Per capita income and institutions: an
empirical analysis
Renda per capita e instituições: uma
análise empírica
Submission date: 1 may 2024
Approval date: 12 august 2024
1Doutor em Economia pelo CEDEPLAR (UFMG); Bolsista de Produtividade em Pesquisa do CNPq; Associate Professor at UFJF;
Pesquisador do Laboratório de Análises Territoriais e Setoriais - LATES; Vice coordenador do curso de Ciências Econômicas
(UFJF), Juiz de Fora – MG. Correo electrónico: lucianofg@gmail.com. ORCID: https://orcid.org/0000-0002-8321-3022.
Mestre em economia pela UFV, Economista pela UFV, Pesquisador do Departamento de Economia da UFV. Correo
electrónico: andre.werberich@ufv.br. ORCID: https://orcid.org/0009-0008-7837-4615.
2
1
Luciano Ferreira Gabriel
André Dannemann Werberich da Silva 2
Luciano Dias de Carvalho 3
Ele possui Mestrado e Doutorado em Desenvolvimento Econômico pela Universidade Federal do Paraná-UFPR. Atualmente,
é professor adjunto na Universidade Federal de Viçosa-UFV. Foi coordenador do curso de Ciências Econômicas de 2015 a
2018. Tem experiência em Economia, com ênfase em Teoria Macroeconômica Keynesiana e Desenvolvimento Econômico,
atuando principalmente nos seguintes temas: Crescimento Econômico, Macroeconomia Pós-Keynesiana e Sistemas
Dinâmicos Não Lineares. Professor associado na UFV. Correo electrónico: lucianodc@gmail.com. ORCID:
https://orcid.org/0000-0002-9031-4497.
3
Abstract
The main objective of this article is to empirically analyze how political and economic
institutions quality influence the economic development of countries. For this purpose,
a sample of 151 countries in the period from 2000 to 2020 was used for panel data
estimations. The main results suggest that political and legal institutions that suffer from
greater influence of the armed forces as well as the low quality of laws concerning
property rights have a more intense and negative impact on per capita income of
developing countries. In addition, it was verified that the greater income inequality, by
benefiting a few society sectors with greater political representation, negatively
influences the per capita income level.
Keywords: political institutions, economic institutions, per capita income.
JEL codes: C23; E02; O10.
Resumen
El objetivo principal de este artículo es analizar empíricamente cómo la calidad de las
instituciones políticas y económicas influye en el desarrollo económico de los países.
Para este propósito, se utilizó una muestra de 151 países en el período 2000-2020
para estimaciones con datos de panel. Los principales resultados sugieren que las
instituciones políticas y jurídicas que sufren una mayor influencia de las fuerzas
armadas así como la baja calidad de las leyes relativas a los derechos de propiedad
tienen un impacto más intenso y negativo en el ingreso per cápita de los países en
desarrollo. Además, se verificó que la mayor desigualdad de ingresos, al beneficiar a
pocos sectores de la sociedad con mayor representación política, influye
negativamente en el nivel de ingreso per cápita.
Palabras clave: instituciones políticas, instituciones económicas, ingreso per cápita.
2
1. Introduction
According to the institutionalist approach of Douglass North, political and economic
institutions exist to reduce the uncertainties resulting from the process of human
interaction, that is, they dictate laws, contracts, codes, and norms of conduct, both
formal and informal, that govern societies and, thus, being able to determine and
influence, through production, transaction and transformation costs, incentives for
innovation and technological development. In this way, institutions are considered by
the New Institutionalists to be the fundamental factors that determine, through
intermediary channels of influence, the growth trajectory of countries and their per
capita income. The differences observed in economic development processes are a
direct consequence of the different forms of institutional organization in which societies
are structured (AREND, CARIO, and ENDERLE, 2012).
According to Acemoglu and Robinson (2012), the process of economic
development is a consequence of the historical trajectory of transformation of its
political and economic institutions. Therefore, the differences observed in the levels of
GDP and per capita income are a consequence of the different institutional structures
existing in each country. These authors propose that to understand the different
trajectories of growth and economic development observed throughout history, it is first
necessary to understand the historical course that accompanied the development of
societies and how the evolution of their different institutions influenced the growth and
development of their economies.
3
From Acemoglu and Robinson (2012), it can be seen that the discrepancies
observed in the economic growth trajectories of countries throughout history do not
exclusively result from differences in the rates of capital and labor accumulation as
presented by endogenous growth models, but from the existing asymmetry in the
institutional structures adopted by the countries. Furthermore, the authors demonstrate
that the existence of self-sustained growth paths is directly associated with the
consolidation of more inclusive political and economic institutional structures.
Inclusive institutions can be characterized as a set of institutions that allow and
ensure the existence and maintenance of individual rights and freedoms and that
enable and encourage the participation of a large part of society in the political and
economic systems. In addition, due to their nature of freedom and incentives, they allow
the consolidation of plural market structures that stimulate investment in human capital
and the development of new technologies, which, in turn, allows countries to reach self-
sustaining paths of growth economic. Conversely, extractive institutions are institutions
that allow only a small portion of society, usually an elite, to have access to the
country´s political system and participate in and enjoy its economic activities, thus
generating a strong disincentive to development (Acemoglu and Robinson, 2012).
Despite the econometric evidence supporting the results found by Acemoglu and
Robinson (2012) that there is a relationship between the quality of institutions and
economic growth, there is no consensus on the way (or the channels) in which the
4
quality of institutions influences this process. Thus, this paper seeks to analyze how
the quality of political and economic institutions influences per capita income from a
heterogeneous sample of countries from a new set of variables and controls.
The results of this work shed more light on the source of the conflicting findings in
the literature because the impact of political regimes on per capita income cannot be
understood solely in terms of a broad-brush distinction between democratic and non-
democratic regimes.
In addition to this brief introduction, section 2 presents the interrelation between
institutions, inequality, and per capita income level based on the empirical literature.
Furthermore, it presents an empirical discussion on how the quality of political and
economic institutions and their spillover channels can impact a country´s economic
development trajectory. Section 3 presents the database, and the empirical
specification of the model, and section 4 addresses the results and discussion. Finally,
section 5 presents the final considerations.
