Forum: help

RE: Dummy Variables in MicEconSNQP [ Reply ]
By: Arne Henningsen on 2015-12-04 08:08

Dear Sam / Sami

Regarding your questions:

1) What are your "additional dummy variables"? Could they be considered as a fixed input?

2a) Quantities that are zero should remain zero -- replacing them by other values (e.g. one or mean values) will result in biased and inconsistent estimates. Compared to Translog functions, quadratic functions have the advantage that they can deal will zero values. However, if a substantial proportion of the values of a dependent variable is zero, you may have a censoring problem. Different procedures for dealing with censoring in the estimation of equation systems have been suggested in the literature. You may also consider to aggregate family labor and hired labor, in order to avoid problems with censoring at zero. In order to account for differences in the productivity of family labor and hired labor, you could use the share of hired labor as additional explanatory variable (see, e.g. http://purl.umn.edu/114548).

2b) I guess that the price of hired labor is *unknown* (and *not* zero) for some observations (e.g. the observations that do not use hired labor). You could use "multiple imputation" (e.g. using R package "mi") to impute the missing values.

Best regards,
Arne

Dummy Variables in MicEconSNQP [ Reply ]
By: Sami Ullah on 2015-10-03 16:14

[forum:42602]

Dear Arne

I have two questions,
1) I have to ask about addition of dummy variable in SNQP function formula in micEconSNQP package. How could I add dummies in snqProfitEst command.

2) Some of my variables have zero (0) values. For example; few of the respondents in my data do not use hired labour, so quantity of hired labour is zero in that case. How should I deal with it, either replacing labour quantity with one or with mean value. Same is the the case of price of hired labour.
I shall be thankful for your response.

Regards,
Sam