Strategic Behavior in Whiskey Distilling, 1887-1895

by

Karen ClayAssistant Professor

Heinz School of Public Policyand Management

Carnegie Mellon UniversityPittsburgh, PA 15213

kclay@andrew.cmu.edu

and

Werner TroeskenAssociate Professor

Department of HistoryUniversity of PittsburghPittsburgh, PA 15260

troesken@pitt.edu

This Version: May 6, 2002

* - Comments from Sara Ellison, Caroline Fohlin, Rick Geddes, Ed Lazear, Margaret Levenstein,Nolan McCarty, and seminar participants at Caltech, Chicago, Hoover, Stanford, and theEconomic History Association meetings are gratefully acknowledged. The usual disclaimerapplies.

1See, Levenstein, “Price Wars”; Granitz and Klein, “Monopolization”; Porter, “CartelStability”; Genesove and Mullin, “Predation”; Weiman and Levin, “Preying”; and Burns“Predatory Pricing”.

2See Lamoreaux, Merger Movement.

1

1. Introduction

A central issue in industrial organization and economic history is the degree to which

firms use strategic behavior to deter entry and competition. Thus far, the empirical literature on

strategic behavior has documented deviations from static profit maximization (consistent with

dynamic profit maximization), and much of this literature has been historical. There are, for

example, well-known historical studies of collusion and/or strategic behavior in the following

industries: bromine; oil refining; railroads; sugar refining; telephony; and tobacco.1 The

emphasis on history makes sense. Because combinations like Standard Oil and American

Tobacco operated during a time when federal antitrust enforcement was notoriously ineffective,

many such combinations openly engaged in predatory behavior and left a clear trail. Firms today

are not so cavalier. There is, however, a concern about selection bias. All existing historical

studies focus on the trusts that used strategic behavior successfully, even though most trusts

failed.2 As a result, we know quite a lot about American Tobacco and Standard Oil, but very

little about hundreds of less successful combinations. Yet if you want to understand why

strategic behavior emerged and worked in oil, sugar, and tobacco, you also need to understand

why it did not work in other industries.

To this end, we explore strategic behavior in whiskey distilling between 1887 and 1895.

Our goals are twofold. First, like other empirical studies, we present evidence of strategic

3Ellison and Ellison, “Strategic Entry” and Kadiyali, “Entry”.

2

behavior.3 The evidence is both qualitative and econometric. The econometric evidence exploits

an unusual data set that allows us to estimate firm-level demand and cost curves for the Whiskey

Trust, the dominant firm for the period under consideration. These estimates show that the

Whiskey Trust set prices below those that would have been set by a profit-maximizing firm

exploiting all of its market power. The estimates also show that after initiating an exclusive

dealing campaign designed to deter entry, the price-cost markup rose. However, despite some

limited success, these strategies ultimately failed to preempt entry, and the combination was

bankrupt by early 1895.

Which brings us to the second goal of this paper: to understand why strategic behavior

did not work in distilling. The evidence presented below suggests market structure played at

least some role. Because a distillery of minimum efficient scale could be constructed for the cost

of opening a modest restaurant, entry was not easily deterred. Along the same lines, as the data

below show, there were many low-cost substitutes for spirits so that even if narrowly construed,

the trust controlled a large market share, broadly construed, it did not. Compounding the effects

of legal entry and competition from close substitutes, a large federal tax on spirits encouraged the

development of an extensive illicit fringe. Illegal stills cost virtually nothing to construct; appear

to have been widespread; and limited the trust’s market power. Illegal entry increased after the

depression of 1893, which increased the real burden of the tax, and after the federal government

increased nominal tax rates by 22 percent in August 1894. The Whiskey Trust was also subject

to multiple antitrust suits by state and federal authorities. While federal antitrust suits had little

or no impact on market structure and performance, there is some evidence that state antitrust

4See Jenks and Clark Trust Problem and Troesken “Exclusive Dealing”.

5See Downard, Dictionary, pp. 227-28. These figures refer to legal consumption ofspirits. We have no precise knowledge of the consumption of illicit spirits.

3

prosecutions reduced the trust’s market power and brought an end to its use of exclusive dealing.

Except for driving up the real burden of the already large federal excise tax, the depression of

1893 played a negligible role in the demise of the trust.

This paper contributes to the existing literature on the Whiskey Trust in significant ways.

Previous studies employ only price data, and are therefore limited as to what they can tell us

about the trust’s level of market power and the success of its exclusive dealing campaign.4 In

contrast, in this paper, we derive direct estimates of marginal cost, and estimate the trust’s

demand curve directly. To our knowledge, this last point is particularly notable. No other study

of the trusts that we are aware of has assembled enough to data to estimate a firm-level demand

curve.

2. Overview of History and Industry

2.a. A Brief The History of the Whiskey Trust

The focal point of the analysis is the Whiskey Trust, which was among the most

prominent combinations of the late nineteenth century. Three factors contributed to the trust’s

prominence. First, it produced a widely consumed commodity–namely, distilled spirits. In 1880,

the average American adult consumed 2.4 gallons of spirits annually. And even though

Americans consumed more beer (11.1 gallons annually), the majority of absolute alcohol (58

percent) was consumed through spirits because spirits had a much higher alcohol content.5

Second, the Whiskey Trust was large, controlling nearly 90 percent of industry production at its

6See United States, Industrial Commission, p. 77; hereafter cited as IC.

7See Troesken, “Exclusive Dealing”.

8See Downard, Dictionary, p. 213; IC, pp. 76, 168-69; and Lamoreaux, MergerMovement, pp. 99-101.

4

peak.6 Third, rivals accused the Whiskey Trust of using anticompetitive strategies, such as

predatory pricing and vertical restraints.7

The beginnings of the trust date back to the Peoria Pool of the early 1870s, a combination

that was limited to distillers located in central Illinois. A much larger pool, the Western Export

Association, formed in 1881. Industry observers claimed the association was created because

overcapacity had depressed prices. To increase prices, the association set production limits for

members, and members who produced in excess of their limit were required to export the excess

at their own expense. However, from its inception, the pool was beset by cheating, market entry,

and periodic price wars.8

After the pools failed, distillers organized the Distillers and Cattle Feeders’ Trust, better

known as the “Whiskey Trust” in May 1887. Modeled after Standard Oil, the Whiskey Trust was

a bona fide trust so that when a distillery joined the trust it surrendered control of its operations

to a board of trustees. Of the eighty-six distilleries that eventually joined the combination, only

ten or twelve were kept in operation; the remainder were shut down. However, during the 1880s,

state courts raised questions about the legality of trust arrangements. In 1890, fearing dissolution

by state courts, the Distillers and Cattle Feeders’ Trust reorganized as an Illinois corporation, the

Distilling and Cattle Feeding Company. Although no longer a trust in the strict sense of the term,

9IC, pp 75-90; and 171.

10See IC, pp. 78 and 835; and Troesken “Exclusive Dealing”.

11See Troesken “Exclusive Dealing”.

