How to interpret results produced by kinkyreg command?

[Matthew Williams]

Hi all,

I am testing the validity of an instrumental variable (IV) using kinkyreg command (SSC). I have read the paper on this command in the Stata Journal but I am still not sure how to interpret the results produced by this command. Perhaps, Sebastian Kripfganz could you please share your insights? Do the results below provide any suggestions that the IV is valid?

    kinkyreg y x1 x2 x3 (viol = Z), range(-0.75 0.75) small inference(viol)
            
    estat exclusion, twoway(, ylabel(0.01 0.05 0.1)) notable

where viol is a potential endogenous variable and Z is an IV for it.




Thank you.

[Sebastian Kripfganz]

You cannot reject the validity of the instrument if the endogeneity correlation is roughly in the interval (-0.6, -0.1). To reach a conclusion, you need to take a stand on the range of plausible endogeneity correlations. For example, if theory tells you that the correlation of the endogenous regressor with the error term should be positive, then you could reject the validity of instrument.

It is important to keep in mind that there is no free lunch. The KLS procedure replaces one assumption (instrument validity) with another assumption (range of plausible endogeneity correlations). Some assumption has to be made.