Dear Researchers,
I hope this message finds you in good spirits. I am conducting an ex-post analysis of the India-Malaysia CECA. I recently attended Silva Tenreyro's Gravity Clinic session where he strictly advised to avoid OLS, fixed, Random effect and Hausmen test. He suggested to use PPML model with fixed effects (time, country specific and country pair fixed effects) and also introduced to us new state of the art command called PPMLHDFE(high dimensional fixed effects). My second paper's analysis revolves around the India-Malaysia CECA and India's top 25 countries to whom India exports. As per the gravity toolkit (Head and Mayer, 2014), the total number of observations should be 14040 (N * T * (N-1)), i.e., 27 * 20 * 26 = 14040, N (number of countries), and T (time period). It will be great if you just go through my codes, output, and dataset once. Let me also shed some light on some of the dummy variables, which may be completely new for you. The dummy variables TC and TD represent trade creation and trade diversion, respectively. X_AL : - dependent variable and all other are my independent variables. I have limited my analysis to OLS and PPML only, with clustered standard errors. The problem is with the output when I add exporter and importer fixed effects, because of which most of the variables are omitted because of the collinearity, and my policy (TCij, TDi and TDj) variables become insignificant. These variables show the expected sign and are statistically significant only when I add year and country pair fixed effects (id). Please guide me. I have also added my output, dataset and Stata codes for your reference.
Short Note on Variables:-
* In time period "t," the sub-index i identifies the exporting country and j denotes the importing country.
* The variable, GDPi (j) is the income which shows the size of the economy.
* DISij represents distance between country i and country j
* The size of the market is represented by Popi(j), population of exporting country “i” has negative impact while on other
hand, population of importing country “j” positively affects the exports.
* The area of exporting country (Areai) boosts exports and area of importing country (Areaj) curbs export.
* The countrieswith common border and common official language are expected to trade more with each other.
Commborij and Langij Colij are dummies for a shared border and a shared official language respectively, takes value of 1
if they share all these aspects or zero otherwise.
Glimpse of the dataset
I hope this message finds you in good spirits. I am conducting an ex-post analysis of the India-Malaysia CECA. I recently attended Silva Tenreyro's Gravity Clinic session where he strictly advised to avoid OLS, fixed, Random effect and Hausmen test. He suggested to use PPML model with fixed effects (time, country specific and country pair fixed effects) and also introduced to us new state of the art command called PPMLHDFE(high dimensional fixed effects). My second paper's analysis revolves around the India-Malaysia CECA and India's top 25 countries to whom India exports. As per the gravity toolkit (Head and Mayer, 2014), the total number of observations should be 14040 (N * T * (N-1)), i.e., 27 * 20 * 26 = 14040, N (number of countries), and T (time period). It will be great if you just go through my codes, output, and dataset once. Let me also shed some light on some of the dummy variables, which may be completely new for you. The dummy variables TC and TD represent trade creation and trade diversion, respectively. X_AL : - dependent variable and all other are my independent variables. I have limited my analysis to OLS and PPML only, with clustered standard errors. The problem is with the output when I add exporter and importer fixed effects, because of which most of the variables are omitted because of the collinearity, and my policy (TCij, TDi and TDj) variables become insignificant. These variables show the expected sign and are statistically significant only when I add year and country pair fixed effects (id). Please guide me. I have also added my output, dataset and Stata codes for your reference.
Short Note on Variables:-
* In time period "t," the sub-index i identifies the exporting country and j denotes the importing country.
* The variable, GDPi (j) is the income which shows the size of the economy.
* DISij represents distance between country i and country j
* The size of the market is represented by Popi(j), population of exporting country “i” has negative impact while on other
hand, population of importing country “j” positively affects the exports.
* The area of exporting country (Areai) boosts exports and area of importing country (Areaj) curbs export.
* The countrieswith common border and common official language are expected to trade more with each other.
Commborij and Langij Colij are dummies for a shared border and a shared official language respectively, takes value of 1
if they share all these aspects or zero otherwise.
Glimpse of the dataset
Comment