5

the combination was still referred to as the Whiskey Trust.9

The trust entered receivership in January, 1895, and reorganized as the American Spirits

Manufacturing Company, a New Jersey corporation, in the fall of 1895, though this new

combination never realized the market dominance of its bankrupt predecessor. Instead of

reviving a rebate plan it had implemented years earlier, American Spirits chose a new form of

vertical integration and set up a subsidiary company to distribute its products. This more explicit

form of vertical integration had at least two attractive features: it was less susceptible to antitrust

prosecution; and it allowed some difficult monitoring problems associated with the trust’s rebate

program, which is described in detail later in the paper and in other published work.10

The failure of the Whiskey Trust is puzzling when one compares the combination to its

more successful counterparts in other industries. Like Standard Oil and the Sugar Trust, the

Whiskey Trust controlled a large market share; was accused of using strategies to deter entry,

such as predatory pricing, vertical restraints, and violence; and aggressively sought ways to cut

costs. For example, trust-affiliated distilleries were the largest in the industry and were among

the first to adopt the use of copper tubing, which enhanced productivity. The trust also hired a

noted Japanese chemist, Jokichi Takamine, who tried to improve the fermentation process by

using wheat bran rather than barley malt. Why, given all of these similarities with Standard Oil

and the Sugar Trust, which by all accounts were highly successful combinations, did the Whiskey

Trust fail? The analysis below answers this question.11

12See IC, pp. 91, 201-02; and Troesken “Exclusive Dealing”.

6

2.b. Technology and Cost Estimates

The Whiskey Trust produced and sold alcoholic spirits. Manufacturing alcoholic spirits

entailed grinding corn into meal; soaking the meal in water; adding malt, which converted the

corn starch into sugar; and using a small amount of yeast, which initiated fermentation and

converted the sugar into alcohol. After fermentation, the corn mash was twice distilled and

charcoal filtered to yield alcoholic spirits. As this description implies, the manufacture of spirits

employed a fixed-proportions production technology predicated mainly on corn and malt, and

there was little, if any, substitutability across these inputs. Partly because grain was central to the

production of alcoholic spirits, most distilleries located in the Midwest and there was an

especially high concentration of distilleries around Peoria, Illinois. (Fixed costs are discussed

thoroughly later in the paper.)12

Given this technology, we characterize the cost of producing a gallon of alcoholic spirits

(c) as follows:

(1) c = k*P g + J,

where k is the amount of grain, in bushels, needed to produce a gallon of spirits; P g is the price of

grain per bushel; and J is federal tax on spirits per gallon. The price of grain, P g, is a weighted

average of the price of corn and malt, such that,

(2) P g = 1 cP c + 1 mP m,

where, 1 c and 1 m are the cost shares of corn and malt; and P c and P m are the price of corn and

malt, per bushel. Although we do not know the precise cost shares of corn and malt, we do know

that the production of spirits involved more corn than malt per bushel. In keeping with this fact,

13Unfortunately, we only possess data on the aggregate amount of grain used, as opposedto data on the separate amount of corn and malt used. If disaggregated data were available, wecould estimate kc + km individually and construct more precise estimates of marginal cost.

14IC, pp. 90, 185-91, 848.

15IC, pp. 89, 201, and 202. On p. 89, the IC reports: “. . . the special water supply is ofvery great advantage. There is an unlimited quantity of water from wells about 30 feet deep,having a temperature the year around of about 54 degrees. This is used for cooling the warmmash and is much more satisfactory and economical than ice.”

7

the upper-bound cost estimate assumes 1 c = 60 %, and 1 m = 40%, while the lower-bound

estimate assumes 1 c = 95%, and 1 m = 5%.13

Following the observations of industry insiders, we do not consider either labor or fuel as

components in marginal cost; they were, instead, fixed or at least quasi-fixed inputs. Industry

observers considered labor a fixed cost because employment remained the same regardless of

output. Contemporary observers believed that the cost of labor was “not a significant factor” in

distilling, and evidence for this position pointed to the fact that the distilling industry had never

experienced a strike in its entire history through 1900. In any case, there is very little variation in

wages over the period of study period.14 Fuel was much the same. The amount of fuel used did

not vary smoothly with output; and was largely independent of normal variations in production.

According to industry observers at the time, distilleries located in and around Peoria not so much

because of the town’s proximity to coal deposits, but because of its proximity to corn and

underground water sources that provided an “illimitable supply” of pure water at just the right

temperature.15

Table 1 presents the estimates of c and k using annual data. The estimates are constructed

by taking annual average prices and production figures based on monthly data, and then

16See IC, pp. 88-89, 202-5, 215.

8

performing the appropriate multiplication as described in Table 1. Four patterns require

additional comment. First, k was falling over time. This decline reflects the introduction of

copper tubing and chemical innovations hastening the process of fermentation, both of which

improved efficiency in the distilling industry. Second, the federal tax, which was between 90

cents and $1.10 per gallon, was large relative to the before-tax cost of distilling (k*P g ), which

was roughly 10 to 25 cents per gallon. Third, the difference between the upper and lower bound

estimates of the before-tax cost of distilling (k*Pg) is between 20 and 40 percent, while the

difference between the upper and lower bound estimates of the after-tax cost (c) is between 2 and

4 percent. Fourth, the cost estimates derived here do not include other potential determinants of

the marginal cost of distilling, such as coal and yeast. To this extent, in the analysis that follows,

the cost estimates might lead us to overestimate markups and therefore overstate the case for the

trust setting unusually low prices. However, in empirical work elsewhere we show that coal,

yeast, and other miscellaneous components of cost represented a minuscule portion (less than 1

percent) of total marginal cost.16

2.c. Market Definitions

Alcoholic spirits were perfectly homogenous; their content and flavor were identical

across all producers. Almost never consumed directly, alcoholic spirits were usually sold to

rectifiers and distributors, who blended the spirits with water, brown sugar, and other flavors to

create rectified whiskey. In contrast to more expensive straight whiskey, such as bourbon and rye

whiskey, rectified whiskey required no aging and could be consumed immediately.

We suggest two measures of the trust’s market share, one narrow and one broad.

9

According to the narrow definition:

(3) narrow market sharet = (trustqt)/(alcoholic spiritst),

where, trustqt is the trust’s production of alcoholic spirits in year t; and alcoholic spiritst is the

total domestic production of alcoholic spirits in year t (no alcoholic spirits were imported during

the period under consideration). Within this narrowly defined market, substitution was costless;

as noted above, alcoholic spirits were perfectly homogeneous across all producers. According to

a broader market definition:

(4) broad market sharet = (trustqt)/(distilled spiritst),

where distilled spiritst is total domestic production of all types of distilled spirits in year t, which

includes alcoholic spirits, bourbon, brandy, gin, rum, and rye whiskey. Within this broadly

defined market, substitution was possible but not costless. Alcoholic spirits, and their derivative

product, rectified whiskey, were generally cheaper (per unit of alcohol) and poorer tasting than

bourbon, brandy, gin, rum, and rye whiskey. Of course, still broader market definitions are

possible if one considers the production of beer, malt liquor, and wine.

