James-The_Development_of_the_National_Money_Market

Economic History Association
The Development of the National Money Market, 1893-1911
Author(s): John A. James
Source: The Journal of Economic History, Vol. 36, No. 4 (Dec., 1976), pp. 878-897
Published by: Cambridge University Press on behalf of the Economic History Association
Stable URL: http://www.jstor.org/stable/2119244
Accessed: 14/05/2010 16:36
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=cup.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
Economic History Association and Cambridge University Press are collaborating with JSTOR to digitize,
preserve and extend access to The Journal of Economic History.
http://www.jstor.org
The Developmentof the National
Money Market, 1893-1911
HE period after the Civil War in the United States witnessed a
significant narrowing of interregional interest rate differentials
and the gradual merger of local markets into an integrated short-term
capital market. The movement toward a national money market was
one of the majorfinancialdevelopments of the post-bellum period; and
Lance Davis's series on gross and net returns on earning assets for
national banks (which served as proxies for loan rates) has provided
the principal evidence on this phenomenon.'
The trend toward integration of regional markets may be seen in
Figure 1, which presents the differentials between interior regions
and the East of moving averages of semiannual, short-term interest
rates of country national banks, 1888-1911.2 These rates, computed as
the average rate of return on the loan portfolio, offer more accurate
estimates of local interest rates than either Davis's series or Gene
Smiley's calculated interest rates.3 More detailed data, semiannually
by reserve city and by state for country banks, are available from the
author on request. In spite of the fact that much of the convergence
took place in the 1880's (as indicated by Davis's series) and that the
rate of convergence slowed markedly after the turn of the century,
Journal of Economic History, Vol. XXXVI, No. 4 (December 1976). Copyright ? The
T
Economic History Association. All rights reserved.
I Lance E. Davis, "The Investment Market, 1870-1914: The Evolution of a National Market," JOURNAL OF ECONOMIC
HISTORY,25 (Sept. 1965), 360-65.
2 East-Maine,
N.H., Vt., Mass., R.I., Conn., N.J., Pa., Del., Md., D.C. South-Va.,
W.Va., N.C., S.C., Ga., Fla., Ala., Miss., La., Tex., Ark., Tenn., Ky. Midwest-Ohio, Ind.,
Ill., Mich., Wis., Minn., Iowa, Mo., Kans., Nebr., N. Dak., S. Dak. West-Colo., Wyo.,
Mont., Idaho, Nev., Utah, N. Mex., Ariz., Calif., Oreg., Wash., Okla. The regional interest
rate series are weighted averages of the average rate on loans and discounts of country national
banks by state, the weights being the relative size of the loan and discount portfolio.
3 For a discussion of the derivation of the local interest rates, see John A. James, "Banking
Market Structure, Risk, and the Pattern of Local Interest Rates in the United States, 18931911," Review of Economics and Statistics, forthcoming.
Although constructed from the sanie sources, earnings statements and balance sheets of
national banks in the annual reports of the U.S. Comptroller of the Currency, the series here
appear to be more accurate than those of Smiley, although the comparison is difficult since their
construction is not explicitly reported. For one thing, these series are semiannual, rather than
annual. The Smiley series is actually a rate of return on private earning assets, including items
such as stocks and bonds, which are separated out of the series here; similarly, here losses are
extracted from the reported Losses and Premiums account. Moreover, a correction is made for
the growth of the loan portfolio in the derivation here, in order to avoid biasing the computed
rates downward. Gene Smiley, "Interest Rate Movement in the United States, 1888-1913,"
JOURNAL OF ECONOMIC HISTORY, 35 (Sept. 1975), p. 595.
878
National Money Market
879
I-.
2
WlDUE51
~~~~~~~~~~~~~~
\
-
O~~~~~~~~~~~~
WEST
=
Z-2
506
A~~~~~~~~ER
FIGURE
INTERREGIONAL
DIFFERENTIALS
I
SHORT-TERM
WITH
THE
INTEREST
RATE
EASTERN
REGION, COUNTRY BANKS
there is in Figure 1 evidence of a general decline in interregional
interest rate differentials and the gradual evolution of a national
market. The gaps between western and eastern and between midwestern and eastern rates both fall by about one percentage point, or
by about 25 and 50 percent respectively. The southern-eastern differential, on the other hand, is much slower to narrow, falling somewhat less than .5 percentage points.4
In his study of interest rate movements over this period, Smiley
finds no evidence of regional interest rate convergence for nonreserve city national banks, that is, country banks. He uses the
coefficient of variation as a measure of dispersion.5 The coefficient of
variation is, however, an inappropriate measure of convergence here,
since the focus should not be on convergence from above and below
toward some mean interest rate, which has no economic meaning.
4 A similar story is found by examining interest rate movements of the more disaggregated
regions reported by the Comptroller of the Currency. The gap between North Central and New
England states narrows significantly over the late 1890's, being virtually eliminated by 1900.
The differential between the Great Plains and Mountain states and New England narrows by
almost two percentage points, or 40 percent, while that between the Mountain and Pacific states
and New England falls around one percentage point, or about 25 percent.
5 Smiley, "Interest Rate Movement," p. 600.
880
James
Rather, a more suitable measure is the proximity of other regional
rates to those in the most developed market, the East. In Figure 1 it is
clear that interregional interest rate differentials did narrow over the
period and that progress was being made toward the development of a
national market.6
How can we account for the interest rate convergence of the
post-bellum period? What forces were at work in forming a national
money market? In the following pages I examine and test two major
hypotheses about this development:7 an Institutional Change
hypothesis, which emphasizes the expansion of the commercial paper
market in facilitating interregional transfers of funds; and a Market
Power hypothesis, which stresses the lowering of legal barriers to
entry in banking and the consequent erosion of local monopoly positions.
Lance Davis emphasizes the role of institutional change; the westward expansion of the commercial paper market, he says, wove
6
Rather than the coefficient of variation, a more appropriate, although ad hoc, indicator of
the rate of convergence would involve some measure of the diminishing differentials between
other regions and the East. The square root of the average of squared differentials of other
regions with the East,
I/
3
(rf - re)2
is presented here for five year intervals over the period below.
