ArticlePDF Available

State Capacity and Long‐Run Economic Performance

October 2012
The Economic Journal 126(590)

DOI:10.2139/ssrn.2044578

Authors:

Mark Dincecco

University of Michigan

Gabriel Katz

University of Exeter

We present new evidence about the long-run relationship between state capacity – the fiscal and administrative power of states – and economic performance. Our database is novel and spans 11 European countries and 4 centuries from the Old Regime to World War I. We argue that national governments undertook two political transformations over this period: fiscal centralisation and limited government. We find a significant direct relationship between fiscal centralisation and economic growth. Furthermore, we find that an increase in the state’s capacity to extract greater tax revenues was one mechanism through which both political transformations improved economic performance. Our analysis shows systematic evidence that state capacity is an important determinant of long-run economic growth.

Conditional Correlations Between Real Per Capita GDP Growth and ‘Future’ Political TransformationsNotes. Dots represent point estimates for α1,0, the coefficient for Ci,t in (4). Circles are point estimates for α1,τ, τ = 1, … ,q. Solid lines represent 90% confidence intervals.

…

Figures - uploaded by Gabriel Katz

Content may be subject to copyright.

Content uploaded by Gabriel Katz

Content may be subject to copyright.

State Capacity and Long-Run Economic Performance∗

(Short title: State Capacity and Economic Performance)

Mark Dincecco Gabriel Katz

We present new evidence about the long-run relationship between state capacity – the

ﬁscal and administrative power of states – and economic performance. Our database is novel

and spans 11 European countries and 4 centuries from the Old Regime to World War I. We

argue that national governments undertook two political transformations over this period:

ﬁscal centralisation and limited government. We ﬁnd a signiﬁcant direct relationship between

ﬁscal centralisation and economic growth. Furthermore, we ﬁnd that an increase in the state’s

capacity to extract greater tax revenues was one mechanism through which both political

transformations improved economic performance. Our analysis shows systematic evidence that

state capacity is an important determinant of long-run economic growth.

Standard economic theory assumes that states are eﬀective. Namely, states have

enough administrative infrastructure to provide secure property rights, basic market

regulations, and dispute resolution through courts. Yet eﬀective states are only a

∗Corresponding author: Mark Dincecco, Department of Political Science, University of Michigan, 5700

Haven Hall, 505 S. State Street, Ann Arbor, MI 48109, USA. Email: dincecco@umich.edu.

We thank Editor Stephen Machin, two anonymous referees, Elhanan Helpman, and seminar participants at

the 2013 APSA Meeting, Berkeley, Bocconi, Caltech, Copenhagen, the 2012 EHA Meeting, the Juan March

Institute, the LSE, Michigan, the 2012 Nemmers Prize Conference, Stanford, Toulouse, Utrecht, and Yale for

valuable comments, and Rui Esteves, Giovanni Federico, Klas Fregert, Michael Pammer, and Mark Spoerer

for data help. Mark Dincecco also thanks Barry Eichengreen for his hospitality during an extended visit to

the Berkeley Economic History Laboratory, where part of this paper was written. He gratefully acknowledges

ﬁnancial support from National Science Foundation Grant SES-1227237.

(pre-published version)

recent historical development and represent just a fraction of modern nations.1Today’s

developing nations, like their historical predecessors, often confront problems of small

administrative infrastructure. Eﬀective states cannot be taken for granted.

While there is a large econometric literature on the economic eﬀects of democracy,

the corresponding literature on state capacity is small.2This paper tests the long-

run relationship between state capacity and economic performance in Europe, the

birthplace of modern economic growth. To the best of our knowledge, it is among the

ﬁrst papers to systematically test this relationship.3

Medieval European nation-states can be thought of as “mosaic states” (Strayer,

1970, p.53) that were not only built upon a patchwork of well-rooted local institutions,

but competed with them for ﬁscal dominance. This lack of extractive capacity made

it impossible for national governments to provide the administrative infrastructure

necessary to facilitate economic activity. We argue that two political transformations

resolved the long-standing state capacity problems faced by emergent nation-states in

this context. The ﬁrst political transformation was the implementation of uniform tax

systems at the national level, or “ﬁscal centralisation”. This transformation typically

took place from 1789 onward. The second political transformation was the establish-

ment of national parliaments with the ability to monitor state expenditures at regular

intervals, or “limited government”. This transformation typically took place decades

after ﬁscal centralisation over the 1800s. We argue that both ﬁscal centralisation and

1For historical state formation, see Hintze (1906), Mathias and O’Brien (1976), Levi (1988), Brewer

(1989), Tilly (1990), Hoﬀman and Rosenthal (1997), Epstein (2000), O’Brien (2011), Dincecco (2009, 2011),

Karaman and Pamuk (2010), Rosenthal and Wong (2011), and Gennaioli and Voth (2012). For state capacity

problems in sub-Saharan Africa, see Migdal (1988), Herbst (2000), Bates (2001), and Acemoglu and Robinson

(2012). By contrast, states have played important development roles in Asian Tiger nations. See Wade

(1990), Kang (2002), and Rodrik (2005).

2Papaioannou and Siourounis (2008) provides an overview of the democracy and growth literature;

also see Acemoglu et al. (2014). A recent literature expands standard economic theory to incorporate state

capacity. See Acemoglu et al. (2004), Acemoglu (2005), Besley and Persson (2009, 2011, 2013), and Acemoglu

et al. (2011).

3Bockstette et al. (2002) ﬁnds a signiﬁcant positive link between state antiquity and current economic

development. Besley and Persson (2009, 2011) show signiﬁcant positive correlations between past wars and

current ﬁscal and economic outcomes. Dincecco and Prado (2012) ﬁnd a signiﬁcant positive relationship

between current ﬁscal capacity and economic performance. They use historical war casualties to instrument

for current ﬁscal institutions. Michalopoulos and Papaioannou (2013) show a signiﬁcant positive impact of

pre-colonial political centralisation on current economic development in Africa.

limited government increased the national government’s capacity to extract greater tax

revenues: a country like France, for example, could extract nine times more revenues

per capita after the two political transformations than before them. Furthermore, we

argue that greater state capacity had positive economic implications through several

potential channels, including the creation of administrative infrastructure. Section 1

describes our hypotheses at length.

We evaluate our argument in two steps. We ﬁrst test the direct relationship be-

tween political transformations and economic performance. Our econometric analysis

uses a novel annual database that spans eleven European countries from the height

of the Old Regime in 1650 to the eve of World War I in 1913. We model economic

growth as a function of political transformations, country ﬁxed eﬀects that account

for time-invariant country characteristics, year ﬁxed eﬀects that account for common

time shocks, time-varying controls, and country-speciﬁc time trends that account for

unobserved time-varying country-level heterogeneity.

The results show a signiﬁcant direct role for ﬁscal centralisation in long-run eco-

nomic performance. We ﬁnd that ﬁscally centralised regimes saw an average annual

growth rate in real per capita GDP that was 0.17% to 0.43% higher than ﬁscally

fragmented regimes. To put these magnitudes into perspective, the average annual per

capita GDP growth rate among sample countries over the 1650-1913 period was 0.62%.

Our estimates thus suggest that the growth improvements associated with ﬁscal cen-

tralisation were equivalent to 27% to 69% of the average per capita GDP growth rate

over this period, and 8% to 20% of its standard deviation. For another perspective,

Papaioannou and Siourounis (2008) ﬁnd that post-1960 democratic transitions were

associated with a 1% increase in annual real per capita GDP growth during a period

for which the world average annual growth rate was roughly 1.8% (Barro and Sala-i-

Martin, 2004, p.4). This comparison suggests that ﬁscal centralisation was of roughly

the same economic importance historically as some scholars argue that democratisation

is today. Furthermore, we show evidence that the relationship between ﬁscal centrali-

sation and economic growth was not just transitory but long-lasting: the long-run per

capita GDP growth rate for ﬁscally centralised regimes was 0.16 to 0.33 percentage

points higher than for ﬁscally fragmented regimes. While we do not ﬁnd evidence for a

signiﬁcant direct relationship between limited government and economic performance,

the results suggest large interaction eﬀects between the two political transformations.

There are several potential mechanisms through which political transformations

could have improved economic performance. The second part of our econometric anal-

ysis tests one speciﬁc mechanism, greater state capacity, using the same methodology

as before to isolate the within-country correlation between political transformations

and ﬁscal outcomes. The results show that ﬁscally centralised and, to some extent,

limited government regimes extracted per capita revenues at signiﬁcantly higher rates

than ﬁscally fragmented or absolutist regimes, respectively. Furthermore, there is some

evidence for a signiﬁcant relationship between limited government and improvements

in the state’s capacity to productively spend government funds. Finally, we ﬁnd sig-

niﬁcant links between extractive capacity and subsequent economic performance. Our

estimates suggest, for example, that extractive capacity improvements account for

42% to 58% of the diﬀerence in average annual per capita GDP growth rates between

eighteenth-century England and France.

To conclude our analysis, we perform placebo tests that recode political transforma-

tions as if they had taken place decades prior to the actual years. The results provide

additional evidence that reverse causation did not drive the relationships between po-

litical transformations and the economic and ﬁscal outcomes that we ﬁnd. We discuss

potential threats to inference and how we address them at length in Section 3.

Overall, the econometric analysis supports our argument that political transfor-

mations improved economic performance. The direct relationship between ﬁscal cen-

tralisation and economic growth is signiﬁcant, large, and robust. Furthermore, we

ﬁnd evidence that greater state capacity was one mechanism through which both ﬁs-

cal centralisation and, to some extent, limited government played signiﬁcant economic

roles. Our results also highlight the role of non-ﬁscal mechanisms through which ﬁscal

centralisation worked.

Our paper relates to the literature that tests the long-run relationships between

historical institutional factors and economic performance, including Engerman and

Sokoloﬀ (1997), La Porta et al. (1998), Hall and Jones (1999), Acemoglu et al. (2001,

2002, 2005), and Banerjee and Iyer (2005). However, this literature typically highlights

the role of state predation rather than state capacity. Furthermore, it does not often

test speciﬁc mechanisms through which institutional factors persist (see Nunn, 2009;

Dell, 2010, is a recent exception).

Our paper also relates to the literature that views the state as an active participant

in the development of modern capitalist systems, including Gerschenkron (1966), Mag-

nusson (2009), and O’Brien (2011). We provide an econometric counterpart to these

works. Our paper thus contributes to the debate about the institutional origins of the

Industrial Revolution (Mokyr, 2008).

The rest of the paper proceeds as follows. Section 1 outlines the historical back-

ground and develops the ﬁscal and economic implications. Section 2 describes the

data. Section 3 presents the econometric methodology. Section 4 reports the estima-

tion results for the direct relationship between political transformations and economic

performance. Section 5 reports the estimation results for the state capacity mechanism.

Section 6 performs the placebo tests. Section 7 concludes.

1. Historical Overview

This section characterizes two political transformations that we argue resolved key state

capacity problems in European history. Our historical account follows Dincecco (2009,

2011), which also provide sources. We ﬁrst describe the political transformations, and

then discuss potential mechanisms through which greater state capacity could improve

economic performance.

1.1. Political Transformations

1.1.1. Fiscal centralisation

Most European states were ﬁscally fragmented before the nineteenth century. Con-

trary to conventional wisdom, early modern monarchs faced a host of incumbent local

institutions that reduced their extractive powers. Epstein writes that

decades of research on pre-modern political practices. . . has shown how “ab-

solutism” was a largely propagandistic device devoid of much practical sub-

stance. . . The strength of a monarch’s theoretical claims to absolute rule was

frequently inversely proportional to his de facto powers.

(Epstein, 2000, pp.13-14)

One general feature of fragmented states was the close relationship between local tax

control and political autonomy. Local elites had strong incentives to oppose national-

level ﬁscal reforms that threatened their traditional tax rights. The result was a classic

public goods problem. Since each local authority attempted to free-ride on the tax

contributions of other locales, the tax revenues that national governments could extract

per capita were low.

To resolve the problem of local tax free-riding and establish greater extractive

capacity, national governments had to gain the ﬁscal authority to impose standard tax

menus, rather than bargain place by place over individual tax rates. So long as states

equalized tax rates across locales at relatively high levels, extractive capacity rose.

1.1.2. Limited government

Although absolutist monarchs in early modern Europe spent government funds as they

wished, elites in national parliaments exercised tax authority. Hoﬀman and Rosenthal

(1997) argue that the main goal of absolutist monarchs was to ﬁght wars, both for

personal glory and for homeland defence. A key reason was the problem of royal moral

hazard in warfare (Cox, 2011). In Hoﬀman’s words, absolutist monarchs

likely provided far more defense than their average subject would have

wanted. They went on the oﬀensive too, and not just to protect their king-

doms. The reasons were not hard to understand. Victory was. . . a source of

glory or a way to enhance their reputation... . they faced no major downside

risk to their thrones, at least in the larger states, for loss in battle. . .

(Hoﬀman, 2012, p.604)

Since parliamentary elites feared that absolutist monarchs would spend additional

revenues in wasteful ways like foreign military adventures, they demanded the power of

budgetary oversight before raising new taxes. To evade parliament, absolutist monarchs

resorted to ﬁscal predation, which reinforced the fear that they could not be trusted.

Parliamentary elites thus resisted tax requests and government revenues per capita

were low.

Regular control over state budgets established the ﬁscal supremacy of parliament.

The surrender of budgetary control was a credible way to resolve the royal moral hazard

problem in warfare. In turn, extractive capacity rose.4

1.2. Economic Implications

Our historical account suggests that both ﬁscal centralisation and limited government

increased extractive capacity. We now discuss potential channels through which greater

state capacity could have improved economic performance.

Besley and Persson (2013) argue that the state’s extractive capacity is central to

economic development.5They show strong correlations between ﬁscal capacity in-

vestments in administrative infrastructure, high tax levels, and economic prosperity.

4We base this argument on North and Weingast (1989). Scholars argue that factors beyond de jure

parliamentary change, including political coalitions (Stasavage, 2003), de facto institutional reforms (Pincus

and Robinson, 2010), and ministerial responsibility (Cox, 2011), were important to subsequent ﬁscal and

economic outcomes in European history.

5Their argument follows a long tradition that includes Schumpeter (1918) and Kaldor (1963).

Following this lead, we focus on the creation of administrative infrastructure or “in-

frastructural power” (Mann, 1986) as a potential channel linking greater state capacity

with improved economic performance. States can facilitate economic activity in sev-

eral ways, including the provision of secure property rights, basic market regulations,

and dispute resolution through courts. To erect the administrative infrastructure that

facilitates this activity, states require suﬃcient revenues.6For example, Brewer (1989)

relates England’s historical military and economic rise to the establishment of lim-

ited government and greater extractive capacity following the Glorious Revolution of

1688 and the subsequent growth in administrative infrastructure. By contrast, Herbst

(2000) links low revenues in Africa with weak administrative infrastructures and the

lack of basic public services (e.g., security, courts) that facilitate economic activity.

While we emphasize the state’s extractive role, a potential complementary chan-

nel is the state’s capacity to play a productive economic role through the provision of

growth-enhancing public services like education or physical infrastructure (Besley and

Persson, 2013). By resolving the problem of royal moral hazard in warfare as described

in the previous subsection, the establishment of limited government could have led to

greater productive capacity. There is also reason to think that the centralised pro-

vision of particular public services was more growth-enhancing than the decentralised

provision of similar services. For example, historical central government investments in

mass primary education promoted common national languages and cultural identities

(Lindert, 2004; Aghion et al., 2012), which could have facilitated trade and innovation.

Political transformations could have also improved economic performance through

non-ﬁscal mechanisms. Institutional fragmentation in early modern Europe imposed

costs, delays, and risks that atomized domestic economies and reduced economic growth

(Epstein, 2000). Although taxes were low overall, ﬁscal fragmentation led to high tax

rates in sectors under royal control (Hoﬀman and Rosenthal, 1997). To prevent re-

6Prescott (e.g., 2004) argues that higher taxation accounts for the worker productivity shortfall in

Europe relative to the United States. Our analysis does not ﬁnd any evidence of a negative relationship

between greater state capacity and economic performance. Still, there may be reason to think that diﬀerent

tax compositions (e.g., income vs. consumption-type taxes) may aﬀect economic performance when state

capacity is already high (i.e., at OECD-country levels).

source diversion into tax-exempt sectors, absolutist monarchs enacted rigid laws. By

resolving this problem, ﬁscal centralisation increased investment mobility. Similarly,

there are several potential mechanisms linking limited government with improved eco-

nomic performance (Papaioannou and Siourounis, 2008). For example, representative

rule can facilitate sound economic policy through eﬃcient information gathering and

transmission (Sen, 1999). Our empirical analysis accounts for the economic role of

non-ﬁscal mechanisms.

In summary, we argue that political transformations improved economic perfor-

mance through greater state capacity. The rest of the paper tests this argument. We

ﬁrst examine the direct relationship between political transformations and economic

performance. We then examine the role of state capacity, by testing the links from

political transformations to state capacity, and from state capacity to economic per-

formance.

2. Data

2.1. Political Transformations

We deﬁne and code political transformations according to Dincecco (2009, 2011). The

process of ﬁscal centralisation was completed the year that the national government

ﬁrst secured its revenues through a standard tax system with uniform rates throughout

the country. Limited government was established the year that parliament gained the

stable constitutional right to control the national budget on an annual basis. To ensure

stability, parliament’s power of the purse had to hold for at least two consecutive

decades.7

7There is a close correspondence between this coding scheme and the schemes of De Long and Shleifer

(1993), Acemoglu et al. (2005), and the Polity IV database of Jaggers and Marshall (2008). De Long and

Shleifer (1993) use three measures: a binary indicator of absolutist versus non-absolutist regimes, an eight-

point constitutional scale, and the categories of capital versus coercion from Tilly (1990). However, they

code political regimes at 150-year intervals. Acemoglu et al. (2005) use two measures: categories of executive

constraints and protection for capital, both from Jaggers and Marshall (2008). However, they code political

regimes at 50- or 100-year intervals. While Jaggers and Marshall (2008) code executive constraints at yearly

intervals, their data do not start until the 1800s.

Table 1 displays the dates for ﬁscal centralisation across sample countries. The

Norman Conquest of 1066 undercut provincial authority in England and established

a precocious uniformity of laws and customs.8Structural ﬁscal change took place

swiftly and permanently in several parts of continental Europe after the fall of the

Old Regime. The National Assembly transformed the French tax system during the

Revolution (1789-99), a process completed by Napoleon upon taking power in 1799.

The First French Republic conquered the Low Countries in 1795, and the Southern

Netherlands including Belgium became French departments. The Batavian Republic,

the successor to the Dutch Republic, established a national tax system under French

rule in 1806. French conquest at the start of the 1800s was also the major catalyst

for ﬁscal change on the Italian peninsula; France annexed Piedmont in 1802. Prussia

undertook ﬁscal centralisation after a battle loss to France in 1806. Napoleon defeated

Austria in 1805 and invaded Portugal in 1807 and Spain in 1808, but only implemented

incomplete administrative reforms. Fiscal centralisation did not take place in Austria

and Spain until the 1840s and in Portugal until the 1850s.9Traditional ﬁscal structures

remained in Scandinavia well through the 1800s. Fiscal centralisation took place in

Sweden in 1861 and in Denmark in 1903.

Table 2 displays the dates for limited government, which was typically established

decades after ﬁscal centralisation over the 1800s.10 Belgium was established as a consti-

tutional monarchy after declaring independence from the Netherlands in 1830. In the

Netherlands, the new constitution of 1848 called for the executive to submit annual

budgets to parliament.11 Kings Charles Albert of Piedmont and Frederick William

8England conjoined with Wales in 1536. The Act of Union of 1707 conjoined Scotland. A similar Act

conjoined Ireland in 1800 (the Irish Free State was established in 1922). For consistency, we use the term

“England” rather than “Great Britain” or the “United Kingdom” throughout the text.

9Austria and Hungary were the largest territories of the Austrian Empire (1804-67). The Compromise

of 1867 led to the establishment of the Austro-Hungarian Empire (1867-1918). For consistency, we use the

term “Austria” throughout the text.

10 Elster (2000, ch.2) argues that the establishment of many modern constitutional governments was non-

incremental and took place in moments of crisis. Also see Russell (2004, p.106). We thank Barry Weingast

for alerting us to these works.

11 The constitution under King William I (1815-40) gave parliament the right to audit state ﬁnances, but

only at 10-year intervals (van Zanden and van Riel, 2010).

IV of Prussia also granted liberal constitutions in 1848.12 The Compromise of 1867

marked the start of the constitutional era in Austria. Limited government was estab-

lished in France after the capture of Emperor Napoleon III during war with Prussia.13

A stable parliamentary regime was established in Spain in 1876. Limited government

and ﬁscal centralisation took place within a decade of each other in Portugal and

Sweden. A stable constitutional regime was established in Portugal in 1851, while the

1866 parliamentary reform in Sweden established a modern bicameral legislature.14 Fi-

nally, limited government was established well before ﬁscal centralisation in the Dutch

Republic (1572-1795) and in Denmark. The Dutch Republic is typically classiﬁed as

constitutional (De Long and Shleifer, 1993; Acemoglu et al., 2005; Stasavage, 2005),

while King Frederick VII of Denmark established a two-chambered parliament after

the 1848 revolutions.

One concern is the possibility of measurement error induced by the coding scheme

for political transformations. The scheme codes ﬁscally fragmented regimes as wholly

fragmented, even for states where ﬁscal divisions were small. Thus, some regimes coded

as fragmented will include data associated with greater extractive capacity, reducing

average capacity improvements after ﬁscal centralisation. Similarly, the scheme selects

early dates for limited government. Because extractive capacity typically increased

over time, some regimes that were coded as limited will include data associated with

lower extractive capacity, reducing average capacity improvements. The systematic

underestimation of the state capacity impacts of ﬁscal centralisation and limited gov-

ernment should bias the data against our hypotheses that political transformations

improved economic performance by enabling states to eﬀectively fulﬁll their extractive

role.

More generally, the various ways in which early modern states tabulated annual

12 Tilly (1966) and Ziblatt (2006, pp.113-16) alike code the post-1848 regime in Prussia as constitutional,

although Tilly notes that the executive acted without legislative approval of military budgets during the

1860s.

13 The July Monarchy of King Louis Philip (1830-48) was not coded as limited because it endured for less

than two decades.

14 Sweden enacted a constitution in 1809, but the executive retained absolute veto authority and parliament

only met once every ﬁve years.

revenues suggests that, on average, they overestimated the amounts of available re-

sources.15 Pre-transformation regimes will thus appear to have greater extractive

capacity. State accounting practices improved over time, reducing the number and

magnitude of overestimates. These two features also bias the data against our hypothe-

ses about the relationships between political transformations, extractive capacity, and

economic performance.

2.2. Economic Performance

Our economic performance measure is the (logarithmic) annual growth rate of real

per capita GDP from Maddison (2010).16 These data are available for 1600, 1700,

and annually from 1820 to 1913. We linearly interpolate the pre-1820 data to provide

annual observations from 1650 onward for all years for which state capacity data are

also available.

Figure 1 displays the time-demeaned average per capita GDP growth rates around

political transformations. The top panel shows average per capita GDP growth rates

for ﬁfty years before and after ﬁscal centralisation (left panel) and limited government

(right panel). Both political transformations were associated with economic improve-

ments. The average per capita GDP growth rate for ﬁscally centralised regimes was

around 0.90%, but only around 0.20% for ﬁscally fragmented regimes. Similarly, the av-

erage per capita GDP growth rate for limited government regimes (also around 0.90%)

was high relative to absolutist regimes (around 0.35%).

The bottom panel zooms in on average per capita GDP growth rates for ten years

before and after political transformations. We observe a sharp, sustained jump in per

capita GDP growth rates in the decade after ﬁscal centralisation.17 While there was

15 Bonney (1995, pp.423-506) and O’Brien (2011, pp.408-20) discuss the limitations of historical budgetary

data.

16 GDP data from Barro and Urs´ua (2010), a potential alternative, are not widely available before the

1850s. However, the post-1850 trends are similar to the Maddison data.

17 Structural break tests for real per capita GDP levels using Bai and Perron (2003)’s algorithm indicate

that the 95% conﬁdence intervals for the structural breaks encompassed the dates of political transformations

for the majority of countries in the sample.

a smaller jump associated with limited government, it was not sustained. The zoom

view also suggests that per capita GDP growth slowed in the years just before political

transformations, raising the possibility that regular recovery from economic downturns

rather than political change drove the growth improvements that we observe. Section

3 describes how we address this concern.

2.3. State Capacity

Our main state capacity measure uses per capita national government revenues to proxy

for the state’s extractive capacity. The revenue and population data are from Dincecco

(2011) and Dincecco et al. (2011). Our second measure uses per capita non-military

expenditures by national governments to proxy for the state’s productive capacity.

Data that are disaggregated beyond non-military expenditures (e.g., education) are

only available for a reduced number of sample countries. The Appendix describes the

sources and construction methods for the spending data.

The ﬁscal data form an unbalanced panel. We linearly interpolate, but never ex-

trapolate, any missing revenue data to provide annual observations from 1650 to 1913.

We also linearly interpolate population data between census years. The non-military

spending data are not widely available before 1816. We do not interpolate any miss-

ing expenditure observations, because the links between tax bases and government

spending were not always straightforward (e.g., during wartime).

Simple calculations show that the average annual growth rate of per capita revenues

for ﬁscally centralised regimes (1.76%) was more than twice as high as for ﬁscally

fragmented regimes (0.82%). Likewise, the average annual growth rate of per capita

revenues for limited government regimes was 1.60%, but only 1.14% for absolutist

regimes, a diﬀerence of 0.46 percentage points. The share of non-military expenditures

in total expenditures also increased by roughly 13 percentage points after both ﬁscal

centralisation and limited government.

Finally, we use cumulative railway networks in miles from Bogart (2008) as a non-

ﬁscal proxy for the state’s “infrastructural power.” Even if the state does not directly

ﬁnance, build, or operate transport systems, it can play a key role as facilitator (see

Section 1). We thus favor the total railway network to the government-built part of the

network as the proper measure. However, the econometric results for the relationship

between ﬁscally centralised regimes and government-built railway networks were similar

to those reported. Data limitations prevented us from testing this relationship for

limited government regimes.

3. Econometric Methodology

To test the direct relationship between political transformations and economic perfor-

mance, we estimate the benchmark regression equation

∆yi,t =α0+α1Ci,t +α2Li,t +X0

i,t−1α3+µi+λt+i,t.(1)

The dependent variable is the (logarithmic) annual growth rate of real per capita GDP

in country ibetween t−1 and t. The ﬁxed eﬀects µiaccount for time-invariant country

characteristics including geography, the nature and quality of initial political institu-

tions (Bockstette et al., 2002; Acemoglu et al., 2005), initial economic and technological

conditions (Comin et al., 2010), and cultural norms (Greif and Tabellini, 2010), while

the ﬁxed eﬀects λtaccount for common time shocks. X0

i,t−1includes an external con-

ﬂict dummy for each year that a sample country participated in an European war, an

internal conﬂict dummy for each year of civil war, coup, or revolution, and a control

for annual population growth. Ci,t and Li,t are dummy variables that take the value

1 upon ﬁscal centralisation and limited government, respectively, in country i(and

take the value 0 beforehand). The coeﬃcients α1and α2thus provide within-country

estimates of the relationships between the two political transformations and economic

performance.

Equation 1 addresses several common limitations of the cross-country growth litera-

ture (Durlauf et al., 2005). Country ﬁxed eﬀects account for time-invariant characteris-

tics and initial conditions that may inﬂuence economic and political development alike.

Furthermore, by including lags of the dependent variable among the regressors in some

speciﬁcations, we not only account for autocorrelation and growth persistence, but can

also quantify the short- and long-run relationships between political transformations

and economic and ﬁscal outcomes.18

However, a set of methodological challenges still remains. A ﬁrst concern is reverse

causation. Economic growth may promote political reforms and not vice versa (e.g.,

Glaeser et al., 2004; see Acemoglu et al., 2008, for a counterargument). Beyond classic

reverse causation, the growth estimates will be upwardly biased if political transforma-

tions took place during an economic upswing, because the coeﬃcients α1and α2would

reﬂect this trend. Similarly, if political transformations took place during an economic

downturn, then the estimates may simply reﬂect the impacts of regular recovery that

would have taken place even without political change.

We address this concern in several ways. To test whether political transformations

took place during economic upswings or downturns, we allow for time-varying impacts

of political transformations (Giavazzi and Tabellini, 2005; Laporte and Windmeijer,

2005). We also perform placebo tests (Bertrand et al., 2004; Stasavage, 2013) that

recode political transformations as if they had taken place decades prior to the actual

dates. Finally, we implement the Granger-style causality test proposed by Angrist and

Pischke (2009, ch.5).

A second concern is omitted variable bias. While ﬁxed eﬀects control for constant

unobserved country-level heterogeneity, they do not account for the potential omission

of relevant time-varying explanatory variables that could be correlated with political

transformations. To address this possibility, we modify our benchmark model to include

country-speciﬁc time trends.19 Similarly, we allow for income level diﬀerences and

18 The bias induced by the inclusion of a lagged dependent variable in least square dummy variable (LSDV)

models decreases with the panel’s time dimension T(Nickell, 1981). Judson and Owen (1999) show that

the LSDV estimator outperforms alternative dynamic panel data models when Tis 30; for our unbalanced

panel with annual observations, Tranges from 50 to 264, with an average value of 194. Furthermore, Judson

and Owen show that the bias for the coeﬃcients of the other regressors, including the those of our variables

of interest, Ci,t and Li,t, is small. For robustness, we used lagged observations dated t−2 and earlier to

instrument for ∆yi,t−1following Bond et al. (2010), and implemented the bias-corrected LSDV estimator

proposed by Bruno (2005). The results were similar to those reported.

19 Country-speciﬁc time trends also help account for the non-stationarity of the political reform dummies,

control for convergence dynamics by including lagged real per capita GDP in X0

i,t−1

in some speciﬁcations.

There are other concerns related to the unique historical nature of our database.

Our benchmark model uses cluster-robust standard errors that accommodate het-

eroskedasticity and within-cluster correlation. However, the number of sample coun-

tries (i.e., 11) is small, which may bias downward the standard errors (Donald and

Lang, 2007; Cameron et al., 2008). For robustness, the regression tables also report

p-values for the two-sided Wald hypothesis tests computed according to Cameron et

al. (2008)’s wild bootstrap-t procedure. This procedure is useful for small sample sizes

like ours because it does not rely on asymptotic approximations.20

Although we use a wide variety of techniques to address the various methodological

concerns, our within-country estimates do not necessarily capture the causal eﬀects

of political transformations on economic and ﬁscal outcomes. The historical record

indicates that the causes and consequences of political transformations were the result

of complex interactions between a broad range of factors, some of which our econo-

metric framework invariably omits. While our estimates are not causal in nature, we

believe that our analysis highlights novel data patterns about the relationships between

political transformations, state capacity, and economic performance.

4. Political Transformations and Economic Performance

Table 3 displays the benchmark estimation results for the direct relationship between

political transformations and annual real per capita GDP growth.21 Column 1 reports

since once sample countries adopted centralised and limited regimes they typically did not revert.

20 To account for cross-sectional dependence beyond that captured by time ﬁxed eﬀects, we also computed

Driscoll-Kraay (1998) standard errors, which are robust to general forms of spatial and temporal dependence

regardless of the size of N. The results were similar to those reported.

21 We tested the time series properties of the main variables by performing common panel-data unit root

tests (e.g., Maddala and Wu, 1999; Levin et al., 2002; Im et al., 2003) with and without country-speciﬁc

trends. All tests failed to reject the hypothesis that the (log) level of per capita GDP contains unit roots for

every sample country. However, the null was always rejected when this variable was ﬁrst-diﬀerenced. These

tests also suggest that the state capacity measures are stationary. Still, given the well-known reservations

about the power and reliability of these tests (Baltagi, 2005), we report results in Sections 4 and 5 for a broad

range of speciﬁcations that take into account both theoretical considerations and time series properties.

the results for the static panel model with country ﬁxed eﬀects. There were signiﬁcant

positive growth improvements after both ﬁscal centralisation and limited government.

Fiscally centralised regimes saw an average annual per capita GDP growth rate that was

0.66% higher than ﬁscally fragmented regimes, while limited government regimes saw

an average annual per capita GDP growth rate 0.32% higher than absolutist regimes.

Column 2 controls for common time shocks by introducing year ﬁxed eﬀects. The

coeﬃcient for Ci,t falls to 0.22, but is still more than two standard deviations greater

than zero. The coeﬃcient for Li,t is no longer signiﬁcant. A potential explanation is

that, because the establishment of limited government typically took place just before

or during the Industrial Revolution (Mokyr, 1998), it may be diﬃcult to disentangle

the two events. To account for unobserved time-varying country-level heterogeneity,

Column 3 introduces country-speciﬁc time trends to the Column 2 speciﬁcation. There

is a small increase in the magnitude of the coeﬃcient for ﬁscal centralisation, which

remains signiﬁcant.

Columns 4 and 5 test potential interactions between the two political transforma-

tions. In this context, the individual coeﬃcients for Ci,t and Li,t measure the economic

impacts of ﬁscal centralisation and limited government, respectively, when the trans-

formation in question was undertaken in the absence of the other, while the coeﬃcient

for Ci,t ×Li,t measures the non-additive impact of the two political transformations

combined. Column 4 shows that the coeﬃcients for Ci,t and Ci,t ×Li,t are positive,

but not signiﬁcant. It is well-known that time ﬁxed eﬀects impose large costs in terms

of lost degrees of freedom for very long panels like ours (Wooldridge, 2003; Cameron

and Trivedi, 2010). To address this potential problem, Column 5 repeats the previous

speciﬁcation after replacing the year ﬁxed eﬀects with a linear time trend to account for

common time shocks. Now the coeﬃcient for ﬁscal centralisation is signiﬁcant, with a

point estimate of 0.25. Furthermore, the coeﬃcient for the interaction term is 0.31 and

is signiﬁcant. This result suggests that the joint impact of political transformations

was notably greater than the sum of the individual impacts. Holding all else constant,

the average annual per capita GDP growth rate for a country in our sample where

limited government followed ﬁscal centralisation was 1%, but only 0.63% if the second

political transformation was never undertaken.

Columns 6 and 7 repeat the speciﬁcations in Column 2 (country and year ﬁxed

eﬀects) and Column 3 (plus country-speciﬁc time trends) for the dynamic panel model

that introduces two lags of the dependent variable to account for growth persistence.22

The coeﬃcient for ﬁscal centralisation remains signiﬁcant, with an increase in the

annual per capita GDP growth rate of 0.29% to 0.34%. Furthermore, the dynamic

panel results indicate that ﬁscally centralised regimes not only saw a short-run increase

in growth, but also a signiﬁcant increase in long-run per capita GDP growth rates of

0.22 to 0.25 percentage points per year.

Table 4 tests the robustness of the main results. Column 1 introduces the controls

for external and internal conﬂicts and population growth to the static panel model

with country and year ﬁxed eﬀects. Column 2 adds country-speciﬁc time trends to this

speciﬁcation. The magnitude and signiﬁcance of the coeﬃcients for ﬁscal centralisation

are similar as before. None of the controls have systematic growth impacts.23

There is the possibility that our previous results conﬂate the direct economic im-

pacts of ﬁscal centralisation with the advent of modern industrial growth, which took

place throughout Europe from the 1850s onward (Mokyr, 1998). Column 3 thus re-

stricts the previous speciﬁcation to the period before 1845. The magnitude of the

coeﬃcient for Ci,t increases to 0.32 and remains signiﬁcant.

The next three columns report results for the dynamic panel model. Column 4

repeats the Column 2 speciﬁcation with country-speciﬁc time trends. The magnitude

and signiﬁcance of the coeﬃcient for ﬁscal centralisation resembles previous estimates.

22 As shown, both lags are statistically signiﬁcant. A model with a single lag, the structure favored by

commonly-used information criteria (AIC, BIC), leads to nearly identical estimates for the political reform

indicators. We also tested longer lag structures of up to 8 lags to mimic other speciﬁcations present in the

long-run growth literature (e.g., Gemmell et al., 2011). The results for α1and α2were unchanged. Finally,

allowing for diﬀerent lag lengths for each sample country did not signiﬁcantly alter these results.

23 If we allow for contemporaneous correlations between the time-varying controls and ∆yi,t, then the

negative coeﬃcients for external conﬂict and population growth become statistically signiﬁcant. However,

the estimated relationship between ﬁscal centralisation and economic growth is unchanged. We report the

results that exclude contemporaneous correlations to address concerns about the possible endogeneity of the

controls (Jones, 1995; Gemmell et al., 2011).

To control for conditional convergence, Column 5 includes the three-year lag of log

per capita GDP, ln yi,t−3. The coeﬃcient for Ci,t remains positive, but is no longer

signiﬁcant. Column 6 thus repeats the previous speciﬁcation using a linear time trend

as justiﬁed previously. Now the coeﬃcient for ﬁscal centralisation is highly signiﬁcant,

with a point estimate of 0.43.

Columns 7 to 9 show the results that use averaged rather than annual data obser-

vations. Some scholars highlight the merits of using annual data in growth regressions

(Attanasio et al., 2000; Papaioannou and Siourounis, 2008), while others argue that ob-

servations averaged over longer periods are more appropriate for growth determinants

that change slowly or infrequently (Durlauf et al., 2005). Taking data averages can

also ﬁlter out business cycle ﬂuctuations and adjustments to occasional shocks (Islam,

1995; Beck et al., 2000), and help attenuate the eﬀects of (transient) measurement

errors (Bond et al., 2010). Column 7 thus estimates the static model with country and

year ﬁxed eﬀects using ﬁve-year data averages.24 Columns 8 and 9 use 10- and 25-year

averages, respectively. The main results obtained using the annual data hold for the

averaged data. Fiscal centralisation saw signiﬁcant growth improvements, while the

coeﬃcients for limited government are statistically indistinguishable from zero.

To better assess the evolution of the relationships between political transformations

and economic performance, we relax the assumption that the impacts of ﬁscal central-

isation and limited government are constant over time and replace Ci,t and Li,t with

ﬁve “pulse” dummies each. The regression equation that we now estimate is

∆yi,t =α0+

j=1

α1,j e

i,t +

j=1

α2,j e

i,t +X0

i,t−1α3+µi+λt+i,t,(2)

where the ﬁrst four pulse dummies span non-overlapping ﬁve-year intervals before and

after each political transformation: e

i,t,e

i,t = 1 for years 6 to 10 before; e

i,t,e

i,t = 1

for years 1 to 5 before; e

i,t,e

i,t = 1 for years 0 to 4 after (including the transformation

24 Since the average number of observations for the ﬁve-year averaged dataset is 39, the LSDV estimator

is still preferable to alternative models (Judson and Owen, 1999). While the generalized method of moments

(GMM) estimator (Arellano and Bond, 1991) gives similar results to those reported, these results should be

interpreted with caution due to the well-known weaknesses of the GMM estimator for small N panels.

year itself); and e

i,t,e

i,t = 1 for years 5 to 9 after. To measure the long-run relation-

ships between political transformations and economic performance, e

i,t and e

i,t take

the value 1 from the tenth year post-transformation onward, and 0 otherwise.25

Figure 2 displays the results of this exercise. Neither “pre-treatment” dummy is

signiﬁcant, which suggests that there were no systematic economic diﬀerences between

the decade prior to political transformations and the benchmark period (i.e., more

than a decade prior). To put it diﬀerently, the positive relationship between ﬁscal

centralisation and economic performance that we ﬁnd in Tables 3 and 4 does not appear

to be driven by regular recovery from any pre-transformation economic downturn, or

by anticipatory eﬀects. Nor do we ﬁnd evidence for any signiﬁcant economic changes

in the ﬁrst ﬁve years post-transformation. This result may highlight the importance of

“institutional consolidation” before any impacts of political transformations could be

realized. By contrast, the coeﬃcient for e

Ci,4shows evidence for medium-run economic

improvements. The average annual real per capita GDP growth rate was 0.48% higher

5 to 9 years after ﬁscal centralisation. This coeﬃcient is signiﬁcant at the 10% level.

Similarly, ﬁscally centralised regimes saw highly signiﬁcant economic improvements

from the tenth year post-transformation onward, with a long-run per capita GDP

growth gain of 0.24 percentage points per year. The size of this coeﬃcient resembles

the long-run estimates from the dynamic speciﬁcations in Tables 3 and 4.26

Overall, the analysis in this section shows an important direct role for ﬁscal centrali-

sation in economic performance. The coeﬃcient estimates for Ci,t are signiﬁcant, large,

and robust: ﬁscally centralised regimes grew faster than ﬁscally fragmented regimes by

an average of 0.17% to 0.43% higher per year. Given that the average annual real per

25 For robustness, we tested shorter and longer pulse dummy lengths and diﬀerent numbers of pulse

dummies overall. We also tested a variation of Equation 2 that replaced just one political reform indicator

with pulse dummies but held the other indicator constant. The results of these tests consistently indicated

a highly signiﬁcant positive long-run impact of Ci,t on per capita GDP growth.

