The antebellum tariff on cotton textiles revisited

MIT LIBRARIES 3

9080

01917

6632

'f

"'i;r

mmimimmmmmi^

/(feiiifilif

ppli

piiiliSBii

HI'mJij/ii! I;'!!};;;;«if//;;/!'ii j,

-It

Digitized

by

the

Internet

Archive

in

2011

with

funding

from

Boston

Library

Consortium

IVIember

Libraries

DEWEY

Massachusetts

Institute

of

Technology

Department

of

Economics

Working

Paper

Series

THE

ANTEBELLUM

TARIFF

ON

COTTON

TEXTILES

REVISITED

Douglas

A.

Irwin

Peter

Temin

Working

Paper

00-19

September

2000

Room

E52-251

50

Memorial

Drive

Cambridge,

MA

02142

This

paper

can

be

downloaded

witiiout

charge from

the

Social

Science

Research Network Paper

Collection

at

,; MASSACHJSETTSINSTITUTE

OFTECHNOLOGY

OCT

3 2000

L.„„„._„„,

Massachusetts

Institute

of

Technology

Department

of

Economics

Working

Paper

Series

THE

ANTEBELLUM

TARIFF

ON

COTTON

TEXTILES

REVISITED

Douglas

A.

Irwin

Peter

Temin

Working

Paper

00-19

September

2000

Room

E52-251

50

Memorial

Drive

Cambridge,

MA

02142

This

paper

can

be

downloaded

without

charge from

the

Social

Science

Research Network Paper

Collection

at

The

Antebellum

Tariff

on Cotton

Textiles Revisited

Douglas

A. Irwin

Department

of

Economics

6106

Rockefeller Hall

Dartmouth

College Hanover,

NH

03755

and

NBER

Peter

Temin

Department

of

Economics

50

Memorial

Drive

MIT

Cambndge,

MA

02142

and

NBER

July 20,

2000

Abstract

Recentresearch has suggestedthat the antebellum U.S. cotton textile industry

would

have been

wiped

outhad it not receivedtariffprotection.

We

reaffirmTaussig's

judgment

thatthe U.S. cottontextile industry

was

largelyindependentofthe tariff

by

the 1830s.

American

andBritish

producers specialized inquite different types oftextileproducts that

were

poorsubstitutes for

one another.

The

Walker

tariffof 1846, forexample, reduced theduties on cotton textiles

from

nearly

70

percentto 25percent and imports soared as aresult, butthere

was

little change in

domestic production.

Using

data

from 1826

to 1860,

we

estimate the responsiveness ofdomestic

productionto fluctuations in import prices and concludethat the industrycould have survived

even ifthetariffhad been completelyeliminated.

Acknowledgments

:

We

thankparticipants at the

DAE

meeting ofthe

NBER

Summer

Institute for

-1-1. Introduction

America's industrializationin the earlynineteenth century

was

spearheaded

by

the

New

England

cotton textile industry.

The

industry received astrong initial impetus

from

the

Jeffersonian

embargo

and thewar-time disruptionsto

commerce

from

1807 to 1815. Afterthe

war, the industrypressed forandreceivedhigh tariffs toblockimport competition and domestic

outputcontinuedto expand.

The

extentto

which

theindustry

depended upon

thetariff, however, has been asource of

debate. Taussig (1931, pp. 135-36)opined that"the tariffof

1816

may

beconsideredajudicious

application ofthe pnnciple of protection to

young

industries" inpartbecause domestic producers

became

strong

enough

to survive withouttariffs "almost certainly

by

1832."

By

thistime,

Taussig concluded, "the cotton manufacture

was

inthe

main

independent ofprotection, andnot

likely to be

much

affected, favorably or unfavorably,by changes in duties."

Bils (1984) and Harley (1992)recentlychallenged Taussig's view.

They

went beyond

Taussig's informedifinformal discussion

by

explicitlyconsideringthe costs ofproducing cotton

textiles in

America

andBritain in the 1830s and 1840s. Bilsconcludedthat, even after

two

decades ofprotection, the domestic industry

"was

stillunable tostand

on

its

own.

. . asof 1833,

removing

protection

would

have eliminatedthe vastmajority of value

added

in the cottontextile

industry."

With

additional cost evidence, drawingprincipally

on

Montgomery

(1840), Harley

concurred and conjectured thatremoval ofthe tariff

would

have shrunk domestic output to about

10 percent ofdomestic

consumption

even as late as the 1850s.

One

problem

withthe Bils and

after all dependson comparative and not absolute advantage, andprovide littleinformation on

the characteristics of

American

andBritishfabrics, akeyconsideration indetermining theimpact

ofthe tariff.'

This paper

examines what

previous researchershave largely ignored

-

namely, the

historical recordof

how

domestic production responded tochanges in the relative price of

imports. Rather thaninferringthe impact ofthetariff

from

cost data,

we

allow time series data

on

prices, domesticproduction, andimports from 1826 to 1860 tospeak on the issueofthe

tariff's importance.

These

data allow us toestimate the sensitivityofdomestic outputto changes

in the relative price of imports, whetherdue to changesin the prices ofgoods,the exchangerate,

or the tariff rate.

We

conclude thatthe findings ofBils and Harleyexaggerate the importance ofthe tariff

after 1830. Rather,

American

andBritish cotton textile producers specialized in quite different

varieties of cotton goods, and domestic producers wereinsulated

from

foreigncompetition

by

the

different characteristics oftheirproducts.

The

relative unimportance ofthe tariffis demonstrated

by

the reduction inthe cotton textileduties in

1846 from

nearly

70

percent to 25 percent.

Imports soared

by

a factorofthree and increasedtheirshare oftheU.S. market

from

about 7

percent to about 15 percent, but there

was no

declinein domestic output. This simpleexperiment

suggests thatU.S. cotton textile producers

were

not dependent

upon

the tariffand thus supports Taussig's view that the industry

was

well established

by

thistime.

