[Jojo Melchi]

Prof Huang,
thank you for your advices. I face the same problem with the "xthreg" command. There are still error messages as follow:

 xthreg DZ L(1/0).(gdp slope_10y_2y_res change_sm dhp) L.(change_fsi Liq Eff exlendg exlendg2) s1-s3, rx(L.change_f
> edrate) qx(L.tgap5) th(1) bs(300) trim(0.01) nobslog robust
Estimating  the  threshold  parameters:   1st ......  Done
Boostrapping for threshold effect test:   1st ......                 bgrid():  3200  conformability error
                thestm():     -  function returned error
                 <istmt>:     -  function returned error
r(3200);

Could you help to detect where the problem is?
Thank you in advance

[River Huang]

Please visit https://alexbick.weebly.com/publications.html, and check out the bottom three papers (and Gauss code). I don't exactly recall which one is the one you need.

[sedki zn]

Dear River Huang

i'm studying the effect on CEO stock option pay on firm performance and i would aply the threshold regression for my panel data. Should i find the fixed effect to use "xthreg" ? and if i is the random effect ? what should i do ?

Best regards
Sedki

[River Huang]

Dear sedki, Arguably, it is relatively uncommon to use random effect approaches in corporate finance. Given the fact that, the fixed effect estimators are consistent, irrespective of whether there is correlation between the explanatory variable and the individual-specific effects, so please proceed to use "xthreg" command.

[sedki zn]

Thank you very much Dear Huang

Best regards
Sedki

[sedki zn]

Dear Huang,

when i try to install "xthreg", i get this message
what wrong with it ? ( i'm using stata 12.0)

best regards
Sedki

[sedki zn]

dear all,

any help please about my last comment ?

thank you

[sedki zn]

Dear all

please i need your help about the means of threshold variables ;
xtreg i q1 q2 q3 d1 qd1, rx(c1) qx(d1)
c1 and d1, especially

best regards
sedki

[Mario Bustamante]

Please drop `unemp_current' from the linear part.

xthreg reelect nom unemp_prev, rx(unemp_current) qx(unemp_current) thnum(1) trim(0.05)

What happen if we use the same variable in rx and qx? because Hansen use different

[Tamara Esther]

Hello,

I have been following the conversation closely. I wanted clarification on what rx and qx stand for. My understanding is that one is the threshold variable and the other is the regime dependent variable. In one example above these are treated as the same thing. Kindly assist , when I try to put different variables I get an error. However, when I put the same variable for rx and qx I get results. Its this approach correct

Secondly, I am not sure how to interpret the last portion of the results which show the threshold variable as a dummy.

regards

Tamara

[River Huang]

Hi, Mario: It is perfectly OK that the same variable can be both the regime-switching as well as the threshold variable.

[Le Trung]

Dear Professor River Huang

Is there any problem when i do not set threshold variable as one among Independent-Regime Variables or Dependent-Regime Variables (means Threshold variable is out of model and has not any coeffient) !

When i tried run model with this case, Stata still had estimation results .

Thank you, Prof.,

[River Huang]

Dear Le, It is perfectly OK.

[Le Trung]

Sorry Professor River Huang for lated Thanks !

My model PTR has 02 thresholds. Pls help me to interprete the results.

Which are proper first threshold, second threshold of OEAR and proper Dependent-Regime coefficients of OEAR that Lower, Between and Higher of Regimes follow by 02 estimated Thresholds ?

1. Result for estimate 1st threshold

xthreg ROE S CA DIA LDR LPCLR MSG GDPG INF CON, rx(OEAR) qx(S) thnum(1) trim(0.01) grid(400) bs(300)

Threshold estimator (level = 95):
-----------------------------------------------------
model | Threshold Lower Upper
-----------+-----------------------------------------
Th-1 | 10.3320 10.3031 10.3762
-----------------------------------------------------

Threshold effect test (bootstrap = 300):
-------------------------------------------------------------------------------
Threshold | RSS MSE Fstat Prob Crit10 Crit5 Crit1
-----------+-------------------------------------------------------------------
Single | 0.6734 0.0026 42.57 0.0000 17.4830 19.9575 25.7016
-------------------------------------------------------------------------------

Fixed-effects (within) regression Number of obs = 270
Group variable: BANK Number of groups = 30

R-sq: within = 0.4560 Obs per group: min = 9
between = 0.2938 avg = 9.0
overall = 0.2884 max = 9

