Wars, Taxation, and Increases in Representation

Yves here. VoxEU regularly posts on studies that mine historical data, and here I mean of long ago history, either to draw lessons for today or to show that seemingly “so-old-how-can-it-still-matter” events still cast a shadow. This analysis is of a shift towards more representation in the medieval era, as in what is commonly thought of as rule by nobles. It finds that increases in wars lead to increases in representation on local councils, which in a longer time frame, meant an acceleration of the trend of medieval constitutionalism.

So in this time of generally dispiriting news, there may be a small silver lining. The ferocious drive towards the rearmament of Europe may require a big reduction in its democratic deficit to get done….or won’t get done much at all due to the need for at least some level of voter buy-in.

By Sascha O. Becker, Xiaokai Yang Chair of Business and Economics Monash University and Full-time Professor of Economics University Of Warwick; Andreas Ferrara, Assistant Professor of Economics University of Pittsburgh; Eric Melander Assistant Professor of Economics University Of Birmingham; and Luigi Pascali, Associate Professor Associate Professor Pompeu Fabra University and Kenen Fellow in the Department of Economics Princeton University. Originally published at VoxEU

In the grand sweep of history over the last millennium, political divergence between Europe and contemporaneous world civilisations preceded economic divergence. This column describes how the development of the institutional frameworks of medieval constitutionalism – characterised by representative assemblies, rule of law, and substantial fiscal and spending capacity – in European towns and cities paved the way for Western modern states. In a current international climate where democratic institutions increasingly appear to be on the back foot, history offers a sobering perspective on the hard-fought moves towards greater representation.

Europe in the Middle Ages was a place marred by near-ceaseless conflict between myriad weak rulers, particularly in a strip of land ranging from the Flemish and Dutch coast, via the German lands, to Northern Italy. Figure 1, which plots the territorial boundaries of European polities between 1290 and 1710, demonstrates the sheer density of borders in this region of Central Europe. Yet it was in this same region that the institutional package of medieval constitutionalism emerged, in towns and cities that enforced the rule of law, granted its citizens political representation, and developed increasingly sophisticated fiscal machineries.

Figure 1 De facto territorial boundaries in Europe, the Middle East and North Africa, 1290-1710

Note:  De facto territorial boundaries in Europe, the Middle East and North Africa, 1290–1710. De facto territorial boundaries from the Centennia Historical Atlas (Clockwork Mapping 2018) overlapped at decadal intervals for the period 1290–1710 (light grey). Modern state boundaries also shown (black).

It has long been hypothesised that this “vortex of near-permanent war … forced [rulers] to strike deals that meant sharing political power” (Hoffman and Norberg 1994). But so far it has been difficult to empirically tease out the direction of causation. In the words of Stasavage (2016): “the evidence suggests some causal link between warfare and representative institutions, although we do not know in which direction causality runs.” Our recent paper (Becker et al. 2025) is the first in this context to present systematic evidence in favour of one direction of causality, from wars to representation.

New Data and Novel Evidence on War and Representation

Our work is underpinned by a massive effort to mobilise detailed data on German towns and cities for over five centuries that span the period of medieval constitutionalism. Drawing on the Deutsches Städtebuch (Keyser 1939-1974), a multi-volume encyclopaedia of German cities, we create a panel dataset documenting the conflicts in which cities were involved, characteristics of their councils, and the sophistication of their tax systems.

With these data in hand, we first document that exposure to conflicts appears to be related to the expansion of city councils (with more members indicating representation of wider strata of the population). Following conflict, these councils are also increasingly likely to be elected by citizens without the interference of the local lord. Figure 2 shows this result descriptively using the first conflict to which territories in our sample are exposed and tracking the evolution of council elections and sizes before and after the ‘event’.

Figure 2 Territory-level event study

Note: Evolution of political outcomes for territories before and after the first conflict to which they are exposed. The probability of citizens electing their local city  council equals 1 if citizens can hold such elections without the interference of the local ruler. Council size is the number of members on the city council.