Table 2 presents data on production and market share. As Table 2 shows, a narrow

market definition suggests the Whiskey Trust’s market share peaked at 87 percent in 1893, while

a broad definition suggests the trust’s market share never rose above 45 percent and was as low

as 20 percent in 1894 and 1895. Competition from substitutes for spirits, such as beer, appear to

have played a role in disciplining the trust. This can be seen in Tables 3 and 4, which report

demand curve estimates for the trust. Although other aspects of these demand curve estimates

will be discussed in detail later in the paper, they indicate that the cross-price elasticity of

demand was such that a 10 percent reduction in the price of beer (as proxied by the price of hops)

17See Jenks and Clark, Trust Problem, and Jenks’ discussion of the Whiskey Trust in IC,p. 47.

10

would have lead to a 3 to 9 percent reduction in the demand for trust-distilled spirits. The cross-

price elasticities of demand with other alcoholic products, such as wine and bourbon, were

probably similar.

3. Did the Whiskey Trust Price Strategically?

According to industry observers, the Whiskey Trust initially used (unusually) low prices

to deter future entry and to encourage existing firms to join the trust.17 We refer to this period of

low prices as the low-price regime. The low-price regime begins with the formation of the trust

in May, 1887 and ends in June, 1890, when the trust began its exclusive dealing program. The

analysis proceeds as follows. We first present a formal econometric test, which yields evidence

that the trust set prices lower than those that would have been set by a firm practicing static profit

maximization. We then supplement the econometric evidence with quotations from industry

insiders and observers consistent with the idea that the trust was setting unusually low prices.

To test for unexpectedly low prices, we conducted the following three-part test. The test

is based on the hypothesis that the trust refrained from exploiting all of its short-run market

power because it feared inducing entry in the long run and because it wanted to induce existing

firms outside the trust to join the combination. First, we estimated the trust’s short run demand

curve. We then calculated the implied (static profit-maximizing) price-cost markup for the trust.

That is, the markup that would have occurred had the trust behaved as a static profit maximizer.

Finally, we compared the implied markup to the observed markup. The observed markup is

calculated using cost estimates derived from equations (1) and (2). We calculate the markup as:

18Kadiyali, “Entry”.

11

[p-c]/c, where p is price per unit, and c is cost per unit. This three-step procedure is similar to

Kadiyali’s study of entry deterrence in the photographic film industry. Kadiyali estimates price

and advertising elasticities and then argues that, given these elasticities, observed prices and

advertising expenditures by Kodak were inconsistent with static profit maximization.18

How this procedure works for distilling can be seen by noting that a (static) profit-

maximizing firm, would set price ( p) so that:

(5) p = 1 (1 - 1/0 f )

c,

where, c equals marginal cost; and 0 f equals the price elasticity of the firm’s demand curve.

From this expression, a static profit maximizer confronting a demand curve with an elasticity of

2 would set prices 100 percent above marginal cost, while a firm confronting a demand curve

with an elasticity of 10 would set prices 11 percent above cost. Suppose that the Whiskey Trust

faced a demand curve with an elasticity of 10. If the trust regularly set prices that were less than

11 percent above marginal cost, this would be inconsistent with static profit maximization, and

suggestive of strategic behavior.

Firm-level data on output prices, sales, and factor prices are employed in estimating the

trust’s demand curve and in calculating the observed price-cost markup. The data are monthly,

and extend from April, 1888 through June, 1890, which covers nearly all of the trust’s low-price

regime. Output prices are the list price of alcoholic spirits for trust-affiliated distilleries located

in Peoria, Illinois, where the production of spirits was centered. Data on quantity sold are based

on the trust’s sales of alcoholic spirits. As instruments, we use the federal tax and the price of

19Price and sales data are from IC, pp. 212-15; and 818-25. Corn and malt prices are fromIC (pp. 818-25); United States, Bureau of Labor Statistics, pp. 446-47; United States, Census ofManufacturers; United States, Agriculture, pp. 66-67). All price data have been adjusted forchanges in the general price level using McCusker, How Much.

12

corn and malt per bushel.19

Three different demand specifications are estimated: linear; log-linear; and exponential.

This helps ensure that the results are not driven by the imposition of a particular functional form.

Also, to control for changes in income and changes in the price of close substitutes, a monthly

index of industrial production and the price of hops (a primary input in beer) are included as

exogenous variables in the demand equations.

Table 3 reports regression estimates, the implied markup, and relevant confidence

intervals for the low-price regime. The results are robust to alternative functional forms, with all

three demand specifications yielding elasticity estimates of roughly 4.4. Point estimates imply

that, had the trust behaved as a static-profit maximizer, it would have set prices 29-31 percent

above cost. Using upper-bound estimates of elasticity (those equal to the 95th and 99th

percentiles), the price-cost markup would have been no less than 13 percent. Observed markups,

however, generally fall well below these thresholds, averaging between 4 and 8 percent during

the low-price regime. Figures 1 and 2 plot the observed markups during the low-price regime

using upper and lower-bound cost estimates. These figures also plot the lower-bound estimate of

the implied markup (indicated by the horizontal line at .13). According to these figures, the

observed markup nears the estimated lower-bound implied markup only briefly and is usually 16-

17 percent of the point estimate of the implied markup. The fact that observed markups are so

much smaller than the implied markup constitutes evidence the trust was setting prices

20Demand curves are also estimated using quarterly (as opposed to monthly) data. Theelasticity estimates and implied markups are very similar, and yield the same conclusion: thetrust generally refrained from exercising all of its market power.

21The quotations are from IC, pp. 81, 76 and 81, respectively. Government investigatorsbased these observations and conclusions on the testimony of John McNulta, who acted as thecourt appointed receiver of the trust after it had entered bankruptcy proceedings; Martin R. Cook,a wholesaler liquor dealer; and Charles C. Clarke, the owner of Peoria distillery that was part ofthe trust before its demise.

13

strategically in an effort to deter future entry, to encourage existing firms to join the trust, or for

some other reason.20

Although we cannot rule out the possibility that the trust’s price-setting behavior was the

result of ignorance of the demand elasticity, qualitative evidence does suggest that market entry

was a serious concern for the trust, as well as the pools that preceded it, and that market entry

was prompted by efforts to raise price above marginal cost. A government investigation of the

industry conducted in 1899, found that only if distillers “kept prices low” would they not

“provoke competition.” For example, when the pools that preceded the trust tried to raise prices,

they not only induced cheating among internal members of the pool, they also induced outsiders

to build their own distilleries and enter. Outside entry played a central role in the demise of the

aforementioned Western Export Association, which dominated the distilling industry during the

1880s. Similarly, after the trust abandoned its low-price strategy and began raising prices, “new

distilleries were erected, and it was found impossible to maintain . . . high prices.” Finally, the

same government investigation solicited testimony that the trust often cut prices in an effort to

induce outside firms to join the trust.21

14

4. Exclusive Dealing and the Whiskey Trust

4.a. Overview of the Exclusive Dealing Program

The Whiskey Trust operated an exclusive dealing program between 1890 and 1895. This

program was central to the trust, and engendered much hostility from competitors and

government authorities–federal officials claimed the program violated the Sherman Antitrust Act.