'

The

cost estimates in

Montgomery

(1840) also were disputed at thetime. SeeJustitia

(1841). If

we

followed the approach offocusing on productioncosts today,

we

would

conclude

that the United Slates

would

not import automobiles from

Germany

(such as

Mercedes

and

-3-2.

Background

For

most

ofthe earlynineteenth century, theU.S. tariffon imported cotton cloth

was

a

combmation

of an ad

valorem

rate and a

minimum

valuation.

The

tariffof 1816, forexample,

generallyregardedas thefirst"protective" U.S.tariff, consisted ofa

20

percent ad valoremrate

on importedtextiles along witha 25 centperyard

minimum

valuation. This

scheme

severely

burdened textile importsthat

were

pricedunder25 cents per yard andthus shiftedthe

composition ofimports

toward

highergrade, higher priced products.

The

minimum

valuation

may

not have been binding

on

British products, however, because goods worth less than 25 cents

per yard were nottypical ofBritish products duringthistime (Stettler 1970, p. 212). Taussig

commented

onlythatthe

minimum

excluded coarsercloths; it also excludedAsian cloth while

not affectingimports

from

Britain.

This

was

a deliberatestrategy

by

FrancisLowell.

As

explained

by

hiscolleague,

Nathan

Appleton, the

minimum

was

designed toprotectedthefledgling industryin

New

England

without antagonizing Southerncotton exporters toEngland. Cotton growers in the South

exported

raw

cottonto Britain,

and

they

were opposed

toanytariffthat

would

restrictthe sales of

theirBritishcustomers to theUnited States.

They

worried both about lossofsales and about

further losses dueto possible British retaliation. Lowell's tariffdesign

shows

thatthe sectional

conflictover thetariffthat

would loom

large atmid-century

was

alreadypresent atthe startof

New

England

industrialization

(Temin

2000).

After 1816, however, the

minimum

valuation

became

increasinglybinding asBritish

exportpricesoftextiles fell steadily andasthe

minimum

valuation

was

graduallyraised in

-4-period.

The

minimum

valuation, forexample,

was

raised

from

25 centsto

30

centsin the tariff

of 1824, andthe ad valorem tariff

was

increased

from 20

percent to25 percent. In 1828the

minimum

valuation

was

raised againto 35 cents.

By

the 1840s, the averageBntish export price

to the United States

was

around 13cents peryard, yet the

minimum

valuation on printedcloth

was

30cents peryard. This

minimum

required that the 13 cent per yardimport

would

be

assessedat

30

cents; applying the30percentdutyto that price implies a charge of9cents per

yard,

amounting

to an ad valorem equivalent ofover

60

percent."^

The

most

important change

after thisperiod

was

the

Walker

tariffof 1846,

which

completelyeliminated the

minimum

valuation andcut the ad valoremtariffto 25 percent.

To

determine the effective ad valoremrate

on

imports,the prices ofthe imported goods

must

be ascertained.

While

U.S. trade statistics duringthisperiodreport onlythetotal value of

cotton imports (with

some

category breakdown, by country), Britainpublished statistics on the

value and

volume

ofits cottontextile exports to various markets, includingthe UnitedStates.

These

data allow the unit value ofBritish textileexports to becalculated, a useful

summary

statisticofthe averageexport priceto the United States

and

to othermarkets.-'

Both

price series

^

_As

_{Table 1 indicates, theTariff of 1842 had a}

_minimum

_{valuation of}

₃₀

_{cents per yard}

ofprinted cloth and

20

cents per yard ofwhite cloth. Accordingto U.S. import statistics, the

value ofprinted, stained,or colored cottonmanufactures

from

Britain vastly exceededthat of

white or uncoloredcotton manufactures. See, forexample, U.S.

House

of Representatives,

"Commerce

and Navigation,"

House

Executive

Document

No. 42, 13'^ Congress, 2"*^_Session (Washington, D.C.:

GPO,

1849),p. 150.

^

_The

_{sourcesforthe dataare GreatBritain,}

_House

_of

Commons

_(1847-48),

p. 325,

House

of

Commons

(1951), p. 261,

House

of

Commons

(1856), p. 385, and

Helm

(1869), p. 435.

These

data

were

originally utilizedby Stettler(1970, pp. 137, 168, 216-217), but

we

found

two

apparent errors inhis series on theaverage prices ofBritish textileexports to the UnitedStates. First, he

seems

to havetaken the Bntish export quantity

from

the

column

marked

"cottons" and

show

a steady decline duringthe 1815 to

1860

period, butprices are

somewhat

higheron goods

destinedfortheU.S. market, consistentwith the effectofthe

minimum

valuation in shiftingthe

composition of imports toward finer, highervalued goods. Figure 1 presentsourestimate ofthe

ad valorem equivalent importtariffon British textiles,

which

iscalculated

by

applying thetariff

rates in Table 1 to the exportprice series.

As

the

minimum

valuations rose and British export

pricesfell, the effectivetariffrose tonearly

70

percent

by

the early 1840s.

The

Walker

tariffof 1846 appearedto pull the rug out

from

under domesticproducers

by

eliminating the

minimum

valuation andapplying a simple 25 percent ad valorem tariff.

As

Figure 1 strikingly demonstrates,the effectivetariff

plummeted from 69

percenttojust

25 percent. Imports

from

Britain nearlytripled, soaring

from

37 million yardsin 1846 to 105

million yards in 1847, andthen falling back to 71 million yards in 1848 (Stettler 1970, p. 168).

Yet

what happened

to domestic production? According to theDavis andStettler(1966, p. 221)

series on

New

England

textile output, presented inFigure 2, production stalledbriefly

m

1847

butthen

resumed

its growth. Despite the sharp tariffreduction, there

was no

absolute decline in

output. Figure 3

shows

that theimport market share

jumped

from roughly 5 percent in 1846 to

about 15 percent in 1847 andthen climbedto around

20

percent

by

the mid-1850s.

