## Abstract (may include machine translation)

Gravity type models are very popular in analysing economic phenomena related to the flow of goods and/or services. As these models are frequently used for policy analysis, especially to investigate the effect of trading blocs, for example, it is quite important to get the econometrics 'right' and avoid unnecessary mistakes in the modelling process. The purpose of this paper is to suggest an alternative specification of the gravity models tailored for 'world models' and large data sets and, also, a testing procedure to check one of the basic assumptions of these models. In a recent short paper Matyas (1997) proposed to incorporate into the specification of the gravity models proper local (export), target (import) country and time specific effects. It can be shown that without these effects the parameter estimates of the model can lead to incorrect inference as their values may artificially be inflated or deflated by this misspecification. Matyas (1997) assumed that these effects were observable from the data and therefore adopted a fixed effects specification. In some instances, however, when the number of countries in the data set is large, for example, this is not a parsimonious approach and it would be better to take these effects into account as non-observable random variables. In this case all structural parameters of the model are assumed to be the same across the complete sample. It is then quite important to test the homogeneity of the data, that is whether it is correct to assume that for all countries and all time periods the parameters of the model are unchanged. Here we propose some easy-to-carry-out homogeneity tests in order to check the overall stability of the parameters of interest.

Original language | English |
---|---|

Pages (from-to) | 397-401 |

Number of pages | 5 |

Journal | World Economy |

Volume | 21 |

Issue number | 3 |

DOIs | |

State | Published - 1998 |

Externally published | Yes |