Overview (ACF command)

ACF displays and plots the sample autocorrelation function of one or more time series. You can also display and plot the autocorrelations of transformed series by requesting natural log and differencing transformations within the procedure.

Options

Modifying the Series. You can request a natural log transformation of the series using the LN subcommand and seasonal and nonseasonal differencing to any degree using the SDIFF and DIFF subcommands. With seasonal differencing, you can specify the periodicity on the PERIOD subcommand.

Statistical Output. With the MXAUTO subcommand, you can specify the number of lags for which you want autocorrelations to be displayed and plotted, overriding the maximum specified on TSET. You can also display and plot values at periodic lags only using the SEASONAL subcommand. In addition to autocorrelations, you can display and plot partial autocorrelations using the PACF subcommand.

Method of Calculating Standard Errors. You can specify one of two methods of calculating the standard errors for the autocorrelations on the SERROR subcommand.

Basic Specification

The basic specification is one or more series names.

  • For each series specified, ACF automatically displays the autocorrelation value, standard error, Box-Ljung statistic, and probability for each lag.
  • ACF plots the autocorrelations and marks the bounds of two standard errors on the plot. By default, ACF displays and plots autocorrelations for up to 16 lags or the number of lags specified on TSET.
  • If a method has not been specified on TSET, the default method of calculating the standard error (IND) assumes that the process is white noise.

Subcommand Order

  • Subcommands can be specified in any order.

Syntax Rules

  • VARIABLES can be specified only once.
  • Other subcommands can be specified more than once, but only the last specification of each one is executed.

Operations

  • Subcommand specifications apply to all series named on the ACF command.
  • If the LN subcommand is specified, any differencing requested on that ACF command is done on the log-transformed series.
  • Confidence limits are displayed in the plot, marking the bounds of two standard errors at each lag.

Limitations

  • A maximum of one VARIABLES subcommand. There is no limit on the number of series named on the list.