F(11,229) = 17.45
corr(u_i, Xb) = -0.7261 Prob > F = 0.0000

------------------------------------------------------------------------------
ROE | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
S | .0580283 .0149929 3.87 0.000 .0284866 .0875701
CA | .5021051 .1565733 3.21 0.002 .1935966 .8106135
DIA | 3.42897 .9265167 3.70 0.000 1.603382 5.254557
LDR | .0365282 .0213872 1.71 0.089 -.0056126 .0786689
LPCLR | -1.816878 .6756829 -2.69 0.008 -3.148228 -.4855277
MSG | .1773937 .0805198 2.20 0.029 .0187393 .3360481
GDPG | -.8211841 .8134514 -1.01 0.314 -2.42399 .7816221
INF | .6927778 .1506911 4.60 0.000 .3958595 .9896961
CON | .8981131 .3037999 2.96 0.003 .2995126 1.496714
|
_cat#c.OEAR |
0 | -8.104964 .9996616 -8.11 0.000 -10.07467 -6.135253
1 | -3.056751 1.087291 -2.81 0.005 -5.199124 -.9143787
|
_cons | -1.088484 .3113797 -3.50 0.001 -1.702019 -.4749482
-------------+----------------------------------------------------------------
sigma_u | .06772773
sigma_e | .05427066
rho | .60897933 (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(29, 229) = 5.90 Prob > F = 0.0000

2. Result for estimate 2nd and 3th Thresholds based on given 1st threshold

xthreg ROE S CA DIA LDR LPCLR MSG GDPG INF CON, rx(OEAR) qx(S) thnum(3) trim(0.01 0.01 0.05) bs(0 300 300)thgiven

Threshold estimator (level = 95):
-----------------------------------------------------
model | Threshold Lower Upper
-----------+-----------------------------------------
Th-1 | 10.3320 10.3031 10.3762
Th-21 | 10.1220 10.1216 10.1321
Th-22 | 10.1109 10.0573 10.1138
Th-3 | 12.5578 12.5251 12.5638
-----------------------------------------------------

Threshold effect test (bootstrap = 0 300 300):
-------------------------------------------------------------------------------
Threshold | RSS MSE Fstat Prob Crit10 Crit5 Crit1
-----------+-------------------------------------------------------------------
Single | 0.6734 0.0026 42.57 0.0000 17.4830 19.9575 25.7016
Double | 0.3658 0.0014 219.43 0.0000 17.2635 20.9185 25.7644
Triple | 0.3494 0.0013 12.31 0.6367 47.1150 61.8000 87.2662
-------------------------------------------------------------------------------

Fixed-effects (within) regression Number of obs = 270
Group variable: BANK Number of groups = 30

R-sq: within = 0.4085 Obs per group: min = 9
between = 0.2592 avg = 9.0
overall = 0.2886 max = 9

F(13,227) = 12.06
corr(u_i, Xb) = -0.5801 Prob > F = 0.0000

------------------------------------------------------------------------------
ROE | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
S | .0686349 .0157334 4.36 0.000 .0376328 .0996371
CA | .0675752 .175757 0.38 0.701 -.2787486 .4138989
DIA | 3.231239 .9711021 3.33 0.001 1.317712 5.144766
LDR | .0401024 .0223915 1.79 0.075 -.0040193 .0842241
LPCLR | -.951709 .7029772 -1.35 0.177 -2.336904 .4334861
MSG | .1505416 .0847104 1.78 0.077 -.0163777 .3174609
GDPG | -.5918369 .863332 -0.69 0.494 -2.293006 1.109333
INF | .5153019 .1592973 3.23 0.001 .2014115 .8291923
CON | .7020234 .3192764 2.20 0.029 .0728991 1.331148
|
_cat#c.OEAR |
0 | -4.040604 1.241235 -3.26 0.001 -6.486419 -1.594788
1 | -7.746548 2.279599 -3.40 0.001 -12.23843 -3.254667
2 | -6.841496 1.01337 -6.75 0.000 -8.83831 -4.844682
3 | -9.602793 1.629526 -5.89 0.000 -12.81372 -6.391861
|
_cons | -1.033142 .329809 -3.13 0.002 -1.68302 -.3832632
-------------+----------------------------------------------------------------
sigma_u | .05409345
sigma_e | .05683923
rho | .47526345 (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(29, 227) = 4.55 Prob > F = 0.0000


Many thanks Prof
Le,

[River Huang]

Dear Le, This is what you should look at:

Th-21 | 10.1220 10.1216 10.1321
Th-22 | 10.1109 10.0573 10.1138