Monetization and Growth in Colonial New England, 1703-1749

Peter L. Rousseau� and Caleb Stroup

Vanderbilt University

Revised March 2011

Abstract: We examine econometrically the real e¤ects of paper money�sintroduction into colonial New England over the 1703-1749 period. Departingfrom earlier analyses that focus primarily on the depreciation of paper moneyin the region, we show that expansion of the money stock promoted growthin modern sector activity and not the other way around. We also �nd that billsemitted for seigniorage purposes had a positive e¤ect on the modern sector, whilebills issued through loan banks did not.

Keywords: Paper money; bills of credit; seigniorage; land bank; colonialAmerica; �nance and growth; vector error correction model.

*Corresponding author: Department of Economics, Vanderbilt University,Nashville, TN 37235, USA. Tel: (615) 343-2466. E-mail: Peter.L.Rousseau@Vanderbilt.Edu.

1. Introduction

In the hundred years that followed its founding in 1620, New England was transformed

from a small subsistence farming collective to the commercial hub of the New World. The

rise of trade and commerce in the region was consistent with the pursuit of its comparative

advantage, given that poor soil and climate, at least when compared to that of the South and

the Chesapeake regions, limited the production of cash crops such as indigo and tobacco.

Yet the rise of commerce was inhibited for some time by the lack of a viable medium of

exchange. In this study, we examine whether the introduction and subsequent expansion of

paper money by colonial legislatures had quanti�able e¤ects on real activity in colonial New

England. The emphasis on the relationship between monetization and growth di¤ers from

earlier work that tends to focus on adherence to the quantity theory of money, the in�ationary

e¤ects of paper money, and how the New England case typi�ed its mismanagement (e.g.,

West, 1978; Smith, 1984, 1985; Michener, 1987; Grubb, 2005; O¢ cer, 2005).1

The main hypothesis is simple: by reducing transactions frictions and increasing the

availability of entrepreneurial credit, monetization with paper issues facilitated the emergence

and development of New England�s commercial sector. This seems reasonable given the

economic environment that prevailed there in the 18th century. In particular, New England

emerged as a hub for a diverse and sophisticated network of non-agricultural (i.e., modern)

industries such as processing, trade, distribution and marketing, as well as a variety of small-

scale production activities. Such activities required a widely-accepted, readily-available and

storable medium of exchange.

While book credit, barter, and inland bills of exchange were serviceable for some transac-

tions, a lower-cost method of exchange was desirable for trades made at arms length. Gold

or silver coins could have served this function, but the colonies had been prohibited from

minting their own and settlers could only bring limited quantities of specie across the At-

lantic. Further, England�s mercantilist policies tended to drain specie from the colonies �so

much so that coins were driven from circulation in New England by 1717 (O¢ cer, 2005, p.

115). At a time when specie was scarce or non-existent, issues of paper money encouraged

trade and commerce.

1Our views are consistent with Lester (1938, 1939), who focused on the negative e¤ects ofin�ation in the colonies but admitted to the possibility that monetary expansion had somepositive real e¤ects.

1

In this article we show that paper money had measurable e¤ects on modern sector activity

using the tools of modern macroeconomics. In particular, estimates from vector error cor-

rection models (VECMs) and the associated impulse responses indicate that money played

a leading role in commerce and not the other way around. Our methodology allows for sep-

arate identi�cation of short and long-run e¤ects, and we �nd the e¤ect of monetization on

real activity to be particularly potent in the long-run. This is consistent with monetization

stimulating growth and not just temporary increases in spending. We also �nd that direct

money issues by colonial legislatures, used mostly to �nance military con�icts, had stronger

real e¤ects than indirect emissions through public loan banks.

This is not to say that the New England colonies enjoyed a smooth path to monetization.

Rather, the region had considerable di¢ culties in providing an ample medium of exchange

for its growing population (see Rousseau, 2006, Fig. 2, p. 104). But our evidence indicates

that government issues of paper money were a useful and growth-promoting expedient at a

time when barter and its variants represented the set of alternatives.

Section 2 reviews the economic environment in colonial New England and describes key

moments in its history of paper money. We discuss the framework for estimation in Section

3, and describe our data in Section 4. We test for a link between money and real outcomes

in Section 5. Section 6 examines the robustness of our �ndings to modeling choices, and

Section 7 concludes.

2. Historical background

2.1. The rise of a commercial center

Not long after establishing the Massachusetts Bay Colony in 1629, New England colonists

discovered that their climate was not well suited to large-scale farming.

Some manufacturing did emerge during the 17th century but was for the most part

unsuccessful due to high �xed costs and a lack of skilled labor (McCusker and Menard, 1985

p. 97). When a dramatic fall in immigration during the 1640s led to plummeting prices,

bankruptcies and foreclosures (Canny, 1994, p. 41), provincial leaders tried to legislate

recovery by providing tax breaks, monopoly rights and subsidies to preferred industries, but

2

these interventions did not have the desired e¤ects.2 Moreover, many of the non-agricultural

sectors that did eventually succeed required technologically advanced inputs from England,

and poor infrastructure initially hampered distribution of these inputs to interior towns such

as Greenwich and New Salem.3

Despite the di¢ culties faced by some sectors, infrastructure improvements initiated by

New England legislatures in the 1640s and 1650s, such as quality controls, inspections and,

most importantly, the construction of road and bridge networks, began to facilitate arm�s

length market interactions among interior townships and between townships and the coastal

merchants who provided British wares.4 Inland traders quickly exploited these networks

to trade, in exchange for household surpluses, the advanced products that were necessary

for diverse small-scale manufacturing and processing industries such as rum distilling, iron

smelting and molasses production. The existence of these trading networks promoted the

expansion of industries around New England�s older export sectors. For example, the �shing

and grain trades required caskmaking, shipping, and middlemen services, while metal bands,

produced by ironworks, were required for caskmaking. Shipbuilding required timber and

milling, and development of the latter led to cheaper inputs for caskmakers. Production

diversity led to development of the required preparation and packing warehouses, as well as

accounting and insurance functions for traders and merchants (Middleton, 2002, p. 208).

As a result, New England settlers soon became pro�cient in providing transportation,

2 For example, in 1641 Boston�s General Court set out to develop ironworks (Woodman,1975). The implementation was carried out by John Winthrop Jr., who made several at-tempts at Braintree, Massachusetts in 1641 and later near the town of Lynn in 1646, but alack of skilled labor led to the project�s failure. Cases such as this demonstrated that ironsmelting was best kept limited to small-scale bloomaries that produced mostly elementarytools (Middleton, 2002, p. 90).

3For example, rum could not be re�ned without mechanical equipment imported fromBritain, and New England�s nascent processing sector relied on British machinery such ascrankshafts, gears, cogs, axles, hammers, nails, �les, chisels, knives and plows (Middleton,2002).

4Connecticut and Massachusetts general courts, for instance, committed funds to completeinter-town infrastructure and large projects like the Hampton ferry and a bridge across theSaugus river near Lynn. Various towns also required inhabitants to construct local roadsand bridges collectively (Newell, 1998, p. 55).

3

storage, distribution and mercantile services. This once again re�ected their comparative

advantage. Unlike the South�s homogenous plantation exports, which were simple to manage

and naturally outsourced to Dutch, English, Scottish and later New England merchants, the

diverse production and processing industries that emerged in the northern colonies by the

late 17th century required sophisticated complementary functions. New England increasingly

provided these functions for herself, the southern and middle colonies, and the Caribbean.

By the 1750s, New England was the New World�s center for processing, transportation,

storage, distribution, marketing and mercantile services, with the areas around Boston and

Newport having over 100 distilleries and rum re�neries. By the 1770s, there were more active

sugar re�neries and rum distilleries in Massachusetts, Connecticut and Rhode Island than

in the other ten colonies combined (McCusker, 1970, pp. 431-437; McCusker and Menard,

1985, pp. 290-293). New Yorkers even complained that their wheat was purchased by Boston

traders, processed in Massachusetts, and sold back to New Yorkers at a pro�t (Newell, 1998,

p. 69). Yet instead of being monopolized by Boston merchants, the majority of New Eng-

landers managed to participate in some aspect of the market system. The region�s economic

development was su¢ ciently pronounced that English merchants soon became concerned

about competition from the Northern economy: �[New England] is already the rival and

supplantress of her mother�(from A Comparison Between the British Sugar Colonies and

New England, 1732, pp. 8-9).

