We propose a new test against change in correlation at an unknown point time based on cumulated sums of empirical correlations. The does not require that inputs are independent and identically distributed under the null. derive its limiting null distribution using functional delta method argument, provide formula for local power particular types structural changes, give some Monte Carlo evidence finite-sample behavior, apply it to recent stock returns.

Summary This paper studies the asymptotic properties of endogeneity corrections based on nonlinear transformations without external instruments, which were originally proposed by Park and Gupta (2012) have become popular in applied research. In contrast to original copula-based estimator, our approach is a nonparametric control function does not require conformably specified copula. Moreover, we allow for exogenous regressors, may be (linearly) correlated with endogenous regressor(s). We...

Summary Most of the literature on change point analysis by means hypothesis testing considers hypotheses form H0:θ1=θ2 versus H1:θ1≠θ2, where θ1 and θ2 denote parameters process before after a point. The paper takes different perspective investigates null no relevant changes, i.e. H0:‖θ1−θ2‖⩽Δ, ‖·‖ is an appropriate norm. This formulation problem motivated fact that in many applications modification statistical might not be necessary, if difference between small. A general approach to...

We propose a specification test for wide range of parametric models the conditional distribution function an outcome variable given vector covariates. The is based on Cramer–von Mises distance between unrestricted estimate joint data and restricted that imposes structure implied by model. procedure straightforward to implement, consistent against fixed alternatives, has nontrivial power local deviations order n − 1/2 from null hypothesis, does not require choice smoothing parameters. In...

For a bivariate time series $((X_i,Y_i))_{i=1,...,n}$ we want to detect whether the correlation between $X_i$ and $Y_i$ stays constant for all $i = 1,...,n$. We propose nonparametric change-point test statistic based on Kendall's tau derive its asymptotic distribution under null hypothesis of no change by means new U-statistic invariance principle dependent processes. The depends long run variance tau, which an estimator show consistency. Furthermore, assuming single change-point, that...

We propose a nonparametric procedure to test for changes in correlation matrices at an unknown point time. The new requires constant expectations and variances, but only mild assumptions on the serial dependence structure, has considerable power finite samples. derive asymptotic distribution under null hypothesis of no change as well local results apply stock returns.

The article suggests a CUSUM‐type test for time‐varying parameters in recently proposed spatial autoregressive model stock returns and derives its asymptotic null distribution as well local power properties. As can be seen from Euro Stoxx 50 returns, combination of modelling change point tests might allow superior risk forecasts portfolio management.