Cigarette Money and Black-Market Prices
during the 1948 German Miracle
Vincent Bignon*
February 2009
‘To paraphrase an old maxim – “money isn’t everything” –
but in Frankfurt, cigarettes are’
New York Herald Tribune1, 5/17/1947
‘Monetary history records a bewildering variety of
commodities that have served as mediums of exchange-from
the wampum beads used by American Indians to the
cigarettes and cognac used in Germany after World War II.’
Friedman, 1951, p. 204
Abstract
This paper is an empirical study of the distribution of black-market prices among 120 Bavarian locations at two dates, the
beginning of July 1947 and the end of June 1948. It shows huge differences in the liquidity of the sample goods when
measured by the coefficient of variation or by the number of locations in which those goods were traded. The main finding is
that the liquidity of cigarettes was high when measured by either the coefficient of variation or the number of counties that
traded them. This made cigarettes special, even when compared with a pure fiat object such as the U.S. dollar. Consistent with
the modern theory of money, the high liquidity of cigarettes is indicative of their use as currency.
Drawing on the work of nineteenth-century economists such as Menger (1892), monetary
theory had recently claimed that the greater liquidity of money is linked more to its role as a
medium of exchange than to its issuance by the state (Kiyotaki and Wright, 1989). Yet no
empirical proof of this theoretical claim has emerged. This is hardly surprising given that,
throughout modern history, most monetary objects had been issued by a state or a sovereign.
One exception, mentioned by Friedman (1951), was the use of cigarettes as money in post-war
*
I thank participants of seminars at the Bank of Japan, Barcelona, Oxford, of the 2007 APHES conference in
Lisbon and of the 2009 ASSA meeting in San Francisco for their comments and insights. I especially thank Marc
Flandreau and Mike Bordo for very useful discussions and comments. I am also grateful to François Bertrand,
Régis Breton, Florence Jusot, Pilar Nogues-Marco, Mariana Rojas-Breu. The usual disclaimer applies. This draft
was completed while I was visiting the Mercatus Center at George Mason University.
Associate professor IUFM– University of Paris 12, Research fellow EconomiX – University of Paris 10 –
Nanterre. Address for correspondence : Université Paris 10, Economix, Bâtiment K, 200 avenue de la République,
92001 Nanterre cedex. E-mail: [email protected]. Webpage: http://economix.uparis10.fr/en/membres/index.php?id=54
1
in “Cigarette is King In Frankfurt Black Market”, by Edwin Hartrich.
1
Germany that were neither issued nor promoted by any state. One contribution of this paper is
to study the cigarette money episode in post world war II Germany to provide the first
empirical assessment of the statement that money’s liquidity stems from its role as medium of
exchange and not from the specificity of its issuer.
The description by Radford (1945) of the cigarette money episode in the prisoners of war’s
camp of Nazi Germany is the most popular and oft-cited examples used to introduce the
advantage of using a money rather than barter. Many textbooks refer to it (e.g. Mankiw, 2007,
p. 339; Mishkin, 2008, p. 51). Despite its entertaining content, Radford’s evidence is entirely
anecdotal. Moreover the use of cigarette as money in Europe of the 1940s was much more
pervasive than suggested by this example. Indeed authors claimed that they were accepted in
payments in post–World War II Europe (Senn 1951), notably in Austria (Rosen 1947), and
Germany (Friedman 1951; Klein 1976; Kindleberger 1984).2 The German case is especially
interesting as it involves the whole German economy and lasted at least three years. Yet our
knowledge of it rests either on sparse mentions or on qualitative or scattered quantitative
evidence. This paper proposes the first comprehensive history of the post-war German episode
and provides quantitative evidences on it. To this end I use new qualitative sources and an
original data set of black-market prices for eight goods in 120 (of 143) Bavarian markets. The
sample allows showing that the liquidity of cigarettes was far higher than that of any other
good.
To provide suggestive evidence of the monetary character of cigarettes, I make use of the
claim that money is the most liquid good (object) of the economy and compute various
measures of the liquidity of the sample goods. Two main criteria are used to measure liquidity:
(1) a coverage ratio, measuring the probability that a given good is sold on a given black
market; and (2) the price dispersion across locations. On one hand the first criterion is a proxy
2
The following quote indicates that Germany was unique: “While the American cigarettes will fetch a high price
in almost every country of Europe, it is only in Germany that it has been elevated to the dignity of a medium of
exchange” (in “The Cigarette Standard,” published 28 February 1947 by the Herald Tribune).
2
for liquidity as the mere existence of a price quote on a local market indicates that a significant
number of trades in that location involved this good. It indicated that cigarettes, like butter and
coffee, benefited from wide coverage; in contrast, U.S. dollars and the military payment
coupon (the two fiat objects of the sample) were quoted in only half of the districts. On the
other hand, the second criterion is a proxy for liquidity if agents arbitraged among markets. It
indicated that the price dispersion was far less for cigarettes than all other goods, even fiat
objects that were perfectly homogenous in quality. This shows that using cigarettes to trade on
a local black market reduced two types of risk: (1) the risk of being unable to sell the holding
and (2) the risk of paying too much because of price dispersion.
This paper also tackles two other important related issues. First, it provides an explanation
of why Germans switched from using a fiat currency (the Reichsmark) to an economy with
barter and commodity money. Second, interpreting the level of price dispersion as an indicator
of liquidity is viable only if the black markets were integrated (i.e., people arbitraged among
markets). The last section therefore tests whether the variance in our sample of prices could be
explained by some type of market segmentation. The results clearly allow us to reject this
hypothesis.
A few other papers have studied the cigarette money episodes, but they mainly rested on
qualitative evidence (Radford 1945) or gave scant coverage to actual price ranges (Schmölders
1973; Botting 1985; Bub 2004). Rosen (1947) is the only paper that also uses black-market
prices to study the demand for cigarettes in 1945 Austria. Drawing on time-series data, Rosen
showed that the evolution of cigarette prices on Vienna’s black market correlated with the price
of the U.S. dollar before the Austrian monetary reform of December 1945 but not after. He
interpreted this switch as evidence of a change in the dynamics behind cigarette demand—a
change explained by cigarettes losing their monetary function during the stabilization.
Although interesting, this analysis is somewhat questionable because we have no proof that, in
3
Austria, the dollar was actually a medium of exchange and not merely an asset. Hence the
correlation could simply mean that cigarettes were an “asset” before the stabilization but not
after, which could have been the case if people chose to keep their wealth in real goods (or
foreign currency) rather than fiat money. To avoid this difficulty, we take a different approach
and measure the liquidity of goods on a spatial basis. This approach reveals that the behavior of
the prices of cigarettes (and dollars) in 1948 Germany differed sharply from that in 1945
Vienna, which is only 400 kilometers away.
The rest of the paper is organized as follows. Section 2 uses the theoretical literature to
derive the indicators used to measure liquidity. Section 3 provides new facts on the emergence
of black market trades and commodity money. Section 4 presents the dataset and computes the
measures of liquidity. Section 5 rules out the hypothesis that local markets were segmented,
and section 6 concludes.
2. What makes money special? Money and the liquidity of the goods That money should have greater liquidity than any other good is an idea that dates back at
least to Carl Menger (1892) and his idea of comparing the saleableness of commodities. The
point that is most relevant to our analysis is that “when the relatively most saleable
commodities have become ‘money’, the event has in first place the effect of substantially
increasing their originally high saleableness” (p. 250). Menger further argued that this greater
saleableness affected relative prices—in other words, that the relative prices of commodities in
terms of the commodity that is used as money experienced smaller variations than the relative
prices expressed in other commodities.3 He also noted that a precondition was “a sufficient
supply of [precious metals] had been collected and introduced into commerce” (p. 254).
3
The exact quote is as follows: “This development [of the use of precious metals as money] was materially helped
forward by the ratio of exchange between the precious metals and the other commodities undergoing smaller
fluctuations, more or less, than that existing between most other goods, a stability which is due to the peculiar
4
Search models of money have recently revivified Menger’s intuition that money is the
most liquid good (Kiyotaki and Wright 1989, 1993). The notion of liquidity in those models
hinges particularly on the good’s acceptability within the population: on the proportion of
agents in the economy who accept it in payment. Building on this concept, Kiyotaki and
Wright show that specialization in productive activities makes agents willing to trade their
goods against an intermediary of exchange that is widely accepted by the population (i.e., the
most liquid good). They further prove that favorable intrinsic properties are not required for a
good to be used as money. Later models used alternative specifications of the intrinsic property
components and showed that, provided the intrinsic component is not too bad, this basic result
holds (Renero 1999; Cuadras Morato 1997). This indicates that the notion of liquidity
(understood as the acceptability of goods by a population) is, to a certain extent, a substitute for
the intrinsic component of the goods traded.