2. Institutions, Inequality, and Per Capita Income Level
There is a great number of empirical works that confirm the existence of the
interrelationship between the quality of institutions in a country and its level of economic
growth and per capita income, such as Barro (1996), Tavares and Wacziarg (2001),
Rivera-Batiz (2002), Ali and Crain (2002) and Acemoglu, Johnson, Robinson , and
Yared (2008), among others. In particular, Barro (1996) proposes one of the first works
5
in the area that seeks to analyze the impact of a series of variables, among them
democracy (used as a proxy to measure the quality of political institutions) on the
economic growth trajectory of a country. As a result, the author shows that democracy,
measured in terms of the degree of political rights of the population, has a positive
effect on the economic growth of a country.
One of the first works to address the relationship between inequality and
economic growth was proposed by Glaeser, Scheinkman, and Shleifer (2003).
According to this research, there is a negative relationship between inequality,
measured through the Gini index, and economic growth, measured in terms of per
capita income growth because in highly unequal societies there is the possibility of
subversion of the institutional apparatus in favor of extractive elites.
The main conclusion of Glaeser, Scheinkman, and Shleifer (2003) explains that
in societies where a small share of the population is rich enough, the economic elite
can use their wealth to subvert, through mechanisms such as bribes and donations to
fund, political campaigns, the judiciary, legislative and executive systems for their
benefit. As a consequence of institutional fragility, an environment is created in which
fundamental rights, such as property rights, are no longer guaranteed, which
discourages investment levels in physical capital and, consequently, the country´s
economic growth.
6
A more sizable literature looks at the effects of democracy on redistribution and
inequality and is reviewed and extended in Acemoglu et al. (2015). As redistribution is
better and inequality decreases, the political power of an extractive elite reduces, in the
same vein as Glaeser, Scheinkman, and Shleifer (2003). In this case the political power
is composed of political power from resource distribution, since groups that have
financial resources have greater ease in solving their collective problems and imposing
their will on society.
Tavares and Wacziarg (2001), Glaeser, Scheinkman, and Shleifer (2003),
Gradstein (2007) and Acemoglu and Robinson (2012), among others, explain that
extractive regimes are formed, regardless of the political regime adopted, by low-quality
institutions in which a small share of the population, an elite, uses its privileges to
subvert institutions for its benefit and extract, through mechanisms such as rent-
seeking, the wealth and income of the rest of the population. As a consequence, there
is a significant increase in inequality indexes in society, in which a highly wealthy elite
enriches at the expense of an impoverished society, in addition to a strong disincentive
to investment in physical capital and technological development, which, in turn,
prevents these countries from reaching self-sustaining trajectories of economic growth.
Ali and Crain (2002) propose one of the first works to separately analyze the
impact of political and economic institutions on a country´s economic growth trajectory.
The choice of separating the variables aims to deepen the discussion on the role of
institutions in economic development and propose a methodological alternative to the
7
current models. The authors’ main argument for this resides in the fact that works that
investigate only the relationship between economic growth and the type of government,
that is, the level of political rights of society, such as, for example, the works of Tavares
and Wacziarg (2001) and Rivera-Batiz (2002) do not provide conclusive and robust
results on the subject.
As a result, Ali and Crain (2002) show that only the variable used to measure
the quality of economic institutions can impact per capita income growth. According to
the authors, these results can be explained by the inability of proxies that take into
account only the type of government (that is, the levels of political rights of a society)
to determine the impact of political institutions on economic growth because other
political factors directly impact the development trajectory of countries such as, for
example, the ability of governments to adopt (regardless of the type of the political
regime) economic institutions with a greater or lesser degree of institutional quality.
Acemoglu, Johnson, Robinson, and Yared (2008) find similar results to those of
Ali and Crain (2002), by showing, through the use of a cross-section model with fixed
effects, that there is no causal relationship between changes in income level of a
country and changes in its political institutions, measured in terms of the level of political
rights of the population. Acemoglu et al. (2008) show, however, that there is a direct
relationship between the quality of political institutions in terms of their influence on
economic institutions on per capita economic growth.
8
At the same time, Acemoglu et al. (2008) suggest that this relationship is
because more inclusive political institutions, that is, with higher levels of institutional
quality (such as in full democracy), allow the emergence of economic institutions
capable of fostering the economic growth of a country. Therefore, the political
institutions would only be able to impact economic growth when associated with
changes in the quality of the country´s economic institutions. On the other hand, the
military´s involvement in politics, for example, even at a peripheral level, decreases
institutional accountability and may privilege a small share of society. Over the long
term, a system of military government, a full autocracy, or a deficient democracy will
certainly diminish effective governmental functioning or create an uneasy environment
for national and foreign entrepreneurs, which negatively influence per capita income
level and growth (Bacha, 2023).
Recently, Acemoglu et al. (2019) examined 184 countries from 1960 to 2010,
which moved between political regimes. For this sample of countries, there were 122
cases of democratization and 71 of reversals to authoritarianism (theocracies or
autocracies with military intervention). They found that countries that moved to
democratic regimes experienced 20% gains in GDP over 25 years, compared to what
would have happened had they remained autocratic
1
. Moreover, their results suggest
that democracy increases future GDP by encouraging investment, increasing
1
Acemoglu et al. (2019)´s empirical strategy rely on a dichotomous measure of democracy
coded from several sources to reduce measurement error and controls for country-fixed effects
and the rich dynamics of GDP, which otherwise confound the effect of democracy on economic
growth.
9
schooling, inducing economic reforms, improving public good provision, and reducing
social unrest.
In Young and Sheehan´s (2014) view, institutional quality is an essential
ingredient for economic growth. The authors note that institutional quality is one
channel through which aid flows may affect economic growth. Specifically, their
study provides evidence on the different dimensions of institutional quality as
likely channels through which aid affects growth. A consensus is that weak
institutional infrastructure is a fundamental constraint on countries ability to
accumulate productive factors (e.g. physical and human capital) and to innovate
and adopt new technology (North, 1990).
Put more clearly, weak institutions lead to expropriation activities because
of a lack of proper checks and balances mechanisms on political power, judicial
manipulation, entry barriers to new entrepreneurs and technologies, corruption,
and inefficient bureaucracy (Slesman et al., 2015).