2.2. The origins and uses of a paper currency

New England�s diverse production mix and complex trading system required a viable

transactions medium, such as specie. Specie was widely accepted, stable in value, and easier

to carry around than more bulky commodity monies, yet the crown for the most part banned

the export of specie and bullion to the colonies. This meant that coins were in short supply,

especially outside of Boston, and this increased transactions costs (Nettles, 1934, pp. 125-

126, 138-141).

Without adequate specie for day-to-day transactions, merchants, traders and producers

used a variety of substitutes such as barter, commodity monies, shop notes, book credit, and

bills of exchange. Each of these alternatives created transactions frictions because trades

could not be completed until some �double coincidence of wants�was ultimately identi�ed

4

and settled.5 Trade within towns was often coordinated by shopkeepers who, in addition to

supplying �nished goods, also kept transactions records for the residents.6 But shop notes,

which were issued by shopkeepers in exchange for goods and were in turn often paid out

by employers to laborers for redemption at the store, tended to limit consumption to items

that a particular shopkeeper had on hand.7 Book credit, a close relative of shop notes, could

remain unused for months until a shopkeeper or customer presented an item for exchange

that the other desired.

Barter was also common between towns and country traders. For example, Thomas

Fitch, a wealthy Boston merchant, purchased imports with a variety of goods brought from

the west. Fitch often advanced goods to traders on credit and required speci�c products

in return. For example, in a letter dated April 1, 1703, he notes that he provided nails to

a country trader who agreed in return to deliver �Turpentine in August or Septem. next.�

(from Thomas Fitch Letterbook).

Country traders traveled extensively to supply creditors with speci�c products and would

often travel to multiple destinations before �nding the desired matches. For example, an

inland trader might take furs from Lancaster, credit a seller�s account, carry the furs (and

5In an account of the business practices of Boston merchant Thomas Hancock, Baxter(1945, p. 23) notes that �the frequency with which goods like �sh and meat were shuttledto and fro suggests that they were not being traded for reasons that we should now considernormal. Instead of taking them for his own use or for resale at a pro�t, a merchant probablytook them for the simple reason that he had no hope of receiving payment in any otherform, and could reasonably expect in turn to pass them on to his own creditors.� TheHancock records show that Thomas and his contemporaries fell back on �commodity money�frequently and that it was in everyday use among traders long after it had been abandonedas an o¢ cial means of payment.

6For example, shopkeepers kept records of day labor owed by one settler to another.

7In a 1721 reply to a �Gentleman,�pampleteer John Colman writes in regard to barter,shop notes and lack of paper currency that �Again (the gentleman) tells us that there arebills enough to buy up all the produce of the country ... but were that true, is there enoughto buy and sell with in the shops, is there enough to pay laborers and tradesmen, withoutforcing them to take goods, which they do not know what to do with, except to put themon their backs; for which some people are very angry, and say they go beyond their degree;whereas the people would not spend it in so many ways, if it were at their own disposal; andthe merchants cannot pay them otherwise than by shop notes, because the shops can�t sellfor money; and subsequently can�t pay money to the merchants; and thus is trade miserablyembarrassed, and the poor oppressed for want of a medium�(Davis, 1911, Vol. 2, p. 75).

5

other goods) to the Atlantic, exchange them for British imports in Boston, and �nally travel

back to Lancaster (trading along the way) where the original seller would take goods equal

in value to his credit. All of this involved risk because traders might not survive the return

journey or acquire suitable goods in exchange. Inland bills of exchange could have facilitated

some of these transactions and compensated for a lack of money, but the extent to which

such bills were used by less wealthy traders is unclear.

The introduction of paper money relaxed bartering constraints, even though the �rst and

many subsequent issues were intended to �nance government expenditures and especially mil-

itary con�icts. The Massachusetts legislature approved the �rst colonial paper money issue

in 1690 to �nance a failed military expedition in Quebec. These tax anticipation bills were

not printed for use as a general transactions medium, but rather as a way to defer payment

to the returning militia. The initial plan was to levy a special tax the next year that could

be paid in real goods, specie or tax bills, but by the time of their scheduled redemption the

bills had become circulating currency and were considered �better than money�(Hutchin-

son, 1936, Vol. 1, pp. 340-341).8 By 1710 Thomas Fitch, though he continued to conduct

most of his business in inland bills, was regularly accepting Massachusetts bills of credit in

payment of debts.9

The popularity of these tax bills (or �tenor�as they were called) led to the �rst peacetime

emission in 1702, and in 1709 the Hartford government issued Connecticut bills to combat

the �great scarcity of money�and, conveniently, to �nance an expedition against the French

(Hutchinson, 1936, Vol. 2, p. 289). New Hampshire also issued its own bills, followed by

Rhode Island in 1710. The tenor of all four colonies circulated at par with one another

and crossed borders freely within New England (McCallum, p. 149), and in 1712 became

8A 1721 pamphlet by a certain �Amicus Patriae�o¤ers a list comparing 1720 prices ofvarious goods in bills of credit with prices at which these goods �formerly sold for money,�noting the bills could be �esteemed by us on par with money�(Davis, 1911, vol. 2, p. 178).To the extent that we do not know whether �money�in this context meant silver coins orcommodity money, the comparison may suggest that bills of credit were supplanting barteras a means of payment.

9For example, a letter dated 26 June 1710 from Fitch to a Mr. Edward Shippen statesthat �a few days past I received of Thom. Mault in part of his bond one hundred and fortypounds in bills of credit of this province.�Another note dated 17 July 1710 to Mr. ThomasPhipps states that �this serves to advise that I received eleven pounds for your account inbills of credit�(Thomas Fitch Letterbook).

6

legal tender for public and private debts. While bills were ostensibly redeemable at future

dates, tax levies were not commensurate with emissions and the supply of bills in circulation

increased throughout the �rst half of the 18th century.

Paper money got into circulation in two ways. First, seigniorage emissions were made to

�nance government debts directly, the bulk of which were related to military expeditions but

could also be used for government salaries, infrastructure projects, fort repairs and day-to-day

administrative expenses. The second method of emission was through public �loan banks.�

Under this system legislatures disbursed bills to town leaders who selected citizens to whom

the bills would be loaned, using land as collateral.10 To the extent that loan bank bills were

based on underlying assets, they did not increase the money supply permanently.11 Smith

(1985, pp 549-551) argues, on the other hand, that loan bank money was only imperfectly

tied to underlying asset values and therefore may have had more permanent e¤ects.

New England�s money supply increased dramatically over the period 1703-1749, even

after accounting for currency depreciation. As Table 1 shows, there were £ 112,800 old tenor

10These public loan banks are not to be confused with private schemes to start �banks�that would lend their own notes on security of land. There was considerable agitation amongcolonists to start such private banks between 1685 and 1732, none of which were successful(Davis, 1911, Vol. 1). Though the e¢ cacy of public versus private loan banks was a hotlydebated issue among contemporaries, the public schemes ultimately won the day. Goldberg(2011) recounts the failure of the �rst private scheme for a land bank in Massachusetts in1685. Starting in 1733 groups of merchants managed to issue private notes in Connecticut,New Hampshire and Massachusetts, but legislative restrictions and unsuccessful operationsensured that private notes would never exceed 15.1 percent of the region�s money supplyand would average only 6.7 percent over the 11 years (1733-43) that such notes circulated(O¢ cer, 2005, p. 104).

11Benjamin Franklin, still not long out of New England in 1729, writes in regard to loanbank money that �if it should be objected that emitting it at so low an interest and on sucheasy terms will occasion more to be taken out than the trade of the country really requires,it may be answered that there can never be so much of it emitted as to make it fall belowthe land it is founded on; because no man in his senses will mortgage his estate for whatis of no more value to him than that he has mortgaged. And if it should ever become soplenty by indiscreet persons continuing to take out a large overplus, above what is necessaryin trade, so as to make people imagine it would become by that means of less value thantheir mortgaged lands, they would immediately of course begin to pay it in again to theo¢ ce to redeem their land, and continue to do so till there was no more left in trade thanwas absolutely necessary.�(from A Modest Enquiry into the Nature and Necessity of a PaperCurrency, reprinted in Davis (1911, Vol. 2, p. 354)).

