tag:blogger.com,1999:blog-6883247404678549489.post8606547027851810399..comments2021-12-06T05:31:59.957-08:00Comments on EViews: Beveridge-Nelson FilterIHSEViewshttp://www.blogger.com/profile/04703437003033046408noreply@blogger.comBlogger22125tag:blogger.com,1999:blog-6883247404678549489.post-59465818226092552402021-07-10T20:31:29.186-07:002021-07-10T20:31:29.186-07:00Dear Prof James Morley,
Hopes you are well. I hav...Dear Prof James Morley,<br /><br />Hopes you are well. I have some problem with the BN Filter add in using Eviews9. After i use this BN filter dialog, then i hit OK. There is error message "@CVAR is an illegal or reserved name".Btw, after i hit OK for the error message, my bn_cycle only show one row1,column1 with value 0. There are also dy_m,dy_s,delta,lag,s1,s2,window,structural break. My question, where is the cycle? What is mean by dy_m?Kesavanhttps://www.blogger.com/profile/05671671700045773881noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-59686570679015863532021-07-10T19:24:24.663-07:002021-07-10T19:24:24.663-07:00Dear Sir,
I am also encountering the same issue n...Dear Sir, <br />I am also encountering the same issue now. Could you pls let me know how did you solve this issue. I could not find the answer in this bloh.<br />Thank you<br />KesavarajahKesavanhttps://www.blogger.com/profile/05671671700045773881noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-18613961534053153282020-09-14T18:57:57.742-07:002020-09-14T18:57:57.742-07:00It has now been fixed. Thanks, again, Davaa.It has now been fixed. Thanks, again, Davaa.James Morleyhttps://www.blogger.com/profile/18112171881288613309noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-46716692608991676742020-09-14T18:57:22.453-07:002020-09-14T18:57:22.453-07:00This has now been fixed, so the timing is correct ...This has now been fixed, so the timing is correct for version 2.2 of the add-in. When the BN filter is applied to a series like log real GDP from, say, 1947Q1 to 2020Q2, it will produce an output gap for 1947Q2 to 2020Q2. Also, the default option in version 2.2 is now to apply the dynamic mean adjustment, which is useful for many time series that have changes in long-run drift.James Morleyhttps://www.blogger.com/profile/18112171881288613309noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-16263514379320793792020-06-24T19:17:52.047-07:002020-06-24T19:17:52.047-07:00Dear Benjamin and Davaajargal great post, thank yo...Dear Benjamin and Davaajargal great post, thank you. How easy is working these techniques in this software. BestJMhttps://www.blogger.com/profile/03600767482433769445noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-74661340767567026552020-05-06T01:29:19.260-07:002020-05-06T01:29:19.260-07:00Hello! I am the new reader of your blog. I really ...Hello! I am the new reader of your blog. I really appreciate the work you do. Your analysis is very professional and interesting. Thanks!Daniel Johnsonhttp://www.confiduss.com/en/services/corporate/accounting/noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-15787782393842731992020-04-20T10:20:42.627-07:002020-04-20T10:20:42.627-07:00Dear Prof James Morley,
Hopes you are well. I hav...Dear Prof James Morley,<br /><br />Hopes you are well. I have some problem with the BN Filter add in using Eviews9. After i use this BN filter dialog, then i hit OK. There is error message "@CVAR is an illegal or reserved name".Btw, after i hit OK for the error message, my bn_cycle only show one row1,column1 with value 0. There are also dy_m,dy_s,delta,lag,s1,s2,window,structural break. My question, where is the cycle? What is mean by dy_m?<br /><br />Hope you can help meAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-72589415212046080882020-03-04T19:05:12.987-08:002020-03-04T19:05:12.987-08:00Thanks, Davaa. And thanks for all the work on the ...Thanks, Davaa. And thanks for all the work on the add-in.James Morleyhttps://www.blogger.com/profile/18112171881288613309noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-67480338318856246542020-03-03T18:48:56.251-08:002020-03-03T18:48:56.251-08:00Thank you Prof. James for pointing the shift. It w...Thank you Prof. James for pointing the shift. It will be fixed soon. Davaahttps://www.blogger.com/profile/04216241732619542147noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-3445273118175548532020-03-01T22:06:14.994-08:002020-03-01T22:06:14.994-08:00Just a note that version 2.1 of the add-in, which ...Just a note that version 2.1 of the add-in, which applies the BN filter to the levels of the series to be detrended, shifts the timing of the cycle one period back. That is, it suggests there is a cycle in the first period that the levels series is available and does not report a cycle in the last period. However, the cycle should be shifted forward one period (i.e., no cycle in the first period that the series is available because the series needs to be differenced when applying the filter and a cycle in the last period). Hopefully this can be fixed in the next version of the add-in. But, in the meantime, it is important for users to shift the cycle forward (e.g., "genr bn_cycle_s_true=bn_cycle_s(-1)") to be correctly timed.James Morleyhttps://www.blogger.com/profile/18112171881288613309noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-90351536162819102862020-02-27T21:15:32.809-08:002020-02-27T21:15:32.809-08:00Ah I needed to add "impose=1" - thanks f...Ah I needed to add "impose=1" - thanks for your help!reyalshttps://www.blogger.com/profile/11609012679704418774noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-47666950243650038592020-02-27T01:25:22.205-08:002020-02-27T01:25:22.205-08:00Thanks. My post below is not relevant after this e...Thanks. My post below is not relevant after this explanation.