In contrast to the popular claims that democracy is bad for growth at early stages
of economic development, Acemoglu et al. (2019) find no heterogeneity by level of
income. There is some heterogeneity depending on the level of human capital, but,
according to them, these effects are not large enough to lead to negative effects of
democracy for countries with low human capital.
10
Acemoglu et al. (2019) find evidence that democratizations take place in regional
waves: a country is more likely to transition to democracy or nondemocracy when the
same transition recently occurred in other countries in the same region. We exploit this
source of variation to identify the effect of democracy on GDP. Using regional waves
as an instrument for democracy, we corroborate our finding that democracy increases
GDP.
When taking into count municipalities Nakabashi et al. (2013) findings
suggest that an increase by one point in the average quality of the institutions
can increase the average GDP per capita by around 20 percent. This means that
each point of increase in the quality of the municipality institutions can increase
the municipality´s GDP per capita by R$1,000 (around US$600). Furthermore,
according to Nakabashi et al. (2013) institutional quality seems to be more
essential in greater municipalities. One potential explanation for this result is that
informal institutions matter in small municipalities because people know each
other, while in bigger cities formal institutions have a more important role. On the
contrary, human capital is more important in small ones (Nakabashi et al., 2013).
According to Bacha (2023) twelve countries grew by 7% or more yearly after the
Second World War for at least 25 years. Common features of these countries were the
following: (i) they were fully connected economies by foreign trade; (ii) they maintained
11
macroeconomic stability; (iii) they generated high savings and investment rates; (iv)
they allowed markets to allocate resources, and (v) had governments committed,
credible and capable, but not necessarily democratic. Considering the classification of
the Matrix of Democracy of the University of Würzburg, seven of these countries are
now democratic: Botswana, South Korea, Indonesia, Japan, Malaysia, Malta, and
Taiwan; three have hybrid regimes: Singapore, Hong Kong, and Thailand; and two are
autocracies: China and Oman.
Bacha (2022) identified twelve countries that in the postwar period made the
transition from middle-income to high-income considering institutional features. These
countries are Singapore, South Korea, Hong Kong, and Taiwan, Israel, Spain, Greece,
Ireland and Portugal, Australia, Norway, and New Zealand. A common feature to them
is its high degree of openness to foreign trade, a medium or small population (5 to 50
million inhabitants), and low inequality of income distribution. Except for Singapore and
Hong Kong, which are hybrid regimes, none of the other ten countries are autocracies.
Based on the above discussion, one can conclude that: i) the relation between
democracy, economic growth, and per capita income is not linear or direct; ii) the
interrelationship among political institutions and economic institutions matters for
catching up; and iii) different proxy variables on political and economic institution can
result in different conclusions. Therefore, in the next section , two distinct sets of
variables were used to quantify the isolated effect of the quality of political and
12
economic institutions to shed more light on what qualitative dimensions of them may
influence per capita income level.
3. Database and Empirical Specification
Based on the discussion in the previous sections and to determine how institutions can
affect the per capita income of a heterogeneous sample of countries, a log-linear
specification for panel data was used.
The use of a log-linear model allows to test how the per capita income is
econometrically influenced in terms of the political and economic institutional proxies
over time. In this way, the following model is specified:
(1)
in which . The s
are the parameters to be estimated for each group of independent variables, explained
below. The dependent variable is the the per capita income level of each
country i in the analyzed period t in terms of its natural logarithm; is the set of
variables that capture the quality of economic institutions in each country; is the
set of variables that capture the quality of political institutions for each country; are
the control variables; is the specific effect of time; captures the unobserved effects
of each country i that are time-invariant and is the idiosyncratic error term.
13
The proxy variables that seek to determine the quality of the Political Institutions
(PI) are Gini Index —gini—, military intervention —militaryinter— and the fulfillment of
contracts —legalenforce.
The first variable, the Gini index, is an indicator that determines, on a scale from
0 to 1, how far the income distribution is from an egalitarian condition, that is, the index
shows how unequal a country is. Income inequality depends mainly on the actions of
the political class in the approval of laws related to regressive/progressive taxation,
provision of public goods, such as, for example, educational institutions, wages in the
public sector, and the role of unions in the private sector, this index measures, even if
indirectly, how the political classes act in the sense of guaranteeing less extractivism
(or less social unrest of workers) about income distribution.
Thus, the inclusion of the Gini index to measure the quality of political institutions
makes it possible to determine whether a country´s economic growth results in
improvements in society´s levels of well-being or whether this effect is captured by an
extractive elite that subverts political institutions and enriches itself at the expense of
the rest of society.
The second variable called militaryinter —military interference in the rule of law
and politics— is an indicator that measures, on a scale from 0 to 10, the level of
involvement in politics of the armed forces of a given country. The indicator considers
that, as armed forces officers are not elected through universal suffrage, any level of
14
political involvement negatively impacts a country´s political freedom and may, in the
long run, affect the level of international trust, and the full functioning of the government
and increase levels of corruption. Thus, the lower (higher) the political involvement of
the armed forces, the higher (lower) the country scores. The variable is constructed
based on information from the International Country Risk Guide.
The third and final variable used to determine the quality of the Political
Institutions (IP) is the enforcement of contracts —Legal Enforcement of Contracts—,
which is an indicator that varies from 0 to 10 and seeks to measure the time and
associated costs to be able to collect a debt through the use of the judicial system. The
variable is constructed based on the aggregation of two different subcomponents, th e
first measuring the time spent between the opening of the process until the moment of
payment of the debt and the second, the financial costs related to the process.
The variables used to determine the quality of Economic Institutions (IE) in a
country are the opening of new businesses —scoresb— and property rights —
proprights. The first variable scoresb —starting a business score— is an indicator,
which varies from 0 to 100, and seeks to measure the amount of time and cost to open
a new business in a given nation. Countries that require more time and/or greater
capital investment receive lower scores. The variable is constructed based on five
different World Bank indicators, the first measuring the number of procedures required
to open a new business, the second the time, measured in days, and the third the
monetary costs involved in opening. The fourth indicator measures reforms in
15
legislation related to opening new businesses and the last one measures the cost of
the minimum wage in the country.