1888
1893
1898
3.737 percent
2.955 percent
3.353 percent
1903
1908
1911
2.748 percent
2.341 percent
2.451 percent
The rate of convergence is not uniform, to be sure, but the trend over the period is clearly
toward narrowing differentials with eastern rates. Moreover, the rate of convergence slows after
1900, as Davis also finds.
7 Richard Keehn in his study rejecting the Sylla hypothesis attributed the decline in differentials to declines in transfer costs between markets due to technological improvements in
transportation and communication. Domestic exchange rates may be taken as a measure of the
transfer costs of funds. Even in the 1830's, a period before significant advances in transportation
and communication, the highest quoted exchange rate was lower than the western-eastern and
southern-eastern differentials over 1888-1911. Those quoted in Bradstreet's in the late
nineteenth century were substantially smaller yet. Consequently transfer costs could account
for only a small part of the total observed differential.
Jeffrey Williamson emphasizes demand rather than supply factors in explaining the movement of the differentials. A relative decline in the demand for funds in the Midwest in the late
nineteenth century caused rates there to decline relative to eastern rates and after the turn of
the century an increase in demand for funds there led to a relative rise in rates. Thus, the
explanation for capital market integration is found in real forces exogenous to the capital market.
However, this explanation is limited to the Midwest and does not account for the movement of
western and southern differentials. Richard Keehn, "Federal Bank Policy, Bank Market Structure, and Bank Performance: Wisconsin, 1863-1914," Business History Review, 48 (Spring
1974), 1-27. Jeffrey G. Williamson, Late Nineteenth-Century American Development (Cambridge: Cambridge University Press, 1975), pp. 130-32.
National Money Market
881
segmented local capital markets together into a national one.8 It was
an open market in which borrowers and lenders from many different
locations could participate, so its geographical expansion, drawing
more and more borrowers and lenders together into one market, was
a force making for interest rate equality.
In contrast to Davis's picture of widespread barriers to interregional flows of funds, Richard Sylla views the national banking system
as a very efficient mobilizer of funds. Indeed, he argues, "It promoted
industrial investment and growth with a ruthless efficiency."9 If the
national banking system was so efficient, then, why were there interest rate differentials over this period? Sylla answers by pointing to
legal barriers to entry established in the National Banking Acts. The
tax on state bank notes left national banks in an advantageous position
after the Civil War, while relatively high minimum capital requirements and the prohibition of mortgage lending severely restricted
entry in rural areas. As a result, many country bankers enjoyed
quasi-monopolistic positions. The observed interregional and citycountry interest rate differentials are thus evidence of local monopoly
power. 10
This local power in turn is said to have been eroded by the passage
of the Gold Standard Act of 1900, which lowered minimum capital
requirements for national banks to $25,000 in towns of population less
than 3,000. Lowering the legal barriers to entry after 1900 resulted in
a rapid increase in the number of national banks in rural areas and,
consequently, in a diminution of monopoly premiums and local loan
rates primarily in western and southern regions."
These hypotheses concerning the development of a national money
market have not, however, been tested systematically in a theoretical
model.12 To do so, I shall examine the forces underlying regional
interest rate convergence, 1893-1911, using a model of local interest
rate determination.
THE MODEL
In order to assess the performance of local capital markets, a model
of local interest rate determination, or in other words, a model of
8
Davis, "The Investment Market, 1870-1914," pp. 370-73.
9 Richard Sylla, "The United States, 1863-1913," in Rondo Cameron, ed., Banking and
Economic Development (New York: Oxford University Press, 1972), p. 258.
10 Richard Sylla, "Federal Policy, Banking Market Structure, and Capital Mobilization in the
United States, 1863-1913," JOURNAL OF ECONOMIC HISTORY, 29 (Dec. 1969), p. 685.
11 Ibid., pp. 664-70.
12 Smiley has considered these two hypotheses by an inspection of the data, but not within a
well-defined theoretical model. Smiley, "Interest Rate Movement," pp. 607-10.
882
James
bank asset selection, is needed. I use a mean-variancemodel of
nineteenth-centurybank portfolio selection based on the capitalasset-pricingmodel.'3 Since the influence of local monopolyon interest rates is to be assessed, it is assumed that the bank faces a
downward sloping demand curve for loans. A portfolio selection
modelwith an imperfectlycompetitiveasset market,the loan market,
is developed, fromwhich it followsthat the expected returnon loans
may be expressed as the sum of the standardportfoliotheory risk
premiumanda monopolypremium.Transformedinto estimableform
the equation may be written
Rt = a,1CPt + a2LSt + a3LSt_1 + a4VARt + a5MPt
+ a6BCt + a7SEASt +
a8
+
Et
(1)
where Rtis the averagereal rate of returnon commercialloans, using
the Warren-Pearson
wholesaleprice indexas a deflator.LStis the loss
rateon commercialloansin periodt. VARtis the measureof loan risk,
the samplevarianceof the loss rateover the previousfive yearsor ten
periods, so its sign shouldbe positive. As riskincreases,localinterest
rates should increase. MPt is the monopolypower measure, a bank
densityindex-the totalnumberof banks(national,state, and private)
per capitain the area, so its sign shouldbe negative. As bankdensity
increases, local interest rates should decrease. BCt is the business
cycle index, the deviationfrom the trend of bank clearingsoutside
New YorkCity. SEAStis a semiannualseasonaldummy.
The equation was estimated for country banks by state and for
reserve cities on semiannual data from 1893 to 1911 using the
Cochrane-Orcuttiterative technique as a correctionfor autocorrelation. The majorityof the estimatedcoefficientsof majorvariablesare
significant,and the fits of the regressionsare quite good. In many
southern and midwesternstates over 90 percent of the varianceis
explained, while on average poorer fits are found in the East and
13 For the derivation see John A. James, "Portfolio Selection with an Imperfectly Competitive Asset Market," Journal of Financial and Quantitative Analysis, forthcoming.
