This feature is exclusively available to Starter, Research, and Pro. Upgrade when you need this workflow, review pricing, or send a question before changing plans.
What you keep on Free
How do you build a complete quantitative accounting of an entire economy -- consumption, investment, trade, government, labor markets -- in a single internally consistent framework?
Jan Tinbergen built the first macroeconometric model for the Netherlands in 1936, then extended the approach to the US economy for the League of Nations in 1939. Lawrence Klein refined the methodology at the Cowles Commission in the 1950s, producing the Klein-Goldberger model: a system of roughly 20 simultaneous equations estimated by limited-information maximum likelihood. Klein won the 1980 Nobel Prize for this work. The approach matured into the large-scale models of the 1960s-70s: the Wharton model (Klein and Evans), the Brookings model, the MPS model at the Federal Reserve, and Ray Fair's multicountry system. These models contained 50-500 equations covering consumption, investment (by sector), housing, international trade, government budgets, wage-price dynamics, and financial markets.
The core mechanism: each behavioral equation represents one economic relationship -- a consumption function, an investment accelerator, a Phillips-curve wage equation, an import demand function. The equations are simultaneous because endogenous variables appear on the right-hand side of other equations. GDP is determined by the national accounts identity (C + I + G + NX), but C depends on GDP through the consumption function, and I depends on GDP through the accelerator. This circularity makes single-equation OLS biased. The Cowles Commission's solution was simultaneous-equations estimation: two-stage least squares (2SLS), three-stage least squares (3SLS), or full-information maximum likelihood (FIML).
The Lucas critique (1976) challenged the entire enterprise. If behavioral parameters shift when policy changes, a model estimated under one regime cannot predict outcomes under another. Sims (1980) attacked from a different angle: the identifying restrictions in macroeconometric models are 'incredible' -- the choice of which variables to exclude from which equations is arbitrary and drives the results. These critiques triggered the profession's pivot toward VARs and DSGE. But large-scale models never disappeared from institutional forecasting. The Federal Reserve's FRB/US, the Bank of England's NIESR model, the European Commission's QUEST III, and Moody's Analytics macro model are all descendants of the Cowles tradition, updated with rational expectations, error correction, and flexible functional forms.
Modern macroeconometric systems differ from their 1970s ancestors in three ways. First, expectations formation is more sophisticated: some equations use model-consistent forward-looking expectations, others use VAR-based or survey-based expectations. Second, long-run properties are anchored by error-correction mechanisms or calibrated steady states, preventing the model from drifting at long horizons. Third, stochastic simulation replaces deterministic baseline-plus-multiplier analysis. The model is shocked with draws from the estimated residual distribution to produce fan charts and probability assessments.
Inputs span the national accounts and beyond. Real GDP components (consumption, fixed investment by type, inventory investment, government spending, exports, imports), price indices (GDP deflator, CPI, PPI, import prices), labor market variables (employment, unemployment, wages, labor force participation, hours worked), financial variables (short and long interest rates, equity prices, exchange rates, credit aggregates), fiscal variables (tax revenues by type, transfer payments, government debt), and external variables (world GDP, commodity prices, foreign interest rates). A typical modern system has 30-60 behavioral equations and 100-300 identities.
The model divides into blocks. The demand block determines GDP components through behavioral equations (consumption function, investment equations, trade equations) and the national accounts identity. The supply block determines potential output, wages, and prices through production functions and wage-price spirals. The financial block links interest rates, credit, and asset prices to the real economy. The fiscal block computes government revenues, expenditures, and debt dynamics endogenously. The external block determines exchange rates and trade flows. Each block feeds into others: demand determines output and employment, which determines wages and prices, which feed back to demand through real income and competitiveness.
Estimation proceeds equation by equation or block by block, using 2SLS or GMM to handle simultaneity. Some parameters are calibrated (tax rates, depreciation rates, accounting identities) rather than estimated. The model is then solved simultaneously: given exogenous inputs and lagged endogenous variables, a nonlinear equation solver (Newton-Raphson or Fair-Taylor iterative method) finds the values of all endogenous variables that satisfy all equations simultaneously. When forward-looking expectations are present, the solution requires an outer loop iterating until expectations are consistent with the model's own projections.
The Federal Reserve's FRB/US model is the direct descendant of the MPS model (Modigliani, Phelps, et al.). It contains approximately 60 stochastic equations and 300+ identities covering the US economy. The Board staff uses FRB/US for the Tealbook forecast (now Summary of Economic Projections support), policy-rule simulations, and alternative-scenario analysis. Every FOMC meeting cycle includes FRB/US simulations showing what happens under different rate paths. The Congressional Budget Office's macroeconomic model, used for scoring fiscal legislation, is a macroeconometric system in the same tradition.