The second and last variable used to determine how inclusive the economic
institutions of a country are is called proprights —protection of property rights—, it is an
indicator that measures, on a scale from 0 to 10, the quality of laws and the institutions
that protect and secure property rights in a given country, where the higher the score,
the greater the protective quality of laws and institutions.
Based on the empirical literature discussed in section 2 and 3 several control
variables were used. The following controls were included in the model: gross physical
capital formation as a GDP share —fbkf—, which measures the increase in physical
capital and inventories in a given country each year; inflation —infla— which represents
the inflation rate of a country, per year, through the percentage variation of the average
costs of acquiring a certain basket of products and services; government spending as
a GDP share —govexp—, which determines how much government spending was
allocated to purchase goods and services; trade openness as a percentage of GDP —
openness— which measures yearly the total imports and exports of a given country
divided by GDP, the technological gap —techgap—, which measures the existing
technological gap between a given country and the technological frontier (USA),
following the methodology of Verspagen (1993); and the population —pop—, which
represents the total population of a given country in each year in thousands of
16
inhabitants, thus controlling the results based on a proxy for the size of the country. All
variables used, abbreviations, and sources are shown in table 1
2
.
Table 1.Variables used to measure the quality of political and economic institutions
and as a control
Variable
Abbreviation
Meaning
Source
Real per capita
income (in n.
logarithm)
logpcGDP
Real per capita income (i.e., in
constant 2015 US dollar)
Penn World
Table 10.0
Gini index
Gini
The Gini index measures the area
between the Lorenz curve and a
hypothetical line of absolute equality,
expressed as a percentage of the
maximum area under the line. Thus, a
Gini index of 0 represents perfect
equality, while an index of 100 implies
perfect inequality.
World Bank
Military
interference
militaryinter
Measures the level of involvement of
the armed forces in politics and rule of
law. The index ranges from 0 to 10.
Economic
Freedom of the
World Index -
International
Country Risk
Guide – Fraser
Institute
Fulfillment of
contracts
legalenforce
It measures the time and costs to
collect a debt through the court
system. The index ranges from 0 to
10.
Economic
Freedom of the
World Index -
Fraser Institute
Starting a New
Business
Scoresb
It measures the time and cost of
opening a new business in a country.
The index ranges from 0 to 100.
World Bank
Property rights
proprights
It measures the quality of laws and
institutions that protect and secure
property rights. The index ranges from
0 to 10.
Economic
Freedom of the
World Index -
Fraser Institute
2
In addition to data from the Fraser Institute, World Bank, and Penn World Tables, were
considered the democracy indexes consolidated by the Polity Project and The Freedom House
as possible proxies to measure the quality of political institutions. However, none of the
indicators showed enough variability to be used in econometric estimations. In addition, they
had higher data missing problems.
17
Gross physical
capital
formation
Fbkf
Country´s level of investment as a
GDP share.
World
Development
Indicators
Inflation
Infla
Represents the inflation rate of a
country. It is measured by the year-
over-year percentage change.
World
Development
Indicators
Government
spending
Govexp
Measures government spending on
the consumption of goods and
services. It is measured as a
percentage of GDP.
World
Development
Indicators
Degree of
international
trade
openness
Openness
Total Imports and Exports in relation
to GDP.
World
Development
Indicators
Technological
gap
Techgap
Technological gap (G) between
countries following Verspargen (1993)
methodology, i.e,
, where
aproduc is the overall average work
productivity for each country.
Own elaboration
based on Penn
World Table
10.0 data
Population
Pop
Measures the total population of a
country. It is measured in numbers of
inhabitants.
Penn World
Table 10.0
Source: own elaboration.
The final sample of analyzed countries is formed by 151 countries of which 36
are considered, according to the classification of the World Bank, developed
3
and 115
non-developed
4
and comprises the period from 2000 to 2020 (longer period available).
3
Australia, Austria, Belgium, Bulgaria, Canada, Croatia, Cyprus, Czech Republic, Denmark,
Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Japan, Latvia,
Lithuania, Luxembourg, Malta, Netherlands, New Zealand, Norway, Poland, Portugal,
Romania, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, United Kingdom, United
States.
4
Albania, Algeria, Angola, Argentina, Armenia, Azerbaijan, Bahamas, Bahrain, Bangladesh,
Barbados, Belarus, Belize, Benin, Bhutan, Bolivia, Bosnia and Herzegovina, Botswana, Brazil,
Burkina Faso, Burundi, Cape Verde, Cambodia, Cameroon, Chad, Chile, China, Colombia,
Congo, Dem. Rep., Congo, Rep., Costa Rica, Cote d'Ivoire, Dominican Republic, Ecuador,
Egypt, Arab Rep., El Salvador, Eswatini, Ethiopia, Fiji, Gabon, Gambia, Georgia, Ghana,
Guatemala, Guinea, Guinea- Bissau, Guyana, Haiti, Honduras, Hong Kong, India, Indonesia,
Iran, Islamic Rep., Iraq, Israel, Jamaica, Jordan, Kazakhstan, Kenya, Korea, Rep., Kuwait,
Kyrgyz Republic, Lao PDR, Lebanon, Lesotho, Liberia, Madagascar, Malawi, Malaysia, Mali,
Mauritania, Mauritius, Mexico, Moldova, Mongolia, Montenegro, Morocco, Mozambique,
18
4. Results and Discussion
It is necessary to define which estimator is the most suitable for the econometric
specification of equation (1). One way to compare random effects estimates with fixed
effects estimates is by using the Hausman test. In the context of this work, it is possible
to observe that the Hausman test has the value of , with value
. Therefore, the null hypothesis is rejected, with the fixed effects model
being the most appropriate for both the group of developed countries and the group of
developing countries.
Applying the Wooldridge test for autocorrelation in panel data for the complete
model given by equation (1), the null hypothesis of no first-order autocorrelation was
rejected with 5% statistical significance. Furthermore, the modified Wald test for
groupwise heteroskedasticity in the fixed effect regression model rejected the null
hypothesis of homoskedasticity with 5% statistical significance. Additionally, it was
applied the Collin test for multicollinearity in the complete model. Usually, individual
variance inflation factor (VIF) greater than 10 should be inspected and average VIF
greater than 6 suggests caution and search for correctional procedures. Indeed,
individual VIF was less than 1.70 and average VIF was less than 2.05 in all
specifications.
