1
European Economic Integration in Econometric Modeling –
Concepts and Measures[(]
Christiane Krieger-Boden, Rüdiger Soltwedel,
The Kiel Institute for the World Economy, Kiel, Germany
Abstract:
Processes of regional economic integration have been shaping the economic relations between countries significantly during the last decades. In addition, an increasing integration of the national economies into the global economy has affected these economic relations, too. In an effort to operationalize these integration processes for the purpose of empirical analyses, this paper reviews concepts and actual measures of European integration and globalization. In particular, it discusses how to separate the effects of European integration from those of globalization. The paper searches for regional integration and globalization indices that avoid the endogeneity problem, and it discusses collinearity between them. Operational indicators are suggested that are then used for analysis in an illustrative gravity model.
1 Introduction 1
2 Understanding regional integration and globalization 2
2.1 Concepts of regional integration and globalization 3
2.2 Distinguishing EU integration from globalization 7
3 Measuring regional integration and globalization 9
3.1 Measures of European integration 11
3.2 Measures of globalization 13
3.3 Correlations between the EU integration and the globalization indices 20
4 Illustration: A gravity model for the European Union 25
4.1 Empirical model and data 25
4.2 Results 27
5 Conclusion 29
References 30
Appendix 33
1 Introduction
Processes of regional economic integration have been shaping the economic relations between countries significantly during the last decades, prominent examples being the European Union (EU), the North American Free Trade Agreement (NAFTA), and the Asia Pacific Economic Cooperation (APEC). In addition, an increasing integration of the national economies into the global economy has affected these economic relations, too. Both kinds of economic integration reduced transactions costs, though perhaps in different ways and to different degrees, and among a given regional subset of countries in the one case, and among all countries worldwide in the other. Conventional wisdom has it, and economic theories of trade and growth show it, that both kinds of integration processes are likely to have had a tremendous impact on the intensity of trade and the division of labor between the countries involved, as well as on wages and incomes within them.
However, empirical studies trying to quantify the impact of these integration processes face major problems: Generally, it is by far not trivial to determine indicators that operationalise integration processes in order to use such indicators as explanatory variables for various aggregate economic variables in econometric models (cf. Bosker, Garretsen 2007). More particularly, it is even more demanding to distinguish separate indicators for the two kinds of integration processes, regional integration and globalization, where the effects of both may be observational equivalent.
In the empirical literature, to the best of our knowledge, integration indicators have not yet been used for quantifying effects of regional integration. Various studies have instead used a simple before-after dummy variable (e.g., Bun and Klaassen 2007; Rose 2000; Glick and Rose 2002), a time trend (e.g., Traistaru, Nijkamp, Longhi 2002), or changes of physical transportation costs (e.g., Glaeser, Kohlhase 2005; Schürmann, Talaat 2000). However, such approaches neglect the gradual non-monotonous nature of the integration process over a long period of time and as a result of several subsequent small integration steps, or they neglect the numerous facets by which integration processes manifest. Empirical studies that would take into account this gradual and multifaceted nature could be expected to yield more accurate estimates of the effects of integration processes than those that do not. The present paper investigates to what extent the available statistical indices that aim at quantifying the progress of regional integration can be used as explanatory variables in econometric models.
Moreover, most empirical studies also neglect the problem of observational equivalence between regional integration processes and globalization. There is a good chance that effects of globalization are misinterpreted as effects of the regional integration processes, if they are not controlled for explicitly in time-series or panel regression analyses. The present paper investigates to what extent the available statistical indices that aim at quantifying the progress of globalization can be used as control variables in panel regressions that aim at assessing the economic effects of regional integration processes.
To adequately specify indicators of regional integration and globalization requires getting straight which purpose the indicators are supposed to serve. In particular, analyses adopting a microeconomic perspective, i.e., assessing the effects of economic integration on economic agents, require detailed indicators that take different values for different agents. In this paper, however, a macroeconomic perspective is adopted, and such analyses, assessing the effects of economic integration on countries, require aggregate indicators that take only one value for a whole country.
Besides being adequately specified and being able to distinguish between regional integration and globalization, operational indicators of regional integration and globalization should satisfy further, practical requirements. The preferred indices should be exogenous to the response variable in respective empirical investigations. If the appropriate indices cannot be assumed to be exogenous, the additional problem of finding appropriate instruments for these indices arises. The present paper will search for regional integration and globalization indices that avoid the endogeneity problem. Also, the paper will discuss the collinearity between potential indicators.
The paper is organized as follows. Section 2 surveys concepts of regional integration processes and globalization in order to distil operational indicators for econometric models. It investigates to what extent regional integration processes can be distinguished conceptually from global integration processes. Section 3 surveys empirical approaches of measuring regional integration and globalization, and investigates by means of a correlation analysis to what extent indices of regional integration can be distinguished empirically from those of globalization. Section 4 presents an illustrative estimation of a panel gravity model that uses the indices of regional integration and globalization to identify the effects of the EU integration on intra-European trade intensities. Finally, Section 5 concludes.
2 Understanding regional integration and globalization
Several concepts of regional integration and globalization exist in the literature, and usually both terms are defined rather vaguely. This section shortly goes over such concepts in order to arrive at indicators of regional integration and of globalization that are operational for econometric models. In a first part, the high complexity of these processes is demonstrated. For meaningful econometric analysis, it would be highly desirable to have this complexity reflected in respective integration indicators. As this paper adopts a macroeconomic perspective, the required indicators need to reflect this complexity in one value for a whole country. In a second part, we will investigate the differences between regional and global integration processes. Only to the extent that the two processes differ conceptually from each other, their differences can be exploited in a meaningful way in empirical studies.
2.1 Concepts of regional integration and globalization
Pure models of trade theory (particularly NEG models, e.g. Fujita et al 1999, Baldwin et al. 2003) analyse the impact of economic integration on trade, wages and the division of labor by defining economic integration to be the inverse of transportation costs.[1] In these models, integration is assumed to reach from autarky (no integration at all) to unrestricted freeness of trade (complete integration). Usually, no distinction is made between regional integration and globalization. Moreover, in such models, integration usually refers to the freeness of exchanging goods and services only.
When analysing the effects of economic integration empirically, however, this definition turns out to be too narrow to capture the complexity of the actual phenomenon of economic integration, and in fact there are several much broader concepts of economic integration. Thus, following Balassa (1961) and Oman (1996), economic integration is often understood as a process, which is initiated by some driving forces and results in a state where all sorts of economic transactions beween countries have increased, or in other words, following Cairncross (1997), where the economic distance between countries has declined. In this broader sense, the integration process consists of three constitutive elements: the forces driving the process of economic integration, the transmission channels through which economic integration affects the economies of the integrating countries and the consequences of economic integration (Figure 1). Following Oman (1996) and van Liemt (1998), the driving forces include technological progress in information, communication and transportation (ICT) technologies (that are interpreted to take place autonomously) as well as institutional progress by deliberate national or supranational policy measures. Moreover, there may be other driving forces such as innovations in industrial organization that direct the focus of business activities towards global rather than local (national) markets and locations, or innovations on financial markets that increase the volatility of financial flows. The impetus from these driving forces is transmitted via reductions of transaction costs, both spatial transaction costs and natural or administrative border impediments between one country and other countries. And following Bhagwati (2004) and Schulze and Ursprung (1999), these reductions of transaction costs affect trade of goods and services, capital flows, flows of information or knowledge, or migration of workers. The effects of lower transaction costs on trade or migration then finally yield the consequences in the form of increased trade intensity, of adjustments in the international division of labor, and of changes of income, employment and growth in the countries involved in the process.
Figure 1: Schematic relationship of driving forces and consequences in processes of economic integration
Source: Own illustration.
