Quantitative modelling
Statistical models, econometrics, model comparison, and hypothesis-driven analysis of structured data.
Quantitative Scientist
Jesper Armouti-Hansen
Economics PhD applying statistical modelling, econometrics, and machine learning evaluation to structured data.
I am a quantitative scientist with a background in economics and a PhD from the University of Cologne. In my current role, I work with machine learning models in an operational context, focusing on how model outputs can be used reliably in practice.
My work includes understanding confidence levels, defining thresholds, and supporting decisions on when results can be used directly and when human review is required.
My academic research covers theoretical and empirical topics in microeconomics, with a focus on behavioral economics, personnel economics, organizational economics, formal modelling, and careful interpretation of results.
Focus
- Econometrics and statistical modelling
- Machine learning model evaluation
- Behavioral and organizational research
Profile
Quantitative work grounded in economic research
My work connects applied data science with formal research training: building models, evaluating evidence, and communicating uncertainty clearly enough to support decisions.
Applied data science
Machine learning workflows with attention to confidence, error analysis, and the practical conditions under which model outputs are useful.
Microeconomic research
Academic work on behavioral economics, incentives, social preferences, and organizational decision-making.
Selected work
Projects
RAG Search Engine
CLI-based RAG search engine for movie data combining BM25, semantic and hybrid search, multimodal CLIP search, Gemini query enhancement, reranking, and generation.
View projectEconomic Theories and Machine Learning
Analysis code connected to research on evaluating economic theories using machine learning techniques.
View projectCitiBike Demand, Risk, and Net Flow Analysis
Demand, risk, and net-flow analysis of CitiBike NYC from 2023 to 2025, using trip and collision data to examine usage patterns, station-level risk, and operational dynamics.
View projectBuild AI Agent
A simple Gemini-powered coding agent that can inspect files, run Python scripts, and write code through tool-calling.
View projectResearch
Selected publications
Publication 2024
Efficiency Wages with Motivated Agents
Games and Economic Behavior, 145, pp. 66-83
Abstract
Many jobs serve a social purpose beyond profit maximization. This paper uses a modified principal-agent gift-exchange game with positive externality to study how workers' prosocial motivation interacts with efficiency wages in stimulating effort. The results show that prosocial motivation and efficiency wages are independent in stimulating effort, while principals offer higher wages in the prosocial treatment because they underestimate reciprocity in the standard gift-exchange environment.
Publication 2024
Managing Anticipation and Reference-Dependent Choice
Journal of Mathematical Economics, 112, 102988
Abstract
The paper develops a model of reference-dependent choice in which the reference point may be any convex combination of possible outcomes under a consumption lottery. It introduces solution concepts, characterizes them on choice data, and identifies the model's parameters.
Publication 2020
Optimal Contracting with Endogenous Project Mission
Journal of the European Economic Association, 18(5), pp. 2647-2676
Abstract
The paper studies how organizations can choose a project mission to attract, incentivize, and screen workers. It analyzes how contractual environments shape the optimal distance between the organization's preferred mission and the agents' preferred mission.
Toolkit