The benefits of Empire? Capital market integration in Central Europe, 1350-18001
By David Chilosi, Max Schulze and Oliver Volckart
[Preliminary version: please don’t quote without permission]
Abstract. By analysing a newly compiled dataset of interest rates on urban annuities, this
paper finds that the capital markets were better integrated in pre-modern Central Europe
than in Italy. It argues that the institutions of the Holy Roman Empire are a credible, albeit
partial, cause of this difference.
1. Introduction
With the aim of studying the Smithian origins of modern economic growth, research has in
recent years increasingly turned to the analysis of pre-modern market integration. However,
to date, the focus has been on grain markets2 and there is a lack of studies of a large set of
financial markets over long periods of time. In consequence, only tentative generalisations
on their long-term development are possible.3 This neglect is unfortunate. Capital had a
1
We would like to thank the Leverhulme Trust for funding the research, and Alexandra Sapoznik and
Angela Ling Huang for their data collection.
2 The emerging consensus is that there was progress in both regional and continental integration over
time in pre-modern Europe (Persson, 1999; Epstein, 2000; Federico, 2011, 2012; Bateman, 2011,
2012; Chilosi et al., 2013; Federico et al., 2015; Clark, 2015).
3 Keene (2000) demonstrates extensive financial interactions between London and the rest of
England between 1300 and 1600, focussing on the debt market. Chilosi and Volckart (2011) find that
the money market in Central Europe integrated during the fifteenth century. Pezzolo and Tattara
(2006: 1112-1117) generally find a high level of integration across sixteenth-century Italian money
markets. Using high-frequency data on exchange rates, Li (2012) detects much lower transaction
costs for the triangle Amsterdam-Hamburg-London in the 1670s than those that obtained in the 1540s
for London-Antwerp. However, Kugler (2011, 2013) and Bernholz and Kugler (2011) find that already
by 1600 the London-Amsterdam pair was better integrated than Seville-Medina del Campo, where
transaction cost were in line with late medieval Basel. Yet the latter were significantly higher than
those measured by Volckart and Woolf (2006) between Flanders, Lübeck and Prussia at around the
same time. Lothian (2002), looking at short-term interest rates, and Neal (1985, 1987, 1993),
focussing on stock markets, find that London and Amsterdam were well-integrated in the eighteenth
century. In Neal’s and Quinn’s (2001) view, this was favoured by the rise of London as a financial
centre from the later seventeenth century. Consistent with this perspective, Schubert (1988) identifies
the same period as key for the progressive integration of European money markets, like London,
1
more favourable weight-value ratio than practically any other good. Capital market
integration is therefore closely related to the evolution of transaction costs (Coase, 1937).
These are seen by new institutional scholars as the main determinant of market formation
and therefore growth in pre-modern Europe (North, 1981; Greif, 2000), but are notoriously
difficult to measure (North, 1987: 427-428; Wang, 2003).4
This paper analyses long-term capital market integration from a novel direction: we examine
an original dataset of spreads between interest rates on urban annuities between 1350 and
1800.5 The focus is on Central Europe, since it pioneered the development of this financial
instrument. However, an open market for urban annuities developed also in Italy from the
sixteenth century, thus providing a benchmark for comparison. This comparison is germane
to investigate the political and financial roots of transaction costs: while Italy was the most
financially advanced area of Medieval and Renaissance Europe, it was more politically
fragmented than Central Europe, which was ruled by the Holy Roman Empire.6 The analysis
also draws on capital flows between cities for robustness and to identify clusters of urban
markets.
After presenting the interest rates data (Section 2), Section 3 examines trends in their
spreads. It finds long-term progress only in Central Europe; this was concentrated in the
sixteenth century, while there was some divergence in the seventeenth century. A similar
pattern is observed in Italy, but there integration was much weaker and the spreads
Amsterdam and Hamburg, though Paris remained peripheral. By the mid-eighteenth century panEuropean monetary links centred around Amsterdam, London and to a lesser extent Paris and
Hamburg are found by Flandreau et al. (2009) (see also Sperling, 1962).
4 The cost of capital represents, of course, the main measure of transaction costs used in pre-modern
settings. Thus, a number of studies (North and Weingest, 1989; van Zanden, 2009; Zujderduiijn,
2009; Dincecco, 2009, 2011; Stasavage, 2011; Chilosi, 2014) emphasise that the cost of public
borrowing signalled institutional quality. Others, however, stress how this cost crucially depended on
financial development and the opportunity cost of investing in the public debt (Epstein, 2000;
Gelderblom and Jonker, 2004, 2011).
5 Research on pre-modern annuity markets has until now never considered our perspective. The only
and very partial exception is Bateman (2012: 120), who detects low spreads in the cost of capital
across Genoa, the Low Countries, England and Germany by the fifteenth century.
6 The Holy Roman Empire only nominally ruled over Northern Italy in our period.
2
remained more resilient than in Central Europe. The analysis of the origin of inter-city
investors and the role of distance in Central Europe (Section 4) demonstrates that the
majority of investments were within short distances, both before and from the sixteenth
century. However, the distance of investments, as well as their size, increased at the same
time as the spreads between distant places declined. The regional analysis (Section 5) finds
that the Central European capital market was a polycentric network and the cluster of cities
around the Hanseatic ports of Lübeck and Hamburg was particularly developed. Long-term
progress in Central Europe was driven by integration between initially peripheral markets in
Holland, Flanders and Upper Germany and the rest. Section 6 interprets the findings. It
argues that the institutional competition between cities is the proximate cause of an
integrated capital market in Central Europe. The Holy Roman Empire is a credible, but
partial, deep cause. Section 7 concludes.
2. The interest rates
We employ so far unused primary data, i.e. interest rates on annuities, and econometric
methods, exploiting the insight that in a perfectly integrated market interest rates converge to
identity. To ensure comparability, we examine interest rates paid on public urban annuity
markets, the so-called perpetuities and life annuities. These were widely used in Central
Europe (modern Germany, Austria, Switzerland, Benelux and North-Eastern France). As
their price was not age-related,7 the ratio between the yearly pay-out and the purchase price
is sufficient to derive the cost of capital.8
7
We found only very few exceptions to this rule and they have been excluded. Although the sources
were not always clear on how many lives were covered, an effort has been made to exclude life
annuities on multiple lives, for comparability’s sake.
8 As detailed below, the great majority of interest rates were computed thus.
3
In the analysis of the spreads between cities we consider only nominal interest rates and
neglect real ones, for three reasons. Firstly, cross-sectional differences in real interest rates
are bound to be imprecisely measured, given that inflation figures are only available for a
few selected cities. Secondly, differences in inflation are not expected to matter much: premodern price movements tended to be similar in the long-run over relatively large areas, as
common shocks, like the “price revolution” of the sixteenth century, trumped synchronic
differences (cf. Allen, 2001; Malanima, 2002). Finally, price changes were at best poorly
anticipated by pre-modern investors (Chilosi, 2014: 903). Therefore nominal interest rates
provide more reliable guidance to their decisions than real ones.
A related issue is that at times annuities, both on the primary and, especially, on the
secondary markets, were not sold at par. Yet, as shown by Chilosi (2014: 889-91; 2015) for
pre-modern Italy, this issue is much less serious for openly sold annuities than it is for forced
loans, which have therefore been excluded from the analysis. We expect the issue to be
even less problematic for Central Europe. When, as in Amsterdam and Lille, we were able to
compare official rates from the edicts and those from actual transactions there was invariably
an exact match, suggesting that annuities tended to be issued at par on the primary markets.
In Germany, we only rely on recorded sales, rather than edicts, and hence, unless discounts
were concealed, our rates can be considered as markets’ yields. Moreover, it is apparent
from the sources that there annuities were, as a rule, issued at all times, rather than only at
times of need. In consequence, selling annuities only when the market was buoyant was
rarely an option for the German cities and thus there should be a close match between
primary and secondary yields. Indeed, since often contracts gave the right to the buyer to
redeem the capital from the city under relative loose conditions, 9 the incentive to sell on the
secondary market at below par were comparatively weak in the area. The little direct
9
Gilomen (1984: 195-6), Hanover City Archives, NAB 8242; Urkunden Abteilung III.
4
evidence that is available on the Central European secondary markets also suggests that
urban annuities were mostly sold at close to par. We have been able to compare 23 yields
from the same place and year on the primary and the secondary markets and the differences
tend be very small, 0.28 percentage points, on average.10 Feenstra (2014: 15-16), too, find
that for bonds issued by the Dutch province of Zeeland in the second half of the eighteenth
century the secondary market yields were very close to the official rates.
Another issue concerning the comparability of data is that sometimes authorities taxed the
pay-out (Steuer, Losung, Schoss), or, on the contrary, offered benefits to the buyers, such
as exemptions from civic duties (e.g. Dingpflicht, Dinstpflicht, Wachtpflicht, Meinwerk, Bede).
