Kenneth Rich

I am very excited and proud to announce that I will be transitioning from a visiting to a tenure-track Assistant Professor at the University of Mississippi this upcoming Fall!

@RebelEconProf @RPShah_econ Is this an appropriate time to say “Come to the Sip?”

@RPShah_econ They give away that they know the answer because they made K so narrow.

Obviously K.

@RPShah_econ H 💀

@nd_econ Congrats! Also very interesting JMP!

Congratulations to Alex Houtz, who just successfully defended his dissertation. Alex will be joining Hope College as an assistant professor this summer.

Christopher Sims, RIP. A giant of macroeconomics has left us. Here is his homepage: princeton.edu/~sims/

New draft out! Must be spring break — another paper update 😅 Conventional VARs can't recover fiscal shocks when agents anticipate policy changes. I demonstrate that an MS-BVAR solves this by showing that regime switching is both the source of foresight and the key to identification. Key Results: 1) Spending multipliers are large and state-dependent (1.4–1.8 in recessions, ~1.0 in expansions). 2) Tax multipliers: consistently small across all regimes — much smaller than conventional estimates suggest. papers.ssrn.com/sol3/papers.cf…

@RosanaFerrero Absolutely agree. This list is a good tasklist for any carefully executed Bayesian exercise.

@KennethMRich Misinterpretations of BF arise from poorly justified priors, insufficient design, unregulated stopping rules, and poorly specified models. The solution is to integrate BF with pre-registration, simulation/calibration, sensitivity analysis, and explicit evaluation of test severity

The Bayesian approach is the clearer way.

@RosanaFerrero But if the goal is measuring evidence, the Bayes Factor does that directly. Conflating the two is what creates the apparent gap.

@RosanaFerrero Both points are true but not unique to Bayes Factors. A p-value from a poorly designed experiment is equally uninformative. Experimental design is prior to the choice of inferential tool. Error rate control matters for decision procedures.

@RosanaFerrero I agree that the replication crisis is about discipline rather than tool selection. But that actually reinforces the Bayesian case. Forcing researchers to be explicit upfront about the hypotheses they wish to test does exactly that.

@KennethMRich -The Bayes factor tells us how much relative evidence there is between two hypotheses. -The frequentist approach (error control and severity) tells us how demanding the test was in detecting flaws, if any existed. Mayo 2025. linkedin.com/posts/rosanafe… linkedin.com/posts/rosanafe…

@Damiano_Difra Great suggestion! A quick read actually answers one of the questions a previous referee brought up to us. I’ll give it a more careful reading today. Thanks for sharing the paper!

@KennethMRich Nice work! One small thing: it may be worth looking at this paper, which shows how to separate size and sign nonlinearity in LPs without confounding the two: dropbox.com/scl/fi/c7op8jj…

New draft out! Govt spending multipliers depend on the SIZE of the shock. Much attention has been paid to whether multipliers vary with the business cycle or the sign of the shock--- we show that shock size is an overlooked but crucial dimension. papers.ssrn.com/sol3/papers.cf…

@lucasian76 Also, remember your job as the presenter is to present your narrative to the audience. Showing more information than necessary risks you losing control of the narrative and letting others form their own.

@lucasian76 Only include the results that you want to focus on. The audience will thank you for not overwhelming them. The interested reader will see your other results in the paper.

Is acceptable to have a 4x3 in impulse-responses in a Slide? Many things that need to be shown, but I am aware that people want 3x3

Marco Duarte and I will be organizing Industrial Organization Society sessions at the @SouthernEconJ Conference in Houston during Nov 21-23, 2026. If you’d like to join us, please sign up using the form in the replies. Marco and I will do the rest. Feel free to share!

@BanilowMary I plan on uploading my lecture notes to my webpage after the semester! I want to work out the kinks with the course first before uploading 🙂

Excited to start teaching a new course on Bayesian Estimation of DSGE Models!

@cmatthes_econ @40yoap That’s a lot better way to say that 😅

@40yoap @KennethMRich “Economically sensible calibration”

@40yoap 🤣

@KennethMRich "Select economically sensible priors" 😏

@aeimit These notes look great! I will definitely draw inspiration from them for this course as well as for my Applied Time Series course. Thank you for sharing :)

@KennethMRich I used to teach a related course at MSU. ec831.weebly.com the notes are all on there but obviously some of them will be outdated.