An autoregressive process is a time-series process in which the current value depends on one or more past values of the same series.
$$$$
A simple example
An AR(1) process is:
$$ y_t = c + \phi y_{t-1} + \varepsilon_t $$
If (|\phi|<1), shocks fade over time and the process is stationary. If (\phi) is close to 1, the series is highly persistent.
Why economists use it
Many economic variables show momentum. Inflation, interest rates, and output often depend partly on their recent history. Autoregressive models provide a simple way to capture that persistence.
Interpretation
The coefficient on the lagged value tells economists how strongly the past carries into the present. A larger coefficient implies slower adjustment after shocks.
Related Terms
- Autocorrelation
- Autoregressive Moving Average (ARMA (p, q)) Model
- Stationary Process
- Time Series Data
Knowledge Check
### In an autoregressive process, current values depend on:
- [x] past values of the same series
- [ ] only current tax rules
- [ ] a completely unrelated variable only
- [ ] no previous information
> **Explanation:** The defining feature is dependence on the series' own lags.
### In an AR(1) model, a larger \(\phi\) typically means:
- [x] more persistence in the series
- [ ] less relation to the past
- [ ] lower variance by definition
- [ ] no role for shocks
> **Explanation:** A stronger lag coefficient means shocks fade more slowly.
### Why are autoregressive models common in economics?
- [x] Because many macro and financial variables evolve gradually over time
- [ ] Because all economic data is independent over time
- [ ] Because only cross-sectional models matter
- [ ] Because they remove the need for inference
> **Explanation:** Persistence is a widespread feature of economic time series.