We have detailed and comparable information on which asset was taxed only for Nuremberg
and Brunswick. For Nuremberg the extent to which taxation made a difference varied over
time; in fact, only in the last decade of the fifteenth century it was stationary, implying a
normalising ratio between taxed and untaxed rates of 1.21; otherwise we excluded all those
instances where taxation made a difference which could not be precisely determined. In
Brunswick, on the other hand, exemption from taxation and civic duties recorded for life
annuities had a small impact on the interest rates: we compared 186 interest rates on life
annuities between 1396 and 1590 and found that the interest rates with an exemption from
taxes and civic duties on average only differed by 0.37 percentage points from those life
annuities not including those benefits. In the other places, we had to assume that the payout was not affected by taxes or benefits.
10
The sale of annuities is already documented in sources from the 13 th century; in Lüneburg,
Hamburg and Bern urban public annuity (Stadtrente) letters gave the right to sell on the secondary
market since the second half of the 14th century (Gilomen, 1984: 193-195). Our yields are from
Nuremberg, Hanover and Münster between 1412 and 1680. Hanover City Archives, NAB 8242, p.
127/3 & 202/2, 441/1; Münster City Archives A IX 139 & 392, 393 & 395; State Archives Nuremberg,
Bestand Losungsamt, vol. 69, fol. 66v, 80v, 81v, 91r, 133v, 217r, 219r, 220r, 222r; vol. 70, fol. 9r-v,
11r, 93v, 106r, 109r, 115v, 117r, 121r.
5
While the analysis of dispersion is often accompanied to those of co-movement and of the
speed of adjustment, our data are ill-suited for the latter analyses: the yields exhibit little
volatility, their frequency is yearly and the panel is unbalanced. At the same time, since the
spreads are mainly determined by transaction costs rather than market efficiency their
analysis is arguably more telling (Federico, 2012). Though popular, the use of yields’
dispersion as a measure of capital market segmentation suffers from two main limitations.
For one, similar conditions of supply and demand produce similar interest rates also in the
absence of arbitrage. Hence, it is advisable to complement the analysis of interest rates’
dispersion with that of capital flows. Reassuringly, the results turn out to be very similar11
and capital flows highlight that although most transactions were local, our Central European
cities were part of the same trading network: they were at least indirectly linked through
relatively dense ties. Thus, the twelve best-covered cities in our sample show a network
density (42 per cent), or proportion of actual over potential links, much higher than the mideighteenth century European money market (12 per cent, Flandreau et al., 2009: 160) and in
the same order of magnitude as that obtaining between nation-states in the present day
bond market (57 per cent, Schiavo et al., 2010: 392). A network distance value of 1.6 signals
that most cities were connected either directly or through a third place.12 The second issue is
that differences in yields may reflect differences in risk, rather than lack of capital integration.
This implies bias not only in the levels, but also and rather more harmfully, if risk changes
over time, in the trends. Most analyses agree in identifying political regimes as the major
11
See Section 3 and Section 4.
Analysis based on the 12 cities with at least 80 observations on the origin of foreign investors
(Basel, Bremen, Brunswick, Erfurt, Frankfurt, Halle, Hamburg, Hanover, Hildesheim, Leiden,
Luneburg and Nuremberg). Although this sample is biased towards North Germany it also includes
important centres from other areas and mitigates the negative bias on the connectivity implied by
failure to record existing links. To compute the network distance figure, Leiden had to be excluded, as
we found only indirect links (through Amsterdam) between this city and the main network.
12
6
determinant of risk in pre-modern Europe.13 Our focus on urban annuities implies a similar
institutional setting across places14 and typically political regimes only rarely change, but we
cannot exclude that, for example, changes in the degree of urban autonomy may affect
differences in risk between cities over time.
Altogether we use about 29000 observations on both perpetuities15 and life annuities16. More
than 80 per cent of the data are from archival sources. The data collection focused on what
we expected to be the richest sources; their main types are: annuity letters, ledgers, urban
accounts and edicts (Table 1).
13
See footnote 4.
Even if for the Italian capitals the interest rates refer to the regional states’ debts, it seems that most
creditors were from the capital city, rather than the wider territory, not only in republics like Venice but
also in principalities like the Grand-duchy of Tuscany and, to some extent, the Papacy (Felloni, 1971:
145; Masini, 2007: 205; Stumpo, 2007). Otherwise the label “urban annuity” is stretched somewhat
only for some observations from Holland, which refer to the provincial debts.
15 Referred to as ewiggeldt, ewiger zins [tinß], ewigrente, erffrente, jargulde, jarlik/ jerlich gült/ rente/
zins, jerlich widderkeuflich/ wiederkouffig zcinse, losrenten/ rente tor lose/ ten lossen, wedderkop
rente, volstendiger wiederkauff, jarlik wedderschacke/ wedderschat, rente heritable, luogo non
vacabile. Term annuities were excluded in Central Europe as they could not always be easily
distinguished from short-term loans and as such the impact on the interest rate was idyosincratic.
1087 observations from Italy refer to monte di pietà deposits, term annuities, tax alienations and
censi; their rates have been normalised to perpetuity as detailed in Chilosi (2014: 892-893). Censi,
absent in that database, are considered here to be equivalent to annuities, as they were characterised
by very similar conditions of purchase (Munro, 2003).
16 Referred to as liffgedinghe, (jarlik) lijfftucht (rente), jarliche lijffrente/ liiftuchtige rente, onloßbair/
losbair lijfrente, jerlicher leipzins, jarlix renthe so lange he lyvet und levet, pension ad vitam, rente
viagère, luogo vacabile.
14
7
Table 1: observations, by source type and annuity type
Annuity Type
Total
Perpetuity
Life Annuity
Letters
10601
9179
1422
Ledgers
7581
6862
719
Urban Accounts
7821
5001
2820
Edicts
438
360
78
Source Type
Historian
2627
2313
314
Total
Sources: see the Appendix.
29068
23715
5353
The types of sources vary in the quality and quantity of information they provide, though the
delimitations between them are not always sharp. Letters offer the most complete
information as they state the complete contract in text form; margin and dorsal notes
sometimes give additional information, e.g. on later changes and redemptions. Ledgers are
the most heterogeneous group of sources as they include all books and documents kept to
administer urban annuities. Consequently, ledgers give information to a varying degree of
accuracy. At best, annuity ledgers record probably all sold annuities, briefly listing the
conditions of the contract, as well as transfers, conversions and redemptions (e.g.
Nuremberg ledgers). However, mostly ledgers give only partial information on the contract
(and its history). Pay-out ledgers, for example, sometimes do not record the original interest
rate, but only a later conversion. In a few cases, we had to make assumptions on the dating
of the conversion, by comparing different type of sources for consistency. Urban accounts
(German:
Kämmereirechnungen/Stadtrechnungen;
Dutch:
Stadsrekeningen;
French:
Compte de la ville) record sold annuities as part of the revenues of a town. As several
departments in urban administration could sell annuities, the main accounts (that were
predominantly used by us) do not necessarily document all annuities sold in a year. They
give only very brief information on the contract, stating the name, capital and pay-out or
interest rate, sometimes the date, but mostly only the financial year. In such cases we had to
assume that the annuity was sold in the first year (e.g. for 1525/26, we assumed 1525). We
only collected data from edicts reporting the official rate and (in varying levels of detail) the
8
conditions of purchase in the Netherlands and Italy, which highlights general differences
between these areas and the German territories.
As only in some cases we were able to identify the timing of the transaction with greater
precision than the year, to wash out noise and enable panel analysis, we rely on yearly
means. Figure 1 shows their distribution over time, distinguishing between life annuities and
perpetuities.
Figure 1: Interest rate observations (yearly means), 1240-1809: temporal distribution
30
25
20
15
10
5
0
1240
1340
1440
1540
Perpetuities
1640
1740
Life annuities
Sources: see the Appendix.
Altogether there are 4893 yearly means for perpetuities and 1520 for life annuities. While in
the very early stages life annuities were the main instrument used, their popularity fell over
time, perhaps reflecting a learning process: perpetuities did not require the authorities to
periodically check that the named beneficiary was still alive and therefore implied lower
transaction costs than life annuities. It is possible to analyse yields’ dispersion from the first
half of the fourteenth century; the size of the dataset increases during the fifteenth century
and, especially, during the sixteenth century, when it is augmented by Italian data. However,
9
it falls during the eighteenth century, when there are fewer surviving data from Central
Europe than before, since as cities lost their autonomy they ceased raising urban debts.
Figure 2 shows the geographical distribution of the yearly means.
Figure 2: Interest rate observations (yearly means), 1240-1809: geographical distribution
a) Life annuities
10
b) Perpetuities
Sources: see the Appendix.
11
The interest rates are from 106 cities; 73 cities can be classified as Central European and 32
as Italian.17 While, unavoidably, the coverage is very uneven, thanks to archival research,
the data-set is much broader than those used by previous comparisons, of either urban
annuities or pre-modern financial integration in Europe. The coverage is particularly good for
the German area, especially the North, and good to a lesser extent for Central and Northern
Italy. The figure also shows that life annuities were clearly much less popular in Italy than
they were in Central Europe. For this reason, life annuities are examined only for the whole
sample and Central Europe. Having presented the sources, let us now compare trends in
Central Europe and Italy.
3. Central Europe and Italy
Figure 3 shows how the cost of capital evolved in Central Europe and Italy, by looking at the
nominal yield on perpetuities.18
17
While Chambery is beyond the Alps, it belonged to the Savoyard state. Thus, data from there are
used to investigate patterns of capital integration both in Italy and Central Europe. Barcelona,
Toulouse and Valencia belong to neither Italy nor Central Europe.
