Trading Costs in Early Securities Markets: The Case of the Berlin Stock Exchange 1880–1910
ABSTRACT Based on daily prices (amtliche Kurse) we estimate effective spreads of securities traded at the Berlin Stock Exchange in 1880, 1890, 1900 and 1910. Several extensions of the Roll measure are applied. We find surprisingly tight effective spreads for the historical data, comparable with similar measures of the MDAX and DAX at the end of the 20th century. Copyright Oxford University Press Science+Business Media, LLC 2006
DISCUSSION PAPER SERIES
Available online at: www.cepr.org/pubs/dps/DP5827.asp and www.ssrn.com/abstract=944041
TRADING COSTS IN EARLY
SECURITIES MARKETS: THE CASE
OF THE BERLIN STOCK EXCHANGE,
Caroline Fohlin and Thomas Gehrig
TRADING COSTS IN EARLY
SECURITIES MARKETS: THE CASE
OF THE BERLIN STOCK EXCHANGE,
Caroline Fohlin, Johns Hopkins University
Thomas Gehrig, Universität Freiburg and CEPR
Discussion Paper No. 5827
Centre for Economic Policy Research
90–98 Goswell Rd, London EC1V 7RR, UK
Tel: (44 20) 7878 2900, Fax: (44 20) 7878 2999
Email: email@example.com, Website: www.cepr.org
This Discussion Paper is issued under the auspices of the Centre’s research
programme in FINANCIAL ECONOMICS. Any opinions expressed here are
those of the author(s) and not those of the Centre for Economic Policy
Research. Research disseminated by CEPR may include views on policy, but
the Centre itself takes no institutional policy positions.
The Centre for Economic Policy Research was established in 1983 as a
private educational charity, to promote independent analysis and public
discussion of open economies and the relations among them. It is pluralist
and non-partisan, bringing economic research to bear on the analysis of
medium- and long-run policy questions. Institutional (core) finance for the
Centre has been provided through major grants from the Economic and
Social Research Council, under which an ESRC Resource Centre operates
within CEPR; the Esmée Fairbairn Charitable Trust; and the Bank of
England. These organizations do not give prior review to the Centre’s
publications, nor do they necessarily endorse the views expressed therein.
These Discussion Papers often represent preliminary or incomplete work,
circulated to encourage discussion and comment. Citation and use of such a
paper should take account of its provisional character.
Copyright: Caroline Fohlin and Thomas Gehrig
CEPR Discussion Paper No. 5827
Trading Costs in Early Securities Markets: The Case of the Berlin
Stock Exchange, 1880-1910*
Based on daily prices (amtliche Kurse) we estimate effective spreads of
securities traded at the Berlin Stock Exchange in 1880, 1890, 1900 and 1910.
Several extensions of the Roll measure are applied. We find surprisingly tight
effective spreads for the historical data, comparable with similar measures of
the MDAX and DAX at the end of the 20th century.