In models with fiat money, the liquidity component of money is explained by a strategic
interactions argument, i.e. agents use a good to pay their purchases because others also do. This
bootstrap effect has the virtue of speeding up the circulation of money and therefore to
accelerate the completion of trades. This is precisely the reason why in equilibrium an object
can have relatively bad intrinsic properties and still be used as money. Strategic interactions
therefore create some indeterminacy on the choice of an object/goods (among the set of goods
available) used as money. A natural question is therefore to inform on the criteria that can
influence this choice, i.e. on the reasons that explained why an object is selected as a money.
Theoretically, this question maps into the question of the uniqueness of an equilibrium in
which a particular good is used as money. In a model with fiat money, Li and Wright (1997)
show that the mere existence of a sufficiently high proportion of agents – they called
government – who is stuck to the strategy of always accepting the fiat object in payment can
circumstances attending the production ,consumption, and exchange of the precious metals, and is thus connected
with the so-called intrinsic grounds determining their exchange value” (p. 254).
5
eliminate the barter equilibrium (that always exists because of the bootstrap effect) in some
regions and therefore make the monetary equilibrium the unique equilibrium. This argument
also applies to models with commodity money as long as the prior demand for the commoditymoney (i.e. the one for consumption purposes) is sufficiently high. Some models did study the
criteria that influence the selection of a good as money in equilibrium. Clearly the demand of a
good for consumption purposes has an impact on its selection as a mean of payments.
The liquidity component specific to money depends on the proportion of agents that
demand it (Cuadras Morato and Wright 1996). In a commodity money model, this demand
have two components, either consumption purposes or to use it as an intermediary, as a mean
of payments. The combination of the consumption motive with the demand linked to the
monetary character of the good makes it the most widely acceptable among agents and can
even induce agents to renounce its consumption. Drawing on the work of Burdett, Trejos, and
Wright (2001), Bignon (2004) extended this model to analyze how the use of such commodity
money can be linked to the proportion of that good’s consumers in the population and their
utility in consuming it. The results shows that if a sufficiently high proportion of the population
accepts one good (say, cigarettes) as a consumption good, then nonsmokers will have an
incentive to use the good as money in order to reduce search costs. But smokers will renounce
consumption only if it offers sufficiently low utility. If both conditions (widespread acceptance
and low utility) hold, then a commodity could become money: the whole population could use
it to pay for purchases of other goods.
The key element in such models is the explicit modeling of reasons that explain the
demand for a medium of exchange, that is, for using a (indirect) trade strategy instead of direct
barter.4 In the Kiyotaki and Wright (1989, 1993) models, specialization in productive activity
4
Another important feature of those models is their insistence on the difficulty of using payment schemes other
than monetary exchange. These include gift-giving economies (Kocherlakota 1999), where a record-keeping
device is set up to keep track of the completion of trades, or arrangements whereby credit is repaid with credit. We
6
reduced the return of direct barter by making it less likely to succeed on the marketplace. This
results hinges, then, on modeling the scarcity of double coincidences of wants between traders.
Thus agents become willing to substitute their holding against a medium of exchange, a good
that has a greater acceptability in the trade process. The literature shows that other frictions
impeding direct barter can be incorporated into the decentralized search framework to account
for the demand for money as a medium of exchange (for a survey, see Shi 2006). Such frictions
include imperfect information on the quality of the goods produced (Williamson and Wright
1994); inefficiency in the bargaining process (Engineer and Shi 2001; Berentsen and Rocheteau
2002); and impediments to the use of alternative payment arrangements (such as IOUs) when,
as in a “dated goods” economy, there is a limited commitment to repay the debt through direct
barter (Kiyotaki and Moore 2004).
These considerations underscore that the equilibrium liquidity of the good(s) used as
money must be greater than the liquidity of all the other goods. According to the literature just
reviewed, two measures can be used to proxy a good’s liquidity: (1) a measure of price
dispersion to account for differences in the relative prices; and (2) an indicator of the good’s
coverage—the ease with which people could buy and sell the good.
3. Cigarette money in post­war Germany The emergence of the cigarettes as money has roots in the institutional and economic
background of the German economy. On the institutional side, the Allies continued the system
introduced during the war to deal with shortages of civil goods and to prevent inflation. Two
major shifts adversely affected the efficiency of this system and led to the emergence of a
barter economy. First, the rationing system failed to provide an adequate standard of living and
lead to a surge of black-market trades (section 3.1.). Second, the fiat money lost its value—not
because of inflation but rather because price controls reduces dramatically the quantity of
do not discuss this dimension here because the very nature of black-market trade makes such alternatives
extremely costly.
7
goods sold on the official market (section 3.2). However and contrary to what we might
observe in modern monetary economy, the use of cigarettes as money did not rule out the use
of barter or other commodity money (section 3.3) although there is much qualitative evidence
of a special role for cigarettes (section 3.4.)
3.1. Caloric value in the rationing system and the rise of black-market transactions
Most food and clothes were distributed through a rationing system. Consumers gained the
right to purchase a predetermined quantity within a specified network of shops against a given
amount of rationing tickets and a money payment in Reichsmarks. The quantities of the
rationed goods sold were set each month by an authority that allocated the available goods
among the various groups of consumers. This means that, with food, the caloric value
distributed depended on the level of production and the producers’ ability and willingness to
deliver. Figure 1 shows that, between August 1945 and June 1948, the caloric value of a
normal consumer rarely exceeded 1,500 calories a day.5 As a consequence, in June 1947 only
22% of polled city dwellers in the U.S. zone agreed that the food supply was adequate (versus
64% of those living in a small village).6
1600
Caloric value, US zone
1500
1400
1300
1200
1100
1000
Effective ration distributed to a normal consumer
900
1945:08 1946:02 1946:08 1947:02 1947:08 1948:02
Figure 1: Caloric value of the monthly ration of a normal consumer, U.S. zone
(Source: Various issues of the statistical annex of the Monthly Report of the Military Governor of the U.S. zone)
5
The other consumer groups included heavy workers such as miners, pregnant women, and various categories for
children and teenagers. A typical normal consumer was a white-collar worker in a service industry.
6
See OMGUS report #363. ICD.
8
The insufficient calories distributed through the rationing system were one factor leading to
the development of illegal exchanges. Various types of black-marketing flourished as early as
the summer of 1945. Indeed, a U.S. Army note (dated 30 July 1945) reported that farmers
refused to deliver food to the rationing system even when the German officer was accompanied
by U.S. soldiers. The note further indicated that this entailed “a noticeable loss of food from the
normal market channels” because of “the people from the city going to nearby farms and
buying on the spot.”7 According to a May 1946 poll of the U.S. occupation authority, 18% of
the sample said they travelled regularly to the countryside in order to complete their food
ration.8 Illegal markets for foods also emerged in the central places of big cities and near the
train station of each town.9
The market share of illegal markets was large for foodstuffs, clothes, and shoes. In
November 1945, an official report of the U.S. Army gauged that 20% of all exchanges for
foodstuffs were illegal;10 six months later, illegal exchanges amounted to 50% of the trading
for some goods.11 Experts in the U.S. military government believed that barter trade was used
for between a third and a half of all business transactions (Menderhauser 1949, p. 655).
Ostrander, head of the price control section of the U.S. economic division, estimated in May
1947 that “the amount of food lost to control distribution through such channels is substantial,
perhaps 20%” and that “the black market activity [has] assumed increasing significance during
the last year” (Herald Tribune, 21 May 1947).
3.2. The Reichsmark, price controls, and inflation
The money stock—currency and deposits—had surged during the war, increasing from
56.4 billion marks in 1938 to about 300 billion in May 1945 (Hansmeyer and Caesar 1976, p.
7
Report entitled “BM operations in the American occupied zone” (07/30/1945), OMGUS archives, file 1/177 3/8
shelf mark BICO C+J.
8
See Merritt (1970), “Public Opinion in Occupied Germany, the OMGUS Survey,” University of Illinois Press.
9
See for example the various reports of the Bavarian food ministry at the Bayerische Hauptstaatsarchiv (Munich).
File MELF 1118.
10
Monthly Report of the Military Governor of the U.S. Army in Germany #5, December 1945, section “Trade and
Commerce,” available at the Institut für Zeitsgeschichte, Munich.
11
Monthly Report of the Military Governor #11.
9
418). In 1945, the money stock (M2) amounted to 230% of 1944 GDP while the stock of
currency represented 40% of 1944 GDP. Various measures reduced M2 by about 127 billion
marks, and during spring of 1947 it was gauged to be 173 billion (including 79 billion marks of
currency).12
To avoid the development of hyperinflation, the Allies continued to work with the system
of price controls set up by the Nazi regime in 1936 and reinforced during the war. Two main
changes were instituted. First, there was no central (German) authority responsible for
managing the system; instead it was managed by an Allied Control Authority (ACA) while the
(German) regional price offices remained responsible for ruling on price increases.13 Second,
the Allies strictly monitored the regional price offices and their implementation of regulations.
