LECTURE 2
Useful Numerical Methods
In many Bayesian applications, one cannot analytically solve for the full posterior
density as we did in our simple example. In that case, we might be able to generate a
random sample from the posterior to draw inference
Lecture 3
The Linear Regression Model
For many econometric models, Gibbs sampling algorithms allows us to sample from
the conditional distributions of the posterior and draw inferences. The key task then is to
write out the conditional posteriors for each
Economics 7632
Terrell
Lecture 5
Seemingly Unrelated Regressions (SUR)
Consider a set of n equations of the form,
y1 = x1 1 + u1
y 2 = x 2 2 + u2
M
y n = xn n + un
,
where
yi are Nx1 vectors of observations for the dependent variable in equation i,
xi are
Lecture 4
The Stochastic Frontier Model
The goal of the stochastic frontier model is first to estimate the cost function of an
efficient firm in an industry, which is called the cost frontier. Deviations from that frontier are
used to measure inefficiency
Economics 7632
Lecture 1
You begin econometrics or statistics by noting that there are three views of probability:
1. Classical - assumes equally probable outcomes, P(success)= #success/#outcomes
Example: Die roll, six possible outcomes.
P(1)=1/6.
P(odd)=
1RWHV
&HQVRUHG 5HJUHVVLRQ 7RELW 0RGHOV
&RQVLGHU WKH EDVLF PRGHO
y i* = ' xi + i , where i ~ N (0, 2 )
0 if y i* 0
yi = *
*
y i if y i > 0.
7KH YDULDEOH \ LV D
ODWHQW YDULDEOH
HFRQRPHWULFLDQ LQVWHDG REVHUYHV WKH
H[FHHGV DQG RWKHUZLVH
XQREVHUYHG E\ WKH H