Not toolkit, but related and hopefully interesting to you. Building Econ Theory in Lean, and on the side to also prove which algorithms solve which models. You can join the LeanEconomics org and contribute.
LeanEconomics
https://t.co/U4Y1n6Wl6o
Current: prove existence of stationary general eqm for Aiyagari 1994, plus uniqueness for mu=1 (already known, now Lean formalized).
Please feel free to join the Github org and contribute to Econ in Lean.
Explain: https://t.co/VNw6X5wRdQ
Claude + VFI Toolkit = Superpower
Wrote a replication of core result of Huggett & @King_ofSweden (2000). Four OLG models in under three hours!!!
Key: Toolkit provides structure, code needed is only a few hundred lines. Claude loves this.
Details: https://t.co/0pi9p37YRp
VFI Toolkit v2.5: Breaks all infinite-horizon models.
Just find-and-replace-all "_Case1" with "_InfHorz"
https://t.co/DICVUC5VpC
Tees up some cool future functionality, with Entrepreneurial-Choice later this month, and aggregate shocks later this year...
Publication using VFI Toolkit:
https://t.co/nqSrQw0IBt
OLG model with two endogenous states, assets and human capital. Toolkit commands solve the OLG and calibrate eqm to match life-cycle microdata. Then perform optimal tax exercise.
VFI Toolkit handles it all comfortably!
Online Video Workshop on VFI Toolkit
https://t.co/MMpDmg3CKA
Original was taught in-person at LSE in May 2025.
Videos cover toolkit concepts and walk through codes. All slides & codes available to play with. Great intro to VFI Toolkit.
OLG models for policy institutions? The Office of Budget Responsibility (UK) and the Reserve Bank of New Zealand both use VFI Toolkit to build OLG models for policy analysis and advice. VFI Toolkit makes these models much simpler to build, modify, and analyse.
(New) Life-Cycle Model 27
Discretize earnings dynamics 'Model 5' of Guvenen, Karahan, Ozkan & Song (2021), use in exo-labour life-cycle model.
All discretization methods are extensions of Tanaka-Toda method. Lots of gaussian-mixtures :D
https://t.co/RKwNlrSj4t
Human Capital!
Life-Cycle Model 42: Female Labor Force Participation History
Life-Cycle Model 43: Uncertain Human Capital
https://t.co/U0o4fE9wl4
(two new life-cycle models in the Intro to Life-Cycle Models, both have assets + human capital)
(New) Example: Huggett, Ventura & Yaron (2006)
https://t.co/cFRO0FKrhd
Does two things:
i) use PSID data in Matlab to estimate life-cycle profiles
ii) calibrate life-cycle model with human capital to reproduce the life-cycle profiles
GMM Estimation of Life-Cycle models
https://t.co/cwCcqvRUVL
Life-Cycle models 45-50 in the Intro to Life-Cycle models show how to do GMM estimation in VFI Toolkit with just 5-10 lines of code (after model and data work). I'm excited about this! :D
@AlexZevelev@robertdkirkby I would love to see it, but would require someone to code it. VFI toolkit is just one person, and I am already busy with the Matlab code. As a historical note, VFI Toolkit was started when Julia was beta and did not do GPUs, so was not an option at that time.
Breaking update to getFredData()
Output now uses Matlab datetime format. https://t.co/4aizU9Advy(:,1) used to be dates, now in X.Dates; https://t.co/4aizU9Advy(:,2) used to be data values, now in https://t.co/4aizU9Advy. Inputs unchanged, can use datetime as start/enddate.
Matlab replaced datenum with datetime as standard way to handle time/dates (in R2022). getFredData() has now be updated to reflect this by using datetime.
A legacy version of getFredData() using datenum is available in github repo: getfreddata-matlab-legacy
v2.2: Same same, but faster! 🔥
VFI Toolkit is now much faster thanks to implementing Tan improvement for all agent distribution computations (it is so fast the FnsToEvaluate got embarassing, so these were overhauled for speed too)
Huge shout to Eugene Tan for this genius idea!
Behavioural Life-Cycle Models
Impatience, Temptation and Self-Control, Loss Aversion and Ambiguity Aversion can now all be handled in life-cycle models! (and OLG ;)
https://t.co/8bnPU7qSgz