18 Although the trends are vulnerable to be biased by composition effects, controlling for this factor
with fixed effects does not alter the results. The rates paid on life annuitities also detect similar
patterns.
12
Figure 3: Nominal interest rates on perpetuities in Central Europe and Italy, 1296-1809
(yearly means, in percentage)
12
10
8
6
4
2
0
1296
1396
1496
Central Europe
1596
1696
1796
Italy
Sources: see the Appendix.
The figure highlights that, firstly, even if there was an obvious long-term decline in nominal
interest rates in Central Europe, contrary to received wisdom (Homer and Sylla, 2005), this
was clearly not steady. In particular, two periods of decline can be distinguished: a late
medieval fall, lasting from the beginning of the fourteenth century until the early fifteenth
century and another one concentrated in the second half of the seventeenth century. In
between and after these periods the rates stagnated. Secondly, while Italy shared in the
decline of the second half of the seventeenth century, its previous development sharply
differed, as it witnessed a rapid fall in the sixteenth century, mirroring that experienced by
Central Europe before the mid-fifteenth century. Thirdly, in spite of Italy’s financial
development, the average cost of capital at the beginning of the sixteenth century was much
higher than in Central Europe. It is during the sixteenth century that it converged towards the
Central European norm, to which it remained closely aligned until the end of our period.
The standard measure of divergence from the law of one price used in the historical study of
market integration is the coefficient of variation, which captures the level of dispersion
around the mean normalised by it, so as to enable inter-temporal comparison; the mean
13
estimates the “market price” and the smaller the coefficient variation, the greater the level of
financial integration. In this context, the limitations of this measure are obvious: the panel is
heavily unbalanced and the results are therefore vulnerable to be biased by composition
effects. To address this issue, fixed effects panel analysis of the yearly rate of change of the
spread around the mean is used, instead (Table 2).19
Table 2: Long term trends in capital integration: fixed-effects panel analysis of the spread
around the mean
N
Years
Initial ratio
Beta*100
Delta*100
Sample
Asset
All
Perpetuity
1344-1805
4863
1.192
-0.002
-1.07
All
Life annuity
1320-1794
1364
1.135
-0.006
-2.88
Central Europe
Perpetuity
1344-1802
3407
1.166
-0.013***
-6.03
Central Europe
Life annuity
1320-1794
1273
1.139
-0.010**
-4.44
Italy
Perpetuity
1493-1800
1367
1.250
0.009*
*=Significant at the 10 percent level.
**=Significant at the 5 percent level.
***=Significant at the 1 percent level.
Notes: N=sample size; ratio=fitted value; Beta=yearly rate of change; Delta=cumulated change.
Sources: see the Appendix.
2.79
The yearly rates of change and the associated cumulated changes show that in Central
Europe but not in Italy there is evidence of long-term progress.20 In addition, the initial values
show that spreads were much larger in Italy than in Central Europe. Long-term trends
obscure non-linear patterns of convergence and divergence, which are expected, particularly
if one investigates very long periods of time, as we do here. To examine non-linear trends,
following on the previous analysis, we look at fixed-effects panel analyses of the dispersion
19
Specifically, the dependent variable here is the absolute value of the natural logarithm of the ratio
between the interest rate in each place and the average interest rate. Here and subsequently,
“spread” or “ratio” refers to the exp of this variable, unless otherwise indicated. The use of the
coefficient of variation or pairwise spreads, instead, produces qualitatively very similar results.
20 For Italy, stagnating or even diverging spreads tallies with Chilosi’s (2014: 900) analysis of the
spreads between the capitals.
14
around the mean against 10-years dummies (Bateman, 2011). Since this type of analysis is
particularly demanding in terms of sample size, we consider perpetuities only (Figure 4).21
Figure 4: Trends in capital integration: fixed-effects panel analysis of the spreads around the
mean (log-scale)
0,4
0,35
0,3
0,25
0,2
0,15
0,1
0,05
0
1340
1390
1440
1490
All
1540
1590
Central Europe
1640
1690
1740
1790
Italy
Sources: see the Appendix.
The graph shows that the spreads in Italy were much higher than in Central Europe
throughout the early modern era. It also confirms that only Central Europe saw long-term
progress: while the spreads declined in the sixteenth century and increased in the
seventeenth century across all areas the swings were much less marked for Central Europe
than for the other two samples.
21
For the same reason, in the analysis of Italian trends we neglect the few isolated instances of issue
from before the 1520s.
15
In Italy significant differences in risk, with assets issued in the Spanish territories (the Duchy
of Milan, and the Kingdoms of Naples and Sicily) being particularly risky, contributed to
keeping large gaps (Chilosi, 2014). However, that the Italian spreads were comparatively
large holds even after excluding observations from these territories.22 The difference
between the Italian and Central European spreads is especially startling if one considers that
the average distance between the Central European cities in the sample (379 km) is much
larger than that between the Italian cities (280 km). That Central Europe was better
integrated than Italy is also confirmed by the origin of the investors.23 In the middle ages,
bonds issued by the Italian city-states were mostly locally held (Molho, 1995: 107-8;
Pezzolo, 2005). Capital flows across Italian cities intensified in the early modern era, but the
majority of bonds remained in local hands: the Italian city where foreigners owed the
greatest share of the debt was probably Venice where by the later seventeenth century
about one-third of the debt was owned by people from outside the republic, mostly from
Genoa (Felloni, 1971).24 In contrast, a large part of the urban debt was owned by strangers
22
In some respects the Italian sample is also comparatively heterogenous in terms of assets; thus, it
was not always possible to identify the term of the monte di pietà deposits; yet excluding these assets
and the censi as well does not significanty alter the patterns observed.
23 That links across the Alps remained overall weak is consistent with evidence on the origin of the
investors, too. Thus, most recorded investors in Italian annuities came from the peninsula, though
there were exceptions and Central European investors (from Switzerland and Germany), in particular,
have been noticed in the Venetian market in the second half of the sixteenth century (Chilosi, 2014,
Online Appendix: 4). Similarly, although the Genoese invested large sums in the national debts in
France and Austria (Felloni, 1971), in our dataset (cf. Section 4 for details) there is only one instance
of an Italian investor in Central European urban annuities: in 1417 an investor from Milan bought an
annuity issued by Basel.
24 Italian historians emphasises the role of Genoese capitalists, who became particularly active within
the peninsula from the seventeenth century, in integrating the Italian markets (Felloni, 1971; Pezzolo,
1995). Our trends suggest that capital investments from Genoa intensified in spite of rising transaction
costs and were driven by the supply shock implied by the severance of the Genoese’s links with
Spain, following the defaults of the crown, and a growing demand, in the wake of expanding debts
(Chilosi, 2014: 897).
16
from the very beginning across Northern France, Flanders, Holland and Germany.25 Capital
flows are further considered in the next section.
4. Capital flows and distance
For 52 Central European places, the sources allowed the collection of data on the origin of
investors from other places. After excluding trivial links with villages within the hinterland,26
the sample consists of 4541 intercity links from 915 pairs. In most cases (4095), we were
also able to identify and convert the capital invested into the cost of yearly “respectable”
subsistence baskets.27 A first way to gain insights into how capital.flows evolved is to
investigate trends in the distance of investments and their size. To control for changes in the
25
The first documented case of an urban annuity in 1228 at Troyes involved the sale of several life
annuities to financiers from Arras and St. Quentin. Subsequently, in December 1232, 26 of the 32 life
annuities sold by Troyes were bought by Rheims financiers. At almost the same time, in 1235,
Auxerre also sold its issues chiefly to Rheims financiers. In around 1275 Ghent found most of its
purchasers in Arras (Munro, 2007: 10). Indeed, in the thirteenth century the Flemish towns sold life
annuities only to outsiders (Fryde and Fryde, 1963: 540). In 1346-7 Ghent sold virtually all the
annuities in the neighbouring Duchy of Brabant (Munro, 2007: 21-2). In the late fourteenth century
about half the debt of Douai was held by strangers (Fryde and Fryde, 1963: 529). In Leiden, although
the great of majority of the debt was locally owned at the beginning of the fifteenth century, from the
mid-fifteenth century ownership became more or less evenly spread between local investors and
capitalists from Flanders, Brabant, Zeeland, Sticht and Holland; in Harleem, investors from the same
provinces owned most of the debt throughout the fifteenth century (Zuijderduin, 2009: 178-9). At
Cologne between 1370 and 1392 perpetual annuities were mainly sold at Lübeck, Mainz and
Augsburg and about 60 percent of the annuities of Mainz in 1444 were owned by foreigners. Between
1370 and 1400 capitalists from Lübeck became the main creditors of Lüneburg, and between 1426
and 1453, at times also controlled more than half of the Hamburg’s debt. Lenders from Lübeck and
other Hanseatic cities like Lüneburg, Stade and Wismar similarly dominated the market in fifteenthcentury Bremen (Fryde and Fryde, 1963: 528-529, 545, 547, 553). Although the early modern
geography of ownership is not as well-studied, the results presented in Section 4 suggest that intercity capital flows intensified.
