser. "Money, Credit, and Prices in a Real Business Cycle." American Economic Review, June 1984, 363-80.

Malone, Dumas and Basil Rauch. The Republic Comes of Age: 1798-1841. New York: Appleton-Century-Crofts, 1960.

Martin, Joseph E. A. Century of Finance, The Boston and Money Markets from January 1798 to January 1898. Boston: Published by Author, 1898.

Massachusetts Legislature. "A True Abstract of the Statements of Several Bank Corporations in the Commonwealth of Massachusetts," Broadsides Printed by Order of Honorable Senate, Boston: 1803-1812.

McMaster, John B. A History of the People of the United States, From the Revolution to the Civil War, Vol. III. New York: Appleton & Company, 1903.

North, Douglass C. The Economic Growth of the United States, 1790 to 1860. Englewood Cliffs, N.J.: Prentice-Hall, 1961.

Pitkin, Timothy. A statistical View of the Commerce of the United States of America. Hartford, Connecticut: Charles Hosmer, 1816.

Smith, W. B. and A. H. Cole. Fluctuations in American Business, 1790-1860. Cambridge, Massachusetts: Harvard University Press, 1935.

Williamson, Stephen D. "Financial Intermediation, Business Failures, and Real Business Cycles." Journal of Political Economy, December 1987, 1196-216.

Professor of Economics and Finance, Towson State University, and Professor of Economics, the University of South Alabama, respectively. We would like to thank J. Ronnie Davis, James Swofford, the editors and anonymous referees for valuable comments. The usual caveat applies.
COPYRIGHT 1990 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1990 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Fenstermaker, J. Van; Filer, John E.
Publication:Economic Inquiry
Date:Jan 1, 1990
Previous Article:Long-run equilibrium in the empirical study of monopoly and competition.
Next Article:On integrating the Ricardian equivalence theorem and the IS-LM framework.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters