Methodology
This research is based on the unique data base which was compiled from multiple sources for the period of time from 1990 to 2005. The data consists of 34 variables and 1161О observations. It was collected in three steps, specifically for this work. First of all, we gathered the data on exports from the country-member of BRICS to every country that it traded with from the International trade statistics. This variable consists of one-way export values. Secondly, we used the NSF-Kellogg Institute Data Base on Economic Integration Agreements to extract the data on entering the economic integration agreement and the type of this agreement. In contrast with most other similar data bases which use а binary variable (0-1) to index the presence of an EIA between two countries, this data base uses а multichotomous index (0-4), with the following interpretations: О indicates no existing Economic Integration Agreement, 1 indicates а One-Way Preferential Trade Agreement, 2 indicates а Two-Way Preferential Trade Agreement, 3 indicates а Free Trade Agreement and 4 indicates а Customs Union between BRICS country and the country that it traded with in the period of time 1990-2005. Finally, we gathered gravity data for all countries from The СЕРП Gravity Dataset. The main variables here are the area of the country, population, GDP of each country and other indexes. All of the above-mentioned variables are crucial for obtaining the empirical evidence of the existence of the economically and statistically significant effects of entering the economic integration agreements on the dynamics of export of countries that form BRICS and of the relationship between trade liberalizations under the different types of EIAs and the trade flow. Data management and analysis was performed using Stata 12 (2011).
ln recent empirical literature, various methods have been developed and introduced in the academic literature to measure the effects of entering the EIAs. Existing evidence suggests that export flows expand after the signing of different integration agreements. But usually papers employ small number of country pairs and years and do not adjust their models to varying types of these agreements. The methodology that we apply in this work differs in the following way. First of all, we develop the basic gravity model including quantitative control variables such as GDP, population and area, and we comment on the results of this regression. Secondly, we create separate variables for each type of the integration agreements and establish the individual gravity model for types 1-4 and also build а single model for trading without integration agreement between countries. Then we ran several specifications to obtain precise results for each type of the economic integration agreements. Lastly, we interpret the obtained variables of interest, give а numerical estimation and then describe meaning and economic foundations behind the effects.
The main method that we adopt for this investigation is the gravity model with several specifications. We take the log of export values as а dependent variable for our regression because measuring the effects of various variables on export flows is one of the most important goals of this research. The independent variables serve as unique specifications of the standard gravity equation model which we introduce in this work. These variables include EIA variable which was described earlier, quantitative variables such as population, GDP and area for both the exporting country (member country of BRICS) and its trading partner. Also we introduce several dummy variables (О if not, 1 if yes) such as common official language, acceptance to the European Union, GATT/WTO membership and some other factors.
Another goal of this study is to propose the possible solution for the problem of the endogeneity bias. We use the Durbln-Wu-Hausman test that helps to evaluate whether а statistical model corresponds to the data. After conducting this test, we apply panel techniques and more precisely we add the fixed-effects specification to the gravity equation to estimate the partial effects of EIAs on pairs of countries' export flows with accounting for the endogeneity bias. It is important to note that the trade variables such as the area of the country are time-invariant and fixed effect is needed to capture all time-invariant unobservable factors that can influence export flows and affect the accuracy of the estimation. This method is an effective way of the elimination of the problem of endogeneity and also one of the most practical ways of estimating precisely the effects of EIAs on export flows for а large set of countries.