7

outstanding in 1710. By 1749 the outstanding total of £ 4,033,700 implied a cumulative

increase of 3,475 percent. Of special interest is that while this paper money did lose value, it

only depreciated by 694 percent over the period. Most of this depreciation occurred between

1745 and 1749 when colonies printed large amounts of bills to fund the Louisburg expedition

during the War of Jenkin�s Ear. The incomplete price adjustment implies that paper money

emissions led to increases in the real money stock.

Anecdotal evidence described in Brock (1975, 1992) and in Davis (1911) shows prolonged

and intense discussions regarding the relative costs and merits of the New England monetary

experiments. With regard to the currency debate, detractors, many of whom were creditors

(i.e., Boston merchants) emphasized the consequences of currency devaluation, while sup-

porters argued that transactions media stimulated trade and commerce and that existing

money stocks were not su¢ cient to support the region�s growing economy.12

It is well known that New England paper money often depreciated rapidly, especially

after 1740. The New England colonies, and Rhode Island in particular, are often held up as

classic examples of overissuance. While not disputing that the issues were often in�ationary,

we believe that the existing discussion, by focusing primarily on the costs of in�ation, ignores

an important dimension of paper money issues: monetization. In a world with better data,

one would ideally like to estimate the relative in�ationary costs and monetization bene�ts

of colonial paper money. In absence of comprehensive economic data, however, we take the

more modest approach of investigating whether increases in the real money stock and paper

money stimulated economic activity. The next section sets out the quantitative framework

that we use to do this.

3. Framework for estimation

Our analysis begins with the well-known quantity theory of money and its equation of

exchange

MV = PY; (1)

12Again, Franklin writes in 1729 that �there is a certain proportionate quantity of moneyrequisite to carry on the trade of the country freely and currently; more than which wouldbe of no advantage in trade, and less, if much less, exceedingly detrimental to it� (Davis,1911, Vol. 2, p. 336).

8

where M is the money supply, V is the velocity of money or the number of times that a

single currency unit is used over some �xed period of time, P is the price level, and Y is a

measure of real economic activity. Because V is not observable, assuming it to be constant

over time is su¢ cient to make Eq. (1) estimable. This approach, as adopted by West (1978)

after also assuming constancy of Y and taking logs, generates the usual empirical test of the

quantity theory

ln(P ) = �0 + �1 ln(M) + �2 ln(Y ) + "; (2)

where a positive and signi�cant coe¢ cient on �1 is typically viewed as evidence that increases

in nominal money are re�ected in higher prices. Our empirical analysis relaxes the assump-

tion of a constant Y as described in Rousseau (2006, 2007, 2010). In that formulation, the

release of pent-up demand for a low-cost transactions medium draws more transactions into

the market sector of an economy and has an additional positive spillover on that sector�s pro-

duction. Still assuming constant velocity, rearrangement of (1) thus suggests a relationship

between real balances and economic activity:

Y = f

�M

P

�: (3)

Following Rousseau (2007), we develop an empirical representation of Eq. (3) that allows

for the possibility that MPdetermines Y and also that Y determines M

P: One way to test for

long-run comovement among these variables is to use the cointegration framework developed

in Johansen (1991). If a pair of nonstationary variables are cointegrated, there exists a

linear combination of them that is stationary. Engle and Granger (1987) further show that

there would be a valid �error correction�mechanism through which the variables are related.

This vector error correction model (VECM) is formed by embedding the stationary linear

combination in an otherwise standard vector autoregression (VAR) in �rst di¤erences. For

a VAR with k lags, the VECM takes the form

4Yt = �1 +k�1Xi=1

�1;i4Yt�i +k�1Xi=1

�1;i4�M

P

�t�i+ 1

�aYt�1 + b

�M

P

�t�1

�+ "1;t; (4)

9

4�M

P

�t

= �2 +

k�1Xi=1

�2;i4Yt�i +k�1Xi=1

�2;i4�M

P

�t�i+ 2

�aYt�1 + b

�M

P

�t�1

�+ "2;t: (5)

Note that if Y and MPhave unit roots and are cointegrated, all of the terms in the

regression are stationary, including the residuals "1;t and "2;t. Consider equation (4), which

regresses 4Y on a constant, a series of its own lags, lags of 4�MP

�, and the cointegrating

relationshiphaYt�1 + b

�MP

�t�1

i. The lag di¤erences capture short-run adjustments up to lag

k � 1 in both 4Y and 4�MP

�: The cointegrating vector (i.e., [a,b]0) speci�es the long-run

equilibrium relationship. When used with the data to form the linear combination in the

square brackets of Eq. 4, the coe¢ cient on this error correction term (ECT), 1, gives the

speed at which 4Y adjusts to deviations in the long-run relationship. With a normalized

to unity, the signs of b and allow the investigator to deduce the direction of any long-run

statistical e¤ect of �uctuations in real money on Y: Similarly, the coe¢ cient 2 in Eq. 5

gives the speed at which 4�MP

�adjusts to deviations in the ECT.

To see how the estimated coe¢ cients indicate the direction of statistical linkages, consider

the case where none of the short-run parameters are signi�cant and 1 and 2 are both

signi�cant. This would imply that MPleads Y in a long-run statistical sense and vice versa.

In this case the model cannot identify a unidirectional e¤ect.

Suppose, however, that 1 is statistically signi�cant but 2 is not. This would indicate

that Y responds to �uctuations in the equilibrium relationship between MPand Y , but that

MPdoes not. If b and 1 were both negative, this would imply that negative deviations from

the equilibrium relationship generated by increases in MPwould force the dependent variable

Y upward. An econometrician would say that MPis �weakly exogenous for Y ,�meaning that

MP�causes�Y in a long-run statistical sense.

The tests must be interpreted cautiously, however, since as a statistical device a rejection

of the null hypothesis does not necessarily imply causality in an economic sense. In particular,

the validity of the tests is predicated on the inclusion of the full information set in the

underlying VAR, which is violated in any �nite regression framework. It also requires that

there are no variables omitted from the system that a¤ect bothM and Y but to which either

M and Y simply responds more quickly. For these reasons, our results can be seen only as

suggestive of the economic linkages operating within each system.

Since a stable VECM is an algebraic re-formulation of a cointegrated VAR in levels, we

can also gain information about the overall adjustment process by estimating the underlying

10

VAR and calculating the associated impulse responses.13 The VAR takes the form

Yt = �1 +kXi=1

�1;iYt�i +kXi=1

�1;i

�M

P

�t�i+ "1;t; (6)

M

P t= �2 +

kXi=1

�2;iYt�i +

kXi=1

�2;i

�M

P

�t�i+ "2;t; (7)

where the coe¢ cients on the regressors in levels allow the investigator to trace out the overall

e¤ects (i.e., combining the short and long-run) of Y to shocks in MPand of M

Pto shocks in

Y with impulse response functions.

4. Data

We begin with a measure of New England�s money supply constructed from various

sources. This includes treasury-issued and loan bank tenor (bills of credit), privately-issued

merchants�notes, and the stock of specie.

The amounts of government-issued paper money in circulation for each of the four New

England colonies (i.e., Massachusetts, Connecticut, New Hampshire, and Rhode Island) are

the cumulative sum of issues less any retirements from Carter et al. (2006, Vol. 5, Series

Eg302-Eg305, pp. 692-3, Historical Statistics hereafter).14 The paper money quantities are

constructed from data on bills issued directly by treasuries (hereafter �seigniorage bills�)

and those issued on loan through public loan banks. Later we will examine, in addition to

the e¤ect of the total real money supply on proxies for Y , the relative e¤ects of these two

components. We obtain the loan bank component by using, for every instance in which a

colonial legislature approved a loan bank emission, the value of paper money emitted and

the pre-speci�ed redemption schedule to calculate the balance outstanding in each year.

