Interpretation of xtabond2 output

[andrea costa]

Dear Statalist-users,

I am using Stata 13.0 to estimate the following equation

dlogtfpt=b0+b1logtfpt-1+b2logFDI+b3log(FDI*X)+b4logX+eijt

where tfp is the total factor productivity
log is the natural logarithm
FDI is the variable of foreign presence
X is the set of control variables
and FDI*X are the interaction variables

code:

xi: xtabond2 d.ltw l.ltw l(0/2).lghor l(0/0).(lhohfd lhotg lhors
lhordf) (lgkl lgrd lgrdf lgtg lgsi lghfd i.year) if duf==0,
gmmstyle(l.ltw l2.d.ltw l.(lgkl lgtg lhohfd lghor), lag (8 8)c) gmm(lgsi,
lag(12 12)c) iv(l2.d.ltw lgrd lgrdf lghfd lhotg lhordf lhors i.year,
equation (level)) twostep robust



Stata output is:

Group variable: plantid Number of obs 46850
Time variable : year Number of groups 4685
Number of instruments = 30 Obs per group: min 10
Wald chi2(26) = 3589.87 avg 10.00
Prob > chi2 = 0.000 max 10



D.ltw Coef. Std. Err. z P>z [95% Conf. Interval]
ltw
L1. -1.377941 .0659355 -20.90 0.000 -1.507172 -1.24871
lghor

--. 0 (omitted)
L1. .0939822 .0192933 4.87 0.000 .0561681 .1317963
L2. .0190475 .0086882 2.19 0.028 .0020189 .036076

lhohfd 0 (omitted)
lhotg -.0880144 .0205217 -4.29 0.000 -.1282361 -.0477926
lhors .0279592 .0059063 4.73 0.000 .016383 .0395354
lhordf -.0139597 .0171669 -0.81 0.416 -.0476063 .0196868
lgkl .8065759 .0415438 19.42 0.000 .7251515 .8880003
lgrd 0 (omitted)
lgrdf 0 (omitted)
lgtg -.0130485 .0495371 -0.26 0.792 -.1101395 .0840425
lgsi 0 (omitted)
lghfd .370614 .0270918 13.68 0.000 .3175149 .423713
_Iyear_1998 5.372316 .4395289 12.22 0.000 4.510855 6.233776
_Iyear_1999 5.444025 .4451601 12.23 0.000 4.571527 6.316522
_Iyear_2000 5.393613 .445117 12.12 0.000 4.5212 6.266027
_Iyear_2001 5.3964 .4413826 12.23 0.000 4.531306 6.261494
_Iyear_2002 5.375868 .4400378 12.22 0.000 4.51341 6.238327
_Iyear_2003 5.396753 .4407484 12.24 0.000 4.532902 6.260604
_Iyear_2004 5.459598 .446981 12.21 0.000 4.583531 6.335665
_Iyear_2005 5.415188 .4475532 12.10 0.000 4.537999 6.292376
_Iyear_2006 5.394337 .4426592 12.19 0.000 4.526741 6.261933
_Iyear_2007 5.397805 .4430694 12.18 0.000 4.529405 6.266205



Am I correct to assume to interpret the results as the Total Factor productivity being in first differences and the independent variables being in levels, and thus the interpretation of xtabond2 output is the percentage point increase in the TFP due to a 1% increase in the independent variables?

Or, what I get is the second differences for the TFP and the first differences for the independent variables?

Thank you in advance!

Andrea Costa

[Carlo Lazzaro]

Andrea:
welcome to the list.
Two asides off the main target:
- you do not have to use -xi- prefix with -fvvarlist- notation: is old-fashioned and redundant;
- in order to make what you typed and what Stata gave you back readable to those interested in help you out, please put all the stuff within CODE delimiters (see the FAQ on that and other relevant topics about posting more efficiently). Thanks.

[andrea costa]

Thank you Carlo. Will do.


Kind regards,

Andrea

[andrea costa]

Dear Statalist-users,

I am using Stata 13.0 to estimate the following equation

dlogtfpt=b0+b1logtfpt-1+b2logFDI+b3log(FDI*X)+b4logX+eijt

where tfp is the total factor productivity
log is the natural logarithm
FDI is the variable of foreign presence
X is the set of control variables
and FDI*X are the interaction variables

xtabond2 d.ltw l.ltw l(0/2).lghor l(0/0).(lhohfd lhotg lhors
lhordf) (lgkl lgrd lgrdf lgtg lgsi lghfd i.year) if duf==0,
gmmstyle(l.ltw l2.d.ltw l.(lgkl lgtg lhohfd lghor), lag (8 8)c) gmm(lgsi,
lag(12 12)c) iv(l2.d.ltw lgrd lgrdf lghfd lhotg lhordf lhors i.year,
equation (level)) twostep robust
   
Group variable: plantid

Number of obs
46850

Time variable : year

Number of groups
4685

Number of instruments = 30

Obs per group: min
10

Wald chi2(26) = 3589.87

avg
10.00

Prob > chi2 = 0.000

max
10

  D.ltw
         Coef.
    Std. Err.
   z
  P>z
              [95% Conf. Interval]

ltw







L1.
-1.377941
.0659355
-20.90
0.000
-1.507172
-1.24871

lghor

  			






--.
0
(omitted)





L1.
.0939822
.0192933
4.87
0.000
.0561681
.1317963

L2.
.0190475
.0086882
2.19
0.028
.0020189
.036076









lhohfd
0
(omitted)





lhotg
-.0880144
.0205217
-4.29
0.000
-.1282361
-.0477926

lhors
.0279592
.0059063
4.73
0.000
.016383
.0395354

lhordf
-.0139597
.0171669
-0.81
0.416
-.0476063
.0196868

lgkl
.8065759
.0415438
19.42
0.000
.7251515
.8880003

lgrd
0
(omitted)





lgrdf
0
(omitted)





lgtg
-.0130485
.0495371
-0.26
0.792
-.1101395
.0840425

lgsi
0
(omitted)





lghfd
.370614
.0270918
13.68
0.000
.3175149
.423713

_Iyear_1998
5.372316
.4395289
12.22
0.000
4.510855
6.233776

_Iyear_1999
5.444025
.4451601
12.23
0.000
4.571527
6.316522

_Iyear_2000
5.393613
.445117
12.12
0.000
4.5212
6.266027

_Iyear_2001
5.3964
.4413826
12.23
0.000
4.531306
6.261494

_Iyear_2002
5.375868
.4400378
12.22
0.000
4.51341
6.238327

_Iyear_2003
5.396753
.4407484
12.24
0.000
4.532902
6.260604

_Iyear_2004
5.459598
.446981
12.21
0.000
4.583531
6.335665

_Iyear_2005
5.415188
.4475532
12.10
0.000
4.537999
6.292376

_Iyear_2006
5.394337
.4426592
12.19
0.000
4.526741
6.261933

_Iyear_2007
5.397805
.4430694
12.18
0.000
4.529405
6.266205

Am I correct to interpret the results as the Total Factor productivity (TFP) being in first differences and the independent variables being in levels, and thus the interpretation of xtabond2 output is the percentage point increase in the TFP due to a 1% increase in the independent variables?

Or, what I get is the second differences for the TFP and the first differences for the independent variables?

Thank you in advance!

Andrea Costa