tag:blogger.com,1999:blog-6883247404678549489.post6190230199154664730..comments2021-01-15T11:44:57.505-08:00Comments on EViews: AutoRegressive Distributed Lag (ARDL) Estimation. Part 3 - PracticeIHSEViewshttp://www.blogger.com/profile/04703437003033046408noreply@blogger.comBlogger169125tag:blogger.com,1999:blog-6883247404678549489.post-46467907386398276142021-01-15T11:44:57.505-08:002021-01-15T11:44:57.505-08:00Hi EViews team. Thank you for the wonderful and co...Hi EViews team. Thank you for the wonderful and concise work in this three-part post on ARDL models. <br /><br />I've been wondering about this tricky issue for years now, maybe someone can help me out:<br /><br />As we know, the unrestricted CEC regression and the ECM regression are estimated via OLS. As so, I believe that one should be able to replicate the coefficients estimated within the EViews' ARDL window in the Least Square window. Note that for the ECM regression, one should run the long-run equation, save its residuals (Proc/Make Residual Series) and use its lagged form to run the ECM regression (i.e., it is a two-step estimation and I know researchers that did exactly that way before EViews 9 provided the ARDL tool). <br /><br />To make it clearer, for an ARDL (2,2) model, one should be able to replicate the coefficients running an OLS as: d(y) d(y(-1)) d(x(-1)) d(x(-2)) resid(-1), where 'resid' is the residual series obtained from the long-run equation. However, it turns out that all coefficients, standard errors, and most diagnostics are totally different when comparing the built-in ARDL method to the two-step OLS approach. I've tried this several times in the last years, with different data. I’ve tried all OLS, FMOLS, and DOLS and got anywhere near the actual coefficients displayed in the built-in method. For the sake of comparison, there are similar packages in other languages that precisely replicate the ARDL coefficients with the more basic LS regression methods. Any thoughts on that? <br /><br />If you are asking why would I want to run the CEC and ECM regressions via the OLS window, it is because we can’t compute Wald tests on the coefficients of the CEC regression. Note that in View\Representations, the user has access to coefficients from the basic estimation (the one that pops out as soon as you regress the model), which I believe represents the 'intertemporal dynamic estimation' as seen in part 1. In fact, everything one can do in terms of coefficient diagnostics (confidence intervals, variance inflation factor, etc) refers to these coefficients, not to those seen at either CEC or ECM windows. Lucas Lourencohttps://www.blogger.com/profile/08476018564448129846noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-50089310677033846392020-12-02T20:40:04.843-08:002020-12-02T20:40:04.843-08:00For example, how if my gdp didn't appear in EC...For example, how if my gdp didn't appear in ECM, but it does appear in long-run ?<br />Anonymoushttps://www.blogger.com/profile/07810933654594860257noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-5776165488558797032020-11-19T05:38:59.366-08:002020-11-19T05:38:59.366-08:00Did any of you get any solution for this problem y...Did any of you get any solution for this problem yet? if yes, can you let me know how to solve this issue?<br />Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-70737716358289947942020-09-28T17:22:17.698-07:002020-09-28T17:22:17.698-07:00By the way, when I run your example I get "Co...By the way, when I run your example I get "Cointegrating representation from ARDL equation ardlno<br />ardlno.makecoint" when I run mine I get "Cointegrating representation from ARDL equation <br />{%equation}.makecoint"<br /><br />Ray DVhttps://www.blogger.com/profile/12874509594945887668noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-91497504871816149372020-09-28T16:19:23.880-07:002020-09-28T16:19:23.880-07:00Dear Eviews team,
I've already tested de dat...Dear Eviews team, <br /><br />I've already tested de data and the residual for my regresion, it doesn't appear to have any problems whatsoever. However when I tried to visualize the fit of the long-run equation (by generating the "Cointegrating Relationship") I get big numbers (around 7 and all positive terms). Do you have any idea what could be causing this? Thanks in advance! <br />Ray DVhttps://www.blogger.com/profile/12874509594945887668noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-19203043940066588822020-08-12T05:49:38.534-07:002020-08-12T05:49:38.534-07:00Hullo, my ARDL F-statistic = 328.914.
