Abstract:We consider bootstrap-based testing for threshold effects in non-linear threshold autoregressive (TAR) models. It is well-known that classic tests based on asymptotic theory tend to be oversized in the case of small, or even moderate sample sizes, or when the estimated parameters indicate non-stationarity, as often witnessed in the analysis of financial or climate data. To address the issue we propose a supremum Lagrange Multiplier test statistic (sLMb), where the null hypothesis specifies a linear autoregressive (AR) model against the alternative of a TAR model. We consider a recursive bootstrap applied to the sLMb statistic and establish its validity. This result is new, and requires the proof of non-standard results for bootstrap analysis in time series models; this includes a uniform bootstrap law of large numbers and a bootstrap functional central limit theorem. These new results can also be used as a general theoretical framework that can be adapted to other situations, such as regime-switching processes with exogenous threshold variables, or testing for structural breaks. The Monte Carlo evidence shows that the bootstrap test has correct empirical size even for small samples, and also no loss of empirical power when compared to the asymptotic test. Moreover, its performance is not affected if the order of the autoregression is estimated based on information criteria. Finally, we analyse a panel of short time series to assess the effect of warming on population dynamics.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to use the bootstrap - based test method to detect the effectiveness of the threshold effect in the Threshold Autoregressive (TAR) model. Specifically, the paper focuses on the fact that when the sample size is small or the model parameters are close to the non - stationary state, the traditional asymptotic theory test methods tend to produce an overly large test scale (that is, the test results tend to be too conservative or too aggressive). To solve this problem, the author proposes a supremum Lagrange multiplier test statistic (sLMb) based on the recursive bootstrap method and theoretically proves its effectiveness. In addition, the author also verifies the performance of this test method in small samples through Monte Carlo simulation. The results show that the bootstrap test has the correct empirical scale and no loss of empirical efficiency, even when the autoregressive order is unknown.
### Main contributions of the paper
1. **Proposing a new test method**: The author proposes an sLMb test statistic based on the recursive bootstrap method to detect the differences between the linear autoregressive (AR) model and the TAR model.
2. **Theoretical proof**: The author proves that under the null hypothesis, the test statistic of the bootstrap distribution is consistent with the asymptotic distribution of the likelihood ratio test proposed by Chan (1990), that is, a function of a central Gaussian process.
3. **New theoretical framework**: To prove the effectiveness of the bootstrap method, the author derives the non - standard bootstrap law of large numbers and the bootstrap functional central limit theorem. These results can be used as a general theoretical framework for dealing with other situations, such as the regime - switching process of exogenous threshold variables or the structural break test.
4. **Empirical analysis**: The author verifies the performance of the bootstrap test in small samples through Monte Carlo simulation and applies it to actual data to analyze the impact of temperature changes on the population dynamics of moth larvae.
### Key formulas
- **TAR model**:
\[
X_t=\phi_0+\sum_{i = 1}^p\phi_iX_{t - i}+\left(\Psi_0+\sum_{i = 1}^p\Psi_iX_{t - i}\right)I(X_{t - d}\leq r)+\epsilon_t
\]
where \(\phi\) and \(\Psi\) are parameter vectors, \(I(\cdot)\) is an indicator function, \(r\) is a threshold parameter, and \(\epsilon_t\) is an independently and identically distributed error term.
- **sLMb test statistic**:
\[
T_n=\sup_{r\in[r_L,r_U]}T_n(r)
\]
where
\[
T_n(r)=\left(\frac{\partial\tilde{\ell}_n}{\partial\Psi}(r)\right)^T\left(I_{n,22}(r)-I_{n,21}(r)I_{n,11}^{-1}I_{n,12}(r)\right)^{-1}\frac{\partial\tilde{\ell}_n}{\partial\Psi}(r)
\]
- **Bootstrap information matrix**:
\[
I_n^*(r)=\begin{pmatrix}
I_{n,11}^*&I_{n,12}^*(r)\\
I_{n,21}^*(r)&I_{n,22}^*(r)
\end{pmatrix}=\frac{1}{\sigma^{*2}}\sum_{t = 1}^nD_t^*(r)D_t^{*T}(r)
\]
### Application example
Finally, the paper applies the sLMb test method to analyze 12 short - time - series data, which reflect the quantity changes of moth larvae populations under different temperature conditions. The study finds that temperature changes have a significant threshold effect on population dynamics, and an appropriate model can explain the asymmetric cycle characteristics of 6 weeks.
Overall, through theoretical and empirical analysis, this paper...