10%
160
150
8%
140
130
4%
120
2%
110
Baseyear 1936
Monthly inflation rate
6%
100
Jun-48
Apr-48
Feb-48
Oct-47
Dec-47
Jun-47
Aug-47
Apr-47
Feb-47
Oct-46
Dec-46
Jun-46
Aug-46
Apr-46
Feb-46
Oct-45
Dec-45
Jun-45
Aug-45
Apr-45
-2%
Feb-45
0%
90
80
-4%
70
Monthly inflation rate
Consumer price index (without housing prices and rents)
Figure 2: Evolution of the Bavarian consumer price index and of the monthly inflation rate of prices
(Source: Bayerische Statistisches Landesamt)
On 6 February 1946, the ACA issued a statement that “prices are to be maintained at the
level before occupation” and “price increases shall only be permitted as an exception”. Figure 2
shows the evolution of the Bavarian consumer price index (CPI), which indicates that regional
12
This includes 73 billion in currency in 1945 (70.3 billion in notes, 1.6 billion Reichsmarks in divisionary coins,
and 1.1 billion Rentenmarks) and 12 billion in banknotes issued by the occupation authorities (Allied
Reichsmarks), from which one must subtract 6 billion in banknotes that were destroyed (Samuelson 1972, pp.
171–2). Petrov (1966) disputed the figures released on Allied Reichsmarks issued and argued that the Soviets had
much inflated the money stock in order to destabilize the economy of the Western part of Germany. I found no
document in the archives that corroborates this theory; moreover, the public budget was balanced already by 1947.
13
See Menderhauser (1949, pp. 647–8).
10
authorities were more lax about price increases than recommended by the Allies.14 The reason
was that “exceptions” allowed price increases to firms that encountered losses. Under the first
three years of Allied rule (from May/June 1945 to June 1948), the CPI increased by 27%
compared to an increase of only 20% between 1936 and May 1945. The annual inflation rate of
official prices hardly qualifies as hyperinflation, since prices went up by only 13%, 4%, and
10% during the first three years of occupation.15
3.3. Black-market traders preferred barter or commodity moneys to fiat money
The low level of calories delivered by the rationing system motivated people to forage for
food in the countryside, even though it was illegal (Diskant 1989, p. 555). As remarked in the
November 1946 Report of the U.S. Military Governor: “Far more than elsewhere, black
markets in Germany are sporadic and erratic, consisting almost wholly of many isolated
transactions.”16 Various goods were used to pay for black-market purchases.
That consumers abandoned fiat money does not come as a surprise. The combination of a
huge inflation in the money stock and tight control of prices naturally resulted in a very low
trade value for the Reichsmark. Although the official rationing system was still characterized in
terms of the fiat money, even the U.S. occupation authority recognized in April 1947 that “The
RM is not yet completely worthless but its value is greatly undermined. Only the fact that the
RM retains a certain usefulness in connection with the ration coupon gives it some continuing
value.”17 As a consequence, Reichsmarks notes traded at a discount on post-war black-markets,
where prices were about a hundred times the legal prices.18
In addition to trading their Reichsmarks at a (huge) discount on the black markets, people
also renounced using the fiat money in illegal trades. Sellers and especially farmers insisted on
14
The price offices could approve of price variations if the producer presented evidence that the official price did
not cover the production cost.
15
Computations were done from June of year T to June of year XT+ 1.
16
Monthly Report #17, p. 44. See Bignon (2007) for an account of the various illegal market structures.
17
OMGUS, Econ division, Trade and commerce branch, 04/23/1947. File 1/ 194-1/6, mark BICO US Cust. Gp.
18
Menderhauser (1949).
11
being paid in kind. An April 1947 U.S. Army report indicated that “not only does the German
farmer now have each larger reserves [of Reichsmarks] than the urban population, but at the
same time ration coupons have less meaning for the farmer than for the city dweller, since the
farmer produces his own food.”19 As a result, “Workers no longer have faith in currency, and
whenever they can receive money in payment for their work, they prefer goods.”20
Consequently, the stock of currency held by commercial banks in the U.S. zone increased sixfold between December 1944 and March 1948; from less than 2 billion to more than 12 billion
(Figure 3). For Germany as a whole, the stock of currency was about 79 billion Reichsmarks.
14 000
millions of Reichsmarks
12 000
10 000
8 000
6 000
4 000
2 000
D
ec
Fe -44
b
Ap -45
rJ u 45
n
Au -4 5
gO 45
ct
D -45
ec
Fe -45
b
Ap -46
rJ u 46
n
Au -4 6
gO 46
ct
D -46
ec
Fe -46
b
Ap -47
rJ u 47
n
Au -4 7
gO 47
ct
D -47
ec
Fe -47
b48
0
Figure 3: Value of Reichsmarks notes held by banks in the U.S. zone, December 1944 through March 1948
(Source: Monthly Report of the Military Governor, various issues)
The return on using fiat money for payments was lower than the return on using goods,
even when there was high transaction costs associated with matching up traders. Because the
Reichsmark had lost its function as a medium of exchange, black-market prices were immune
from inflation, at least after the winter of 1946 (Menderhauser 1949, p. 654). Until 1947 this
fact intrigued the U.S. occupation authority, which repeatedly mentioned it in the monthly
reports. Figure 4 shows that, after an initial surge during the first months of the occupation,
there was no positive trend toward inflation on the black market in Frankfurt.
19
Report dated 04/23/1947. OMGUS, Econ division, Trade and commerce branch APO 742, file 1/ 194-1/6 shelf
mark BICO US Cust. Gp.
20
Periodic report for week ending on the 4th September of 1946 from Det. 245, Landshut and Det. 348,
Stadtsteinach. See National Archives of Bavaria, OMGBY file 10/85 3/1.
12
Most of the U.S. reports—as well as contemporary academic articles written by economists
who worked for the Allies in Germany—pointed out that people used “various forms of direct
barter” as substitutes “for normal methods of purchases and sales”.21 Most observers indicated
that each good could have been used as means of payment provided a seller accepted it in
payment of its holdings. Thurnwald (1948) noted that the unlawful nature of those trades made
credit difficult. Sauermann (1950) described economics of the time as a “primitive barter
economy [that] guaranteed nothing more than an animal life” (p. 178). Menderhauser (1949)
remarked on the “economic incongruity” of a “system of reciprocal exchanges of goods”.
Index base 100 in September 1947
160
140
120
100
80
60
40
Butter (kg)
Cigarettes (pack)
Coffee (kg)
Flour (kg)
Sugar (kg)
20
1945:08 1946:02 1946:08 1947:02 1947:08 1948:02
Figure 4: Evolution of black-market prices in Frankfurt am Main, August 1945 through June 1948
(Sources: Aug. 1945-Sept. 1946: OMGUS archives file 1/194-1/6 BICO US cust gp; Sept 1947-Jun 1948: Kropat
Wolf-Arno 1979; Hessen in der Stunde Null 1945/1947)
Some academics (e.g. Friedman 1951, p. 204) observed that cigarettes, chocolate, candy
bars and alcohol were regular means of payment. The exchange pattern between agents
involved many transaction costs, since trades were made in a disorganized fashion, with mostly
decentralized trades in pairs and some kind of random search.22 Indeed, the typical trading
pattern was as follows: City dwellers (buyers) travelled to the countryside—or to any place that
specialized in production of the desired good—with some valuables, cigarettes, or industrial
21
Monthly Report of the Military Governor, August–September 1947, p. 2. According to Lutz (1949), people
“acquired most of the commodities they wanted against commodities they had to offer” (p. 122).
22
See Bignon (2007) for a rationale.
13
items and searched for a farmer (seller) ready to make a deal.23 Often it took some time to find
a seller who would agree to trade food for the buyer’s “currency”. More often than not, the
buyer first encountered a seller unwilling to trade, but fortunate buyers might still glean some
advice on better candidates. Another strategy was to carry goods, such as commodity moneys,
that were more acceptable to farmers.