26 As another robustness check, we relaxed the assumption that the impact of political transformations was

the same for all cross-sectional units. In the spirit of Pesaran and Smith (1995)’s mean group-style estimator,

we estimated the dynamic models in Table 4 for each sample country and then averaged the coeﬃcients. The

point estimate for the long-run relationship between ﬁscal centralisation and economic growth was roughly

0.30 and was signiﬁcant. While we interpret this result with caution due to the small N, it provides further

support for our main conclusions.

capita GDP growth rate among sample countries over the 1650-1913 period was 0.62%,

our estimates indicate that the growth improvements associated with ﬁscal centralisa-

tion were equivalent to about one-quarter to two-thirds of the actual per capita GDP

growth rate over this period, and 8% to 20% of its standard deviation. As noted in the

introduction, these magnitudes suggest that ﬁscal centralisation was roughly as im-

portant historically to economic development as democratisation is sometimes thought

to be today (Papaioannou and Siourounis, 2008). Furthermore, the economic im-

provements associated with ﬁscal centralisation were long-lasting. Fiscally centralised

regimes saw a signiﬁcant increase in long-run per capita GDP growth rates of 0.16 to

0.33 percentage points per year. While the coeﬃcient estimates for limited government

are typically positive, the direct relationship between this political transformation and

economic performance is not signiﬁcant. However, we do ﬁnd some evidence of signiﬁ-

cant interaction eﬀects between the two political transformations, such that the joint

impact of ﬁscal centralisation and limited government combined was much greater than

the sum of each transformation when undertaken independently.

As described in Section 1, there are several potential mechanisms through which

political transformations could have improved economic performance. We now examine

one speciﬁc mechanism: greater state extractive and productive capacity.

5. Role of State Capacity

5.1. Political Transformations and State Capacity

To test for the relationship between political transformations and state capacity, we

use a modiﬁed version of Equation 1 which takes the dependent variable ∆Ei,t, the

(logarithmic) annual growth rate of state capacity in country ibetween t−1 and t.

Table 5 displays the estimation results for growth in per capita revenues, our mea-

sure of the state’s extractive capacity. Column 1 shows the static panel model with

country and year ﬁxed eﬀects. There was a signiﬁcant extractive capacity improvement

after ﬁscal centralisation. Fiscally centralised regimes saw an average annual growth

rate of per capita revenues that was 1.41% higher than ﬁscally fragmented regimes.

The estimate for limited government is positive, but not signiﬁcant.27 Column 2 in-

troduces the standard time-varying controls. The magnitude of the estimate for Ci,t

increases slightly and remains signiﬁcant.

Column 3 introduces country-speciﬁc time trends. The coeﬃcient for ﬁscal central-

isation increases further to 2.93 and remains signiﬁcant. The estimate for Li,t becomes

signiﬁcant (for p-values computed using cluster-robust standard errors) once we re-

place the year ﬁxed eﬀects with a linear time trend to reduce the cost of lost degrees

of freedom as in Column 4.

Column 5 restricts the Column 3 speciﬁcation to the pre-1845 data to further control

for the potential ﬁscal impacts of the Industrial Revolution. The coeﬃcient for Ci,t

roughly doubles in size to 6.61. The coeﬃcient for Li,t is also large and signiﬁcant.28

Columns 6 and 7 repeat the speciﬁcations in Columns 3 and 4 for the dynamic

panel model with two lags of the dependent variable. The estimates for both ﬁscal

centralisation and limited government are similar in magnitude and signiﬁcance as

before. Furthermore, these estimates indicate that ﬁscally centralised regimes saw

a large increase in the long-run growth rate of per capita revenues of 2.89 to 3.02

percentage points per year.

To account for conditional convergence, Column 8 includes the three-year lag of log

per capita revenues, ln Ei,t−3, in the speciﬁcation from Column 6. The coeﬃcient for

Ci,t is again large and signiﬁcant. The results also indicate that ﬁscal centralisation

was associated with a signiﬁcant increase in (log) per capita revenue levels of 37% over

the long run.29

Table 6 displays the estimates for the model with country and year ﬁxed eﬀects,

27 However, if we use log per capita revenue levels as the dependent variable and re-run the Table 5

regressions, then the estimates for limited government are nearly always signiﬁcant.

28 Unlike for economic growth, the Table 5 regressions do not show evidence of signiﬁcant interaction

eﬀects between the two political transformations.

29 We computed the long-run impact on per capita revenue levels as minus the ratio of the coeﬃcients

for Ci,t and ln Ei,t−1, with standard errors obtained using the delta method (Papaioannou and Siourounis,

2008). Re-running the Column 6 regression with ln Ei,t as the dependent variable indicates that ﬁscal

centralisation saw a signiﬁcant short-term annual increase in per capita revenue levels of 4%.

time-varying controls, and country-speciﬁc time trends for our alternative state capac-

ity measures. Columns 1 and 2 show the results for growth in per capita non-military

expenditures, our measure of the state’s productive capacity. In line with the theoretical

implications as described in Section 1, the coeﬃcient for limited government is always

positive and becomes signiﬁcant (for p-values computed using cluster-robust standard

errors) once we control for conditional convergence in the dynamic panel model as in

Column 2.

For contrast, Column 3 shows the results for growth in per capita military expendi-

tures. Unlike the previous results, there is no evidence that limited government regimes

saw increases in the average annual growth rate of per capita military expenditures (the

coeﬃcient for Li,t is negative). This comparison supports the case that the establish-

ment of limited government was associated with changes toward non-military spending.

The estimates for ﬁscal centralisation are never signiﬁcant across these speciﬁcations.

Recall from Section 2 that expenditure data disaggregated beyond military expen-

ditures are not widely available. With this important caveat in mind, Columns 4

and 5 show the results for growth in per capita education expenditures for the static

and dynamic panel models, respectively. The estimates for Li,t are positive, large,

and signiﬁcant. These results suggest that education was one non-military item upon

which limited government regimes spent funds. Again, the estimates for Ci,t are not

signiﬁcant.

Columns 6 and 7 show the results for growth in cumulative railway networks, our

non-ﬁscal state capacity measure. The coeﬃcients for Li,t, while positive, are not sig-

niﬁcant for this measure. By contrast, the coeﬃcient for Ci,t becomes signiﬁcant (for

p-values computed using cluster-robust standard errors) once we account for condi-

tional convergence in the dynamic panel model as in Column 7. This result provides

some further evidence that ﬁscal centralisation enhanced the “infrastructural power”

of states.

In summary, Table 5 shows an important role for political transformations in greater

state capacity. The estimates indicate that ﬁscally centralised regimes and, to some

extent, limited government regimes extracted per capita revenues at signiﬁcantly higher

rates than ﬁscally fragmented regimes or absolutist regimes, respectively. Fiscally

centralised regimes saw an average annual growth rate of per capita revenues that was

1.41% to 4.27% higher than ﬁscally fragmented regimes. Given that the average annual

growth rate of per capita revenues over the 1650-1913 period was 1.36%, these estimates

are sizeable. Furthermore, the extractive capacity improvements associated with ﬁscal

centralisation were enduring. Fiscally centralised regimes saw an increase in long-run

growth rates of per capita revenues of 1.54 to 1.65 percentage points per year. While

Table 6 shows some evidence for a signiﬁcant relationship between limited government

and productive capacity improvements, this evidence is less robust. Taken together,

these results suggest that political transformations had more important consequences

for extractive capacity than for productive capacity. To complete the analysis of the

state capacity mechanism, we now test for the relationship between state capacity and

economic performance.

5.2. State Capacity and Economic Performance

Following Bond et al. (2010), the benchmark regression equation that we estimate is

∆yi,t =β0+

j=1

βj∆Ei,t−j+β3ln Ei,t−3+X0

i,t−1β4+µi+λt+i,t,(3)

where, as before, ∆yi,t is the (logarithmic) annual growth rate of real per capita GDP in

country ibetween t−1 and t, ∆Ei,t−j, j = 1,2, are the ﬁrst two lags of the (logarithmic)

annual growth rate of state capacity, and ln Ei,t−3is the three-year lag of the state

capacity measure in log levels.30

Table 7 displays the estimation results. The ﬁrst ﬁve columns test per capita

revenues, our measure of the state’s extractive capacity. Column 1 shows the results

30 We excluded contemporaneous correlations between ∆yi,t and the state capacity measures to address

endogeneity concerns (Jones, 1995; Gemmell et al., 2011). For robustness, we included either ∆Ei,t, ln Ei,t ,

or both variables as regressors in Equation 3, using past values of the state capacity measures and historical

variables including state antiquity (Bockstette et al., 2002) and protection of capital (Acemoglu et al., 2005)

as instruments. The results were similar to those reported.

for the static panel model with country and year ﬁxed eﬀects. Column 2 introduces the

standard time-varying controls. Column 3 adds country-speciﬁc time trends. There

is a signiﬁcant relationship between extractive capacity improvements and subsequent

per capita GDP growth across all speciﬁcations: the coeﬃcient of interest, ln Ei,t−3,

is always positive and signiﬁcant, with values between 0.11 and 0.15. A comparison of

eighteenth-century England, which had undertaken both political transformations, and

France, which was still ﬁscally fragmented and absolutist, helps put these magnitudes

into perspective. English per capita revenues, which averaged 7.50 gold grams from

1700 to 1788, were more than double the French average over this period, at 3.71 gold

grams. Our estimates thus suggest that, ceteris paribus, this diﬀerence in extractive

capacity was associated with an annual per capita GDP growth rate for England that

was between 0.08 and 0.11 percentage points higher than for France. Given that the

actual average annual per capita GDP growth rate for France over the 1700-88 period

was 0.19, these magnitudes are large.

Column 4 repeats the Column 3 speciﬁcation for the dynamic panel model with two

lags of the dependent variable. The coeﬃcient for ln Ei,t−3remains highly signiﬁcant

and increases in magnitude to 0.18, implying a long-term impact on the per capita GDP

growth rate of 0.13 percentage points per year. Recall that ﬁscal centralisation was

associated with a signiﬁcant long-run increase in per capita revenue (log) levels of 37%

(Column 8 of Table 5). A simple back-of-the-envelope calculation thus suggests that

the impact of ﬁscal centralisation on the long-run annual per capita GDP growth rate

that went through the revenue channel was about 0.05 percentage points, or roughly

15% to 30% of the total long-run per capita GDP growth improvements associated

with this political transformation.

To further assess the relative importance of this mechanism, Column 5 shows the

results for the Column 3 speciﬁcation after introducing the dummy variables for ﬁscal

centralisation and limited government. The magnitude and signiﬁcance of the coeﬃ-

cient for ln Ei,t−3is similar as before. The coeﬃcient for ﬁscal centralisation is also

signiﬁcant, with a similar point estimate (0.26) as the equivalent Section 4 speciﬁcation

(i.e., Column 2 of Table 4). Taken in combination, these results suggest that, while

greater extractive capacity was one mechanism through which ﬁscal centralisation im-

proved long-run economic performance, non-ﬁscal mechanisms as noted in Section 1

(e.g., reductions in internal tariﬀ barriers that increased investment mobility) were also

of great signiﬁcance.

Column 6 shows the results for growth in per capita non-military expenditures. The

coeﬃcient for ln Ei,t−3is negative and not signiﬁcant. We do not report the results for

the other alternative state capacity measures, which were also not signiﬁcant.

Overall, Tables 5, 6, and 7 show evidence that greater state capacity was one

mechanism that linked political transformations with better economic performance.

There is a signiﬁcant positive relationship between ﬁscal centralisation and, to some

extent, limited government and extractive capacity, and between greater extractive

capacity and economic growth. While we did not ﬁnd evidence for a direct relationship

between limited government and economic performance in Section 4, these results

suggest that limited government played some indirect economic role through extractive

capacity improvements. These results also indicate that ﬁscal centralisation improved

economic performance through both ﬁscal and non-ﬁscal mechanisms. To conclude the

analysis, the next section uses placebo tests to further evaluate the robustness of the

main results.

6. Placebo Tests

The historical evidence described in Section 2 highlights the role of critical junctures in

political transformations (Acemoglu and Robinson, 2012, ch.4). Similarly, the results

of the pulse dummy exercise show no evidence of anticipatory eﬀects during the decade

prior to political transformations (Figure 2). For further robustness, we now perform

placebo tests (Bertrand et al., 2004; Stasavage, 2013) that address the possibility that

economic and ﬁscal diﬀerences across political regimes were the result of underlying

trends that preceded political transformations, rather than the transformations them-

selves. We recode political transformations as if they had taken place decades prior to

the actual dates and then re-estimate the static and dynamic models with country and

year ﬁxed eﬀects and time-varying controls. If the coeﬃcients for the placebo transfor-

mations are not signiﬁcant, then this analysis will reinforce our previous results about

the economic and ﬁscal importance of political transformations.

Table 8 displays the estimation results for the placebo tests. Panel A reports the

results when the dependent variable is annual real per capita GDP growth. Column 1

shows the results for the political transformation placebos 25 years prior to the actual

dates, while Columns 2 to 4 increase the placebos to 50, 75, and 100 years prior,

respectively. The coeﬃcients for the ﬁscal centralisation placebos are small, negative,

and not signiﬁcant. For example, the 25-year placebo estimate in the static model is

0.007, versus 0.21 in the original speciﬁcation (Table 4, Column 1).

Panel B repeats the placebo tests when the dependent variable is per capita revenue

growth. The placebo coeﬃcients for ﬁscal centralisation or limited government are

typically not signiﬁcant, and the magnitudes are routinely smaller than in the original

speciﬁcations. More than 80% of the placebo estimates are negative.

Panel C estimates the placebo models when the dependent variable is per capita

non-military expenditure growth. We report the results for 25-, 30-, 35-, and 40-

year placebos (we cannot compute the 50-year placebo due to the lack of non-military

expenditure data prior to 1816). Nearly all of the coeﬃcients for the limited government

placebos are negative, and none are signiﬁcant.

As an even further robustness check for reverse causation, we implement the pro-

cedure proposed by Angrist and Pischke (2009, ch.5). Our main result from Section 4

is that there is a signiﬁcant relationship between ﬁscal centralisation and annual real

per capita GDP growth. If Ci,t impacts ∆yi,t but not the other way around, then the

coeﬃcients for the leads Ci,t+τ,τ= 1, . . . , q, should not be statistically signiﬁcant in a

regression of the sort

∆yi,t =α0+α1,0Ci,t +

τ=1

α1,τ Ci,t+τ+α2Li,t +X0

i,t−1α3+µi+λt+i,t.(4)

Figure 3 displays the results of this regression for two speciﬁcations: the ﬁrst with q= 3

and no controls (left panel), and the second with q= 20, time-varying controls, and

country-speciﬁc trends (right panel). Only the coeﬃcient for Ci,t remains signiﬁcant

across speciﬁcations. The α1,τ s alternate between positive and negative coeﬃcients,

with p-values ranging from 0.13 to 0.81. Using diﬀerent speciﬁcations or qvalues, or

simultaneously including leads for Ci,t and Li,t, leads to broadly similar results.