JEL Classification: D23, G14 and N23
Keywords: effective spreads, market microstructure and price discovery
Department of Economics
Johns Hopkins University
For further Discussion Papers by this author see:
Institut zur Erforschung der
For further Discussion Papers by this author see:
* Without implicating them we thank Will Goetzmann, Cyril Monet, Albrecht
Ritschl, Dick Sylla and especially Jan Annaert for very helpful and constructive
comments. Tobias Brünner, Almira Buzaushina, Seth Freedman, Henry
Gerbrandt and Gwendolin Gürtler provided excellent research assistance. We
are also grateful for the support of Sevtap Kestel and the diligent work of the
data collection team Nikolay Filipov, Mariya Onyshchenko, Patrick Rutkowski,
David Speck and Lei Song – Without their careful efforts this paper would not
exist. Financial support of the Alexander-von-Humboldt Foundation through its
TransCoop Programme as well as the U.S. National Science Foundation is
Submitted 03 August 2006
Price discovery is rather well understood in modern securities markets, yet relatively little is
known about price determination and information aggregation in securities markets of the
past. Securities markets began to emerge by the sixteenth century. These early markets—
such as the stock exchanges in Holland (the Netherlands), England, France, Germany, and the
US—are characterized by the spontaneous emergence of trading institutions in an otherwise
essentially unregulated environment. With the liberalization of incorporation laws throughout
most of the industrializing world in the mid-19th century, and as new technologies and
industries sprang up, a range of equity shares took first place among the securities available in
the marketplaces. These developments naturally attracted increasing attention to the
exchanges, from investors and governments alike. Regulation appeared and developed apace,
as markets grew more formalized, involved more participants, and traded larger volumes of
more securities. Thus, studying early securities markets not only allows us to trace back the
varying trajectories of modern trading institutions; it provides unique real world experiments
The Berlin Exchange between 1870 and 1910 is of particular interest for a number of
reasons. First, the last quarter of the nineteenth century marks the period when active trading
in stocks emerged at the Berlin Exchange—Germany’s dominant securities market of the pre-
World War II era. Not coincidentally, this period also spans the later stages of German
industrialization, the major industries and corporations of which attracted and circulated
significant capital via the securities markets. In this respect, stock markets contributed to the
rapid economic growth in that period. At the same time, the German states unified under a
national system, and that new government promulgated a string of regulations on
incorporation, corporate governance, stock market listing, transaction taxes, and trading. To
some, the most critical of these laws was the comprehensive code of regulation for securities
trading set into force in 1896 (Börsengesetz), which would remain mostly intact until the
Second Financial Markets Directive of the EU in 1994. The 1896 law marked the first
instance of direct government restriction on the operations of the German markets,
exemplified by the prohibition on futures trading in all but a few securities. Debates continue
over the importance of the 1896 law, and tracking changes in effective spreads can provide
critical insights into its net costs on efficiency.
One of the stipulations of the 1896 law separated banking functions from market
making functions, and this point raises a second set of reasons for studying the Berlin market
of the pre-WWI era. In particular, the period in question covers the ascent of the joint-stock
universal banks to a position of national prominence and heavy involvement with certain
sectors of the economy. Already by the first years of the twentieth century, and made famous
by Alexander Gerschenkron in the 1960’s, the universal banks’ role in fueling industrial
growth grew to mythic proportions. The banks’ dominance in post-World War II Germany
solidified their reputation as the key component in German corporate finance. With this
institutional design and historical backdrop in mind, the recent finance literature considers
Germany, and other countries with universal banks, as bank-based financial systems and
draws a stark dichotomy between these and market-oriented (specialized banking) systems.
Moreover, due to Germany’s civil-law tradition, the so-called ‘law and finance’ literature
views the country’s financial-regulatory system as relatively unfriendly to markets; assuming
that the legal system suppresses securities markets, making them small, thin, and
informationally inefficient.1 Thus, the common view over the past few decades displaces
securities markets to the margins of Germany’s corporate finance system.2
1 See, for example, LaPorta et al (1998) on the apparent relationship between legal protections for investors and
the type of legal system in place—common-law versus various types of ‚civil’ law tradition—in a cross-section
of countries. For a compact summary of the debates over ‘bank-based’ versus ‘market-based’ financial systems,
see Levine (2002). For a more complex picture of comparative financial systems, see Allen and Gale (2000).
2 See the recent volume surveying the German financial system, edited by Krahnen and Schmidt (2004).
Ongoing debates over financial system design still tend to view banks and markets as
competitors or substitutes, leading to arguments about whether one or the other is consistently
dominant in certain systems and to the current notion that these positions may have changed
over time.3 To address these debates at the most basic level, we need to assess the functioning
of early markets relative to other markets during the same time and compared with their own
performance over time. This study focuses specifically on the cost efficiency of Germany’s
main corporate securities market during its formative period of development. To be sure, the
topic is narrow; but since we are examining the performance of a market operating within a
system of universal and relationship banking, our results relate to the broader issue of the
relationship between banks and markets: Can securities markets function efficiently in the
presence of universal banks? Is there anything inherent in a civil-law tradition or in universal
banking systems that hampers markets?