3.4. Historical evidence emphasizing the unique role of cigarettes in payments
The claim of most U.S. reports (and some other works) that Germans bartered a lot on the
black markets is at odds with the evidence given by newspapers and German post-war scholars
regarding the special character of cigarettes. To cite just a few references, Hess (1996) and Bub
(2004) are detailed studies of its use as money, and they indicate that cigarettes had a universal
acceptability: everybody accepted them in payment. Moreover, the acceptability of cigarettes
on illegal markets was far greater than that of the Reichsmark, as highlighted by the Herald
Tribune of 28 February 1947: “Anything money will buy, and a good many things it won’t, can
be had for cigarettes in Germany today.” And in December 1945, the military newspaper Stars
and Stripes noticed their “exorbitant prices” and explained them by the fact that “cigarettes buy
potatoes, meat and other necessities because in this fantastic city-market [Reichs]marks are
considered hardly worth more than the paper they’re printed on.”24
As a result of the medium of exchanges role of cigarettes, Botting (1985, p. 179) gauged
that a cigarette changed “hand[s] a hundred times before reaching the end of the line, the
smoker.” The Herald Tribune reported in May 1947 that “cigarettes today are too expensive for
the average German to smoke but they provide a portable, easily negotiable medium of
exchange” (21 May). Some evidences also points that cigarettes were also a unit of account
because people used it to post or negotiate the price of a deal. In an article published during
23
As mentioned in the Herald Tribune article of 28 February 1947 entitled “The Cigarette Standard”: “A lot of
them [cigarettes] undoubtedly end up in the hands of farmers because cigarettes are a way of getting food.”
24
“Big Business: New York Fifth Avenue is a Bargain Deal Compared to Berlin’s Lush Black Market,” by Jack
Caldwell (Stars and Stripes, 30 December 1945).
14
May 1947, the Herald Tribune wrote that “all prices in this shadowy world of commerce are
either based or quoted in terms of American cigarettes […] when shopping around for items, a
visitor to “Black market alley” [in Frankfurt] will find dealers constantly quoting prices in
Stangen (cartons). “Eine Stange” for this; “Zwei Stangen” for that” (17 May).
Two types of cigarettes were traded on the black market, German (made of dark tobacco) or
American (made with tobacco of the Virginia type). As a result, it was easy for traders to
recognize the quality of those cigarettes. Moreover, as reported by Botting (1985), on big
black-market places, intermediaries facilitated deals by assessing the quality and weight of the
cigarettes traded. On the black market, German cigarettes traded at a 20% discount to U.S.
cigarettes. The most generally used were however the American ones.
The enactment of a new regulation – promulgated on 26 May 1947 by the U.S. Army – that
prohibited the free import of cigarettes by U.S. soldiers is another evidence of their widespread
use as money.25 The ban applied to the “mailing or shipment of these commodities through
army post offices, international mail or commercial channels”, and the War Department
coupled its action with an appeal to the general public for cooperation “in a strong move to
wipe out black market trading in American cigarettes and tobacco”.26 However, the newspaper
Tagespiegel reported that the main consequence of this new regulation was not to inhibit the
use of cigarettes as money but rather to increase their prices (4 August 1947).27 During the
preparation of this regulation, the U.S. occupation authority mentioned for the first time (in the
Monthly Report of the Military Governor) the monetary character of cigarettes, stating that “to
an increasing degree the Reichsmark prices of good have become subsidiary or have even been
25
“Zigaretten, die Vermögen Liquiedierten,” (Die Wirtschaftsspiegel, 15 July 1947).
“German Food Crisis Is Laid to Black Market and Cigarette,” by John Elliott (Herald Tribune, 21 May 1947).
27
The original quote is: “Die «Zigarettenwährung» ist nicht abgeschafft, sie ist nur aufgewertet worden, und
weitern die gestiegenen Zigarettenpreise haben alle anderen Schwarzmarktpreise mitgezogen. Nicht allein der
Kaffee auch die Butter ist teurer geworden, der Zuckern die Feuersteine, viele Dinge, die ihrer Herkunft nach nicht
ausländischen Quellen entstammen, sich jedoch offensichtlich nach der Zigarettenlage orientieren.”
26
15
eliminated entirely from transactions in favour of a variety of direct barter ratios between
goods, especially between cigarettes and other goods.”28
Finally as further evidence of the value of cigarettes and of the huge disequilibrium
between supply and the demand (estimated at 80,000 tons by the newspaper Die Welt), a new
type of entrepreneur emerged. Some clever people established small factories to manufacture
cigarettes, buying their raw material from the Kippensamler (literally, collectors of butt-ends)
who collected cigarette butts from waiters and maids at places where they were thrown in
abundance: the cinema, the entrance to messes, and soldiers’ clubs (Botting 1985, p. 179).29
That cigarette was the good selected in equilibrium to replace the Reichsmark as a mean of
payment (and not the dollar or coffee) need some further qualification. Of course, cigarettes
had some relative advantage over the other goods either in terms of divisibility or durability.
The following quote from the Herald Tribune of 28 February 1947 sums them up:
In Germany, cigarettes lubricate the trade. For instance, if a Berliner has a large radio that he
has decided to sell, he cannot conveniently lug it out into the country in search of a farmer
willing to give him butter for it. Instead, he trades it to a black marketer for cigarettes and
takes the cigarettes to the farmer. This has had advantage that he can dispose of the cigarettes
bit by bit instead of having to accept a whole radio’s worth of butter at one time. Meanwhile,
the black marketeer takes the radio to an American officer who gives him more cigarettes.
With these he can get another radio or butter or whatever he needs to carry on his trade. Then
he can sit down and smoke his profit—or a part of it. (Herald Tribune, 28 February 1947)
But cigarettes had other advantages. Their demand for consumption purposes was quite
high as compared to fiat objects. Their consumption especially increased during and after the
war as they were considered both as an appetite suppressant and a substitute of medicine for
sore throat (Proctor, 1999). They had moreover some relative advantage over the U.S. dollars
or the M.P.C. in terms of perceived lawfulness. Although U.S. cigarettes were not sold in the
rations (and consequently this made illegal to buy some), they were distributed freely as part of
28
OMGUS #21, February and March 1947, section Price control, p. 19. The Herald Tribune copied this sentence
in its issue of 21 May 1947.
29
It was estimated that the waiters of the Café Wein in Berlin earned an average of 50 Reichsmarks (about $5) per
day for selling 75 to 100 butts to a Kippensamler. Botting (1985) indicated that it took 7 butts to manufacture a
new cigarette, so the 100 butts collected from the Café Wein waiters were enough to manufacture about 15 new
cigarettes.
16
the CARE packages to the numerous people that lost everything during the war (e.g. displaced
persons from some former part of Germany or refugees from other countries). This very fact
provides an excuse to their holders if the police asked them to justify their holdings, which
thereby reduce the risk of them being confiscated. This was not the case for dollars or MPC.
It was also easier for suppliers to smuggle them than with U.S. dollars for two reasons.
First, the opportunity cost of importing dollars for a U.S. soldier must have been greater than
the cost of importing cigarettes as there were much more opportunities to spend dollars in the
U.S. as compared to cigarettes. Second capital controls make dollars harder to obtain while it
was quite easy for U.S. soldiers – the most prominent suppliers of American cigarettes – to
acquire them either at the military PXs or by order directed to U.S.-based firms (eventually
using family and friends as middlemen). The U.S. military government for Germany30
estimated these imports during spring of 1947 by checking the contents of some of the 200,000
packages sent to soldiers between March and June; they found cigarettes in 95% of the checked
parcels.31 Hess (1996) gave a lower estimate: “half of the 3 million packages which arrived
every month by military mail from the US were cigarette shipments.”
Capital controls make also easier the supply of cigarettes by civilian traffickers either from
neighboring countries or the diversion of U.S. Army cigarettes stocks. Smuggling by traffickers
came from Czechoslovakia, Hungary, Belgium, and the Netherlands or through the Hamburg
haven (Hess 1996). They also diverted some of the imports made by the U.S. Army. As an
example, newspapers reported that by April 1948 10% of the 210 million American cigarettes
stored in a Frankfurt warehouse in October 1947 had disappeared.32
Building on those evidences of the use of cigarettes as money, the next section uses a
sample of black market prices to test whether their liquidity was significantly different from
that of the other goods of the sample.
30
New York Herald Tribune, 17 May 1947.
Rhein Neckar Zeitung, 29 July 1947.
32
Frankfurter Rundschau, 21 April l948.
31
17
4. Data and measures of the liquidity of sample goods The first section presents the data (section 3.1). The second subsection studies of the
relative liquidity of a sample of black-market goods in Bavaria (Section 3.2).
4.1. The sample of black-market prices
This paper uses black-market prices in 120 out of 143 Bavarian districts. These prices are
available at the county (Landkreis) level for two dates, 1 July 1947 and 30 June 1948. All
prices are denominated in Reichsmarks in 1947 and in Deutsche Marks in 1948. The greatest
distance between any two districts is 488 km and the least distance is 8.5 km. Prices were
collected in September 1948 by the economics division of the U.S. occupation army of Bavaria
(OMGB), which asked their correspondent to give representative black-market prices for the
beginning and end of the period covered by the survey (available at the Bavarian archives). The
questionnaire and the answers are in English. The sources used by each local office could have
been the soldiers (who were active at these offices) or the German local price police (who
monitored those markets). Both sources knew the phenomena well.
