3.4 ESTADÍSTICAS DESCRIPTIVAS
4.1.2 Sesgo de Selección
Table 5.3 shows the dependent and independent variables used in the modified or augmented gravity models from previous empirical studies, and their hypothesised signs. The most common explanatory variables are Gross National Incomes (GNPs) or Gross Domestic Products (GDPs), per capita GNPs or GDPs, bilateral distance, adjacency, and trade bloc membership. However, more explanatory variables are added into the model such as differences of per capita GDPs of the exporting and importing countries, population, remoteness, exchange rate, common language, price index, production index,
complementarity index, extra-bloc trade openness, etc. The dependent variables include the values of export flows, import flows or total trade flows (exports plus imports).
The GNPs or GDPs of exporting and importing countries are included in the gravity model as an explanation of the economic size of the countries. The coefficients of exporter and
importer GNPs or GDPs are hypothesised to be positive. A high level of exporter income results in an increase in production investment, which increases the availability of goods for export. A high level of income in the importing country leads to higher imports. Larger exporting and importing countries tend to trade more (Bergstrand, 1989; Frankel and Wei, 1998; Krueger, 1999; Endoh, 2000; Elliott and Ikemoto, 2004; Jayasinghe and Sarker, 2004;
Table 5.3 Variables used in the gravity equation for bilateral trade flows and the hypothesised signs
Variables Types of variables
Expected signs
Estimated signs by previous studies Bergstrand
(1989)
Frankel and Wei (1998) Krueger (1999) Endoh (2000) Dascal et al. (2002) Dependent Variable Continuous Disaggregated
Export Aggregated Export Aggregated Export Aggregated Export Disaggregated Export Import Explanatory Variables
- GDP of the exporting country Continuous (+) (+) (+) (+) (+)
- GDP of the importing country Continuous (+) (+) (+) (+) (+)
- Per capita GDP of the exporting country Continuous (+)/(-) (+)/(-) (+) (+) (+) (+)
- Per capita GDP of the importing country Continuous (+)/(-) (+)/(-) (+) (+) (+) (+)
- Differences of per capita GDP Continuous (+)/(-)
- Population of the exporting country Continuous (+)/(-) (-)
- Population of the importing country Continuous (+)/(-) (-)
- Bilateral geographical distance Continuous (-) (-) (-) (-) (-)
- Remoteness of the exporting country Continuous (+) (+) (+) (+)
- Remoteness of the importing country Continuous (+) (-) (+) (-)
- Adjacency Dummy (+) (+)/(-) (+) (+) (+)
- Landlockedness Dummy (-)
- Bilateral exchange rate Continuous (+)/(-) (+)/(-) (+) (-)
- Common language Dummy (+) (+) (+) (+)
- Exporter wholesale price index Continuous (+) (+)/(-) (-)
- Importer wholesale price index Continuous (-) (+)/(-) (-)
- Production index (production capacity) Continuous (+)/(-) (+) (-)
- Complementarity index Continuous (+)
- Intra-RTA/FTA trade bias Dummy (+) (+)/(-) (+)/(-) (+)/(-) (+)/(-) (+) (+)
- Extra-RTA/FTA import openness Dummy (-) (+)/(-) (+)/(-) (+)/(-)
- Extra-RTA/FTA export openness Dummy (-) (+)/(-)
Table 5.3 Variables used in the gravity equation for bilateral trade flows and the hypothesised signs (continued)
Variables Types of variables
Expecte d signs
Estimated signs by previous studies Elliott and
Ikemoto (2004)
Jayasinghe and Sarker (2004)
Liu (2004) Grant and Lambert (2005)
Kisu (2006)
Ram and Prasad (2007) Dependent Variable Continuous Aggregated
Import Disaggregated Total Trade Aggregated Import Disaggregated Import Aggregated Total Trade Aggregated Total Trade Explanatory Variables
- GDP of the exporting country Continuous (+) (+) (+) (+) (+) (+) (+)
- GDP of the importing country Continuous (+) (+) (+) (+) (+)/(-) (+) (+)
- Per capita GDP of the exporting country Continuous (+)/(-) (+) (+)/(-) (+)/(-) (+)
- Per capita GDP of the importing country Continuous (+)/(-) (+) (+)/(-) (+)/(-) (+)
- Differences of per capita GDP Continuous (+)/(-) (+)/(-)
- Population of the exporting country Continuous (+)/(-) (+)
- Population of the importing country Continuous (+)/(-) (+)
- Bilateral geographical distance Continuous (-) (-) (-) (-) (+)/(-) (-) (-)
- Remoteness of the exporting country Continuous (+)
- Remoteness of the importing country Continuous (+)
- Adjacency Dummy (+) (+) (+) (+) (+) (+)
- Landlockedness Dummy (-) (+)/(-) (-)
- Bilateral exchange rate Continuous (+)/(-) (-) (-)
- Common language Dummy (+) (+) (+) (+)
- Exporter wholesale price index Continuous (+)
- Importer wholesale price index Continuous (-)
- Production index (production capacity) Continuous (+)/(-)
The per capita GDPs of exporting and importing countries in the aggregated gravity model also explain the economic size of the countries. The coefficients of per capita GDPs are positive (Frankel and Wei, 1998; Krueger, 1999; Elliott and Ikemoto, 2004; Ram and Prasad, 2007). But in the disaggregated gravity model, the coefficients of per capita GDPs of
exporting and importing countries can be interpreted in terms of microeconomic theory. The coefficients of per capita GDPs of exporting and importing countries would have both positive and negative signs. A positive coefficient for an exporter per capita GDP indicates that the considered product tends to be capital intensive in production, and a negative coefficient for an exporter per capita GDP indicates that the considered product tends to be labour intensive in production. In cases of an importer per capita GDP, a positive coefficient indicates that the considered product is normal goods, and a negative coefficient for an importer per capita GDP indicates that the considered product is inferior goods (Jayasinghe and Sarker, 2004; Grant and Lambert, 2005).