26 Each place is considered to belong to the hinterland of the closest city with at least 5000 inhabitants
sometimes during the period covered here (1228 to 1802). Most places are very close to the selected
hinterland: within 20 km in 85 per cent of the cases, and 70 km at most.
27 The price of a consumption basket is that of Strasbourg, since for this city the longest series of
consumer price index in Central Europe is available (1326-1875). The series has been extrapolated
backwards to 1264 on the basis of London data. As the price index is very volatile, for the
conversions, we used a smoothed trend obtained with an Epanechnikov kernel. Both price indices
are from the Allen (2001) database.
17
composition of the sample over time, as before, we run fixed-effect panels of the natural
logarithm of the yearly means against time (Table 3).
Table 3: Inter-city investment in pre-modern Central Europe: trends in average distance (in
km) and capital (in subsistence baskets) (fixed effects panel analysis)
Initial
Final
Variable
Years
N
Beta*100
Delta*100
value
value
Distance
1228-1802
1434
0.159***
148
47
116
Capital
1328-1798
1320
0.222***
183
31
87
*=Significant at the 10 percent level.
**=Significant at the 5 percent level.
***=Significant at the 1 percent level.
Notes: N=sample size; Initial and final values=fitted values; Beta=yearly rate of change;
Delta=cumulated change.
Sources: see the Appendix.
Clearly, both the distance and the capital figures detect significant increases over the longterm, thus confirming increased capital integration for Central Europe: the rises are both
statistically and economically significant, with average values over twice as big at the end of
the period as compared to the beginning. The figures also show that while inter-city
investments tended to be within relatively short distances, their size was all but small: on
average, the yearly return on an investment in perpetuities was sufficient to feed, clothe and
house about 3 people, i.e. roughly a family.
To gauge when the distance and average size of inter-city investment increased, we regress
the average figures against 50-years dummies, until 1650-1699.28 The trends are then
compared with that of the Central European spreads. To facilitate interpretation the latter are
inverted (so that in all cases a higher value signals increased capital market integration)
(Figure 5).
28
As these variables are much more volatile than the spreads more observations are needed to
produce reliable estimates, which is why 50-years rather than 10-years dummies are used here.
Unfortunately for the eighteenth century, the number of observations is not sufficient to produce
reliable estimates. Thus, for distance, there are only 18 means in 1700-1749 and 8 means in 17501799 as compared to 68 in 1650-1699 and 297 in 1550-1599.
18
Figure 5: Inter-city investment in pre-modern Central Europe: trends in distance (in km) and
capital (in subsistence ratios) compared to the inverted spread (fixed-effects panel analysis)
120
0,92
0,91
100
0,9
0,89
80
0,88
60
0,87
0,86
40
0,85
0,84
20
0,83
0
1350
0,82
1400
1450
Distance
1500
1550
Capital
1600
1650
Spread inverted
Sources: see the Appendix.
The three measures are overall stable before the sixteenth century; there is also agreement
across measures in identifying increased integration in the sixteenth century; in the second
half of the seventeenth century, the capital figures, like the spreads (but not distance) detect
disintegration. Another way to gain insights into the evolution of inter-city investment is to
compare the distributions of distance and capital before and from 1520, as estimated by
Epanechnikov kernels (Figure 6). As well as marking the beginning of sixteenth-century
integration from the standpoint of the spreads, this cut-off date is convenient from the
perspective of data availability, as it splits the sample in more or less equally-sized subsamples: there are 2545 observations from before 1520 and 1998 from that date onwards.
19
Figure 6: Inter-city investment in pre-modern Central Europe: the distribution of distance (in
km) and capital (in subsistence ratios) before and from 1520
0
.005
Density
.01
.015
a) Distance
0
500
1000
dist
1500
Before 1520
2000
From 1520
.006
.004
0
.002
Density
.008
.01
b) Capital
0
500
1000
1500
2000
c1
Before 1520
Sources: see the Appendix.
20
From 1520
2500
The patterns are remarkably similar across time and indeed variables. Most investments
were within 200 km and only rarely did they exceed 200 consumption baskets, both before
and from 1520. While the core remained unchanged, longer right tails from 1520 show that
the reach of the market increased at the high margin: investment reached sizes and
distances that were previously unattained. It is this dynamic that drove the average values
upwards from the sixteenth century.
We now investigate whether a key role for long-distance investment in driving capital
integration is confirmed by the interest rate data.29 Since we identify a sharp drop in the
frequency of inter-city investments after c. 200 km, it is logical to use this cut-off point to
compare long-term trends, once again using fixed effects panel estimation. The results are
shown in Table 4.
Table 4: The integration of capital markets in pre-modern Central Europe: the role of
distance (in km)
Distance
N
Years
Beta*100
Ratio 0
Ratio T
Delta*100
<200
6258
1333-1795
-0.023***
1.225
1.100
-10.181
>200
15183
1320-1804
-0.042***
1.360
1.109
-18.435
*=Significant at the 10 percent level.
**=Significant at the 5 percent level.
***=Significant at the 1 percent level.
Notes: N=sample size; Ratio=fitted pairwise spread; 0=beginning; T=end; Beta=yearly rate of change;
min=minimum; max=maximum; Delta=cumulated change.
Sources: see the Appendix.
29
To make a good use of the available interest rates data, here and in the susequent analysis we
extrapolate missing years from the perpetuities’ series on the basis on the rates paid on life annuities
in the same place at around the same time. Normalising yields on assets with different terms to
maturity is standard practice in financial history (e.g. Flandreau and Flores, 2009; Chilosi, 2014). In
this particular context, it becomes a safe procedure from around the mid-fifteenth century, as else
normalising rates tend to be very idiosyncratic, presumably reflecting the fact that the pricing became
systematic after an initial phase of experimentation. In addition, we linearly interpolate missing
observations when both the adjacent years are covered. Such interpolations are expected to be very
precise, given that the series exhibit very little volatility. In these ways we augment the overall sample
of yearly means by about 12 per cent. Reassuringly, the overall trends detected by the perpetuity and
the augmented samples are identical.
21
The key role of long-distance integration is strongly confirmed by the results: the yearly rates
of change and the associated cumulated changes demonstrate that the progress of longdistance integration was almost twice faster than that of local integration; in consequence,
the fitted spreads show, by the end of the period, pairwise spreads between close places
were no bigger on average than between distant places. We also investigate when spreads
between long-distant places shrank with fixed-effects panel analysis of the pairwise spreads
across distance groups against 10-years dummies (Figure 7).
Figure 7: Pairwise spreads by distance in pre-modern Central Europe: fixed-effects panel
trends (log-scale)
0,5
0,45
0,4
0,35
0,3
0,25
0,2
0,15
0,1
0,05
0
1320
1370
1420
1470
1520
1570
<200 km
1620
1670
1720
1770
>200 km
Sources: see the Appendix.
Again we find substantial accord with the previous analysis. Not much happened in the
fourteenth century. During the fifteenth century only local integration progressed. Important
gains for long-distance integration were made in the sixteenth century, with spreads between
place distant over 200 km steadily decreasing from the 1520s. In consequence, from around
1610 there was hardly any difference in the spreads between close and distant places;
henceforth, in spite of some disintegration during the seventeenth century, for both distance
22
groups the interest rates differed little, about 10-15 per cent, on average. Otherwise put,
given at the time the cost of capital was about 4 per cent, we are looking at spreads of about
half a percentage point, which, by any standard, suggest an integrated market. We now
develop the geographical analysis with an investigation of patterns of integration within and
between Central European regions.
5. Regional integration
The Central European regions are endogenously identified on the basis of the capital flows.
In particular, we rely on the model-based clustering with an ultrametric space developed by
Schweinberger and Snijders (2003). This method shares with the “block analysis” used by
Flandreau et al. (2009) the assumption that observed links are the product of a stochastic
process, and it is therefore suited to our context, where measurement error is expected. 30 It
also has two advantages. Firstly, it is specifically designed to identify analogous groups,
rather than groups of equivalent members (which may be groups with intense capital flows,
but also, say, groups of cities with little or no flows with another group, but that do not
necessarily exchange capital amongst each other). Secondly, it takes into account the
strength of the ties and estimates it across clustering levels, thus making it possible to fully
exploit the information provided by the capital flows.
We can be reasonably confident that the recorded links between cities approximate the
actually existing links only for those cities for which there is a good coverage. Yet,
geographical spread is a concern, too. With this trade-off in mind, we focus on those cities
for which we have at least ten observations on investments made from cities within other
hinterlands and consider other samples for robustness. Furthermore, recorded capital flows
30
This features distinguishes this approach from another popular clustering procedure, that
developed by Newman and Girvan (2004).
23
are bound to increase with coverage. This sample bias is econometrically addressed, by
regressing capital flows between city-pairs against the log of the sum of their sample sizes,
using a negative-binomial specification, since over-dispersion turns out to be an issue and
count regressions have desirable properties for estimating the determinants of bilateral flows
(Silva and Tenreyro, 2006, 2011; Burger, Van Oort and Linders, 2009). The capital flows that
would obtain if all the linked cities were evenly covered are then estimated with the residuals
normalised to take values between zero and one. Hence, our measure of network intensity is
capital flows compared to that of the pair with the highest value: Lübeck-Lüneburg, where we
find that capital flows (18799 consumption baskets) were 17514 consumption baskets
greater than expected on the basis of their combined coverage (698 observations).
