The quality of institutions: A genetic programming approach☆
Introduction
In recent decades, the new institutional economics (NIE) has constituted a program of research that has propelled the return of institutions into the agenda of mainstream economics. The coasean notion of transaction costs (Coase, 1937, Coase, 1960) and the northian notion of institutions (North, 1990) established the foundations for the theoretical framework of the NIE. Political rules, informal norms and enforcement mechanisms constitute the “rules of the game” of a society and these rules establish an incentives structure that affects the level of transaction costs and the efficiency in the economy.
The NIE is a young program that is in a stage of development and it includes some academic debates and controversies, but has already allowed significant advances in different areas such as economic history, economics of organization, law and economics, policy analysis and development economics (Willliamson, 2000, Ménard and Shirley, 2005). The progress of the NIE is generated via a “guerrilla action” (Coase, 1999) that stems from several social sciences, and that was propelled by the award of the Nobel Prize to Ronald Coase in 1991 and to Douglass North in 1993. Since then, the NIE has experienced a growing process in which its analytical abilities have been recognized (Caballero, 2001, Caballero, 2002). Nevertheless, although North (2005) already proposed an extension of the NIE, this program continues requiring efforts in the theoretical and applied work. In fact, debates on definition (for example, the notion of institutions as an equilibrium (Greif, 2006) versus the view of institutions as rules (North, 1990)), methodology and measurement are present, and we need small pieces of work that expand the stock of knowledge on institutions and economy. In this sense, empirical work is the best way to improve this knowledge.
The contribution of institutions in determining income levels around the world has been one of the main programs of empirical research that has been developed since the last decade (Knack and Keefer, 1995, Knack and Keefer, 1997, Mauro, 1995, Hall and Jones, 1999, Rodrik, 1999, Acemoglu et al., 2001, Rodrik et al., 2004). There is now widespread agreement among economists studying economic growth that institutional quality holds the key to prevailing patterns of prosperity around the world (Rodrik, 2004). In this way, economics understands the relevance of analysing the quality of institutions and its determinants.
The study of the quality of institutions includes papers such as La Porta et al. (1999) or Islam and Montenegro (2002). Traditionally, this program of research, which analyzes the effect of a set of variables on the quality of institutions, has adopted a parametric perspective; therefore a specific functional form (usually linear) is assumed and the unknown parameters are later estimated using some optimization procedure as ordinary least square (OLS). The theoretical validity of the model is easily analysed considering the signs of the coefficients, the statistical significance of the parameters estimated and some fit criterion such as the Adjusted R-Square. However, assuming a parametric point of view might cause misspecification problems and, in consequence, originate a bias in the results, a loss of predictive ability and an absence of generalization of the model in the face of new observations.
Nowadays, the great advances done in the field of Computer Science allow us to develop, improve and apply powerful and sophisticated techniques for the estimation and prediction of different phenomena. One of these techniques, called Genetic Programming (GP), is inspired by Genetics and by the Darwinian theories of natural selection and survival (Holland, 1975, Koza, 1992, Mitchell, 2001). The method has already been used satisfactorily in different scientific areas, including economics (Koza, 1995, Beenstock and Szpiro, 2002), finance (Álvarez-Diaz and Álvarez, 2003, Álvarez-Diaz and Álvarez, 2005) and environmental economics (Álvarez-Diaz and Domínguez-Torreiro, 2006). This increasing and intense spread of GP is mainly due to its advantages. Firstly, they do not have any initial restriction on the functional form underlying in the data. Moreover, unlike other methods based on Computer Science, the GP also offers explicitly a mathematical equation which allows a simple ad hoc interpretation of the results. However, as opposed to these advantages, these techniques usually have the difficulty of being computationally intensive and the construction of confidence intervals and hypothesis contrasts is not trivial.
This paper studies the relationship between the quality of institutions and a set of historical, economical, geographical, religious and social variables via a genetic programming application. In this sense, we intend to verify the existence of a bias motivated by employing a parametric perspective; therefore, we try to detect possible misspecifications problems associated to the traditional parametric models. In our opinion, this verification is crucial in an empirical application and it should be always done in order to verify and corroborate the adequacy of the parametric results. In our specific application we use a Genetic Programming called DARWIN (Álvarez et al., 2001) to realize this verification and, additionally, to model what factors explain the institutional quality in different countries. To this purpose, we compare the GP results with those obtained from the traditional parametric point of view and analyse their differences and similitude.
The paper is presented as follows. After this introduction, Section 2 presents a brief explanation of the method used in our study. In Section 3, the data are described. In Section 4, the results obtained for each method are presented. Finally, in Section 5, we draw our conclusions.
Section snippets
Genetic programming
Genetic Algorithms, originally developed by Holland (1975) and later spread by Goldberg (1989) and Mitchell (2001), enclose a whole series of computing procedures inspired in biological concepts based on the Theory of Evolution of Species: survival of the fittest individuals, reproduction, and birth of offspring with a good genetic heritage. The basic characteristic of these procedures is to use some evolutionary rules observed in nature as inspiration for solving certain mathematical
Data
This paper analyses the functional relation between the quality of institutions and a set of historical, geographical, economical, religious and social variables. In Table 1 a brief description of the considered variables is showed. The General Institutional Quality Index (IG) is the endogenous variable, while the other nine variables are considered as explanatory variables. In this way, the database constructed for our study contains complete information about 117 countries.
