Dear statalists:
I want to use prodest to estimate TFP,but my production function is a translog-form,as the following equation showing:
lny=w+a*lnl+b*lnk+(lnl)^2+c*(lnk)^2+d*lnl*lnk
The code is as fllowing,I have found that tfp calculated by prodest+acf+translog is totally different from tfp calculated by prodest+acf,the former tfp is really too big.I wonder if you can figure mistakes in my code.
----------------------- copy starting from the next line -----------------------
------------------ copy up to and including the previous line ------------------
I want to use prodest to estimate TFP,but my production function is a translog-form,as the following equation showing:
lny=w+a*lnl+b*lnk+(lnl)^2+c*(lnk)^2+d*lnl*lnk
The code is as fllowing,I have found that tfp calculated by prodest+acf+translog is totally different from tfp calculated by prodest+acf,the former tfp is really too big.I wonder if you can figure mistakes in my code.
----------------------- copy starting from the next line -----------------------
Code:
* Example generated by -dataex-. To install: ssc install dataex
clear
insheet using "https://raw.githubusercontent.com/GabrieleRovigatti/prodest/master/stata/data/prodest.csv", names clear
prodest log_y, method(lp) free(log_lab1 log_lab2) proxy(log_materials) state(log_k) valueadded acf translog id(id) t(year) reps
> (50)
.........10.........20.........30.........40.........50
lp productivity estimator translog PF
ACF corrected
Dependent variable: value added Number of obs = 2544
Group variable (id): id Number of groups = 497
Time variable (t): year
Obs per group: min = 1
avg = 5.1
max = 11
------------------------------------------------------------------------------
log_y | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
log_lab1 | .887048 .0024949 355.54 0.000 .8821581 .8919379
log_lab2 | .7611732 .0016125 472.05 0.000 .7580128 .7643337
log_k | -.5397494 .0018904 -285.52 0.000 -.5434546 -.5360443
var_1_1 | .1001688 .0032244 31.07 0.000 .0938492 .1064885
var_1_2 | .0163688 .0024564 6.66 0.000 .0115544 .0211832
var_1_3 | -.0623013 .0015516 -40.15 0.000 -.0653424 -.0592602
var_2_2 | .0797299 .002452 32.52 0.000 .0749241 .0845357
var_2_3 | -.0491968 .0015186 -32.40 0.000 -.0521732 -.0462204
var_3_3 | .0410398 .0020477 20.04 0.000 .0370265 .0450532
------------------------------------------------------------------------------
Wald test on Constant returns to scale: Chi2 = 4205.71
p = (0.00)
Estimated parameters displayed. To see estimated input elasticities, type predict, parameters
. predict tfp_translog,residual
. prodest log_y, method(lp) free(log_lab1 log_lab2) proxy(log_materials) state(log_k) valueadded acf id(id) t(year) reps(50)
.........10.........20.........30.........40.........50
lp productivity estimator Cobb-Douglas PF
ACF corrected
Dependent variable: value added Number of obs = 2544
Group variable (id): id Number of groups = 497
Time variable (t): year
Obs per group: min = 1
avg = 5.1
max = 11
------------------------------------------------------------------------------
log_y | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
log_lab1 | .5486727 .0662128 8.29 0.000 .418898 .6784474
log_lab2 | .4122355 .1238101 3.33 0.001 .1695721 .6548989
log_k | .4765666 .0766743 6.22 0.000 .3262877 .6268455
------------------------------------------------------------------------------
Wald test on Constant returns to scale: Chi2 = 43.36
p = (0.00)
. predict tfp_acf,residual
.
. ****the tfp calculated by prodest+acf+translog is too big,I wonder if you can tell me causes.
. su tfp_translog tfp_acf
Variable | Obs Mean Std. Dev. Min Max
-------------+---------------------------------------------------------
tfp_translog | 2,544 16.4528 1.713113 9.93541 24.53713
tfp_acf | 2,544 5.779031 .8841653 2.97516 9.391429
end
