Static models freeze the economy to measure it.
GAGE runs it forward, year by year.
GAGE is a full general equilibrium model of Georgia. Prices move, wages move, capital accumulates, and people decide whether to move here — all of it at once, every year, for fifteen years. A policy lands in year one; you get the path the state actually takes, the band around it, and who along the income distribution ended up better off.
Licensed users sign in. Not licensed yet? Request access →
Deviation from the no-policy control path. Central migration response, housing-congestion margin on. Ranges are the published elasticity band, not a confidence interval. Illustrative results — the model itself runs server-side.
An IMPLAN multiplier assumes prices never move, workers are never scarce, and everything happens at once. Georgia's economy does none of the three.
Fixed coefficients are the right instrument for one project's supply chain — that is exactly what our own Georgia Input-Output Model is built for, and it says so on its own front page. They are the wrong instrument for a policy that moves wages, moves prices, and changes where people choose to live.
A shock goes in. An economy comes out — year by year.
GAGE is a computable general equilibrium model: every market clears, every price is an outcome rather than an assumption, and the solution in one year becomes the starting stock for the next. That is the difference between an estimate of what a project is worth and a forecast of what a policy does.
Define the scenario
Stack as many levers as the policy actually has: productivity, export prices, final demand, the state income tax rate, transfers, government purchases, labor supply — statewide or in one of 26 sectors. How you pay for it is a choice you make, not an assumption we bury.
Solve the equilibrium
375 equations resolve simultaneously — one Georgia equilibrium, one year. A full run repeats that solve across sixteen years and three migration-elasticity paths, with the nine-tier household system on top: 27,296 equations, 64 equilibria. Producers substitute between labor and capital, households re-optimize, importers and exporters respond to relative prices, and the state budget moves with the tax base. Nothing clears by assumption; everything clears by price.
Trace the path
Then it runs again, and again. Capital accumulates out of investment net of depreciation, the labor force grows with demographics and with migration, and rising housing costs push back on the people the boom attracted. Fifteen years of adjustment, reported one year at a time.
Eight things you cannot get from a multiplier.
Each of these is a question a legislator, a board, or an opposing analyst will ask. GAGE answers all eight from the same set of accounts, in the same run.
When the benefit arrives is usually the argument
An external investment of $100M a year raises Georgia output by $65M in year one and $151M by year fifteen — because capital and workers take time to arrive. A static model reports one of those numbers and calls it the answer. Both are on the page here.
Every headline ships with its range
Elasticities come from published literature with published ranges. GAGE reports the band as standard output — never a single number dressed up as precision.
Housing costs bite back
Population growth pushes up relative housing costs, which damps further in-migration. Without that margin, every long-run population answer is an upper bound. Most models leave it out.
Nine income tiers, at any year on the path
Aggregate growth is not the question a legislator gets asked at a town hall. GAGE re-solves a nine-household version of the model at any year of the run and reports the money-metric gain for each income tier — so "who benefits" has an answer with a number attached.
Where the money comes from changes the answer
Demand funded from Georgia savings crowds out investment; demand funded from outside does not. A tax cut can be deficit-financed or offset. You choose; the model does not choose for you.
Show your assumptions, then test them
Eight labor-market and macro closure rules ship as switches. Run the same policy under each and show a client exactly how much of the answer is economics and how much is assumption.
The county is the atomic unit
State results allocate down to all 159 counties, then sum back up into whatever region the engagement needs — a regional commission, an MSA, a twelve-county service area plus rest-of-state, or a custom set you draw. The reporting geography is a setting, not a rebuild.
Zero parameters tuned to a target
Every elasticity, share and behavioral parameter traces to a published source and carries a documented range. Nothing was adjusted to make a result come out right — a standing rule of the build, and the reason the sensitivity bands mean what they say.
Built in the open, from accounts you can audit.
Economic Impact Group built GAGE from published federal data and published method. There is no black box to take on faith, and no licence that expires. If a client's analyst wants to interrogate an equation, there is an equation to interrogate.
A Georgia social accounting matrix
The foundation is a balanced SAM: EIG's Georgia input-output accounts extended with factor accounts, institution accounts, real federal-state-local tax and transfer flows, a savings and investment account, and separate external accounts for the rest of the U.S. and the rest of the world. Built from BEA regional accounts, Census government finance, and BLS wage records; balanced with cross-entropy methods.