The model assumes that links within a cluster are symmetric and that clusters are not
overlapping at a given clustering level. These assumptions are bound to be violated to some
extent, but are shared by the other clustering procedures and imply that groups where the
assumption of direct arbitrage between cities is closely approximated are identified. The
clusters can be identified either with a maximum likelihood estimator or with a Bayesian
estimator. The latter provides a more elegant model selection procedure, but, in this case
yielded unstable and at times un-plausible settings, suggesting a poor fit with the data. We
therefore use the maximum likelihood estimator. As our measure of strength is continuous
we assume a Gaussian distribution. To investigate possible non-convergence, we run ten
sequences. Given that there are 28 cities in the sample we allow up to seven different levels
of clustering. The results are reported in table 5, where the second to the seventh columns
report the expected network intensity at each clustering level for each number of levels.
24
Table 5: Network clustering: expected network intensity by level and number of levels
Level/N. of
levels
7
6
5
4
3
2
1
1.000
1.000
1.000
1.000
1.000
0.275
2
0.476
0.364
0.364
0.364
0.206
0.013
3
0.297
0.163
0.163
0.150
0.010
4
0.148
0.025
0.039
0.009
5
0.009
0.005
0.008
6
0.009
0.000
7
0.009
Log likelihood
-0.491
Sources: see the Appendix.
-0.516
-0.520
-0.531
-0.734
-1.3191
The log-likelihood function is maximised at seven levels, but the difference is big only when
compared to two or three levels and inspection of the output reveals that non-convergence
may be an issue, as evidenced by unstable maxima of the log-likelihood function, for five
levels or more. Moreover, the results are qualitatively almost identical for four to six levels:
all detect very similar values at levels one to three and sharp drops in network intensity
subsequently. Hence, no important information is lost by only considering four levels, and we
therefore focus on this model. The clusters are shown in Figure 8.
25
Figure 8: Network analysis of capital markets in pre-industrial Central Europe: clusters of
cities
Sources: see the Appendix.
The analysis detects that capital flows between Lübeck and Lüneburg were exceptionally
high. The expected flows drop by two thirds at the following level, which identifies two
separate clusters: the main Hansa cities (Lübeck, Lüneburg, Hamburg and Brunswick) and
the Frankfurt-Mainz pair. Expected capital flows remain relatively high at the following level;
this is however much more inclusive than the previous two and detects six clusters, which
are conventionally named Breisgau (Basel, Colmar, Freiburg), Hansa (Hamburg, Lübeck,
Lüneburg, Brunswick, Bremen, Hannover, Hildesheim), Upper Germany (Augsburg, Munich,
Nuremberg), Hessen (Frankfurt, Mainz, Worms), Saxony (Erfurt, Leipzig, Hall), and
Westphalia (Münster, Wesel). As said, network intensity sharply drops at the next level.
26
Central Europe emerges as a polycentric network: consistent with the previous analysis of
the role of distance (cf. Section 4) all the cities with strong links between them are located
within relatively short-distances, with clusters developing around important financial centres,
like Frankfurt, Leipzig, Nuremberg and Hamburg; outside these clusters investment tended
to be much less frequent. In spite of the similarities, the clusters around the Hansa port cities
of Lübeck and Hamburg reached particularly high levels of capital flows and were
significantly wider than the others at the same level of network intensity.31 Running the same
analysis with cities with at least eighty observations of investments from other hinterlands, all
the cities in the sample, before 1520 and from 1520 does not alter these conclusions; the
sole qualification needed is that enlarging the sample reveals a wide cluster around
Nuremburg, which developed strong links with Salzburg and Maribor.
To examine regional integration the remaining cities are allocated to the named clusters on
the basis of geographical proximity, adding two separate groups for Holland and Flanders,
where the latter group includes also the relatively few observations from Northern France.
Table 6 shows long-term trends in the dispersion around the mean within clusters and
between cities in
each cluster and Central Europe; as before, we rely on fixed-effects
estimation.
31
Node degree analysis also finds that capital markets were particularly developed in Lübeck (before
1520) and Hamburg (from 1520).
27
Table 6: The integration of capital markets in pre-modern Central Europe: within and
between clusters
Within integration
N
Years
Beta*100
Ratio 0
Ratio T
Delta*100
Breisgau
241
Hansa
1591
1396-1741
-0.016*
1.093
1.035
-5.318
1351-1750
-0.022***
1.152
1.056
-8.283
Upper Germany
210
1388-1551
0.023
1.092
1.133
3.764
Hessen
108
1550-1760
-0.004
1.038
1.028
-0.922
Saxony
61
1497-1621
0.002
1.052
1.055
0.259
Westphalia
236
1350-1760
-0.002
1.078
1.068
-0.992
Holland
364
1520-1795
0.002
1.082
1.087
0.499
Flanders
167
1392-1775
0.028***
1.027
1.143
11.218
N
Years
Beta*100
Ratio 0
Ratio T
Delta*100
Breisgau
447
1383-1791
-0.008
1.136
1.101
-3.081
Hansa
1653
1320-1804
-0.004*
1.138
1.114
-2.112
Upper Germany
402
1382-1804
-0.037***
1.255
1.071
-14.611
Hessen
232
1410-1797
-0.009
1.116
1.077
-3.516
Saxony
196
1320-1698
-0.011
1.119
1.073
-4.04
Westphalia
415
1350-1780
-0.001
1.108
1.106
-0.225
Holland
501
1422-1795
-0.062***
1.389
1.100
-20.771
Between integration
Flanders
346
1360-1795
-0.060***
1.381
1.063
-23.056
*=Significant at the 10 percent level.
**=Significant at the 5 percent level.
***=Significant at the 1 percent level.
Notes: N=sample size; Ratio=fitted ratio; 0=beginning; T=end; Beta=yearly rate of change;
Delta=cumulated change.
Sources: see the Appendix.
Consistent with the result that market integration was more dynamic over long than short
distances (cf. Section 4), the evidence of integration is stronger between than within clusters.
The within analysis detects very low spreads with the mean – with differences mostly within
10 per cent –, both at the beginning and the end. In other words, by and large, clusters were
integrated already in the late middle ages and remained so. The only and partial exceptions
are the Hansa, where there is some evidence of progress over time, and Flanders, for which
we find disintegration, possibly as a result of divergences between the French and the other
cities.
Between integration mainly concerned a few regions where the spreads were comparatively
high at the beginning. Thus, the integration of the Hansa with Central Europe was high from
28
the late middle ages and progressed little. By contrast, there were initially significant barriers
with Upper Germany and especially Holland and Flanders which were eroded over time.
What should we make of these patterns? The next section interprets our findings.
6. The benefits of Empire?
Let us begin by returning to the contrast between a highly integrated Central Europe and a
poorly integrated Italy. Our analysis (cf. figures 6a and 8) confirm that under pre-modern
conditions information and hence monitoring costs sharply increased with distance
(Stasavage, 2011). Yet, geography clearly did not favour Central Europe, large parts of
which were land-locked. Indeed, Chilosi et al. (2013) find that the early modern Central
European grain markets were less integrated than the Italian ones. Equally, it is
uncontroversial that the Italians pioneered the use of sophisticated financial instrument, the
bill of exchange and bank giro transfers, enabling cashless capital transfers a low cost (de
Roover, 1963; Lopez, 1976; Day, 1987: 141-161; Felloni, 2008). Hence, financial
development falls short of accounting for the observed difference, too.
The origins of large spreads on the Italian capital market had to be institutional in nature.
The roots of a fragmented capital market can be traced back to the times of forced loans in
the middle ages, when the purchase of bonds was imposed on the local well-to-do and there
were institutional restrictions on foreigners’ participation in the secondary markets: outsiders
could purchase such bonds only if they had special exemptions granted as a form of
privilege. It is noteworthy that these restrictions were partly designed to prevent reprisal in
case of default (Sieveking, 1905: 29; Molho, 1995: 107-8; Pezzolo, 2005). In other words,
because medieval Italian cities valued autonomous public finances – the right to transform a
loan into a tax at will -, over access to foreign capital, urban collective liability hindered rather
than helped the development of inter-city financial links. Keeping high returns for the local
oligarchs was arguably another reason for closure (cf. Chilosi, 2014, 2015). The restrictions
on foreigners’ participation were eliminated as markets opened up in the sixteenth century:
29
at that time the contribution of foreigners was often explicitly called for32 and the spreads
shrank (see Figure 4). Nevertheless, the gains were short-lived and foreign investment
remained costly compared to local investment, as strangers were discriminated against
when there were partial defaults, and with taxation, forced loans and liquidations (cf.
Pugliese, 1924: 339-76; Felloni, 1971: 146-147, 214-217, 289, 304-306, 315-317; Calabria,
1991: 128-9).