Thus, the current paper relates to the law and finance literature, and to the larger
questions of complementarity between banks and markets. Given what we know about the
German financial system of this period, we start out with the view that banks and markets can
work as complements, and that some level of systemic complexity may well improve
efficiency in the mobilization of financial capital.4 While the analysis here cannot be
construed as a direct test of a particular position in these debates, a finding of well-
functioning markets would run counter to the notion that strong, universal-relationship banks
supplant or inhibit securities markets.
The recent finance literature has developed a range of methods to test for the costs—
implicitly the efficiency—of securities markets. We focus here on measuring effective
3 See, most notably, Rajan and Zingales (2003), who coined the term ‘great reversals,’ to capture the idea that in
many continental (civil law) European countries, financial development—including that of securities markets—
superceded that in the U.S. and other common-law countries prior to World War I and that those positions
reversed during the depression and World War II. Of course, such a characterization is necessarily broad-brush
and not focused on institutional detail.
4 See Fohlin (2006) for an in-depth discussion of the evolution of the German corporate finance system prior to
World War I, in particular emphasizing complexity as well as balance and even symbiosis among segments of
the system—especially universal banks and securities markets.
trading costs, or spreads, in the Berlin market, using the Roll (1984) measure and refinements
thereof. Once we understand market functioning on average, we can determine how
regulation affected price discovery later on, analyze the determinants of spreads at the firm
level, and draw comparisons with other markets in other countries or later periods. If the
“bank-domination” line of reasoning proves correct, then securities markets in universal bank-
based systems may well produce higher trading costs. In particular, if dominant banks
suppress market functioning and make for less efficient trading, then Germany should have
higher estimated bid-ask spreads than the US, regardless of time period.5
On the contrary, our results indicate that trading costs were low in the Berlin market in
the decades before World War I, even by modern US standards and certainly by recent
German standards. These findings suggest two conclusions: First, universal banking in itself
does not suppress market functioning or make them inefficient; using only recent data creates
a skewed impression of the connection between universal banking and poor market
performance. Second, securities markets functioned very well even 100 or more years ago.
Indeed, given our modern comparison, German markets may have functioned more efficiently
then than now. Modern-day economists often assume that national and international markets
operated less efficiently in the distant past. The findings in this paper should help to change
that view; demonstrating that market efficiency is not a recent phenomenon, and that market
evolution has been highly uneven across countries and over time.
The results here underscore the usefulness of historical data for providing a much
longer horizon over which to study the performance of markets and the role of institutions.
While high-frequency data now available ostensibly provide statistical significance, economic
phenomena evolve over much longer time-spans. In some respects—particularly international
market integration, rapid technological change, and government regulatory stance—the pre-
5 Of course, prior to the passage of the Glass-Steagall Act in 1933, the United States permitted financial
institutions to combine investment and commercial banking services. Nonetheless, few would categorize the
U.S. financial system as bank-dominated at any point over the past 100 to 125 years—especially not in
comparison to Germany.
World War I economy and financial markets are more similar to their modern counterparts
than were those of the inter-war and early post-World War II eras. Thus, a very long-run
perspective offers modern finance the opportunity to make far more robust claims about the
performance of securities markets and how efficiency might be promoted in the future.
2. The Berlin Stock Exchange and German Corporate Finance before WWI
Despite Germany’s reputation as a bank-dominated corporate financial system, the German
stock exchanges played an important role in the financial system during industrialization and
long after. Indeed, the German stock exchanges trace their history back far earlier than the
universal banks; the Cologne exchange having been founded in 1553.6 Most governments
heavily restricted charters for new share companies until the second half of the nineteenth
century, meaning that few shares existed for trading on exchanges. The same was true in
Germany, and the exchanges therefore listed primarily debt securities—heavily municipal and
national, but also corporate—during this time. Banks, insurance companies, and railroads
comprised the principle sources of equities on the exchanges until the 1860’s and early
1870’s, when, like many other countries in that era, the government liberalized incorporation
laws. Even by 1870, a cumulative total of approximately 200 share corporations
(Aktiengesellschaften) had formed in Prussia.7 But after the 1870 company law standardized
incorporation procedures and made the right nearly universal, hundreds of companies
incorporated: over 900 between 1871 and 1873, more than half of which formed in 1872
6 See Chapter 8 in Fohlin (2006, forthcoming) and the shorter historical review in Fohlin (2002). The primary
American market, the New York Stock Exchange, began around the same time (1792). See the historical review
by the New York Stock Exchange, www.nyse.com.