Is this acce...Hullo, my ARDL F-statistic = 328.914.<br />Is this acceptable for concluding cointegration?Ronald Ochenhttps://www.blogger.com/profile/12193329186454674420noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-28388906516885209612020-07-21T23:32:10.234-07:002020-07-21T23:32:10.234-07:00how about the short-run coefficient in eviews 11?how about the short-run coefficient in eviews 11?Anonymoushttps://www.blogger.com/profile/07962801405581527050noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-78446649534206672762020-06-17T23:30:41.068-07:002020-06-17T23:30:41.068-07:00This comment has been removed by a blog administrator.saivenkathttps://www.blogger.com/profile/07265662654934474648noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-71339983402845815082020-05-25T08:02:43.772-07:002020-05-25T08:02:43.772-07:00Hello,
Thank you for the great efforts you put enh...Hello,<br />Thank you for the great efforts you put enhancing Eviews.<br />but I still have a few doubts about ARDL estimation, although i have read all of posts n comments in each post. <br /><br />you have said that "Everything else outside the CointEq is considered to be a short-run variable." but there exist three alternative representations so there will be three different results. <br /><br />my question are:<br />1. what's the difference between parameter estimation in intertemporal dynamic estimation and in conditional error correction (CEC) representation, i know that CEC is the reduce form, i mean the difference in how to interpret the parameter<br />2. what is the correct parameter to interpret as a short run parameter, using parameter in intertemporal dynamic estimation or in conditional error correction (CEC) representation? because i have read many journal and got the different result about using short run parameter. and as we know that the result is also different. <br /><br />I would be appreciate it if Eviews team can guide me here!Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-81328544922254015872020-05-21T15:45:03.085-07:002020-05-21T15:45:03.085-07:00Hi , your post is really special I am using annua...Hi , your post is really special I am using annual time series data of 27 years of 5 variables(using eviews 9). The f stat shows there is cointegrating relationship in the variables. ECT is also negative and significant. My only question us that are the values of f stat reliable for a small sample like mine. To date I haven't read eviews or Ardl methodology mentions that sample should be greater greater or equal to 30.<br />Cheers!Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-12227793364859022020-03-31T11:55:23.796-07:002020-03-31T11:55:23.796-07:00HI Eviews team,
thank you for your great efforts....HI Eviews team, <br />thank you for your great efforts.<br />i have found evidence of co-integration since my F-statistic is beyond the upper bound. however when i generate long run coefficients through ( long run form and bound test) i find large coefficients:<br /> Variable 1(-45.26939<br />Variable 2 (46.50257)<br />Variable 3(-22.63170)<br />Variables 4 (57.91694)<br />Variable 5 (-82.49271)<br />Variable 6 (15.40217)<br />Variable 7 (-73.66744)<br />Constant 459.2406<br />is this normal, provided that the regression passed all tests<br />thank you<br /> hamza belfhttps://www.blogger.com/profile/07141282914341444548noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-45821653858805438402020-03-12T10:21:31.890-07:002020-03-12T10:21:31.890-07:00Hello EViews team.
Please, this is just for co...Hello EViews team.<br /> Please, this is just for confirmation, Am I right or am I missing something? "if we have case 4 or case 2 in ARDL modeling, we can have a degenerate cointegration but can not have a nonsensical cointegration".<br />CordiallyEcointelligencyhttps://www.blogger.com/profile/09764228437124629520noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-4433595641054648562020-03-12T03:13:19.954-07:002020-03-12T03:13:19.954-07:00IHSEviews, thank you for your reply!