Myanmar, Namibia, Nepal, Nicaragua, Niger, Nigeria, North Macedonia, Oman, Pakistan,
Panama, Papua New Guinea, Paraguay, Peru, Philippines, Russian Federation, Rwanda,
Saudi Arabia, Senegal, Serbia, Seychelles, Sierra Leone, Singapore, Sri Lanka, Syrian Arab
Republic, Tajikistan, Tanzania, Thailand, Trinidad and Tobago, Tunisia, Turkey, Uganda,
Ukraine, United Arab Emirates, Uruguay, Venezuela, RB, Vietnam, Yemen, Rep., Zambia,
Zimbabwe.
19
Given these test results, one of the most common ways to correct the
heteroscedasticity of the errors that is consistent with the existence of correlation in the
data is through the incorporation, in the model, of robust standard errors. According to
Hoechle (2007), one of the most used methods for correcting this type of violation of
assumptions is through the application of the Generalized Method of Moments (GMM).
It is important, however, to observe that although the GMM method produces robust
estimators, they do not take into account the effects caused by the correlation of cross-
sectional-spatial-samples. Thus, this type of method is based on the assumption that
the residuals are correlated between the elements of the same sampling unit, but not
between cross-sections, which can reduce the inference capacity of the model.
One of the first methodological attempts to simultaneously include, in the
analysis, the effects of temporal and spatial correlations are through the FGLS
estimator —feasible generalized least-squares— the model, however, tends to
underestimate the standard error of the sample. An alternative to the FGLS estimator,
which corrects the sample underestimation problem, is the use of pooled models with
standard error correction through the PCSE method —panel— corrected standard
errors. However, Hoechle (2007) points out that for short panels, where N > T, both
estimators will be inefficient.
Driscoll and Kraay (1998) propose a non-parametric estimator for the covariance
matrix capable of producing efficient estimators that consider both the effects caused
by temporal and spatial correlation that remain valid for short and long panels with
heteroscedasticity.
20
Table 2 presents the estimation results of model (1) using Driscoll and Kraay
(1998) estimators splitting the sample between developing and developed countries. In
both country samples, the variables related to the quality of economic institutions,
scoresb, and proprights are positive and statistically significant. Taking into account the
complete model, the impact of opening a business and protecting property rights is
stronger in the sample of non-developed countries. The first variable has an impact
1.80 times greater and the second is 6.39 times greater on the per capita income of
this sample of countries.
It can be noticed that while the economic institutions´ quality variables affect
more undeveloped countries when compared to the developed ones, the control
variables associated with the technological gap and degree of economic openness
affect this last group of countries to a greater extent. These results suggest that
international trade and the distance from the technological frontier, captured by the
control variables, affect developed economies more intensely, while the quality of laws
and institutions that protect and ensure property rights, as well as how the time and
cost of starting a new businesses affect non developed countries the most.
These results are in line with what Acemoglu and Robinson (2012) explained,
which shows that economic institutions are only able to influence the growth trajectory
of a given country when they create an environment of stimulation and protection that
enables the promotion of economic activity. Thus, the higher the quality of a country´s
economic institutions, the closer they are to institutions capable of guaranteeing a
higher level of per capita income.
21
The significance of scoresb and proprights both for developed countries and
developing countries is due, according to Acemoglu and Robinson (2012), to the fact
that the existence of inclusive economic institutions is not necessarily conditioned to
the existence of inclusive political institutions, that is, countries with lower institutional
quality can develop stimulus and protection environments that allow the promotion of
economic activity and the development of new technologies. The result is also
corroborated by Ali and Crain (2002) that when analyzing the effects of political and
economic institutions on the growth trajectory of countries, show that the economic
growth and per capita income of a country are independent of the type of government,
but more dependent on the quality of their economic institutions. Therefore, regardless
of the type of government, countries can adopt economic institutions with a greater or
lesser degree of freedom and quality capable of promoting, to a greater or lesser extent,
the economic growth of a country.
Regarding the proxies variables used to measure the quality of political
institutions —gini, militaryinter, and legalenforce— the corrected model shows that for
developing countries the variables gini and militaryinter were statistically significant and
showed an inverse relationship, that is, increases in these variables result in decreases
in per capita income levels of up to 0.664% per year and 1.45% per year, respectively.
For this group of variables in the sample of developed countries, the only variable that
showed significance was legalenforce, which also showed an inverse relationship, that
is, the time and costs associated with recovering liabilities in the judicial system can be
reduced by up to 2.99% per year the per capita income level.
22
For this last variable, the longer the time and associated costs to be able to fulfill
contracts and collect a debt through the use of the judicial system, the scarcer
resources are reallocated outside of productive activity for longer periods, in such a
way that their impact on per capita income is negative. Furthermore, the greater the
financial costs related to the process, the greater its negative effects tend to be.
