Steven Hoekstra

PhD Candidate in Economics
University of Amsterdam & Tinbergen Institute

I study how household balance sheets shape consumption and macroeconomic forecasts, the welfare effects of housing policy, and when estimated models can be trusted to evaluate policy, including when microdata on households and firms improves their counterfactuals. I build agent-based models with optimising households, calibrated to microdata and estimated with simulation-based inference.

Research interests

Macroeconomics
Household balance sheets
Inflation
Complexity
Agent-based modelling
Heterogeneous-agent models
Bounded rationality
Networks
Forecasting
Housing and labour markets
Simulation-based inference
Microdata
High-performance computing

Research

Forecasting with a Dutch Agent-Based Model and Optimally Behaving Households — with Cars Hommes and Frank Pijpers, working paper
Two-asset HANK households optimise under subjective expectations inside an ABM. Twelve-quarter household-consumption forecast errors are 40.5–54.2% lower than matched univariate benchmarks, with the gain concentrated in the 2013–14 consumption trough and not statistically significant with 28 forecast origins; investment gains of 14.5–15.5% are. Runner-up, CEF 2026 Graduate Student Paper Award.
BeforeIT.jl: High-Performance Agent-Based Macroeconomics Made Easy — working paper, with Meligrana, Devetak, Reiter, Minnella, Poledna, Bardoscia and Glielmo
The Bank of Italy's open-source Julia framework for macroeconomic agent-based modelling. I implemented the CANVAS and growth-rate model variants and the multicountry forecasting pipeline with its validation tools.
Identification, Estimation and Misspecification and Their Effects on Forecasting and Policy: Wage–Price Pass-Through in a Dutch Agent-Based Model — single-author working paper
Known-truth experiments show when estimation separates a change in cost pass-through from a wrong wage rule; the framework is applied to the 2021–23 Dutch inflation episode.
Shortage or Misallocation? The Welfare Loss from Dutch Housing Policy, 2015–2035 — work in progress
A spatial macroeconomic ABM with housing, labour-market mobility, demography and firm dynamics, designed to compare policies that expand or reallocate the housing stock. Calibration and welfare comparisons are ongoing.

Education

2023 – present: PhD in Economics, University of Amsterdam
Supervisors: Cars Hommes and Frank Pijpers
Expected thesis submission: May 2027; defence: September 2027
2021 – 2023: MPhil in Economics and Econometrics, Tinbergen Institute (cum laude)
Specialisation: Complexity, Mathematics and Computation
2020 – 2021: MSc Econometrics, University of Amsterdam (cum laude)
Specialisation: Complexity and Economic Behaviour
2019 – 2020: MSc Economics, London School of Economics
Specialisation: Public Economics
2015 – 2019: BSc Econometrics and BSc Economics and Business Economics, University of Groningen
Exchange at RMIT University, Melbourne, 2018 – 2019

Experience

2025: PhD Intern, De Nederlandsche Bank
2023: Thesis Intern, Statistics Netherlands (CBS)
Calibrated the Poledna et al. (2023) agent-based model with Dutch register and national accounts data
2022: Visiting Scholar, International Institute for Applied Systems Analysis (IIASA)
Hosted by Sebastian Poledna
2020 – 2021: Research Assistant, SEO Amsterdam Economics

Award & computing allocations

Runner-up, Society for Computational Economics Graduate Student Paper Award, CEF 2026
For Forecasting with a Dutch Agent-Based Model and Optimally Behaving Households. Award announcement
NWO compute grants
Three awarded grants of one million CPU hours each on Snellius, the Dutch national supercomputer.

References

Prof. Cars Hommes University of Amsterdam (CeNDEF), Bank of Canada, Tinbergen Institute
c.h.hommes@uva.nl
Prof. Frank Pijpers University of Amsterdam (Korteweg-de Vries Institute), Statistics Netherlands
f.p.pijpers@uva.nl

Contact

Email: s.g.hoekstra@uva.nl
Address: Amsterdam School of Economics, University of Amsterdam
Roetersstraat 11, 1018 WB Amsterdam, The Netherlands
Online: github.com/SGHoekstra · LinkedIn