18
Map 1: Post-WWII Germany, the Länder, and the occupied zones
The survey asked for the prices of eight goods: butter, sugar, coffee, cigarettes, meat, flour,
U.S. dollar, and military payment coupon (MPC). The MPCs were actually denominated in
dollars and were originally intended for use by soldiers to pay in military shops. All prices are
for 1 kilogram except for cigarettes (one pack) and for the dollar and MPC.
Of these eight goods, three were produced locally: butter, flour, and meat. A survey
conducted by the Bavarian statistical office gives us a good idea of the degree of agricultural
19
specialization in Bavaria.33 The main crop was wheat, which accounted for 52% of the
cultivated surface in the sample of counties studied. The area devoted to sugar (13,705 acres)
represented only 0.78% of the agricultural surface; at that time, sugar was mainly imported
from outside Bavaria. This was also obviously the case for MPCs, the U.S. dollar and coffee.
Two types of cigarettes were traded on the black market, those made with dark tobacco and
those of the Virginia type. The German production consisted of dark tobacco, which was
mostly grown in two regions close to Bavaria (Swabia and Palatinate; cf. Ruland 1968, p. 59).
This production amounted to 30,000 tons per year and was used to distribute cigarettes to
consumers as part of their ration.34
4.2. Comparing the liquidity of cigarettes to that of other goods
Two criteria are used to measure liquidity: (1) a coverage ratio measuring the probability
that a given good is sold on those black markets;35 and (2) the coefficient of variation of prices
across locations. The first criteria indicates that cigarettes benefited (as did butter and coffee)
from wide coverage whereas U.S. dollars and MPCs (the two fiat objects) were quoted in only
half of the counties. The coefficient of variation of relative prices measures the risk of using a
particular good as payment for trade. This measure is especially accurate when agents
arbitraged between districts with goods rather than with Reichsmarks.
It is notable that the coefficient of variation is always lower when the relative prices are
computed in cigarettes, even when compared with fiat objects (currency) of perfectly
homogenous quality. Therefore, using cigarettes as money enabled traders to benefit from a
lower spatial variation of prices than when using some other good. This suggests that the best
strategy when using indirect barter was to pay with cigarettes.
Table 1: Descriptive statistics of the sample of Bavarian black-market prices
33
The book is part of the “Beitraege zur Statistik Bayern” series and is entitled Bodennutzung und Ernteergebnisse
1946 und 1947 (number 143).
34
The ration of cigarettes was forty a month, which was valued at about two thirds of the monthly wage of a
secretary in Berlin (Herald Tribune, 21 May 1947).
35
This probability is equal to the number of observations divided by 120 (the number of sample districts).
20
Butter
Coffee
Sugar
Cigarettes
Meat
Flour
M.P.C.
U.S.D.
Max
July 1, 1947
500
1300
400
260
360
120
400
600
Min
70
80
22,5
60
12
5
45
40
Range (= max/min)
7,14
16,25
17,77
4,33
30
24
8,89
15
# of observations
120
119
119
120
116
110
71
66
100 %
99,17 %
99,17 %
100 %
96,67 %
91,67 %
59,17 %
55 %
U.S.D.
Cover ratio
Butter
Coffee
Sugar
Cigarettes
Meat
Flour
M.P.C.
Max
June 30, 1948
32
50
10
7
12
6
15,5
30
Min
2,5
5
1
2,5
1,75
0,3
4,5
7,5
Range (= max/min)
12,8
10
10
2,8
6,85
20
3,44
4
# of observations
116
115
76
117
103
96
78
72
96,67 %
95,83 %
63,33 %
97,50 %
85,83 %
80 %
65 %
60 %
Cover ratio
Table 1 gives the descriptive statistics for each good. The cover ratio measures the number
of counties in which there were a price quote for a good. It is a proxy that measures liquidity to
the extent that agents arbitraged among counties and mainly used barter (rather than fiat
money). Thus agents had to take into account the ease of selling their holding in other districts.
In 1947, butter, cigarettes, coffee, and sugar were sold in nearly all districts but the dollar and
MPC were sold in only 55% and 59% of the districts, respectively. In 1948, the cover ratio of
sugar, meat, and flour dropped by 36%, 11% and 11.5% (respectively) while the one of other
goods remained stable.
Figure 5: Coefficient of variation for black-market prices in 1947 and 1948
Price dispersion was high for all goods, whether produced locally or “imported”. This is to
be expected, given the trade pattern within illegal markets (described in Section 1). Although
various alternative measures of price dispersion have been applied in the literature, two are
especially accurate: the coefficient of variation and the range. The range is computed as the
21
ratio of the highest to the lowest price; hence it is independent of scale but is not independent
of the size of the sample. In contrast, the coefficient of variation—defined as the standard
deviation divided by the mean—has the advantage of being both scale-independent and
unaffected by changes in the number of observations.
The range of prices is large for all goods, with a minimum ratio of 4 (for cigarettes in
1947 and for MPCs and dollars in 1948; see Table 1). But the highest price of butter in 1947
was 7 times greater than the lowest price, and for sugar and coffee it was more than 16 times
greater. This would seem to indicate a huge spatial difference. However, since those figures
could have been driven by some outliers, it is worth analyzing the data in terms of the
coefficient of variation (CV), which is quite high both in 1947 and in 1948 (Figure 5). A pack
of cigarettes had the lowest CV both in 1947 and in 1948, with respective values of 21.5% and
14%. Excluding cigarettes, in 1947 the coefficient of variation ranged from about 30% for
butter, coffee, MPCs, and U.S. dollars to 46% for sugar and to more than 77% for meat. These
figures change sharply in 1948: meat is characterized by a big drop in the CV to 39%, and flour
increases from 52% in 1947 to about 70% in 1948.
Table 2: Coefficient of variation for relative prices
Coefficient of variation
of relative prices (RP)
in 1947
CV of cigarettes
CV of butter
CV of coffee
CV of sugar
CV of meat
CV of flour
CV of M.P.C.
CV of dollar
RP in
cigarettes
RP in
butter
RP in
coffee
RP in
sugar
RP in
meat
RP in
flour
RP in
M.P.C.
RP in
dollar
0
0.37
0.35
0.48
0.7
0.53
0.35
0.37
0.56
0
0.46
0.55
0.8
0.64
0.51
0.56
0.59
0.53
0
0.74
0.68
0.72
0.54
0.51
0.67
0.65
0.58
0
1.56
0.67
0.95
0.83
0.78
0.63
0.78
0.84
0
0.70
0.71
0.62
0.68
0.70
0.63
0.61
0.74
0
0.68
0.65
0.40
0.54
0.45
0.58
0.82
0.77
0
0.29
0.85
1.04
0.52
0.85
0.92
0.76
0.25
0
As mentioned in Section 3, most trades were done either through barter or cigarette money.
Computing the coefficient of variation of relative prices is then a more accurate measure of
price dispersion; see Table 2. The CVs of these barter ratios were computed by dividing the
22
price of each item in one district by the price of the benchmark good in the same district. Two
remarks are in order.
On one hand, the coefficient of variation for relative price gives a measure of the risk
associated with the use of barter or cigarette money. This measure is especially accurate if
agents arbitraged between districts with goods other than Reichsmarks. The notable feature of
Table 2 is that the coefficients of variation are always lower when the relative prices are
computed in cigarettes. For example, in row 3 the CV for butter shows that its prices are less
dispersed when the benchmark good is cigarettes (0.37, or 37%) but are more dispersed when
the benchmark good is coffee (0.53) or dollars (1.04). Likewise, the other rows in the table
show that the coefficients of variation are the lowest for cigarettes (column 1). This means that
an agent who took cigarettes to trade in another county faced the lowest possible spatial
variation of prices (provided the district was randomly chosen). We again conclude that the
best strategy for indirect barter was to use cigarettes for payment.
CV of the relative prices in cigarettes
CV of the nominal prices
77,8%
70%
90%
80%
70%
60%
50%
40%
53 %
51,7%
48 %
37%
33%
35 %
32,5%
46 %
37 %
35 %
37,05%
31,7 %
30%
20%
10%
0%
Butter
Coffee
Sugar
Meat
Flour
MPC
USD
Figure 6: Comparison of the CV of relative prices in cigarettes and the CV of nominal prices
On the other hand, the coefficient of variation for the relative prices in cigarettes did not
differ sharply from those computed for nominal prices (Figure 6). This indicates that cigarettes
did not lower price dispersion as compared to Reichsmarks. One possible explanation for this
low variation is as follows. Agents usually paid for their purchases with cigarettes, so it could
be that most of the quotes observed by local price offices were denominated in cigarettes. But
23
since the survey asked for price quotes in terms of fiat money, it’s likely that some price offices
converted the price in cigarettes to money by using the cigarette/Reichsmark exchange rate.