Succession Disputes as a Source of Random Variation in Conflict

To isolate variation in conflict exposure that mimics a randomised experiment, we introduce a second dataset to construct an instrumental variable for conflict. Using data on European noble families compiled by the ambitious Peerage project, we reconstruct the aristocratic network at each point in time and identify central individuals within this network. These nobles were key to war and peace in this period, and we home in on one aspect in particular: succession.

In the German context, a large number of conflicts were succession wars, fought over women’s right to inherit as rulers. By identifying the gender of important nobles’ firstborn children, we construct our instrument for conflict. The logic is the following: upon having their first child, nobles enter into a lottery in which either realisation (a male or a female child) is equally likely. But firstborn girls open up the possibility of challenges to succession and, hence, succession wars. The lottery thereby randomises some observations in our dataset into higher conflict proneness than others. It is this exogenous variation in conflict that we then exploit to identify the causal effect of conflict on political and fiscal institutions.

The War of the Succession of Landshut (1503-1505) illustrates how succession wars led to the rise and spread of medieval constitutionalism. Duke George of Bavaria-Landshut did not have a male heir, and named his firstborn daughter, Elisabeth, as his successor. This breached an agreement with the House of Wittelsbach, according to which Duke Albert of Bavaria-Munich would have been the legal successor. Albert refused to concede, and a destructive war ensued.

In Figure 3, we plot the evolution of political and fiscal institutions around the outbreak of the war of succession, comparing outcomes for Bavaria-Munich to those for the rest of Bavaria and other German regions. Cities in Bavaria-Munich saw an immediate increase in the probability that citizens elected their city councils and in their size. A few decades after the conflict, these cities had developed increasingly sophisticated tax systems, consistent with temporary war taxes evolving into more far-reaching fiscal capacity, based on income and wealth taxes.

Figure 3 Territory case study: Bavaria-Munich and the Landshut Succession War

Note: Evolution of political and taxation outcomes for cities in the territory of Bavaria-Munich, other cities in Bavaria and other cities in the German lands in the century before and after the Landshut Succession War (indicated with vertical dashed lines). The probability of citizens electing their local city council equals 1 if citizens can hold such elections without the interference of the local ruler. Council size is the number of members on the city council. Complex taxes refer to those taxes that require a register or administrative infrastructure to levy them because the taxed items are not easily observed, which includes income, wealth, or business taxes.

We then go on to confirm our findings from this single example in our full sample of cities and time periods. On average, after a war triggered by the birth of a female firstborn, cities see a 2-3 percentage point increase in the probability that their councils are elected by citizens and an expansion of the council by approximately two members. This is followed in subsequent decades by a shift towards more sophisticated forms of taxation.

Long-Term Implications of Medieval Constitutionalism

In the final section of the paper, we investigate the long shadow that the institutional package of representation and high fiscal capacity – developed in the context of medieval constitutionalism – had on the subsequent rise of modern states. Previous work highlights the importance of high tax capacity in state formation (Tilly 1975, Gennaioli and Voth 2015) and a recent paper by Cantoni et al. (2024) documents empirically the importance of centralised territorial fiscal systems in the early modern era. Our exercise here complements these findings by highlighting the parallel importance of city-level tax systems until at least 1700.

We show that the cities and territories in our sample that developed high fiscal capacity in the context of medieval constitutionalism were more likely to come out on top in the process of territorial consolidation from the 17th century. Concretely, we look at instances where cities switch hands from one territory to another, and calculate the difference in tax capacity between the gaining and losing territory. From the year 1600 onwards, there is a pronounced increase in the tax advantage of conquerors over the conquered.

Conclusions

All told, our findings underscore the role played by wars in the emergence of crucial elements of the blueprints for modern states. In particular, we re-emphasise – now equipped with systematic causal evidence – conclusions from previous qualitative and historical literatures that the development of representative institutions was not a foregone conclusion, but rather a silver lining of the bargain that conflict forced rulers to strike with their subjects.

See original post for references