tag:blogger.com,1999:blog-2198942534740642384.post4455949870988589047..comments2023-10-24T03:16:41.009-07:00Comments on Econometrics Beat: Dave Giles' Blog: ARDL Models - Part IDave Gileshttp://www.blogger.com/profile/05389606956062019445noreply@blogger.comBlogger109125tag:blogger.com,1999:blog-2198942534740642384.post-54768821423553621322019-04-15T07:01:48.589-07:002019-04-15T07:01:48.589-07:00Yes, it's OKYes, it's OKDave Gileshttps://www.blogger.com/profile/05389606956062019445noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-64274259137744427142019-04-14T17:36:44.803-07:002019-04-14T17:36:44.803-07:00Dear Professor,
I have two series of daily futures...Dear Professor,<br />I have two series of daily futures' returns data for about 10 years making it about 2500 data points. But both the series are stationary in nature. Is it okay to run an ARDL model to measure the short term and long term relationship between the returns ? Ipsitahttps://www.blogger.com/profile/06343374486184937486noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-78576841005289738562019-04-14T07:19:23.973-07:002019-04-14T07:19:23.973-07:00On the face of it, you should be O.K.On the face of it, you should be O.K.Dave Gileshttps://www.blogger.com/profile/05389606956062019445noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-70391535361378454722019-04-13T11:13:03.923-07:002019-04-13T11:13:03.923-07:00Thank you very much Professor. And one more questi...Thank you very much Professor. And one more question, I have 60 observations (Quarterly) , and after running ARDL with 4 independent variables , using Schwarz criteria, ARDL(1,0,1,0,0,3) was chosen. Bound test shows co-integration and there is no serial correlation in residuals, so can I trust this results? , considering that some of my independetn variables are I(0). Thank you again.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-38236351541150144782019-04-13T06:20:50.807-07:002019-04-13T06:20:50.807-07:00ARDL models are estimated by OLS (& not just i...ARDL models are estimated by OLS (& not just in EViews). OLS will be biased (for small samples) in any model that has lagged values of the dependent variable as regressors, so that includes ARDL models. However, it is a consistent estimator (as long as the errors are independent), so the bias vanishes for large samples. You shouldn't use an ARDL model with a very small sample. And keep in mind that the long-run relationship is, again, just that - it needs plenty of observations in the sample to be meaningful. Finally, if there is a long-run cointegrating relationship, then OLS is a really good choice for estimating its parameters. The reason is this. Under cointegration, OLS is "super-consistent". The estimates converge to the true parameter values at the square of the usual rate as the sample size (n) grows. In standard models this rate of convergence is SQRT(n), but the rate is the same as "n" itself under cointegration.Dave Gileshttps://www.blogger.com/profile/05389606956062019445noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-41250925741040368962019-04-13T05:26:47.361-07:002019-04-13T05:26:47.361-07:00Hello Professor Giles, I want to know, If OLS esti...Hello Professor Giles, I want to know, If OLS estimation of ARDL model gives Biased results then how can we rely on the Long-run or short run coefficients? given by the ARDL model? . When runing ARDL model, does Eviews use OLS to estimate it? Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-37473971225596743762017-08-23T09:56:03.920-07:002017-08-23T09:56:03.920-07:00The same way that you would for any other regressi...The same way that you would for any other regression model - you need either more observations or fewer lags.Dave Gileshttps://www.blogger.com/profile/05389606956062019445noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-35217468166698152742017-08-23T09:42:46.668-07:002017-08-23T09:42:46.668-07:00Thank you for your post. Is it possible to advise ...Thank you for your post. Is it possible to advise how can I solve (Singular matrix) error using Eviews to apply ARDL.<br /><br />ThankyouAnonymoushttps://www.blogger.com/profile/03831606437334478470noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-25332992421404853742017-06-07T07:58:30.872-07:002017-06-07T07:58:30.872-07:00Dhanusha - I don't work for EViews - I suggest...Dhanusha - I don't work for EViews - I suggest you contact them through their User Forum at http://forums.eviews.com/Dave Gileshttps://www.blogger.com/profile/05389606956062019445noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-63487898114028056682017-06-07T07:17:37.558-07:002017-06-07T07:17:37.558-07:00Dear Professor Giles, the ARDL model I prepared on...Dear Professor Giles, the ARDL model I prepared on eview is not providing output when the forecast is run, and nor does the ols. I have changed the data range after I prepared both ardl and ols. but when the forecast is run the output is not provided. can you kindly help me to solve this issue please? thank you so much in advance.Dhanushanoreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-29497529649105547612017-05-11T14:51:09.991-07:002017-05-11T14:51:09.991-07:00So ARDL is basically a one-equation version of a V...So ARDL is basically a one-equation version of a VAR model?