The difference with Central Europe is stark. There, collective liability and associated
institutions fostered foreign investment during the middle ages and persisted into the early
modern era (Fryde and Fryde, 1963: 528-529, 533; Zuijderduin, 2009: ch. 3; Boerner and
Ritschl, 2002, 2005). Indeed, Fryde and Fryde (1963: 528, 533) argue that, as a result of
collective liability, in Central Europe foreign investment was less risky than domestic
investment. The sources also highlight how the Central European cities actively competed to
attract foreign capital by meeting the needs of the investors. Thus, urban officials, as well as
private agents, acted as intermediaries to facilitate inter-city investment from the early
fifteenth century,33 and at times cities arranged that the pay-out would be delivered at the
buyer’s place of residence, at seemingly no extra cost.34 Less often it is stipulated in the
contract that the payment of the annual pay out and/or the capital in case of redemption
should take place in a place convenient for both parties, mostly at financial fairs, like those of
To make an example, on the 14th of March 1599, the Florentine Senate boasted that: the “buyers …
can be … subjects as well as foreigners … of whatever fate, grade or condition” (ASF, Monte
Comune o delle Graticole, parte I, pezzo 3: 261).
33 Nuremberg State Archives, Bestand: Losungsamt Vol. 69, fol. 21r (no.117), 26v (no. 136), 28r (no.
137), Hanover City Archives, NAB 7228, fol. 19v, Zuijderduijn (2009: 113-115).
34 GAD 1, no. 434, fol. 50v-53r, Brunswick City Archives, B I 11 Leibgedingebücher, vol. 4, fol. 36v39r), Hanover City Archives, NAB 8242_Stadtobligationsbuch_1387-1533, 108/2, 110/1, 118/12,
130/1, 137/3, 140/2, 141/4,156/1, 163/2, 164/2, 177/1, 184/1, 192/1, 197/1, 227/3, 267/2, 268/1,
275/1, 284/1, 287/2, 364/2, 434/1, 437/2, 542/2, 554/2, 571/2, 755/1; NAB Nr. 7228 p.2; Urkunden
Abteilung 3 - Schuldurkunden des Rates, nos. 43, 46, 51, 55, 56, 57, 60, 132, 172, 248, 269, 275,
293, 298, 307, 309, 310, 314, 315, 317, 318, 321, 323, 334, 340, 342, 346, 347, 350, 354, 357, 358,
361, 363, 366, 367, 373, 537, Albers (1930: 49-50), Zuijderduijn (2009: 115).
32
30
Frankfurt and Leipzig, or seemingly specialised markets like Hildesheim.35 By contrast, in
early modern Italy, Genoese capitalists investing in other cities had to pay a fee to a private
agent to have the pay-outs delivered to them (Felloni, 1971: 97).
In short, the Central European cities found it easier to credibly commit to the protection of
foreign investment and compete for foreign capital than the Italian ones, where shorttermism and inter-city rivalries prevailed. The functioning of collective liability in Central
Europe was backed by imperial law (Boerner and Ritschl, 2005: 12), and the institutional
differences between Italy and Central Europe point to an important role for the Holy Roman
Empire in addressing coordination failure and facilitating inter-city investment (cf. Chilosi and
Volckart, 2011). That the cost of long-distant investment mainly fell during the sixteenth
century also sits well with this interpretation: in around 1500 the Empire underwent major
institutional reforms aimed at political and economic integration (Angermeier, 1984: passim;
Whaley, 2012a: ch. 2). Notably, institutions that developed under the aegis of the reformed
Reich included initiatives directly aimed at fostering credible commitment and reducing the
risk of public lending to other polities: thus, imperial debt commissions that organised
creditors’ control over the fiscal policies of defaulting polities were established (Herrmann,
1999; Ackermann, 2002; Whaley, 2012b: 64), implying falls in monitoring costs. Since, as
mentioned earlier, these sharply increased with distance, their reduction is expected to
mainly affect long-distance investment.
35
Hanover City Archives, NAB 8242, Stadtobligationsbuch (1387-1533), 118/12; Abteilung 3 Schuldurkunden des Rates nos. 43, 46, 51, 57; Nuremberg State Archives, Bestand: Losungsamt,
vol. 69, fol. 97r-v (no. 444), 98r (no. 447), 99r (no. 450), 103r (no. 468), 112r (nos. 499-500), 126r (no.
567); vol. 70, fol. 125r (no. 386); Erfurt City Archvies, 1-1/21 10 Libri ordinationum, vol. 1, fol. 1v, 8v;
1-1/21-12/1 Obligationen, 70ff, 75ff, 147ff.; 1-1/22, 2 Hauptrechnungen no. 1, 0-1/ 4- 121 (1);
Luneburg City Archives, AB 55 Kopie von Rentenbriefen (1441-1492), fol. 20r, 76r f., 90r, 93r-94r, 98v
f.; Landeshauptarchiv Magdeburg, Copiar der Obligationen der Stadt Halle, Cop. 395a,fol.
27v,29v,38v, 315r; Cop. 396, fol. 30r, 131r, 134v, 159r; Klinger, 2011: p. 336 ff., fol. 146r-v (no. 570);
Archives de la ville Strasbourg, Série IV No. 71, p. 148; Munster City Archives, Ratsarchiv A IX,
Findbuch zu den Rentenverschreibungen aus Abt. A IX des "Alten Archivs", no. 43.
31
Yet the results of the regional analysis point to two difficulties and imply that the institutions
of the Holy Roman Empire are not sufficient to explain inter-city coordination in Central
Europe. Firstly, the dominant role of the Hansa shows that capital markets developed the
most where traditionally the empire was weak (Whaley, 2012a: 532). Spruyt’s (1994)
argument that city leagues promoting cooperation developed in Central Europe but not in
Italy because the Italian cities enjoyed monopolies in high value trade, had enough
resources to act independently and their autonomy was not threatened by feudal lords allied
with the emperor thus fits the evidence better. While this interpretation implies a key role for
other factors, it is nevertheless hasty to conclude that the Empire opposed or was not
relevant for urban cooperation: the promotion of peace and prosperity defined its role and
shifting alliances rather than a simple opposition characterised the relationship between the
Central European cities, on the one hand, and feudal lords and Empire, on the other. The
interest Charles IV took in the Hansa in the 1370s is a case in point (Dollinger, 1981: 151 f);
another one is the Swabian League of 1488, formed by both cities and feudal lords and
actively promoted by Frederick III to resist the expansionary ambitions of the Bavarian dukes
(Whaley, 2012a: 29). Moreover, integrated capital markets in Central Europe survived the
demise of the city-leagues. Indeed, our results suggest that Ogilvie’s (1992) negative
assessment on the effects of the “seventeenth-century crisis” for the Central European
market needs to be qualified: inter-state rivalries appears to have had a much less disruptive
effect on the Central European than on the Italian market (cf. figure 4), in spite of the fact
that by then the conditions which according to Spruyt (1994) prevented inter-urban
cooperation in medieval Italy had lost much of their force. Evidently, political boundaries
remained relatively porous in Central Europe and it is reasonable to assume that the Empire
played a role in ensuring it: research emphasises the resilience of imperial structures and
influence in all but the largest states, even after Westphalia (1648) (Whaley, 2012a: 1-14).
Secondly, the regions mostly responsible for increased integration in the long-run, Flanders
and Holland, were also located in areas where imperial institutions were weak, and indeed
32
were becoming increasingly so from the sixteenth century (Press 1986; Whaley, 2012a: 2022), when this integration was taking place. These patterns would instead point towards a
primacy of specifically financial developments in explaining long-term progress: the sixteenth
century saw the development of the bill of exchange into a fully negotiable financial
instrument; in consequence, effectively the bill became a form of paper money, thus greatly
decreasing the cost of transferring capital and facilitating long-distance transactions. Closely
matching our patterns of integration, this financial innovation was centered in Antwerp, from
whence it spread to Northern Europe (Munro, 2003: 553 ff.). Still, we also observe increased
integration with Upper Germany, where obviously the Empire was a strong presence.
Moreover, to a large extent, the rise of Antwerp as the dominant financial centre of the
sixteenth century was tied to the fact that it was the place of choice for Charles V to remit
American silver across the Empire and raise loans from the Fuggers of Augsburg and their
associates (Braudel, 1982: 150-151). Hence, the “material life” of the Empire, if not its legal
and fiscal structures, was conducive to the development of the inter-regional links that we
observe.
7. Conclusion
In summary, we find that the capital market was much better integrated in pre-modern
Central Europe than in Italy. Moreover, only in Central Europe do we detect long-term
integration: while in both areas there was progress in the sixteenth century, in Italy the gains
were short-lived. Within Central Europe, even if most investment remained local, increased
integration mainly concerned distant rather than close places. The size and depth of the
market reached comparatively high levels around the Hansa port cities of Lübeck and
Hamburg, and it was initially poorly integrated clusters of cities in Flanders, Holland and
Upper Germany that improved their integration with Central Europe over time. We argue that
the proximate cause of higher integration in Central Europe than in Italy was institutional
competition: more effectively than the Italian cities, the Central European cities credibly
33
committed not to discriminate against foreign investors and competed for foreign capital. The
institutions of the Holy Roman Empire are a credible deep cause of this contrast, though
spatial patterns of integration within Central Europe imply that they fall short of being
sufficient.