Several features of nineteenth-century banking greatly simplified the portfolio problem. The
benign influence of the real bills doctrine ensured that loans were short-term and thus eliminated the term structure problem. Moreover, liquidity constraints do not seem to have been
considered.
The assumption of risk aversion in the mean-variance model seems to be a reasonable one.
There is no evidence of a well-developed market in bank stock, so it may be assumed that banks
were generally closely held by the entrepreneur, who is usually taken to be risk-averse. The
Sylla and Keehn models of bank behavior, on the other hand, portray banks as firms, equating
marginal revenues and marginal costs on assets; such a characterization misses the essence of
choice among varying risks in portfolio selection. John A. James, "The Evolution of the National
Money Market, 1888-1911," unpublished MIT Ph.D. thesis, 1974, pp. 301-7.
National Money Market
883
West. Space limitations preclude detailed presentation of the regression results, but they are available elsewhere.14 Here we shall be
concerned with testing hypotheses in this framework.
The formulation allows the role of risk to be taken into account
explicitly. Indeed, Stigler has criticized Davis insofar as he identifies
the existence of interest rate differentials with capital market imperfections.15 They might reflect, for example, real transactions and
information costs. Even in a perfect capital market, rates would not
be uniform if loans differed in the degree of risk attached to them
across regions. In that case declining interest rate differentials would
be evidence of decreases in the relative risk of western and southern
loans. 16
In time series regressions of equation (1) the estimated risk
coefficient, that of VAR, is significantly positive (as expected) in about
one-third of the states. Moreover, my analysis of variance test on the
rate of decrease of the sample variance led me to reject the hypothesis
-at the 95 percent level-that there were no regional differences;most
of the regional variation was provided by the western states.17
In spite of the fact that loan risk may have decreased in interior
regions over this period, changes in risk alone can not account for the
observed regional interest rate movements. In a cross-section time
series regression of country banks across states the estimated risk
coefficient was found to be quite small and not significantly positive.
In addition, the monopoly power index was significant, indicating that
changes in local market structure and monopoly power did have a
significant influence on local interest rates.18 In brief, I could reject
14 For detailed information about the transformation of the
theoretical risk-return relationship into estimable form, construction of the variables, and regression results see James, "Banking
Market Structure."
15 George Stigler, "Imperfections in the Capital Market,"Journal of Political Economy, 75
(June 1967), 288.
16 Smiley suggests that his finding of no interest rate convergence over this period may have
been due to an increase in risk aversion of bankers, although risk is not explicitly measured.
Smiley, "Interest Rate Movement," pp. 600-5.
17 The estimated coefficients from a simple regression of the VAR series by state on a time
trend were regressed in turn on regional dummies.
VAR =-.00045
DEAST
(-.06)
R2 = .17
-
.0016 DSOUTH
(0.23)
-
.00085 DMW - .0267 DWEST
(-.11)
(-3.47)
F(3,41) = 2.99
Even though the groupings are not significant, the magnitudes of the coefficients across regions
are consistent with out expectations, the smallest decline being in the East, then the Midwest,
the South, and finally the West with the largest decline.
18 In a cross-section time series regression for reserve cities, however, the estimated VAR
coefficient was significantly positive and of substantial magnitude, perhaps indicating that closer
884
James
the hypothesis that a perfect market existed; in order to have some
degree of local monopoly power, capital markets must have been
geographically separated to some extent.
Decreases in risk differentials were not sufficient to explain the
observed narrowing of interregional interest rate differentials, even
though such changes may have been important in such areas as the
West. Capital market integration was a complex process, and while
declining risk played a role in that process, the national money
market must have been in some degree imperfect. Barriers to capital
mobility must have existed or else capital flows would have equalized
rates. Consequently, we must examine the Institutional Change and
Market Power hypotheses, both of which are based on the existence
of an imperfect national capital market.
THE INSTITUTIONAL
CHANGE
HYPOTHESIS
According to the Institutional Change hypothesis, the westward
spread of the commercial paper market broke down the barriers to
capital mobility and reduced interregional interest rate differentials.
But the westward expansion of the commercial paper market was not
as clear-cut and easily identifiable a process as the one Davis has
pictured by chronicling the westward march of commercial paper
house openings.19 Interior banks were never really isolated from the
commercial paper market since they could always have participated in
it through the New York correspondent bank that virtually every
country bank had. Travelling salesmen were also a common method
of dealing with distant customers. In addition, it is very difficult to
identify the influence of the commercial paper house in any particular
community. Just because a house was established in a city in a given
year, it does not follow that it began to exert an influence on loan rates
immediately. In Minneapolis or St. Paul, a note broker was in operation as early as 1875, but it was not until 1900 that banks in
moderate-sized Minnesota towns had become purchasers of commercial paper.20 The rate of dissemination of commercial paper from the
city into the countryside is difficult to determine.
Nevertheless, in testing this hypothesis, I rely on tests of the
timing of the spread of the commercial paper market because this is
market integration among cities allowed banks to reduce risk. For the complete regression
results see James, "Banking Market Structure."
19 Davis, "The Investment Market, 1870-1914," p. 372.
20 Albert 0. Greef,
The Commercial Paper House in the United States (Cambridge: Harvard
University Press, 1938), p. 39.
National Money Market
885
the only developmenton whichwe have information;detaileddataon
the volume of commercialpaper transactionsin local marketsdo not
seem to have been compiled. As a result, I employ a post hoc, ergo
propterhoc analysiswhichcomparesthe timingof the introductionof a
commercialpaperhouse into an areaor the date at which such paper
began to be purchasedlocally in relativelysubstantialamountswith
the pattern of local interest rates.21
The introductionof commercialpaperin an area might affectlocal
loan ratesthroughboth demandand supplyeffects.An increasein the
supply of commercialpaper meant that local firms had alternative
formsof financing,enablingthem to escape a localbankingmonopoly.