Overall, the results of the placebo and Angrist-Pischke tests provide additional ev-

idence that reverse causation does not drive the relationships between political trans-

formations and the economic and ﬁscal outcomes that we ﬁnd. They thus reinforce our

previous ﬁndings.

7. Conclusion

This paper presents new evidence about the long-run relationship between state capac-

ity and economic performance. We focus on Europe, the birthplace of modern economic

growth. National governments in European history were typically ﬁscally fragmented

and absolutist. We argue that both ﬁscal centralisation and limited government in-

creased the state’s capacity to extract greater tax revenues, and that greater state

capacity had positive economic implications through the creation of administrative

infrastructure and other channels.

To test this argument, we perform a panel data analysis on a novel database that

spans eleven countries and four centuries. Our analysis accounts for potential biases

including simultaneity, omitted variables, and unobserved heterogeneity. Placebo tests

allow us to further evaluate the validity of our argument. The results suggest that ﬁscal

centralisation rather than limited government was the most consequential political

change to occur in Europe from the Old Regime to World War I. We ﬁnd a signiﬁcant

direct relationship between ﬁscal centralisation and economic growth. Furthermore, we

ﬁnd that greater state capacity was one mechanism through which ﬁscal centralisation

and, to some extent, limited government played signiﬁcant economic roles. To the best

of our knowledge, these results are among the ﬁrst to show systematic evidence that

state capacity is a signiﬁcant determinant of long-run economic growth.

Our analysis indicates that ﬁscal centralization operated through both ﬁscal and

non-ﬁscal mechanisms. One extension is to identify the main non-ﬁscal mechanisms.

We believe that our results take a ﬁrst step that can guide future research in this

direction.

Data Appendix

The database and do-ﬁle will be available upon publication from the website https:

//sites.google.com/site/mdincecco/.

Data for per capita tax revenues in gold grams from 1650 to 1913 are from Dincecco

(2011, appendices A.1, A.2, A.3). See Section 2 for further details.

Data sources for per capita military and education expenditures are listed ahead.

Disaggregated expenditure data in home currencies were converted into gold grams fol-

lowing the methodology in Dincecco (2011, appendix A.2). Data for total expenditures

and population are from Dincecco (2011, appendices A.1, A.2) unless otherwise stated.

Per capita non-military expenditures were computed as per capita total expenditures

minus per capita military expenditures.

Austria. Military spending data are from Pammer (2010). Education expenditure

data were not available.

Belgium. Military spending data are from Singer (1987). They were downloaded from

the Correlates of War website as the National Military Capabilities Dataset, Version

4.0. Education expenditure data were not available.

Denmark. Military spending data are from Singer (1987). They were downloaded from

the Correlates of War website as the National Military Capabilities Dataset, Version

4.0. Education expenditure data were not available.

England. Military and education spending data are from Mitchell (1988, public ﬁnance

table 4). To compute military expenditures, spending for the Army and Ordnance and

for the Navy were summed. Education expenditures uses the category for Education,

Art, and Science.

France. Military and and education spending data are from Fontvieille (1976, tables

CVXI-XXXV).

Netherlands. Military spending data are from van Zanden (1996, table 4) for 1816-

41. Van Zanden provides data averages for 1816-20, 1821-4, 1825-9, 1831-4, 1835-9,

and 1841-50. The average for 1816-20 was used for 1816, the average for 1821-4 for

1821, and so on. The military spending shares closely match those from van Zanden

and van Riel (2010, table 2.1). Total expenditure data from this source were used

in combination with the information on shares to back out military expenditures. For

1816-30 we divided these ﬁgures by the expenditure share for the Southern Netherlands

(i.e., Belgium, Luxembourg, and their hinterlands) according to van Zanden (1996,

table 5) to derive military expenditures for the (Northern) Netherlands, as data for

total expenditures from Dincecco (2011) exclude the Southern Netherlands. The source

for the 1816-41 data does not provide education spending, which is included under the

expenditure category for Home Aﬀairs. Military and education spending data are from

van Zanden and van Riel (2010, table 2.3) for 1850-1913. They provide data shares at

10-year intervals for 1850, 1860, 1870, 1880, 1890, 1900, and 1913. Total expenditure

data from this source were used in combination with the information on shares to back

out military expenditures.

Piedmont. Military spending data are from Dincecco et al. (2011) for 1830-59, the

Uﬃcio Storico (1980, pp. 508-9) for 1861-9 (Italy), and Hobson (1993) for 1870-1913

(Italy). Education expenditures are from Felloni (1959) for 1830-59 and Brosio and

Marchese (1986, table 4a) for 1861-1913 (Italy). Population data are from Dincecco et

al. (2011) for 1830-59.

Portugal. Military and education spending data are from Silveira (1987, table 8)

for 1816-27, Mata and Val´erio (2001, table 1) for 1832-45, and Mata (1993, table

1) for 1851-1913. To compute military expenditures, spending by the Ministerio da

Guerra (after 1827; Exercito beforehand) and the Ministerio da Marihna were summed.

There was no education ministry over this period. Education expenditures thus uses

the category for the education burden (Encargos cum Instru¸c˜oes). Since the total

military spending calculation matches well with the Encargos cum Difesa category

(and perfectly from 1884 onward), we are conﬁdent that the same holds for education.

Dincecco (2011) does not provide total expenditures for Portugal for 1833; this data

point was taken from Mata and Val´erio (2001, table 1).

Prussia. The German Reich (1871-1945) was a federal system and a great deal of taxing

and spending was done at the state (e.g., Prussian) level. The federal government was

responsible for military expenditures and welfare (Ziblatt, 2006). Spoerer (2010, table

4.1) provides Prussian military and welfare expenditures for 1847 and 1867. After

uniﬁcation there are only Reich data available for these categories. These data were

not used because there was no clear way of integrating the pre-1871 Prussian series with

the post-1870 Reich one. Spoerer’s data for Prussia were supplemented with 1820 data

for military defence from Ziblatt (2006, table 3.1). Here total Prussian expenditures

from 1821 were used due to data availability.

Spain. Military spending data are from Carreras and Tafunell (2006), table 12.8 for

1816-42 and table 12.13 for 1845-1913. To compute military expenditures, spending by

the Ministerio de Guerra (through 1842; the Ministerio de Defensa from 1845 onward)

and the Ministerio de Marina were summed. The sources for the 1816-99 data do

not provide education spending, which is included under the expenditure category

for the Ministerio de Estado through 1842 and the Ministerio de Fomento from 1845

onward. Education spending data for the Ministerio de Fomento are displayed for

1900-13 (Ministerio de Educaci´on y Ciencia).

Sweden. Military spending data are from Krantz and Sch¨on (2010, table XI). At the

central government level, there were no separate expenditure categories for infrastruc-

ture or education.

University of Michigan

University of Exeter

Submitted: 9 November 2012

Accepted: 15 April 2014

References

Acemoglu, D. (2005). ‘Politics and economics in weak and strong states’, Journal of

Monetary Economics, vol. 52(7), pp. 1199-226.

Acemoglu, D., S. Johnson, and J. Robinson (2001). ‘Colonial origins of comparative

development: an empirical investigation’, American Economic Review, vol. 91(5),

pp. 1369-401.

Acemoglu, D., S. Johnson, and J. Robinson (2002). ‘Reversal of fortune: geography

and development in the making of the modern world income distribution’, Quarterly

Journal of Economics, vol. 117(4), pp. 1231-94.

Acemoglu, D., S. Johnson, and J. Robinson (2005). ‘The rise of Europe: Atlantic trade,

institutional change, and economic growth’, American Economic Review, vol. 95(3),

pp. 546-79.

Acemoglu, D., S. Johnson, J. Robinson, and P. Yared (2008). ‘Income and democracy’,

American Economic Review, vol. 98(3), pp. 808-842.

Acemoglu, D., S. Naidu, P. Restrepo, and J. Robinson (2014). ‘Democracy does cause

growth’, NBER Working Paper 20004.

Acemoglu, D. and J. Robinson (2012). Why Nations Fail, London: Proﬁle.

Acemoglu, D., J. Robinson, and T. Verdier (2004). ‘Alfred Marshall lecture – klep-

tocracy and divide-and-rule: a model of personal rule’, Journal of the European

Economic Association, vol. 2(2-3), pp. 162-92.

Acemoglu, D., D. Ticchi, and A. Vindigni (2011). ‘Emergence and persistence of ineﬃ-

cient states’, Journal of the European Economic Association, vol. 9(2), pp. 177-208.

Aghion, P., T. Persson, and D. Rouzet (2012). ‘Education and military rivalry’, NBER

Working Paper 18049.

Angrist, J. and J.S. Pischke (2009). Mostly Harmless Econometrics, Princeton: Prince-

ton University Press.

Arellano, M. and S. Bond (1991). ‘Some tests of speciﬁcation for panel data: Monte

Carlo evidence and an application to employment equations’, Review of Economic

Studies, vol. 58(2), pp. 277-97.

Attanasio, O., L. Picci, and A. Scorcu (2000). ‘Saving, growth, and investment: a

macroeconomic analysis using a panel of countries’, Review of Economics and Statis-

tics, vol. 82(2), pp. 182-211.

Bai, J. and P. Perron (2003). ‘Computation and analysis of multiple structural change

models’, Journal of Applied Econometrics, vol. 18(1), pp. 1-22.

Baltagi, B. (2005). Econometric Analysis of Panel Data, West Sussex, UK: Wiley and

Sons.

Banerjee, A. and L. Iyer (2005). ‘History, institutions, and economic performance: the

legacy of colonial land tenure systems in India’, American Economic Review, vol.

95(4), pp. 1190-1213.

Barro, R. and X. Sala-i-Martin (2004). Economic Growth, Cambridge, MA: MIT Press.

Barro, R. and J. Urs´ua (2010). ‘Barro-Urs´ua macroeconomic dataset’, http://rbarro.

com/data-sets/ (last accessed: 1 November 2012)

Beck, T., R. Levine, and N. Loayza (2000). ‘Finance and the sources of growth’, Journal

of Financial Economics, vol. 58(1-2), pp. 261-300.

Bates, R. (2001). Prosperity and Violence, New York: Norton.

Bertrand, M., Duﬂo, E., and Mullainathan, S. (2004). ‘How much should we trust

diﬀerence-in-diﬀerences estimates?’ Quarterly Journal of Economics, vol. 119(1),

pp. 249-75.

Besley, T. and T. Persson (2009). ‘The origins of state capacity: property rights,

taxation, and politics’, American Economic Review, vol. 99(4), pp. 1218-44.

Besley, T. and T. Persson (2011). The Pillars of Prosperity, Princeton: Princeton

University Press.

Besley, T. and T. Persson (2013). ‘Taxation and development’, in (A. Auerbach, R.

Chetty, M. Feldstein, and E. Saez, eds.), Handbook of Public Economics, pp. 51-110,

Amsterdam: Elsevier.

Bockstette, V., Chanda, A., and L. Putterman (2002). ‘States and markets: the ad-

vantage of an early start’, Journal of Economic Growth, vol. 7(4), pp. 347-69.

Bogart, D. (2008). ‘Engines of development?’, Working paper, University of California,

Irvine.

Bond, S., A. Leblebicioglu, and F. Schiantarelli (2010). ‘Capital accumulation and

growth: a new look at the empirical evidence’, Journal of Applied Econometrics,

vol. 25(7), pp. 1073-99.

Bonney, R. (1995). Economic Systems and State Finance, Oxford: Oxford University

Press.

Brewer, J. (1989). The Sinews of Power, London: Unwin Hyman.

Brosio, G. and C. Marchese (1986). Il potere di spendere, Bologna: Il Mulino.

Bruno, G. (2005). ‘Approximating the bias of the LSDV estimator for dynamic unbal-

anced panel data models’, Economic Letters, vol. 87(3), pp. 361-66.

Cameron, A., J. Gelbach, and D. Miller (2008). ‘Bootstrap-based improvements for

inference with clustered standard errors’, Review of Economics and Statistics, vol.

90(3), pp. 482-96.

Cameron, A. and P. Trivedi (2010). Microeconometrics Using Stata, College Station:

Stata Press.

Carreras, A. and X. Tafunell (2006). Estad´ısticas Hist´oricas de Espa˜na, Madrid: Fun-

dacin BBVA.

Comin, D., W. Easterly, and E. Gong (2010). ‘Was the wealth of nations determined

in 1000 BC?’, American Economic Journal: Macroeconomics, vol. 2(3), pp. 65-97.

Cox, G. (2011). ‘War, moral hazard, and ministerial responsibility: England after the

Glorious Revolution’, Journal of Economic History, vol. 71(1), pp. 133-61.

De Long, B. and A. Shleifer (1993). ‘Princes and merchants: European city growth

before the Industrial Revolution’, Journal of Law and Economics, vol. 36(3), pp.

671-702.

Dell, M. (2010). ‘The persistent eﬀects of Peru’s mining mita’, Econometrica, vol. 78(6),

pp. 1863-1903.

Dincecco, M. (2009). ‘Fiscal centralization, limited government, and public revenues in

Europe, 1650-1913’, Journal of Economic History, vol. 69(1), pp. 48-103.

Dincecco, M. (2011). Political Transformations and Public Finances, Cambridge, UK:

Cambridge University Press.

Dincecco, M., G. Federico, and A. Vindigni (2011). ‘Warfare, taxation, and political

change: evidence from the Italian Risorgimento’, Journal of Economic History, vol.

71(4), pp. 887-914.

Dincecco, M. and M. Prado (2012). ‘Warfare, ﬁscal capacity, and performance’, Journal

of Economic Growth, vol. 17(3), pp. 171-203.