23
Table 2. Developing and Developed Countries with Driscoll and Kraay standard
errors for the period from 2000 to 2020
Development countries
Developed
(1)
(2)
(3)
(4)
(5)
(6)
logpcGD
P
logpcGD
P
logpcGDP
logpcGDP
logpcGD
P
logpcGD
P
Legalenforce
0.0120 *
-0.0209
-0.0164
-0.0444 **
-0.0391 **
-0.0299 **
(2.59)
(-1.49)
(-1.13)
(-3.31)
(-3.16)
(-3.29)
Militaryinter
-0.00841
-0.0326 ***
-0.0145 *
-0.115 *
-0.0934 *
-0.00539
(-2.05)
(-4.66)
(-2.87)
(-2.31)
(-2.27)
(-0.41)
Scoresb
0.00486 ***
0.00531 ***
0.00317 ***
0.00574 **
0.00666 ***
0.00176 *
(10.67)
(9.31)
(6.28)
(4.05)
(6.24)
(2.52)
Proprights
0.0521 ***
0.0842 ***
0.0568 ***
0.0475 ***
0.0422 **
0.00888 *
(4.84)
(6.78)
(5.82)
(5.53)
(3.53)
(2.57)
Gini
-0.0137 ***
-0.00664 **
0.00332
-0.00192
(-5.41)
(-3.51)
(1.06)
(-1.03)
Fbkf
-0.00114
-0.000634
0.00724 ***
0.00469 **
(-0.78)
(-1.56)
(4.84)
(3.41)
Infla
-0.000443
-0.000150
-0.00665 **
-0.00112
(-1.22)
(-1.43)
(-3.22)
(-0.60)
Govexp
-0.00384
-0.00199
-0.0122 ***
-0.00665 *
(-1.29)
(-0.73)
(-5.17)
(-2.91)
Openness
-0.000901 *
0.00110 *
(-2.32)
(2.81)
Techgap
-0.0282 ***
-0.232 ***
(-11.54)
(-9.12)
Pop
0.0105 **
0.0134 ***
(3.86)
(10.29)
_Cons
7,502 ***
8,332 ***
8,669 ***
10.78 ***
10.48 ***
10.49 ***
(78.23)
(98.94)
(158.47)
(29.33)
(33.52)
(52.02)
No
1474
887
887
739
713
703
Note. t statistics in parentheses; *p < 0.05, **p < 0.01, ***p < 0.001
Source: own elaboration, 2022 (from STATA 15 output).
24
The statistical significance of the variables military intervention —militaryinter—
and the gini index only for developing countries is in line with the literature presented
in sections 2 and 3. For this country group the minimum score of militaryinter was 1.11,
with a standard deviation (average) of 2.31, approximately, between panels
(considering all years). As explained in the last section, the lower (higher) political
involvement of the armed forces, the higher (lower) the country scores. Therefore,
lower scores of militaryinter are negatively associated with per capita income for this
sample of countries.
As developing countries are, for the most part, nations that have low-quality
institutional factors such as the interference of the armed forces in politics and rule of
law, as well as considerable levels of income inequality, can be used as instruments to
subvert political institutions, thus creating institutions that use extractive mechanisms
that benefit a small political and economic elite that enriches itself at the expense of the
rest of society. Since in these societies, there are no consolidated enough checks and
balances that prevent the political system from being used excessively, arbitrarily, or in
a way that benefits an individual or even a small group of society, these institutions
hamper the increase in the level of per capita income.
The statistical significance of the Gini index for this group of countries implies
that where there is not enough institutional maturity, the economic elite can use their
wealth to subvert, through bribes and illegal donations to campaign funds and political
institutions for their own benefit (Glaeser, Scheinkman and Shleifer, 2003). In addition,
according to Gradstein (2007), the general institutional quality of a country is directly
associated with the level of inequality in society, that is, countries with higher levels of
25
inequality are countries with lower institutional quality, since in societies with low levels
of inequality the economic elite cannot use its wealth to subvert, for its benefit, the
political institutions.
Concerning developed countries, the lack of significance of the Gini index (in the
model with all controls) and militaryinter can be explained by the high institutional
quality of this sample of economies, since there are consolidated mechanisms capable
of preventing the subversion of political institutions to benefit just a small share of the
society (Acemoglu and Robinson, 2012)
5
. Thus, unlike what was found for developing
countries, there is no mutual reinforcement between the level of involvement of the
armed forces in politics and the rule of law (which tends to privilege smaller shares of
society) and the level of economic inequalities. In other words, when considering
several important covariates that influence per capita income it is harder to isolate
inequality effects on it
6
.
In order to control for individual unobserved characteristics of the sample that
affect the dependent variable and the possible endogeneity of independent variables
we use the methodology of dynamic panel (GMM) in equation (1) for developing and
developed countries (See Appendix A)
7
. Thereby, we applied the system GMM by
5
It´s noteworthy that the minimum value for militaryinter for developed countries was 7.84, with
a standard deviation of 0.847 (between panels). Therefore, the score is higher and with lower
variability when compared to non-developed countries.
6
It should be said that when time dummies are included for the periods 2007-2009 (subprime
USA crises) and 2019 (SARS-CoV-2 pandemic) both are statistically significant, harming per
capita income for developing and developed countries. Notwithstanding, it is beyond the
objective of this paper to analyze how these crises affected long-term growth trajectories.
7
Endogeneity implies the correlation between the covariates and the error term, that is,
. In the dynamic model will be taking into account logpcGDP lagged effects on the
present, so the conventional method (OLS) to panel data leads to inconsistent estimates since
this variable is correlated with the error term . Moreover, the traditional sources of
26
Arellano and Bover (1995) and Blundell and Bond (1998). This method creates a
system of regressions in difference and level. The instruments of the regressions in the
first difference remain the same as in the GMM difference. The instruments used in the
regressions in level are the lagged differences of the explanatory variables.
When controlling for possible endogeneity scoresb, propright and militaryinter are
statistically significant. These results reinforce what was found before, i.e., economic
institutions are only capable of impacting income levels when associated with the
protection of property rights, which enables the promotion of economic activity and
technological progress.
In all estimations reported in Appendix A, it is not rejected the null hypothesis
that overidentified restrictions are valid at the 1% level of significance. Similarly, it does
not reject the null hypothesis that there is no autocorrelation for higher order.
Furthermore, with the two-step estimations it was obtained efficient and robust
parameters for any standard of heteroscedasticity, whereas Windmeijer´s (2005)
standard errors avoided the downward bias for the standard errors in the estimators.
To make the last results clearer, Figure 1 presents the relationship between per capita
income and the variable militaryinter. The first thing to note is that there is no linearity
between these two variables, which helps in explaining the econometric results. The
second, and more important issue, closer to per capita income of eleven thousand
dollars (approximately) one can observe, respectively, low (closer to 7.2) and high
(closer to zero) military involvement in politics and rule of law. Notwithstanding, after
endogeneity are due to dynamic effects such as cited, simultaneity between variables, omitted
variables, or measurement errors of variables (Greene, 2012).
27
this threshold of income, only countries with greater scores, i.e., lower political
involvement of the armed forces in politics and rule of law achieve higher income.
These results suggest that when analyzing just the two variables, the greater per capita
income is associated with regimes closer to full democracies, in which military forces
are not involved in politics and the rule of law.