5.
Ruling out the case of segmented black markets
The observed level of price dispersion indicates that the “law of one price” (LOP) hardly
held in those markets. Besides consumer search incentives, two other explanations could
account for the high coefficient of variation. First, it could be that the LOP held in each
(separate) region but that interregional differences account for the level of price dispersion; we
test for this possibility in Section 4.1. The second alternative explanation relies on a
segmentation of local markets, with prices reflecting the local supply and demand. In Section
4.2 we rule out this possibility by correlating the prices with the characteristics of each district.
5.1. Regional market segmentation?
Here we examine the sample’s spatial dimension to explore whether the level of price
dispersion computed via the coefficient of variation can be explained by some regional
clustering of prices. It is certainly possible that some goods (say, butter) were more expensive
in some regions than in others. Provided that the price differences among Bavarian regions
were large enough, this could explain the level of price dispersion.
As an introduction to the spatial distribution pattern of our sample prices, Map 2 shows the
spatial distribution of the price of butter (other maps are in Appendix 3 and 4). Inspection of
these maps suggests little spatial clustering of prices, with many adjacent districts characterized
by huge discrepancies in prices.
To systematize the exploration of such spatial clustering, we shall compute a measure of
spatial autocorrelation. Measures of spatial autocorrelation are especially useful for
investigating whether prices in one district are similar to those in neighboring districts. Among
the most widely used measures are Moran’s (1950) I statistic, which uses the whole sample to
24
derive local measures of spatial autocorrelation that give a spatial association coefficient for a
particular locality i (see also Anselin 1995).36
Map 2: Price distribution of butter in Bavaria on 1 July 1947
Moran’s I statistic measures the covariance of prices in connecting districts relative to the
variance of the price across districts. Moran’s I will differ from zero to the extent that prices in
connected districts are relatively similar. More precisely: There will be positive spatial
correlation (with a value of I close to unity) if prices are similar among neighboring districts; a
negative spatial correlation implies dissimilar prices. Table 3 gives the value of the Moran I
36
See e.g. Shiue and Keller (2004) and Keller and Shiue (2007).
25
statistic for each good, both in nominal prices and in relative prices in cigarettes. It shows that
no good was spatially correlated, since the values ranged from −0.1 to 0.1.
Table 3: Moran's I statistic for nominal and relative prices in 1947 and 1948
Moran’s I
Butter
statistic
For nominal
-0.0926
prices in 1947
For RP in
0.0184
cigarettes in 1947
For nominal
- 0.0305
prices in 1948
For RP in
0.1028
cigarettes in 1948
Coffee
Sugar
Cigarette
Meat
Flour
M.P.C.
U.S.D.
-0.0227
0.1156
0.0311
0.0255
0.0615
0.0635
0.0036
0.0906
0.1284
-
0.0166
0.0855
/
/
0.0437
-0.0952
0.0871
0.0794
0.1009
0.1184
0.1557
0.0907
-0.0582
/
0.206
0.0764
0.1591
0.1723
The absence of spatial correlation is striking and requires some qualification. A statistical
test was run to decide whether these spatial patterns deviated significantly from a random
pattern. Table 8 (in Appendix 2) indicates that most of the sample goods exhibit no spatial
autocorrelation at the 5% significance level. However, in 1948 for meat, dollars, and MPC
there is a positive association—that is, expensive counties tend to be located near other
expensive counties.
Figure 9 gives a graphical representation of Moran’s I for two examples, the density of
dairy cows and the price of butter in 1947. The former is spatially correlated but the latter is
not. The I statistic (indicated at the top of each graph) is equal to the slope of the regression line
of the scatter plot. The horizontal axis represents the actual value of the variable, and the
vertical axis plots this value weighted by the value of its neighbors (see Appendix 2) for the
weighting method). As a result, the four quadrants in the graph provide a classification of four
types of spatial autocorrelation: high-high (upper right) and low-low (lower left) for positive
spatial autocorrelation; high-low (lower right) and low-high (upper left) for negative spatial
autocorrelation. The variables are standardized so that the units correspond to standard
deviations. Inspecting the graph reveals that districts with a small density of dairy cows tend to
26
have neighboring districts with a small density of dairy cows (and vice versa), whereas the
graph for butter prices shows that no such relation exists.
Figure 9: Moran's scatter plots of the density of dairy cows and the 1947 prices of butter
Our investigation of the pattern of price dispersion on the Bavarian black market finds no
spatial autocorrelation,37 which means that there were no regions in which the prices of a good
tended to be higher than in another region. Hence the price heterogeneity was not ordered
according to a spatial pattern.
The remaining alternative is that markets were segmented, so prices differed only according
to local supply and demand. We test for this possibility in the next section.
5.2. Perfect segmentation of markets?
We have seen that the dispersion of prices, as measured by the coefficient of variation, is
large for each good of the sample. Moreover, Moran’s I statistics and the maps in Appendix 3
indicate that prices can have varied quite largely even between contiguous markets. One
possible explanation of such a pattern is that markets were segmented.
Many black trades implied city-dwellers that forage for foods in farmers’ houses in the
countryside. There is ample anecdotal evidence of this. Often people took the train and carryon some valuables or other commodity to trade for foodstuffs. Those travels were often
completed on a local basis but it was not uncommon for a German living in the British zone of
37
The maps in Appendix 3 and 4 show the geography of the prices of butter, meat, dollar, and sugar. Inspecting
these maps reveals that prices seem not to be spatially positively correlated, since two contiguous districts are
often characterized by a large difference in the price of goods.
27
occupation to travel hundreds of kilometres by trains to find fruits or shoes in Bavaria (see
archives). Given the shortages of gasoline, many travels occurred using the train transportation
system38. As such this is an evidence of arbitrage between counties, regions. Figure 1 plotted
for each month a measure of the use of trains (the number of passengers in a train) with a proxy
for the incentive to black trade (the caloric value of the official ration). This indicates that
agents did adjusted to changes in the incentive to trade illegally because reducing the official
ration increase the incentives to travel to farmers’ house.
The caloric value distributed through the rationing
system and the use of passengers trains.
Daily caloric value of food ration
scales (each month, the 10th)
2200
Feb. 47 to June 48
July 48 to Dec. 48
2000
Linear (Feb. 47 to June 48)
Linear (July 48 to Dec. 48)
1800
2
R = 0,4446
1600
1400
2
R = 0,442
1200
1000
250
350
450
550
650
# of passengers in a passenger train (monthly
mean)
Figure 5: the incentives to trade illegally and the use of trains39.
A formal test of market segmentation consist in trying to explain the distribution of blackmarket prices in terms of the districts’ characteristics in order to check whether each district’s
price was correlated with our proxies of the local demand and supply for that good. The
38
In fact, travelling by train was very cheap before the monetary reform as train tickets were paid in Reichsmarks
(which according to sources had lost most of its purchasing power except on the rationed market).
39
All figures are index, basis 100 in January 1948. Sources: The caloric value come from various statistical annex
of the OMGUS monthly reports (IfZ). The number of passengers per passenger train were calculated using the
figures on transportation also publish in the monthly reports. The caloric value is taken at the 10th of each month
and is the true caloric value that was distributed through the rationing system and not the one announced. Such
discrepancies come from the fact that there were food shortages that prevents to distributed all of the rations
announced. The index on passenger train use is computed using the numbers of passengers and the number of
trains that have circulated in the US and UK zone during the month considered.
28
following example explains the intuition. If markets were integrated, then the demand on a
local black market might have consisted of two elements: the demand expressed by local agents
and the demand from agents that traveled to the district. Thus,
Ddistrict = Dlocal + Dtravelers
If, however, agents chose not to arbitrage, then Ddistrict must be equal to Dlocal.
From the supply perspective, the supply in a given district could come either from local or
traveling producers. If producers do not choose to arbitrage, then (as with demand) the supply
on a local black market is equal to the supply of local producers. In the case of segmented
markets we have pij(Dj) = pij(Sj), where i is an indicator of the good traded and j indicates the
district. Thus prices must be positively correlated with the local demand and negatively
correlated with the local supply.
Since neither the quantities sold on each local black market nor the production of each
district is known, these variables are proxied using indicators of production capacity in the area
considered. If markets were segmented, then the local supply should be negatively correlated
with the price of butter. The demand side is approximated by two sets of indicators: population
density (an indicator of urbanization) and mean industrial wages (an indicator of potential
black-market demand). The intuition for this is as follows. People working in industry had to
exchange on the black market in order to obtain the goods they needed. Hence, if black markets
were segmented, then the more urbanized a district was the higher its black-market prices must
have been. We also expect that market segmentation would have resulted in a positive
correlation between prices and the mean industrial wage of the district.
