Nowcasting inflation, Knotek and Zaman (2014,2017,2024)

If you are seeking RATS code for implementing a particular technique or replicating results from a paper, post your request here. Be sure to include complete citations for any papers or books.
hardmann
Posts: 218
Joined: Sat Feb 26, 2011 9:49 pm

Nowcasting inflation, Knotek and Zaman (2014,2017,2024)

Unread post by hardmann »

Dear Tom:

Federal Reserve Bank of Cleveland has long been nowcasting inflation. The customized MF model developed by Knotek and Zaman (2017,2024), which relies on deterministic model switching (DMS). The DMS has been shown to be highly effective in producing timely and accurate nowcasts of inflation. I don't know what kind of econometric model DMS belongs to? How to code to replicate it?


Knotek, Edward S. II, and Saeed Zaman, 2014. “Nowcasting
U.S. Headline and Core Infl ation,” Federal Reserve Bank of Cleveland,working paper no. 14-03.https://DOI 10.26509/frbc-wp-201403r
Knotek II, Edward S. and Saeed Zaman. 2024. "Nowcasting Inflation." Working Paper No. 24-06. Federal Reserve Bank of Cleveland. https://doi.org/10.26509/frbc-wp-202406.

Best Regard
Hardmann
Last edited by hardmann on Mon Jan 13, 2025 10:06 pm, edited 2 times in total.
TomDoan
Posts: 7686
Joined: Wed Nov 01, 2006 4:36 pm

Re: nowcasting inflation, Knotek and Zaman (2017,2024)

Unread post by TomDoan »

There's nothing really to that. It's just a model which switches based upon observable criteria, in this case, which data are available. (In particular, they use a different model after CPI has been released but not yet CPE, and another when both have been observed).

The complication is collecting the right kind of data to estimate those types of models. The current historical data on those (and other variables involved) have been revised and re-adjusted sometimes multiple times, which will likely give different statistical behavior to the data that were available at the time. Generating a whole string of separate data sets with the data that were available at the time would probably offer a more realistic model, but that can be a lot of work, which is why it typically is done by central banks.
hardmann
Posts: 218
Joined: Sat Feb 26, 2011 9:49 pm

Re: nowcasting inflation, Knotek and Zaman (2014,2017,2024)

Unread post by hardmann »

Dear Tom:

I had found thier old working paper "Nowcasting U.S. Headline and Core Inflation" in 2014 and revised version in 2015.
https://www.clevelandfed.org/center-for ... -inflation
Their data come from the St. Louis Fed’s Archival Federal Reserve Economic Data (ALFRED).

I dont know the DMS how to switch underlying the tow model. I implore you to write the basic model or demonstrative model without data or simulated data. Or you can provide me with detailed information about the indicator data, and I will collect the relevant data.



Best Regard
Hardmann
TomDoan
Posts: 7686
Joined: Wed Nov 01, 2006 4:36 pm

Re: Nowcasting inflation, Knotek and Zaman (2014,2017,2024)

Unread post by TomDoan »

Instead of finding old working papers, have you tried contacting the authors to see if they have more technical details about how they collect and use the data? They might even be able to send you a copy of their data set.
hardmann
Posts: 218
Joined: Sat Feb 26, 2011 9:49 pm

Re: Nowcasting inflation, Knotek and Zaman (2014,2017,2024)

Unread post by hardmann »

Dear Tom:
I have contacted the authors, they did not respones. we try directly fetch data form online FRED rather than real-time dataset prepared in advance. We focous on the DMS itself,leting aside the high freqent time series, eg, oil and food. I'm not sure how to switch between two basic models, do I rely on code to implement it, or do I rely on internal switching mechanisms like the regime switching model.

Please help me review my code. Also, there seems to be a problem with my if statement.

Best Regard
Hardmann
Attachments
US DMS 0.06.RPF
(1.76 KiB) Downloaded 10 times
tclark
Posts: 94
Joined: Wed Nov 08, 2006 3:20 pm

Re: Nowcasting inflation, Knotek and Zaman (2014,2017,2024)

Unread post by tclark »

Sometimes the authors in question are busy and need time to reply. You might also try checking journal websites, especially with journals that encourage or require code and data. In the case of the Knotek and Zaman work, the JMCB data archive has a replication package with their code (in Matlab) and data associated with their 2017 paper in that journal. See https://economics.osu.edu/sites/default ... cation.zip
hardmann
Posts: 218
Joined: Sat Feb 26, 2011 9:49 pm

Re: Nowcasting inflation, Knotek and Zaman (2014,2017,2024)

Unread post by hardmann »

Dear Clark:
Thanks a million, Dr. Clark.

Best Regard
Hardmann
Post Reply