The details of the trust’s exclusive dealing program are described in a previously published

paper, and there is no need to retell these details in all of their complexity here. Suffice it to say

that the trust used rebates to encourage distributers to carry only whiskey made with trust-

distilled spirits. Between 1890 and 1891, the trust set the rebate at five cents per proof gallon.

Between 1891 and the summer of 1894, it raised the rebate to seven cents per proof gallon.

During the fall and winter of 1894, the trust reduced the rebate to two cents per proof gallon.

Compared to the price of spirits, the rebate was substantial. At the time, the before-tax price of

spirits ranged from ten to thirty cents per gallon; the after-tax price of spirits ranged from one

dollar to $1.20.

Some economists argue that firms can use exclusive dealing to increase their rivals’ costs

and deter entry. In the usual anticompetitive story, manufacturers use exclusive dealing to

foreclose scarce distribution outlets. This makes it costly for new firms to enter. If new

companies decide to enter, they must do so as vertically-integrated enterprises, operating both as

manufacturers and as distributors, because incumbents have tied-up all distributing outlets.

Alternatively, new entrants could try to offer similar incentives as the incumbent to encourage

distributors to carry their product. But this too could be costly if pre-existing contractual

arrangements with the incumbent firm impose penalties on distributors who deal with competing

22See Krattenmaker and Salop, “Competition”; and Salop and Scheffman “Raising”.There is a competing view which holds that exclusive dealing promotes efficiency. Among otherthings, exclusive dealing might protect relationship specific investments and prevent competitorsfrom free riding on a firm’s advertising expenditures. See Bork Antitrust Paradox, pp. 193-207;and Posner Antitrust, pp. 171-205. Given the nature of distilling, it seems highly unlikely thatthis Chicago School view of exclusive dealing could have applied to the Whiskey Trust.

15

producers.22

4.c. Testing for Anticompetitive Effects

We perform two tests to see if the trust’s rebate program had anticompetitive effects.

First, we estimate the trust’s demand curve during periods when it offered rebates, and periods

when it did not. If the rebate program prevented entry and competition, the trust’s demand curve

would have become more inelastic after the rebate program began. This tests exploits the same

firm-level data described above, except that these data cover the exclusive dealing regime which

extends from June, 1891 through July, 1894. Second, using the same data, we examine the

behavior of the price-cost markup before and after the implementation of the trust’s rebate

program. If the rebate had anticompetitive effects, the price-cost markup would have risen after

the rebate program began. A caveat is in order, however: to the extent the trust initiated the

rebate program because it had lost (gained) market power, these procedures will understate

(overstate) its anticompetitive effects.

Table 4 reports the demand estimates for the exclusive-dealing regime. The difference in

the estimated elasticity of demand during the limit pricing and exclusive-dealing regimes, are

trivial, both economically and statistically. However, stronger evidence that exclusive dealing

had an effect emerges when we look at the behavior of the price-cost markup over time. Figures

1 and 2 plot the observed monthly mark-up across three regimes, where regime shifts are

23This confirms the results of a previous paper, Troesken “Exclusive Dealing”, whichshowed that after the initiation of the trust’s exclusive dealing regime, there was a significantbreak in the trend in the price of spirits, with prices trending upward at a steeper rate thanpreviously.

16

indicated by vertical lines. The first line indicates the transition from the trust’s limit pricing

regime to the exclusive dealing regime; and the second line indicates the transition from the 90

cent tax rate to $1.10. (Discussion of this final tax-related transition is given later in the paper.)

Figure 1 plots the markup using the upper-bound cost estimate, and figure 2 plots the markup

using the lower-bound cost estimate. In both figures, the data have been smoothed slightly using

Cleveland’s running-line smoother (bandwidth = .2). The figures suggest that the exclusive

dealing program inhibited entry and gave the trust greater market power. Notice that with the

introduction of the rebate, the markup began to trend upward, and this upward trend did not stop

until the tax increase in 1894.23

5. Did Antitrust Enforcement Cause the Failure of the Whiskey Trust?

The evidence just presented suggests the Whiskey Trust behaved strategically in an effort

to deter competition, yet the trust was bankrupt by early 1895. This raises the question: why did

strategic behavior fail to protect the Whiskey Trust from competition? Sections 5, 6, 7, and 8

consider four possible explanations: antitrust enforcement; the depression of 1893; market

structure; and changes in the federal tax on spirits.

The Whiskey Trust was prosecuted for violating both state and federal antitrust laws. The

federal cases against the trust were inspired by wholesalers dissatisfied with the trust’s rebate

system. Wholesalers claimed that the trust refused to redeem their rebate vouchers after they

purchased alcohol from companies not associated with the trust, and lobbied federal officials to

24New York Times, December 22, 1892, p. 9; CT, January 8, 1892 p. 5; In re Greene, 52F. 104 1892; In re Corning et. al., 51 F. 205 1891; U.S. v. Greenhut et. al., 50 F. 469 1892;hereafter all citations to the New York Times are given by NYT.

25See Troesken “Before Sherman” and “Did the Trusts?”.

17

take action against the rebate program. Federal prosecutors responded, and indicted the trust for

violating the Sherman Antitrust Act. According to prosecutors, the rebates constituted a restraint

of trade because they prevented wholesalers from purchasing spirits from competing distillers.

Federal courts, however, consistently ruled that the rebate program did not violate the Sherman

Act.24

State courts were less sympathetic to the trust. In June1890, the Nebraska Supreme Court

sustained a lower court ruling and ordered the Nebraska Distilling Company, a trust-affiliated

distillery, to remove itself from the Whiskey Trust. Anticipating the supreme court’s final

decision, the owners of the Nebraska distillery had reconfigured the plant to produce cereal and

malt several months earlier. Significantly, Nebraska won its case against the Whiskey Trust

without the aid of either state or federal antitrust laws. In prosecuting the trust, the Nebraska

attorney general used what was known as a quo warranto proceeding. Through such

proceedings, state officials were able to revoke the charters of corporations that had violated

provisions of their corporate charters by joining or forming monopolistic combinations.

Attorneys general in California, Illinois, Louisiana, New York, and Ohio used the same legal

device to successfully attack the Cotton Oil Trust; the Chicago Gas Trust; the Standard Oil Trust;

and the Sugar Trust.25

The actions of the Nebraska courts, as well as the adverse decisions against other trusts,

concerned the managers of the Whiskey Trust and prompted them to reorganize as an Illinois

26IC, p. 171.