The spreadof commercialpaperinto an areashouldhave reducedthe
monopolypower of local banks.
On the demandside, the commercialpaper marketallowedbanks
to diversifytheir portfoliosacrossregions. Smallcountrybankspreviously restricted to makingloans in local areas, where default risks
were likely to be highly correlated,could then in effect makeloansin
other regions and industries, where the risks were independent or
only weakly correlatedwith those on local loans. Giving the bank a
wider geographicalareaover whichto loan enabledit to avoidsome of
the local risk, such as weather conditions,thereby reducingthe risk
on its total portfolio.There should, as a result, have been a decrease
in the loss rate and its varianceand thus in the interest rate.
First let us considerthe demandside effects. I performeda Chow
test on the estimated VAR coefficient in equation (1) to determine
whether the introductionof the paper marketresulted in a decrease
in portfoliorisk. The date at which local banks began to purchase
open-marketpaper can be identified for two reserve cities; banksin
both St. Louis and KansasCity did not start investing substantial
sums in open-marketpaperuntil after 1897. Similarly,countrybanks
in several midwesternstates-Minnesota, North Dakota, Nebraska,
and Kansas-began purchasingcommercialpaper aroundthe turn of
the century.22The Chow test results are reportedin Table 1.
The results are inconclusive.There seems to be a definite shift in
the KansasCity coefficient,but on the other hand, no effect can be
seen in St. Louis. For countrybanks, there is evidence of a shift in
Minnesotaand North Dakota, but not in Kansasand Nebraska.
21
Smiley suggests that the expansion of the commercial paper market over the period
1907-1909 may have led to interest rate convergence, but does not explicitly test the proposition. Smiley, "Interest Rate Movement," pp. 607-9.
22
Greef, The Commerical Paper House in the United States, p.48.
886
James
TABLE
1
CHOW TEST ON VAR COEFFICIENT FOR THE EFFECTS
OF PORTFOLIO DIVERSIFICATION
St. Louis
Kansas City
Minnesota
Nebraska
Kansas
North Dakota
a
b
F(1,21)
F(1,21)
F(1,21)
F(1,21)
F(1,21)
F(1,16)
=
=
=
=
=
=
.966
10.054a
7.319a
3.553
2.301
4.711b
Significant at 1 percent level.
Significant at 5 percent level.
On the supply side, the introductionof a commercialpaper dealer
into an area enabled a local firm to borrowfrom thousandsof banks
across the United States by selling its paper to the note broker,
thereby limiting the monopoly premium which local banks could
extract.There shouldhave been a smallerdegree of monopolypower
after the introductionof the commercialpaper market, so there
shouldbe a fall in the magnitudeof the MP coefficient.I did a Chow
test to determine whether the MP coefficienthas shifted for every
reservecity in whichI couldidentifythe date of the firstsales of openmarketpaper.23Those test results appearin Table 2. As the reader
TABLE 2
CHOW TEST ON MP COEFFICIENT AROUND BEGINNING
OF COMMERCIAL PAPER SALES
Cities
Year of First Sale
San Francisco
Portland
Columbus
Dallas
1900
1906
1907
1910
F statistic
F(1,31)
F(1,8)
F(1,7)
F(1,3)
= 21.785a
= 3.096b
= 1.719
= 1.793
a Significant at 1 percent level.
b
Significant at 10 percent level.
can see, the San Franciscoregressionclearly indicatesa shift in the
MP coefficient;evidence fromthe other regressions,however, is less
conclusive.
Since my tests thus far have relied on the exacttimingof the introduction of the commercialpaper market(which might be measured
ratherimprecisely),I also decided to use a weaker, less restrictive
test. It is clear that even in cities in which the open marketwas
established,there shouldhave been a decline in the monopolypower
of local banks if the size of the marketwas growing. Thus, the MP
23 For identification of the starting date of commercial paper sales in various cities see Greef,
p. 39. Chester A. Phillips, Bank Credit, (New York: Macmillan Co., 1920), p. 135.
National Money Market
887
coefficientshouldhave become more positiveover time. To test this,
equation (1) was re-estimated including a slope dummy, MPDM,
which was one over the second halfof the period. The MP coefficient
growsmorepositive over time if the monopolypower associatedwith
a given marketstructureis decreasing;therefore, the slope dummy
should be positive. This test does not depend on the exact timing of
the spreadof the open market,only on the fact that a given market
structureshouldresultin less marketpowerover time, if competition
from commercialpaper was increasing. Table 3 presents the estimates of the slope dummy coefficients.
The signs are mixed, but negative signs, the "wrong"sign, predominate. Only three cities have significantlypositive coefficients.
TABLE 3
ESTIMATED MP SLOPE DUMMY COEFFICIENTS
RESERVE CITY BANKS
(T STATISTICS IN PARENTHESES)
East
BOS
NYC
ALB
PHILA
MP Dummy
South
NO
14.94a
(2.81)
9.206
(.96)
-3.815
(-1.06)
2.563
SAV
HOU
DAL
BALT
-1.539
(-.92)
3.364
BROOK
3.901
(.97)
CIN
-.1594
CLEV
(-.13)
4.346
(.69)
-2952
West
SF
CHI
DET
-1.577
(-.69)
-1958
(-.31)
-6.034
PORT
LA
-18.69
(-1.15)
- 7.337a
(-2.17)
MP Dummy
-9.166a
(-2.70)
- 10.77a
(-2.72)
-22.95
(-1.10)
15.03
(1.23)
MILW
ST L
(-.70)
(.49)
WASH
Midwest
(-.63)
(.49)
PITT
MP Dummy
-6.132
(-.91)
-11.86a
(-2.07)
KC
OMA
ST J
-.2206
(-.10)
_3.144a
(-2.78)
-1.395
(-.87)
MINN
-.8751
(-.20)
ST P
-.2465
(-.04)
-1.643
DM
(-.90)
LINC
IND
COL
KK
7.423a
(2.92)
4.877
(.96)
23.30a
(4.06)
.4849
(.09)
a Significant at 1 percent level.
