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

ARDL / bounds testing Model

Autoregressive distributed lag model for short-run dynamics and long-run relationships among variables that may have mixed integration orders.

OverviewGraphMathReferences

Graph workspace

Error-correction dynamics and long-run equilibrium (2x1 panel)

Graph workspace

The ARDL's core output is the long-run equilibrium relationship and the speed at which deviations are corrected. The top panel shows yty_tyt​ and the estimated equilibrium path βxt\beta x_tβxt​. The bottom panel shows the error-correction term ECTt\text{ECT}_tECTt​ and the adjustment path after a displacement, making the convergence speed πy\pi_yπy​ directly visible.

Chart axis metadata
xTime (periods)quarters · 0 to TyLevel (top) / Deviation from equilibrium (bottom)variable units / same units · data-dependent / centered at zero
Series and its long-run attractor static chart preview showing Dependent variable, Long-run attractor
Disequilibrium being corrected static chart preview showing Error-correction term, Equilibrium
xTime (periods)quarters · 0 to TyLevel (top) / Deviation from equilibrium (bottom)variable units / same units · data-dependent / centered at zero

Current results

Half-life of disequilibrium

Explain: Half-life of disequilibrium
h0.5=ln⁡(0.5)/ln⁡(1+πy)h_{0.5} = \ln(0.5) / \ln(1 + \pi_y)h0.5​=ln(0.5)/ln(1+πy​)

Number of periods for the ECT to halve. Short half-lives (2-4 quarters) indicate rapid equilibrium restoration; long half-lives (10+) indicate sluggish adjustment.

2.41

periods

Long-run multiplier

Explain: Long-run multiplier
β^1\hat{\beta}_1β^​1​

Equilibrium response of y to a permanent unit change in x. The central estimate of interest in most ARDL applications. Blank when the error-correction speed is zero: without cointegration there is no long-run relation to read it from.

1.50

variable units per unit of x

Mean absolute ECT

Explain: Mean absolute ECT
T−1∑∣ECTt∣T^{-1} \sum |\text{ECT}_t|T−1∑∣ECTt​∣

Average distance from equilibrium. Large values relative to y's scale indicate that the long-run relationship, while present, allows substantial short-run deviations.

0.6750

variable units

Error-correction speed

Explain: Error-correction speed
π^y\hat{\pi}_yπ^y​

Direct reading of the adjustment coefficient. Values between -0.5 and -0.1 are typical in macro data. Values near zero mean very slow correction, and zero means no cointegration.

-0.2500

per period

Controls

Outcome moves more
Explain: Long-run effect of the input

Long-run coefficient: 1.5

The lasting effect on the outcome of a permanent one-unit rise in the input.

Changes the slope of the dashed red equilibrium line in the top panel.

Typical adjustment
Explain: How fast the gap closes

Error-correction speed: -0.25

The share of the gap from equilibrium closed each period; more negative means faster.

Controls how tightly yty_tyt​ hugs the equilibrium in the top panel and how fast the ECT decays in the bottom panel.

Changes partly continue
Explain: Momentum in recent changes

Short-run AR(1) on Δy: 0.2

Whether recent changes tend to continue when positive, or reverse when negative.

Higher values make yty_tyt​'s approach to equilibrium more sluggish, adding inertia to the convergence.

0.4
Explain: Size of random shocks

How big the random disturbances hitting the relationship are in each period.

Higher values increase the amplitude of fluctuations around equilibrium in both panels.

200
Explain: How many periods to simulate

The number of time periods the simulation generates.

Longer samples show more mean-reversion cycles and make the long-run relationship more visible.

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