18

corporation in February, 1890. Oddly, managers of the Whiskey Trust pointed to the suits

against the Sugar Trust as more worrisome than their own legal travails.26 Once organized as an

Illinois corporation, the Whiskey Trust functioned much like a holding company, owning a

majority interest in all of its constituent distilleries. Except for the removal of the Nebraska

Distilling Company from the Whiskey Trust, the distilling industry was no less concentrated than

it had been before the Nebraska ruling. This can be seen more clearly in Table 2, which shows

the trust’s market share over time. Using a narrow market definition, notice that between 1889

(the year of the Nebraska decision) and 1890, market share fell only slightly from 79 to 76

percent, but rebounded in subsequent years. Large reductions in market share did not occur until

1894.

In the summer of 1893, the Illinois attorney general filed suit against the Whiskey Trust,

claiming that the combination was in violation of the Illinois antitrust law. The attorney general

sought to revoke the charter of the trust. Two years later, the Illinois Supreme Court sustained a

lower court, and ordered the dissolution of the Whiskey Trust. Writing for the court, Justice

Bailey argued that by acquiring distilleries for the purpose of “crushing out competition and of

establishing a monopoly,” the trust had “misused and abused the powers granted by its charter,”

and as such, “rendered itself liable to prosecution” and ouster from the state.

The ultimate effects of the Illinois antitrust suit are unclear. On the one hand, it is

possible that the trust discontinued the rebate program because of the Illinois antitrust suit. The

lower-court ruling against the trust was handed down in late-September, 1894; two weeks later,

in early October, the trust announced that it would stop the rebate program. On the other hand,

27A circular issued by the trust explained: “At the request of the patrons of this company,we have decided to discontinue the issuance of all rebate vouchers. . . . While we believe thethat the rebate system has been a great advantage to wholesalers around the country, we willinglyaccede to the wishes of our customers.” According to the Chicago Tribune: “abolishing therebates is regarded by many as a severe blow to the trust, but the opposing forces were sopowerful that the company did not dare refuse their demands.” See Chicago Tribune, October12, p. 8.

28William Newberger sued and won when the trust refused to honor his vouchers. SeeNew York Times, December 21, 1892, p. 1. Similarly in Gottschalk Company v. Distilling &Cattle Feeding Company, 62 F. 901 (1894), a Chicago court allowed Gottschalk to redeem hisvouchers despite the trust’s claims that Gottschalk had purchased from non-trust houses. If theplaintiff had, the court said, it was accidental. Gottschalk had accumulated $35,000 worth ofrebate vouchers.

19

by the time the lower Illinois court ruled in September, 1894, the trust was already in serious

financial trouble. Rumors of bankruptcy and receivership for the trust were heard as early as

1893. By the time the Illinois Supreme Court ruled in June, 1895, the trust had already been in

receivership for several months. Also, the trust’s management claimed that it was customer

dissatisfaction that lead it to abandon its rebate program.27

In addition to the antitrust rulings narrowly construed, however, they might have been

another avenue through which the courts mattered. When the trust refused to pay out on rebates

it had issued, several whiskey wholesalers sued, demanding payment. The trust claimed that it

did not have to pay the rebates because the wholesalers had purchased spirits from distilleries not

affiliated with thrust. The wholesalers claimed otherwise. In one decision issued in 1892 and a

second decision issued toward the end of the trust’s existence in 1894, the courts agreed with the

wholesalers and forced the trust to redeem the rebates. If the courts refused to enforce the

exclusivity of these rebates they would have lost whatever power they possessed to undermine

entry and competition.28

29IC, p. 834.

20

6. Did the Depression of 1893 Cause the Demise of the Trust?

Another plausible explanation for the demise of the trust focuses on the depression of

1893. During that year, industrial production and wholesale prices fell by 25 and 18 percent.

The depression of 1893 might have contributed to the demise of the trust through two avenues.

First, through an income effect, the depression might have reduced the demand for alcoholic

spirits. This assumes alcoholic spirits were a normal good, an assumption consistent with both

qualitative and econometric evidence: during the 1890s, industry observers claimed that more

alcohol was consumed in “good times than in bad”;29 and the regression results reported in tables

3 and 4 suggest industrial production (income) was positively correlated with quantity demanded.

The second avenue, referred to as the “tax effect”, is less direct. By reducing the general price

level by nearly 20 percent, the depression drove up the real burden of the tax (which was fixed at

90 cents per gallon), increasing the cost of spirits and reducing quantity demanded. In addition,

increasing the tax might have caused production costs for the trust to rise relative to illicit

(untaxed) producers. This in turn would have increased black market activity and reduced

demand for legally-produced spirits.

Figure 3, which plots monthly data on quantity sold for the trust, suggests the depression

of 1893 might have been an important factor in the demise of the trust. Before January 1893,

there appears to be a mild upward trend in quantity sold, but after that month, there is a clear

break in the data and quantity sold drops sharply. However, from this graph alone, it is

impossible to disentangle the income effect from the tax effect. To identify the relative impact of

the income and tax effects, we estimate quantity demanded under three counterfactual scenarios.

30To arrive at these counter-factual estimates, we first estimate a demand curve underactual conditions. We then use the resulting coefficient estimates to create counter-factualworlds. In essence we ask, holding everything else constant, what would have happened todemand if the real burden of the tax rose, but industrial production did not fall? What wouldhave happened if the real burden of the tax remained constant, but industrial production fell? And so on.

21

Scenario one identifies the effect of tax changes, independent of the reduction in income

associated with the depression of 1893. This calculation effectively answers the question,

“holding income constant, how much would have been demanded, if the tax (in real terms) had

never risen above the level it reached in December 1892?” Scenario two identifies the effect of

income reductions, independent of any increases in the tax. In effect, this calculation answers the

question, “holding the tax rate constant, how much would have been demanded if industrial

production had never fallen from the level it reached in December 1892?” Scenario three

identifies the joint effect of changes in income and taxes.30

Figure 4 plots predicted quantity demanded under the three counterfactual scenarios and

under actual events. The results suggest the income effect was small relative to the effect of tax

changes. If taxes had remained constant, quantity demanded would have dipped modestly and

recovered strongly in 1894. If industrial production had remained constant, quantity demanded

would have continued to fall, and dropped sharply during 1894 when the government increased

the nominal tax from $.90 to $1.10 per gallon. The effects of tax changes are discussed further

in section 9.

7. Did Market Structure Cause the Demise of the Trust?

In distilling, fixed costs included labor, fuel, the cost of containers, and plant and capital.

That labor and fuel were quasi-fixed inputs is explained above. By law, distillers were required

31IC, pp. 92, 228-30.

32IC, pp. 254-55.

33On the cost of constructing a distillery, see IC, pp. 184, 203, 248-249. On the costs ofopening a restaurant, see Pittsburgh Post Gazette, April 13, 1999, p. F.1.

