Help regarding TFP calculation

[Daniyal Ahmed]

I am calculating TFP using OLS and I took sales as output, and (salaries of all workers, short term investment, stock trading in market, long term investment and net assets) as input. After running regression the results error term was calculated and a variable of tfp was calculated. The equation is as follows
gen tfp1 = lSales- (0.1041)*( lsalaries ) -(0.0230)*( lshort_inv )-(0.396)*( lstock_trading)-(-0.0080)*( lLong_inv)-(0.217)*( lnet_assets)

But the TFP comes out to be negative for a number of observations. Please help

[Nick Cox]

I have no idea what TFP is (we're really not all economists in your field) but if you're seeking an empirical equation of a quantity that can't be negative, use a logarithmic link function. Poisson regression should work well, for example, even with measured variables.

[Daniyal Ahmed]

Thanks