Summary

Michael Manolakis

I am a Chicago-based computational economist and data analyst with nine years of prior professional experience in regulated real estate valuation and quality-control auditing. I hold a B.A. in Quantitative Economics (Cum Laude, GPA 3.55) and have completed all coursework and my capstone defense for my M.A. in Economics (GPA 3.51; degree conferral December 18, 2026) at the University of Hawaiʻi at Mānoa.

My work bridges empirical economics, reproducible data engineering, and regulatory risk review. Across my graduate capstone and professional appraisal career, I have built multi-language ETL and imputation pipelines across 16,297 properties and more than 3,000 U.S. counties, authored more than 500 USPAP-compliant valuation reports, and conducted quality-control review audits issuing approval or rejection determinations for appraisal management companies.

Currently, I run an independent local inference benchmarking program evaluating more than 60 open-weight Large Language Models (45+ base architectures) for quality, reliability, and latency on unified-memory hardware, alongside experimental economics research in Python/oTree. This website serves as a portfolio of my empirical research, computational pipelines, and professional background.

Technical Stack & Methodologies

  • Languages & Tools: Python (pandas, GeoPandas, LightGBM, miceforest, oTree), R (tidyverse, sf), Julia, SQL, Stata, MATLAB, Advanced Excel, Git/GitHub, Docker, Linux, Google Cloud Platform, and LaTeX.
  • Econometrics & Spatial Data Engineering: Spatial econometrics (GIS, OpenStreetMap, Cadastral and Zoning point-in-polygon joins), Multiple Imputation by Chained Equations (MICE), LightGBM, OLS/Panel Regression, and Rubin's Rules pooling across independent Python, R, and Julia pipelines.
  • AI Evaluation & Reproducible Infrastructure: Local inference benchmarking of 60+ open-weight AI model variants using llama.cpp on 128GB AMD Ryzen AI Max+ 395 unified-memory hardware, driven by an 18-prompt execution-graded coding suite in Python, R, Julia, and Java under Git and Docker.

History

Education

  • University of Hawaiʻi at Mānoa
    • M.A. in Economics (GPA 3.51; all coursework complete, capstone defended Sept. 3, 2026, degree conferral Dec. 18, 2026) — The Ostrich Effect (GitHub)
    • B.A. in Economics, Quantitative Economics Concentration, Cum Laude (GPA 3.55, May 2025)
  • International HPC Summer School (IHPCSS) 2026
    • Selected Participant (fully funded via ACCESS), Perth, Australia (July 2026). Completed coursework in HPC Python, Applied Machine Learning at Scale, OpenMP, Parallel I/O, and Scientific Visualization.
  • Independent Technical Coursework (100+ Hours)
    • University of Helsinki (mooc.fi): Python Programming, Data Analysis with Python, Building AI, Elements of AI, and Introduction to Cyber Security.

Professional Experience & Former Credentials

  • Harani Appraisals, Honolulu, HI (Aug 2016 – Dec 2025)
    • Certified Residential Appraiser (Jan 2023 – Dec 2025): Authored 500+ USPAP-compliant valuation reports, merged tax, MLS, and municipal zoning datasets, maintained a multi-year Excel reconciliation ledger, and conducted USPAP-compliant quality-control review audits of external appraisers' reports for AMCs.
    • Earlier Progression: State Licensed Residential Appraiser (Oct 2020 – Jan 2023), Residential Appraiser Trainee (Jan 2019 – Oct 2020), and Appraisal Assistant (Aug 2016 – Jan 2019).
  • Former Professional Licenses (State of Hawaiʻi DCCA)
    • Certified Residential Appraiser — CRA-1543 (Former; 01/25/2023 – 12/31/2025; previously FHA-approved)
    • State Licensed Appraiser — SLA-1410 (Former; 10/07/2020 – 12/31/2023)
    • Real Estate Appraiser Trainee — REAT-1346 (Former; 01/04/2019 – 12/31/2021)

Teaching & Leadership

  • Graduate Assistant (Teaching Assistant), UH Mānoa (Aug 2025 – Aug 2026)
    • Principles of Microeconomics (ECON 130): Instructed and mentored 256 undergraduates across two sections (147 and 109 students), led supplemental review sessions, held weekly office hours, and graded examinations and quantitative problem sets with written feedback.
  • Mānoa Economics Association (MEA) (Aug 2024 – May 2026)
    • President (Aug 2025 – May 2026) · Vice President (Jan 2025 – Aug 2025) · Treasurer (Aug 2024 – Dec 2024)
    • Built the organization's Excel budget model and pitched funding proposals to university fee boards (ASUH and SAPFB), securing a 100% approval rate across all funding cycles.
    • Designed and delivered workshops on graduate research, reproducible LaTeX/Overleaf templates, and AI model resources; recruited department faculty to present research.

Completed Work

M.A. Capstone: The Ostrich Effect

Ignoring the Opportunity Cost of U.S. Golf Courses

Defended Sept. 3, 2026

This capstone quantifies the macroeconomic opportunity cost and spatial misallocation of recreational land across 16,297 U.S. golf courses and 3,000+ counties, paired with a parcel-level cadastral and zoning validation study in Honolulu County (1,072 Tax Map Key parcels across 36 courses).

