The BackShift Operator


The backshift operator $B$ is used to rewrite auregressive models as factorable polynomials. The backshift operator shifts a time series back one time step.

$$ B X_t = X_{t-1} $$

When the backshift operator shifts more or less than one time step, this is indicated with a exponent. $B^n$ represents the n$th$ backshift. Thus, more generally

$$ B^k X_t = X_{t-k} $$

For a single step back in time $B^1$, the exponent it typically omitted.

Usage Examples


The following are real application of using backshift notation.

The AR(1) Model

The zero-meam AR(1) is expressed as

$$ X_t - \phi X_{t-1} = a_t $$

This can be rewritten in backshift notation as

$$ (1 - \phi B) X_{t} = a_t $$

The 1st Order Difference Filter

The first order difference filter is expressed as

$$ X_t = Z_t - Z_{t-1} $$

This can be rewritten in backshift notation as

$$ X_t = Z_t (1 - B) $$

The Moving Average Filter

The 3 point moving average filter can be expressed as

$$ X_t = \frac{Z_{t+1} + Z_t + Z_{t-1}}{3} $$

This can be rewritten in backshift notation as

$$ X_t = Z_t \frac{B^{-1} + B^0 + B^1}{3} $$

References


  • [1] W. Woodward and B. Salder, "Autoregressive Models and Filtering", SMU, 2019