A Comparison of Sequential and Information‐based Methods for Determining the Co‐integration Rank in Heteroskedastic VAR Models

In this article, we investigate the behaviour of a number of methods for estimating the co‐integration rank in VAR systems characterized by heteroskedastic innovation processes. In particular, we compare the efficacy of the most widely used information criteria, such as Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) , with the commonly used sequential approach of Johansen [Likelihood‐based Inference in Cointegrated Vector Autoregressive Models (1996)] based around the use of either asymptotic or wild bootstrap‐based likelihood ratio type tests. Complementing recent work done for the latter in Cavaliere, Rahbek and Taylor [Econometric Reviews (2014) forthcoming], we establish the asymptotic properties of the procedures based on information criteria in the presence of heteroskedasticity (conditional or unconditional) of a quite general and unknown form. The relative finite‐sample properties of the different methods are investigated by means of a Monte Carlo simulation study. For the simulation DGPs considered in the analysis, we find that the BIC‐based procedure and the bootstrap sequential test procedure deliver the best overall performance in terms of their frequency of selecting the correct co‐integration rank across different values of the co‐integration rank, sample size, stationary dynamics and models of heteroskedasticity. Of these, the wild bootstrap procedure is perhaps the more reliable overall as it avoids a significant tendency seen in the BIC‐based method to over‐estimate the co‐integration rank in relatively small sample sizes.     

Bootstrap Co‐integration Rank Testing: The Effect of Bias‐Correcting Parameter Estimates

Bootstrap‐based methods for bias‐correcting the first‐stage parameter estimates used in some recently developed bootstrap implementations of co‐integration rank tests are investigated. The procedure constructs estimates of the bias in the original parameter estimates by using the average bias in the corresponding parameter estimates taken across a large number of auxiliary bootstrap replications. A number of possible implementations of this procedure are discussed and concrete recommendations made on the basis of finite sample performance evaluated by Monte Carlo simulation methods. The results show that bootstrap‐based bias‐correction methods can significantly improve the small sample performance of the bootstrap co‐integration rank tests.     

Short run and long run causality in time series: inference

We propose methods for testing hypothesis of non-causality at various horizons, as defined in Dufour and Renault (Econometrica 66, (1998) 1099–1125). We study in detail the case of VAR models and we propose linear methods based on running vector autoregressions at different horizons. While the hypotheses considered are nonlinear, the proposed methods only require linear regression techniques as well as standard Gaussian asymptotic distributional theory. Bootstrap procedures are also considered. For the case of integrated processes, we propose extended regression methods that avoid nonstandard asymptotics. The methods are applied to a VAR model of the US economy.     

Bootstrapping cointegrating regressions

The paper investigates the usefulness of bootstrap methods for small sample inference in cointegrating regression models. It discusses the standard bootstrap, the recursive bootstrap, the moving block bootstrap and the stationary bootstrap methods. Some guidelines for bootstrap data generation and test statistics to consider are provided and some simulation evidence presented suggests that the bootstrap methods, when properly implemented, can provide significant improvement over asymptotic inference.     

BOOTSTRAPPING A STABLE AD MODEL: WEAK VS STRONG EXOGENEITY*

Through Monte Carlo experiments the effects of a feedback mechanism on the accuracy in finite samples of ordinary and bootstrap inference procedures are examined in stable first- and second-order autoregressive distributed-lag models with non-stationary weakly exogenous regressors. The Monte Carlo is designed to mimic situations that are relevant when a weakly exogenous policy variable affects (and is affected by) the outcome of agents’ behaviour. In the parameterizations we consider, it is found that small-sample problems undermine ordinary first-order asymptotic inference procedures irrespective of the presence and importance of a feedback mechanism. We examine several residual-based bootstrap procedures, each of them designed to reduce one or several specific types of bootstrap approximation error. Surprisingly, the bootstrap procedure which only incorporates the conditional model overcomes the small sample problems reasonably well. Often (but not always) better results are obtained if the bootstrap also resamples the marginal model for the policymakers’ behaviour.     

BOOTSTRAPPING THE TRACE STATISTIC IN VAR MODELS: MONTE CARLO RESULTS AND APPLICATIONS

ABSTRACT This paper investigates through Monte Carlo experiments both size and power properties of a bootstrapped trace statistic in two prototypical DGPs. The Monte Carlo results indicate that the ordinary bootstrap has similar size and power properties as inference procedures based on asymptotic critical values. Considering empirical size, the stationary bootstrap is found to provide a uniform improvement over the ordinary bootstrap if the dynamics is underspecified. The use of the stationary bootstrap as a diagnostic tool is suggested. In two illustrative examples this seems to work, and again it appears that the bootstrap incorporates the finite-sample correction required for the asymptotic critical values to apply.     

