Chow test for stability.

Carries out Chow breakpoint or Chow forecast tests for parameter constancy.

Syntax

chow(options) obs1 [obs2 obs3 ...] @ x1 x2 x3

You must provide the breakpoint observations (using dates or observation numbers) to be tested. To specify more than one breakpoint, separate the breakpoints by a space. For the Chow breakpoint test, if the equation is specified by list and contains no linear terms, you may specify a subset of the regressors to be tested for a breakpoint after an “@” sign.

Options

f | Chow forecast test. For this option, you must specify a single breakpoint to test (default performs breakpoint test). |

p | Print the result of test. |

Examples

The commands:

ls m1 c gdp cpi ar(1)

chow 1970Q1 1980Q1

perform a regression of M1 on a constant, GDP, and CPI with first order autoregressive errors, and employ a Chow breakpoint test to determine whether the parameters before the 1970’s, during the 1970’s, and after the 1970’s are “stable”.

To regress the log of SPOT on a constant, the log of P_US, and the log of P_UK, and to carry out the Chow forecast test starting from 1973, enter the commands:

ls log(spot) c log(p_us) log(p_uk)

chow(f) 1973

To test whether only the constant term and the coefficient on the log of P_US prior to and after 1970 are “stable” enter the commands:

chow 1970 @ c log(p_us)

Cross-references

See
“Chow's Breakpoint Test” for further discussion.

See
Equation::facbreak,
Equation::breaktest,
Equation::ubreak, and
Equation::rls for related equation object views.