A number of implications follow. To begin with, as in other fields, the experience of London
and Amsterdam is atypical rather than representative of pre-modern Europe: while most
historians agree that these cities became financially integrated from the later seventeenth
century,36 the timings of capital integration in Central Europe and Italy sharply differed.
Similarly, an imperfect match with the integration of money markets in our areas (cf. Chilosi
and Volckart, 2011; Pezzolo and Tattara, 2006) suggests caution in generalising from the
public bonds’ market to other markets. Still, institutional competition between the Central
European cities calls into question the idea that state formation is necessary to avert rentseeking and the formation of barriers to entry on the part of the urban elites (cf. Volckart,
1999, 2002; Stasavage, 2011, 2014). Generally, capital integration in Central Europe raises
doubts on the argument that transaction costs are necessarily much lower within territorial
states than between cities (cf. Spruyt, 1994; Epstein, 2000). Finally, weak capital market
integration was associated with economic decline in early modern Italy. However, we traced
the origin of weak integration in Italy to the middle ages, when the peninsula was
economically leading, and it does not look as if pre-modern Central Europe performed much
better than Italy overall (Malanima, 2009). Hence, our results offer only weak support to the
new institutional view that low transaction costs were the main engine of growth in premodern Europe.
36
See footnote 3.
34
References
Ackermann, Jürgen. 2002. Verschuldung, Reichsdebitverwaltung, Mediatisierung. Eine
Studie zu den Finanzproblemen der mindermächtigen Stände im Alten Reich: Das
Beispiel der Grafschaft Ysenburg-Büdingen 1687–1806. Marburg: Selbstverlag des
Hessischen Landesamtes für geschichtliche Landeskunde.
Allen, Robert C. 2001. “The great divergence in European prices and wages from the Middle
Ages to the First World War”, Explorations in Economic History, Vol. 38, No. 4, pp.
411-47.
Albers, Hermann. 1930. Die Anleihen der Stadt Bremen vom 14. bis zum 18. Jahrhundert. In
Veröffentlichungen aus dem Staatsarchiv der freien Hansestadt Bremen, 3, pp. 1163.
Bateman, Victoria N. 2011. “The evolution of markets in Early Modern Europe, 1350-1800: a
study of wheat prices,” Economic History Review, Vol. 64, No. 2, pp. 447-71
Bateman, Victoria N. 2012. Markets and Growth in Early Modern Europe. London: Pickering
& Chatto.
Bernholz, Peter, and Peter Kugler. 2011. “Financial market integration in the early modern
period in Spain: results from a threshold error correction model”, Economic Letters,
Vol. 110, No. 2, pp. 93-96.
Braudel, Fernand. 1982. Civilization and Capitalism, 15th-18th Centuries, Volume III: The
Perspective of the World. Trans. by Siân Reynolds. Oakland, CA: University of
California Press.
Boerner, Lars, and Albrecht Ritschl, 2002. "Individual enforcement and collective liability in
pre-modern Europe: a comment”, Journal of Institutional and Theoretical Economics,
Vol.158, No. 1, pp. 205-213.
Boerner, Lars, and Albrecht Ritschl, 2005. "Making financial markets: contract enforcement
and the emergence of tradable assets in Late Medieval Europe”, Mimeo.
Burger Martijn, Van Oort, Frank, and Linders, Gert-Jan. 2009. “On the specification of the
gravity model of trade: zeros, excess zeros and zero-inflated estimation”, Spatial
Economic Analysis, Vol. 4, No. 2, pp. 167-190.
Calabria, Antonio. 1991. The Cost of Empire: The Finances of the Kingdom of Naples in the
Time of Spanish Rule. Cambridge: Cambridge University Press.
Chilosi, David. 2014. “Risky institutions: political regimes and the cost of public borrowing in
Early Modern Italy”, Journal of Economic History, Vol. 74, No. 3, pp. 887-915.
35
Chilosi, David. 2015. “Relecture des notes sur l’histoire du taux d’intérêt: les marchés
obligataires dans la République de Gênes”, Ressources Publiques et Construction
Étatique en Europe. XIIIe-XVIIIe Siècle, Katia Béguin (ed.). Paris: Institut de la gestion
publique et du développement économique, Comité pour l’histoire économique et
financière de la France, pp. 209-226.
Chilosi, David, and Oliver Volckart. 2011. “Money, states, and empire: financial integration
and institutional change in Central Europe, 1400-1520,” Journal of Economic History,
Vol. 71, No. 3, pp. 762-791
Chilosi, David, Tommy E. Murphy, Roman Studer, and Coskun Tuncer. 2013. “Europe’s
many integrations: geography and grain markets”, Explorations in Economic
History, Vol. 50, No. 1, 46-68
Clark, Gregory. 2015. “Markets before economic growth: the grain market of medieval
England”, Cliometrica, Vol. 9, No. 3, pp. 265-287.
Coase, Ronald. 1937. “The nature of the firm”, Economica, 4:16, pp. 386-405.
Day, John. 1987. The Medieval Market Economy, Oxford: Basil Blackwell.
De Roover, Raymond. (1963/79): The Organization of Trade, in: M.M. Postan, E.E. Rich and
E. Miller, eds., The Cambridge Economic History of Europe, vol. 3: Economic
Organization and Policies in the Middle Ages, Cambridge: Cambridge University
Press, pp. 42-118.
Dincecco, Mark. 2009. “Political regimes and sovereign credit risk in Europe, 17501913”, European Review of Economic History, Vol. 13, No. 1, pp. 31-63
Dincecco, Mark. 2011. Political Transformations and Public Finances: Europe, 1650-1913.
Cambridge: Cambridge University Press.
Dollinger, Philippe. 1981. Die Hanse. 3 ed. Stuttgart: Kröner.
Epstein, Stephan R. 2000. Freedom and Growth: The Rise of States and Markets in Europe,
1300-1750. London and New York: Routledge.
Federico, Giovanni. 2011. “When did European markets integrate?,” European Review of
Economic History, Vol. 15, No. 1, pp. 93-126.
Federico, Giovanni. 2012. “How much do we know about market integration in Europe?,”
Economic History Review, Vol. 65, No. 2, pp. 470-97.
Federico, Giovanni, Max-Stephan Schulze, and Oliver Volckart. 2015. European goods
market integration in the very long-run: from the Black Death to the First World War.
Mimeo.
Feenstra, Alberto. 2014. “Keeping the ship of state afloat: Zeeland’s sovereign debt
management, 1600-1800”, Paper presented at the Leiden International Conference
in Political History.
Felloni, Giuseppe. 1971. Gli Investimenti Finanziari Genovesi in Europa tra il Seicento e la
Restaurazione. Milan: Giuffrè Editore.
36
Felloni, Giuseppe. 2008. “Dall’Italia all’Europa: il primato della finanza italiana dal medioevo
alla prima età moderna”, Storia d’Italia, Annali 23, La Banca, Alberto Cova, Salvatore
La Francesca, Angelo Moioli, et al. (eds.). Torino: Einaudi, pp. 91-149.
Flandreau, Marc, and Juan H. Flores. 2009. “Bonds and brands: foundations of sovereign
debt markets”, Journal of Economic History, Vol. 69, No. 3, pp. 646-684.
Flandreau, Marc, Christopher Galimard, Clemens Jobst, and Pilar Nogués-Marco. 2009.
“Monetary Geography before the Industrial Revolution”, Cambridge Journal of
Regions, Economy and Society, Vol. 2, No. 2. pp. 149-71.
Fryde, Edmund Boleswal and Matthew M. Fryde. 1963. “Public Credit, with Special
Reference to North-Western Europe”, M. Postan, E. E. Rich, and E. Miller (eds.)
The Cambridge Economic History of Europe, III: Economic Organization in the
Middle Ages. Cambridge: Cambridge University Press, 430-553.
Gelderblom, Oscar, and Joost Jonker. 2004. “Completing a financial revolution: the
Finance of the Dutch East India trade and the rise of Amsterdam capital market”.
Journal of Economic History (64), no. 3, 641-672.
_______. 2011. “Public finance and economic growth: the case of Holland in the
seventeenth century”. In Journal of Economic History (71), no. 1, 1-39.
Gilomen, Hans-Jörg. 1984. Der Rentenkauf im Mittelalter. Habilitationsschrift, Universität
Basel. Available at:
http://www.hist.uzh.ch/fachbereiche/mittelalter/emeriti/gilomen/publikationen/Rentenk
auf.pdf (Consulted on the 5th November 2014).
Greif, Avner. 2000. “The fundamental problem of exchange: a research agenda in historical
institutional analysis”, European Review of Economic History, Vol. null, No. 3, pp. 251284.
Herrmann, Susanne. 1999. “Die Durchführung von Schuldenverfahren im Rahmen
kaiserlicher Debitkommissionen im 18. Jahrhundert am Beispiel des Debitwesens der
Grafen
Montfort”,
in
Reichshofrat
und
Reichskammergericht:
Ein
Konkurrenzverhältnis, edited by Wolfgang Sellert. Köln: Böhlau, pp. 111-128.
Homer, Sidney and Sylla, Robert. E. 2005. A History of Interest Rates. 4th ed. Hoboken, N.
J.: Wiley.
