BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//AMSE//Event Calendar//FR
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
UID:event-10353@www.amse-aixmarseille.fr
DTSTAMP:20260422T113202Z
CREATED:20260422T113202Z
LAST-MODIFIED:20260422T113202Z
STATUS:CONFIRMED
SEQUENCE:0
SUMMARY:phd seminar - Daniela Horta Sáenz*\, Baptiste Meunier**
DTSTART:20230912T090000Z
DTEND:20230912T101500Z
DESCRIPTION:*In Colombia\, numerous anti-drug policies have been implemente
 d\, including the controversial use of glyphosate through aerial spraying a
 s a national policy. However\, the potential impact of this policy on the c
 ivil population remains largely unknown. We investigate the effect of aeria
 l eradication on educational outcomes. We use a sharp regression discontinu
 ity design that combines school census data with newly digitized maps from 
 the Integrated Monitoring System of Illicit Crops. Our preliminary findings
  indicate that aerial spraying leads to an 7% increase in the dropout rate 
 and a 6% increase in the fail rate\, compared to the control group. Further
 more\, we observe that this shock primarily affects early child human capit
 al accumulation. We document that the income shock is the likely mechanism 
 for the observed effect.**We nowcast world trade using machine learning\, d
 istinguishing between tree-based methods (random forest\, gradient boosting
 ) and their regression-based counterparts (macroeconomic random forest\, gr
 adient linear boosting). While much less used in the literature\, the latte
 r are found to outperform not only the tree-based techniques\, but also mor
 e “traditional” linear and non-linear techniques (OLS\, Markov-switchin
 g\, quantile regression). They do so significantly and consistently across 
 different horizons and real-time datasets. To further improve performances 
 when forecasting with machine learning\, we propose a flexible three-step a
 pproach composed of (step 1) pre-selection\, (step 2) factor extraction and
  (step 3) machine learning regression. We find that both pre-selection and 
 factor extraction significantly improve the accuracy of machine-learning-ba
 sed predictions. This three-step approach also outperforms workhorse benchm
 arks\, such as a PCA-OLS model\, an elastic net\, or a dynamic factor model
 . Finally\, on top of high accuracy\, the approach is flexible and can be e
 xtended seamlessly beyond world trade.\\n\\nContact: Lucie Giorgi: lucie.gi
 orgi[at]univ-amu.frRicardo Guzman: ricardo.guzman[at]univ-amu.frNatalia Lab
 rador: natalia.labrador-bernate[at]univ-amu.frNathan Vieira: nathan.vieira
 [at]univ-amu.fr\n\nPlus d'informations: https://www.amse-aixmarseille.fr/en
 /events/daniela-horta-s%C3%A1enz-baptiste-meunier-0
LOCATION:Îlot Bernard du Bois - Amphithéâtre\, AMU - AMSE\, 5-9 boulevar
 d Maurice Bourdet\, 13001 Marseille
URL;VALUE=URI:https://www.amse-aixmarseille.fr/en/events/daniela-horta-s%C3%A1enz-baptiste-meunier-0
CONTACT:Lucie Giorgi: lucie.giorgi[at]univ-amu.frRicardo Guzman: ricardo.gu
 zman[at]univ-amu.frNatalia Labrador:&nbsp\;natalia.labrador-bernate[at]univ
 -amu.frNathan Vieira: nathan.vieira[at]univ-amu.fr
TRANSP:OPAQUE
END:VEVENT
END:VCALENDAR