Medieval European traditions in representation and state capacity today

See Stasavage (2016) and the references cited therein.

North et al. (2009) describe the ability of a government to centralize and consolidate the control of violence as one of the “doorstep conditions” leading towards what they call an “open access society”, where all citizens are respected equally and governed under the rule of law. If the powerful and centralized-yet-limited conundrum cannot be solved, then it is hard to imagine the centralization and consolidation of violence as one the relevant doorsteps.

Friedman et al. (2000) find no evidence that higher tax rates are associated with greater shadow economy size. Rich economies are the ones that generally have higher tax rates.

Murphy (2019) provides a “state economic modernity index” that aims to measure “a state’s willingness and ability to perform its core public goods functions, and the state’s size relative to the economy” (p. 73). He demonstrates that this index displays statistically significant correlations with both income levels and growth rates across countries; however, once country fixed effects are included in the regressions, those correlations are no longer significant.

Nunn (2009) provides an excellent overview of the early studies in this literature. La Porta et al. (2008) review the legal origins literature specifically.

By attempting to transfer their norms and values, members of an older generation impose costs and/or benefits on members of a younger one; likewise, members of a younger generation generate costs and/or benefits for members of an older one when they make efforts to adopt the latter’s norms and values.

These assemblies had antecedents in the tribal customs of Germanic barbarian groups that succeeded the Western Roman Empire (Downing 1989; Barnwell and Mostert 2003; van Zanden et al. 2012). See Young (2015) for a discussion of these early Germanic institutions circa 50 BC-50 AD and how they were influenced by Julius Caesar’s conquest of Gaul.

The weakest of European monarchs faced difficulties in convoking assemblies in the first place. Boucoyannis (2015) argues that only relatively strong European monarchs (e.g., those of England in the High Middle Ages) were able to compel members of the estates to attend their assemblies. As Stasavage (2016, p. 158) puts it: “[A] ruler’s authority must be sufficiently weak that he cannot simply extract what he wants without consent [but where] some degree of authority is necessary in order to get the people to show up at all.” This is also consistent with the fact that weaker (and particularly early medieval) monarchs resorted traveling incessantly to gain consent for their policies from the leading men of their realms (Marongiu 1968, p. 23; Bernhardt 1993; de Jouvenel 1993 [1947], p. 7; Ertman 1997, pp. 44, 236; Heather 2009, p. 529).

This is consistent with Barzel’s (2000) theory of rulers who guarantee property rights (and, in doing so, relinquish some power) to their subjects in exchange for the increased wealth (and thus extraction) that the property rights provide the subjects with incentives to create.

See Marongiu (1968, pp. 33–37) and Stasavage (2016, pp. 150–152) for a concise discussion of the development and application of different variants of the original maxim.

See Hanson and Sigman (2013) and Lindvall and Teorell (2017) for a recent reviews of this literature and expositions on the subject.

As with some of the other capacity indicators, we exclude the year 2016 despite its availability due to a lack of 2016 data in most of our controls.

Due to data limitations on other control variables Stasavage’s (2010) empirical analysis includes only 24 of these polities. For a full list of the secondary sources see the online appendix to Stasavage (2010): http://​politics.​as.​nyu.​edu/​docs/​IO/​5395/​onlineappendix.​pdf. An alternative data set on 32 European assemblies for a comparable time period is assembled by van Zanden et al. (2012).

Since the averages (ASSEMBLY_AVG; FREQUENCY_AVG) are taken over intervals (out of a possible 11) during which the polity is observed (existed), the final term in the product (PERIODS_IN_SAMPLE) adjusts the measure upward for polities that are observed during a larger portion of the overall 550 year period.

Note that we here speak of an initial assignment of assembly experiences to 15 present-day European countries. This is because those initial assignments are not ultimately the Assembly Experience values corresponding to these countries in the estimations. See below and, in particular, footnote 19.

Taking an unweighted average is arbitrary; in particular cases (e.g., historical France and Burgundy each taken equally in relation to present-day France) it may seem questionable. However, what would be a weighted average alternative that can be applied generally is unclear. We have therefore decided to take the most straightforward approach.

