As part of my econometrics project, I am investigating wage structures for female workers. The econometric model I am using follows the Mincer earnings function. The dataset available includes variables, including:
'EDAGE' this is the age at which the worker left full time education 'Potential Experience' this is the age of the worker minus EDAGE, potexp = age-EDAGE
As part of preliminary research, I find out that female workers are more likely to have non-continuous labor market participation. Thus, I figure that there will likely be an estimation error in the potential experience variable, as it does not take into account periods of absence (e.g for maternity/paternity leave, unemployment, re-entering education and etc.).
To remedy this, I thought maybe I could simply create the a variable for AGE (which is not available in the dataset) by summing the variables 'Potential Experience' and 'EDAGE'. As this would still be appropriate for the Mincer Earnings function.
However, I am worried I will over complicate the model, risk specification errors and violate Gauss-Markov theorem.
Should I just stick with Potential Experience and simply acknowledge the likelihood of estimation error, or should I construct the different variable for age.
Please let me know, thank you very much.