But

this

growth in imports occurred without a decline in domestic production.

The

uniformity ofthe growth in

New

England

textile outputinFigure 2 is striking.

The

tariff

may

have affected the level of

American

textileproduction to a slight extent, butit doesnot

value ofall export categories (including cotton thread, twist, and yam).

While

the otherexport

categories are

much

less important than "cottons," it doesraise thecalculated averageprice and

thusreduce the calculatedaveragetariff. Second, even usingthe

columns

that Stettlerselected,

the unit value isnot

what

hereports itto bedue toan apparent arithmetic error forthe years prior

-6-seem

to have affectedits growth.

As

Taussig (1931,p. 140) noted,

"when

the actof 1846

was

passed, the protectionistspredicted disaster; butdisaster

came

not, eitherforthe countryatlarge

orforthecotton industry" asproduction steadily increased.

3. Estimatingthe

Impact

ofTariff

Changes

To

explore further theimportance ofthetarifftothe domesticcotton industry,

we

employ

a simplereduced-form model, based on

Grossman

(1986), to determinethe effectsofchanges in

the relative price ofimports on domesticoutput. In this model, the priceofdomestic goodsis

endogenous, as is thequantity produced,

and

this enables us toestimate the effectofthetariffon domestic production without

assuming

that the domestic price

was

unaffected

by

the tariff. This

framework

enables us to

examine

thedatain a

more

systematic way.

On

the supply side,

we

assume

that cotton textiles areproduced with

raw

cotton (C),

labor (L), andcapital (K).

The

production function takes thefollowing

Cobb-Douglas

form:

(1)

Q

=

Ae'"C^'L^^K^\

where

_Q

is the quantityoftextileproduction,

n

isthe rate ofHicksneutral technological

progress, and t representstime."*

Cotton is

assumed

to be atraded input andavailable to the industryat the

exogenous

price _Pc and laboris also

assumed

tobe supplied attheexogenous pricew.

The

quantityof

cotton usedand labor

employed

isfound

by

settingthe marginal value productofeach equal to

its price:

"*

While

the

Cobb-Douglas

specification imposes certain restrictions onthe coefficients, the estimatedreduced

form

is

more

general and doesnot implyany restrictions on the

(2)

C

=

(a,pQ/pc),

(3)

L

=

(a.pQ/w),

where

p is the priceoftextileoutput, _p^ is the priceofcotton, and

w

is thewage. Capital is a

non-traded factor

whose

supply

grows

atthe

exogenous

trend rate (including depreciation) of5

percentperyear:

K

=

Ke^'

On

the

demand

side, domestic textilessubstituteimperfectly with foreign-produced

textiles,

which

is perfectly elastic in supply and can be imported at

exogenous

price p*.

Domestic

textiles also substitute imperfectlywith the aggregate basket ofdomestic goods so that

demand

ischaracterized as:

(4)

Q

=

Be^"[Ep*(l-hT)/p]''[pVp]''-Y",

where

\\i isthe rate of secular

demand

shift,

E

is theexchange rate, p* is the (foreign)import

priceoftextiles, tisthe ad valorem tariffrate

on

importedtextiles,_p^ is the price ofthe

aggregate basket ofdomestic goods,

and

Y

is real national income.

The

domestic price and production oftextiles areboth

endogenous

variables, and

domestic production isthe variable that

we

are

most

interestedin explaining. Therefore, solving

out for the reduced

form

equation

and

taking logs (and suppressing time subscripts)

we

have:

(5) log

Q

=

tto H- a, log (Ep*(l-f-T)/pcw) -i-Ojlog (Pa/PcW)

+

aj log

Y

+

₀₄1 -h s.

Thus, domestic textile production is an increasing function ofthe tariff-inclusive importprice,

theprice of aggregate goods, and national income, and adecreasing function oftheprice of

cotton andthe

wages

paidin the textile industry.

We

would

expecta^

>

0, a2

>

0, ttj

>

0, while

tto and a^could beofeither sign. In thisequation, thedegree ofimport competitionis fully

-8-does notappearbecause import

volume

is

endogenous

todevelopmentsin theUnited Statesand

elsewhere.

The

domesticprice oftextiles also doesnot appearbecause it too is an

endogenous

variablethat is represented

by

the

exogenous

priceof cotton and

wage

rate in the industry.

This equationis estimatedusing annual data

from

1826 to 1860.

The

quantity of

domestictextile production (in yards)

comes

from

Davis and Stettler(1966, p. 221). This series

is theaggregate output oftextiles in

New

England,but

New

England

accountedforover

70

percent oftotalU.S. output aroundthis time

(Temin

2000, p. 122).

New

England

alsocontained

virtually all ofthe large-scalecotton production inthecountrythathas been thefocus of

historical attention.

The

U.S. import price ofBritish textiles istaken as theexportunit value, as

described above.

The

tariffrate is based onour calculation in Figure 1.

We

ignoretransportation

costs,

which

as Harley (1992, p. 566) noted

would

add lessthan halfacent per yardto the landed

cost ofBritish cloth; any changes intransportation costs are picked

up by

the time trend orare

relegatedtothe eiTortemi.

The

dollar-poundsterling

exchange

rateis thatofthe 60-daybills of

exchange

provided

by

Lawrence

Officer(forpublication inthemillennial edition of Historical

Statistics_ofthe UnitedStates). U.S. laborcosts intextileproduction isthe average annual earnings perfull-time

worker from

Layer (1955, pp.46-47),

which

is highlycorrelated with the

lesscomplete

wage

series in Zevin (1971).

The

aggregate price index isthe

consumer

price

index

from

David

andSolar (1977) andreal U.S.

GNP

is

from

Berry (1988).

Table 2 presentsthe estimationresults.

