Please Help Me...

[Henry Nguyen]

I use "Qregpd" and run the equation follow:
equation 1:
qregpd e_ta_bk mblag lagtang proflag lagsize lagblev, id(id) fix(year) optimize(mcmc) noisy draws(1000) burn(100) arate(0.5) quantile(.1)
Result:

qregpd e_ta_bk mblag lagtang proflag lagsize lagblev, id(id) fix(year) optimize(mcmc) noisy draws(1000) burn(100) arate(0.5) quantile(.1)
Adaptive MCMC optimization
................................................. 50: f(x) = -491.682055
................................................. 100: f(x) = -341.294361
................................................. 150: f(x) = -329.841774
................................................. 200: f(x) = -325.344571
................................................. 250: f(x) = -310.753205
................................................. 300: f(x) = -304.755536
................................................. 350: f(x) = -267.094713
................................................. 400: f(x) = -258.626104
................................................. 450: f(x) = -256.360916
................................................. 500: f(x) = -247.309495
................................................. 550: f(x) = -232.739689
................................................. 600: f(x) = -229.698686
................................................. 650: f(x) = -229.698686
................................................. 700: f(x) = -227.903506
................................................. 750: f(x) = -227.903506
................................................. 800: f(x) = -230.524269
................................................. 850: f(x) = -226.760101
................................................. 900: f(x) = -228.97304
................................................. 950: f(x) = -228.97304
................................................. 1000: f(x) = -228.899609


Quantile Regression for Panel Data (QRPD)
Number of obs: 36108
Number of groups: 3454
Min obs per group: 1
Max obs per group: 17

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

mblag .0003999 .000106 3.77 0.000 .0001921 .0006078
lagtang .0235918 .0030777 7.67 0.000 .0175597 .0296239
proflag .0347651 .0027136 12.81 0.000 .0294465 .0400836
lagsize -.0026297 .000184 -14.29 0.000 -.0029904 -.002269
lagblev .0023893 .0011837 2.02 0.044 .0000693 .0047093

No excluded instruments - standard QRPD estimation.


MCMC diagonstics:
Mean acceptance rate: 0.074
Total draws: 1000
Burn-in draws: 100
Draws retained: 900
Value of objective function:
Mean: -258.4611
Min: -344.4691
Max: -226.7601
MCMC notes:
*Point estimates correspond to mean of draws.
*Standard errors are derived from variance of draws.

and equation 2
qregpd c1bk mblag lagtang proflag lagsize lagblev, id(id) fix(year) optimize(mcmc) noisy draws(1000) burn(100) arate(0.5) quantile(.1)
note that -(e_ta_bk) = c1bk
qregpd c1bk mblag lagtang proflag lagsize lagblev, id(id) fix(year) optimize(mcmc) noisy draws(1000) burn(100) arate(0.5) quantile(.1)
Adaptive MCMC optimization
................................................. 50: f(x) = -1485.30362
................................................. 100: f(x) = -910.460328
................................................. 150: f(x) = -774.740706
................................................. 200: f(x) = -636.863033
................................................. 250: f(x) = -562.688539
................................................. 300: f(x) = -416.551356
................................................. 350: f(x) = -393.861994
................................................. 400: f(x) = -394.787211
................................................. 450: f(x) = -383.245927
................................................. 500: f(x) = -383.245927
................................................. 550: f(x) = -383.245927
................................................. 600: f(x) = -381.003299
................................................. 650: f(x) = -370.73573
................................................. 700: f(x) = -366.675186
................................................. 750: f(x) = -350.241635
................................................. 800: f(x) = -308.686663
................................................. 850: f(x) = -279.434892
................................................. 900: f(x) = -266.704798
................................................. 950: f(x) = -237.271468
................................................. 1000: f(x) = -234.880424


Quantile Regression for Panel Data (QRPD)
Number of obs: 36108
Number of groups: 3454
Min obs per group: 1
Max obs per group: 17
------------------------------------------------------------------------------
c1bk | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
mblag | .0179996 .0018409 9.78 0.000 .0143914 .0216077
lagtang | -.3846238 .0184168 -20.88 0.000 -.4207201 -.3485275
proflag | -.7282647 .0483179 -15.07 0.000 -.822966 -.6335633
lagsize | .0768478 .0073139 10.51 0.000 .0625128 .0911827
lagblev | -.4521848 .0147003 -30.76 0.000 -.4809969 -.4233727
------------------------------------------------------------------------------
No excluded instruments - standard QRPD estimation.


MCMC diagonstics:
Mean acceptance rate: 0.067
Total draws: 1000
Burn-in draws: 100
Draws retained: 900
Value of objective function:
Mean: -411.2478
Min: -910.4603
Max: -234.2407
MCMC notes:
*Point estimates correspond to mean of draws.
*Standard errors are derived from variance of draws.

But I get the results very different. While I check the result in reg pool, xtreg (fixed effect, random effect) it just changes the sign of independent variables.

Another thing, after I run quantile regression again (qreg, xtqreg, qregpd), I'll get the other results...

I really don't understand why. Does anyone can help me, please?
Thanks so much for any your reply.

[Nick Cox]

if you don't get a response, do

* start a new thread making qregpd as part of the title

* close this one with a link to the new thread.

* explain where qregpd comes from.

(The question title is not good: almost all threads imply the same request.)

[Henry Nguyen]

Dear Professor Nick Cox,
Thank so much for your response.
Best Wishes,