Testing for structural breaks in GARCH models

Abstract

We study the ability of traditional diagnostic tests and LM and CUSUM structural break tests to detect a range of different types of breaks in GARCH models. We find that Wooldridge's (1990) robust LM tests for autocorrelation and ARCH have no power to detect structural breaks in GARCH models. However, CUSUM- and LM-based structural break tests have excellent size when the data is Gaussian, but the CUSUM tests tend to overreject even in quite large samples when returns have fat tails. However, the LM-based tests have approximately the correct size and exhibit impressive power to detect a range of breaks in the dynamics of conditional volatility. We apply these tests to a range of financial time series using returns starting only in 1990 and find that many GARCH models that pass standard specification tests fail the structural break tests.

Conditional coskewness and asset pricing

“…17 The test requires that we omit a fraction of the initial and final observations for the test to be well behaved, and we arbitrarily set this to be ten percent. 18 Note that Smith (2004) finds in a Monte Carlo study that the expLM test tends to overreject the null of no structural break in GARCH models in finite samples.…”

Are There Structural Breaks in Realized Volatility?

“…We test for multiple structural breaks using the change-point (or structural Examples include Chu (1995), Hillebrand (2005), Hwang and Pereira (2004), Lamoureux and Lastrapes (1990), Mikosch and Starica (2004), and Starica and Granger (2005) among others. Andreou and Ghysels (2002), Pastor and Stambaugh (2001), Rapach and Strauss (2006), Ray and Tsay (2002), and Smith (2006). For instance, Engle and Rangel (2005).…”

Price Discovery and Memory Effects in Infant African Stock Markets: Evidence from Tanzania

“…-100 0 100 1 1 /1 0 /0 8 1 /5 /0 9 3 /2 /0 9 4 /2 7 /0 9 6 /2 2 /0 9 8 /1 7 /0 9 1 0 /1 2 /0 9 1 2 /7 /0 9 2 /1 /1 0 3 /2 9 /1 0 5 /2 4 /1 0 7 /1 9 /1 0 9 /1 3 /1 0 1 1 /8 /1 0 1 /3 /1 1 2 /2 8 /1 1 4 /2 7 /1 1 CUSUM 5% Significance 1 1 /1 0 /0 8 1 /5 /0 9 3 /2 /0 9 4 /2 7 /0 9 6 /2 2 /0 9 8 /1 7 /0 9 1 0 /1 2 /0 9 1 2 /7 /0 9 2 /1 /1 0 3 /2 9 /1 0 5 /2 4 /1 0 7 /1 9 /1 0 9 /1 3 /1 0 1 1 /8 /1 0 1 /3 /1 1 2 /2 8 /1 1 4 /2 7 /1 1 CUSUM of Squares 5% Significance breakpoints on June 26 th 2008 and July 11 th 2008 for sub-sample 1, while June 17 th 2009 (CRDB listing) is tested for sub-sample 2. For the Quandt-Andrews approach we acknowledge the fact that structural shift tests tend to be sensitive to sample sizes or number of observations (see for example, Inclan & Tiao, 1994;Smith, 2008). Therefore, we do not only test the whole sample, but also each sub-sample is tested separately for any unknown breakpoints.…”