The gravity equation is probably the most important tool in international economics to explain and estimate trade flows. However, since the gravity equation is important for political decisions, it is very important to achieve reliable results from its empirical application. Thus, it is necessary to employ the gravity equation using a theoretically and empirically proper methodology. One important discussion addresses the implausibly high measures for the impact of trade cost proxies on exports that frequently appear, especially in older works. This problem became known as the "border puzzle" (Obstfeld and Rogoff, 2001). The aim of the study is to contribute to the discussion about the suitability of the gravity equation's empirical applications. The basic idea is that trade costs between two countries could additionally depend on the exports between these two countries and not only on the (more or less) exogenous proxy variables for trade costs, as they are normally used. In this study, a new theory of endogenous trade costs is provided which shows that iceberg trade costs are likely to depend on exports. An interaction between exports and trade costs (or the gravity function and a trade cost function) leads to a simultaneity problem. Moreover, this theory can be confirmed after estimating the gravity equation with an alternative econometric strategy: A simultaneous equation system using a theory-based index to compensate for the directly immeasurable trade. A further target of the study is in its use of the comprehensive trade cost index to compute "multilateral resistances" of countries to trade, introduced in the trend-setting work by Anderson and van Wincoop (2003). These multilateral resistances are necessary to retrieve unbiased results from empirical gravity equations. A methodology was developed to make the heretofore unknown index of multilateral resistances visible. The result of the simultaneity approach and the use of constructed data for bilateral and multilateral trade costs is that the estimated direct effects of variables influencing exports decrease. The proposed methodologies of this study could help to achieve more plausible and reliable results from the gravity equation as the "workhorse for empirical studies" (Eichengreen and Irwin, 1998) of international trade.

The gravity equation is probably the most important tool in international economics to explain and estimate trade flows. However, since the gravity equation is important for political decisions, it is very important to achieve reliable results from its empirical application. Thus, it is necessary to employ the gravity equation using a theoretically and empirically proper methodology. One important discussion addresses the implausibly high measures for the impact of trade cost proxies on exports that frequently appear, especially in older works. This problem became known as the "border puzzle" (Obstfeld and Rogoff, 2001). The aim of the study is to contribute to the discussion about the suitability of the gravity equation's empirical applications. The basic idea is that trade costs between two countries could additionally depend on the exports between these two countries and not only on the (more or less) exogenous proxy variables for trade costs, as they are normally used. In this study, a new theory of endogenous trade costs is provided which shows that iceberg trade costs are likely to depend on exports. An interaction between exports and trade costs (or the gravity function and a trade cost function) leads to a simultaneity problem. Moreover, this theory can be confirmed after estimating the gravity equation with an alternative econometric strategy: A simultaneous equation system using a theory-based index to compensate for the directly immeasurable trade. A further target of the study is in its use of the comprehensive trade cost index to compute "multilateral resistances" of countries to trade, introduced in the trend-setting work by Anderson and van Wincoop (2003). These multilateral resistances are necessary to retrieve unbiased results from empirical gravity equations. A methodology was developed to make the heretofore unknown index of multilateral resistances visible. The result of the simultaneity approach and the use of constructed data for bilateral and multilateral trade costs is that the estimated direct effects of variables influencing exports decrease. The proposed methodologies of this study could help to achieve more plausible and reliable results from the gravity equation as the "workhorse for empirical studies" (Eichengreen and Irwin, 1998) of international trade.