13Examples of using the VAR methodology to elucidate the direction of linkages betweenmoney and output in the economic history literature include Rousseau and Wachtel (1998)and Rousseau and Sylla (2005).

14The data in Historical Statistics are corrected and re�ned versions of series that appearin Brock (1975).

11

The underlying accounts of loan bank issues are from Brock (1975, pp 23, 26, 38, 44-45,

47-49). Seigniorage bills are obtained by subtracting the outstanding loan bank bills from

the total stock of government-issued paper money. Given par circulation of bills across the

New England colonies prior to 1750, the total amount of government-issued paper money

for the region is the sum of the outstanding balances for the four colonies.

For privately-issued notes, we use the series compiled by O¢ cer (2005, pp. 114-16). These

include Boston Merchant�s notes, New London Society notes, New Hampshire Merchant�s

notes, and notes of the Massachusetts-chartered private �Land Bank�and �Silver Bank.�

No comprehensive measure of the specie stock exists for the period, but O¢ cer (2005, pp.

116-17) constructs a proxy in Massachusetts tenor equivalents from the existing fragmen-

tary sources, and we use his estimates here.15 Table 1 lists the total money stock and its

components along with the other series used in our main analysis.

[Insert Table 1 about here]

To convert the nominal stock of paper money into real terms, we need a measure of P.

This presents a challenge because there is no comprehensive price index for New England over

our sample period. One approach, taken by McCallum (1992) and Smith (1985), assumes

sterling prices to be stable and uses the exchange rate to convert tenor to sterling values.

We prefer to measure the price level using real goods actually traded in Massachusetts,

and build a new price index to do so. There are only three available series for the period:

the prices of a bushel of wheat (1703-1749) and a gallon of molasses (1720-1749) from Cole,

(1938, Table 36, p. 117), and a hundredweight of merchantable cod�sh (1703-1749) from

Historical Statistics (Vol. 5, Series Eg248, p. 675). Since the price series for molasses begins

15O¢ cer�s underlying sources include a direct estimate of £ 200,000 at the end of 1699 fromDavis (1900). There are also two estimates of the relative amounts of specie to bills. Brock(1992, p. 7) states that: �in 1710 there was perhaps as much silver in circulation as therewere bills.�Davis (1911, vol. 4, p. 157) cites an anecdotal account from 1743 which placesbills in circulation at three times silver balances in 1712. The �nal observation is a generally-accepted estimate of 1717 as the year in which no specie remained in circulation. O¢ certhen measures the amount of specie in circulation by linearly interpolating these points usingthe trade de�cit as a weighting factor.

12

in 1720, we backcast it to 1703 using a �rst-order autoregressive model.16 The �nal index,

presented in Table 1, averages the three price series after normalizing each to unity in 1703.17

Since we are testing for e¤ects of monetization on activity in the modern sector, it is

not obvious that the ideal measure of Y would be gross domestic product, given its large

agricultural component, even if it were available (Rousseau and Sylla, 2005, p. 11). On the

other hand, imports from England in constant 1700-1702 sterling are continuously available

for the period of our study from Historical Statistics (Vol. 5, Series Eg437, pp. 710-11) and

may serve as a proxy for New England�s modern sector so long as they are representative

of activity therein.18 These data, presented in the �nal column of Table 1, were compiled

under supervision of the British Inspector General and are su¢ ciently comprehensive because

the majority of New England trade occurred between Boston and London, rather than with

Southern Europe, the East Indies, South America or other British colonies (McCusker, 1978).

It may seem that there is some disconnect in relating the money stock, and indeed paper

money only after 1717, to English imports given that imports were ultimately paid for with

specie bills of exchange or real goods. But it is likely that increases in real cash balances

improved the e¢ ciency of local production and promoted local trade. This in turn would have

increased income and/or foreign trade credits and so increased the demand for imports.19

Consistent with Michener (1987), it is also possible that increases in cash balances freed

16This involves estimating a univariate AR(1) model for molasses prices running the databackwards from 1749 to 1721 and using the estimated coe¢ cients to backcast a value forthe series in 1720. We then insert the value for 1720 into the dataset and repeat the processuntil all values to 1703 are backcasted.

17West (1978) used only wheat prices. O¢ cer (2005) adds molasses prices after 1720 toprovide a broader index but does not treat the years before and after 1720 symmetrically.

18Exports from New England are a less satisfactory proxy because they were in�uencedheavily by Navigation Laws, especially after 1740, and this series aggregates both NewEngland exports with re-exports from the Caribbean and other colonies.

19Some contemporaries certainly believed this to be the case. In a 1734 pamplet titled �AModest Apology for Paper Money,� for example, the writer states that it is �most certainthe richest trading countries have the most paper money; they cannot carry on so great atrade without it, and have got part at least of their riches from it. It has facilitated theirown commerce among themselves at home, and that is for ever the one and true way to havea large and gainful trade abroad�(Davis, 1911, Vol. 3, p. 93).

13

specie credits to obtain more imports.

Our analysis ends in 1749 because the Currency Act of 1751, which forbade the New

England colonies from issuing further legal tender bills of credit, and anticipation of the Act,

seems to have broken down the de facto system of par circulation of paper monies within

the region. Massachusetts had begun to reform its currency by 1750, issuing interest-bearing

Treasury notes that e¤ectively put it on a specie standard for the remainder of the colonial

period.20 Other colonies were slower to reform, with Connecticut doing so in 1756, and New

Hampshire and Rhode Island delaying until the early 1760s. Given that reforms came at

di¤erent times, it is di¢ cult to imagine that the New England paper monies continued to

circulate against each other at par after the Currency Act.21

5. Main results

We �rst check the stationarity properties of real money and imports from England (Y ).

The �rst column of Table 2 presents Augmented Dickey-Fuller test statistics for these series,

which do not reject the null hypothesis of a unit root in any of the money series. We therefore

go on to test for a cointegrating relationship between the two series. The Johansen (1991)

statistics for the trace and maximum eigenvalue tests presented in the right panel of Table

2 indicate that at least one of the tests is consistent with a single cointegrating relationship

between English imports and each component of MP. By inserting the cointegrating vector

from the Johansen procedure into Eq. 4 and Eq. 5, we can then form a VECM to identify any

long-run e¤ects that are present. A lag order (i.e., k) of three was chosen for the underlying

levels VARs based upon a series of nested likelihood ratio tests, so that the corresponding

20See the discussion in Brock (1975, pp. 244-56). With regard to a regime shift, McCallum(1992, pp. 154-155) observes that �students of colonial money have agreed that the paperissued by Massachusetts during this period (i.e., 1754-61) did not circulate as a medium ofexchange.�

21McCusker (1978, pp. 136, 152) reports an exchange rate of 26:1 for Rhode Island paperagainst the pound sterling in 1760, which is ten years after Massachusetts had returnedto a specie standard. Massachusetts paper traded at nearly 11:1 against sterling in 1749,just before reforms were enacted, and its Treasury notes �uctuated between 1.3:1 and 1.4:1against sterling in the 1750s.

14

VECMs use two �rst di¤erences of the variables in each system.

[Insert Table 2 about here]

Our main results are presented in the �rst two columns of Table 3, labeled �Total Money

Supply,�which present estimates for Eq. 4 and Eq. 5. The coe¢ cient on the error correction

term (ECT) in the imports (Y ) equation is statistically signi�cant at the one percent level

with a negative sign. When combined with the negative loading for b in the cointegrating

vector from the Johansen test (presented in the �nal row of the table), this implies a strong

positive response of English imports to low frequency movements in the money supply. The

absence of statistical signi�cance for the ECT in the money equation indicates that this long-

run statistical link is unidirectional. The positive e¤ect of money on imports is not evident

in the VECM�s short-run parameters. The Durbin-Watson statistics o¤er no indication of

serial correlation in the residuals from either equation.