Differences in per capita GDPs of exporting and importing countries are included in the aggregated gravity model to test the Linder and Hecksher-Ohlin hypotheses. A negative coefficient of differences in per capita GDPs would support the Linder hypothesis: two countries with similar income tend to increase trade between them. A positive coefficient of differences in per capita GDPs would support the Hecksher-Ohlin hypothesis: two countries with differences in income tend to increase trade between them (Elliott and Ikemoto, 2004; Ram and Prasad, 2007).
Population is a measurement of country size. The population of exporting and importing countries would have negative effects on bilateral trade flows. A larger population in an exporting country implies a larger domestic market which results in a reduction in exports. However, a larger population in an importing country leads to more varieties of domestic output and less dependence on imports (Endoh, 2000). While Liu (2004) found that the exporter and importer population have a significantly positive impact on aggregated bilateral import.
Bilateral geographical distance is used as a proxy for transportation cost which would have a negative effect on bilateral trade flows. A greater distance leads to a higher transportation cost and a reduction in trade (Bergstrand, 1989; Frankel and Wei, 1998; Krueger, 1999; Endoh, 2000; Elliott and Ikemoto, 2004; Jayasinghe and Sarker, 2004; Liu, 2004; Grant and Lambert, 2005; Kisu, 2006; Ram and Prasad, 2007). In addition, the remoteness of the exporting and importing countries is added into the gravity model as a measurement of how far an exporting or importing country is from other countries. An exporter’s and importer’s remoteness would
have positive effects on bilateral trade flows. Both remote countries tend to trade more with each other (Frankel and Wei, 1998).
Adjacency and common language are dummy variables which are equal to one when a pair of trading countries shares a land border or common language. The coefficients of the dummies would be positive. If two countries are neighbouring countries, the transportation cost
between them will decrease and they will trade more with each other (see Bergstrand, 1989; Frankel and Wei, 1998; Krueger, 1999; Endoh, 2000; Elliott and Ikemoto, 2004; Liu, 2004; Grant and Lambert, 2005; Kisu, 2006; Ram and Prasad, 2007). Similarly, if a pair of countries communicates the same language, communication and transaction costs will decrease and trade between them will increase (see Frankel and Wei, 1998; Krueger, 1999; Endoh, 2000; Liu, 2004; Grant and Lambert, 2005; Ram and Prasad, 2007). On the other hand, the dummy variable for landlocked has a negative effect on bilateral trade flows. If a pair of trading countries is landlocked and does not have any ocean port, their transportation costs will increase and their trade will decrease (Grant and Lambert, 2005; Ram and Prasad, 2007). Exchange rate is expected to have both positive and negative effects on bilateral trade flows depending on the exporter or importer views. The appreciation of the real exchange rate of the exporting country tends to reduce its exports while the appreciation of the real exchange rate of the importing country tends to increase its imports (Bergstrand, 1989; Dascal et al., 2002; Liu, 2004; Kisu, 2006).
Wholesale price index (WPI) is a proxy of the price of goods. The coefficient of the exporter wholesale price index is expected to have a positive effect on bilateral trade flows while the coefficient of the importer wholesale price index is expected to have a negative effect on bilateral trade flows. For example, if there is an increase in the price of a specific product in the exporting country, the exporter tends to produce and export more. If a product price in the importing country increases, import demand for the product will decrease (Bergstrand, 1989; Dascal et al., 2002).
A dummy for intra-RTA/FTA trade bias is included in the gravity model to capture trade creation of RTA/FTA. Dummies for extra-RTA/FTA import openness, and extra-RTA/FTA export openness are added to examine trade diversion. The coefficient of intra-RTA/FTA trade bias is expected to be positive which indicates trade creation of RTA/FTA. The presence
tend to trade less with non-members (see Frankel and Wei, 1998; Krueger, 1999; Endoh, 2000; Elliott and Ikemoto, 2004; Jayasinghe and Sarker, 2004; Liu, 2004; Grant and Lambert, 2005). Some studies capture only trade creation with a dummy for intra-RTA/FTA trade bias while some studies capture both trade creation and trade diversion effects which add two or three dummies into the gravity model.