888
James
On the whole, my conclusionis that the commercialpaper market
was not responsible for long-run declines in monopoly power and
interest rates.
Thus introductionof the commercialpaper marketin an areadoes
not seem to have had a significantimpacton localinterest rateseither
throughdemandinfluences (portfoliodiversificationand risk reduction)or throughsupplyinfluences(thatis, reductionin the monopoly
power of local banks).To be sure, the evidence is somewhatmixed
and in certain local marketscommercialpaper may have had a significanteffect. Nevertheless, there does not appearto be a systematic
patternof shifts in the risk or monopolypower coefficientsafter the
appearanceof a commercialpaper dealer.
Furthermore,the general patternsof regionalinterest rate movementsare not consistentwith the pace of expansionof the commercial
paper market.ConsiderFigure 1. The commercialpaper marketdid
not reach the PacificCoast until 1900. Substantialrelative interest
rate declines, however, took place in the western states before that
date, and hence cannotbe accountedfor by the influenceof commercial paper.24On closerinspection,the patternof regionalinterestrate
movementsis not consistentwith the timing of the expansionof the
commercialpaper market.
THE MARKET POWER
HYPOTHESIS
Cross-sectionregressionsfor selected dates indicate, however, that
local bankingmarketstructureor monopolypower was quite important in accountingfor interstate variationin country bank interest
rates. In addition, in the time series regressionsof equation (1) by
state for country banks, the estimated MP coefficients are almost
alwayssignificantfor states outside the East.25Statisticalsignificance
alone revealslittle about the explanatorypower of a variable.In this
case, however,the ceteris paribussimulatedeffects of changesin the
MP variableover the period accountfor almost all of the observed
interest rate declines of countrybanksin most states, especially for
the Southand Midwest.26Decreases in local monopolypowerappear
24 This case is strengthened by examining more finely drawn regions. For example, it was
noted that country banks in the Great Plains began buying paper around 1900: however, there
also were substantial declines in interregional differentials before 1900.
25 In the South and Midwest, for example, virtually every state has a negatively signed and
significant MP coefficient. For the complete list of estimated regression coefficients see James,
"Banking Market Structure."
26 For the simulated magnitudes see James, "The Evolution of the National Money Market,
1888-1911," p. 542.
889
National Money Market
to have been the most importantfactorsaccountingfor the narrowing
of inter-regionalinterest rate differentials.In this section we shall
examine why local monopolypower was eroded.
As we have noted, the Syllahypothesisfocuseson the effectsof the
Gold StandardAct of 1900; by lowering entry barriers,it brought
abouta rapidincreasein the numberof nationalbanksin ruralareas.
But consider Figure 1 again. Most of the decline in interregional
interest rate differentialsactuallyoccurredprior to 1900. The Sylla
hypothesis cannot explain the substantialnarrowingof differentials
before 1900. The patternof interestrate differentialspredictedby his
hypothesisis, in fact, the opposite of the patternthat can be seen in
Figure 1.
Whateffectsdid the Gold StandardAct have on bankingstructure?
Figure 2, which is based on National MonetaryCommissionstatisFIGURE
2
N.M.C. SERIES OF PRIVATE, STATE,
AND NATIONAL BANKS, 1877-1909
12,000
-
-
-
-
~-
11, 000
Xfa
10,000W<
-
9,000
-
-
-
--
-
8,000
-
-
_
--
L
X
7,000
6, 000
_
-07
-0
5,000
State
-
Private
_Yl
Natsonal
v-
------------
Barks
-
-
Year
1877
Banks
.-
4,C000,-
3, 13
Banks
1881
1885
-
-1889
.L.X
1893
11897
1901
1905
1909
Source: George E. Barnett, State Banks and Trust Companies since the Passage of the National
Bank Act (Washington: Government Printing Office, 1911), p. 202.
tics, presentsthe growthof national,state, and privatebanksbetween
1877 and 1909. Over the 1890's the number of nationalbanks remained quite steady, while the number of state banks almost doubled. After 1900 both state and nationalbanks grew rapidly, and
nationalbankscertainlydid not have a dominantrole in the general
27
expansion.
27 The increase in the rate of formation of both state and national banks after 1900 suggests
that they were affected by some common influence such as the business cycle. If lowered legal
890
James
The Sylla hypothesis overstates the influence of the Gold Standard
Act of 1900 on the net rate of bank formation. Many of the new
national banks were merely conversions or reorganizations of existing
state banks. Thornton Cooke in a 1901 article pointed out that the law
had been anticipated; in the six months following the passage of the
Gold Standard Act, not one new (as opposed to converted) national
bank was established in North Dakota. In South Dakota and Missouri
only one was added, in Nebraska and Kansas only two. In these five
states the new law resulted in a net addition of six banks over the six
month period; in contrast, eighty-five new state banks and ten new
private banks were organized during the same months. Even though
the new law had substantial effects in some states in which state bank
minimum capital requirements were relatively high,28 its impact on
local market structure is likely to be greatly exaggerated if one
examines only the rate of national bank formation.
This discussion and Figure 2 point up the fact that one of the
principal difficulties with the Sylla hypothesis is its relative neglect of
the role of state banks. The argument is focused on changes in the
number of national banks, leaving state banks on the periphery; this is
done in spite of the fact that the state institutions far out-numbered
national banks and were growing more rapidly at this time. Except for
1901, national banks always totaled less than one-third of all newly
created banks-despite the passage of the Gold Standard Act in
1900.29 In addition, the greatest growth in state banks was in the
interior regions, the areas where monopoly was supposedly the
strongest. In New England the number of state banks in 1909 was
exactly the same as in 1879, and in the eastern states as a whole the
number of state banks increased between 1879 and 1909 by only 105
percent; on the other hand, in the middle western and western states
the increases were 1, 160 percent and 1, 198 percent, respectively.30
Banking services were becoming much more readily available than
the Sylla hypothesis indicates. In 1908 it was observed, "Almost every
barriers to entry in the National Banking Act had been the principal influence, no influence on
the formation of state banks should be observed. If anything, the rate of state bank formation
should have decreased with the lowering of national bank capital requirements.