22

to store and ship alcoholic spirits in barrels. Distillers typically owned the barrels and viewed

them as capital, using the same barrel repeatedly.31 Economies of scale were exhausted at fairly

low levels of output. Industry observers argued that minimum efficient scale was obtained when

a distillery reached the capacity to process three to five thousand bushels of corn a day (about 5

percent of the market). As one distiller explained, “after this limit is reached,” a large distillery

“can not produce any more liquor per bushel of corn” than a small distillery.32 In terms of the

costs of plant and capital, a distillery with the capacity to supply 10 percent of the market could

have been constructed for roughly $2 million 1999 dollars, less than it costs to open a small

restaurant.33

Given the ease of entry it is probably not surprising that the trust’s efforts to raise prices

failed, even if the efforts to increase prices were accompanied by an aggressive exclusive dealing

campaign. When the trust raised prices it induced new firms to enter. These firms circumvented

the trust’s exclusive dealing campaign by entering as vertically integrated enterprises. By doing

so, they created their own distribution outlets and therefore did not have try to break into the

existing distributer-ships locked into carrying trust-distilled spirits. Press accounts and previous

research suggest that as these new vertically-integrated firms entered, they undercut the trust and

gradually eroded its market share and profitability.

34IC, pp.825 and 41.

23

8. Tax Changes and Illicit Entry

8.a. The Costs of Illicit Entry

A still with the capacity to produce a barrel a day could have been constructed for as little

as $1400 (1999 dollars). Although there are no precise data on the production of stills, industry

observers suggested illicit production was as high as 30-40 million gallons per year, or about 40-

50 percent of legal production. In addition, illicit production was not limited to remote,

Appalachian locations, as common folklore suggests. Government investigators in 1900 found

that illicit production “was all over the county,” including major cities such as New York and

Chicago.34

There was also a class of distilleries known as “registered stills.” These were very small

distilleries, located mostly in the South, that were licensed by the government. At large

distilleries (like those in the trust), IRS officials monitored production daily, and they monitored

everything, how much grain was used; how many employees there were; how many barrels were

used; how much alcohol was in the barrels, etc. At registered stills production was only

estimated, and the taxes were paid based solely on these estimates. In the South alone, there

were about 6,500 registered stills. The typical registered still had the capacity to process about 5

bushels of grain per day, while the very largest distilleries in the industry had the capacity to

process 10 to 15 thousand bushels per day. Despite the vast size difference, large distilleries had

a relatively small cost advantage–the pre-tax cost of distilling a gallon of spirits at a distillery

with a 15,000 bushel capacity was no more than 20 percent lower than the pre-tax cost at a

35IC, p. 844.

24

distillery with a 5 bushel capacity.35

8.b. The Tax Change and Illicit Entry

For our time period, the federal government changed the tax on spirits only once: in

August, 1894, the federal government increased the nominal tax from $.90 to $1.10 per gallon. If

this change in the federal tax induced increased entry and production from illicit producers, the

tax changes might have contributed to the demise of the trust. To see if increases in the tax, and

corresponding increases in illicit production, contributed to the demise of the trust, we conduct

two tests. The first test merely establishes the plausibility of the hypothesis that illicit production

increased with the tax. Although there are no data on the actual output of illicit stills, there are

data on the number of stills seized each year by IRS agents. To the extent still seizures reflect

illicit production, one would expect seizures to increase as the tax increased. A simple

regression indicates that following the tax increase, seizures increased by about 40 percent. The

central concern with this approach, however, is that IRS enforcement also probably rose with the

increase in the tax, and it is impossible (with the data available) to disentangle the effects of

increased illicit production and increased enforcement.

In the second test, we estimate the elasticity of the trust’s demand curve following the tax

increase. If the tax increase induced illicit entry, the trust’s demand curve would have become

more elastic in the face of increased competition. Table 5 reports the regression results.

Elasticity estimates are between 52 and 74 for the high-tax period, compared to estimates

between 4 and 5 for the low-price and exclusive dealing regimes. Unfortunately, there are only

seven observations available to estimate demand during the high-tax period. Given this,

25

industrial production and the price of hops have been dropped as exogenous regressors.

Dropping these regressors makes it difficult to compare the elasticity estimates in Table 5 to the

estimates in Tables 3 and 4. To address this concern, the price of hops and industrial production

are dropped from the demand estimates for the low-price and exclusive-dealing regimes, making

it easier to construct meaningful comparisons across regimes. When hops and industrial

production are dropped, elasticity estimates during the low-price and exclusive regimes rise only

slightly, to between 5 and 6. As in the second test, there is also a concern that the effect of

terminating the rebate program might be conflated with the effect of the tax increase. However,

the estimated demand elasticity was not significantly affected by the introduction of the rebate

program–as noted above, point estimates of elasticity are nearly identical across the low-price

and exclusive-dealing regimes.

Section 9. Conclusions

We find evidence the Whiskey Trust behaved strategically. Initially, the trust set prices

low in an effort to deter entry and to induce existing competitors to join the trust. Evidence of

this “low-price” regime is based on the observation that the trust charged prices well below those

implied by static profit maximization. When low prices failed to deter entry, the trust turned to

exclusive dealing. The anti-competitive effects of the trust’s exclusionary rebate program is

confirmed by the behavior of the price-cost markup over time. After the trust initiated the

program, the markup rose steadily.

Despite the trust’s many efforts to deter entry and competition, it ultimately failed to do

so. The trust’s low-price regime might have deterred some entry, but after the trust began to

steadily raise prices during its exclusive dealing regime, new distilleries grew-up, in the words of

36Clearly, many of our ultimate conclusions mirror the arguments Jenks presented manyyears ago in the Trust Problem and his write-up in IC, see especially pp. 45-48. Nonetheless, wealso believe we have extended Jenks in many ways. As noted earlier, Jenks did not have offerany of the evidence we have produced on demand elasticities, price-cost markups, violence, orantitrust enforcement. Nor did Jenks frame his analysis using the terms and tools of modernindustrial organization as we have.

26

one observer, “like mushrooms in the night.”36 Initially, new entrants had to vertically integrate

(into distilling, rectifying, and wholesaling) to get around the trust’s rebate program, but over

time, the trust’s market position eroded to such a point that even firms that were not vertically

integrated were able to enter. In addition to legal entry, the trust confronted competition from a

large illicit fringe, a fringe that was made even larger following increases in the real tax rate in

1893 and 1894. Evidence that the illicit fringe rose with the tax rate comes from an examination

of the price-cost markup and demand elasticities over time. After the tax rate rose, demand

elasticity increased significantly.

There were several factors that undermined the ability of the trust to dominate its market.

First, while federal antitrust enforcement played a minor role in the demise of the trust and its

rebate scheme, state antitrust regulation was probably important. Federal courts ruled that the

rebates did not violate the Sherman Antitrust Act, and had no impact on market structure or firm

profitability. In contrast, state courts forced the trust to adopt new organizational forms; might

have helped prompt the trust to discontinue its exclusionary rebate program; and reduced firm

profitability, at least in the short run. Second, entry barriers in distilling were very low, and the

trust was able to do little to raise them. As a result, it was subject to frequent entry whenever it

tried to raise prices above competitive levels. Third, there was some competition from an illicit

fringe, but it is unclear how important that was. Fourth, competition from close substitutes, such

37See Genesove and Mullin, “Predation”.