[Insert Table 3 about here]

We now quantify the dynamics of this adjustment. In particular, we would like to know

if shocks to money are associated with discrete increases in English imports at some speci�c

horizons, or if the e¤ects accumulate gradually. Because the VECM is a re-formulation of

a standard VAR in levels, stationary and serially uncorrelated errors in the VECM imply

that the corresponding VAR in levels will also be stationary. This allows us to evaluate

adjustment dynamics by estimating Eq. 6 and Eq. 7 and using the coe¢ cients to compute

the responses of each variable to a one standard-deviation shock to the other.

Panel A of Table 4 presents estimates for Eq. 6 and Eq. 7. The �rst two rows present the

individual coe¢ cient estimates. Centered below these are the signi�cance levels of F -tests of

the null hypothesis that the coe¢ cients on the �rst three lags of a given variable are jointly

zero. The results are consistent with those found in the VECM: real money balances lead

imports but not the other way around.

[Insert Table 4 about here]

15

The upper panel of Fig. 1 shows the corresponding impulse responses along with 90 per-

cent con�dence bands. The �gure on the left plots the percentage change in English imports

(Y ) over time to a one standard deviation shock to MP. The e¤ect becomes statistically

signi�cant (i.e., the lower 90 percent con�dence band rises above zero) after two years and

persists until fourteen years after the initial shock. On the right we plot the response of MPto

a one standard deviation shock to English imports: Here, the e¤ect is not signi�cant, again

con�rming that money leads imports in a statistical sense but not the other way around.

Overall, results from the VECMs, VARs and impulse response functions o¤er strong evidence

that New England�s monetization as measured by the total money stock led to permanent

increases in economic activity as measured by imports from England.

[Insert Figure 1 about here]

We now examine the e¤ects of government paper money emitted through each of New

England�s two mechanisms: bills emitted for seigniorage by treasuries and bills authorized

by legislatures to be loaned on collateral to colonists. We denote these two series by S and

LB, respectively. Columns 3 through 6 of Table 3 present estimates from Eq. 6 and Eq. 7.

Results for the VECM system with seigniorage bills are similar to those for the total money

stock in that the coe¢ cient on the ECT is negative and statistically signi�cant and there is a

negative loading on b in the cointegrating vector. Low frequency �uctuations in seigniorage

bills therefore have a strong statistical e¤ect on English imports in the long-run.

The �nal two columns of Table 3 report results for loan bank bills. Here, the ECT

is signi�cant in the LBPequation at the one percent level, indicating that legislatures may

have emitted loan bank bills in response to low frequency �uctuations in real economic

activity. The overall sign of this e¤ect, given by the negative coe¢ cient on the ECT and

positive loading for a in the cointegrating vector, is negative, indicating that decreases in

English imports preceded increases in loan bank emissions. This may be because economic

downturns led colonists to pressure legislatures for new emissions. Turning to the imports

(Y ) equation, the ECT is weakly signi�cant and the estimated e¤ect of land bank money

on English imports is small and negative. This suggests that loan bank emissions may not

have been conducive to modern sector development.

The associated VARs in Table 4 bear out these �ndings. The F -tests in Panel B indi-

16

cate that seigniorage bills weakly lead English imports at the 16 percent level, but Panel

C indicates the presence of bidirectional links between loan bank emissions and imports.

The corresponding impulse response functions in Fig. 1 show a protracted e¤ect of seignior-

age issues on English imports that begins by the third year after emission and continues

through the eleventh year. Moreover, there is no measurable e¤ect of imports on the timing

of seigniorage emissions. This is similar to our �ndings for the entire money stock.

By comparing Panel A with Panel B in Fig. 1, we can see that the e¤ect of shocks to

the total money stock had a larger and more protracted e¤ect on our proxy for Y than

did seigniorage issues. This is consistent with the view that paper money, while e¤ective

in monetizing the region, may not have been as e¤ective as specie would have been, the

latter not being subject to in�ation and being more widely accepted outside the region. The

impulse responses for land bank emissions in Panel C of Fig. 1 also indicate no measurable

e¤ect of loan bank bills on English imports but repeated and long-term e¤ects of imports on

loan bank bills.

Our �nding that seigniorage issues led economic activity but that loan bank issues did

not calls for explanation. Franklin (1929) o¤ers some clarity with respect to loan banks. He

reasoned that if a land owner borrowed money from a loan bank, using his land as collateral,

and if money issues made his land more valuable by promoting economic activity, he would

repay the loan quickly because the collateral would soon exceed the value of the money

borrowed. Thus land bank issues would play a stabilizing role in the economy (see fn. 11

above). At the very least, loan bank money was o¤ered at interest to colonists and had a

well-de�ned repayment schedule, meaning that these issues were temporary in nature.

Seigniorage bills, on the other hand, were less predictable in both the quantities and

timings of emission and well as the extent to which they would actually be sunk through

taxes. Thus, when a war required an unexpected issue of bills, they would immediately

enter the economy to purchase goods and stimulate economic activity. And if these bills

were not fully removed from circulation before new emissions were made, the real money

stock would grow so long as prices did not fully adjust, and we already know that they

did not. This mechanism highlights why relaxing the assumption of a constant Y can be

crucially important when testing the quantity theory of money so that, rather than being

purely in�ationary, exogenous shocks to the money supply through seigniorage bills can draw

transactions into a still under-monetized market sector of the economy.

17

6. Robustness

Along the way we have made decisions about the relevant sample period and how best to

measure the real money supply and modern sector activity. While we have already addressed

the economic motivation for our choices in Section 4, we now explore the robustness of

our main result to various alternatives. To this end, each component of Table 5 reports

the coe¢ cients on the ECTs in VECM systems, along with Durbin-Watson statistics and

loadings on the cointegrating vectors, after altering the sample period and/or replacing one

of the data series with a reasonable substitute.

[Insert Table 5 about here]

Panel A of Table 5 reports on robustness related to the choice of de�ator for the money

stock. In the main analysis we used an index constructed from wheat, molasses and cod�sh

prices. An alternative used by Smith (1985) and others assumes that English prices were

stable so that the exchange rate of tenor against sterling can be used to de�ate the money

supply. The �rst regression in Panel A presents results from this estimation exercise. The

estimated coe¢ cients on the ECTs retain the same sign and level of statistical signi�cance

as in the left panel of Table 3, with money leading English imports (Y ) and not the other

way around.

The second VECM in Panel A considers robustness to proxying New England prices

with the index of English prices constructed by O¢ cer (2005) with data originally drawn

from O�Brien (1985, pp. 788�789, 793�794). This alternative assumes that transport costs

were su¢ ciently small so that purchasing power parity held between England and the New

England colonies. If true, goods prices net of transport costs should be the same in both

locations so that English prices would be an acceptable proxy. As before, the ECT is

signi�cant in the English imports (Y ) equation but not in the real money equation.

Our own approach used actual Massachusetts prices of wheat, molasses and cod�sh, but

we needed to backcast the incomplete molasses series from 1720 to 1703. The third regression

in Panel A uses instead the simple average of the two series for which continuous data do

exist from 1703 to 1749. Once again our main result holds.

Another possibility is that �uctuations in imports and real money balances were driven

18

by di¤erentially-timed responses to other exogenous variables, and particularly to war and

peace. If bills of credit were printed to �nance wars and imports from England declined

during wars, a subsequent return to normal levels of imports and a contraction of the money

stock after a war might produce a statistical link from money to imports that appears to

be causal but is really just a result of the particular sequencing of the variables and the

dynamics of war. During our sample period there were two wars involving New England:

Queen Anne�s War (1703-13) and the War of Austrian Succession (1742-49). As a check on

the robustness of our VECM results, we re-estimate using subsamples that exclude these two

events and report the results in Panel B of Table 5. For all subperiods (1703-42, 1713-49, and

1713-42) the ECT in the imports equation retains the anticipated sign and is statistically

signi�cant at the �ve percent level. In the �rst regression in Panel B (1703-42) the ECT

in the money equation is also statistically signi�cant at the �ve percent level. Though this

o¤ers evidence of a bidirectional linkage between money and English imports, it would not

be surprising to see real money balances rise at times to accommodate increases in real

activity.22 Rather, our statistical technique simply shows that in most cases the link from

money to real activity is dominant.