Testing log-linear models with inequality constraints: a comparison of asymptotic, bootstrap, and posterior predictive p-values

An important aspect of applied research is the assessment of the goodness-of-fit of an estimated statistical model. In the analysis of contingency tables, this usually involves determining the discrepancy between observed and estimated frequencies using the likelihood-ratio statistic. In models with inequality constraints, however, the asymptotic distribution of this statistic depends on the unknown model parameters and, as a result, there no longer exists an unique p -value. Bootstrap p -values obtained by replacing the unknown parameters by their maximum likelihood estimates may also be inaccurate, especially if many of the imposed inequality constraints are violated in the available sample. We describe the various problems associated with the use of asymptotic and bootstrap p -values and propose the use of Bayesian posterior predictive checks as a better alternative for assessing the fit of log-linear models with inequality constraints.     

Parametric bootstrap tests for continuous and discrete distributions

Statistical procedures based on the estimated empirical process are well known for testing goodness of fit to parametric distribution families. These methods usually are not distribution free, so that the asymptotic critical values of test statistics depend on unknown parameters. This difficulty may be overcome by the utilization of parametric bootstrap procedures. The aim of this paper is to prove a weak approximation theorem for the bootstrapped estimated empirical process under very general conditions, which allow both the most important continuous and discrete distribution families, along with most parameter estimation methods. The emphasis is on families of discrete distributions, and simulation results for families of negative binomial distributions are also presented.     

Bootstrap and Asymptotic Tests of Long-run Relationships in Cointegrated Systems

Hypothesis testing on cointegrating vectors based on the asymptotic distributions of the test statistics are known to suffer from severe small sample size distortion. In this paper an alternative bootstrap procedure is proposed and evaluated through a Monte Carlo experiment, finding that the Type I errors are close to the nominal signficance levels but power might be not entirely adequate. It is then shown that a combined test based on the outcomes of both the asymptotic and the bootstrap tests will have both correct size and low Type II error, therefore improving the currently available procedures.     

Two-stage stationary bootstrapping for bivariate average realized volatility matrix under market microstructure noise and asynchronicity

Under the two important modern financial market features of noise and non-synchronicity for multiple assets, for consistent estimators of the integrated covariations, we adopt the two-time scale average realized volatility matrix (ARVM) which is a matrix extension of the two-time scale realized volatilities of Zhang et al. (2005). An asymptotic normal theory is provided for the two-time scale ARVM and resulting realized covariations. The asymptotic normality is not directly applicable in practice to construct statistical methods owning to nuisance parameters. To bypass the nuisance parameter problem, two-stage stationary bootstrapping is proposed. We establish consistencies of the bootstrap distributions, and construct confidence intervals and hypothesis tests for the integrated covariance, regression coefficient and correlation coefficient. The validity of the stationary bootstrap for the high frequency heterogeneous returns is proved by showing that there exist parameters of the stationary bootstrap blocks so that the bootstrap consistencies hold. The proposed bootstrap methods extend the i.i.d. bootstrapping methods for realized covariations by Dovonon et al. (2013), that are confined to synchronous noise-free sampling. For high frequency noisy asynchronous samples, a Monte-Carlo experiment shows better finite sample performances of the proposed stationary bootstrap methods based on the two-time scale ARVM estimator than the wild blocks of blocks bootstrap methods of Hounyo (2017), based on pre-averaged truncated estimator.     

Controlling the Overall Significance Level of a Battery of Least Squares Diagnostic Tests*

Double bootstrap methods are used to control the overall significance level of a battery of diagnostic tests applied to a regression model estimated by ordinary least squares. Monte Carlo evidence on the finite sample performance of the bootstrap methods is reported and discussed.     

Bootstrapping cointegrating regressions

In this paper, we consider bootstrapping cointegrating regressions. It is shown that the method of bootstrap, if properly implemented, generally yields consistent estimators and test statistics for cointegrating regressions. For the cointegrating regression models driven by general linear processes, we employ the sieve bootstrap based on the approximated finite-order vector autoregressions for the regression errors and the first differences of the regressors. In particular, we establish the bootstrap consistency for OLS method. The bootstrap method can thus be used to correct for the finite sample bias of the OLS estimator and to approximate the asymptotic critical values of the OLS-based test statistics in general cointegrating regressions. The bootstrap OLS procedure, however, is not efficient. For the efficient estimation and hypothesis testing, we consider the procedure proposed by Saikkonen [1991. Asymptotically efficient estimation of cointegration regressions. Econometric Theory 7, 1–21] and Stock and Watson [1993. A simple estimator of cointegrating vectors in higher order integrating systems. Econometrica 61, 783–820] relying on the regression augmented with the leads and lags of differenced regressors. The bootstrap versions of their procedures are shown to be consistent, and can be used to do asymptotically valid inferences. A Monte Carlo study is conducted to investigate the finite sample performances of the proposed bootstrap methods.     

Bootstrap inference in systems of single equation error correction models

We analyze OLS-based tests of long-run relationships, weak exogeneity and short-run dynamics in conditional error correction models. Unweighted sums of single equation test statistics are used for hypothesis testing in pooled systems. When model errors are (conditionally) heteroskedastic tests of weak exogeneity and short run dynamics are affected by nuisance parameters. Similarly, on the pooled level the advocated test statistics are no longer pivotal in presence of cross-sectional error correlation. We prove that the wild bootstrap provides asymptotically valid critical values under both conditional heteroskedasticity and cross-sectional error correlation. A Monte-Carlo study reveals that in small samples the bootstrap outperforms first-order asymptotic approximations in terms of the empirical size even if the asymptotic distribution of the test statistic does not depend on nuisance parameters. Opposite to feasible GLS methods the approach does not require any estimate of cross-sectional correlation and copes with time-varying patterns of contemporaneous error correlation.     