Production, trade, and a very open state
Leontief intermediates with CES value-added nests in production; Armington import demand and CET export supply governing trade. Georgia trades far more freely with Tennessee than the U.S. does with Germany, and the substitution elasticities are set accordingly — the single biggest behavioral difference between a state model and a national one, and the one most often gotten wrong.
Recursive dynamics
Capital carries forward as last year's stock net of depreciation plus this year's investment. The labor force grows with demographic trend and with net migration. Productivity trends forward on measured sectoral growth. Each year's equilibrium is the next year's starting point — which is what produces an adjustment path instead of a jump.
Migration on the published REMI form
Net migration responds to Georgia's relative employment opportunity and its relative after-tax real wage, in the functional form Treyz and colleagues published in the 1990s — measured as a deviation from the control path, so baseline trends stay where they belong. After-tax matters: a policy that leaves take-home pay higher is a reason to move to Georgia, and a model built on gross wages is structurally blind to it. The response parameters carry a literature range, and that range is the dominant source of the long-run band.
A housing-congestion margin
In-migration raises population, population raises relative housing costs at a rate set by the metro housing-supply elasticity, and higher housing costs cut into the real wage that attracted the migrants. Atlanta is among the more elastic large metros, so the damping is real but modest — which is itself a finding, and a defensible one, because the elasticity came from the literature rather than from tuning.
Validation without a reference model
Benchmark replication to machine precision. Homogeneity and Walras' law, which catch equation errors ruthlessly. Convergence on EIG's input-output multipliers when the behavioral margins are switched off — the bridge between the model economists already trust and this one. And systematic sensitivity sweeps across every key elasticity, which is where the reported bands come from.
Every new margin has to prove it changed nothing else
The model grows by documented margins, never by patches. A new behavioral channel ships only when three things hold: switched off, it reproduces the previous vintage bit for bit; every parameter it introduces carries a published source and a published range; and the results it moves are reported both ways, with the superseded figures kept as documented bounds rather than quietly replaced. That is precisely how the housing-congestion margin entered the model, and how the after-tax migration signal cleared the same gate after it — both on by default now, both with the figures they superseded still on the record beside them. It is the reason a band from GAGE means something.
simultaneously, every year
across the state economy
reported on demand
to hit a target
Who gains — and how long they wait.
The same run, re-solved across nine household income tiers. Below: the money-metric welfare gain from a permanent one-percent statewide productivity improvement, as a percent of each tier's own consumption, in year one and again in year fifteen. Every tier gains — and the shape of the gain is the conversation.
Read it two ways. Everyone is better off, which is the case for the policy. The gain is roughly twenty times larger at the top than at the bottom, which is the case for pairing it with something else. GAGE will not tell you which argument to make — it will make sure both of them are numerate.
Three questions. Three instruments. One set of accounts.
Economic development analysis asks three different things, and using one tool for all three is how defensible numbers go wrong. Each of ours answers the question it was built for, and they reconcile because they are built on the same Georgia accounts. GAGE is the one you reach for when the question is about the state economy rather than a single project.
What is this project worth?
Regional multipliers across 402 BEA industries for any Georgia geography. Output, jobs, labor income, value added — direct, indirect, induced, with uncertainty shown.
What does it do to the local budget?
Revenue and cost of service, jurisdiction by jurisdiction, over the life of a project — including what an incentive actually costs the taxing bodies that grant it.
What does the policy do to Georgia?
The whole state economy, adjusting over fifteen years — jobs, income, population, capital, and state revenue, with bands and a distributional read.
- 1Statewide policy. A tax change, an incentive program, an infrastructure package, a workforce initiative — anything whose effect runs through prices, wages, and migration rather than through one project's supply chain.
- 2Anything with a time dimension. When the question is not "how much" but "how soon", "for how long", and "what does year ten look like" — GAGE is built to answer in years, not totals.
- 3Anything that will be contested. Documented sources, reported ranges, switchable assumptions, and a methodology you are allowed to show. Built for the analysis that gets read by someone hoping to find a hole in it.
Run the policy you'll have to defend in a hearing.
GAGE is available to state agencies, statewide organizations, and institutional clients under annual licence, and to sponsors as part of a commissioned study. Every engagement includes the technical documentation.
Tell us the question. We'll tell you whether GAGE is the right instrument.
Licensing, scoping a commissioned analysis, or a walkthrough of the methodology before you commit to anything — write to the economists who built it.