The

first

column

focuses onthe

OLS

results.

The

coefficient a, is estimated tobe 1.52,

which

indicates that a ten percentreduction in the relative

price ofimports

would

be expectedto reducedomestic output

by

about 15 percent. Thisis a very

-9-from 69

percentto 25 percentdue largelyto the elimination ofthe

minimum

valuations. This

would

beequivalent to a26percent decline in the priceofimported textiles, calculated as

(1.25-1.69)/1.69=-0.264.

Such

a changein relative prices

would

implya

40

percent declinein

domestic output,

nowhere

near

what

was

actuallyobserved duringthisperiod. Inaddition, the

coefficient a2 carried the

wrong

sign andperversely impliesthat a decreasein the relative priceof

other

consumer

goods

would

increase domestic textile output.

The

coefficient

on

the logof

income

plausiblyimplies that

demand

is

income

elastic.

However,

the

Lagrange

multipliertest statisticfor first-orderserial correlationis

significant. Serial correlation impliesthat

OLS

is notefficient, thatthe coefficients

may

be

biased, and thetest statistics are invalid.

Column

₍₂₎presents the generalleast squares

(GLS)

estimates thatcorrect forserial correlation

by

allowing8,

=

ps,.i

+

e[.

These

estimates are not unbiasedbut are consistent and asymptotically

more

efficient than

OLS.

When

this is done, the

coefficient on the relative priceofimportsfalls to 0.19. It also is estimated

more

precisely,

allowingus torule out thehypothesis thatthe coefficientis large. It isnotsignificantly different

from

zero (noeffect ofthetariff) andis significantlylowerthan one (proportional effect ofthe

tanff).

Column

(3) presents the results for afirst difference specification,

which

eliminate

most

oftheserial correlation whilepreserving the

OLS

standarderrors and test statistics. In this case,

the coefficient

on

the relativeprice ofimportsis 0.27, similartothe

GLS

estimate, andalso

insignificandydifferent

from

zero. If

we

takean elasticityaround0.25 asbeing ourcentral

estimate, thenthe

Walker

tariffwith its

26

percentreduction in the relative price ofimports

-10-we

cannotrule outthat the effectis zero.

This

modest

effectis

more

consistentwith the observed growth indomestic output of 1

percentin 1847.

What

accounts for thedifference between theestimatedandthe actual change

inoutput? In

column

(2), the key determinantofdomestic outputis simplythe timetrend,

which

indicatesthat output

would

increase4 percenta year, on average, holding otherfactors fixed.

Similarly, in

column

(3), the coefficient on time (here aconstant in thefirst difference

specification) indicates an 10 percent annual increasein domesticoutput. Regardless ofthe

relative price of imports, or even changes innational income, there

were

strongfactors pushing

American

outputhighereveryyear.^ Therefore, the effectofthe tariff-inducedchangeinthe

relative price

would

be almost completelyoffset

by

the trend increase in domestic production.

What

would

bethe impact ofacompletetariffelimination? In thiscase the relative price

ofimports

would

fall 41 percent.

Given

an elasticity ofabout 0.25,this

would

translate into a 10 percent decline in domestic output. This, in

some

sense, is the worst case scenario

from

the

standpoint ofdomestic producers.

While

imports

would

surge as a result, and theimport market

share

would

increase dramatically, the level ofdomestic output

would

not fall substantially. This

result is alsoquite different

from

Harley,

who

conjectured that free trade

would

have reduced domesticoutput

by 90

percent, that it

would

have reduced the U.S. share ofthe marketto about

10percent ofconsumption.

Do

we

actuallyobserve such largechanges in relativeprices?

The

answer

is yes: inthe

1840sthe relativeprice ofimports fell

by

nearly

50

percent. Thus, such largechanges are partof

thetime series datawith v^'hich theestimated response is calculated.

-11-An

alternative approach to solving outthe

endogenous

variablesand estimatingareduced

form

isto accountfor the

endogenous

variables

by

two-stage leastsquares estimation. This approach

would

take

Ep*(l+T)

as

exogenous

but treatthe domestic price oftextiles (p) as

endogenous

and use asinstruments such variables as the price of cotton andthedomestic

wage

rate in the textileindustry.

The

domestic textile priceis that ofRussian

brown

sheeting, in

New

York, taken

from

Zevin (1971,p. 134); the results are essentially

unchanged

if

we

usethe

alternativecotton sheeting price series in theU.S.

Bureau

of

Census

(1975, seriesE-128).

Table 3 presents

some

econometric results

from

this specification.

The

first is a simple

2SLS

regression

which

yields anelasticity ofdomestic output with respectto the relative price of

imports ofabout0.6. This implies thatthe

Walker

tariffof1846

would

reduce domestic output

by

16 percent. Yet once again this specification is afflictedwith serial correlation.

Column

(2)

and

column

(3)present theinstrumented

GLS

and

first difference specifications that yieldan

elasticity (againstatistically insignificantly different

from

zero) ofabout 0.05. Thiselasticity

impliesthatthe

Walker

tariff

would

reducedomestic output

by

just 1 percent,

which

is very

small, butperhaps accurate in lightofthe actual historical experience.

Due

to the smallness ofthe estimatedelasticity,

we

cannot ruleout acase in

which

changes in the relative price ofimports

would

have

no

effect on domesticproduction. In results

we

do

notreport, however,

we

find that changes in the relative priceofimports didhave a large

and statisticallysignificant effect

on

the

volume

of imports. (Theestimated elasticityis around

2.5.) This impliesthatthetariff

was

quitesuccessful atreducing import volume, butthat a

significant increaseinimport

volume would

nothave detracted

much

from

the

demand

for

-12-2 and3 (as

amounts

andas sharesofU. S. consumption).

But

the growthofU. S. production,

also

shown

inFigure 2, continuedunabated.