7 See Horn (1979), p. 136.
alone. The population of AGs exceeded 3,000 by 1890 and surpassed 5,000 later that
Securities exchanges appeared in most areas of the German empire by the late
nineteenth century, though many traded primarily in local issues. With the unification of the
German empire in 1871 and the recognition of Berlin as the capital city, Berlin’s exchange
became the dominant German market—a position it retained until its demise in World War
II.9 Trading volume on the Berlin exchange averaged an estimated nine and twelve times that
of Frankfurt and Hamburg, respectively, in the three decades before World War I.10 Tax
receipts (for both new issues and turnover) and numbers of shares listed reinforce the clear
impression that the provincial exchanges lagged behind Berlin.11 In the early years, even
before its emergence as the primary German market, virtually all of the registered
Aktiengesellschaften listed their shares in Berlin. Listings grew rapidly over the century,
though less quickly than the overall boom in company flotations: the Berlin exchange listed
approximately half of all joint-stock firms by the early 1870's, and still almost a third for most
of the 1890's and early 1900's.
Underwriting new issues and admission of shares to official trading
While new issues of equity shares could be floated by direct subscription of shares, it was far
more common by the late 19th century to use underwriting by a universal bank, in which case
the bank purchased the full block of new shares and then sold them off to investors
(Übernahmegründung).12 This practice stemmed at least in part from the incorporation law of
1870 that stipulated full subscription of shares and minimum levels of payments for those
8 Fohlin (2005) chronicles the post-World War I development of the German stock exchanges. Listings
continued to increase rapidly during and shortly after the war. For an overview of several areas of recent
German financial history, see the edited volume by Krahnen and Schmidt (2004).
9 See Marx (1913), and Gömmel (1992).
10 Wetzel (1996), p. 431 (Appendix VI).
11 Note that Frankfurt listed nearly as many foreign securities as did Berlin. See also Wormser (1919), p. 229,
for tax receipts in Berlin, Frankfurt, and Hamburg between 1900 and 1913.
12 Interestingly, the railroad boom in Germany (1840’s to 1860’s) was largely financed by the subscription
method, that is, “successive issue.”
stakes. Firms also needed to insure subscription of shares within a specified timeframe and a
minimum level of attendance at the initial meeting of shareholders. Failure to meet the
regulations meant large losses to the company founders and justified the expense of floating
shares through an informed underwriter—often a Berlin-centered universal bank—who would
hold the shares on its own account, if necessary.
Government regulation of corporate finance typically followed on the heels of crises.
The boom of the early 1870’s ended as abruptly as it started; the massive losses starting in the
mid-1870’s provoked a popular outcry and demands for greater protections for shareholders
via regulations on incorporation, corporate governance, and securities market activities. The
1884 company law, along with exchange regulations of 1881 and 1884, increased the
responsibilities of firms and their governing bodies to perform well, to oversee operations
responsibly, and to reveal more and better information to potential shareholders. To the
extent that these stipulations improved both the quality of listed shares and the information
available on them, the laws may well have enhanced investors’ trust of the securities markets,
lowered trading costs, and encouraged more active trading.
The economy emerged from stagnation in the late 1880’s, prices rose again, and, in
1890 and especially 1891, they fell again. This time, price declines hit agriculture especially
hard, and the large agrarian sector blamed their losses on futures trading in commodity
markets. In 1892, the national government formed a commission to investigate the operation
of the exchanges and recommend remedies. The new stock exchange law that resulted in
1896 imposed a number of new restrictions—fully paid-in equity shares, waiting periods, and
published financial statements, for example–on companies wishing to officially list shares on
the exchange. The 1896 law also created new governing institutions for the stock
exchanges—a commissioner (Staatskommissar), the Ehrengericht (a judiciary body), and the
Börsenausschuss (a committee of experts)—in order to insure closer scrutiny of new issues.