Regarding the
Results
In order to study the institutional determinants and their temporal dynamics, we estimate our model for different years of the dependent variable. Table 4 depicts the results obtained using OLS regression and GP. At a first glance, we should highlight the temporal consistency in the results for both methods. In spite of considering different years, the results in terms of Adjusted R-Square, the explanatory variables finally chosen and their effects on quality index do not show a huge temporal
Conclusions
The general procedure to model the quality of institutions has been based almost exclusively on a linear and parametric point of view. Therefore, a-priori and rigid functional forms are discretionally imposed by the researcher rather than observed in the data. This leads to a possible misspecification problem and, in consequence, a bias in the results. In order to validate the results obtained in a parametric framework, it is relevant to open an avenue of research on the determinants of the
References (38)
- et al.
DARWIN—an evolutionary program for nonlinear modeling of chaotic time series
Computer Physics Communications
(2001) - et al.
Specification search in non-linear time-series models using genetic algorithms
Journal of Economic Dynamic & Control
(2002) - et al.
The colonial origins of comparative development: an empirical investigation
American Economic Review
(2001) - et al.
Forecasting exchange rates using genetic algorithms
Applied Economic Letters
(2003) - et al.
Genetic multi-model composite forecast for non-linear forecasting of exchange rates
Empirical Economics
(2005) - et al.
Using genetic algorithms to estimate and validate bioeconomic models: the case of the Ibero-Atlantic sardine fishery
Journal of Bioeconomics
(2006) La Nueva Economía Institucional
Sistema
(2001)El programa de la Nueva Economía Institucional: lo macro, lo micro y lo político
Ekonomiaz
(2002)The nature of the firm
Economica
(1937)The problem of social cost
Journal of Law and Economics
(1960)
The task of the society
ISNIE Newsletter
Africa's growth tragedy. Policies and ethnic divisions
Quaterly Journal of Economics
Tropics, germs, and crops: how endowments influence economic development
Journal of Monetary Economics
Genetic algorithms in search, optimization, and machine learning
Institutions and the path to the modern economy
Why do some countries produce so much more output per worker than others?
Quaterly Journal of Economics
Adaptation in Natural and Artificial Systems
What Determines the Quality of Institutions?
Genetic programming prediction of stock prices
Computational Economics
Cited by (21)
Estimation of state changes in system descriptions for dynamic Bayesian networks by using a genetic procedure and particle filters
2014, Economic ModellingCitation Excerpt :First, we organize the relation equation between the basic DAG model and the change of the DAG shape (Fearnhead, 2006; Friedman et al., 1998; Nielsen and Nielsen, 2008; Punskaya et al., 2002; Wang et al., 2008, 2011). We describe the method for estimating the DAG shape change and variable state transition by using a genetic approach (Alvarez-Diaz and Miguez, 2008; Chen and Duan, 2011; Chi and Tang, 2007; Tokinaga and Ikeda, 2011; Wong et al., 1999). In this case, in the estimation of DAG shape change, we do not use the entire process included in the crossover procedure of Genetic Programming (GP), but use methods based on mutations, such as changing the direction of a branch of the DAG, in order to maintain consistency in the process (called Evolutionary Programming: EP in Wong et al., 1999) (Alvarez-Diaz and Miguez, 2008; Antoci et al., 2012; Chen and Duan, 2011; Chi and Tang, 2007; Tokinaga and Ikeda, 2011; Wong et al., 1999).
Corruption and growth: The role of governance, public spending, and economic development
2014, Economic ModellingCitation Excerpt :The quality of governance is manifested through exogenous cost parameters faced by the agents for corruption, tax evasion, and monitoring. Notably, Álvarez-Díaz and Caballero (2008) find that the parametric perspective about institutional quality can be treated as accurate. Intuitively, a better quality of institutions leads to higher costs imposed on illicit behaviour.
Evolutionary Computation and Genetic Programming
2013, Engineered BiomimicryTransport energy demand forecast using multi-level genetic programming
2012, Applied EnergyCitation Excerpt :Alvarez-Diaz and Miguez used GP approach to evaluate quality of the institutions. They employed GP to study the functional relation between the quality of institutions and a set of historical, economical, geographical, religious, and social variables [11]. Karabulut and his colleagues applied GP to forecast long term electrical power consumption in Turkey.
On dichotomous choice contingent valuation data analysis: Semiparametric methods and Genetic Programming
2010, Journal of Forest EconomicsInstitutions and management of fishing resources: The governance of the Galician model
2008, Ocean and Coastal ManagementCitation Excerpt :It is quite clear that an institutional approach in studying fishery resources can clearly not be limited to issues of property rights. New institutional economics [17–19] has drawn up a broader theoretical framework that includes references among others to informal norms, social capital, path dependence, public policies, beliefs and transaction costs [20–23]. Thus the theoretical basis for new institutional economics has enabled progress to be made on issues put forward by Jentoft [12]: “Institutions in fisheries: What they are, what they do, how they change”.
- ☆
This paper was presented at the Annual Conference of the International Society for New Institutional Economics (Barcelona, 2005), the Annual Meeting of Public Choice Society (New Orleans, 2006), the Meeting of Applied Economics (Jaen, 2006) and the Simposium of Economic Analysis (Oviedo, 2006). A previous version was presented at FUNCAS as a working paper (Colección de Documentos de Trabajo de la Fundación de las Cajas de Ahorro (FUNCAS), Número 239, Madrid, 2006). We thank the anonymous referees for their useful comments.