Another sample-period robustness check relates to the monetary regime under study. Our

analysis focuses on the monetary regime from 1703 to 1749, yet data do exist for the New

England colonies up to 1774. We have already noted that the Currency Act of 1751 placed

Massachusetts on a specie standard for the remainder of the colonial period and that the

other New England colonies resisted reforming their currencies for many years. This means

that the par circulation of bills that the New England colonies enjoyed before 1750 would

have broken down quickly after the Act, and that any time series covering the full time span

from 1703 through 1774 would likely be subject to a regime shift in 1750. To verify that

there was indeed a regime shift, we run the models in Table 3 using data from 1703-74 and

compute a Chow test for parameter stability across the 1703-49 and 1750-74 subperiods for

all of New England and for Massachusetts only. In both cases the null hypothesis of no

regime shift is rejected in both the imports and money equations at the one percent level.

22As a further check of the alternative wartime theory, we plotted real balances and Englishimports over the period from 1703 to 1714 and found that both of the series trended upward,inconsistent with the hypothesis that wartime money emissions coupled with collapses intrade are driving our results.

19

This o¤ers econometric as well as economic justi�cation for ending our analysis in 1749.

We proceed to consider alternative measures of modern sector development that include

available data for Scotland and report the results in Panel C of Table 5. This procedure has

an advantage and a disadvantage. The advantage is that there may have been substitution

between Scottish and English imports or exports. On the other hand, the Scottish series

only go back to 1740, so if there was substantial unrecorded trading before then, adding in

the observed Scottish data may bias our estimates. In any event, the �rst regression in Panel

C uses the combined real imports from Scotland (Historical Statistics, Vol. 5, series Eg453,

p. 715) and England as the proxy for Y . We �nd that the results are qualitatively identical

to those obtained with our main speci�cation in Table 3 that used only English imports.

We believe that imports are the best available proxy for Y , but exports could also be

used so long as re-exports from the Caribbean, which are included in the exports series, are

uncorrelated with the money supply. This condition is not testable, however, because we

do not have a series for re-exports that could be used to form a measure of �fundamental�

exports from New England to the mother country. Nevertheless, the second regression in

Panel C re-estimates the VECM using the combined exports of New England to Scotland

(Historical Statistics, Vol. 5, series Eg444, p. 714) and England (Historical Statistics, Vol.

5, series Eg430, pp. 710-11) as the proxy for Y . Once again our main result holds.23 The

third regression in Panel C proxies for Y with total trade volume (I+X �the sum of imports

and exports including Scotland). The results for this case are similar to those obtained with

our other proxies for Y .

Finally we examine robustness to changes in various components of the money supply,

returning to our earlier use of English imports as the proxy for Y . The �rst regression in

Panel D of Table 5 considers only the paper money supply. The argument for omitting silver

in circulation is that these data are based on weak and fragmentary evidence. This presents

a dilemma, however, in that if silver was indeed in circulation until 1717, as is generally

accepted, omitting it from the measured money supply from 1703 to 1717 would incorrectly

force an excessively rapid rise in the series over these years. To avoid this, we run our VECM

model using data from 1718 to 1749 �the period when paper represented the entire money

23We obtain similar results when using exports and the sum of imports and exports toEngland only.

20

supply. This adjustment of the sample period does not a¤ect our main result.

The second and third regressions in Panel D repeat the VECM models in Table 3 for

seigniorage and loan bank bills excluding the period from 1703 to 1713. This not only

controls for Queen Anne�s War but allows us to compare the responses of English imports

to seigniorage and loan bank issues at a time when bills from both sources circulated. The

coe¢ cient on the ECT in the equation for English imports is smaller than over the full

sample but retains the expected sign and is statistically signi�cant at the �ve percent level.

The coe¢ cient on the ECT in the imports equation with loan bank bills is not statistically

signi�cant but, as in Table 3, low frequency increases in English imports do lead increases

in loan bank emissions.

7. Conclusion

Our primary contribution is to use the available historical data on money, prices and economic

activity to show that the New England colonies�experiments with paper money in the early

18th century in�uenced the real economy of the region in a quantitatively meaningful way.

We do this using the macroeconometric tools of vector autoregression, vector error correction,

and impulse response analysis. These tools help us to show that monetary expansions at this

time in colonial history had their strongest impact on growth in the medium to long term �so

much so that linkages in this direction dominated the more traditional channel through which

increases in economic activity generate a need for additional �nancial assets. Much of the

impact of money on economic activity seems to have been driven by relatively unpredictable

and permanent seigniorage issues by the colonial treasuries designed to fund government

expenditures such as wars. Direct government loans to colonists on land collateral, being

more predictable and temporary, appear to have a played a more stabilizing role in the

regional economy.

At a broader level, our study o¤ers support for the view that, in addition to causing some

in�ation, New England�s currency-issuing experiments facilitated the emergence of a modern

economic sector. This is reminiscent of the argument made in Rothenberg (1992, p. 99) for

the hundred years following 1750 whereby monetization helped to break down an archaic

system of market-places with �xed prices and opened Massachusetts to the advantages of

a market economy. To the extent that English imports re�ect such activity, our estimates

21

provide quanti�able historical evidence that the existence of a widely-accepted medium of

exchange, namely �at paper money, had put the process of monetization and economic mod-

ernization into motion half a century earlier. Though New England was to su¤er setbacks

with in�ation in the 1740s, a reluctant return to specie in the 1750s and 1760s, hyperin�a-

tion of the Continental during the Revolutionary War, and weak monetary regimes under

the Articles of Confederation, the process of monetization could not be repressed for long,

returning with renewed vigor as the Federal Constitution took e¤ect and the transition to

the dollar commenced (see Rousseau, 2006).

To be sure, we are focused on real growth along the extensive margin, which averaged 5.3

percent per year for English imports and 3.2 percent for the money stock. But our �ndings

must be reconciled with less dramatic growth on the intensive margin, which averaged 2.4

percent per year for per capita imports and 0.4 percent for both per capita exports and

the money stock. Indeed, the levels of per capita real money balances and English imports,

exports, and total trade were on average lower after 1730 than they were from 1703-29 (see

Rousseau, 2006, Fig. 5, p. 108). It may be, however, that evidence of modern per capita

growth in the monetary and commercial aggregates is simply too much to demand from a

region with nearly three percent annual growth in population! Rather, the ability of the

region to absorb a rapidly increasing labor force and keep it productive in its still largely

agrarian pursuits was perhaps a feat in itself, and one that bene�ted from having fewer real

resources diverted to the task of facilitating daily transactions.

Most scholarly work on money in the British North American colonies has focused on

how monetary emissions a¤ected the speed and extent to which the outstanding money stock

lost value. By departing from this literature to answer a broader question about the rela-

tionship between monetization and economic development, we conclude that it is important

to consider the bene�ts of �at paper money in addition to the costs when evaluating those

monetary regimes that employed it.

8. References

A Comparison between the British Sugar Colonies and New England, 1732. London.

Andrews, C.M., 1964. The Colonial Period of American History, Vol. 1-4. Yale University

Press, New Haven, CT.

22

Baxter, W. T., 1945. The House of Hancock: Business in Boston, 1724-1775. Harvard

University Press, Cambridge, MA.

Brewster, F., 1695. Essays on Trade and Navigation, in Five Parts. Printed for Tho.

Cockerill: London.

Brock, L.V., 1975. The Currency of the American Colonies 1700-1764. Arno Press, New

York.

Brock, L.V., 1992. The colonial currency, prices, and exchange rates. Essays in History

34, 70-132.

Canny, N., 1994. Europeans on the Move: Studies on European Migration, 1500-1800.

Oxford University Press, New York.

Carter, S.B., Gartner, S., Haines, M., Olmstead, A., Sutch, R., 2006. Historical Sta-

tistics of the United States: Earliest Times to the Present, Millennial Edition. Cambridge

University Press, New York.

Cole, A.C., 1938. Wholesale Commodity Prices in the United States, 1700-1861. Harvard

University Press, Cambridge, MA.

Davis, A. M., 1900. Currency and Banking in the Province of Massachusetts Bay.

Macmillan, New York.

Davis, A. M., 1911. Colonial Currency Reprints. Prince Society, Boston.

