引力模型中研究中国和其他国家的贸易，如何进行time and country fixed effect？

5967

收藏 2015-02-04

悬赏 3 个论坛币已解决

如题，要研究的是中国和其他国家的贸易，面板数据，时间跨度2001到2013，看了许多文献，里面如果用OLS回归的话就要加进去很多控制变量，尤其是中国的GDP，可是用OLS回归出来的结果特别的不好。。。另外就是用固定效应，如果是time and country fixed effect的话就还需要加时间的dummy。目前我就用xtreg的fe回归，结果比ols好，但是还没有尝试过再把时间也固定住。求问个位大神到底怎么样比较好一点？

最佳答案

sc13572468 查看完整内容

扫码加我拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

全部回复

sc13572468

2015-2-4 23:54:16

OLS is out-dated and theoretically wrong (See Anderson and Van Wincoop 2003). The norm now is to include fixed effects. My intuition is that if you have a panel, you want to control time-destination fixed effect (interaction between time and destination, not only time FE and destination FE). This is because the destination fixed effects capture general equilibrium objects, which should vary from year to year.

See
Head, K., and T. Mayer, 2014, "Gravity Equations: Workhorse,Toolkit, and Cookbook", chapter 3 in Gopinath, G, E. Helpman and K. Rogoff (eds), vol. 4 of the Handbook of International Economics, Elsevier: 131–-195.

Thierry Mayer has stata code posted on his website.

扫码加我拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

sinliulsy

2015-2-8 14:25:33

sc13572468 发表于 2015-2-6 23:39
OLS is out-dated and theoretically wrong (See Anderson and Van Wincoop 2003). The norm now is to inc ...

wow! thank you so much! But what I\'m considering is that my tutor suggested me to use the simple OLS first...and that\'s why I hesitate...but I should thank you all the same!