The transaction costs are at the core of the process of economic integration, linking driving forces and consequences, and being closest to the concept of economic integration in pure models of trade theory. To precisely determine these transaction costs would represent a major improvement for assessing the effects of economic integration. However, this is difficult to achieve given the high degree of heterogeneity of transaction costs. Transaction costs vary considerably according to the various dimensions they include. To list but a few of such dimensions: First, as mentioned above, transaction costs differ according to the objects-to-be-moved: Moving goods, people, capital, or information requires different means of transport with different amounts of costs. Second, transaction costs differ according to the distance dependencies incorporated:[2] Some costs are closely related to all forms of bridging geographical distance (spatial transaction costs). Others are independent of distance (one-stop border impediments); they may arise from tariffs or non-tariff barriers, technical standards, market and product regulations, consumer preferences, differences in law, culture and language, or other institutional settings. Third, transaction costs differ according to different components included: The material component relates to the costs of physically moving objects from one place to another, which in itself may vary according to the transport means employed. The non-material component relates to costs of, e.g., getting permissions and allowances, paying tariffs, closing contracts between the agents involved (potentially under different legal and cultural systems and in different languages), insuring freight, processing payments, opening up foreign markets for the objects in question.
Fourth, transaction costs differ according to the spatial validity of the rules under which they accrue, e.g., within a country, within a regional integration zone like the EU, or worldwide. Costs for land transport, for instance, often depend on the specific rules within each country; certain cross-border transaction costs depend on rules agreed upon bilaterally or within a regional integration zone; marine or air transport costs for goods and persons depend on the rules of global markets. Each country has, in principle, the sovereignty of deciding to what extent it participates in such bilateral or multilateral rules. It may take measures to unilaterally reduce the transaction costs of international trade or factor flows to a greater extent than other countries, participate in multilateral agreements on jointly reducing transaction costs, or exempt itself from economic integration processes.[3] Each country may even prevent its economic agents from adopting new ICT technologies and thus shape its transaction costs. Fifth, transaction costs differ according to time: Usually they are assumed to have declined continuously and gradually over at least the last fifty years, yet there may also be sudden leaps such as the fall of the iron curtain, or drawbacks. [4]
As each of these dimensions may occur in combination with almost each other, they span a whole range of different possible transaction costs (not all combinations are equally likely, however). In a strict sense, transaction costs are thus to be defined for a specific transaction between a specific pair of countries at a specific point in time only. Also, any reductions of transaction costs thus refer to specific transactions between a specific pair of countries and for a specific time interval only.
When analysing empirically the effects of economic integration, a reflection of this immense heterogeneity of transaction costs in respective integration indicators would be highly desirable. This would assure that the effects of integration are specified adequately in time series and panel regression estimates. However, obviously, it is hard to imagine operational indicators that would allow for all dimensions of transaction costs. Moreover, for macroeconomic analyses, all information has to be condensed into aggregate country-specific and year-specific indices. In search for operational indicators, attention is therefore drawn to the driving forces of economic integration processes that underlie the transaction cost reductions. These driving forces may serve as proxies of transaction costs facilitating the procedure of determining adequate indicators. Moreover, they are more likely to be exogenous to the response variables of respective empirical investigations than any transaction cost indicators.
The driving forces of economic integration intervene into all combinations of transaction costs at different points. The progress in ICT technologies, for instance, is expected to extend the opportunities to interact with other countries by bridging distance. Major improvements of this sort are, e.g., the invention of automobiles and expressways, of container traffic, or of the telecommunication technology from the telegraph to the internet. Primarily, the progress in ICT technologies can be expected to reduce the material component of the transaction costs and to influence distance-dependent costs. However, progress in ICT technologies, particularly of telecommunication technologies, may also facilitate the handling of legislative regulations, of administrative and cultural border barriers, etc., thereby reducing also non-material, one-stop transaction costs.