27

as beer, limited the trust’s ability to monopolize the relevant market. Finally, the depression of

1893 played a negligible role in hastening the demise of the trust. Demand estimates indicate

that even if the depression had never occurred, the trust still would have failed.

The implications of these results are fourfold. First, as one of several empirical studies of

strategic behavior, this paper documents the use of strategic behavior in a specific industry.

Second, the results here complement the recent findings of Genesove and Mullin.37 Genesove

and Mullin show that the rate of return to predatory behavior in sugar refining was as high as 60

percent; the results of this paper suggest predation had a much lower rate of return in distilling.

The differences in the effectiveness of predation across the whiskey and sugar industries

probably stems from the fact that entry barriers are much higher in sugar than whiskey. Also, the

stark contrast between sugar refining and distilling highlights the importance of considering the

trusts that failed (like the Whiskey Trust) as well as those that succeeded (like the Sugar Trust).

Third, the results here suggest black markets might play a role in limiting the market power of

large, dominant firms.

28

References

Bork, Robert H. The Antitrust Paradox: A Policy at War with Itself. New York: The FreePress, 1993.

Burns, Malcolm R. 1986. “Predatory Pricing and the Acquisition Cost of Competitors,” Journalof Political Economy, 266-96.

Chicago Tribune, various issues, cited in text as CT.

Downard, William L. 1980. Dictionary of the History of the American Brewing and DistillingIndustries. Westport, CT: Greenwood Press.

Ellison, Glenn and Sara Fisher Ellison. 1999. “Strategic Entry Deterrence and the Behavior ofPharmaceutical Incumbents Prior to Patent Expiration.” MIT.

Farrell, Maurice L. 1972. The Dow Jones Averages, 1885-1970. New York: Dow Jones andCompany.

Genesove, David and Wallace P. Mullin. 1998. “Testing Static Oligopoly Models: Conduct andCost in the Sugar Industry, 1890-1914,” Rand Journal of Economics, 29:355-77.

Genesove, David and Wallace P. Mullin. 1999. “Predation and Its Rate of Return: The SugarIndustry, 1887-1914.” Department of Economics, Hebrew University; and Department ofEconomics, Michigan State University.

Granitz, Elizabeth and Benjamin Klein. 1996. “Monopolization and Raising Rivals’ Costs: TheStandard Oil Case,” Journal of Law and Economics, 39:1-49.

Kadiyali, V. 1996. “Entry, Its Deterrence and Its Accommodation: A Study of the U.S.Photographic Film Industry,” Rand Journal of Economics, 27:452-78.

Krattenmaker, Thomas G. and Steven C. Salop. 1986. “Competition and Cooperation in theMarket for Exclusionary Rights,” American Economic Review 76: 109-13.

Lamoreaux, Naomi R. The Great Merger Movement in American Business, 1895-1904. Cambridge: Cambridge University Press, 1985.

Levenstein, Margaret C. 1996. “Do Price Wars Facilitate Collusion? A Study of the BromineCartel before World War 1,” Explorations in Economic History 33:107-37.

McCusker, John J. 1992. How Much is That in Real Money? A Historical Price Index for Useas a Deflator of Money Values in the Economy of the United States. Worcester:

29

American Antiquarian Society.

Miron, Jeffrey A. and Christina D. Romer. 1990. “A New Monthly Index of IndustrialProduction, 1884-1940,” Journal of Economic History, 50:321-37.

New York Times, various issues, cited in text as NYT.

Porter, Robert H. 1983. “A Study of Cartel Stability: The Joint Executive Committee, 1880-1886,” Bell Journal of Economics, 14:301-14.

Posner, Richard A. Antitrust Law: An Economic Perspective. Chicago: University of ChicagoPress, 1976.

Salop, Steven C. and David T. Scheffman. 1983. “Raising Rivals’ Costs,” American EconomicReview 73: 267-71.

Thorelli, Hans B. 1955. The Federal Antitrust Policy: Origination of an American Tradition. Baltimore: Johns Hopkins University Press.

Troesken, Werner. 1995. “Antitrust Enforcement Before the Sherman Act: The Break-up of theChicago Gas Trust Company,” Explorations in Economic History, 32:109-36.

Troesken, Werner. 1998. “Exclusive Dealing and the Whiskey Trust, 1890-1895,” Journal ofEconomic History, 58:755-78.

Troesken, Werner. 2000. “Did the Trusts Want A Federal Antitrust Law? An Event Study ofthe Sherman Antitrust Act and State Antitrust Regulation,” in Public ChoiceInterpretations of American Economic History. New York: Kluwer Academic Press.

Troesken, Werner and Karen Clay. 1999. “Did the Trusts Have Market Power? Evidence fromDistilling, 1881-1899.” University of Pittsburgh and Carnegie Mellon University.

United States. Annual Report of the Commissioner of Internal Revenue. Washington: Government Printing Office. Various years, 1881-1899. (Cited in text as IRS.)

United States. 1900 Department of Commerce and Labor. Bulletin of the Bureau of Labor. No.29 July, 1900 Washington: Government Printing Office. (Cited in text as BLS 1900.)

United States. 1900. Department of the Treasury. Bureau of Statistics. Monthly Summary ofCommerce and Finance of the United States. January, 1900. 56th Congress, 1st Session. House of Representatives. Doc. No. 15, Part 7. Washington: Government PrintingOffice.

30

United States. 1901. Department of the Treasury. Bureau of Statistics. Statistical Abstract ofthe United States, 1900. Washington: Government Printing Office.

United States. 1900. Industrial Commission. Reports. Volume 1. Trusts. 56th Congress, 1st

Session. House Documents. Volume 93, No. 476, Part I. (Cited in text as IC.)

United States. 1902. Department of Commerce and Labor. Bulletin of the Bureau of Labor. No. 39. March, 1902. Washington: Government Printing Office. (Cited in text as BLS1902.)

United States. 1902. Census of Manufacturers. Part III. Special Reports for Selected Industries. Washington: Government Printing Office.

United States. 1904. Department of Commerce and Labor. Bulletin of the Bureau of Labor. No. 53. July, 1904. Washington: Government Printing Office. (Cited in text as BLS1904)

United States. 1937-38. Department of Agriculture. Agricultural Statistics. Washington: Government Printing Office. Various years, 1937-1938.

Warren, George F., Frank A. Pearson, and Herman M. Stoker. Wholesale Prices for 213 Years,1720 to 1932. Ithaca, NY: Cornell University, Agricultural Experiment Station, Memior412.

Weiman, David F. and Richard C. Levin. 1994. “Preying for Monopoly? The Case of SouthernBell Telephone Company: 1894-1912,” Journal of Political Economy, 102:103-26.

