We are on the way to applying the r* estimation of Holston-Lawbach-Williams model and IMF WEO (2nd chapter of April 2023 WEO) to Japan's r*.
However, our results showed r* are unusually large negative numbers. Probably we should in advance solve the estimation biases that standard deviation of the innovations to z (component of r* other than trend growth) and g (trend growth rate) are biased toward zero (so-called the "pile-up problem" discussed in Stock 1994). HLW explained that they used Stock and Watson's (1998) median unbiased estimator to obtain estimates of the ratio λg=σg/σy* and λz=ar(σz/σy) by maximum likelihood. When estimating the entire signal equations and state space equations, how can we obtain those two variables in the first estimation by using maximum likelyfood estimation. We would appreciate if someone can provide the code of EVIEWS13 about this estimation.
Our models are as follows for your convenience. sv1=r*, sv2=z, sv3=y* (potential growth rate), and sv4=g. Attached please find the workfile.
@signal 100*Y = C(1)*100*(Y(-1)-SV3_1) + C(2)*100*(Y(-2)-SV3_2) + C(3)/2*(r(-1)-SV1_1)+C(3)/2*(r(-2)-SV1_2)+100*SV3
@signal pi = c(4)*pi(-1)+c(5)/3*pi(-2)+c(5)/3*pi(-3)+c(5)/3*pi(-4)+(1-c(4)-c(5))/4*pi(-5)+(1-c(4)-c(5))/4*pi(-6)+(1-c(4)-c(5))/4*pi(-7)+(1-c(4)-c(5))/4*pi(-8)+c(6)*100*(Y(-1)-SV3_1)+c(7)*pi(-1)+c(8)*im(-1)
@state sv2 = sv2(-1) +[var = exp(c(9))]
@state sv4 = sv4(-1) + [var = exp(c(10))]
@state sv1 = c(12)*sv4(-1) + sv2(-1)
@state sv1_1 = sv1(-1)
@state sv3_1 = sv3(-1)
@state sv3_2 = sv3_1(-1)
@state sv1_2 = sv1_1(-1)
@state sv3 = sv3(-1)+sv4(-1)+[var = exp(c(11))]
@mprior svec0
@vprior svar0
How to solve biased estimation of r*
Moderators: EViews Gareth, EViews Moderator
How to solve biased estimation of r*
- Attachments
-
- data0712.wf1
- (43.46 KiB) Downloaded 1099 times
Who is online
Users browsing this forum: Google [Bot] and 1 guest