28 For example, in Texas where state banks were still forbidden, thirty-four new national
banks were organized over the same period. In Iowa, where the state minimum capital
requirement was $25,000, the same as the national one, thirty-two new national banks were
organized. Thornton Cooke, "The Effect of the New Currency Law on Banking in the West,"
QuarterlyJournal of Economics, 15 (Feb. 1901), 278-80.
29 All-Bank Statistics, p. 32.
30 George E. Barnett, State Banks and Trust Companies since the Passage of the NationalBank Act (Washington: Government Printing Office, 1911), p. 202.
National Money Market
891
municipalcommunitythroughoutthe length andbreadthof our country has its bankinginstitutionand in manyof the smallaggregationsof
not more than 1,000 population we find more than one banking
"31 Indeed, towns as small as Battle Creek,
institutionprospering.
Iowa, with a population of 688 in 1915, had three banks; North
Liberty, Iowa, with a populationof under 200, had two banks in
1918.32This proliferationof bankswas not looked upon favorablyby
some of the bankersthemselves:"So numeroushave countrybanks
become that a state of competitionexists, so excessive as seriouslyto
threaten the welfare of the banks, and, to a materialdegree, demoralize the people through too cheap credit."33Another banker
complainedthat "morethan a fair averageprofit cannot be realized
... It is whether... suitableamendmentsto statelawsshouldnot be
interposed to prevent more than a limited number of banks to the
local assessed valuation."34
How can we accountfor this impressivegrowthof statebanksin the
post-bellumperiod?One possibilityis that the shiftin preferencesfor
holding money in the form of deposits rather than specie, as evidenced by the rapid rise of the deposit-currencyratio, lessened the
disadvantage at which the non-note-issuing banks operated immediatelyafterthe CivilWar.35In his studyof post-bellumWisconsin
banking,RichardKeehn finds that the rapidspreadof deposit banking made state banks close substitutesfor nationalones.36
Table 4 presents the minimum capital requirements for state
banks in 1895 and 1909. If state and nationalbankswere relatively
close substitutes, the lowered national bank capital requirements
should have had relativelylittle influence on bankformation.In the
great majorityof states the minimumcapitalrequirementwas below
$25,000.37 In New England states in which banks were incorporated
31
A. E. Padgett, "The Multiplication of Banks," Proceedings of the South Carolina Bankers'
Association, 1908, p. 126.
32 Howard H. Preston, History of Banking in Iowa (Iowa City: State Historical Society of
Iowa, 1922), p. 354.
33 "Country Credit Methods," Proceedings of the Maryland Bankers' Association, 1914, p.
47.
34 T. H. Hinchman, Banks and Banking in Michigan (Detroit: M. Graham, 1887), p. 170.
35 Milton Friedman and Anna Schwartz, A Monetary History of the United States (Princeton:
Princeton University Press, 1963), p. 122. On the other hand, state banks might not have been
severely handicapped by the loss of the privilege of note issue even relatively early in the
post-bellum period. In 1881, the earliest observation date, checks rather than currency constituted over 70 percent of total receipts of national banks in all rural regions except the Pacific
states. U.S. Comptroller of the Currency, Annual Report, 1881 (Washington: Government
Printing Office, 1881), pp. 17-19.
36 Keehn, "Federal Bank Policy," p. 27.
37 Private banks, unincorporated banks not generally subject to state regulations, should also
TABLE 4
MINIMUM CAPITAL REQUIREMENTS FOR STATE BANKS
1909
1895
State
Change
EAST
Maine
New Hampshire
Vermont
Massachusetts
Rhode Island
Connecticut
New York
New Jersey
Pennsylvania
Delaware
Maryland
Virginia
West Virginia
North Carolina
South Carolina
Georgia
Florida
Alabama
Mississippi
Louisiana
Texas
Arkansas
Tennessee
Kentucky
Ohio
Indiana
Illinois
Michigan
Wisconsin
Minnesota
Iowa
Missouri
Kansas
Nebraska
North Dakota
South Dakota
Colorado
New Mexico
Arizona
Utah
Wyoming
Montana
Idaho
Nevada
Washington
Oregon
California
Oklahoma
$
$
_-a
b
b
b
50,000
b
b
b
b
b
b
b
25,000
50,000
25,000
25,000
50,000
50,000
b
0
0
-25,000
b
50,000
SOUTH
10,000
25,000
b
0
25,000
15,000
50,000
0
100,000
10,000
-40,000
10,000
25,000
5,000
0
15,000
15,000
15,000
10,000
10,000
0
0
0
-10,000
0
-35,000
+ 10,000
-90,000
10,000
__a
0
b
0
15,000
0
MIDWEST
25,000
25,000
25,000
15,000
25,000
10,000
25,000
10,000
5,000
5,000
5,000
5,000
25,000
25,000
25,000
20,000
10,000
10,000
25,000
10,000
10,000
10,000
10,000
10,000
0
0
0
+5,000
WEST
30,000
30,000
10,000
30,000
-20,000
0
0
b
0
0
0
+5,000
+5,000
+5,000
+5,000
0
b
10,000
25,000
-15,000
10000
b
20,000
20,000
0
10,000
10,000
b
b
10,000
25,000
-15,000
10,000
b
25,000
5,000
b
+20,000
10.000
a
No state banks
Capital individually determined
Source: U.S. Comptroller of the Currency, Annual Report, 1985 (Washington: Government
Printing Office, 1896), pp. 38-60, 111-169. George E. Barnett, State Bank and Trust
Companies (Washington: Government Printing Office, 1911), p. 43.
b
892
National Money Market
893
by specialact, the capitalwas fixedforeach bankby the legislature.In
the eastern states and the most easterly midwestern states (Ohio,
Indiana, and Illinois), the minimum capital requirementwas relatively high, usuallyaround$25,000. On the other hand, in the less
dense, more rural states of the South and West, the minimum requirementwas usuallymuch lower, mostly rangingbetween $10,000
and $15,000.38 On average between 1895 and 1909 there was a
decreasein capitalrequirementswhere they had been relativelyhigh
in 1895. On the other hand, severalstates in the GreatPlains,where
requirements-$5,000-were quite low in 1895, raised them to
$10,000 after finding bankswith a very small capitalunsatisfactory.