As

the nextsectionpointsout,this isbecause the

product

mix

ofdomestic andforeignproducers

was

quite different, implying thatthere

were

limited opportunitiesto substitute theproducts foroneother in consumption.

4. Interpretingthe Results

The

results oftheprevioussection supportthe conclusion that atariffreductionreduced

the relative price of imports,and that this reduction, while stimulatinggreaterimports,had onlya

modest

effecton domestic production. That import competition had a

minor

effecton domestic

production is supported

by

our observation ofwhat

happened

in 1846

when

the cotton duties

were

cut drastically, imports surged, and domestic outputremained high.

There is

no

contradiction inthefinding that U.S. imports of cotton textiles aresensitive to

the relative price ofimports whereas domesticproduction isnot. This apparentparadox merely

suggests thatdomesticallyproducedtextiles and importedtextiles

were

different products that

were

far

from

perfect substitutes forone another. Britain specialized in finercotton goods while

America

specialized in heavier, standardcloths.

As

Zevin (1971,pp. 126-127)noted, "imports

from

Britain and theproducts of

New

England

mills tended to fall into quite distinctproduct

classifications. . .

The

imports

were

largelyginghams,

woven

in intricatepatterns to

which

the

power looms had

yetbeen adopted.

New

England

power looms were

supplyingplain

weaves

—

sheeting, shirting, and,

somewhat

later, twills

—

usually

made

oflower count

yams

than the

British cloths." In fact,the

mean

countof

yam

spun in

England was

over50in theearly 1830s,

-13-counts spun in the

two

countries: over

90

percent oftextileproduction in Lancashire

was above

20

count, while

more

than 80 percentof outputin

New

York

was below 20

countat thistime

(Temin

1988, p. 896).

As

aresult, any growth inimports did not necessarily

come

atthe expense

ofdomestic production.

Thisdoes not

mean

that theU.S. industry

was

completely unaffected

by

the surgein

imports

from

Britain as aresultofthe

1846

tariffreduction.

The

tariffaffected the

mix

of traded

goods

atthemargin,forcing

some

domestic adjustments.

Domestic

producers ofhigherquality,

highercount goods wereforced to adapt tothe

new

competition

by

shifting theirproduct

mix

toward lower count, lowerquality goods. Stettler(1970, p. 224) reportedthattheratio of

low

count to high count yards produced

by

his sample of firms rose

from

2.07in 1843-44to 2.58 in

1847-48.

He

notedthathigh countmillsincreasedtheiroutput rapidly in 1843-44 (afterthe tariff

increase in 1842) and

much

lessrapidlyin 1847-48 (afterthetariffreduction in 1846). But, for

the industry as a whole, theimports triggered only

minor

adjustmentsbecause

most

U.S. firms

were

notproducingfinergoods.^

Fortheirpart, British producers did not shift

away

from

specializingin finerproductsand

begin producingcoarserproductsjustbecause the

American

market

opened

up. (The United

States only took about 10 percent ofBritain's textileexportsduringthis period.) Britain

*" _There

may

_{have been}

a slightlygreaterconcentration offiner

good

producersin

Pennsylvania,

which

would

have implied

some

regional variation in theadjustmenttothelower

tariff.

The

number

ofyardsperpound, an indicationofthe cloth quality,

was

similarin all states in 1831, except forPennsylvania

(Temin

1988, p. 895). There

was

also a slightfall in

Pennsylvania's share of U.S. cotton manufactures between the

1840

and

1850

censuses. But

Scranton's account ofthe textileindustryin Philadelphiadoes not assign a large role to thetariff there.

One

prominent cotton mill closed atthe

end

ofthe 1840s, but its

owner

died, and itis hard

-14-apparently did attemptto export

some

plainer cloth,but the venture did notsucceed/

The

price

of cotton textiles exportedtotherest ofthe world

was

about 5 centslowerthanthose exported to

the United Statesin the late 1830s, but

were

only about 2cents cheaperin the late 1840s. This

indicates thatBritain

was

notexportingto othermarkets large quantities oflower count goods

that

American

producers had specialized in. Therefore the scope for British producersto adjust

theirexport

mix

totheUnited States

by

shifting

down

tolower count goods

was

limited. Finally,

although textile imports

from China

andIndiaroseafter the 1846 tariffreduction, the

amounts

were

very small in comparison to imports

from

Britain.

Our

conclusion also isin accord with theprevioushistorical literature,

which

was

skeptical ofthe tariffsimportance tothe industry.

Ware

(1931, p. 106)concluded that coarse

goods "neverreallyneeded protection, and since 1833 even high tariffadvocates had admitted

thattheycould stand alone, forordinary 'domestics' could be

made

sufficientlycheaplyin the

UnitedStates to completesuccessfully with the producerof other countriesin foreign markets."

One

contemporary observed in 1850that "thebusiness of cotton manufactures

was by

thattime

so firmly established as tobe little affected by changesin legislationin regardtoduties on the

coarserfabrics requiredfordomestic consumption, to

which American

machinery

had

been

adapted" (quoted in

Copeland

1912, p. 15).

This conclusion

had

been the consensus for

many

decadesuntil the recent

work

ofBils

andHarley.

As

discussedin the introduction,using cost datato determine thecompetitive

^

_Stanwood

_(1903,

II, pp. 90-91)writes that "it is admitted thatthe

new

tariffcaused little

injuryto cotton manufactures . . . large quantities ofplain cloth

were

importedin 1847, but the

quality

was

sodistinctly inferiorto thatofthedomestic goods that the venture resulted in a loss.

In fact,except for fine goods, laces,embroidery, and the like, the

American

manufacturer

had

the

-15-position ofthedomestic industrycan be problematic. Indeed, afterexamining the

Montgomery

costdatathat Harley relies on,Batchelder (1863,p. 91)

conceded

that costs appearedto be lower

in Britain than in

American

foreverything but

motor

power, yet still concluded that "it is

questionable whether

heavy

goods, such as drillmgand sheeting,

which

make

up

averylarge

proportion ofthe

consumption

ofthiscountry, can beproduced cheaperthan inthe United

States."