While these are the best estimates available and have been used in the economic development literature (e.g., Comin et al. 2010) there are clearly conceptual and practical issues that go along with them. The main appendix to Putterman and Weil (2010) does a good job of discussing these issues: http://​www.​brown.​edu/​Departments/​Economics/​Faculty/​Louis_​Putterman/​Appendix%20​to%20​1500%20​Origins%20​Matrix%20​1.​1.​doc. A recent paper by Easterly and Levine (2016) assembles data on European population shares during the time of colonization, which is distinct from the Putterman and Weil descendancy shares (where, e.g., someone who moves from the UK to the US today can still potentially count towards the UK descendancy share in the US). Easterly and Levine show that colonization shares are important in accounting for income levels today even when controlling for descendancy shares. However, our use of the descendancy shares is appropriate because we hypothesize that long-institutional memory of assembly activity is transmitted through the generations, such that descendants from circa 1500 populations bring it with them to their destinations regardless of whether they reach those destinations in 1500 or 1900.

This is the case for all of the countries in our sample including the 15 present-day European countries that matched to the Stasavage polities. (For example, then the Assembly Experience of France that enters our estimations is not the assembly experience of medieval/early modern France. To the extent that there are descendants of individuals from the other 14 European countries circa 1500 in present-day France, the historical assembly of those 14 countries also factor in.) Therefore, we construct Assembly Experience consistently across all countries in our sample, European or otherwise. Doing so is appropriate given our hypothesis of long institutional memory that is transmitted across through both time and space.

In “Appendix 1” we provide detailed examples of how assembly experiences are calculated for a present-day country that matches to historical Stasvage polities (France) and a present-day country that does not (Canada).

The seven categories include East Asia and Pacific, Europe and Central Asia, Latin America and the Caribbean, Middle East and North Africa, North America, South Asia, and Sub-Saharan Africa. See https://​datahelpdesk.​worldbank.​org/​knowledgebase/​articles/​906519-world-bank-country-and-lending-groups for details.

See Gwartney et al. (2016, pp. 275–277) for details on the methodology employed in combining data from these sources into a country’s score.

This is the measure that Dincecco et al. (2019) primarily focus their comparative study of whether “war makes states” holds historically in Africa vis-à-vis Asia and Europe.

http://​www.​ggdc.​net/​maddison/​Historical_​Statistics/​vertical-file_​02-2010.​xls (last accessed July 21, 2017).

Recent studies have empirically linked aspects of culture to economic development outcomes (e.g., Guiso et al. 2009; Williamson 2009; Tabellini 2010; Gorodnichenko and Roland 2011, 2012, 2017; Davis 2016) and measures of institutional quality (e.g., Licht et al. 2007; Tabellini 2008; Williamson and Kerekes 2011; Klasing 2013; Alesina and Giuliano 2015; Davis and Abdurazokzoda 2016; Tarabar 2017).

Our region classification comes from World Bank Classifications (https://​datahelpdesk.​worldbank.​org/​knowledgebase/​articles/​906519-world-bank-country-and-lending-groups). Regions include: (1) East Asia and Pacific, (2) Europe and Central Asia, (3) Latin America and the Caribbean, (4) Middle East and North Africa, (5) North America, (6) South Asia, and (7) Sub-Saharan Africa.

When considering the effects on military resources and the consolidation of violence, one concern is the North American and European countries that have a common and credible defense policy via NATO. We re-ran the Table 7 estimations excluding NATO countries; the results are not meaningfully different. (Those results are available upon request.)

The only statistically significant result for legal system and property rights reported in Table 13 is for the monopoly on force measure (5% level). The point estimate is somewhat smaller than the benchmark (0.630 vs 0.903).

See Kerekes and Williamson (2012) for a discussion of Icelandic institutions during the 930–1262 period.

Also recall that about a third of the countries in our analysis have zero assembly experience values. For these countries, going to the average assembly experience of the Stasavage countries (3.463) would be associated with changes in the shadow economy, monopoly on force, and military personnel measures of − 0.042, 1.277, and − 2.493 respectively (35%, 62%, and 32% of the relevant standard deviations).

Brambor et al. (2006) emphasize that coefficients in models with an interaction term cannot be interpreted separately and straightforwardly. In particular, the conditional standard error for Assembly Experience is here [Var(β₁) + Var(β₃) × (historical conflict 1400–1799)² + 2 × Cov(β₁,β₃) × (historical conflict 1400–1799)]^1/2, where β₁ is the coefficient on Assembly Experience and β₃ is the coefficient on its interaction with Historical Conflict 1400–1799.

E.g., referencing Table 3, an English descendancy share is given positive weight; while a German descendancy share receives 0 weight.