Structural Break Test (Chow Test)

The Chow test is an F test that checks whether the coefficients of a regression model differ before and after a known break point, indicating a structural break.

Chow 검정은 사전에 알려진 시점 전후로 회귀모형의 계수가 달라졌는지, 즉 구조변화가 있는지를 검정하는 F 검정이다.

How do I test whether a relationship changed after a specific date or event?

특정 시점이나 사건 이후 관계가 달라졌는지 어떻게 검정하나요?

The Chow test compares the residual sum of squares from one pooled regression with that of separate regressions estimated before and after a pre-specified break point. It fits time series or grouped data when the date of a possible structural change, such as a policy or crisis, is known in advance.

When to use Structural Break Test (Chow Test)

• A policy change, crisis, or reform occurred at a known date
• Testing whether trend or regression coefficients differ between two periods
• Comparing regression coefficients between two groups (the same F test)

When to use something else

• The break date is unknown: use the Quandt-Andrews (sup-F) or Bai-Perron multiple break tests
• The goal is the causal effect of an intervention on a single series: use interrupted time series or regression discontinuity in time

Assumptions and requirements

• Break point specified in advance, not chosen from the data
• Equal error variances in both subperiods
• Independent, normally distributed errors
• Enough observations in each subperiod to estimate the model

What the analysis reports

• F statistic with degrees of freedom and p-value
• Coefficients estimated for each subperiod
• Residual sums of squares for pooled and split models
• Plot of the series with the break point marked

How to report it

A Chow test indicated a structural break at [date], F([df1], [df2]) = [value], p = [value].

Related methods

• Bai-Perron Multiple Break Test
• Interrupted Time Series Analysis
• CUSUM Test

Run Structural Break Test (Chow Test) on Quorant