Over the period there was thus a convergence toward minimum
capitalrequirementsof $10,000-$15,000.
Did these low minimum capital requirements, especially the
$5,000 minimumin the Great Plainsstates, actuallyencouragebank
formationin small towns and erode local monopolies?In North and
South Dakotavirtuallyall state bankswere located in towns with a
populationof 1,000 or less. In North Dakota47 out of 71 state banks
had a capitalof $10,000or less; in SouthDakota,48 out of 76. In both
Kansasand Nebraskathree-quartersof the state bankshad capitalsof
under $25,000, the level to which the nationalbankrequirementwas
loweredin 1900. Thirteentownsin Nebraskawith a populationof less
than one hundredhad state banks,and there were eight such villages
in Kansas.39Clearly,low minimumcapitalrequirementsdid encourage the development of banking in small towns. High capital requirements prevented the establishmentof nationalbanks in these
towns, but bankingservices were provided by state institutions.
To determine the effect state minimumcapitalrequirementshad
on the structureof interest rates, the countrybank interest rates of
be mentioned. On the whole they were quite small and provided banking facilities in very small
towns which could not support a larger bank. For example, in Wisconsin in 1890 one-third of
private banks had capital of less than $5,000 and two-thirds had capital of less than $15,000.
How could any bank possess monopoly power as long as private banks could be established with
no minimum capital? Apparently private banks were not perfect substitutes for chartered banks.
The public seemed to put a premium on soundness and felt more comfortable with chartered
banks as a result. Nevertheless, private banks did to some extent act as an offset to high capital
requirements. In the period immediately after the Civil War private banks grew very rapidly,
providing alternatives to the relatively difficult-to-establish national banks; later, in the 1880's,
state banks became viable alternatives to national ones, and the growth of private banks
essentially stopped. Theodore A. Andersen, A Century of Banking in Wisconsin (Madison: State
Historical Society of Wisconsin, 1954), pp. 62-63.
38 In 1909 only four states had no minimum capital requirements for state banks-Arizona,
Arkansas, South Carolina, and Tennessee. In these states state banks were organized under the
business incorporation law and like other corporations the size of capital was at the discretion of
the incorporators. Barnett, State Banks, p. 36.
39 Thornton Cooke, "Distribution of Small Banks in the West," Quarterly Journal of
Economics, 12 (Oct. 1897), 71.
894
James
1909 were regressed on 1909 state minimum capital requirements
(representing barriers to entry) and on the distance of the major city
in the state from New York City (representing the ease of capital flow
from the national to the local market). The cross section was weighted
by the proportion of state loans and discounts to total loans and
discounts as a correction for heteroscedasticity. The regression results
appear below.
R = .0892 CPTL + .00296 DIST + .01608
(11.87)
(7.68)
(1.92)
R2 = .834
F(2,45) = 113.19
(2)
The CPTL coefficient is positive and highly significant, so that lower
state minimum capital requirements are associated with lower local
interest rates, ceteris paribus. The DIST coefficient has a positive
sign, as expected, and is also highly significant. State minimum
capital requirements definitely seem to be important factors in explaining cross-sectional differences in interest rates.
The effects of state bank minimum capital requirements should
appear in the time series also. I regressed the change in interest rates,
1888-1911, by state, on the 1909 minimum state capital requirements;
the results are:
AR = .0767 CPTL
(2.39)
R2 = .110
-
3.097
(-5.20)
F(1,46) = 5.704
(3)
The CPTL coefficient is positive and significant. States in which there
were lower bank capital requirements experienced a larger fall in
interest rates, ceteris paribus.
Lower state bank capital requirements are associated with lower
interest rates in the cross section and in the time series. Clearly,
lower requirements allowed more banks to be established in small
towns because they lowered entry barriers; as a result, local
monopoly power should have been diminished. If so, there should be a
relationship between state bank capital requirements and the estimated MP coefficients from equation (1): the more negative the MP
coefficient, the larger the degree of monopoly power. The estimated
MP coefficients for country banks by state were regressed on 1909
minimum capital requirements for state banks and state population
density (which was included to correct for differences in population
structure across states). Other things being equal, a bank in a lower
density region should have more monopoly power since the im-
National Money Market
895
mediateareacan supportfewer banks.The regressionis weighted by
the t statisticsof the MP coefficientsto emphasizethe more accurate
estimates.
MPC = .0810 DNS - 1.349 CPTL - .1687
(1.15)
(-3.12)
(-1.13)
R2 = .195
F(2,45) = 5.446
(4)
The lower the capital requirement, the smaller the degree of
monopoly power should be and hence the more positive the MP
coefficient. The CPTL coefficient is in fact significantlynegative,
indicatinga definiterelationshipbetween minimumstatebankcapital
requirementsand the degree of monopolypower.
Lower state bank capitalrequirementsencouragedthe growth of
state banks, as did an importantchange in the institutionalsetting.