The data needed to construct these variables were collected in the 1947 statistical book
edited by the Bavarian Statistical Office.40 This document gives the agricultural and total area
of each district, its working and total population, its mean industrial wage, and the type and
40
In Bavaria there are two types of administrative districts, Landkreise and Stadtkreise (also called “free cities”).
When a Stadtkreise was part of a Landkreis, I merged the two numbers.
29
number of livestock it supported. We use the following type of ordinary least squares (OLS)
regression test for the market integration:
Pij(1947) = α + β * supply +χ * Demand + ui
This equation explains 1947 butter prices by using a proxy for the local production capacity
of butter (the density of dairy cows in 1947) together with a proxy for the intensity of the
demand for this good (the district’s population density and mean industrial wage). For sugar
and flour, the proxy for the supply-side variable is the percentage of the cultivated area devoted
to sugar and wheat. The local production of meat is approximated by the density of pigs in the
districts. If markets were segmented, then the sign of supply-side variables should be negative
and that of demand-side variables positive. The results are reported in Tables 4 and 5. They
indicate few relationships between prices and local variables; and even when a relation is
significant, the adjusted R2 is close to zero.
Table 4: Regression results for nominal prices in 1947
Butter
-0,072
0,418
0,295
224,1
no
0,00%
116
Nominal prices, July 1947
Population density
Mean wage
Supply side variable
Constant
Jarque Berra test
Adjusted R²
# of observations
Coffee
0,136
-1,105
719,04
yes
0,00%
116
Sugar
0,0294
0,093
-0,012
124,64
yes
0,00%
116
Cigarettes Meat
Flour
-0,013
-0,0011 0,0145
0,189**
-0,106 -0,0187
0,431 -69,82*
79,46
76,32
39,71
yes
yes
yes
1,90%
0,00% 7,16%
116
116
110
MPC
-0,026
0,366
Dollar
-0,114
1,416
50,91
yes
1,60%
71
-47,69
yes
9,15%
66
* significant at 1%; ** significant at 5%; *** significant at 10%
Table 5: Regression results for nominal prices in 1948
Nominal prices, June 1948
Population density
Mean wage
Supply side variable
Butter
Coffee
Sugar
0,0004
0,0331
0,064
0,005
-0,041
-0,0017
0,022**
0,0001
Cigarettes
Meat
Flour
Constant
0,768
29,15
-0,046
3,53
2,38
1,05
Jarque Berra test
Adjusted R²
# of observations
yes
0%
116
yes
0%
115
yes
0,47%
76
yes
1,20%
116
no
0,90%
103
yes
0%
96
-0,00016 0,0012 0,0002
0,007*** 0,042*** -0,00046
0,007
0,722
* significant at 1%; ** significant at 5%; *** significant at 10%
Tables 6 and 7 (in Appendix 1) show that using relative prices in cigarettes as the left-handside variable did not change the results. Thus it seems that there was no relation between the
30
local demand or supply and the level of prices. One way to explain this is by supposing that
prices were not spatially autocorrelated but that the proxies for demand and supply were (see
Table 9).
5. Conclusion
The German economy experienced a huge change in means of payments at the end of
World War II. The Reichsmark was replaced as the money of the black markets by a
combination of barter and commodity money such as cigarettes. This paper uses a new data set
of black-market prices to provide evidence that the liquidity of cigarettes was far higher than
the liquidity of any other good in the sample. To my knowledge this is the first empirical proof
of the old intuition that money of whatever commodity has the greatest liquidity. The result
hinges on the computation of various measures of liquidity, such as the probability of selling a
good in a given district and the dispersion of prices across districts. In checking whether price
dispersion is indeed a viable measure of liquidity, I demonstrate that local black markets were
not segmented; in other words, despite great spatial variation in prices, they must be considered
as a network of interconnected markets. Because all districts quoted cigarettes and because the
CV of black-market prices was lower for cigarettes than for any other commodity, traders were
strictly better-off paying for their purchases with cigarettes. In doing so they limited the risk of
losses due to price dispersion.
References
Anselin L., 1995, Local Indicators of Spatial Association, Geographical Analysis 27, 93-115.
Baye M., Morgan J. and P. Scholten, 2006, Information, Search and Price Dispersion,
Handbook on Economics and Information, Elsevier, T. Henderschott ed.
Bayerisches Statistisches Landesamt, 1947, Bayerisches Statistisches Jahrbuch für den
Freistaat Bayern, Munich.
Bayerisches Statistisches Landesamt, 1947, Bodennutzung und Ernteergebnisse 1946 und
1947, Beiträge zur Statistik Bayern heft 143.
Berentsen A. and G. Rocheteau (2002), On the Efficiency of Monetary Exchange: How
Divisibility of Money Matters, Journal of Monetary Economics 49 1621-49.
31
Bignon V., 2004, Une théorie de l’élection de la cigarette comme monnaie, Revue Economique
55, 383-93.
Bignon V., 2007, Black and Grey Markets for Illegal Exchanges in post WW II Germany,
manuscript.
Botting D., 1985, In the Ruins of the Reich, George Allen and Unwin: London
Bub K. , 2004, AEG, Siedlerstolz und Schwarzmarkt – Von der Zigarette in der
Nachkriegszeit, in Der Schuß aus dem Bild. ed. by Topitsch K. and Brekerbohn A., 28-36
Burdett K, Trejos A. and R. Wright, 2001, Cigarette Money, Journal of Economic Theory 99,
117-42
Cuadras Morato X., 1997, Can Ice Cream be Money? Journal of Economics 66, 103-25.
Cuadras-Morato X. and R. Wright, 1997, On Money as a Medium of Exchange When Goods
Vary by Supply and Demand, CARESS Working Papers 97-1, Univ. of Pennsylvania.
Diamond P. A., 1971. A model of price adjustment, Journal of Economic Theory 3. 156-168
Diskant J. A., 1989, Scarcity, Survival and Local Activism: Miners and Steelworkers,
Dortmund 1945-8, Journal of Contemporary History 24(4), 547-573.
Eckard, F. W., 2004, The ‘law of one price’ in 1901, Economic Inquiry 42 101-110
Engineer M. and Shi S., 2001, Bargains, Barter and Money, Review of Economic Dynamics 4,
188-209
Friedman M., 1951, Commodity Reserve Currency, Journal of Political Economy 59 203-32
Hansmeyer, K. H. and R. Caesar, 1976, Kriegswirtschaft und Inflation (1936-1948), Währung
und Wirtschaft in Deutschland 1876-1975. ed. Deutche Bundesbank, Fritz Knapp Gmbh,
Frankfurt am Main, 367-429.
Hess H., 1996, The other Prohibition: The Cigarette Crisis in Post-war Germany, Crime, Law
and Social Change 25, 43-61
Keller W. and C. Shiue, 2007, The Origins of Spatial Interaction: Evidence from Chinese Rice
Markets 1742-1795, Journal of Econometrics 140 304-32.
Kindleberger C., 1984, A Financial History of Western Europe, Routledge: London & N.Y.
Kiyotaki N. and J. Moore, 2004. Evil is the Root of all Money, Clarendon Lectures 1.
Edinburgh School of Economics Discussion Papers 110, University of Edinburgh.
Kiyotaki N. and R. Wright, (1989), On money as a medium of exchanges, Journal of Political
Economy 97 927-54
Kiyotaki N. and R. Wright, (1993), A Search-Theoretic Approach to Monetary Economics, The
American Economic Review 83, 63-77
Klein, B., 1976, Competing Monies: Comment, Journal of Money, Credit & Banking 8, 513-9
Li, Y. and R. D. Wright, 1998, Government Transaction Policy, Media of Exchange, and
Prices, Journal of Economic Theory 81(2), pp. 290-313
Lutz F., 1949, The German Currency Reform and the Revival of the German Economy,
Economica 16, 122-142
Mankiw Gregory N., 2008, Macroeconomics, Worth Publishers
Mishkin Frederic, 2006, The Economics of Money, Banking and Financial Markets, Pearson
Addison Wesley Publishing Company; 8th edition.
Menderhauser H., 1949, Money, Prices and the Distribution of Goods in post-war Germany,
The American Economic Review 39, p. 646-72
Menger K, 1892, On the Origin of Money, The Economic Journal 2, 239-55
32
Merritt A. J. and R. L. Merritt, 1970, Public Opinion in Occupied Germany: The OMGUS
Surveys, University of Illinois Press: Urbana, Chicago, London.
Moran, P.A.P., 1950, Notes on Continuous Stochastic Processes, Biometrika 37: 17-23.
