Hello everyone,

I am estimating a cross-country IV regression that uses internal instruments which exploit the presence of heteroskedasticity in the model’s residuals. My variables of interest are bank credit and bank credit^2(squared). I want to find the point estimate where the effect of bank credit on economic growth becomes negative (e.g. when bank credit reaches 90% of GDP). I tried nlcom command but it does work with it. Can somebody recommend me another option?

Note: Bank credit is measured as total credit extended to the private sector divided with GDP

ivreg2h gr linitial log_infl log_trade log_govsize log_school ( prcreditB prcreditB2=) , gmm2s robust

Too few excluded instruments: standard IV model not estimable

IV with Generated Instruments only

Instruments created from Z:
linitial log_infl log_trade log_govsize log_school

2-Step GMM estimation
---------------------

Estimates efficient for arbitrary heteroskedasticity
Statistics robust to heteroskedasticity

Number of obs = 83
F( 7, 75) = 18.74
Prob > F = 0.0000
Total (centered) SS = 156.8713703 Centered R2 = 0.4629
Total (uncentered) SS = 537.6248214 Uncentered R2 = 0.8433
Residual SS = 84.26147015 Root MSE = 1.008

------------------------------------------------------------------------------
| Robust
gr | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
prcreditB | 6.83556 1.817279 3.76 0.000 3.273759 10.39736
prcreditB2 | -3.64533 .8717815 -4.18 0.000 -5.35399 -1.936669
linitial | -1.001398 .1821731 -5.50 0.000 -1.358451 -.6443452
log_infl | .0567211 .1284435 0.44 0.659 -.1950236 .3084658
log_trade | .3784561 .1323573 2.86 0.004 .1190405 .6378717
log_govsize | -.9833192 .3495389 -2.81 0.005 -1.668403 -.2982355
log_school | 1.779788 .4935776 3.61 0.000 .8123935 2.747182
_cons | 6.083197 .9213752 6.60 0.000 4.277335 7.889059
------------------------------------------------------------------------------
Underidentification test (Kleibergen-Paap rk LM statistic): 17.347
Chi-sq(9) P-val = 0.0435
------------------------------------------------------------------------------
Weak identification test (Cragg-Donald Wald F statistic): 3.441
(Kleibergen-Paap rk Wald F statistic): 6.071
Stock-Yogo weak ID test critical values: 5% maximal IV relative bias 18.76
10% maximal IV relative bias 10.58
20% maximal IV relative bias 6.23
30% maximal IV relative bias 4.66
10% maximal IV size 29.32
15% maximal IV size 16.16
20% maximal IV size 11.65
25% maximal IV size 9.31
Source: Stock-Yogo (2005). Reproduced by permission.
NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors.
------------------------------------------------------------------------------
Hansen J statistic (overidentification test of all instruments): 12.227
Chi-sq(8) P-val = 0.1414
------------------------------------------------------------------------------
Instrumented: prcreditB prcreditB2
Included instruments: linitial log_infl log_trade log_govsize log_school
Excluded instruments: prcreditB_linitial_g prcreditB_log_infl_g
prcreditB_log_trade_g prcreditB_log_govsize_g
prcreditB_log_school_g prcreditB2_linitial_g
prcreditB2_log_infl_g prcreditB2_log_trade_g
prcreditB2_log_govsize_g prcreditB2_log_school_g
------------------------------------------------------------------------------