Skip to main content

Cookies on Macro by Mark.We use essential cookies for sign-in, locale, and security. With your permission, we also use optional cookies for preferences, for analytics, and for advertising measurement with the Meta (Facebook) Pixel. Read our cookie policy.

Macro by MarkMacro by Mark
  • Home
Loading…
Loading…
← Models
MacroConceptsPolicyModels

Macroeconometric system Model

Larger central-bank-style macroeconometric forecasting systems built around behavioral equations and policy reaction functions.

OverviewGraphMathReferences

Understand the mathematics

Follow the same equations and assumptions one section at a time or in a continuous reading.

Hover or focus a symbol defined in this model to find its exact matches. Select it to keep the highlight; Escape clears the selection.

Simultaneous Equations and the Identification Problem

A macroeconometric system is a collection of G structural equations in G endogenous variables. Each equation represents one behavioral relationship or identity. The structural form expresses each equation with contemporaneous endogenous variables on both sides.

The identification problem: given the reduced-form parameters (which can be estimated consistently by OLS on the reduced form), can we recover the structural parameters uniquely? The answer depends on the exclusion restrictions - which variables are absent from which equations.

The rank and order conditions provide necessary and sufficient conditions for identification of each structural equation.

BYt=ΓXt+ut\mathbf{B} Y_t = \mathbf{\Gamma} X_t + \mathbf{u}_tBYt​=ΓXt​+ut​
B\mathbf{B}BΓ\mathbf{\Gamma}Γ

Structural form. YtY_tYt​ is the G×1G \times 1G×1 vector of endogenous variables. XtX_tXt​ is the K×1K \times 1K×1 vector of predetermined variables (exogenous + lagged endogenous). BBB is G×GG \times GG×G (contemporaneous endogenous coefficients). Γ\GammaΓ is G×KG \times KG×K (predetermined variable coefficients). utu_tut​ is the G×1G \times 1G×1 structural error vector.

Yt=B−1ΓXt+B−1ut=ΠXt+vtY_t = \mathbf{B}^{-1} \mathbf{\Gamma} X_t + \mathbf{B}^{-1} \mathbf{u}_t = \mathbf{\Pi} X_t + \mathbf{v}_tYt​=B−1ΓXt​+B−1ut​=ΠXt​+vt​
Γ\mathbf{\Gamma}ΓΠ\mathbf{\Pi}Π

Reduced form. Π=B−1Γ\Pi = B^{-1} \GammaΠ=B−1Γ is the reduced-form coefficient matrix. vt=B−1utv_t = B^{-1} u_tvt​=B−1ut​ is the reduced-form error. Π\PiΠ can be estimated consistently by OLS equation by equation.

Step 1

βi\beta_iβi​ is the iii-th row of BBB (coefficients on endogenous variables in equation iii). γi\gamma_iγi​ is the iii-th row of Γ\GammaΓ (coefficients on predetermined variables). Some elements of βi\beta_iβi​ and γi\gamma_iγi​ are zero by exclusion restrictions.
Write the i-th structural equation: βi′Yt=γi′Xt+uit\text{Write the } i\text{-th structural equation: } \beta_i' Y_t = \gamma_i' X_t + u_{it}Write the i-th structural equation: βi′​Yt​=γi′​Xt​+uit​

Step 2

KiexcludedK_i^{\text{excluded}}Kiexcluded​ is the number of predetermined variables excluded from equation iii. GiG_iGi​ is the number of endogenous variables included in equation iii. The equation must exclude at least Gi−1G_i - 1Gi​−1 predetermined variables for identification. This is necessary but not sufficient.
Order condition (necessary): Kiexcluded≥Gi−1\text{Order condition (necessary): } K_i^{\text{excluded}} \geq G_i - 1Order condition (necessary): Kiexcluded​≥Gi​−1
KiexcludedK_i^{\text{excluded}}Kiexcluded​

Step 3

Form the submatrix of Π\PiΠ whose rows correspond to the endogenous variables included in equation iii and whose columns correspond to the predetermined variables excluded from equation iii. If this submatrix has rank Gi−1G_i - 1Gi​−1, the equation is identified. This is consistent with the order condition (Step 2): the submatrix has KiexcludedK_i^{\text{excluded}}Kiexcluded​ columns, so achieving rank Gi−1G_i - 1Gi​−1 requires Kiexcluded≥Gi−1K_i^{\text{excluded}} \geq G_i - 1Kiexcluded​≥Gi​−1. The rank condition ensures the exclusion restrictions provide enough independent information to distinguish equation iii from all linear combinations of other equations.
Rank condition (necessary and sufficient): rank(Πincl,iexcl)=Gi−1\text{Rank condition (necessary and sufficient): } \text{rank}(\mathbf{\Pi}_{\text{incl},i}^{\text{excl}}) = G_i - 1Rank condition (necessary and sufficient): rank(Πincl,iexcl​)=Gi​−1