Pratt, J. W., Wise D. A & Zeckhauser R., 1979. Price Differences in Almost Competitive
Markets, The Quarterly Journal of Economics 93, 189-211
Radford R.A., 1945, The Economic Organization of a P.O.W. Camp, Economica, November
Renero J. M., 1999, Does and Should a Commodity Medium of Exchange have Relatively Low
Storage Costs, International Economic Review 40, 251-64.
Rosen M., 1947, The Demand for Cigarettes in Austria, Southern Economic Journal 14, 186-91.
Ruland B., 1968, Geld wie Heu und Nichts zu Fressen: Karussel einer verrückter Zeit, Hestia
Verlag: Bayreuth.
Samuelson A., 1972, Le Mark : histoire de la monnaie allemande, Didier: Paris.
Senn P., 1951, Cigarettes as currency, The Journal of Finance 6, 329-32
Schmölders G., 1973, Die Zigaretten-Währung, Festschrift für Schmölders. Brinkmann: Berlin.
Scholten, P.S. and S.A. Smith, (2002) Price Dispersion Then and Now: Evidence from Retail
and E-Tail Markets. Advances in Microeconomics: Economics of the Internet, 11 63-88.
Shi S., 2006, Viewpoint: A Microfoundation of Monetary Economics, Canadian Journal of
Economics 39, 643-688.
Shiue C. H. and W. Keller, 2004, Markets in China and Europe on the Eve of the Industrial
Revolution, NBER Working Papers 10778, National Bureau of Economic Research.
Steiger O., 1992, Cigarette currencies, New Palgrave Dictionary of Money, 354-5
Stigler G., 1961, The economics of information, Journal of Political Economy 69 213-25.
Thurnwald H., 1948, Gegenwarts- Probleme Berliner Familien, Eine soziologishe
Untersuchung an 498 Familien, Weidmannsche Verlagsbuchhandlung, Berlin
Williamson S. and R. Wright, 1994, Barter and Monetary Exchange under Private Information,
American Economic Review 84, 104-123
Appendix 1: Relative prices in 1947
Table 6 gives the relative price of each sample good in terms of the other goods. It was computed by dividing the
average of the ratio of the market price for good j in market i by the market price of benchmark good in market i.
We interpret the result as follows. The figures in each row of Table 6 are the expected quantity of a good that a
consumer had to carry as barter for 1 kg of good i. Such an interpretation assumes implicitly that this type of
action did not change the observation.
This exercise shows that with 1 kg of butter a buyer could have purchased about 0.5 kg of coffee, 2.5 kg of
sugar, or 12 kg of flour. Conversely, with 1 kg of flour an agent could have bought only 120 grams of butter, 60
grams of coffee, or 30% of a pack of cigarettes. Clearly the prices of flour at that date were so low that people
could not have used it for barter without carrying a huge quantity. In contrast, using cigarettes (or butter or coffee)
to pay for purchases entailed carrying lesser amounts of goods and so made it easier to arbitrage between the
various markets.
Table 6: Matrix of the (average) quantity of goods that could be bought on black-markets in 1947
Relative prices (RP)
RP of butter (1 kg)
RP of coffee (1 kg)
RP of sugar (1 kg)
RP of cigarettes (1
pack)
RP of meat (1 kg)
RP of flour (1 kg)
in
butter
in coffee
in sugar
1,00
2,27
0,55
0,53
1,00
0,27
2,41
4,84
1,00
in
cigarettes’
packs
2,79
5,78
1,40
0,43
0,21
0,93
0,27
0,12
0,13
0,06
0,63
0,24
in meat
in flour
in M.P.C.
in U.S.D.
5,67
12,05
2,84
11,80
23,75
5,55
1,90
3,85
0,96
1,45
2,60
0,67
1,00
2,27
4,45
0,69
0,52
0,69
0,30
1,00
0,58
2,64
1,00
0,49
0,22
0,32
0,15
33
RP of M.P.C.
RP of dollar
0,66
0,98
0,32
0,47
1,50
2,19
1,65
2,36
3,31
5,03
6,65
9,52
1,00
1,50
0,71
1,00
In the matrix, the differences between the computed relative prices for the same couple of goods hinges on the fact
that the relative prices cannot be computed for some pairs because of because no price was quote in some district
for the reference good.
Appendix 2: Computing Moran’s I statistics
Computing the Moran statistic requires that we define a matrix of the interrelations between districts. This is a
weight matrix {wij} in which wij = 1 if the ith and the jth district are spatially connected and wij = 0 otherwise.
Spatial connectedness is here defined at the first order of contiguity relationships, which implies that wij is set to 1
when two districts are neighbors and to 0 when they are not.41 For a given year and good, Moran’s I statistic (or
the spatial autocorrelation coefficient) is given by
is
where i and j are two distinct districts, n is the number of districts, xi is the price of the good in district i, and
the mean of prices. Under the null hypothesis that the xi are identically and independently distributed normal
variables, the expected value of the Moran’s statistic is E[I] = (N − 1)−1. Table 3 reports the results of Moran’s I.
To test whether the spatial distribution of prices was random or ordered, I ran a statisticial test whose logic
runs as follows. If the number of regions is large, the sampling distribution of I under the null hypothesis of no
spatial pattern approaches the normal distribution; then the mean and the variance of I can be used to create the
statistic Z = (I − E[I])/
, whose value can be compared with the critical value of the normal table. At a
significance level of 5%, a value of Z greater than 1.96 or less than −1.96 indicates that the pattern is characterized
by spatial autocorrelation.
Table 7: Z statistic of spatial autocorrelation
Z(I)
For nominal prices in 1947
For RP in cigarettes in 1947
For nominal prices in 1948
For RP in cigarettes in 1948
Butter
-1.46
0.14
0.31
1.36
Coffee
-0.46
1.24
0.50
1.21
Sugar
1.55
1.63
-1.56
-1.04
Cigarettes
0.34
/
1.14
/
Meat
0.25
0.10
1.01
3.03
Flour
0.81
1.16
1.33
0.98
M.P.C.
0.70
U.S.D.
-0.17
1.53
2.08
2.00
2.33
Table 9: Moran’s I statistics for right-hand-side variables
density of dairy
cow
Moran’s I statistic
0,4296
E(I) = 1/(n-1)
0,0087
Standard deviation
0,0718
Z(I)
5,862
Density of pigs
0,5515
0,0087
0,0671
8,089
Density of
population
0,1267
0,0087
0,0584
2,021
Mean industrial
Share of wheat
wage
0,1927
0,5259
0,0087
0,0087
0,065
0,0698
2,831
7,410
Share of sugar
0,4099
0,0087
0,0618
6,492
As described in Section 4.1, the four quadrants in each of the following graphs provide a
classification of four types of spatial autocorrelation.
Moran’s scatterplot for the relative price of meat in cigarette packs (left: 1947; right: 1948)
41
All spatial statistics have been computed using Geoda (https://www.geoda.uiuc.edu/).
34
Moran’s scatterplot for the relative price of coffee in cigarette packs (left: 1947; right: 1948)
Appendix 3: Results of OLS regression for relative prices in cigarettes
Table 8: Results of OLS regression with the relative prices in cigarettes as the dependant variable
Relative prices, July 1947
Population density
Mean wage
Supply side variable
Constant
Jarque Berra test
Adjusted R²
# of observations
Butter
Coffee
Sugar
Meat
Flour
-0,0007
0,0012
0,0103
0,0016
-0,023*
0,0005
-0,0018
-0,00023
1,1 E-5
-0,0021
0,0032
0,0002
-0,0008
-0,754*
1,602
0,902
0,485
yes
0%
116
yes
0%
116
yes
9,62%
96
2,481
no
0%
116
yes
4,30%
116
Meat
Flour
Relative prices, July 1948 Butter Coffee Sugar
Population density
0,0003 0,0009 -0,0042
0,002
3,32 E-5
Mean wage
0,0094*** -0,011 4,79 E-5 0,0017 -9.147 E-5
Supply side variable
0,0023
0,0049 0,0102***
0,136
Constant
Jarque Berra test
Adjusted R²
# of observations
1,01
6,57
-0,042
0,484
0,232
yes
1,46 %
116
yes
0%
115
yes
0,89 %
76
No
1,46 %
103
yes
0%
96
35
Appendix 4: Spatial distribution of prices in 1947
Price of butter in cigarette packs
Price of coffee in cigarette packs
Price of flour in cigarette packs
Price of meat in cigarette packs
Price of sugar in cigarette packs
36
Appendix 5: Spatial distribution of prices in 1948
Price of butter in cigarette packs
Price of coffee in cigarette packs
Price of flour in cigarette packs
Price of meat in cigarette packs
Price of sugar in cigarette packs
Price of dollar in cigarette packs
37