The law also dictated greater independence of the committee tasked with admitting securities
to the exchanges (the Zulassungstelle); stipulating, for example, that half the members must
not be listed in the stock exchange register, a third must not be involved in securities trading,
and nobody involved in a new issue would be permitted any say in the acceptance of that
issue to trading.13 The legislation reinforced the liability clauses of the 1884 law, holding
underwriters accountable for the accuracy of information they provided to investors.
The most notorious provision of the 1896, by far, was the prohibition on futures
trading in the securities of most non-financial corporations as well as in a wide range of
agricultural commodities (grain and mill products). Popular opinion had connected price
volatility to futures trading, but many argued that speculation in futures actually stabilized
prices.14 Debate over the proposed ban raged, but the agrarians and other proponents
ultimately prevailed over industrialists and financiers. The futures ban essentially shut down
the Berlin commodities exchange; it may also have hindered the operation of the spot
securities market.15 By some accounts, however, the courts enforced the provision unevenly,
and, even for restricted securities, some futures trading persisted after 1896.16 The 1908
novelle to the stock exchange law then rescinded the blanket prohibition of futures trading.
Price setting on the exchanges
The pricing process at the German securities exchanges changed little over the half century
before World War I. Throughout the period, the exchanges operated as call auctions, and
brokers set a final, binding price only once per day in order to maximize volume for each
security. In the early years, it is thought that bankers played an active role in price discovery,
even directly performing brokerage functions, particularly in thin markets.
13 See Wiener (1905).
14 Prion (1910, 1929) and Meier (1993).
15 Fohlin (2002) finds little evidence that the new law reduced market activity, but some evidence that tax levies
a few years earlier diminished turnover to some extent.
16 Buss (1913), and Bund der Landwirte (1908).
Legal changes in 1884 and 1896 did tighten provisions regarding the exchange brokers
and attempted to insure their independence. The first step, in 1884, formalized the
appointment official brokers (vereidigte Maklern), stipulated life terms for them, and
prohibited them from trading on their own accounts or with other brokers.17 Despite these
regulations, bankers or other interested parties could possibly influence price setting, if only
through purchases and sales of securities themselves. Critics continued to call for further
reform of the official broker system, and complaints grew particularly loud in the aftermath of
the 1891 crisis.
As with other provisions in the 1896 stock exchange law, the government attempted to
appease the investing public by formalizing the official brokers (called Kursmaklern) and
attempting to further insulate the price discovery process from outside influences. According
to the new law, exchange directors would set official prices solely in conjunction with the
commissioner, secretary, brokers, directors, and representatives of other trades prescribed by
exchange regulations. Long after the law went into effect, commentators still noted the
involvement of banks—particularly the universal banks—in price setting on the exchanges.18
Thus, banks may well have taken part in price discovery in the Berlin market and may have
substantially affected the levels or volatility of prices there. At the same time, their
involvement in price setting would have simultaneously limited the banks’ ability to exploit
their privileged access to information.
3. Methodology: Measuring Effective Spreads
The empirical analysis is based on variants of the Roll measure for effective bid-ask spreads
(Roll, 1984), and a measure of total transactions costs developed by Lesmond, Ogden, and
17 Some brokers apparently often did trade on their own accounts nonetheless.
18 See especially Passow (1920), who described bankers (underwriters) as Schutzpatronen (literally, patron
saints) of their client firms’ shares. It may be inaccurate to assume his commentary applies to the pre-war era.
denote the transactions return on a security i in period t, where
is the stock price.19 Then the Roll measure
is an estimate of security
i’s effective spread. Note that the effective spread is smaller than a quoted spread, since
transactions regularly occur at prices between the quoted bid and the ask prices.