An improved bootstrap test of stochastic dominance

We propose a new method of testing stochastic dominance that improves on existing tests based on the standard bootstrap or subsampling. The method admits prospects involving infinite as well as finite dimensional unknown parameters, so that the variables are allowed to be residuals from nonparametric and semiparametric models. The proposed bootstrap tests have asymptotic sizes that are less than or equal to the nominal level uniformly over probabilities in the null hypothesis under regularity conditions. This paper also characterizes the set of probabilities so that the asymptotic size is exactly equal to the nominal level uniformly. As our simulation results show, these characteristics of our tests lead to an improved power property in general. The improvement stems from the design of the bootstrap test whose limiting behavior mimics the discontinuity of the original test’s limiting distribution.     

Size and Power of the Error Correction Model Cointegration Test. A Bootstrap Approach

The size and power of the ECM cointegration test are investigated by using the 'bootstrap critical values'. The purpose of this paper is to show the ability of the bootstrap technique to produce critical values which are much more accurate than the asymptotic ones. The properties of the test have been studied, using Monte Carlo methods, for three different data generating processes. As regards the size of the test, we find that the ECM cointegration test together with the bootstrap critical values perform better than the ECM cointegration test based on the asymptotic critical values. While as regards the power of the tests, the results prove to be similar for the different versions.     

A bootstrap algorithm for testing cointegration rank in VAR models in the presence of stationary variables

In this paper, a bootstrap algorithm for a reduced rank vector autoregressive (VAR) model which also includes stationary regressors, is analyzed. It is shown that the bootstrap distribution for estimating the rank converges to the distribution derived from the usual asymptotic framework. Because the asymptotic distribution will typically depend on unknown parameters, bootstrap distributions are of considerable interest in this context. The result of an application and some Monte Carlo experiments are also presented.     

Bootstrapping Financial Time Series

It is well known that time series of returns are characterized by volatility clustering and excess kurtosis. Therefore, when modelling the dynamic behavior of returns, inference and prediction methods, based on independent and/or Gaussian observations may be inadequate. As bootstrap methods are not, in general, based on any particular assumption on the distribution of the data, they are well suited for the analysis of returns. This paper reviews the application of bootstrap procedures for inference and prediction of financial time series. In relation to inference, bootstrap techniques have been applied to obtain the sample distribution of statistics for testing, for example, autoregressive dynamics in the conditional mean and variance, unit roots in the mean, fractional integration in volatility and the predictive ability of technical trading rules. On the other hand, bootstrap procedures have been used to estimate the distribution of returns which is of interest, for example, for Value at Risk (VaR) models or for prediction purposes. Although the application of bootstrap techniques to the empirical analysis of financial time series is very broad, there are few analytical results on the statistical properties of these techniques when applied to heteroscedastic time series. Furthermore, there are quite a few papers where the bootstrap procedures used are not adequate.     

Monitoring risk in a ruin model perturbed by diffusion

We construct a nonparametric sequential test for the ruin probability and a corresponding change-point test in a risk model perturbed by diffusion. Some limiting properties are derived, which extend and improve on recent results of Conti (Stat Prob Lett 72:333–343, 2005) and Jahnke (Diploma thesis, University of Cologne, 2007). It is shown that the monitoring procedures can be designed such that the tests have an asymptotic prescribed false alarm rate (size) α and power 1. Some results from a small simulation study are also presented.     

Asymptotic and bootstrap prediction regions for vector autoregression

Small sample properties of asymptotic and bootstrap prediction regions for VAR models are evaluated and compared. Monte Carlo simulations reveal that the bootstrap prediction region based on the percentile-t method outperforms its asymptotic and other bootstrap alternatives in small samples. It provides the most accurate assessment of future uncertainty under both normal and non-normal innovations. The use of an asymptotic prediction region may result in a serious under-estimation of future uncertainty when the sample size is small. When the model is near non-stationary, the use of the bootstrap region based on the percentile-t method is recommended, although extreme care should be taken when it is used for medium to long-term forecasting.     

A nonparametric test for equality of distributions with mixed categorical and continuous data

In this paper we consider the problem of testing for equality of two density or two conditional density functions defined over mixed discrete and continuous variables. We smooth both the discrete and continuous variables, with the smoothing parameters chosen via least-squares cross-validation. The test statistics are shown to have (asymptotic) normal null distributions. However, we advocate the use of bootstrap methods in order to better approximate their null distribution in finite-sample settings and we provide asymptotic validity of the proposed bootstrap method. Simulations show that the proposed tests have better power than both conventional frequency-based tests and smoothing tests based on ad hoc smoothing parameter selection, while a demonstrative empirical application to the joint distribution of earnings and educational attainment underscores the utility of the proposed approach in mixed data settings.