Note. Figure 1 results were weighted by all years.
Source: own elaboration (based on STATA 15 output).
Figure 1. Per capita income (in 2015 US dollars) and militaryinter –
2000 – 2020 – broad sample
05000 10000 15000 20000 25000 30000
Per capita income (pcGDP)
0 1 2 3 4 5 6 7 8 9 10
Military intervention in politics and rule of law (militaryinter)
28
Finally, the statistical significance of the variable legalenforce only for the group
of developed countries can be explained by the fact that political institutions only can
influence the per capita income when associated with changes in the quality of
governance, that is, the determination of per capita income level is directly linked to the
quality of effective legislation for the enforcement of contracts by the judicial system,
which is very influenced by the political system. As the quality of the latter is worse in
non-developed countries, according to Rivera-Batiz (2002), this helps in explaining the
lack of statistical significance for this sample of countries.
It is noteworthy that for a different sample of countries and methods, Bacha
(2023) found similar results to the results discussed above. For a sample of 164
countries and data for the year 2019 and using the democracy index from Würzburg
University (Democracy Matrix, 2019)
8
, the author found that only full democracies have
very high per capita income levels
9
. Moreover, there are no moderate autocracies with
very high income levels. The other three types of political regimes classified as deficient
democracies, hybrid regimes, and full autocracies (organized by income levels) present
full autocracies with income levels of around US$10,000. In particular, according to
Bacha (2023, p. 5), there are just three full autocracies with very high per capita income,
all of them oil producers in the Middle East, such as Saudi Arabia, Qatar, and the United
Arab Emirates.
8
This index ranges from zero to one, where lower values indicate autocracies full, followed by
moderate autocracies, hybrid systems, democracies disabled, and, at the highest levels,
functioning democracies.
9
Measured in constant 2019 US dollar.
29
Bacha (2023) considers until US$ 10,000 in per capita income levels, there is
slight variation in the democracy index with values are around 0.5 (representing
disabled democracies, hybrid regimes, and moderate autocracies). However, from per
capita incomes greater than US$10,000 there is a positive relationship between per
capita income and democracy.
Figure 2 presents Gini index and militaryinter scores for a broad sample of
countries. As seen in Figure 1, per capita income greater than US$ 10,000 is just
verified for countries with militaryinter scores greater than 7,2, approximately. Figure 2
shows a negative relation between military influence in politics and the rule of law for
most of the curve. More importantly, increasing equality countries (Gini index < 35) are
associated with less involvement of the armed forces in politics and the rule of law
(militaryinter < 7,2).
30
Note. Figure 1 results were weighted by all years.
Source: own elaboration (based on STATA 15 output).
Figure 2. Gini index and militaryinter score – 2000 – 2020 – broad sample
5. Final Remarks
The main objective of this work was to examine the relationship between the quality of
political and economic institutions and the level of per capita income through the
application of a log-linear panel data model. Unlike other works, new dimensions of
political and economic institutions were incorporated. In this way, it was possible to
determine, in isolation, the real impact of the quality of political and economic
institutions of heterogeneous samples of developed and non-developed countries.
5.5 6 6.5 7 7.5 8 8.5 9 9.5 10
Military intervention in politics and rule of law (militaryinter)
20 30 40 50 60
Gini index
31
As a result, the log-linear panel data model with fixed effects and Driscoll-Kraay
estimator showed that for the variables used to measure the quality of political
institutions —gini, militaryinter, and legalenforce—, for the set of developing countries,
the variables gini and militaryinter were significant and showed a negative relationship.
Thus, the higher (lower) level of economic inequality and the higher (lower) involvement
of the armed forces in politics result in decreases (increases) in the level of per capita
income.
The statistical significance of the gini and militaryinter variables only for
developing countries is in line with the fact that most developing countries have
institutions of low functional quality in terms of governance, transparency, and
accountability, where inequality and greater military involvement in governments can
be used as instruments to subvert political institutions due to the lack of consolidated
containment mechanisms (checks and balances) that prevent this system from being
used arbitrarily, excessively or in a way that benefits a small share of society, as
explained by Glaeser, Scheinkman and Shleifer (2003) and Gradstein (2007).
For the variables used to measure the quality of economic institutions scoresb
and proprights, the results suggest that both for the group of developed countries and
for group of developing countries, both variables are significant and present a positive
relationship, that is, the improvement of the business environment and the guarantee
of property rights positively influence the level of per capita income.
The statistical significance of both variables for both groups of countries shows
that economic institutions are only capable of impacting income levels when associated
with the protection of property rights, which enables the promotion of economic activity
32
and technological progress. The results also suggest that the level of per capita income
depends on the degree of influence of the armed forces on political activity, since more
authoritarian governments can adopt economic institutions with a lower degree of
economic freedom and guarantee benefits only to an extractive elite.
In general terms, and from a broader perspective, the results suggest that the
economic development of a country depends on the existence of an economic
environment that guarantees the existence of incentives and legal protection for
economic activity, or as defined by Acemoglu and Robinson (2012), of inclusive
economic institutions. In the political dimension, institutions have the function of
preventing a share of society from subverting the economic system for their benefit,
increasing inequality.
References
Acemoglu, D.; Johnson, S. Robinson, J.A. and Yared, Pierre. (2008) Income and
Democracy. American Economic Review, Sl, v. 98, no. 3, p. 808-842, Jun. DOI:
10.1257/aer.98.3.808
Acemoglu, D. and Robinson, J. A. (2012). Why Nations Fail: the origins of power,
prosperity, and poverty. New York: Crown Business, 529 p.
Acemoglu, D., Naidu, S., Restrepo, P. and Robinson, J.A. (2015). Democracy,
Redistribution, and Inequality. In: Handbook of Income Distribution, vol. 2B,
edited by Anthony B. Atkinson and François Bourguignon, 1885–1966.
Amsterdam: Elsevier
33
Acemoglu, D., Naidu, S., Restrepo, P. e Robinson, J. (2019). Democracy does cause
growth. Journal of Political Economy, vol. 127, n. 1, pp. 47-100.
https://doi.org/10.1086/700936
Ali, A. M. and Crain, W. M. (2002). Institutional Distortions, Economic Freedom and
Growth. Cato Journal. SL, p. 415-426.