Engle, R.F., Granger, C.W.J., 1987. Co-integration and error correction: Representation,

estimation, and testing. Econometrica 55, 251-276.

Goldberg, D., 2011. Why was America�s �rst bank aborted? Journal of Economic History

71, 211-222.

Grubb, F., 2005. Two theories of money reconciled: The colonial puzzle revisited with

new evidence. NBER Working Paper No. 11784.

Hutchinson, T., 1936. History of the Colony and Province of Massachusetts Bay, 1764-

1828, Vol. 1-3. Harvard University Press, Cambridge, MA.

Johansen, S., 1991. Estimation and hypothesis testing of cointegration vectors in Gaussian

vector autoregression models. Econometrica 59, 1551-1580.

Lester, R., 1938. Currency issues to overcome depression in Pennsylvania, 1729 and 1729.

Journal of Political Economy 46, 324�375.

Lester, R., 1939. Currency issues to overcome depression In Delaware, New Jersey, New

York and Maryland 1715-1737. Journal of Political Economy 47, 182-217.

23

McCallum, B., 1992. Money and prices in colonial America, a new test of competing

theories. Journal of Political Economy 100, 143-161.

McCusker, J.H., 1970. The Rum Trade and the Balance of Payments of the Thirteen

Continental Colonies, 1650-1775. Ph.D. Dissertation, University of Pittsburgh, Pittsburgh,

PA.

McCusker, J.H.,1978. Money and Exchange in Europe and America, 1600-1775: A Hand-

book. University of North Carolina Press, Chapel Hill, NC.

McCusker, J.H., Menard R.R., 1985. The Economy of British America, 1607-1789. Uni-

versity of North Carolina Press, Chapel Hill, NC.

Michener, R., 1987. Fixed exchange rates and the quantity theory in colonial America.

Carnegie-Rochester Series on Public Policy 27, 233-307.

Middleton, R., 2002. Colonial America, A History: 1565-1776. Blackwell Publishing

Company, Malden, MA.

Nettles, C.P., 1934. The Money Supply of the American Colonies before 1720. University

of Wisconsin Press, Madison, Wisconsin.

Newell, M., 1998. From Dependency to Independence: Economic Revolution in Colonial

New England. Cornell University Press, Ithaca, NY.

O¢ cer, L. H., 2005. The quantity theory in New England 1703-1749: New data to

analyze an old question. Explorations in Economic History 41, 101-121.

Osterwald-Lenum, M., 1992. A note on the fractiles of the asymptotic distribution of

the maximum likelihood cointegration rank test statistics: Four cases. Oxford Bulletin of

Economics and Statistics 54, 461-478.

Rothenberg, W. B., 1992. From Market-Places to a Market Economy: The Transforma-

tion of Rural Massachusetts, 1750-1850. University of Chicago Press, Chicago, IL.

Rousseau, P. L., 2006. A common currency: early US monetary policy and the transition

to the dollar. Financial History Review 13, 97-122.

Rousseau, P. L., 2007. Backing, the quantity theory, and the transition to the US dollar,

1723-1850. American Economic Review

97, 266-270.

Rousseau, P. L., 2010. Monetary policy and the dollar. In: Irwin, D.A., Sylla, R. (Eds.),

Founding Choices: American Economic Policy in the 1790s. University of Chicago Press,

Chicago, pp. 121-149.

24

Rousseau, P. L., Sylla, R., 2005. Emerging �nancial markets and early US growth.

Explorations in Economic History 42, 1-26.

Rousseau, P. L., Wachtel, P., 1998. Financial intermediation and economic performance:

Historical evidence from �ve industrialized countries. Journal of Money, Credit and Banking

30, 657-678.

Smith, B.D., 1984. Money and in�ation in colonial Massachusetts. Federal Reserve Bank

of Minneapolis Review 8, 1-14

Smith, B. D., 1985. American colonial monetary regimes: The failure of the quantity

theory and some evidence in favor of an alternate view. Canadian Journal of Economics 3,

531-565.

Thomas Fitch Letterbook, 1702/3-11. American Antiquarian Society.

West, R. C., 1978. Money in the colonial American economy. Economic Inquiry 16, 1-15.

Woodman, W.C., 1975. Saugus Iron Works 1647. Saugus, MA,

http://www.asme.org/Communities/History/Landmarks/Saugus_Ironworks_1647.cfm.

25

Table 1Components of the real money supply and English imports, New England 1703-1749.

Seigniorage Loan bank Private Silver in Money Price Imports inbills bills notes circulation supply level 1700-02

Year £ tenor £ tenor £ tenor £ tenor £ tenor 1703=1 £ sterling

1703 6,400 0 0 173,848 180,248 1 59,6081704 17,700 0 0 167,834 185,534 0.980 74,8961705 29,500 0 0 160,329 189,829 0.946 62,5041706 31,100 0 0 153,798 184,898 0.064 57,0501707 40,800 0 0 144,399 185,199 1.093 120,6301708 57,000 0 0 130,725 187,725 1.066 115,5051709 69,400 0 0 114,802 184,202 1.146 120,3491710 112,800 0 0 96,414 209,214 1.158 106,3381711 92,900 50,000 0 83,337 226,237 1.053 137,4211712 188,900 25,000 0 70,260 284,160 1.131 128,1051713 219,500 0 0 57,676 277,176 1.281 120,7781714 146,400 50,000 0 45,092 241,492 1.383 121,2881715 174,500 80,000 0 32,508 287,008 1.166 164,6501716 71,000 170,000 0 19,925 260,925 1.126 121,1561717 135,800 175,000 0 7,340 318,140 1.212 132,0011718 145,300 163,500 0 0 308,800 1.358 131,8851719 138,500 152,000 0 0 290,500 1.377 125,3171720 126,300 150,500 0 0 276,800 1.256 128,7671721 68,600 239,000 0 0 307,600 1.223 114,5231722 126,000 237,500 0 0 363,500 1.321 133,7221723 150,700 236,000 0 0 386,700 1.340 176,4861724 181,200 234,500 0 0 415,700 1.432 168,5071725 213,400 233,000 0 0 446,400 1.501 201,7681726 266,400 211,500 0 0 477,900 1.713 200,8821727 269,800 190,000 0 0 459,800 1.654 187,2771728 256,400 266,000 0 0 522,400 1.647 194,5891729 265,300 242,000 0 0 507,300 1.852 161,1021730 277,000 218,000 0 0 495,000 1.886 208,1961731 285800 260,000 0 0 545,800 1.648 183,4661732 290,000 242,000 0 0 532,000 1.793 216,6001733 284,750 375,250 14,904 0 674,904 1.948 184,5701734 483,600 346,000 122,906 0 952,506 2.119 146,4601735 495,550 316,750 144,145 0 956,445 2.457 189,1251736 529,700 297,500 46,341 0 873,541 2.495 222,1581737 518,300 282,700 31,073 0 832,073 2.763 223,9231738 454,000 359,000 33,613 0 846,613 2.537 203,233

26

Table 1, continued.Components of the real money supply and English imports, New England 1703-1749.

Seigniorage Loan bank Private Silver in Money Price Imports inbills bills notes circulation supply level 1700-02

Year £ tenor £ tenor £ tenor £ tenor £ tenor 1703=1 £ sterling

1739 562,700 329,600 32,221 0 924,521 2.379 220,3781740 459,300 472,900 19,555 0 951,755 2.775 171,0811741 587,981 458,818 186,837 0 1,233,637 3.535 198,1461742 526,063 544,736 92,811 0 1,163,611 3.341 148,8991743 626,045 510,655 19,555 0 1,156,255 3.126 172,4611744 775,627 636,573 0 0 1,412,200 3.006 143,9821745 1,430,809 602,491 0 0 2,033,300 3.590 140,4631746 2,698,591 568,409 0 0 3,267,000 4.663 209,1771747 3,319,772 534,327 0 0 3,854,100 6.165 210,6401748 3,540,554 494,245 0 0 4,034,800 8.580 197,6821749 3,589,536 444,164 0 0 4,033,700 8.129 238,286

Sources. Seigniorage and land bank bills are from Historical Statistics Series Eg302-Eg305 and Brock (1975, pp 23, 26, 38, 44-45, 47-49). See Section 4 of text for detailson construction. Private notes and silver in circulation are from O¢ cer (2005, pp. 114--117). The money supply is the sum of seignorage, land bank and private notes, andsilver in circulation. See Section 4 for derivation of our price level index from wheatand molasses prices from Cole (1938, Table 36, p. 117) and cod�sh prices fromHistorical Statistics (Vol. 5, Series Eg248, p. 675). Imports from England (in constant1700-02 sterling) are from Historical Statistics (Vol. 5, Series Eg437, pp. 710-11).