Donald, S. and K. Lang (2007). ‘Inference with diﬀerence-in-diﬀerences and other panel

data’, Review of Economics and Statistics, vol. 89(2), pp. 221-33.

Driscoll, J. and A. Kraay (1998). ‘Consistent covariance matrix estimation with spa-

tially dependent data’, Review of Economics and Statistics, vol. 80(4), pp. 549-60.

Durlauf, S., P. Johnson, and J. Temple (2005). ‘Growth econometrics’, in (P. Aghion

and S. Durlauf, eds.), Handbook of Economic Growth, pp. 555-677, Amsterdam:

Elsevier.

Elster, J. (2000). Ulysses Unbound, Cambridge, UK: Cambridge University Press.

Engerman, S. and K. Sokoloﬀ (1997). ‘Factor endowments, institutions, and diﬀeren-

tial paths of growth among New World economies’, in (S. Haber, ed.), How Latin

America Fell Behind, pp. 260-304, Palo Alto: Stanford University Press.

Epstein, S. (2000). Freedom and Growth, London: Routledge.

Felloni, G. (1959). ‘Le entrate degli Stati Sabaudi dal 1825 al 1860’, Archivio economico

dell’Uniﬁcazione Italiana, volumi 34, fascicolo 2.

Fontvieille, L. (1976). ´

Evolution et croissance de l’ ´

Etat francais, 1815-1969, Paris:

Institut des Sciences Mathmatiques et Economiques Appliques.

Gemmell, N., R. Kneller, and I. Sanz (2011). ‘The timing and persistence of ﬁscal policy

impacts on growth: evidence from OECD countries’, ECONOMIC JOURNAL, vol.

121(550), pp. F33-58.

Gennaioli, N. and H.J. Voth (2012). ‘State capacity and military conﬂict’, Working

paper, Pompeu Fabra University.

Gerschenkron, A. (1966). Economic Backwardness in Historical Perspective, Cam-

bridge, MA: Belknap Press.

Giavazzi, F. and G. Tabellini (2005). ‘Economic and political liberalizations’, Journal

of Monetary Economics, vol. 52(7), pp. 1297-330.

Glaeser, E., R. La Porta, F. Lopez-de-Silanes, and A. Shleifer (1999). ‘Do institutions

cause growth?’, Journal of Economic Growth, vol. 9(3), pp. 271-303.

Greif, A. and G. Tabellini (2010). ‘Cultural and institutional bifurcation: China and

Europe compared’, American Economic Review Papers and Proceedings, vol. 100(2),

pp. 135-40.

Hall, R. and C. Jones (1999). ‘Why do some countries produce so much more output

per worker than others?’, Quarterly Journal of Economics, vol. 114(1), pp. 83-116.

Herbst, J. (2000). States and Power in Africa, Princeton: Princeton University Press.

Hintze, O. (1906). ‘Military organization and the organization of the state’, in (F.

Gilbert, ed.), The Historical Essays of Otto Hintze [1975], pp. 178-215, New York:

Oxford University Press.

Hobson, J. (1993). ‘The military-extraction gap and the wary titan: the ﬁscal-sociology

of British defence policy, 1870-1913’, Journal of European Economic History, vol.

22(3), pp. 466-507.

Hoﬀman, P. (2012). ‘Why was it Europeans who conquered the world?’, Journal of

Economic History, vol. 72(3), pp. 601-33.

Hoﬀman, P. and J.L. Rosenthal (1997). ‘The political economy of warfare and taxation

in early modern Europe: historical lessons for economic development’, in (J. Drobak

and J. Nye, eds.), The Frontiers of the New Institutional Economics, pp. 31-55, St.

Louis: Academic Press.

Im, K., M. Pesaran, and Y. Shin (2003). ‘Testing for unit roots in heterogeneous panels’,

Journal of Econometrics, vol. 115(1), pp. 53-74.

Islam, N. (1995). ‘Growth empirics: a panel data approach’, Quarterly Journal of

Economics, vol. 110(4), pp. 1127-70.

Jaggers, K. and M. Marshall (2008). ‘Polity IV project: political regime characteristics

and transitions, 1800-2007’, Center for Systematic Peace, George Mason University.

Jones, C. (1995). ‘Time series tests of endogenous growth models’, Quarterly Journal

of Economics, vol. 110(2), pp. 495-525.

Judson, R. and A. Owen (1999). ‘Estimating dynamic panel data models: a guide for

macroeconomists’, Economic Letters, vol. 65(1), pp. 9-15.

Kaldor, N. (1963). ‘Taxation and economic development’, Journal of Modern African

Studies, vol. 1(1), pp. 7-23.

Kang, D. (2002). Crony Capitalism, Cambridge, UK: Cambridge University Press.

Karaman, K. and S¸. Pamuk (2010). ‘Ottoman state ﬁnances in comparative European

perspective, 1500-1914’, Journal of Economic History, vol. 70(3), pp. 593-629.

Krantz, O. and Sch¨on, L. (2010). Swedish Historical National Accounts, 1800-2000.

Department of Economic History, Lund University.

La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vishny (1998). ‘Law and ﬁnance’,

Journal of Political Economy, vol. 106(6), pp. 1123-55.

Laporte, A. and F. Windmeijer (2005). ‘Estimation of panel data models with binary

indicators when treatment eﬀects are not constant over time’, Economic Letters, vol.

88(3), pp. 389-96.

Levi, M. (1988). Of Rule and Revenue, Berkeley: University of California Press.

Levin, A., C. Lin, and C. Chu (2002). ‘Unit root tests in panel data: asymptotic and

ﬁnite sample properties’, Journal of Econometrics, vol. 108(1), pp. 1-24.

Lindert, P. (2004). Growing Public, Cambridge, UK: Cambridge University Press.

Maddala, G. and S. Wu (1999). ‘A comparative study of unit root tests with panel data

and a new simple test’, Oxford Bulletin of Economics and Statistics, vol. 61(S1), pp.

631-52.

Maddison, A. (2010). ‘Statistics on world population, GDP, and per capita GDP, 1-

2008 AD’, http://www.ggdc.net/MADDISON (last accessed: 1 November 2012)

Magnusson, L. (2009). Nation, State, and the Industrial Revolution, London: Rout-

ledge.

Mann, M. (1986). ‘The autonomous power of the state: its origins, mechanisms, and

results’, in (J. Hall, ed.), States in History, pp. 109-36, Oxford: Oxford University

Press.

Mata, M. (1993). As ﬁnan¸cas p´ublicas portuguesas da Regenera¸c˜ao `a Primeira Guerra

Mundial, Lisbon: Bank de Portugal.

Mata, M. and N. Val´erio (2001). ‘As ﬁnan¸cas constitucionais entre duas guerras civis

(1833-1845)’, Revista de Hist´oria Econ´omica e Social, vol. 1, pp. 135-44.

Mathias, P. and P. O’Brien (1976). ‘Taxation in Britain and France, 1715-1810: a

comparison of the social and economic consequences of taxes collected by central

governments’, Journal of European Economic History, vol. 5(3), pp. 601-50.

Michalopoulos, S. and E. Papaioannou (2013). ‘Pre-colonial ethnic institutions and

contemporary African development’, Econometrica, vol. 81(1), pp. 113-52.

Migdal, J. (1988). Strong Societies and Weak States, Princeton: Princeton University

Press.

Mitchell, B. (1988). British Historical Statistics, Cambridge, UK: Cambridge University

Press.

Mitchell, B. (2003). International Historical Statistics: Europe, 1750-2000, New York:

Palgrave Macmillan.

Mokyr, J. (1998). ‘The second Industrial Revolution, 1870-1914’, in (V. Castronono,

ed.), Storia dell’economia Mondiale, pp. 219-45, Rome: Laterza.

Mokyr, J. (2008). ‘The institutional origins of the Industrial Revolution’, in (E. Help-

man, ed.), Institutions and Economic Performance, pp. 64-119, Cambridge, MA:

Harvard University Press.

Nickell, S. (1981). ‘Biases in dynamic models with ﬁxed eﬀects’, Econometrica, vol.

49(6), pp. 1417-26.

North, D. and B. Weingast (1989). ‘Constitutions and commitment: the evolution

of institutions governing public choice in seventeenth-century England’, Journal of

Economic History, vol. 49(4), pp. 803-32.

Nunn, N. (2009). ‘The importance of history for economic development’, Annual Review

of Economics, vol. 1(1), pp. 65-92.

O’Brien, P. (2011). ‘The nature and historical evolution of an exceptional ﬁscal state

and its possible signiﬁcance for the precocious commercialization and industrializa-

tion of the British economy from Cromwell to Nelson’, Economic History Review,

vol. 64(2), pp. 408-46.

Pammer, M. (2010). ‘Public ﬁnance in Austria-Hungary, 1820-1913’, in (J. Cardoso and

P. Lains, eds.), Paying for the Liberal State, pp. 132-161, Cambridge, UK: Cambridge

University Press.

Papaioannou, E. and G. Siourounis (2008). ‘Democratisation and growth’, ECO-

NOMIC JOURNAL, vol. 118(532), pp. 1520-51.

Pesaran M., and R. Smith (1995). ‘Estimating long-Run relationships from dynamic

heterogeneous panels’, Journal of Econometrics, vol. 68(1), pp. 79-113.

Pincus, S. and J. Robinson (2010). ‘What really happened during the Glorious Revo-

lution?’, Working paper, Harvard University.

Prescott, E. (2004). ‘Why do Americans work so much more than Europeans?’, Federal

Reserve Bank of Minneapolis Quarterly Review, vol. 28(1), pp. 2-13.

Rodrik, D. (2005). ‘Growth strategies’, in (P. Aghion and S. Durlauf, eds.), Handbook

of Economic Growth, pp. 967-1014, Amsterdam: North-Holland.

Rosenthal, J.L. and R.B. Wong (2011). Before and Beyond Divergence, Cambridge,

MA: Harvard University Press.

Russell, P. (2004). Constitutional Odyssey, Toronto: University of Toronto Press.

Schumpeter, J. (1918). ‘The crisis of the tax state’, International Economic Papers,

vol. 4, pp. 5-38.

Sen, A. (1999). Development as Freedom, New York: Anchor Books.

Silveira, L. (1987). ‘Aspectos da evolu¸c˜ao das ﬁnan¸cas p´ublicas portuguesas nas

primeiras d´ecadas do s´eculo XIX (1800-27)’, An´alise Social, vol. 97(2), pp. 505-29.

Singer, D. (1987). ‘Reconstructing the correlates of war dataset on material capabilities

of states, 1816-1985’, International Interactions, vol. 14(2), pp. 115-32.

Spoerer, M. (2010). ‘The evolution of public ﬁnances in nineteenth-century Germany’,

in (J. Cardoso and P. Lains, eds.), Paying for the Liberal State, pp. 103-31, Cam-

bridge, UK: Cambridge University Press.

Stasavage, D. (2003). Public Debt and the Birth of the Democratic State, Cambridge,

UK: Cambridge University Press.

Stasavage, D. (2005). ‘Cities, constitutions, and sovereign borrowing in Europe, 1274-

1785’, International Organization, vol. 61(3), pp. 489-526.

Stasavage, D. (2013). ‘Was Weber right? City autonomy, political oligarchy, and the

rise of Europe’, Forthcoming, American Political Science Review.

Strayer, J. (1970). On the Medieval Origins of the Modern State, Princeton: Princeton

University Press.

Tilly, R. (1966). ‘The political economy of public ﬁnance and the industrialization of

Prussia, 1815-66’, Journal of Economic History, vol. 26(4), pp. 484-97.

Tilly, C. (1990). Coercion, Capital, and European States, 990-1990, Cambridge, UK:

Blackwell.

Uﬃcio Storico (1980). ‘L’esercito italiano’.

van Zanden, J.L. (1996). ‘The development of government ﬁnances in a chaotic period,

1807-50’, Economic and Social History of the Netherlands, vol. 7, pp. 53-71.

van Zanden, J.L. and A. van Riel (2010). ‘The development of public ﬁnance in the

Netherlands, 1815-1914’, in (J. Cardoso and P. Lains, eds.), Paying for the Liberal

State, pp. 57-80, Cambridge, UK: Cambridge University Press.

Wade, R. (1990). Governing the Market, Princeton: Princeton University Press.

Wooldridge, J. (2003). Introductory Econometrics, Mason, OH: Thomson South-

Western.

Ziblatt, D. (2006). Structuring the State, Princeton: Princeton University Press.

Table 1

Dates of Fiscal centralisation

Year Event

England 1066 Establishment of uniform rule after Norman Conquest

France 1790 Major administrative reforms during French Revolution

Belgium 1795 Major administrative reforms after French annexation

Piedmont 1802 Major administrative reforms after French annexation

Netherlands 1806 Major administrative reforms under French control

Prussia 1806 Major administrative reforms after French defeat in battle

Spain 1845 Major administrative reforms after Moderate Coup of 1843

Austria 1848 Major administrative reforms during Year of Revolutions

Portugal 1859 Major administrative reforms after Revolutionary Era

Sweden 1861 Abolition of traditional tax system

Denmark 1903 Abolition of traditional tax system

Data source and notes. Dincecco (2011). See text for deﬁnition of ﬁscal centralisation.

Table 2

Dates of Limited Government

Year Event

Netherlands 1572 Establishment of Dutch Republic (1572-1795) after revolt from Spain

1848 Implementation of new constitution during Year of Revolutions

England 1688 Establishment of constitutional monarchy during Glorious Revolution

Belgium 1831 Founded as constitutional monarchy after Revolution of 1830

Piedmont 1848 Establishment of constitutional monarchy during Year of Revolutions

Prussia 1848 Establishment of constitutional monarchy during Year of Revolutions

Denmark 1848 Establishment of constitutional monarchy during Year of Revolutions

Portugal 1851 Establishment of constitutional monarchy after Revolutionary Era

Sweden 1866 Introduction of bicameral legislature

Austria 1867 Establishment of constitutional monarchy after defeat by Prussia

France 1870 Formation of constitutional regime during war with Prussia

Spain 1876 Establishment of constitutional monarchy after civil war

Data source and notes. Dincecco (2011). See text for deﬁnition of limited government.