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-65146447274282832902016-11-18T05:52:16.939-08:002016-11-18T05:52:16.939-08:00If all of the variables are I(1) AND they are coin...If all of the variables are I(1) AND they are cointgrated then you could use this "old fashioned" ARDL model for forecasting legitimately. But not if the AREN'T cointegrated. That's true for ANY OLS regression. Also - see the links at the end of this post to "modern" ARDL modelling.Dave Gileshttps://www.blogger.com/profile/05389606956062019445noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-24631206398761585682016-11-18T05:38:03.260-08:002016-11-18T05:38:03.260-08:00My pricipal focus is forcasting. you have written ...My pricipal focus is forcasting. you have written above that, if there is a co-integrating relationship, you can estimate an ARDL. <br />as i understand , the ardl that we estimate is nothing but an OLS with lags of the variables involved. <br /><br />So, My question to you is can the models in levels with lagged variables be used for forecasting? <br /><br />this is especially important since the variables are non-stationary and hence, running ols wont be valid according to my knowledge.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-7344473968861534162016-11-12T05:36:37.036-08:002016-11-12T05:36:37.036-08:00The last equation won't be legitimate unless t...The last equation won't be legitimate unless the variables are cointegrated. You can estimate an ARDL model if you wish, but forecasting isn't its primary purpose. If the variables are NOT cointgrated you'll need to difference them before estimating the last model you mention. Otherwise use an ECM.Dave Gileshttps://www.blogger.com/profile/05389606956062019445noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-12955369274243305142016-11-12T02:58:12.690-08:002016-11-12T02:58:12.690-08:00Dear Professor Giles,
I would like to study the im...Dear Professor Giles,<br />I would like to study the impact of three variables “X1”, "X2" and “Z” on “Y”. The final aim is to forecast. These series are not stationary. They are all I(1). I would like to ask you if I can use the ARDL model in levels. Is it possible to estimate the following model in levels?<br />Y(t) = a + b1*Y(t-1) + b2*X1(t-1) + b3*X2(t-1) + b4*Z(t-2)<br />Best regards,Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-89639633248171722612016-10-21T12:52:50.885-07:002016-10-21T12:52:50.885-07:00The same as for any regression model - it's th...The same as for any regression model - it's the number of observations minus the number of regressors.Dave Gileshttps://www.blogger.com/profile/05389606956062019445noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-58890065526495028262016-10-21T02:57:57.969-07:002016-10-21T02:57:57.969-07:00Please proof , how to check the degree of freedom ...Please proof , how to check the degree of freedom in ARDL MODEL?amkhttps://www.blogger.com/profile/15867500960790926409noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-79755791032477288992016-10-17T06:06:23.616-07:002016-10-17T06:06:23.616-07:00Only if you have lots of observations. It;s just a...Only if you have lots of observations. It;s just a regression model you need positive degrees of freedom.Dave Gileshttps://www.blogger.com/profile/05389606956062019445noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-61574569729912391112016-10-17T01:22:18.880-07:002016-10-17T01:22:18.880-07:00Dave Giles, i have 9 independent variables and 1 d...Dave Giles, i have 9 independent variables and 1 depended variables, (total 10) i would it be possible to use ARDL Model?<br />amkhttps://www.blogger.com/profile/15867500960790926409noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-62992382305616265142016-09-27T02:38:59.576-07:002016-09-27T02:38:59.576-07:00sir, is normality test necessary for ARDL MODEL?
sir, is normality test necessary for ARDL MODEL?<br />Anonymoushttps://www.blogger.com/profile/03617835587097648812noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-4123947716460594222016-09-02T05:47:16.599-07:002016-09-02T05:47:16.599-07:00ARDL models were "revived" because it tu...ARDL models were "revived" because it turned out that they provide a very useful context for testing for long-run relationships when there is ambiguity about the stationarity/non-stationarity of the data.Dave Gileshttps://www.blogger.com/profile/05389606956062019445noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-6225969508912735582016-09-01T22:44:37.726-07:002016-09-01T22:44:37.726-07:00Anticipating the second a portion of this and pres...Anticipating the second a portion of this and present day employments of ARDL models. I was under the feeling that they were moderately old-school models that were put into the dustbin once ARIMA and ARIMAX models turned out to be anything but difficult to fit.you can also see this link <a href="http://www.anzael.com/vcdb-tools" rel="nofollow">VCDB TOOL</a> which you get cars validation, xml reports,validates ACES data,customize report and more...Anzael LLChttps://www.blogger.com/profile/13064109569322827764noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-83145947824169552562016-06-11T08:25:51.584-07:002016-06-11T08:25:51.584-07:00Just do what you would do with any other regressio...Just do what you would do with any other regression model.Dave Gileshttps://www.blogger.com/profile/05389606956062019445noreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-38334987330934009242016-06-11T08:21:33.749-07:002016-06-11T08:21:33.749-07:00After running the ARDL model, I found that it is h...After running the ARDL model, I found that it is heteroskedastic, what should be done now?Parulnoreply@blogger.comtag:blogger.com,1999:blog-2198942534740642384.post-72665676242091865822016-06-05T07:15:55.099-07:002016-06-05T07:15:55.099-07:00Yes, you can legitimately estimate an ARDL model i...Yes, you can legitimately estimate an ARDL model in this case, but it's not really the right basis for Granger causality testing.Dave Gileshttps://www.blogger.com/profile/05389606956062019445noreply@blogger.com