The passageof generalbankincorporationlaws by manystates stimulated the rapid growth of state banks after the mid-1880's. Bank
incorporationin manystateshad been a difficultprocessin the period
followingthe Civil War. At the extreme, the Texas constitutionof
1876 prohibitedthe charteringof state banks,a ban which remained
in effectuntil 1905.As late as 1870, the specialcharterwas, with a few
exceptions, the only means of incorporatinga bank.40
In Illinois between 1870 and 1888 the state legislaturegrantedno
bankchartersat all becauseof the constitutionalprohibitionof special
charters and the absence of a general banking law. This clearly
encouragedthe establishmentof nationalbanks. A free bankinglaw
was finallypassed in Illinois in 1887.41Manywestern states adopted
free bankingprovisionsin their constitutionsabout this time: North
Dakota,South Dakota, Montana,Wyoming, Idaho, and Washington
in 1889, Utah in 1895. Most southernstates adoptedgeneralbanking
rules at this time. Mississippi did so in 1890, Alabamain 1901,
Virginiain 1902. The New Englandstates resisted the trend toward
free state banking until after the turn of the century. By 1910,
however, all eastern states except Delaware and Marylandhad
adopted such laws.42The passageof these laws in most of the states
greatly facilitatedstate bank formation.
The Sylla hypothesis then is correct in identifyingdifferencesin
local monopolypoweras a principalcause of interregionaldifferences
40
Barnett, State Banks, pp. 23, 32.
Dan M. Dailey, "The Development of Banking in Chicago before 1890," (unpublished
Ph.D. dissertation, Northwestern University, 1934), pp. 309, 355.
42 Barnett, State Banks, p. 31. Leonard C. Helderman, National and State Banks (Boston:
Houghton Mifflin Co., 1931) pp. 161-62.
41
896
James
in interestrates. It is also correctin attributingthe narrowingof those
differentialsto the erosion of that power. The hypothesis does not,
however, correctlyanalyzethe mechanismwhich broughtaboutthat
erosion. The lowering of nationalbank capital requirementsby the
Gold StandardAct of 1900 could not have produced the observed
patternof differentials.The mistakewas in concentratingon national
banksand neglectingthe role of state banks,which were much more
numerousand growing more rapidlyat this time.
High minimumcapitalrequirementsfor nationalbanksrestricted
entry in small towns, but state banksdeveloped as substitutes. Differencesin state bankminimumcapitalrequirementswere important
in accounting for cross-sectionaldifferences in interest rates and
stateswith lower capitalrequirementsalso showedgreaterdeclines in
interestratesover time. The erosionof localmonopolypowerwas due
primarilyto the growth of state rather than of nationalbanks. The
organizationof statebankswas encouragedby lower minimumcapital
requirementsand alsoby the passageof generalbankinglawsin many
states in the late nineteenth century.43
CONCLUSIONS
Over the post-bellum period in the United States interregional
interest rate differentialsnarrowedsubstantially.In this paper two
majorhypothesesconcerningthe forcesbehind this trend have been
examinedand tested in a model of local interest rate determination
with an imperfectlycompetitiveloan market.Thus the effects of risk
andlocalbankmarketstructureon the courseof averagelocal interest
rates could be assessed.
Even though risk on loans declined and may have had significant
effects on local interest rates in some areas (such as the western
states),changesin riskdifferentialscouldaccountfor only a smallpart
of the observed changes in interest rate differentials.The national
money marketin the post-bellumperiod had achieved only partial
integration,since in most stateslocal marketstructureand monopoly
power had a significantinfluenceon local interest rates, a conclusion
supportedby both time series and cross-sectiontime series regressions.
The short-termcapitalmarketwas segmented to some extent, but
43 This conclusion is consistent with Rockoff's suggestion that free banking laws improved the
allocation of bank capital in the antebellum period. Hugh Rockoff, "The Free Banking Era: A
Reexamination," Journal of Money, Credit, and Banking, 6 (May 1974), 157-63.
National Money Market
897
it does not appearthat the expansionof the commercialpapermarket
had much to do with weakeningthe barriersto capitalmobility. In
only a few areasdid the introductionof commercialpaper appearto
have significanteffects. The pattern of the narrowinginterregional
interest rate differentialsdoes not fit well with the timing of the
spreadof the commercialpaper market.
The Sylla hypothesis, which emphasizedmarketimperfectionsin
the nationalbankingsystem and the role of the Gold StandardAct of
1900, does not fit the observed patternof interest rate differentials.
The erosion of local monopoly power was the principalreason for
narrowinginterest rate differentials,but, contra Sylla, this resulted
from the growth of state banks. Lower capital requirements,more
liberalregulations,and, afterthe 1880's,the passageof generalbanking laws which made incorporationmuch easier encouragedthe formation of state banks rather than of nationalbanks. These institutionswere primarilyresponsibleforbreakingdownlocal monopolies.
Both the InstitutionalChange and the MarketPower hypotheses
are based on the assumptionthat local monopolypower existed. In
the formercase it was diminishedby the introductionof commercial
paper and the resultantcapitalinflowfrom East to West. Decreased
barriersto capitalmobilityallowed a larger flow of capitalinto high
interest rate areas, thereby promotinginterest rate convergence.On
the other hand, the MarketPower hypothesis focuses on the local
marketstructureratherthan on increasesin the supply of funds. If
state bankswere establishedwith localfunds, then there would have
been a decline in localinterest rateswithoutany capitalinflow.44My
analysis indicates that reductions in the barriers to interregional
capital mobility were much less significantthan were changes the
states made in the legal frameworkof banking;these latter changes
encouragedthe growthof state banksand were, apparently,the most
importantfactorsaccountingfor the decreasesin interest rate differentials.
JOHN A. JAMES, University of Virginia
" The Davis hypothesis emphasizing the spread of the commercial paper market postulates a
net flow of funds from East to West, while, on the other hand, Sylla emphasizes the flow from
the countryside to the cities through the mechanism of bankers' balances. It can be shown that
unless the eastern interregional balance of trade surplus was enormous, close to $1 billion per
year over the period 1900-1910, the direction of short-term capital flow must have been from
West to East, a result consistent with the observations of contemporaries concerning the
concentration of funds in New York through the system of bankers' balances and also consistent
with the function of New York as a financial intermediary. The Market Power hypothesis fits
quite well with this picture of capital flows, but the Institutional Change hypothesis does not.
James. "The Evolution of the National Money Market, 1888-1911," pp. 200-12.