Weighted Smooth Transition Regressions

Abstract

SUMMARYA new procedure is proposed for modelling nonlinearity of a smooth transition form, by allowing the transition variable to be a weighted function of lagged observations. This function depends on two unknown parameters and requires specification of the maximum lag only. Nonlinearity testing for this specification uses a search over a plausible set of weight function parameters, combined with bootstrap inference. Finite-sample results show that the recommended wild bootstrap heteroskedasticity-robust testing procedure performs well, for both homoskedastic and heteroskedastic data-generating processes. Forecast comparisons relative to linear models and other nonlinear specifications of the smooth transition form confirm that the new WSTR model delivers good performance.

The Role of a Changing Market Environment for Credit Default Swap Pricing

“…Times of positive macroeconomic news are characterized by greater fractions of chartists while fundamentalists dominate during downturns. Related approaches with multivariate transition functions can be found in Medeiros and Veiga () and Becker and Osborn () where transition is driven by different lags of the endogenous variable. Multivariate transition functions for threshold models are considered by Massacci ().…”

Spatial agglomeration and productivity in Italy: A panel smooth transition regression approach

“…In this literature, Hurn and Becker () and Becker and Osborn () have recently dealt with the problem of heteroscedasticity and the distortions the latter causes to the size of the test in small samples. Dealing with heteroscedasticity in non‐linearity tests can be problematic: on the one hand, neglecting heteroscedasticity may lead to reject the null of linearity when it is not the case; on the other hand, robustification can remove most of the test power (Lundbergh and Teräsvirta ).…”

International R&D Spillovers, Absorptive Capacity and Relative Backwardness: A Panel Smooth Transition Regression Model

“…They showed that AveLM, ExpLM or wLM are always more powerful than the SupLM and the Taylor expansion-based tests. Hurn and Becker (2009) and Becker and Osborn (2010), instead, dealt with the problem of heteroskedasticity and the distortions the latter causes to the size of the test in small samples. Indeed, dealing with heteroskedasticity in nonlinearity tests can be problematic.…”