If I may ask ...IHSEviews, thank you for your reply!<br />If I may ask one final question. Can I test the stability or the stationarity of the EC model by testing the inverse roots (AR Structure). Kindly can you advice me how to?<br />Thank you again!<br />AmirAmir https://www.blogger.com/profile/08355544087004838174noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-16719859837898962292020-03-11T06:03:20.697-07:002020-03-11T06:03:20.697-07:00This comment has been removed by the author.Ecointelligencyhttps://www.blogger.com/profile/09764228437124629520noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-26022551392093063432020-03-08T23:05:14.192-07:002020-03-08T23:05:14.192-07:00Hello Eviews Team, I am currently facing the same ...Hello Eviews Team, I am currently facing the same problem as indicated by Lucas. Could you please provide a layman's interpretation of the shortrun coefficients of the variables indicated as thus: ** Variable interpreted as Z = Z(-1) + D(Z). Thank youAlhassanunoreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-8436893645200322042020-03-07T08:59:05.434-08:002020-03-07T08:59:05.434-08:00Hello,
Sorry, I worked with the Eviews 9.5 and no...Hello, <br />Sorry, I worked with the Eviews 9.5 and now I just want to confirm these something:<br /><br />1)- How many cases-ARDL in EViews 9.5?<br />2)- Are the critical values of Bounds test-ARDL in case 4 (with TREND) on EViews 9.5 the same with case 5 on EViews 10?<br /><br />B.WEcointelligencyhttps://www.blogger.com/profile/09764228437124629520noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-34705986549036304652020-03-07T08:39:21.659-08:002020-03-07T08:39:21.659-08:00This comment has been removed by the author.Ecointelligencyhttps://www.blogger.com/profile/09764228437124629520noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-29472855141402817152020-03-02T11:25:35.991-08:002020-03-02T11:25:35.991-08:001) At the moment, we don't have the CUSUM stat...1) At the moment, we don't have the CUSUM statistic implemented.<br /><br />2) The most we can say in the long run (if there is evidence of cointegration) is that both the independent and dependent variables are governed by common forces. This implies that if you do run a regression in the long run, that regression will be valid and whatever relationship exists at that point can be interpreted in the usual way.<br /><br />3) If if F-statistic is greater than the I(1) bound then you reject H0: no cointegration. Thus, fail to reject that the series are cointegrated. In the next step, if you run the test on the long-run coefficient and you reject significance, it means that the relationship between x and y is non-sensical. Please read Part 2 of the ARDL blog series for further details.IHSEViewshttps://www.blogger.com/profile/04703437003033046408noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-4913779777282866482020-03-02T08:42:55.210-08:002020-03-02T08:42:55.210-08:00Hello,
Thank you for the great efforts you put enh...Hello,<br />Thank you for the great efforts you put enhancing Eviews. <br />I have a few doubts about ARDL estimation.<br />I would be appreciate it if Eviews team can guide me here!<br />1- Is CUSUM test performed on the ECM of the long run equation?<br />2- Can the long run coefficients be interpreted as positively/ negatively impact the independent variable in the long run?<br />3-In a two-variable (y and x) ARDL model, if the F-Statistics in the Bounds test is greater than I(1) bound, but the long run coefficient is not significant, can one say there is a long run relationship between x and y?<br />Thank you!<br />A<br />Amir https://www.blogger.com/profile/08355544087004838174noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-6663099178641220382020-02-18T01:21:18.195-08:002020-02-18T01:21:18.195-08:00Hello,
Ok, what is the result of "t-Boun...Hello, <br /> Ok, what is the result of "t-Bounds test" for the coefficient of Yt-1?Ecointelligencyhttps://www.blogger.com/profile/09764228437124629520noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-73411518799259347412020-02-18T01:20:40.008-08:002020-02-18T01:20:40.008-08:00And how if my stability test show that the model i...And how if my stability test show that the model is stable, but the ect <-2?Wonnhttps://www.blogger.com/profile/07622921979231937008noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-1470581922273948492020-02-18T01:03:15.225-08:002020-02-18T01:03:15.225-08:0033Wonnhttps://www.blogger.com/profile/07622921979231937008noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-15173891587017638232020-02-11T02:55:16.810-08:002020-02-11T02:55:16.810-08:00Hello,
Please, are you in case 1, 2, 3, 4 or 5?Hello,<br /> Please, are you in case 1, 2, 3, 4 or 5?Ecointelligencyhttps://www.blogger.com/profile/09764228437124629520noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-46353406921501442952020-02-11T02:55:03.221-08:002020-02-11T02:55:03.221-08:00This comment has been removed by the author.Ecointelligencyhttps://www.blogger.com/profile/09764228437124629520noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-75811854977625289242020-02-11T00:32:07.630-08:002020-02-11T00:32:07.630-08:00My F bound test shows that the cointegration is ex...My F bound test shows that the cointegration is exist, but the ect in ecm is not significant. How should i explain this?Wonnhttps://www.blogger.com/profile/07622921979231937008noreply@blogger.com