5.

The

Tariff of

1816

We

have

shown

that thetariff

was

not cntical to the

American

cotton industry in the

1830s and 1840s. But Appleton claimedthattheTariffof 1816

was

a crucial factorin the

industry's earlierdevelopment.

What

can

we

say aboutthis earlier

penod?

We

cannot use

econometric techniques for the earlierperiod both becausethe data are very scantyand because

thevagaries of

war

and peace introduceda lot of noise into any time series.

Some

available dataare

shown

inTable 4.

We

obtainedthe value ofdomestic textile

production

by

takingdomestic production inyards

from

Zevin (1971, p. 123) andmultiplying it

by

the domestic price oftextiles inthe U.S.

Bureau

ofthe

Census

(1975, series E-128). British

imports

were

recorded in variousparliamentaryreports. Imports from othercountries are harder

to find.

They

were

taxed at an ad valorem rate, as

shown

in Table 1, andthe

government

lumped

all importssubject to a given tariffratetogether. In other words,there is

no

way

to distinguish

cottontextile imports

from

India

from

otherimports

from

Indiataxed atthe

same

rate.

American

imports from India and

Chma

were

probably notverydiverse, however, andit is likely that the

-16-cotton textiles, as Stettler(1970, p. 214) assumed.

The

data inTable

4

record the totalimports

from

India and

China

subject to thetariffat therate paid

by

cotton textileimports. Ifinaccurate,

theyare overestimates ofthe cotton imports.

Itis apparent thatthe importswere verysubstantial

compared

to domestic production at

thebeginning ofthecentury.

Most

ofthe cotton textiles

consumed

in theUnitedStates before

the

Embargo

of 1808

were

imported.

Domestic

production outside the Waltham-style firmsin

Massachusetts is omitted

from

Table4,butit cannot havebeen large

enough

to alter this

conclusion. This condition,in anycase,

was

short-hved.

Domestic

productionbegan to riserapidly starting soon afterthe

Embargo

of1808

was

instituted in

December

ofthat yearand continuingunderthe protection afforded

by

wartime

conditions. Imports

from

India and

China

fellin 1809 andfollowing years, although they revived

brieflyin 1810-11 as the

embargo

was

relaxed. In the yearsbefore hostilities

between

Britain

andtheUnitedStates, imports

from

Britain continuedunabated.

They

presumably fell sharply duringthe

War

of 1812. In the protectedenvironment ofinternational hostilities, the

modem

cottontextile mdustryof

New

England

grew by

an order ofmagnitude

from

1809 to 1815.

With

peace, however,

came

disaster,as domestic productionfell

by

two-thirdsin 1816.

Although

most

historians talkofdrastic price falls as British imports

were

sold atauction, the

priceof cotton cloth continuedits

downward

trendwithout anybreak.

The

dramatic fall

was

in

production.

The

New

England

cotton mills produced only about one-third as

many

yards of

fabric in 1816 as in 1815; production collapsedbackto the level of 1811 (Zevin 1971, p. 123).

It is nothard tosee FrancisLowell and the BostonAssociates anticipating disaster andruin.

-17-"thecombination ofthe post-war slump, the

dumping

ofBritish goods on the

American

market

afterthe peace,

and

thecollapse ofthe western currency sent

numbers

ofthe oldproducers tothe

wall. 'Halfthe spindles' in the vicinityofProvidence andFall River

were

said to be idle in

1816."

The

banking andcurrency problems in the western United States cut sharplyinto the

domestic

demand

fortextiles andleft

many

manufacturerswith debts

from

unpaid shipments.

The

import surgethat followedpeace in 1815 exacerbatedthe situation.

The

value ofBritish

cottonfabric exportedto theUnitedStates in 1815

was

over

$20

million, almosthalf the value of

domestic productionin thatyear.

We

have notbeen able to findrecords ofBritish cotton imports

inthe yearsjust before 1815, but theimports fell

by

half

from

1815 to

1816

and stayed far

below

the 1815 level.

The

imports

shown

inTable 4

from

Indiaand

China were

not large

enough

to

have causedthe dramatic contraction in domestic production, but theyclearly

show

theeffect of

peace

by

rising sharply in 1816.

The

dramatic (ifshort lived) decline in U.S. production suggeststhat domesticproducers

were

far

more

responsive toimports than they

were two

or three decades later, butthisis ahighly

tentative conclusion.

Thorp

(1926, pp. 117-118) describedthe year 1815 as

one

of"financial

panic" and 1816 as a "depression."

And

recordedprices did not collapsein 1815 or 1816 even

though they

were

fallingsteadily

from

their

peak on

an annual basis in 1814.

Although

textileproducerscouldnot

do

anything about thegeneral

economic

slump, they

couldtrytostop imports. Lowell appealed toCongressfortariffprotection, as recounted

by

Appleton.

He

got it, andcotton imports

from

Asia fell, although they did not vanish as

shown

in

Table 4. Instead, they fellback

from

their

peak

and did not share in the growth of

demand

as the

-18-theirvaluein 1815.

Even

in 1847,

when

the tariff

was

reduced sharply, imports were only about

half as large in value as theyhad been in 1815, although they didexceedthe 1815 yardagefor

onlythe thirdtime since 1815 (Great Britam 1847-48, p. 325).

As

shown

inTable 4, domestic

production revived quicklyand

resumed

itsrapid expansion.

We

draw

two

highlytentative conclusions from thesefragmentarydata. First, the

American

cotton industry probably

was

protected

by

theTariffof 1816.

