Reverse causality is one of the reasons behind endogeneity. In any model we assume that the DV is caused by the IV. We have reverse causality when we can argue also the opposite: the IV is caused by the DV. In order to check whether your regression suffers from reverse causality, the best solution is to re-estimate the model using a lagged IV so that the DV(t) is a function of IV(t-1): the logic is that if there is no reverse causality the link between the IV and the DV has to be the same irrespective of the time lag. Thus, if the sign of the IV changes and is significant, this means that your regression suffers from reverse causality; if there are no changes in the sign of the IV, then you can rule out reverse causality.
Endogeneity is used in econometrics to refer to bias in regression estimates that is caused by (1) reverse causality, (2) omitted variables and (3) measurement errors.
Reverse causality means two variables are jointly determined. For example, if we have two variables X and Y and they are jointly determined, then it means X causes Y and Y causes X.
Fazul. I forgot to add that in single regression framework, the workhorse of dealing with endogeneity is using instrumental variables. Depending on ones preference, these could take the form of IV, IV2SLS, IV3SLS, GMM.
Reverse causality is one of the reasons behind endogeneity. In any model we assume that the DV is caused by the IV. We have reverse causality when we can argue also the opposite: the IV is caused by the DV. In order to check whether your regression suffers from reverse causality, the best solution is to re-estimate the model using a lagged IV so that the DV(t) is a function of IV(t-1): the logic is that if there is no reverse causality the link between the IV and the DV has to be the same irrespective of the time lag. Thus, if the sign of the IV changes and is significant, this means that your regression suffers from reverse causality; if there are no changes in the sign of the IV, then you can rule out reverse causality.
Endogeneity is said to occur if one or all the independent variable(s) is(are) correlated with the error term. This violates the CIA which is very paramount in cross sectional econometric analysis. One solution is to instrument the X variable (s). Reverse causality is when there is a bi-directional relationship between the DV and IV
Thanks for above post..Its useful to clear the mind about reverse casualty. Anybody could suggest the command to do as I am not sure how to add variable (t-1).