25 Years IWH

Dr Peter Sarlin

Dr Peter Sarlin
Current Position

since 2/17

Executive Chairman and Chief Scientist


since 1/15

Research Affiliate

Halle Institute for Economic Research (IWH) – Member of the Leibniz Association

since 1/14

Associate Professor of Economics

Hanken School of Economics (Helsinki, Finland), and Director of RiskLab Finland

Research Interests

  • systemic risk analytics
  • macroprudential policy
  • machine learning and visual analytics

Peter Sarlin is a Research Affiliate at the IWH since 2015. He is an Associate Professor of Economics at Hanken School of Economics (Helsinki, Finland), and executive chairman and chief scientist of Silo.AI.

Peter Sarlin received his PhD (Econ) from the Department of Information Technologies, Åbo Akademi University (Turku, Finland), in 2013. He has also studied at LSE, at Stockholm School of Economics and at Stockholm University.

Your contact

Dr Peter Sarlin
Dr Peter Sarlin
Mitglied - Department Macroeconomics
Send Message Personal page


Working Papers


Optimizing Policymakers' Loss Functions in Crisis Prediction: Before, Within or After?

Peter Sarlin Gregor von Schweinitz

in: ECB Working Paper Series, No. 2025, 2017


Early-warning models most commonly optimize signaling thresholds on crisis probabilities. The expost threshold optimization is based upon a loss function accounting for preferences between forecast errors, but comes with two crucial drawbacks: unstable thresholds in recursive estimations and an in-sample overfit at the expense of out-of-sample performance. We propose two alternatives for threshold setting: (i) including preferences in the estimation itself and (ii) setting thresholds ex-ante according to preferences only. Given probabilistic model output, it is intuitive that a decision rule is independent of the data or model specification, as thresholds on probabilities represent a willingness to issue a false alarm vis-à-vis missing a crisis. We provide simulated and real-world evidence that this simplification results in stable thresholds and improves out-of-sample performance. Our solution is not restricted to binary-choice models, but directly transferable to the signaling approach and all probabilistic early-warning models.

read publication
Mitglied der Leibniz-Gemeinschaft LogoTotal-Equality-LogoWeltoffen Logo