The

tariffextendedthe

protection afforded

by

embargo

and

war

toenable theindustryto grow. It ispossiblethatthe

industrial growth achievedduring the years ofconflict withBritain could havebeen erased

by

free trade afterthe endofhostilities. SinceLowell

moved

with great dispatchto forestall this

eventualityand general

economic

conditions

improved

thereafter, thiscounter-factual doesnot

have afirmbase. There

was

onlyone yearin

which

the industry facedforeign competition

without ahigh tariffafterits initial growth.^

Second, British imports

were

decreased

by

thetariffof 1816, despiteLowell's cunning

design ofthe

minimum.

The

British appearto have specializedin higherqualityexportsto the

United Statesat leastpartlyin response tothe

American

tariffstructure.

The

differencein quality

ofdomestically producedand imported cotton inthe 1830s and 1840s noted above

may

have

been partly the result ofthe tariff. Ifso, therobust nature ofthe domestic industryinthe faceof

imports inthe 1830s and 1840s

may

have been partlythe result oftheTariffof 1816 in shaping

thenature of

Anglo-American

trade incotton textiles.

^ _Indeed,

_Ware

(1931, p. 72) argued that "Thereis also

no

question that domestic

sheetings were actually in needof

no

protection, forbefore thepassage ofthe [1816] tariffthe

Waltham company

was

enlarging its business in order to meet a rapidly growing

demand."

Zevin (1971, p. 128)concluded that "the tariff

made

no

significant contnbutionto the secular

-19-6.

Conclusions

This paperhas

produced two

strands ofevidencein support ofthefindingthat

American

textilemanufacturers were well established

by

the 1830s. First, there isthe actual response of

the industrytothesharp tariffreduction of 1846. Imports surgedas aresult,butdomestic

production appears tohave been virtuallyunaffected. Second, usingtime series data,the

estimated effectofchanges inthe relative price ofimports on domestic production appearsto be

small.

Both

pieces ofevidence are consistentwith one another

and

suggest, as historical

contemporaries observed, that British

and

American

products

were

quite different

from

one

another.

We

conclude that high tariffs

were

notanessential

component

ofthe survival and

success ofthe laterantebellum domestic cottontextile industry, although theearlycotton

-20-Figure 1:

Average

Ad

Valorem

U.S.Tariffon BritishTextileImports,

1816-1860

0.8 0.7 0) 0.6 ra 0.5 E Q) O 0.4 CO > to 0.3 0.2

-21-Figure 2: Cotton Textiles:

New

England

Production

and U.K.

Imports,

1825-1860

1000000 800000

V

600000 -o

o

400000

p

200000

VS.Production U.K.Imports

Figure 3: ImportsofCottonTextiles

from

United

Kingdom

as Share ofU.S. Market,

1825-1860

-22-Table

1: U.S. Tariff

on Imported Cotton Cloth

Date ofTariffAct Description

Minimum

Valuation

Ad

Valorem

Rate

April27, 1816 Cottoncloth(until 7/19) 250 peryard

25%

Cottoncloth(after7/19) 250 peryard

20%

May

22, 1824 Cottoncloth 300 per yard

25%

May

19, 1828 Cottoncloth 350 peryard

25%

July 14, 1832 Cottoncloth,printed 350 per yard

25%

Cottoncloth, white 300per yard

25%

March 2, 1833 Cottoncloth (after 1/34) 9/10 of excess of1832rateover

20%

Cottoncloth (after 1/36) 8/10 of excess of1832rateover

20%

Cottoncloth (after 1/38) 7/10 of excess of1832rateover

20%

Cottoncloth (after 1/40) 6/10 of excess of1832rateover

20%

Cottoncloth (after 1/42) 3/10of excess of1832rateover

20%

March30. 1842 Cottoncloth, printed 300peryard

30%

Cottoncloth, white 200 peryard

30%

July 30. 1846 Cottoncloth None

25%

March3, 1857 Cottoncloth None

24%

-23-Table

2:

Reduced-Form

Estimates of

Domestic

Cloth

Production

Dependent

Variable:

Log

of

Domestic

Production

Variable (1) (2) (3)

OLS

GLS

First Difference

Constant

Log

(Ep*(l-i-T)/pcW)

Log

_(Pa/Pcw)

Log(Y)

Time

-2.70 11.83*

—

(7.33) (2.03) 1.52* 0.19 0.27 (0.36) (0.16) (0.30) -2.20* -0.27 -0.33 (0.49) (0.24) (0.43) 2.53* 0.13 -0.02 (1.04) (0.31) (0.49) 0.01 0.04* 0.10* (0.05) (0.01) (0.03)

—

0.77* (0.06)

—

35 35 35 0.90 0.98 0.06 0.21 0.32 0.18 8.26* 0.94 2.62

N

Adj.

R2

StandardError

Note: * _{indicates significance atthe 5 percentconfidence level. Standard errors have been}

-24-Table

3: Instrumental-Variable Estimates of

Demand

for

Domestic

Cloth

Dependent

Variable:

Log

of

Domestic

Production

Vanable

(1) (2) (3)

2SLS

2SLS

FirstDifference Constant

Log(Ep*(l+x)/p)

Log

(Y)

Time

-L43

12.64*

—

(7.70) (1.63) 0.59* 0.03 0.08 (0.25) (0.07) (0.14)

L81

-0.04 -0.20

(LIB)

(0.23) (0.30) 0.01 0.04* 0.10* (0.05) (0.01) (0.03)

—

0.78* (0.06)

—

35 35 35 0.86 0.98 0.02 0.30 0.10 0.14 17.7* 2.03 2.18

N

Adj.R^

StandardError

LMr

Note: * indicates significance at the 5 percentconfidence level. Standarderrors have been

corrected forheteroskedasticity. Instruments include logofincome, log ofprice of

raw

cotton,

-25-Table

4: U.S. Cloth

Production

and

Selected Imports, 1805 to

1819

(thousands ofdollars)

New

England

Imports

from

Imports

from

Imports from

Year

Production Britain India

China

1805

978

13,110 1,965 1,716 1806 1,353 22,140 3,524 2,091 1807 1,738 21,214 3,982 1,698 1808 4,073 21,927 4,165 2,510

1809

6,418 29,663

639

452

1810

13,984 32,340 3,277 3,207 1811 15,251 n.a. 3,819 2,723 1812 20,087 n.a.

