Friday, November 27, 2020

2020 EC2 Program Now Posted

Looking great!
31th (EC)^2 Conference: High Dimensional Modeling in Time Series
December 11-12, 2020
Paris, France (Alas, virtually...)
Program, registration, etc. at

Sunday, November 22, 2020

Classification Under Asymmetric Loss

I just read the stimulating new paper by Babii et al. on binary choice / classification w asymmetric loss,

It led me to recall some work of mine with Peter Christoffersen that may be related in interesting ways. The hyperlinked papers are below. We study optimal prediction under asy loss, focusing not only on how the amount of loss asymmetry drives the optimal bias (of course, as in Granger's seminal work), but also focusing on how heteroskedasticity​ (H), interacting with loss asymmetry, drives the optimal bias.  (The optimal bias increases as variance increases, and conversely.)  

We focus on time-series H, but of course cross section H is massively relevant as well, so I wonder how it would all work out in theory and practice in the Babii et al. cross-section classification environment.  Of course everyone talks about H destroying consistency in logit and related models, but that's deeper econometric consistency for marginal effects etc. I don't see why it would destroy consistency for the optimal prediction / classification, which is automatically induced by virtue of the estimation criterion as routinely exploited in the ML literature.

In any event the key recognition is that heteroskedasticity and asymmetric loss interact. Asymmetric loss of course influences the optimal prediction / classification, but it influences it more in regions (cross section) or periods (time series) where / when variance is high.

Christoffersen, P. and Diebold, F.X. (1997), "Optimal Prediction Under Asymmetric Loss," Econometric Theory, 13, 808-817.

, P.F. and Diebold, F.X. (1996)
, "Further Results on Forecasting and Model Selection Under Asymmetric Loss," Journal of Applied Econometrics, 11, 561-571.

(Somewhat) related earlier No Hesitations post:

Saturday, November 21, 2020

Essie Maasoumi Econometric Theory Interview

Check it out here. So fine and so appropriate for Essie.

More generally, seeing the latest reminds me of the invaluable ET Interviews series. Piece-by-piece, thanks to the initiative of Peter Phillips at ET, it is assembling a history of modern econometric thought. May its future be as vibrant as its past!

See here for some background circa 2015.

Monday, November 2, 2020

Russian Holidays Predict Troll Activity 2015-2017

A fascinating new abstract. Timely too. 

Russian Holidays Predict Troll Activity 2015-2017
Douglas Almond, Xinming Du, and Alana Vogel #28035


While international election interference is not new, Russia is credited with “industrializing” trolling on English-language social media platforms. In October 2018, Twitter retrospectively identified 2.9 million English-language tweets as covertly written by trolls from Russia's Internet Research Agency. Most active 2015-2017, these Russian trolls generally supported the Trump campaign (Senate Intelligence Committee, 2019) and researchers have traced how this content disseminated across Twitter. Here, we take a different tack and seek exogenous drivers of Russian troll activity. We find that trolling fell 35% on Russian holidays and to a lesser extent, when temperatures were cold in St. Petersburg. More recent trolls released by Twitter do not show any systematic relationship to holidays and temperature, although substantially fewer of these that have been made public to date. Our finding for the pre-2018 interference period may furnish a natural experiment for evaluati! ng the causal effect of Russian trolling on indirectly-affected outcomes and political behaviors — outcomes that are less traceable to troll content and potentially more important to policymakers than the direct dissemination activities previously studied. As a case in point, we describe suggestive evidence that Russian holidays impacted daily trading prices in 2016 election betting markets.