Moody's Analytics operates one of the largest commercial macroeconometric models, covering 80+ countries with 1,500+ equations per country model. Banks, insurers, and asset managers use Moody's model output for CCAR/DFAST stress testing (Federal Reserve regulatory requirement), commercial real estate valuation, and credit loss forecasting. Oxford Economics and IHS Markit (now S&P Global) run similar global macroeconometric systems for corporate planning and sovereign risk assessment.
Multicountry macroeconometric models (NiGEM at NIESR, QUEST III at the European Commission, the IMF's MULTIMOD successor) are used for analyzing policy spillovers across countries. When the ECB models the impact of a US fiscal expansion on Eurozone GDP, the channel runs through US import demand, dollar depreciation, European export volumes, and Eurozone monetary policy response -- all endogenous within the linked system.
Do not use a macroeconometric system when you need micro-founded welfare analysis (use DSGE), when data availability is limited (the model requires complete national accounts), when the forecast horizon is very short (1-2 quarters -- use nowcasting models or BVARs), or when the institutional capacity for model maintenance is absent (large-scale models require dedicated teams of 5-15 economists for ongoing development and update).
GDP is the sum of consumption, investment, government spending, and net exports. This identity must hold exactly in every solution period.
Aggregate consumption depends on disposable income, wealth, real interest rates, and habit persistence. Estimated by 2SLS with instruments for endogenous income.
Business fixed investment depends on output (accelerator), Tobin's q or user cost of capital, long-term real interest rates, and adjustment costs captured by lagged investment.
Nominal wages depend on expected prices, the unemployment rate (or gap), labor productivity, and lagged wages. Part of the wage-price spiral.
Output price level depends on unit labor costs, import prices, the output gap, and lagged prices. Markup pricing over costs with demand pressure.
Import volume depends on domestic demand, relative prices (domestic vs. foreign), and the real exchange rate. Estimated with trade-weighted price indices.
Tax revenues by type (income, payroll, consumption, corporate) are modeled as functions of their respective tax bases with statutory or effective rates.
Each behavioral equation has a stochastic residual. In forecasting mode, staff set add-factors (non-zero residuals) to incorporate judgment about near-term deviations from estimated relationships.
Each behavioral equation excludes specific endogenous variables that appear in other equations. These exclusion restrictions provide the identifying information needed for consistent estimation of simultaneous equations.
If violated: If exclusion restrictions are wrong (a variable excluded from the consumption function actually belongs there), 2SLS estimates are inconsistent. Instrument relevance and overidentification tests can detect some violations but not all.
Behavioral parameters (marginal propensity to consume, Phillips curve slope, trade elasticities) are stable over the estimation sample and the forecast horizon.
If violated: The Lucas critique: parameters may shift when policy regimes change. Structural breaks in consumption behavior (financial liberalization), trade patterns (globalization), or price-setting (anchored expectations) render pre-break estimates unreliable.
Behavioral equations are linear or log-linear in parameters. Nonlinearities, where present, are specified a priori (e.g., Okun's law with an asymmetric gap).
If violated: Misspecified functional form produces biased forecasts at extreme values. A linear consumption function cannot capture the asymmetric marginal propensity to consume between liquidity-constrained and unconstrained households.
Instruments used in 2SLS are correlated with the endogenous regressors (relevance) and uncorrelated with the structural error (exogeneity). Lagged endogenous variables and exogenous variables serve as instruments.
If violated: Weak instruments produce 2SLS estimates with large bias toward OLS. Invalid instruments (e.g., lagged variables that are correlated with autocorrelated errors) produce inconsistent estimates. The cure (more or better instruments) is rarely available in macroeconomic data.
Each behavioral equation includes all economically relevant determinants. Omitted variables that are correlated with included regressors cause bias.
If violated: In a 300-equation model, specification errors in one block propagate through the simultaneous solution to all other blocks. A misspecified wage equation contaminates the price block, the real income calculation, the consumption function, and ultimately GDP.
The nonlinear simultaneous-equation solver converges to a unique solution for the endogenous variables given exogenous paths and initial conditions.
If violated: Multiple equilibria or non-convergence. With strong nonlinearities or forward-looking expectations, the Fair-Taylor solution may cycle or diverge. This is a practical modeling crisis, not a theoretical curiosity -- modelers sometimes encounter it when imposing extreme fiscal scenarios.