Arellano, M. and Bover, O. (1995). Another look at the instrumental variable estimation
of error-components models. Journal of Econometrics 68: 29–51.
https://doi.org/10.1016/0304-4076(94)01642-D
Arend, M.; Cario, S. Ferraz, A. and Enderle, R. A. (2012). Institutions, Innovations and
Economic Development. Research and Debate, São Paulo, v. 23, no. 1, p. 110-
133, Jan.
Bacha, E. (2022). Fechamento ao comércio e estagnação: por que o Brasil insiste?Em:
Mendes, M. (org.), Para não esquecer: políticas públicas que empobrecem o
Brasil. Rio de Janeiro: Autografia, 2022, pp. 831-853.
Bacha, E. (2023). Democracia e economia. O Instituto de Estudos de Política
Econômica/Casa das Garças (IEPE/CdG). Working Paper (sem número).
Barro, R. (1996) Determinants of Economic Growth: a cross-country empirical study.
Nber Working Paper 5698, [SL], v. 1, no. 1, p. 1-117, Aug. National Bureau of
Economic Research. DOI 10.3386/w5698
Blundell, R. and S. Bond. (1998) Initial conditions and moment restrictions in dynamic
panel data models. Journal of Econometrics 87: 115–143.
https://doi.org/10.1016/S0304-4076(98)00009-8
34
Democracy Matrix (2019). Universidade de Würzburg. Disponível em:
https://www.democracymatrix.com/
Driscoll, J. C.; Kraay, A. C. (1998). Consistent Covariance Matrix Estimation with
Spatially Dependent Panel Data. The Review of Economics Ans Statistics. SL,
p. 549-560. Nov.
Glaeser, E.; Scheinkman, J. and Shleifer, A. (2003). The injustice of inequality. Journal
Of Monetary Economics, SL, v. 50, no. 1, p. 199-222, Jul.
https://doi.org/10.1016/S0304-3932(02)00204-0
Gradstein, M. (2007). Inequality, democracy and the protection of property rights. The
Economic Journal, SL, v. 1, no. 117, p. 252-269, Jan.
https://doi.org/10.1111/j.1468-0297.2007.02010.x
Greene, W. H. Econometric Analysis. 5th ed. Upper Saddle River, NJ: Prentice Hall,
2012.
Hoechle, D. (2007) Robust standard errors for panel regressions with cross-sectional
dependence. The State Journal. SL, p. 281-312.
Nakabashi, L., Pereira, A. E. G. and Sachsida, A. (2013) Institutions and growth: a
developing country case study. Journal of Economic Studies. Vol. 40 No. 5,
2013. pp. 614-634. https://doi.org/10.1108/JES-09-2011-0111
North, D. C. (1990). Institutions, Institutional Change and Economic Performance. Sl:
Cambridge University Press. 152 p.
Rivera-Batiz, F. L. (2002). Democracy, Governance, and Economic Growth: Theory
and Evidence. Review of Development Economics, Sl, v. 2, no. 6, p. 225-247.
Dec. https://doi.org/10.1111/1467-9361.00151
35
Slesman, L., Ahmad, Z. and Wahabuddin, R. (2015). Institutional infrastructure and
economic growth in member countries of the Organization of Islamic
Cooperation (OIC), Economic Modelling, Vol. 51, pp. 214-226.
https://doi.org/10.1016/j.econmod.2015.08.008
Tavares, J. and Wacziarg, R. (2001). How democracy affects growth. European
Economic Review, Sl, v. 45, no. 2001, p. 1341-1378, May.
Verspagen, B. (1993). Uneven Growth Between Interdependent Economies. London:
Averbury.
Windmeijer, F. (2005): “A finite sample correction for the variance of linear
efficient two-step GMM estimators.” Journal of Econometrics 126: 25–51.
Young, A. and Sheehan, K. (2014), “Foreign aid, institutional quality, and
growth”, European Journal of Political Economy, Vol. 36, pp. 195-208.
36
Appendix A
Table A.1. System dynamic panel-data estimation - two steps robust
Developing Countries
Developed Countries
logpcGDP
logpcGDP
l.logpcGDP
0.821***
0.744***
(69.04)
(37.96)
Legalenforce
-0.0105**
-0.00361
(-2.82)
(-1.30)
Militaryinter
-0.00252*
-0.00917*
(-2.95)
(-2.09)
Scoresb
0.0000540*
0.000552*
(2.27)
(2.05)
Proprights
0.0151***
0.00784***
(6.67)
(4.52)
Gini
-0.000982
0.00119
(-1.51)
(1.58)
Fbkf
0.00105***
0.00263***
(4.02)
(7.19)
Infla
-0.00172***
-0.00206***
(-7.23)
(-3.95)
Govexp
-0.00688***
-0.00794***
(-8.52)
(-6.58)
Openness
0.00122***
0.000397***
(8.85)
(5.34)
Techgap
-0.00596***
-0.0741***
(-10.45)
(-11.55)
Pop
0.000719**
0.00123***
(2.80)
(4.47)
_cons
1.623***
2.741***
(15.27)
(12.61)
Arellano and Bond´s test for
AR(1) – A
z = -3.89 Pr > z = 0.000
z = -3.47 Pr > z = 0.001
Arellano and Bond´s test for
z = 0.53 Pr > z = 0.615
z = 0.93 Pr > z = 0.457
37
AR(2) – A
Hansen test of joint validity of
instruments (p-value) - B
chi2(39) = 42.24 Prob >
chi2 = 0.333
chi2(37) = 41.20 Prob >
chi2 = 0.292
Note. t statistics in parentheses; * p < 0.05, ** p < 0.01, *** p < 0.001. Two-step standard errors are
robust to heteroscedasticity (Windmeijer, 2005). The t (s) statistics are in brackets; * p<0.05, ** p<0.01,
*** p<0.001. In A – The null hypothesis: there is no “n” order correlation in the residues. In B – The null
hypothesis: the model is correctly specified and all overidentifications are correct.
Made by the author, 2022 (from STATA 15 output).
