Error Correction Model
A time series model that relates the short-term dynamics of variables to their long-term equilibrium.
What is Error Correction Model?
An Error Correction Model (ECM) is a general framework (of which VECM is the multivariate version) where the change in a variable is modeled as a function of past changes and the 'error' or deviation from a long-run equilibrium in the previous period. The model forces the variables to gradually correct past deviations.
Why Error Correction Model Matters
It resolves the problem of spurious regression with non-stationary variables by explicitly modeling the cointegrating relationship, providing parameters for both the speed of adjustment and the long-run effects.
Example
A researcher studies consumption and income. The ECM shows that if consumption is unexpectedly low relative to income in one year, it will adjust upward in the following year to restore their long-term ratio.
Common Mistakes
- Failing to verify that the error term from the long-run equation is actually stationary before building the ECM.
- Omitting relevant short-run dynamic terms, leading to autocorrelated residuals.