27

Table 2Unit root and cointegration tests for M

Pand imports (Y ).

Unit root tests Johansen testsMaximum

ADF Trace eigenvaluestatistic � 0 � 1 � 0 � 1

Y -2.75*

MP

-0.60 10.627 0.625 10.002 0.625

SP

-1.45 10.368 2.128 6.632 2.128

LBP

-2.20 22.130* 10.502* 11.628 10.502*

Note. The variable Y is English imports. M , S, and LBare the money stock, seigniorage bills, and loan bank bills,respectively. All variables are in log levels. The leftpanel reports ADF test statistics for the null hypothesisof a unit root using a model with constant, time trendand two additional lag di¤erences. The right panelpresents Johansen test statistics for cointegration betweenthe row series and English imports. The column labeled�� 0�tests the null of no cointegration while the rowlabeled �� 1�tests the null of a single cointegratingvector. The Johansen testsuse three lags in the underlyingVAR and include an unrestricted intercept. Critical valuesare from Table 1 in Osterwald-Lenum (1992). * denotesstatistical signi�cance at the 1% level or less.

28

Table 3Estimates from vector error correction models, New England, 1703-1749.

Total money supply Seigniorage bills Loan bank bills4Yt 4

�MP

�t

4Yt 4�SP

�t

4Yt 4�LBP

�t

ECTt�1 -0.436*** 0.074 -0.563*** -0.168 -0.021* -0.352**(0.123) (0.104) (0.167) (0.292) (0.011) (0.139)

4Yt�1 -0.186 -0.178 -0.108 0.032 -0.413*** -0.565(0.145) (0.123) (0.167) (0.293) (0.151) (1.836)

4Yt�2 -0.115 -0.220* -0.056 -0.031 -0.197 -2.419(0.133) (0.113) (0.150) (0.263) (0.149) (1.817)

4(MP)t�1 -0.423** -0.021

(0.179) (0.152)

4(MPI)t�2 -0.002 -0.271

(0.199) (0.169)

4( SP)t�1 -0.157 -0.328*

(0.100) (0.176)

4( SP)t�2 -0.063 0.054

(0.093) (0.163)

4(LBP)t�1 0.008 -0.387***

(0.011) (0.134)

4(LBP)t�2 -0.015 -0.444***

(0.011) (0.132)

R2 0.422 0.186 0.375 0.154 0.293 0.438

Durbin-Watson 2.148 2.069 1.932 1.964 2.111 2.118

Coint. loading [a, b] [1, -0.583] [1, -0.276] [1, 0.531 ]

Note. The variable Y is English imports. M , S, and LB are the money stock, seignioragebills, and loan bank bills, respectively. All variables are in log levels. The �rst two columnsreport estimates for equations (4) and (5) using the total money supply. The third and fourthcolumns report estimates using only seigniorage or land bank bills. Standard errors are inparentheses. *, **, and *** denote statistical signi�cance at the 10%, 5% and 1% levels.

29

Table 4Estimates from vector autoregressive models, New England, 1703-1749.

Yt�1 Yt�2 Yt�3MP t�1

MP t�2

MP t�3 R2

Panel A: Total money supply (M)

English imports (Y ) 0.379*** 0.070 0.118 -0.179 0.427* -0.017 0.70(0.142) (0.144) (0.124) (0.168) (0.237) (0.195)

F-test 24.35 9.311P-value (0.000) (0.025)

MP

-0.0978 -0.044 0.229** 0.908*** -0.235 0.222 0.87(0.120) (0.121) (0.105) (0.142) (0.200) (0.164)

F-test 5.605 101.92P-value (0.132) (0.000)

Panel B: Seigniorage bills (S)

English imports (Y ) 0.332** 0.053 0.059 -0.007 0.090 0.063 0.67(0.152) (0.156) (0.141) (0.088) (0.104) (0.086)

F-test 7.56 5.221P-value (0.006) (0.156)

SP

-0.0878 -0.030 0.091 0.619*** 0.330* -0.057 0.82(0.258) (0.265) (0.239) (0.150) (0.177) (0.146)

F-test 0.249 60.50P-value (0.969) (0.000)

Panel C: Loan bank bills (LB)

English imports (Y ) 0.457*** 0.121 0.079 0.004 -0.017 0.020** 0.67(0.141) (0.152) (0.140) (0.010) (0.011) (0.010)

F-test 23.53 5.018P-value (0.000) (0.170)

LBP

0.120 -0.948 3.53** 0.355*** -0.109 0.383*** 0.76(1.761) (1.891) (1.742) (0.128) (0.140) (0.123)

F-test 5.074 26.99P value (0.166) (0.000)Note. The table includes results for bi-variate VARs with three lags. The rows ineach panel correspond to equations (6) and (7). The dependent variable for eachequation is listed in the left column. All variables are in log levels. Coe¢ cientestimates appear in the columns for the independent variables, given by thecolumn labels, with standard errors in parentheses. The rows labeled �F-test�report test statistics for block exogeneity and the rows labeled �P-value�reportcorresponding tail probabilities for the null hypothesis that the lags of each systemvariable are jointly zero. *, ** and *** denote statistical signi�cance at the10%, 5%, and 1% levels, respectively.

Table 5Robustness of vector error correction results to modeling choices.

4Yt 4�MP

�t

4Yt 4�MP

�t

4Yt 4�MP

�t

Panel A: Alternative price de�ators1) St. exchange rate 2) English prices 3) Two goods only

ECTt�1 -0.304*** 0.017 -0.328*** 0.091 -0.355*** 0.108(0.087) (0.058) (0.101) (0.088) (0.109) (0.086)

Durbin-Watson 2.086 1.957 2.120 1.919 2.054 1.970

Coint loading [a, b] [1, -0.380] [1, -0.591] [1,-0.452]

Panel B: Alternative sample periods1) 1703-1742 2) 1713-1749 3) 1713-1742

ECTt�1 -0.484*** 0.306** -0.442*** -0.026 -0.615** 0.333(0.172) (0.121) (0.169) (0.167) (0.253) (0.227)

Durbin-Watson 2.020 2.093 1.972 2.075 1.844 1.981

Coint loading [a, b] [1, -1.025] [1, -0.557] [1, -0.953]

Panel C: Alternative proxies for Y (including Scotland)1) Imports 2) Exports 3) I +X

ECTt�1 -0.468*** 0.094 -0.298** 0.051 -0.336*** 0.124(0.126) (0.109) (0.117) (0.083) (0.100) (0.105)

Durbin-Watson 2.166 2.070 2.033 2.059 2.224 2.155

Coint loading [a, b] [1, -0.659] [1, -0.063] [1, -0.465]

Panel D: Components of the money supply1) Paper money 2) Seigniorage bills 3) Loan bank bills

1718-49 1713-49 1713-49

ECTt�1 -0.452** -0.061 -0.356** 0.0367 -0.035 -1.142**(0.187) (0.167) (0.173) (0.312) (0.049) (0.467)

Durbin-Watson 2.011 1.644 1.978 2.102 1.959 1.522

Coint loading [a, b] [1, -0.457] [1, -0.335] [1, 0.286]

Note. See text for description of these regressions and the data items used.

Fig. 1. Impulse response functions, New England, 1703-1749.

E¤ect of MPon Y E¤ect of Y on M

P

Panel A: Total money stock

Panel B: Seigniorage bills

Panel C: Loan bank bills

Note: The �gures plot median annual responses of real money and English imports (Y) toone standard deviation shocks to each other constructed from the VAR models with threelags reported in Table 4. The dashed lines are 90 percent con�dence bands.

33