Table 3

Political Transformations and Economic Performance, 1650-1913

(1) (2) (3) (4) (5) (6) (7)

Dependent Variable is Real Per Capita GDP Growth

Fiscal centralisation 0.657 0.222 0.268 0.166 0.249 0.292 0.344

(0.087) (0.086) (0.125) (0.094) (0.099) (0.116) (0.171)

(0.000) (0.041) (0.033) (0.137) (0.030) (0.063) (0.063)

Limited government 0.321 0.053 -0.028 -0.102 -0.108 0.049 -0.080

(0.109) (0.165) (0.146) (0.230) (0.113) (0.225) (0.205)

(0.027) (0.757) (0.837) (0.728) (0.385) (0.836) (0.738)

Fiscal centralisation * Limited government 0.232 0.311

(0.239) (0.126)

(0.433) (0.010)

Lag (1) per capita GDP growth -0.185 -0.195

(0.094) (0.094)

Lag (2) per capita GDP growth -0.168 -0.178

(0.056) (0.053)

R-squared 0.049 0.202 0.207 0.202 0.055 0.244 0.254

Observations 1,772 1,772 1,772 1,772 1,772 1,750 1,750

Notes. Estimation method is OLS. All speciﬁcations include country ﬁxed eﬀects. Columns 2-4 and 6-7 include

year ﬁxed eﬀects and Column 5 includes a linear time trend. Additionally, Columns 3 and 7 include country-speciﬁc

time trends. Cluster-robust standard errors clustered by country in parentheses (ﬁrst line below coeﬃcients) and

p-values for two-sided Wald tests computed according to Cameron et al. (2008)’s wild bootstrap-t procedure also

in parentheses (second line below coeﬃcients for Fiscal centralisation, Limited government, and interaction term).

Table 4

Political Transformations and Economic Performance, 1650-1913: Robustness

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Dependent Variable is Real Per Capita GDP Growth

Fiscal centralisation 0.206 0.246 0.316 0.337 0.216 0.429 0.220 0.272 0.377

(0.096) (0.132) (0.101) (0.170) (0.170) (0.097) (0.073) (0.080) (0.130)

(0.064) (0.086) (0.015) (0.084) (0.238) (0.011) (0.000) (0.000) (0.000)

Limited government 0.080 -0.002 0.020 -0.068 0.045 0.138 -0.055 -0.043 -0.159

(0.175) (0.142) (0.035) (0.198) (0.231) (0.152) (0.191) (0.172) (0.374)

(0.644) (0.983) (0.467) (0.760) (0.845) (0.360) (0.758) (0.807) (0.673)

External conﬂict dummy (lagged) 0.264 0.271 0.072 0.189 0.160 -0.003 -0.066 -0.218 0.310

(0.279) (0.279) (0.054) (0.214) (0.226) (0.106) (0.084) (0.114) (0.450)

Internal conﬂict dummy (lagged) 0.030 0.101 -0.429 -0.038 -0.117 0.144 -0.252 -0.320 -0.534

(0.239) (0.250) (0.120) (0.249) (0.207) (0.169) (0.425) (0.186) (0.411)

Population growth (lagged) 0.008 0.009 -0.000 0.003 0.001 -0.002 0.003 0.029 -0.017

(0.008) (0.007) (0.001) (0.006) (0.006) (0.003) (0.015) (0.028) (0.095)

Lag (1) per capita GDP growth -0.193 -0.185 -0.177

(0.093) (0.092) (0.096)

Lag (2) per capita GDP growth -0.177 -0.170 -0.139

(0.054) (0.058) (0.053)

Lag (3) log per capita GDP -0.612 0.223

(0.832) (0.420)

R-squared 0.205 0.211 0.208 0.256 0.247 0.096 0.425 0.626 0.720

Observations 1,757 1,757 1,027 1,746 1,746 1,746 350 173 65

Notes. Estimation method is OLS. All speciﬁcations include country ﬁxed eﬀects. All speciﬁcations include year ﬁxed

eﬀects except for Column 6, which includes a linear time trend. Additionally, Columns 2-4 include country-speciﬁc time

trends. Column 3 restricts the data to before 1845 and Columns 7-9 use 5-, 10-, and 25-year data averages, respectively.

Cluster-robust standard errors clustered by country in parentheses (ﬁrst line below coeﬃcients) and p-values for two-sided

Wald tests computed according to Cameron et al. (2008)’s wild bootstrap-t procedure also in parentheses (second line below

coeﬃcients for Fiscal centralisation and Limited government).

Table 5

Political Transformations and Extractive Capacity, 1650-1913

(1) (2) (3) (4) (5) (6) (7) (8)

Dependent Variable is Per Capita Revenue Growth

Fiscal centralisation 1.405 1.467 2.929 3.254 6.610 3.770 3.923 4.268

(0.497) (0.531) (0.926) (0.738) (2.526) (1.251) (0.907) (1.081)

(0.001) (0.007) (0.039) (0.025) (0.108) (0.028) (0.006) (0.035)

Limited government 0.438 0.047 0.717 1.215 1.883 1.209 1.601 5.117

(0.564) (0.638) (0.652) (0.631) (0.758) (0.989) (0.807) (2.791)

(0.466) (0.973) (0.348) (0.120) (0.017) (0.254) (0.080) (0.236)

External conﬂict dummy (lagged) -0.774 -0.777 -1.257 0.620 0.957 -0.540 1.526

(1.237) (1.115) (0.508) (1.180) (1.887) (0.692) (1.796)

Internal conﬂict dummy (lagged) 2.697 3.008 1.854 4.190 2.106 1.262 0.499

(0.858) (0.962) (1.587) (1.518) (1.131) (1.823) (2.495)

Population growth (lagged) -0.191 -0.192 -0.064 -0.226 -0.274 -0.134 -0.256

(0.110) (0.110) (0.090) (0.056) (0.171) (0.149) (0.162)

Lag (1) per capita revenue growth -0.173 -0.169 -0.227

(0.026) (0.028) (0.030)

Lag (2) per capita revenue growth -0.132 -0.128 -0.201

(0.049) (0.044) (0.057)

Lag (3) log per capita revenues -11.401

(2.405)

R-squared 0.160 0.162 0.165 0.008 0.176 0.197 0.045 0.235

Observations 1,760 1,748 1,748 1,748 1,019 1,734 1,734 1,734

Notes. Estimation method is OLS. All speciﬁcations include country ﬁxed eﬀects. All speciﬁcations include

year ﬁxed eﬀects except for Columns 4 and 7, which include linear time trends. Additionally, Columns 3-8 include

country-speciﬁc time trends. The sample period is 1650-1845 for Column 5. Cluster-robust standard errors clustered

by country in parentheses (ﬁrst line below coeﬃcients) and p-values for two-sided Wald tests computed according

to Cameron et al. (2008)’s wild bootstrap-t procedure also in parentheses (second line below coeﬃcients for Fiscal

centralisation and Limited government).

Table 6

Alternative State Capacity Measures, 1816-1913

(1) (2) (3) (4) (5) (6) (7)

Dependent Variable is State Capacity Growth Measure

Non-Mil Non-Mil Mil Edu Edu RR RR

Exps Exps Exps Exps Exps Miles Miles

Fiscal centralisation -4.606 -0.429 0.046 -0.006 -0.117 1.224 2.718

(3.405) (3.715) (0.029) (0.068) (0.116) (1.393) (0.920)

(0.417) (0.927) (0.180) (0.877) (0.384) (0.357) (0.128)

Limited government 1.194 7.069 -0.066 0.081 0.138 2.480 4.927

(5.418) (2.933) (0.042) (0.016) (0.011) (4.512) (4.255)

(0.866) (0.139) (0.157) (0.000) (0.000) (0.556) (0.531)

External conﬂict dummy (lagged) -1.070 2.860 -0.149 -0.062 -0.007 -5.338 -0.604

(3.084) (3.748) (0.064) (0.030) (0.045) (3.746) (1.730)

Internal conﬂict dummy (lagged) 3.800 0.036 0.030 -0.040 -0.045 1.784 3.069

(1.893) (2.225) (0.075) (0.070) (0.070) (5.488) (4.778)

Population growth (lagged) 0.423 -0.050 -0.003 -0.011 0.006 -3.221 0.221

(0.264) (0.061) (0.003) (0.047) (0.058) (3.885) (1.081)

Lag (1) dependent variable log level -54.445

(33.094)

Lag (1) dependent variable -0.496 -0.370 -0.202

(0.054) (0.103) (0.060)

Lag (2) dependent variable -0.294 -0.061 -0.134

(0.044) (0.051) (0.033)

Lag (3) dependent variable log level -32.955 -28.164

(3.745) (5.003)

R-squared 0.186 0.348 0.200 0.463 0.515 0.495 0.517

Observations 724 694 728 330 320 435 413

Notes. Estimation method is OLS. All ﬁscal variables are in per capita terms. All speciﬁcations include country

and year ﬁxed eﬀects and country-speciﬁc time trends. The sample period is 1870-1913 for Columns 6 and 7.

Cluster-robust standard errors clustered by country in parentheses (ﬁrst line below coeﬃcients) and p-values

for two-sided Wald tests computed according to Cameron et al. (2008)’s wild bootstrap-t procedure also in

parentheses (second line below coeﬃcients for Fiscal centralisation and Limited government).

Table 7

State Capacity and Economic Performance, 1650-1913

(1) (2) (3) (4) (5) (6)

Dependent Variable is Real Per Capita GDP Growth

Lag (1) per capita revenue growth -0.003 -0.004 -0.003 -0.003 -0.003

(0.002) (0.002) (0.002) (0.002) (0.002)

Lag (2) per capita revenue growth 0.001 0.001 0.001 -0.001 0.001

(0.002) (0.002) (0.002) (0.002) (0.002)

Lag (3) log per capita revenues 0.112 0.117 0.154 0.184 0.159

(0.031) (0.033) (0.064) (0.089) (0.073)

(0.000) (0.000) (0.000) (0.029) (0.023)

Lag (1) per capita non-mil exp growth -0.007

(0.005)

Lag (2) log per capita non-mil exp growth 0.000

(0.008)

Lag (3) log per capita non-mil exps -0.032

(0.380)

(0.880)

Fiscal centralisation 0.260

(0.130)

(0.059)

Limited government -0.043

(0.131)

(0.718)

External conﬂict dummy (lagged) 0.295 0.296 0.216 0.287 0.897

(0.300) (0.298) (0.247) (0.285) (0.880)

Internal conﬂict dummy (lagged) 0.037 0.124 0.000 0.111 0.853

(0.263) (0.285) (0.266) (0.267) (0.284)

Population growth (lagged) -0.000 0.001 0.001 0.001 0.025

(0.006) (0.006) (0.006) (0.006) (0.024)

Lag (1) per capita GDP growth -0.193

(0.092)

Lag (2) per capita GDP growth -0.175

(0.054)

R-squared 0.204 0.205 0.211 0.255 0.211 0.206

Observations 1,736 1,736 1,736 1,736 1,736 699

Notes. Estimation method is OLS. All speciﬁcations include country and year ﬁxed eﬀects. Additionally,

Columns 3-6 include country-speciﬁc time trends. The sample period is 1816-1913 for Column 6. Robust

standard errors clustered by country in parentheses (ﬁrst line below coeﬃcients) and p-values for two-

sided Wald tests computed according to Cameron et al. (2008)’s wild bootstrap-t procedure also in

parentheses (second line below coeﬃcients for Lag (3) Log per capita revenues, Fiscal centralisation,

Limited government, and Lag (3) Log per capita non-military expenditures).

Table 8

Placebo Tests

(1) (2) (3) (4)

Panel A: Dependent Variable is Real Per Capita GDP Growth

25 yrs prior 50 yrs prior 75 yrs prior 100 yrs prior

Static Model

Fiscal centralisation (placebo) 0.007 -0.158 -0.079 -0.037

(0.132) (0.106) (0.074) (0.092)

Limited government (placebo) -0.048 -0.037 -0.035 -0.045

(0.173) (0.107) (0.086) (0.075)

Observations 1,757 1,757 1,757 1,757

Dynamic Model

Fiscal centralisation (placebo) 0.012 -0.188 -0.099 -0.047

(0.163) (0.135) (0.096) (0.127)

Limited government (placebo) -0.018 -0.056 -0.043 -0.071

(0.237) (0.126) (0.103) (0.090)

Observations 1,746 1,746 1,746 1,746

Panel B: Dependent Variable is Per Capita Revenue Growth

25 yrs prior 50 yrs prior 75 yrs prior 100 yrs prior

Static Model

Fiscal centralisation (placebo) -0.232 -0.727 -0.340 -1.868

(0.879) (0.950) (0.812) (0.938)

Limited government (placebo) 0.164 -0.536 -1.034 -0.598

(0.619) (0.937) (1.489) (1.357)

Observations 1,748 1,748 1,748 1,748

Dynamic Model

Fiscal centralisation (placebo) 0.070 -1.025 -0.569 -2.207

(0.720) (1.172) (1.096) (1.367)

Limited government (placebo) 0.671 -0.689 -0.917 -0.471

(1.138) (1.177) (1.819) (2.011)

Observations 1,734 1,734 1,734 1,734

Panel C: Dependent Variable is Per Capita Non-Military Expenditure Growth

25 yrs prior 30 yrs prior 35 yrs prior 40 yrs prior

Static Model

Fiscal centralisation (placebo) 1.018 0.684 3.709 8.720

(2.086) (2.502) (5.273) (12.342)

Limited government (placebo) -2.378 -4.855 -5.861 -3.916

(4.203) (6.402) (8.632) (10.660)

Observations 724 724 724 724

Dynamic Model

Fiscal centralisation (placebo) -2.025 -1.696 -3.142 -5.846

(2.871) (2.892) (5.214) (21.222)

Limited government (placebo) -0.922 -4.036 -4.369 2.116

(3.376) (5.904) (7.541) (6.964)

Observations 694 694 694 694

Notes. Estimation method is OLS. All speciﬁcations include country and year ﬁxed eﬀects

and time-varying controls. The sample period is 1650-1913 for Panels A and B and 1816-1913

for Panel C. Robust standard errors clustered by country in parentheses.