The basic idea of Roll’s spread estimator is that in informationally efficient and
stationary markets all variation in transactions prices is caused by the randomness of buy and
sell orders given and the existence of positive transaction costs. In liquid markets with low
transaction costs, successive individual orders will have little impact on observed transaction
prices. In thin markets, price effects of individual trades may be more pronounced. If
transaction costs are higher, the deviation of transaction prices from true fundamentals will
not be immediately arbitraged, even in efficient markets. Therefore, the covariance of
successive price changes is informative about effective transaction costs.20
Roll argues that the effective spread provides a better estimate of the effective
transaction costs actually incurred by market participants compared to the quoted spread—a
piece of information that may be unavailable in many cases. Indeed, since the Berlin
Exchange operated as an auction market in the period in question, bid-ask spreads would not
have been produced.
To be sure, the Roll measure has well known deficiencies; for example, the
requirement of negative returns covariance. Especially, when (unobserved) securities returns
are positively auto-correlated, transactions returns may also be positively correlated. But even
in the case in which the Roll measure exists, it is downward biased whenever (unobserved)
19 The transactions return is based on observed transactions prices. It should be differentiated from the
unobserved true return of a security, which is based on its hypothetical true value. Transactions returns will
typically differ from true returns, because even in efficient markets transactions costs prevent arbitrage, when
true returns and transactions returns are close enough. In the sequel we will reserve the term “returns”
exclusively for transactions returns.
20 See Madhavan (2000) for a more technical survey on the empirical estimation of transaction costs.
securities returns are positively auto-correlated (George, Kaul, and Nimalendran, 1991).21
While we can assess the likely size of the bias based on Harris (1990), we employ a
refinement of the spread estimates based on George, Kaul, and Nimalendran (1991) to correct
for positively auto-correlated securities returns. The GKN approach runs as follows: let
denote the return of the equal-weighted market portfolio and
1, − denote the
one-period-ahead forecast thereof. Then the projection of itr on
1, − (
r ) yields the residuals
, which are not explained by fundamental valuations. The covariance
measure of these residuals should therefore provide a better spread estimate. The GKN
estimator coincides with the Roll measure when the estimated regression parameters
In many applications, including our own, typically
, in which case both effective
spread estimators are downward biased. The GKN-estimator, however, is less biased than the
Roll estimator. If the expected return of the equal-weighted portfolio,
1, −, is a perfect proxy
for the expected (unobservable) securities return, the GKN measure is unbiased.
Since we will see that both measures are not fully satisfactory for our analysis, we also
implement the estimator suggested by Lesmond, Ogden, and Trzcinka (1999). In contrast to
the earlier measures, the LOT estimator measures the complete cost of a roundtrip transaction,
effective bid-ask spreads plus transaction taxes and commissions. It is based on the idea that
arbitrage will take place only outside the band of effective transaction costs around the
security’s true value. So there are thresholds
α such that the measured return itr
depends on the true return
itr in a linear way
21 See Harris (1990) for an in-depth analysis of the statistical properties of the Roll estimator. Also, note that in
Roll’s own study, only a portion of his sample satisfied this negative covariance criterion.
This is the so-called limited dependent variable model, originally proposed by Tobin
(1958) and Rosett (1959).
The estimated difference
α α −
is a measure of the true roundtrip transaction costs.22
Accordingly, this estimator encompasses not only the bid-ask spread but also commissions
and transaction taxes. Therefore, the LOT estimator overestimates the effective bid ask
spread. It provides, however, an unbiased estimator of the roundtrip transaction costs.
Naturally, due to the structure of trading in the pre-WWI Berlin market, we cannot use other
Roll-measure refinements that require information on bid and ask prices or on the nature of
transactions (buy versus sell).
In order to estimate effective spreads, we need daily closing prices for a range of securities
traded at the Berlin Stock Exchange.23 The data for this exercise comes from the
Norddeutsche Allgemeine Zeitung, which published a daily record of the Berlin market. As a
starting point, we chose four years (1880, 1890, 1900 and 1910), and we continue to gather
22 Note that this model can also be generalized readily to the case of non-zero returns of the underlying market
model. In our framework the zero-return assumption cannot be statistically discarded (see Table 0).
23 As a referee points out it might be preferable to use continuous data. While this desideratum is not questioned
the only data available are the call auction data. Moreover, the method we employ was developed and applied to
daily data as well by Roll (1984). Thus, at least, the results are comparable to some extent across time and