Step 4

When the order condition holds with equality, there is exactly one set of structural parameters consistent with the reduced-form estimates. 2SLS and indirect least squares give identical results.
Exact identification: Kiexcluded=Gi−1  ⟹  unique mapping from Π to structural parameters\text{Exact identification: } K_i^{\text{excluded}} = G_i - 1 \implies \text{unique mapping from } \Pi \text{ to structural parameters}Exact identification: Kiexcluded​=Gi​−1⟹unique mapping from Π to structural parameters
KiexcludedK_i^{\text{excluded}}Kiexcluded​

Step 5

Excess exclusion restrictions imply constraints on the reduced-form matrix Π\PiΠ that can be tested. Sargan's test (overidentification test) checks whether the excluded instruments are uncorrelated with the structural residual. Rejection indicates misspecification - at least one exclusion restriction is invalid.
Over-identification: Kiexcluded>Gi−1  ⟹  testable restrictions on Π\text{Over-identification: } K_i^{\text{excluded}} > G_i - 1 \implies \text{testable restrictions on } \PiOver-identification: Kiexcluded​>Gi​−1⟹testable restrictions on Π
KiexcludedK_i^{\text{excluded}}Kiexcluded​

Step 6

Insufficient exclusion restrictions mean infinitely many structural parameter vectors are consistent with the reduced-form data. The equation cannot be estimated consistently by any method. This is not a data problem; it is a model specification deficiency.
Under-identification: Kiexcluded<Gi−1  ⟹  structural parameters not recoverable\text{Under-identification: } K_i^{\text{excluded}} < G_i - 1 \implies \text{structural parameters not recoverable}Under-identification: Kiexcluded​<Gi​−1⟹structural parameters not recoverable
KiexcludedK_i^{\text{excluded}}Kiexcluded​

Sims's critique

Sims (1980) argued that most exclusion restrictions in macroeconometric models are not credible. Why should government spending be excluded from the investment equation? Why should import prices be excluded from the wage equation? If any exclusion restriction is wrong, the structural estimates are inconsistent. This critique motivated the move to VARs, which impose no exclusion restrictions on the contemporaneous structure (at the cost of not identifying structural shocks without additional assumptions).

Notations

B\mathbf{B}B

Endogenous coefficient matrix

Γ\mathbf{\Gamma}Γ

Predetermined coefficient matrix

Π\mathbf{\Pi}Π

Reduced-form coefficient matrix

KiexcludedK_i^{\text{excluded}}Kiexcluded​

Excluded predetermined variables

About these symbols

Endogenous coefficient matrix: G x G matrix of contemporaneous endogenous variable coefficients in the structural form

Predetermined coefficient matrix: G x K matrix of predetermined variable coefficients in the structural form

Reduced-form coefficient matrix: G×KG \times KG×K reduced-form coefficient matrix: Π=B−1Γ\Pi = B^{-1} \GammaΠ=B−1Γ

Excluded predetermined variables: Number of predetermined variables excluded from structural equation i

    Transition checks

    Forecast origin
    y^T+h∣T=ET[yT+h],h=1,2,…\widehat{y}_{T+h|T}=\mathbb{E}_T[y_{T+h}],\qquad h=1,2,\ldotsy​T+h∣T​=ET​[yT+h​],h=1,2,…

    Conditioning on information at T separates the sample from the projection. The forecast can be smooth even when later realized data will not be.

    The bridge point is shared so the eye can see the handoff. After that point, the red line is a model forecast, not observed history.

    Do not score the forecast segment as realized data or read it as a policy promise.

    Uncertainty propagation
    Var⁡T(yT+h)=σ2∑j=0h−1ψj2\operatorname{Var}_T(y_{T+h})=\sigma^2\sum_{j=0}^{h-1}\psi_j^2VarT​(yT+h​)=σ2j=0∑h−1​ψj2​

    Each future innovation contributes through its impulse weight ψj\psi_jψj​. Unless the future is degenerate, longer horizons carry more conditional variance.

    The band widens when shocks accumulate through the model recursion.

    Do not treat the band edge as an alternative deterministic scenario.

    Variable timing

    Observed history

    yt∈ITy_t\in\mathcal{I}_Tyt​∈IT​

    The history segment is realized data available at the forecast origin.

    Conditional mean

    y^T+h∣T=E[yT+h∣IT]\widehat{y}_{T+h|T}=\mathbb{E}[y_{T+h}\mid\mathcal{I}_T]y​T+h∣T​=E[yT+h​∣IT​]

    The forecast path is a conditional expectation, so it can be smooth while future realizations remain volatile.

    Prediction band

    σh2=Var⁡(yT+h∣IT)\sigma_h^2=\operatorname{Var}(y_{T+h}\mid\mathcal{I}_T)σh2​=Var(yT+h​∣IT​)

    Band width is horizon-specific uncertainty, not a second deterministic transition.

    Keep going

    Read next

    Continue with a related concept, model, or example.

    Theory model

    IS-LM Model

    a well-connected theoretical model two steps from here

    New angle: theoretical model →
    Macro by Mark
    Help
    Privacy PolicyTerms of UseEthics and Compliance

    © 2026 Mark Jayson Nation LLC