SERVING THE QUANTITATIVE FINANCE COMMUNITY

 
User avatar
buckeye
Topic Author
Posts: 5
Joined: August 9th, 2005, 4:21 am

ACD modeling (GARCH with additonal indepedent var)

August 9th, 2005, 1:37 pm

I am trying to implement an Autoregressive Conditional Duration model. According to a paper by Noble Laureate Robert Engle and Joe Lange, the ACD model is given as:EPTIME (t) = omega + alpha1 * PTIME(t-1) + beta1 * EPTIME(t-1) - (A), andEPTIME (t) = omega + alpha1 * PTIME(t-1) + beta1 * EPTIME(t-1) + gamma * spread(t-1) - (B)To estimate parameters omega, alpha, and beta in (A), I use a simple GARCH(1,1) model. How do I incorporate the additonal exogenous lagged variable Spread in (B). I am a newbie and don't know how to do this regression (I am using statistical package Eviews and R, and can get access to Matlab if it is more convenient in Matlab).Any help will be highly appreciated.Thanks.PS: The paper is "Measuring, Forecasting and Explaining Time Varying Liquidity in the Stock Market" by Robert Engle and Joe Lange
 
User avatar
buckeye
Topic Author
Posts: 5
Joined: August 9th, 2005, 4:21 am

ACD modeling (GARCH with additonal indepedent var)

August 9th, 2005, 6:33 pm

How do I delete my previous message? I don't see any "delete message" option.Thanks.
 
User avatar
proximo20
Posts: 52
Joined: January 15th, 2010, 9:54 am

ACD modeling (GARCH with additonal indepedent var)

June 12th, 2010, 9:02 am

ok I am working on the same paper using SAS. I want to use proc lifereg command but I don't understand how you get EPTIME variable. I would really appreciate it if anyone can explain how you estimate EPTIME values.
 
User avatar
msperlin
Posts: 608
Joined: July 10th, 2006, 6:21 pm

ACD modeling (GARCH with additonal indepedent var)

June 16th, 2010, 7:43 am

QuoteOriginally posted by: buckeyeI am trying to implement an Autoregressive Conditional Duration model. According to a paper by Noble Laureate Robert Engle and Joe Lange, the ACD model is given as:EPTIME (t) = omega + alpha1 * PTIME(t-1) + beta1 * EPTIME(t-1) - (A), andEPTIME (t) = omega + alpha1 * PTIME(t-1) + beta1 * EPTIME(t-1) + gamma * spread(t-1) - (B)To estimate parameters omega, alpha, and beta in (A), I use a simple GARCH(1,1) model. How do I incorporate the additonal exogenous lagged variable Spread in (B). I am a newbie and don't know how to do this regression (I am using statistical package Eviews and R, and can get access to Matlab if it is more convenient in Matlab).Any help will be highly appreciated.Thanks.PS: The paper is "Measuring, Forecasting and Explaining Time Varying Liquidity in the Stock Market" by Robert Engle and Joe LangePlease be aware that Garch estimation function for ACD models only works when sample size goes to infinity and when the distribution assumption for the model is the exponential (simplest case).I ahve worked with ACD models in the past and I published some codes for estimations:Rmetrics - fACD packageMatlabIn this packages you can extend the duration process with independent variables so it should work for you.
Last edited by msperlin on June 15th, 2010, 10:00 pm, edited 1 time in total.
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


Twitter LinkedIn Instagram

JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...


GZIP: On