●

Years around political transformation (T=0)

Per capita GDP growth (%)

−1 0 1 2

−50 −25 0 25 50

●

Years around political transformation (T=0)

Per capita GDP growth (%)

−1 0 1 2

−50 −25 0 25 50

●

●●

●

Years around political transformation (T=0)

Per capita GDP growth (%)

−0.5 0 0.5 1

−10 −5 0 5 10

●

Years around political transformation (T=0)

Per capita GDP growth (%)

−1 0 1 2

−10 −5 0 5 10

Fig. 1. Time-Demeaned Average Real Per Capita GDP Growth Around Political Transformations

Notes. Top panel shows 50 years before and after and bottom panel shows 10 years before and after. Left

panel shows ﬁscal centralisation and right panel shows limited government. Circles represent average per

capita GDP growth rates across sample countries. Solid lines represent locally-weighted regression curves

ﬁtted to data. Dashed horizontal lines represent pre-transformation and post-transformation average per

capita GDP growth rates.

●

−1.25 −0.75 −0.25 0.25 0.75 1.25

~1C

~2C

~3C

~4C

●

−1.25 −0.75 −0.25 0.25 0.75 1.25

~1L

~2L

~3L

~4L

Fig. 2. Time-Varying Relationship Between Political Transformations and Real Per Capita GDP Growth

Notes. Left panel shows ﬁscal centralisation and right panel shows limited government. Dots correspond to

point estimates for coeﬃcients of pulse dummies and solid lines represent 90% conﬁdence intervals (Equation

2). e

i,t,e

i,t = 1 for years 6 to 10 before political transformations; e

i,t,e

i,t = 1 for years 1 to 5 before;

i,t,e

i,t = 1 for years 0 to 4 after (including transformation year itself); e

i,t,e

i,t = 1 for years 5 to 9 after;

and e

i,t,e

i,t = 1 from 10th year post-transformation onward.

●

0123

−2 2

α1,0

●

●●

●

●●●

●

●●●

●

0 5 10 15 20

−4 −2 2 4

α1,0

Fig. 3. Conditional Correlations Between Real Per Capita GDP Growth and “Future” Political Transforma-

tions

Notes. Dots represent point estimates for α1,0, the coeﬃcient for Ci,t in Equation 4. Circles are point

estimates for α1,τ ,τ= 1, . . . , q. Solid lines represent 90% conﬁdence intervals.

MENA Region Remains at Crossroads: A Panel Analysis Modelling Political Stability And Economic Freedom To Determine Economic Growth

Article

Full-text available

Aug 2023

This paper seeks to revisit the theoretical and empirical literature to assess the MENA region's economic growth indicators after more than 10 years of the Arab Spring events and it aims to examine how political stability and higher scores of economic freedoms affect economic growth in 14 MENA region countries over a time span of 20 years from 2001-2021. The main econometric estimation uses the Arellano-Bover/Blundell-Bond model via a Generalized Method of Moments (GMM) estimator to address the endogeneity and reverse causality existing between political and economic freedom measures and economic growth. The main findings detect a positive relationship between political stability and economic growth across robust estimations, whereas a positive relationship was confirmed for most of the Economic Freedom Index components to impact economic growth.

MENA region remains at crossroads: A panel analysis modelling political stability and economic freedom to determine economic growth

Article

Full-text available

Jul 2023

Local institutions and human capital formation in pre-industrial societies: evidence from Valencia

Article

Full-text available

Nov 2023

This article analyses the relationship between institutions and human capital formation. We use literacy rates in 1860 at the municipal level in Valencia where the expulsion of the Moriscos in 1609 was followed by the Christian resettlement. Our findings show that male literacy was consistently lower in Morisco areas by mid-19th century. Yet, the analysis also shows the disappearance of this effect at the beginning of the 20th century. We argue that the deployment of the liberal state would have entailed the gradual decoupling of educational outcomes from the institutional heterogeneity characteristic of the Old Regime.

Reconfiguring essential and discretionary public goods

Article

Full-text available

Sep 2023

When is state coercion for the provision of public goods justified? And how should the social surplus of public goods be distributed? Philosophers approach these questions by distinguishing between essential and discretionary public goods. This article explains the intractability of this distinction, and presents two upshots. First, if governments provide configurations of public goods that simultaneously serve essential and discretionary purposes, the scope for justifiable complaints by honest holdouts is narrower than commonly assumed. Second, however, claims to distributive fairness in the provision of public goods also turn out to be more complex to assess.

Administrative capacity and local level development in metropolitan, municipal and district assemblies in Ghana: mediating role of resource capacity

Article

Apr 2023

Purpose This study aims to examine the mediating role of resource capacity on the effect of administrative capacity on local level development. Design/methodology/approach Data is obtained from three Metropolitan, Municipal and District Assemblies (MMDAs) in the Central Region of Ghana using a survey. A structural equation modelling that includes a formatively measured construct (administrative capacity) is used in the analyses. Findings Leadership, human resource management, financial resource management, technical and adaptive capacities are supported as formative constructs of administrative capacity. More so, it is established that administrative capacity has an effect on resource capacity. Also, administrative capacity and resource capacity each has an effect on local level development. The mediation analysis shows that resource capacity mediates the effect administrative capacity has on local level development. Practical implications Administrative capacity is a necessity for local governments as enhancing it together with the resource capacity of MMDAs are vital if local level development is to be achieved. In this regard, administrative and resource capacities of local governments must be pursued by both the central government and decentralised governments for the latter to achieve its primary mandate. Originality/value To the best of our knowledge, this is the first study to empirically ascertain leadership, human resource management, financial resource management, technical and adaptive capacities as the dimensions of administrative capacity using a structural equation model. Given that there is a lack of consensus on the dimensions of administrative capacity, this study contributes significantly to the discourse. The study also highlights the need to pursue and enhance administrative and resource capacities for local level development.

Factors involved in Economic Development and their Impact on State-Building in Malaysia

Article

Dec 2022

The successful history of poverty elimination and impressive economic development in Malaysia for the post-colonial period regarded a worth model for growth in developing world. The question raised in this achievement was; how economic development affected on the state-building in the country? Therefore, the motive of the study is to examine the determinants of economic development on state-building. Autoregressive distributed lag (ARDL) model employed to test economic factors affecting the state-building. The study covered data spanning from 1970-2017. The long-run co-integration in the model has confirmed, and econometric diagnostic tests are robust. The study findings show that human capital, oil rent, and per capita income, positively and significantly affect state-building in the long-run. In the short-run, human capital and oil revenue effects positively and significant. Population growth, democracy, inflation, and financial development have confirmed a negative and significant impact on state-building. Policy implication has highlighted the importance of human capital development, macroeconomic stability, and population growth in Malaysia's state-building. Demographic transition exerts a considerable negative effect on state-building in the long-run, affecting the economic health and steady-state growth. Financial development inserting a negative impact on state-building to promote development. The study recommended that the economic policymakers and relevant bureaucracies correct the fiscal deficit and ease domestic financial institutions into the credits based on the findings. Malaysia is facing a demographic threat in the long run due to the aging population and low fertility rate; therefore, it have to re-address the population policy.

It takes two to Tango: The joint effect of democracy and fiscal capacity on economic growth in Indonesia

Article

Full-text available

Dec 2022

The literature continues to debate the effects of democracy and fiscal capacity on economic growth, both partially and jointly. To remedy the literature puzzle, this study examines the economic growth effects of democracy and fiscal capacity in 34 Indonesian provinces from 2016 to 2021. Using a fixed-effect model, this study documents no evidence of a partial effect; rather, it finds a joint effect of democracy and fiscal capacity on Indonesian economic growth. These findings remain relatively robust even when provincial heterogeneity, COVID-19 pandemic shocks, and sectoral composition are factored into the model. This finding indicates that regions with democracy and strong fiscal capacity possess relatively fast per capita GRDP growth. Based on these findings, the study concludes that democracy and fiscal capacity should exist side by side. Indonesia's sub-national economic growth strategy, like a tango game, requires reforming two types of decentralization: political decentralization to improve the quality of democracy that upholds the merit system and fiscal decentralization to expand local tax capacity to finance public goods productively.

Firms, policies, informality, and the labor market

Preprint

Full-text available

Nov 2023

This paper studies how taxes and regulations affect firms and labor market outcomes in the developing world. We build a general equilibrium model of firm dynamics subject to search frictions, taxes, and imperfectly enforced legislation. This setup leads to informal employment along the intensive and extensive margin. Estimated to match firm and worker-level data from Peru, the model sheds light on the effect of corporate income tax on informality and unemployment. On the one hand, a reduction in corporate taxes concentrates employment in larger and more productive firms, increasing efficiency and reallocating workers to formal jobs. On the other hand, employment shifts to a smaller mass of firms creating higher unemployment duration and higher income inequality. Holding tax revenues constant, we compare two simulated reforms: a reduction in corporate income tax and an equivalent reduction in payroll taxes. We find that contracting corporate income taxes can achieve 0.9% higher output gains, 1 p.p. higher formal employment, and a 1.3 p.p. lower unemployment rate. A cut in payroll taxes generates instead lower and more unequally distributed output gains. A revenue budget-neutral welfare-maximizing policy shifts the burden from corporate income to payroll taxes, reducing informality by 2.2 p.p. and increasing output per capita by 2.4%.

State Capacity and Economic Growth: Cautionary Tales from History

Article

Feb 2023
Natl Inst Econ Rev

Sheilagh Catheren Ogilvie

This paper uses economic history to probe the relationship between state capacity and economic growth during the Great and Little Divergences (c.1500–c.1850). It identifies flaws in the dominant measure of state capacity, fiscal capacity, and advocates instead analysing state expenditures. It investigates five key activities on which states historically spent resources: waging war; providing law and administration; building infrastructure; pursuing industrial policy; and fostering a national culture. The lesson of history, it concludes, is not to build a capacious state. Rather, we need a state that uses its capacity to help (or at least not hinder) market activity.

Terminology Matters: A Review on the Concept of Economic Predation

Article

Dec 2022

This article compares various disciplines to show that the concept of predation is associated with notions that are relevant to the economic domain, such as equilibrium, risks, and hidden societal costs. We review the economic literature in some of the top-ranked economic journals on the subject of predation dating back a hundred years. Our analysis shows that economic academics use the concept of predation differently and at times in an erroneous or overly narrow way. This article argues that economic theory must take into consideration predator-prey dynamics during one-on-one interactions between sellers acting as predators of sorts by using information asymmetry to their advantage, and their prey in the form of naïve clients, who undergo economic/financial harm as a result. We propose that a better understanding of the concept of predation may enhance the drafting of government policies. To our knowledge, this is the first article that examines the concept of predation in its original sense across the literature and across decades and that demonstrates its relevance in economic theory.

As Finanças Constitucionais Portuguesas entre as duas Guerras Civis (1833-1845)

Article

Full-text available

Jun 2001

The Portuguese Constitutional Public Finance and civil wars (1833-1845)

Introduction The Long Road To The Industrial Revolution

Chapter

Jan 2009

Jan Luiten van Zanden

The Institutional Origins of the Industrial Revolution

Chapter

Nov 2008

JOEL MOKYR

Governing the Market: Economic Theory and the Role of Government in East Asian Industrialization

Book

Dec 1990

Robert Wade

The Economic Effects of Constitutions

Book

Aug 2003

The authors of The Economic Effects of Constitutions use econometric tools to study what they call the "missing link" between constitutional systems and economic policy; the book is an uncompromisingly empirical sequel to their previous theoretical analysis of economic policy. Taking recent theoretical work as a point of departure, they ask which theoretical findings are supported and which are contradicted by the facts. The results are based on comparisons of political institutions across countries or time, in a large sample of contemporary democracies. They find that presidential/parliamentary and majoritarian/proportional dichotomies influence several economic variables: presidential regimes induce smaller public sectors, and proportional elections lead to greater and less targeted government spending and larger budget deficits. Moreover, the details of the electoral system (such as district magnitude and ballot structure) influence corruption and structural policies toward economic growth.Persson and Tabellini's goal is to draw conclusions about the causal effects of constitutions on policy outcomes. But since constitutions are not randomly assigned to countries, how the constitutional system was selected in the first place must be taken into account. This raises challenging methodological problems, which are addressed in the book. The study is therefore important not only in its findings but also in establishing a methodology for empirical analysis in the field of comparative politics.

The Formation of National States in Western Europe

Article

Jun 1976
West Polit Q

Railways and the Economic Development of Western Europe, 1830–1914

Book

Jan 1983

Patrick O’Brien

The Persistent Effects of Peru's Mining Mita

Article

Jan 2010

Melissa Dell

This study utilizes regression discontinuity to examine the long-run impacts of the mita, an extensive forced mining labor system in effect in Peru and Bolivia between 1573 and 1812. Results indicate that a mita effect lowers household consumption by around 25% and increases the prevalence of stunted growth in children by around six percentage points in subjected districts today. Using data from the Spanish Empire and Peruvian Republic to trace channels of institutional persistence, I show that the mita's influence has persisted through its impacts on land tenure and public goods provision. Mita districts historically had fewer large landowners and lower educational attainment. Today, they are less integrated into road networks, and their residents are substantially more likely to be subsistence farmers.

The autonomous power of the state: its origins, mechanisms and results.

Article

Jan 1986

M. Mann

Specifies the origins, mechanisms and results of the autonomous power the state possesses in relation to the major power grouping of 'civil society'. The state is first defined, and two aspects of that definition, centrality and territoriality, are discussed in relation to 'despotic' and 'infrastructural' state power. Argues that the state is essentially an arena, a place, and that this is precisely the origin and mechanism of its autonomous powers. -after Author

Economic Backwardness in Historical Perspective

Article

Jul 1963

State Capacity and Long‐Run Economic Performance

Abstract and Figures

Recommended publications

Unlocking Pakistan's Revenue Potential

The Rise of Effective States in Europe

The Rise of Effective States in Europe

Warfare, Fiscal Capacity, and Performance

The Budgetary Origins of Fiscal-Military Prowess