I'm running an instrumental variables regression with a single predictor on a sample size of 120.
When I run the regression, the predictor is significant at the 0.01 level, but the adjusted R-squared is negative.
My interpretation of a negative R-squared value would be that my predictor does not fit the data at all. If that's true, however, I'm not sure why the predictor itself would have such a high level of significance. Could the fact that I'm not running OLS regression be affecting the the results? How do results like this occur?