BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Date iCal//NONSGML kigkonsult.se iCalcreator 2.20.2//
METHOD:PUBLISH
X-WR-CALNAME;VALUE=TEXT:TSE
BEGIN:VTIMEZONE
TZID:Europe/Paris
BEGIN:STANDARD
DTSTART:20251026T030000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
RDATE:20261025T030000
TZNAME:CET
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20260329T020000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:calendar.142380.field_date.0@www.tse-fr.eu
DTSTAMP:20260413T202603Z
CREATED:20260413T101001Z
DESCRIPTION:Karun Adusumilli (University of Pennsylvania\, USA)\, “You’ve G
 ot to Be Efficient :Ambiguity\, Misspecification and Variational Preferenc
 es”\, Econometrics and Empirical Economics Seminar\, TSE\, May 19\, 2026\,
  15:30–16:50\, room Auditorium 4.\n\nThis article introduces a framework f
 or evaluating statistical decisions\nunder both prior ambiguity and likeli
 hood misspecification. We begin\nwith an ambiguity set — a frequentist mod
 el that pairs a possibly misspecified\nlikelihood with every possible prio
 r — and uniformly expand it by a Kullback–\nLeibler radius to accommodate 
 likelihood misspecification. We show that optimal\ndecisions under this fr
 amework are equivalent to minimax decisions with an\nexponentially tilted 
 loss function. Misspecification manifests as an exponential\ntilting of th
 e loss\, while ambiguity corresponds to a search for the least favorable\n
 prior. This separation between ambiguity and misspecification enables loca
 l asymptotic\nanalysis under global misspecification\, achieved by localiz
 ing the priors\nalone. Remarkably\, for both estimation and treatment assi
 gnment\, we show that\noptimal decisions coincide with those under correct
  specification\, regardless of\nthe degree of misspecification. These resu
 lts extend to semi-parametric models.\nAs a practical consequence\, our fi
 ndings imply that practitioners should prefer\nmaximum likelihood over the
  simulated method of moments\, and efficient GMM\nestimators — such as two
 -step GMM — over diagonally weighted alternatives.
DTSTART;TZID=Europe/Paris:20260519T163000
DTEND;TZID=Europe/Paris:20260519T175000
LAST-MODIFIED:20260413T201001Z
LOCATION:TSE\, May 19\, 2026\, 15:30–16:50\, room Auditorium 4
SUMMARY:Econometrics and Empirical Economics Seminar
URL;TYPE=URI:https://www.tse-fr.eu/seminars/2026-youve-got-be-efficient-amb
 iguity-misspecification-and-variational-preferences
END:VEVENT
END:VCALENDAR