I am hiring a 2-year post doc @EconUCL under my @ERC_Research STG "CreMac", job posting here: Suitable for students with a strong econometrics background, macro interest a plus. Please consider applying or sharing with your network!.

Updated Matlab code with all of these new features can be found here: A fast and efficient Stata package will follow soon. Get in touch if you have any questions about how to use the tests! (6/6).

3) Perhaps most importantly, we provide tests for the bias in a *single coefficient* of interest. Often researchers don't care about all coefficients on endogenous regressors. Now, you can test based on just what you care about! (5/6).

2) Offer a second "absolute bias" criterion, identical to Stock & Yogo's. This can also be interpreted as bias relative to the max OLS bias. This provides a novel, intuitive option for researchers even when N=1, as it is new relative to Montiel Olea & Pflueger (4/6).

New in this version, we.1) Show how our test also generalises Sanderson & Windmeijer's (2016) test for weak instruments due to a rank deficient first-stage to be robust to heteroskedasticity and autocorrelation (3/6).

The paper generalises the popular Stock & Yogo (2005) first-stage test for bias arising from weak instruments to be robust to heteroskedasticity and autocorrelation, in the process extending Montiel Olea & Pflueger's (2013) robust test for one endogenous regressor to N>1 (2/6).

The latest version of my paper (with Karel Mertens), "A Robust Test for Weak Instruments for 2SLS with.Multiple Endogenous Regressors", is finally available! We're very excited about several new features ⬇️ (1/6)

I'm looking forward to using this great opportunity to support some (hopefully) exciting papers, the development of young researchers as post-docs and RAs, and fun conferences. Look out for conference announcements and job postings in the years to come! Thank you @ERC_Research!.

The second stream will provide robust inference tools when identification remains weak, for a wide range of identification strategies. In particular, I'll be working on methods tailored to subset inference with possibly many nuisance parameters and/or heteroskedasticity.

CreMac focuses on more precise (and credible) inference methods tailored to macroeconomic policy questions. The first stream will focus on how to extract stronger signals from popular identification strategies.

Thrilled and honoured to announce that my proposal, Credible Inference for Empirical Macroeconomics, has been awarded an @ERC_Research starting grant! #ERCStG @UCLEconomics

A simple idea, but one that we hope was useful as economists and policymakers responded to unprecedented data-related challenges. And another chance to thank the RAs (including Mihir) who made the WEI possible .

Honoured to hear my paper "Measuring real activity using a weekly economic index" with @jimstockmetrics, Karel Mertens, and Mihir Trivedi has been awarded the Journal of Applied Econometrics' Stone Prize

What a sublime conference! Many thanks to @francescofurla8 and @FraRavazzolo for organising.

RT @nberpubs: Estimates of unconditional distribution of marginal propensity to consume (MPC) show a lot of heterogeneity (4–133 percent).….

We are hiring! Come join a great group of colleagues in London!.

RT @CavaliereGiu: Hi #EconTwitter!. The Autums 2023 schedule for the Virtual 𝐓𝐢𝐦𝐞 𝐒𝐞𝐫𝐢𝐞𝐬 𝐒𝐞𝐦𝐢𝐧𝐚𝐫𝐬 (@ovtss) is out!📈. It includes @OtiliaBol….

RT @CavaliereGiu: Hi #EconTwitter!. You use @jimstockmetrics-Yogo F pre-test for weak instruments, and. (i) you worry about heteroskedastic….

Matlab code is available:.Comments, feedback, and coding bugs welcome! (4/4).

We nest both as special cases (up to approx. error). Multiple endogenous regressors is a common challenge when interaction terms are included, for instance. Previously, users wanting a bias-based test (à la Stock-Yogo) for n>1 were forced to assume heteroskedasticity etc. (3/4).