Saltar al contenido principal

Cookies en Macro by Mark.Usamos cookies esenciales para inicio de sesión, idioma y seguridad. Con tu permiso, también usamos cookies opcionales de preferencias, para analítica y para medición publicitaria con el píxel de Meta (Facebook). Lee nuestra política de cookies.

Macro by MarkMacro by Mark
  • Inicio
Cargando…
Cargando…
← Modelos
MacroConceptosPolíticaModelos

Bridge equations Model

Simple regressions that bridge from monthly indicators to a quarterly target like GDP, refreshed as each indicator lands.

ResumenGráficoMatemáticasReferencias

Graph workspace

Nowcast evolution within the quarter

Graph workspace

The bridge equation's distinguishing feature is how the nowcast updates as new monthly data arrives within the quarter. Showing the sequence of nowcasts from the beginning to the end of the quarter - each update triggered by a new indicator release - captures the model's real-time behavior.

Chart axis metadata
xInformation vintage (weeks into quarter)weeks · 0-14yGDP nowcast (q/q %)percentage points · -2 to 4
Nowcast as each month of the quarter
arrives static chart preview showing Nowcast after each release, Nowcast with the full quarter, 90% band, upper, 90% band, lower
Hard indicator's three monthly
readings, observed or not static chart preview showing Series 1
xInformation vintage (weeks into quarter)weeks · 0-14yGDP nowcast (q/q %)percentage points · -2 to 4

Current results

Current nowcast

Explain: Current nowcast
y^Q(v)=c+β1x^1,Q(v)+β2x^2,Q(v)\hat{y}_Q(v) = c + \beta_1 \hat{x}_{1,Q}(v) + \beta_2 \hat{x}_{2,Q}(v)y^​Q​(v)=c+β1​x^1,Q​(v)+β2​x^2,Q​(v)

The GDP growth estimate given information available at the current vintage. Updates each time a new monthly indicator arrives.

0.42%

percent q/q

Information gain

Explain: Information gain
RMSE(v=0)−RMSE(v=current)\text{RMSE}(v=0) - \text{RMSE}(v=\text{current})RMSE(v=0)−RMSE(v=current)

How much the forecast error has been reduced by the monthly data received so far. The RMSE is the bridge residual σy\sigma_yσy​ plus the error the fill-in makes on each month not yet observed (an AR forecast for the hard indicator, the mean for the survey), so the gain never falls as data arrive. Larger gain means the indicators are informative.

0.29 pp

percentage points RMSE reduction

Months observed / total

Explain: Months observed / total
count of observed monthly values2×3\frac{\text{count of observed monthly values}}{2 \times 3}2×3count of observed monthly values​

Fraction of the monthly indicator data for the current quarter that is actually observed (not AR-filled).

1.00

ratio (0 to 1)

Nowcast revision from last update

Explain: Nowcast revision from last update
∣y^Q(v)−y^Q(v−1)∣|\hat{y}_Q(v) - \hat{y}_Q(v-1)|∣y^​Q​(v)−y^​Q​(v−1)∣

How much the nowcast changed from the previous data release. At vintage 0 it is the move from the intercept-only nowcast (no data) to the survey's first reading. Large revisions indicate the new data was surprising.

0.20 pp

percentage points

Controls

moderate influence
Explain: Weight on indicator 1

Bridge coeff. (indicator 1): 0.5

How much the nowcast moves when indicator 1's quarterly average changes by one unit.

Larger β1\beta_1β1​: nowcast moves more when indicator 1 data arrives. The trajectory has bigger jumps at indicator 1 release dates.

moderate influence
Explain: Weight on indicator 2

Bridge coeff. (indicator 2): 0.3

How much the nowcast moves when indicator 2 changes by one unit.

Larger β2\beta_2β2​: nowcast is more sensitive to indicator 2 releases.

partly follows last month
Explain: How missing months get filled

AR coeff. (indicator 1 fill-in): 0.6

How closely the stand-in for a missing month follows the previous month's reading.

Higher ϕ1\phi_1ϕ1​: the AR fill-in carries the last observed month further into the missing ones, so the early nowcast leans harder on it. That helps when the indicator persists and hurts when it turns: in this quarter the hard indicator's first month (-0.85) is followed by a rise, so a higher ϕ1\phi_1ϕ1​ leaves the month-1 nowcast further from the full-quarter one (about 0.40 below it at ϕ1=0.95\phi_1 = 0.95ϕ1​=0.95, 0.31 at ϕ1=0\phi_1 = 0ϕ1​=0).

moderate uncertainty band
Explain: How much GDP stays unexplained

Bridge regression std error: 0.25

The size of the GDP swing the indicators cannot explain.

Wider σy\sigma_yσy​: the gray confidence band expands. Even with all monthly data observed, substantial uncertainty remains.

modest growth baseline
Explain: Baseline growth rate

Intercept (trend growth): 0.4

The nowcast's starting point; roughly trend GDP growth when indicators sit at their averages.

Shifts the entire nowcast trajectory up or down.

3
Explain: Months of hard data observed

How much of the quarter has actually been published. The survey indicator leads the hard indicator by about a month, so at v the survey has v+1 months and the hard series has v - the ragged edge a bridge model exists to handle.

Moves the nowcast along its own release path: earlier vintages lean on the AR fill-in and carry a wider band, later ones on observed data.

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
Ayuda
Política de privacidadTérminos de usoÉtica y cumplimiento

© 2026 Mark Jayson Nation LLC