Feb 5, 2008 - Y_star = Z*delta + e ... When writing the likelihood function, the term (alpha*Z*delta) ... scalar s_uet = (alpha^2)*s_u^2+s_e^2+2*alpha*s_ue.
st: Simultaneous Tobit Models with ml From To Subject Date
"Olivier Paradis-Beland" st: Simultaneous Tobit Models with ml Thu, 27 Nov 2003 12:44:55 -0500
Hi everybody. I want to estimate the following simultaneous Tobit model : H_star = alpha* y + X*beta + u Y_star = Z*delta + e Where H_star and Y_star are latent variables, both observed (i.e. as H and Y) when greater than zero. The difficulty with such a model is obviously that the (endogenous) regressor "y" is a latent variable. We denote "s_u", "s_e" and "s_eu" as the elements of the covariance matrix between error terms. There is indeed for regime in the model (1: H=0 and y=0) (2: H>0 and Y=0) (3: H=0 and Y>0) (4: H>0 and Y>0). Note that we have the following reduced form for H: H= X*beta + e alpha*Z*delta + X*beta + u_et 0
if if
H_star>0 and y=0 H_star>0 and y>0 otherwise
Note that the coefficient associated with "y" in the structural equation (alpha) enters the reduced form associated with "Z*delta" i When writing the likelihood function, the term (alpha*Z*delta) appears in regime 3 and regime 4 and "alpha" enters the Here is my program (note that I test the program by fixing alpha to an arbitrary constant value and it seemed ok) :
capture program drop maxlik program define maxlik args lnf beta delta sig_u sig_e sig_ue quietly capture dropvars xbeta zdelta gen double xbeta=`beta' gen double zdelta=`delta' scalar s_u = exp(`sig_u') scalar s_e = exp(`sig_e') scalar s_ue = exp(`sig_ue') scalar rho = tanh(s_ue/(s_u*s_e)) scalar alpha = xbeta[1,1] scalar s_uet = (alpha^2)*s_u^2+s_e^2+2*alpha*s_ue scalar rho_et = (alpha*s_u^2+s_ue)/(sqrt(s_u^2*(alpha^2*s_u^2+s_e^2+2*alpha*s_ue))) scalar gamma = alpha*zdelta quietly{ capture dropvars regime_1 regime_2 regime_3 regime_4 gen double regime_1 = 0.0 replace regime_1 = binorm(-xbeta/s_e,-zdelta/s_u,rho) if $ML_y1==0 & $ML_y2==0 gen double regime_2 = 0.0 replace regime_2 = norm((-zdelta/s_u-rho*($ML_y1-xbeta)/s_e)/sqrt(1-rho^2))*1/s_e*normden(($ML_y1-xbeta)/s_e) if $ML_y1>0 & $ML_y2== gen double regime_3 = 0.0 replace regime_3 = norm((-gamma-xbeta-rho_et*(s_uet/s_u)*($ML_y2-zdelta))/s_uet*sqrt(1-rho_et^2))*1/s_u*normden(($ML_y2 gen double regime_4 = 0.0 replace regime_4 = (1/(2*_pi*s_u*s_uet*sqrt(1-rho_et^2)))*exp((-1/2*(1-rho_et^2))*( (($ML_y2-zdelta)/s_u)^2-2*rho_et*(($ML_y2 } quietly replace `lnf' = ln(regime_1) if $ML_y1==0 & $ML_y2==0 quietly replace `lnf' = ln(regime_2) if $ML_y1>0 & $ML_y2==0 quietly replace `lnf' = ln(regime_3) if $ML_y1==0 & $ML_y2>0 quietly replace `lnf' = ln(regime_4) if $ML_y1>0 & $ML_y2>0 end use scf86_roc.dta, clear ml model lf maxlik (beta: heures=revimp tn ipe ne nb man sas alb cb age_16_24 age_25_34 age_35_44 age_55_69 homme celib ml maximize I know all this is a bit confused. Feel free to ask for more precisions. Olivier * * * * *
For searches and help try: http://www.stata.com/support/faqs/res/findit.html http://www.stata.com/support/statalist/faq http://www.ats.ucla.edu/stat/stata/
Prev by Date: Re: st: RE: Formaatin the labels of the histogram-bars ? Next by Date: st: RE: Truncating x-axis in kernel density graphs Previous by thread: st: logarithmic scales Next by thread: st: RE: Truncating x-axis in kernel density graphs Index(es): Date Thread
formulas for their stationary species abundance distribution. Here we ... Keywords Biodiversity · Ewens sampling formula · Metacommunity · Neutral model ·.
Traditional road safety is a reactive approach, based on historical ... Direct extrapolation method is difficult because of ... based clustering methods. â ex: leading ...
Improving models of forest nutrient export with equations that predict the nutrient .... production of each tree compartment with its mean nutrient concentration. ...... crowns with application to modelling the foliage area distribution in the crown.
and to observe the adsorption of proteins in liquid phase with these two .... adsorption of antibodies at a concentration of 10 g/ml con- centrations could result in ...
larger than the typical depth, the free surface problem is frequently ... λ the typical wavelength of the waves, by asurf their typical amplitude and by abott.
environment is viewed as a communications channel over which a message .... weight vectors for the output units of the network, and the columns of Ware.
chitectures: mixtures of Gaussians and locally weighted regression. .... brie y review how the statistical approach can be applied to neural networks, as described.
23 janv. 2015 - Contact your local waste disposal department to dispose of the device/batteries in accordance with applicable local laws and regulations. Product Environmental Programs. Information about the Garmin product recycling program and. WEEE
square root of the number of observations, here 23. ... From the table we see that the single equation probit that completely ignores endogeneity has a.
tering or feature extraction on the visual input is not sufficient. Moreover, it is difficult to see ...... Discovering the structure of a reactive environment by exploration.
Neuroimaging techniques can be used not only to identify the neural substrates of attention, but also to test cognitive ... which fMRI and MEG can provide key evidence in answering these questions. ...... cific activity at 160 msec: An MEG study.
Thus, first some background materials and tools are pre- sented briefly and progressively, some new or at least state of the art materials follows: - In section 2 the ...
Refinement Programming by refinement consists of getting an executable pro- .... Some examples Following the above intuitive definition of an interpretation, the .... Mathematical preliminaries Here we introduce some standard notions and ...
Gearbox capacrty. Capacit6 du groupe hydraulique. Aggregatleistung pro 1'. Velocita stelo 1' mm. Rod sp€ed. Vrtesse de Ia tige. Hub Geschwindino- keit mm/1'.
+385-1-3091116. Uvoznik za BiH: GNLD International d.o.o.,. Butmirska cesta 31,71210 Ilidža, Bosna i. Hercegovina. Tel: +387-33-761670. SRB Gel za čišćenje ...
One efficient multi-fusion strategy is to detect targets with one sensor, which are then ... black object of intrinsic luminance L0, the apparent lumi- nance of the ...