31

Table 1Cost Estimates

c*1 c =.95, 1 m =.05 c*1 c =.6, 1 m =.4

year k P c P m J P g k*P g k*P g + J P g k*P g k*P g + J

188118821883188418851886188718881889189018911892189318941895189618971898

.27

.26

.26

.26

.25

.24

.24

.24

.24

.23

.23

.23

.21

.22

.23

.23

.22

.22

.63

.67

.53

.52

.43

.37

.39

.46

.35

.40

.59

.46

.40

.43

.40

.26

.26

.32

1.1.83.78.74.74.76.89.88.76.75.93.80.78.74.69.56.54.62

.9

.9

.9

.9

.9

.9

.9

.9

.9

.9

.9

.9

.9.981.11.11.11.1

.65

.68

.54

.53

.44

.39

.42

.48

.37

.41

.60

.48

.42

.45

.42

.27

.27

.33

.18

.18

.14

.14

.11

.09

.10

.11

.09

.10

.14

.11

.09

.10

.09

.06

.06

.07

1.081.081.041.041.01.991.001.01.991.001.041.01.991.081.191.161.161.17

.80

.73

.63

.61

.55

.52

.59

.63

.51

.54

.72

.60

.55

.56

.51

.38

.37

.44

.22

.19

.16

.16

.14

.13

.14

.15

.12

.13

.17

.14

.11

.12

.12

.09

.08

.10

1.121.091.061.061.041.031.041.051.021.031.071.041.011.111.221.191.181.20

Variable definitions:

k = bushels of grain (corn and malt) required to produce a gallon of spirits.

P c = the price of corn, $’s per bushel.

P m = the price of malt, $’s per bushel.

J = the federal tax on spirits, $’s per gallon.

P g = price of grain, $’s per bushel. Formally, P g = 1 cP c + 1 m P m, where 1 c and 1 m are the costshares of corn and malt. By assumption, 1 c =.95 and 1 m =.05 under the lower-bound costestimate; and 1 c =.6 and 1 m =.4 under the upper-bound cost estimate.

c = the after-tax cost of producing a gallon of spirits. Formally, c = k*P g + J.

Sources: IC, pp. 815-19, and 843; ; BLS (1902, pp. 446-47); United States (1900); and UnitedStates (1938, pp. 66-67).

32

Table 2Production, Market Share, and Yield: 1879-1900

Year ending,March 31

Industry production Trust production & market share

AlcoholicSpirits

(1)

DistilledSpirits

(2)

Alcoholic Spirits

(3)

MarketShare1(3)/(1)

MarketShare2(3)/(2)

1879188018811882188318841885188618871888188918901891189218931894189518961897189818991900

51.1 59.7 65.1 61.1 54.6 53.3 50.2 46.4 45.4 47.2 48.9 52.3 57.3 60.3 59.4 54.7 41.8 44.6 40.3 42.7 50.2 52.0

69.0 86.8 112.5 110.3 83.3 76.2 76.4 80.5 80.0 73.6 86.3 106.1 116.1 118.3 127.9 101.9 84.5 88.0 70.7 78.8 96.0 107.0

30.0 38.7 39.7 47.1 45.5 51.8 30.5 17.1

0.63 0.79 0.76 0.82 0.75 0.87 0.56 0.41

0.410.450.370.410.390.410.300.20

Note: production and output numbers are in millions of gallons.

Sources: IC, pp. 89 and 815; IRS, 1881-1898; and United States (1900, p. 32).

33

Table 3: Demand Estimates: Low-Price Regime

Variablemean

(std dev) linear exponential log-linear

Quantity

ln(quantity)

Price of spirits

ln price of spirits

Price of hops

ln price of hops

Industrial production

ln industrial production

Constant

Price elasticity of demand 95% confidence interval [low,hi] 99% confidence interval [low,hi]Implied price-cost markup implied markup using 95% [hi] implied markup using 99% [hi]

3.05(.61)1.10(.20).83

(.03)-.19(.04).15

(.01)-1.89 (.07)46.0

(4.02)3.83(.09). . .

. . .

. . .

dep.var.. . .

-15.8(5.18)

. . .

16.6(11.9)

. . .

.083(.022)

. . .

15.9(3.61)4.29

[1.37,7.22][0.32,8.27]

.30[.16][.14]

. . .

dep.var.

-5.47(1.66)

. . .

5.44(3.82)

. . .

.028(.007)

. . .

3.53(1.11)4.54

[1.46,7.40][0.65,8.42]

.28[.16][.13]

. . .

dep.var.. . .

-4.70(1.42)

. . .

.929(.612)

. . .

1.28(.330)-.103(.315)4.70

[1.76,7.64][0.71,8.69]

.27[.15][.13]

Observed markup using upper- bound cost estimate

Observed markup using lower- bound cost estimate

mean(std dev)

[95th %tile]mean

(std dev)[95th %tile]

.039(.037)[.103].074

(.039)[.150]

No. of obs.Adjusted-R 2

Estimation

26. . .. . .

26.50

2SLS

26.53

2SLS

26.54

2SLS

Instruments price of corn; price of malt; federal tax; 1888 dummy

note: unless otherwise indicated, variables in parentheses are standard errors.

34

Table 4: Demand Estimates: Exclusive-Dealing Regime

Variablemean

(std dev) linear exponential log-linear

Quantity

ln(quantity)

Price

ln price of spirits

Price of hops

ln price of hops

Industrial production

ln industrial production

Constant

Price elasticity of demand 95% confidence interval 99% confidence intervalNo. of observationsAdjusted-R 2

Estimation

3.15(.73)1.12(.25).87

(.03)-.14(.04).20

(.02)-1.64(.07)55.0(7.2)4.00(.11). . .

. . .

50. . .. . .

dep.var.. . .

-15.2(7.21)

. . .

5.03(4.89)

. . .

.05(.01). . .

11.8(3.17)4.19

[0.19,8.20][0.00,9.53]

50.12

2SLS

. . .

dep.var.

-5.56(2.46)

. . .

1.90(1.67)

. . .

.01(.006)

. . .

4.79(2.37)4.84

[0.54,9.14][0.00,10.6]

50.10

2SLS

. . .

dep.var.. . .

-4.70(2.10)

. . .

.35(.30). . .

.79(.35)-2.12(1.55)4.70

[0.48,8.92][0.00,10.3]

50.13

2SLS

Instruments price of corn; price of malt; tax; 12/92-1/93 dummy

Variables in parentheses are standard errors.

35

Table 5Demand Estimates: High-Tax Regime

Variablemean

(std dev) linear exponential log-linear

Quantity

ln(quantity)

Price

ln price

Constant

Price elasticity of demand 95% confidence interval 99% confidence intervalNo. of obs.Adjusted-R 2

Estimation

1.24(.58).04

(.75)1.03(.01).03

(.01). . .

. . .

. . .

. . .7

. . .

. . .

dep.var.. . .

-63.0(18.5)

. . .

66.4(19.1)52.3

[12.9,91.8][0,114.3]

7.59

2SLS

. . .

dep.var.

-69.4(21.3)

. . .

71.8(22.0)71.5

[15.1,127.7][0,159.8]

7.68

2SLS

. . .

dep.var.. . .

-74.4(23.3)2.54(.80)74.4

[14.4,134.3][0,168.3]

7.66

2SLS

Instruments price of corn; price of malt; federal tax

Variables in parentheses are standard errors.