Interactive: The Opportunity-Cost Cascade

Step through how the $935B gross national headline narrows under developer economics ($608B central / $374–702B range) and binding zoning law — and see the Preservation Paradox on Oahu.

Key Methodologies & Findings

  • Tri-Language Validation Pipeline: Independent Python, R, and Julia implementations benchmarked against each other so an error in one implementation surfaces against the others.
  • Multiple Imputation (MICE): Resolved 28.8% missing acreage using Random Forest / LightGBM-backed MICE (M = 100 per language) pooled via Rubin's rules.
  • National Findings: Estimated $935 billion gross national opportunity cost ($790B observed-only floor across 10,925 doubly-observed courses), narrowing to $374–$702B ($608B central at a 35% discount) after developer economics.
  • Honolulu Micro-Validation: Parcel-level Oahu cadastral analysis across 36 courses (5,810 acres; 1,072 TMK parcels) yielding $28.78B gross opportunity cost, with urban courses valued at roughly 55 times rural courses per acre nationally.
  • Preservation Paradox: 81.0% ($23.30B) of Oahu's golf footprint sits in Preservation or Federal/Military zones where housing is legally prohibited, leaving only $1.37B (4.75%) legally realizable under current zoning.

Current & Ongoing Work

Open-Weight LLM Evaluation & Benchmarking

I develop and execute a local inference benchmarking program evaluating the quality, reliability, and latency of 60+ open-weight Large Language Model and quantization variants (45+ base architectures) with llama.cpp on 128GB AMD Ryzen AI Max+ 395 unified-memory hardware. Models are evaluated against an 18-prompt execution-graded coding suite in Python, R, Julia, and Java with reproducible pipelines maintained under Git and Docker.

Behavioral Economics Experimentation (Python/oTree)

I designed and built a runnable incentivized behavioral economics experiment ("The Lame-Duck's Revenge") in Python/oTree examining responses to loss of a controlling position with endogenous and exogenous removal mechanisms, supported by synthetic LLM-agent pretesting using locally hosted open-weight models.

Journal Publication of The Ostrich Effect

Following my September 2026 capstone defense, I am preparing an extended journal manuscript expanding the parcel-level zoning and tax-assessment validation across additional coastal housing markets.

The Ostrich Effect

A $935 billion number that shrinks when you look at it

The headline is easy to quote and easy to misread. Step through what happens to it once real-world constraints are applied — developer economics, then the law itself.

Q

What we actually measured · Oahu (36 courses · 5,810 acres)

The Preservation Paradox

Of Oahu’s $28.78B in golf-land opportunity cost across 1,072 TMK parcels, this is where the law lets it go — and where it doesn’t.

  • $23.30B Preservation / Federal (81.0%) — housing prohibited
  • $4.11B Agriculture (14.3%) — possible with a zoning amendment
  • $1.37B Resort / Residential / Other (4.7%) — buildable today

If a jurisdiction will allow eighteen holes of irrigation and fertilizer on land it calls “Preservation,” the label has stopped protecting ecology. It is protecting an incumbent.

The finding isn’t the $935B. It’s the distance between it and the sliver — the opportunity cost we’ve zoned ourselves into ignoring.

Where the $935B concentrates

Opportunity cost isn’t spread evenly across the map. A handful of coastal markets carry most of it.

Maps & figures from the paper

The visuals behind the analysis, drawn straight from the capstone.

About

Michael Manolakis

My career sits at the intersection of applied quantitative economics, regulated valuation, and reproducible data engineering. Before relocating to Chicago and completing my graduate degree in Economics at the University of Hawaiʻi at Mānoa, I spent nine years in residential real estate appraisal—progressing from Assistant and Trainee to State-Certified Residential Appraiser.

Working on both sides of the valuation desk shaped how I approach data and modeling. As a producer, I authored more than 500 USPAP-compliant reports grounded in municipal zoning, tax records, and comparable market data. As a reviewer, I audited external appraisers' reports for Appraisal Management Companies, challenging unsupported methodologies and issuing approval or rejection determinations under deadline.

That same verification discipline carries into my computational work: cross-checking spatial econometric pipelines independently across Python, R, and Julia so errors surface against each other, and building execution-graded coding benchmarks to test open-weight AI models for reliability and failure modes before relying on their outputs.

Beyond technical work, I value clear communication and institutional ownership. As President, Vice President, and Treasurer of the Mānoa Economics Association, I built the organization's Excel budget model, secured a 100% funding approval rate across all university fee-board cycles, and ran technical workshops on graduate research and AI tooling. As a Graduate Assistant, I instructed and mentored 256 undergraduates across two microeconomics sections.

Technical Philosophy

I build for reproducibility, data privacy, and verifiable outputs. Whether implementing tri-language spatial pipelines under Git and Docker or benchmarking 60+ open-weight LLMs locally on 128GB unified-memory Linux hardware, I favor open-source tooling, transparent assumptions, and repairable systems where every number can be audited back to source data.

Thank you for visiting my site.

The Bases

All your base are belong to us