227

1813 31,514 n.a. 1,150

478

1814

44,453 n.a. 9 1815 47,160 21,185

4

207

1816

16,355 11,199 2,291 1,985 1817 69,739 8,434

820

1,058 1818 122,600 10,922 1,480 1,032 1819 164,027 4,903 2,812 1,016

Sources:

New

England

production (in yards)

from

Zevin (1971), p. 123, multiplied bythe

domesticprice ofRussian

brown

sheeting

from

U.S. Bureau ofthe

Census

(1975), Series E-128.

Imports

from

Britain:

from

Stettler (1970), p. 108, andGreat Britain (1847-48). Other Imports

from

U.S. Congress (1832-34), Vols. I,H.

-26-References

Batchelder, Samuel. Introduction andEarly Progress ofthe Cotton Manufacture ofthe United

States. Boston: Little

&

Brown,

1863.

Berry,

Thomas

Senior. "Production andPopulation Since 1789, Revised

GNP

Seriesin Constant

Dollars." BostwickPapers No. 6.

Richmond,

VA: The

Bostwick

Press, 1988.

Bils, Mark. "TariffProtection and Production in theEarly U.S. Cotton Textile Industry."

Journal of

Economic

History

44

(December

1984): 1033-1045.

Copeland, Melvin T.

The

Cotton Manufacturing Industryofthe UnitedStates. Cambridge:

Harvard UniversityPress, 1912.

David, Paul, and Peter Solar.

"A

BicentenaryContribution to theHistory ofthe CostofLivingin

America," in Research in

Economic

History. Greenwich,

CT:

JAIPress, 1977.

Davis, Lance, andH. Louis StettlerIH.

"The

New

England

CottonTextile Industry, 1825-1860: Trends and Fluctuations." Output,

Employment,

andProductivity inthe United States after

1800. Studies in

Income

and Wealth, Vol. 30.

New

York: National Bureau of

Economic

Research, 1966.

Great Britain,

House

of

Commons.

AccountsofCotton

Wool

Imported, and ofCotton

Manufacture, and Cotton Twist,

Yam

or

Thread

Exportedin

Each Year

1815 to 1847. Sessional

Papers 1847-48, Vol 58.

GreatBritain,

House

of

Commons.

AccountsofCotton

Wool

Imported, and ofCotton Manufacture, and Cotton Twist,

Yam

or

Thread

Exportedin

Each

ofthe Three Years 1848, 1849. and 1850. Sessional Papers 1851,

Vol

53.

Great Britain,

House

of

Commons.

Retums

of Quantities and Declare Value Imported

from

the

Exported tothe United States. Sessional Papers 1856,

Vol

55.

Harley, C. Knick. "Intemational Competitiveness ofthe

Antebellum American

CottonTextile

Industry."

Joumal

of

Economic

History52 (September 1992): 559-584.

Helm,

Elijah.

"The

Cotton Trade oftheUnited

Kingdom." Joumal

ofthe Royal Statistical

Society 32

(December

1869): 428-437.

Grossman,

Gene.

M.

"Imports as a

Cause

ofInjury:

The

Case

ofthe U.S. Steel Industry."

Joumal

of Intemational

Economics

29

(May

1986): 201-223.

-27-Ameica

. Newburyport:

Marss

and Brewster, 1841.

Layer,Robert G. Earnings ofCottonMill Operatives, 1825-1914. Cambridge: Harvard

UniversityPress, 1955.

Montgomery, James

A.

A

Practical Detail ofthe Cotton ManufactureoftheUnitedStates of

America

.

Glasgow:

J. Niven, 1840.

Scranton, Philip. Proprietary Capitalism:

The

TextileManufacture at Philadelphia, 1800-1885 (Cambridge:

Cambridge

UniversityPress, 1983).

Stanwood,

Edward..

American

Tariff Controversies in theNineteenth Century. 2 vols. Boston:

Houghton,

Mifflin,

&

Co., 1903.

Stettler,

Henry

Louis, in.

Growth

andFluctuations in the

Ante-Bellum

TextileIndustry. Ph.D.

Thesis,

Department

of

Economics, Purdue

University, 1970.

Taussig, Frank

W.

The

TariffHistory ofthe United States. 8*Edition.

New

York: G. P.

Putnam's

Sons, 1931.

Temin,

Peter. "ProductQuality andVertical Integration in the EarlyCottonTextile Industry."

Journal of

Economic

History

48

(December

1988): 891-907.

Temin,

Peter.

"The

Industrialization of

New

England, 1830-80," inPeter

Temin

(ed.). Engines of

Enterprise:

An

Economic

History of

New

England

.

Cambridge:

Harvard

UniversityPress, 2000.

Thorp, WillardL. Business

Annals

.

New

York: National

Bureau

of

Economic

Research, 1926.

U.S.

Bureau

ofthe Census. Historical Statistics ofthe UnitedStates,

From

Colonial

Times

to

1970. Washington, D.C.:

GPO,

1975.

U.S. Congress.

American

State Papers:

Commerce

and Navigation. Washington: Gales and Seaton, 1832-34.

Ware,

CarolineF.

The

Early

New

England

Cotton Manufacture. Boston:

Houghton

Mifflin Co.,

1931.

Zevin,RobertB.

"The

Growth

ofCotton Textile Production After 1815." In Robert Fogel and

Stanley

Engerman

(eds.).

The

Reinterpretation of America'sPast.

New

York: Harper

&

Row,

Date

Due

MIT LIBRARIES