Keene, Derek. 2000. “Changes in London’s economic hinterland as indicated by debt cases
in the court of common pleas”, in James A. Galloway (ed.), Trade, Urban Hinterlands
and Market Integration c.1300-1600. London: Centre for Metropolitan History, pp.5981.
Klinger, J. (ed.) 2011. Das Dresdener Stadtbuch 1477-1495. Edition und Forschung
(Inaugural-Dissertation zur Erlangung der Doktorwürde der Philosophie, Kunst- und
Gesellschaftswissenschaften der Universität Regensburg). Regensburg.
Knipping, Richard. 1894. “Das Schuldenwesen der Stadt Köln im 14. und 15. Jahrhundert“,
Westdeutsche Zeitschrift für Geschichte und Kunst, Vol. 13, pp. 340-397
Kugler, P. 2011. “Financial market integration in Late Medieval Europe: results from a
threshold error correction model for the Rhine gulden and Basle Pound”, Swiss
Journal of Economics and Statistics, Vol. 147, No. 3, pp. 337-52
37
Kugler, P. 2013. “The changing pattern of arbitrage between the Amsterdam and London
foreign exchange market 1600-1912”, Mimeo.
Li, Ling-Fan. 2012. Bullion, bills and arbitrage: exchange markets in fourteenth- to
seventeenth-century Europe. PhD manuscript. Economic History Department, LSE.
Lopez, Robert Sabatino. 1976. The Commercial Revolution of the Middle Ages 950-1350,
Cambridge: Cambridge University Press.
Lothian, James R. 2002. “The internationalization of money and finance and the
globalization of financial markets”, Journal of International Money and Finance, Vol.
21, No. 6, pp. 699-724
Masini, Roberta. 2007 Gli investitori nei titoli di debito pubblico pontificio: categorie sociali,
distribuzione delle quote, motivazioni di una scelta (XVII secolo). In De Luca, G. and
Moioli, A. (eds.) Debito Pubblico e Mercati Finanziari in Italia. Secoli XIII-XX.
Rome: Franco Angeli, 197-214.
Malanima, Paolo. 2002. L’Economia Italiana. Dalla Crescita Medievale alla Crescita
Contemporanea. Bologna: Il Mulino.
Malanima, Paolo. 2009. Pre-modern European Economy. One Thousand Years (10th-19th
Centuries). Leiden and Boston: Brill.
Molho, Anthony. 1995. “The state and public finance: a hypothesis based on the history of
Late Medieval Florence”, Journal of Modern History, Vol. 67, Supplement, pp. 97135.
Munro, J. H. 2003. The Medieval origin of the financial revolution: Usury, rentes, and
negotiability. In International History Review, 25:3, 505-62.
Munro, John H. 2007. “The usury doctrine and urban public finances in Late-Medieval
Flanders (1220-1550): rentes, excise taxes, and income transfers from the poor to
the rich”, Paper presented to the Washington Area Economic History Seminar at the
University of Maryland Baltimore County.
Neal, Larry. 1985. “Integration of international capital markets: quantitative evidence from the
eighteenth to the twentieth centuries”, Journal of Economic History, Vol. 45, No. 2, pp.
219-26.
Neal, Larry. 1987. “The integration and efficiency of the London and Amsterdam stock
markets in the eighteenth century”, Journal of Economic History, Vol. 47, No. 1, pp. 97115
Neal, Larry. 1993. The Rise of Financial Capitalism. International Capital Markets in the Age
of Reason. Cambridge: Cambridge University Press.
Neal, Larry and Stephan Quinn. 2001. “Networks of information, markets, and institutions in
the rise of London as a financial centre, 1660-1720”, Financial History Review, Vol. 8,
No. 1, pp. 7-26
Newman, Mark E. J., and Michelle Girvan. 2004. “Finding and evaluating community
structure in networks”, Physical Review E, Vol. 69, pp. 1-15.
38
North, Douglass C.1981. Structure and Change in Economic History. London and New York:
W. W. Norton & Company.
North, Douglass C.1987. “Institutions, transaction costs and economic growth”, Economic
Enquiry, Vol. 25, No. 3, pp. 419-428.
North, Douglass C. and Barry, R. Weingast. 1989. “Constitutions and commitment: The
evolution of institutions governing public choice in seventeenth-century England”,
Journal of Economic History, Vol. 29, No. 4, pp. 803-32.
Ogilvie, Sheilagh C.1992. “Germany and the seventeenth-century crisis”, Historical Journal,
Vol. 35, No. 2, pp. 417-441.
Persson, Karl Gunnar. 1999. Grain Markets in Europe, 1500-1900: Integration and
Deregulation. Cambridge: Cambridge University Press.
Pezzolo, Luciano. 2005. “Bonds and government debt in Italian city-states, 1250-1650”, The
Origins of Value: The Financial Innovations that Created Modern Capital Markets
edited by William N. Goetzmann, N., and K. Geert Rouwenhorst, 145-64.
Oxford
and New York: Oxford University Press.
Pezzolo, Luciano and Giuseppe Tattara. 2006. ““Una fiera senza luogo”: was Bisenzone an
international capital market in sixteenth-century Italy?”, Journal of Economic History,
Vol. 68, No. 4, pp. 1098-1122.
Press, Volker. 1986. “Die Niederlande und das Reich in der Frühen Neuzeit”, Etat et Religion
aux XVe et XVIe Siècles: Actes du Colloque à Bruxelles du 9 au 12 Octobre 1984,
edited by Wim P. Blockmans and Herman van Nüffel. Bruxelles: Archives générales
du Royaume de Belgique, pp. 321-338.
Pugliese, Salvatore. 1924. “Condizioni economiche e finanziarie della Lombardia nella prima
metà del secolo XVIII”, Miscellanea di Storia Italiana, Terza Serie, Tomo XXI, 1-495.
Turin: Fratelli Bocca Librai di S. M.
Schiavo, Stefano, Javier Reyes, and Giorgio Fagiolo. 2010. “International trade and financial
integration: a weighted network analysis”, Quantitative Finance, Vol. 10, No. 4, pp.
389-99
Schubert, Eric S. 1988. “Innovations, debts, and bubbles: international integration of financial
markets in Western Europe, 1688-1720”, Journal of Economic History, pp. 299-306
Schweinberger, Michael, and Tom A. B. Snijders. 2003. “Settings in social networks: a
measurement model”, Sociological Methodology, Vol. 33, No. 1, pp. 307-41
Sieveking, Heinrich. 1905. “Studio sulle finanze genovesi nel Medioevo e in particolare
sulla Casa di S. Giorgio”. Translated by Onoro Soardi. Atti della Società Ligure di
Storia Patria, Vol. 35, No. 1.
Silva, J. M. C. Santos and Tenreyro, Silvana. 2006. “The log of gravity”, Review of
Economics and Statistics, Vol. 88, No. 4, pp. 641-658.
Silva, J. M. C. Santos and Tenreyro, Silvana. 2011. “Further simulation evidence on the
performance of the Poisson pseudo-likelyhood estimator”, Economics Letters, Vol.
112, No. 2, pp. 220-222.
39
Sperling, J. 1962. “The international payments mechanism in the seventeenth and
eighteenth centuries”, Economic History Review , Vol. 14, No. 3, pp. 446-468.
Spruyt, Hendrik. 1994. The Soverign State and its Competitors. Princeton, New Jersey:
Princeton University Press.
Stasavage, David. 2011. States of Credit: Size, Power, and the Development of European
Polities. Princeton: Princeton University Press.
Stasavage, David. 2014. “Was Weber right? The role of urban autonomy in Europe’s rise”,
American Political Science Review, Vol. 108, No. 2, pp. 337-354.
Stumpo, E. 2007. Città, stato e mercato finanziario: il diverso ruolo del debito pubblico in
Piemonte e in Toscana. In De Luca, G. and Moioli, A. (eds.) Debito Pubblico e
Mercati Finanziari in Italia. Secoli XIII-XX. Rome: Franco Angeli, 147-165.
Van Zanden, J. L. 2009. The Long Road to the Industrial Revolution. The European
Economy in a Global Perspective, 1000-1800. Leiden and Boston: Brill.
Volckart, Oliver. 1999. “Institutional competition: a new theoretical concept for economic
history”, Essays in Economic and Business History, pp. 75-87.
Volckart, Oliver. 2002. “Central Europe’s way to a market economy, 1000-1800”, European
Review of Economic History, Vol. null, No. 3, pp. 309-37.
Volckart, Oliver, and Nikolaus Woolf. 2006. “Estimating financial integration in the middle
ages: what can we learn from a TAR model?”, Journal of Economic History, Vol. 66,
No. 2, pp. 122-39.
Wang, Ning. 2003. “Measuring transaction costs. An incomplete survey”, Ronald Coase
Institute Working Paper Series, Number 2.
Whaley, Joachim. 2012a. Germany and the Holy Roman Empire. Vol. I. Oxford and New
York: Oxford University Press.
Whaley, Joachim. 2012b. Germany and the Holy Roman Empire. Vol. II. Oxford and New
York: Oxford University Press.
Zuijderduijn, C. Jaco. 2009. Medieval Capital Markets. Markets for Renten, State Formation
and Private Investment in Holland (1300-1550). Leiden and Boston: Brill.
40