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-68751075129140920342020-02-27T01:13:52.440-08:002020-02-27T01:13:52.440-08:00y.bnfilter(lag=12, impose=1, delta=0.1)y.bnfilter(lag=12, impose=1, delta=0.1)Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-91737089670080053902020-02-27T01:06:45.432-08:002020-02-27T01:06:45.432-08:00y.bnfilter(lag=12, impose=1, delta=0.1)y.bnfilter(lag=12, impose=1, delta=0.1)Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-69672983842138918252020-02-26T23:00:40.938-08:002020-02-26T23:00:40.938-08:00It seems that delta is setting only in add-in now....It seems that delta is setting only in add-in now. In command line this is not working at this point.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-55691459062914474592020-02-26T17:31:52.929-08:002020-02-26T17:31:52.929-08:00How do you set the smoothing delta in command? I t...How do you set the smoothing delta in command? I tried the code below but no luck:<br /><br />y.bnfilter(lag=12, demean=1, window=40, delta=0.1)<br />reyalshttps://www.blogger.com/profile/11609012679704418774noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-13075920929197835262020-02-26T12:43:35.740-08:002020-02-26T12:43:35.740-08:00Just tried out the updated version of the add-in (...Just tried out the updated version of the add-in (version 2.1) and it works perfectly. Once installed, it can be applied easily by opening a series to be detrended (e.g., y, where "genr y=100*log(realGDP)"), clicking the Proc button, and selecting BN Filter under Add-ins at the bottom of the menu. <br /><br />There is also now an option to impose a smoothing parameter, if desired.<br /><br />Thanks, Davaa for creating the add-in and updating it to work like the HP filter procedure in EViews, while, of course, producing a more reliable estimate of output gap!James Morleyhttps://www.blogger.com/profile/18112171881288613309noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-60059243604148980712020-02-26T07:52:52.569-08:002020-02-26T07:52:52.569-08:00Thanks for bringing it to our attention. We've...Thanks for bringing it to our attention. We've updated the system with the correct version. Things should work as expected now.IHSEViewshttps://www.blogger.com/profile/04703437003033046408noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-47573621623982891702020-02-25T23:09:41.467-08:002020-02-25T23:09:41.467-08:00Probably, something is wrong with this add-in. Ins...Probably, something is wrong with this add-in. Instead of cycle, it gives trending line. So I cannot replicate this stuff in Eviews 11 and 10. Additionally, Eviews freezes while executing bnfilter_blog.prg. Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-88424539669900581472020-02-25T20:58:34.114-08:002020-02-25T20:58:34.114-08:00Thanks, Yamin. It seems that Ben has reorganized h...Thanks, Yamin. It seems that Ben has reorganized his site. The code is available on his <a href="https://sites.google.com/site/benjaminwongshijie/research" rel="nofollow">research page</a>.James Morleyhttps://www.blogger.com/profile/18112171881288613309noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-46146911999993644672020-02-25T18:37:21.146-08:002020-02-25T18:37:21.146-08:00Hi James. I just wanted to give you a heads up tha...Hi James. I just wanted to give you a heads up that both links for the Matlab and R code for the BN filter on Global Perspective's Model Hub currently do not appear to be working. It redirects to Ben's page but gives a 404 error. I just wanted to let you know.Yaminhttps://www.blogger.com/profile/05634086630077783591noreply@blogger.comtag:blogger.com,1999:blog-6883247404678549489.post-56289466683882546492020-02-25T16:34:52.648-08:002020-02-25T16:34:52.648-08:00Thanks, Ben and Davaa. This is very helpful in sho...Thanks, Ben and Davaa. This is very helpful in showing how to use the BN filter in EViews. <br /><br />The comparison with the CBO gap is interesting. But I decided to make a direct comparison between the BN filter output gap (with dynamic demeaning) and the HP filter output gap using the EViews procedures, including the add-in. <br /><br />The result is <a href="https://drive.google.com/open?id=1ClNs7SlHBTgjUjrAQNUXXazu6wTMEpKJ" rel="nofollow">here</a>.<br /><br />As can be seen in the linked figure, the BN and HP gaps have more similar amplitude than with the CBO gap. The HP gap is deeper than the BN gap in the 81-82 recession, while the BN gap is deeper than the HP gap in the Great Recession.<br /><br />Note that I have used an old version of the EViews add-in for which the BN filter needs to be applied to the *first difference* of the series to be detrended, while the HP filter procedure applies directly to the series to be detrended. The blog post seems to suggest that the BN filter can be applied to directly to the series to be detrended, which makes sense since the BN decomposition is a trend/cycle decomposition of y, not dy. So hopefully an updated version of the add-in will allow this. Also, it would be great if users can impose a smoothing parameter delta, much like imposing a lambda for the HP filter. This is possible in the Matlab and R code for the BN filter available on the <a href="https://sydney.edu.au/arts/our-research/futurefix/global-perspectives-on-economic-policy/model-hub.html" rel="nofollow">Global Perspective's Model Hub</a>.<br />James Morleyhttps://www.blogger.com/profile/18112171881288613309noreply@blogger.com