Hi Stata users
I'm running two panel data looking at impact of insurance (as well as banking, stock market) on economic growth (GDP growth - gdpg). This is similar to the work of Levine, Beck & Loayza (2000) "Finance intermediation and growth: causality and causes", and Arena (2006) "Does insurance market activity promote economic growth?".
Annual panel, T=26, N=60 --> xtabond difference GMM
Five-year average panel: T=5, N=59 --> xtabond2 system GMM
Both are unbalanced panel (due to lack of data for all the years).
I use GDP growth (gdpg calculated by difference in log real GDP per capita) as dependent variable.
Expected signs for independent variables (insurance premium/GDP (penetration), banking, stock market turnover) are positive.
However, always get negative sign for all the independent variables in xtabond (annual) and insignificant results in xtabond2 (five year average). Sargan test and A-B second order correlation test are okie (accepted)for both.
. xtabond gdpg pene_t bank stock lgdpi
> govt linf open tot, lags(1) maxldep(25)
> twostep artests(2);
Arellano-Bond dynamic panel-data estimation Number of obs = 902
Group variable (i): country Number of groups = 64
Wald chi2(9) = 4598.67
Time variable (t): year Obs per group: min = 1
avg = 14.09375
max = 25
Two-step results
------------------------------------------------------------------------------
D.gdpg | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
gdpg |
LD. | .0951484 .0108329 8.78 0.000 .0739163 .1163804
pene_t |
D1. | -.359937 .0477571 -7.54 0.000 -.4535392 -.2663349
bank |
D1. | -.0373582 .0049504 -7.55 0.000 -.0470608 -.0276556
stock |
D1. | .005358 .0004958 10.81 0.000 .0043863 .0063297
lgdpi |
D1. | .0735472 .0071024 10.36 0.000 .0596267 .0874676
govt |
D1. | -.0795676 .0095947 -8.29 0.000 -.0983728 -.0607624
linf |
D1. | -.0137611 .0022353 -6.16 0.000 -.0181422 -.00938
open |
D1. | .0097536 .0041788 2.33 0.020 .0015633 .0179439
tot |
D1. | .0149388 .0026216 5.70 0.000 .0098005 .0200771
_cons | -.0007371 .0001217 -6.05 0.000 -.0009757 -.0004984
------------------------------------------------------------------------------
xtabond2:
xtabond2 pgdpg L.pgdpg L(0/1).(ppene_t pbank
> plgdpi popen pgovt plinf ptot) period2-period5
> , gmm(l.pgdpg ppene_t pbank)
> iv(period2-period5) robust twostep artests(2);
Dynamic panel-data estimation, two-step system GMM
------------------------------------------------------------------------------
Group variable: country Number of obs = 183
Time variable : year Number of groups = 57
Number of instruments = 39 Obs per group: min = 1
Wald chi2(18) = 77.62 avg = 3.21
Prob > chi2 = 0.000 max = 4
------------------------------------------------------------------------------
| Corrected
pgdpg | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
pgdpg |
L1. | -.9711892 .6954424 -1.40 0.163 -2.334231 .3918528
ppene_t |
--. | .0131906 .1759554 0.07 0.940 -.3316756 .3580568
L1. | .0052757 .1954426 0.03 0.978 -.3777847 .388336
pbank |
--. | .0021213 .0054946 0.39 0.699 -.008648 .0128906
L1. | -.0058175 .0093167 -0.62 0.532 -.0240779 .0124429
plgdpi |
--. | .0727495 .054013 1.35 0.178 -.0331141 .1786131
L1. | -.0751615 .056488 -1.33 0.183 -.1858759 .0355529
popen |
--. | .0178883 .0125482 1.43 0.154 -.0067058 .0424824
L1. | -.0165761 .0136101 -1.22 0.223 -.0432514 .0100992
pgovt |
--. | -.182051 .1468107 -1.24 0.215 -.4697947 .1056927
L1. | .1615535 .1007263 1.60 0.109 -.0358663 .3589734
plinf |
--. | -.0639689 .0229701 -2.78 0.005 -.1089895 -.0189483
L1. | -.0102448 .0122008 -0.84 0.401 -.0341579 .0136684
ptot |
--. | .2249072 .2673885 0.84 0.400 -.2991648 .7489791
L1. | -.0815871 .1382154 -0.59 0.555 -.3524844 .1893101
period2 | .000801 .0038174 0.21 0.834 -.006681 .008283
period3 | .0033025 .0034863 0.95 0.343 -.0035304 .0101355
period4 | -.0024695 .0020707 -1.19 0.233 -.0065281 .0015891
_cons | .0399558 .0255705 1.56 0.118 -.0101614 .0900731
------------------------------------------------------------------------------
I wonder whether anyone has experienced this before? I have tried to separate the countries into group of developed & developing countries, but still same result. Also I wonder whether my xtabond2 is structured correctly.
Or is it because my N is too small?
Thanks a lot,
Yen Nguyen
Monash University
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
