在 python 中使用面板 ols 进行 Wald 测试

Question

I am trying to test hypothesis that all intercepts coefficients in pooled OLS are equal to zero with: F-test, Likelihood Ratio and Wald-test. I have some problems with the last one.

Here is the code PanelOLS((table1['Return']-table1['RF']), sm.add_constant(table1[['Mkt_RF','SMB','HML']])).fit().wald_test() in statsmodels you can put your hypothesis in wald_test method in string format, however it seems not to work in liner_model PanelOLS package. Am I doing it in a wrong way?

Thanks!

Answer 1

I found it. The problem is in syntax.

For example if I want to check if SMB=0 I do:

PanelOLS((table1['Return']-table1['RF']), sm.add_constant(table1[['Mkt_RF','SMB','HML']]), entity_effects=True).fit().wald_test(formula='SMB = 0')

which is a bit different from statsmodels