Abstract:In a variety of different settings cumulative sum (CUSUM) procedures have been applied for the sequential detection of structural breaks in the parameters of stochastic models. Yet their performance depends strongly on the time of change and is best under early-change scenarios. For later changes their finite sample behavior is rather questionable. We therefore propose modified CUSUM procedures for the detection of abrupt changes in the regression parameter of multiple time series regression models, that show a higher stability with respect to the time of change than ordinary CUSUM procedures. The asymptotic distributions of the test statistics and the consistency of the procedures are provided. In a simulation study it is shown that the proposed procedures behave well in finite samples. Finally the procedures are applied to a set of capital asset pricing data related to the Fama-French extension of the capital asset pricing model.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: in the time - series regression model, how to detect the time points of parameter structural mutations more stably, especially when the mutations occur relatively late. The traditional Cumulative Sum (CUSUM) method performs well for early changes, but is slow to respond to and ineffective for late changes. For this reason, the author proposes an improved CUSUM procedure, aiming to improve the stability of the mutation time and exhibit better finite - sample behavior in various situations.
### Problem Background
The linear regression model is one of the most widely used stochastic models and is widely applied in scientific research and practical problems in various disciplines. However, external shocks (such as financial crises) may cause changes in the structure of the model, resulting in model mismatches. Rapid detection of these changes is crucial to avoid serious economic consequences. Therefore, methods that can quickly detect structural changes need to be developed.
### Research Motivation
The existing Cumulative Sum (CUSUM) method performs well for early changes but has poor performance when the changes occur late. To overcome this limitation, this paper proposes an improved CUSUM method based on the idea of Page (1954) to improve the stability of the change time. In addition, this method also takes into account the practical application requirements in fields such as finance and allows for a certain degree of dependence (for example, the time - series models of error terms and regression variables).
### Solution
The author proposes two improved CUSUM detectors:
1. **Page CUSUM detector based on ordinary residuals** \( \hat{Q}_P(m,k) \)
2. **Page CUSUM detector based on squared residuals** \( \hat{S}_P(m,k) \)
These detectors are implemented by defining the stopping time \( \tau_{\text{Page}}^{\alpha, \gamma}(m) \), which triggers an alarm when the detector exceeds a certain boundary function. The form of the boundary function is:
\[ h_{\alpha, \gamma}(m, k) = c m^{1/2} \left(1 + \frac{k}{m}\right) \left(\frac{k}{k+m}\right)^\gamma \]
where \( 0 \leq \gamma < 1/2 \), and \( c = c(\alpha, \gamma) \) is a constant that satisfies the given significance level \( \alpha \) under the null hypothesis.
### Main Results
- **Asymptotic Distribution**: Under the null hypothesis, the asymptotic distribution of the detector is related to the standard Wiener process.
- **Consistency**: Under the alternative hypothesis, the detector will diverge, indicating that it can effectively detect structural changes.
- **Simulation Study**: A large number of simulation experiments have verified the good performance of the proposed method in finite samples and demonstrated its advantages over the traditional CUSUM method.
### Application
Finally, the author applies the proposed detection method to the data set of the Fama - French capital asset pricing model, further verifying its effectiveness in practical applications.
In summary, this paper aims to improve the detection ability of structural changes in time - series regression models through the improved CUSUM method, especially for late - change situations. This helps to detect model mismatches earlier, thereby reducing potential economic losses and other adverse effects.