How we got here
The architecture has been tested by ministries and central banks since the 1970s. The models running on it today carry a 2024 base year, a full financial system and two classes of AI capital, and are re-solved for every engagement.
The models sit in the ORANI, MONASH and USAGE tradition of computable general equilibrium modelling, which began in Australia in the late 1970s with the first large-scale model of its kind. Finance ministries and central banks have used that family ever since. Dr. Ashley Winston has worked in it for thirty years, advising governments, corporations and NGOs in more than forty countries.
Every sector carries a complete balance sheet: households, firms, government, banks and the rest of the world hold deposits, bonds, credit and equity, and the accounts reconcile every year. Prices and wages are determined rather than assumed, so a rate rise produces both a contraction and disinflation. Capital comes in five classes, two of them AI, split by whether the technology substitutes for labour or augments it. Every mechanism is certified against a registered prediction before it is switched on. The base year is 2024.
Five generations
The first large-scale computable general equilibrium model. Comparative static. Used in dozens of countries and the basis for GTAP.
The first large-scale dynamic CGE model. Recursive dynamics, 113 industries, 115 commodities.
Built for the United States. 500 industries, 700 occupations, every state. Applied across federal agencies on high-profile policy reforms.
Emissions added. A full tax module. Multiple investor classes. Endogenous technical change.
A complete financial system, with a balance sheet for every sector. An explicit nominal anchor. Two classes of AI capital. Registered-prediction certification. Base year 2024. Language models narrate the results; surrogate models make scenarios fast.
Phylleos began with a two-year search for a way to measure carbon emissions at the source, independently of what emitters reported. Emissions are the residue of every industry, input and transaction that produced them, so measuring them honestly requires a model of the economy that generated them.
That search led to Ashley, who had spent decades building exactly that. His models answered far more than the carbon question, and the company was built around what else they could do.
For twenty years this work was delivered as bespoke engagements. A ministry or a board commissioned a model, the report was delivered, and the model went into a drawer. The client kept the conclusions and lost the instrument that produced them.
Phylleos built the core architecture once and extends it model by model, country by country. Clients keep the instrument.
The founders
Dr. Ashley Winston is Co-founder and Chief Economist of Phylleos, Inc. He is an internationally recognised economist and an original author of large-scale dynamic economic models, with 30 years advising governments, corporations and NGOs in more than 40 countries on fiscal, energy, climate, trade, infrastructure and macroeconomic policy.
He was previously Chief Economist and Managing Director at Millstein & Co. and at KPMG, founder and CEO of The MacroDyn Group, and a senior adviser to U.S. federal agencies. He holds a PhD in economics from Monash University and an adjunct appointment at the Centre of Policy Studies, Victoria University.
Taiya spent her career negotiating, across the U.S. government, with our allies and competitors, especially China, and at times with rebel groups. The work taught her how much depends on numbers that are correct and reliable, and the need to have them fast. Analysis that arrives after the room has emptied is worth nothing.
Speed is what led her to AI, and Phylleos is the result: the numbers a decision-maker needs when they need them, with an account of where they came from and where they are going. She held senior roles at the U.S. Treasury and the U.S. Department of State and worked in private advisory before co-founding the firm.
Where we work
Headquartered close to the decisions the models are built for, with colleagues in Auckland.
Phylleos