Gravity model analysis: robust evidence from the Czech Republic and corruption matching
|Author(s):|| Mgr. Eva Michalíková , |
Mgr. Michal Paulus ,
|Type:||IES Working Papers|
|ISSN / ISBN:|
|Published in:||IES Working Papers 32/2014|
|Keywords:||Czech Republic, export, gravity model, fixed effects, LSDV, least trimmed squares|
|JEL codes:||C13, C23, F10, F12, F14|
|Suggested Citation:||Michalíková E., Paulus M. (2014). “Gravity model analysis: robust evidence from the Czech Republic and corruption matching” IES Working Paper 32/2014. IES FSV. Charles University.|
|Grants:||GAČR č. P402/12/0982: Trade Flows in Times of Economic Boom and Slump: Modifying the Gravity Model for Country, Time and Product-Specific Decision-Making SVV 260 113 - Strengthening Doctoral Research in Economics and Finance|
|Abstract:||This paper analyses Czech exports by applying the gravity model to a panel dataset consisting of 177 trade partners during 1995-2011. The model is based on a micro-founded specification derived for panel data estimations. We utilize the fixed effects (FE) and LSDV estimations and the Least Trimmed Squares (LTS). The FE and LSDV methods allow us to deal with multilateral resistance terms derived from micro-economic specification of gravity models. We demonstrate a theoretical bias in estimated coefficients, if the estimation does not take into account the resistance terms.. The heterogeneity between partners is explored by the LTS through which we identify the most important outliers from the perspective of our gravity model.
The results generally confirm that Czech trade is oriented towards European countries and determined primarily by key economic factors of domestic and foreign GDP. The institutional variables remain largely insignificant, except corruption due tothe counterintuitive result that a higher corruption level in partner country should boost mutual trade. We interpret this finding as a result of “corruption matching”. The exclusion of outliers (LTS) significantly increases R-square and extends the number of significant determining factors (e.g. population or other institutional variables). The outliers, according to the LTS, are composed mainly of African, Asia and South or Central America states.