Sign of coef in OLS and IV-Lewbel Model

[LEE CHAN]

I used OLS and IV-Lewbel to analyze the effect of A on B. But on OLS it showed negative sign and IV-Lewbel it showed positive sign. how do i interpret this result? do i need to do Oster (2019) test after OLS? (actually i did it and the delta value is smaller than 1, is that enough for me to not trust the result from OLS and choose IV-lewbel?)

[George Ford]

It's not uncommon for the sign to change with IV. In demand/supply curve estimation, it's often the desired outcome since the signs are often wrong in OLS.