Johan Lyhagen

Likelihood-Based Cointegration Tests in Heterogeneous Panels

This paper presents a likelihood-based panel test of cointegrating rank in heterogeneous panel models based on the mean of the individual rank trace statistics. The existence of the first two moments of the asymptotic distribution of the individual trace statistic is established. Based on this, the asymptotic distribution of the test statistic is shown to be normal. The small-sample size and power properties are investigated using Monte Carlo simulations. An empirical example for a consumption model including consumption, income and inflation is estimated for 23 OECD countries over the period 1960-1994. The results indicate that two cointegrating relations exist in the system: One containing consumption and income and one inflation only.

Inflation, exchange rates and PPP in a multivariate panel cointegration model

New multivariate panel cointegration methods are used to analyze nominal exchange rates and prices in four major economies in Europe: France, Germany, Italy and the United Kingdom for the post-Bretton Woods period. We test for purchasing power parity (PPP) between these four countries and find that the theoretical PPP relationship does not hold. However, the estimated unrestricted relationship is found to be remarkably close to the theoretical one (1, −1.5, 0.9 instead of 1, −1,1). Relevant asymptotic results are stated, proved, and evaluated using Monte Carlo simulations. The asymptotic results are general and may hence be used in similar empirical contexts using the same model structure. Parametric bootstrap inference is used in order to deal with test size distortions. Copyright Royal Economic Society 2008

Inference in Panel Cointegration Models With Long Panels

This article presents a general likelihood-based framework for inference in panel vector autoregressive (VAR) models with cointegration restrictions. The cointegrating relationships are restricted to each cross section while the rest of the model is unrestricted. The homogeneous restriction of common cointegrating space is also considered. Asymptotic distributions of parameter estimators and the test statistics for the cointegrating rank and the homogeneous restriction are derived. The asymptotic distribution for the cointegrating rank is shown to be the convolution of the standard distribution of the trace statistic and the χ2 distribution. The homogeneous restriction test statistic is asymptotically χ2. A Monte Carlo simulation investigates the small-sample properties of the two tests. The empirical size of the test for the cointegrating rank is well above the nominal. A Bartlett-corrected test statistic is shown to have size very close to the nominal. We give an empirical example for a consumption model, including consumption, income, and inflation as well as considering the monetary exchange rate model of Groen and Kleibergen.

A simple linear time series model with misleading nonlinear properties

This paper demonstrates that long memory leads to spurious rejection of the linearity hypothesis, when a STAR specification constitutes the alternative.

Forecasting performance of seasonal cointegration models

Forecasts from seasonal cointegration models are compared with those from a standard cointegration model based on first differences and seasonal dummies. The effects of restricting or not restricting seasonal intercepts in the seasonal cointegration models are examined as well as the recently proposed specification and estimation procedure for the annual frequency by Johansen and Schaumburg (1999). The data generating process used in the Monte Carlo simulation is based on an empirical six-dimensional macroeconomic data set. Results show that the seasonal cointegration model improves forecasting accuracy, compared with the standard cointegration model, even in small samples and if short forecast horizons are considered. Furthermore, the specification suggested by Johansen and Schaumburg seems to work better than the original model presented by Lee (1992). An empirical forecasting example confirm most of the results found in the Monte Carlo study.

Testing for Purchasing Power Parity in Cointegrated Panels

This paper applies the maximum likelihood panel cointegration method of Larsson and Lyhagen (2007) to test the strong PPP hypothesis using data for the G7 countries. This method is robust in several important dimensions relative to previous methods, including the well-known issue of cross-sectional dependence of error terms. The findings using this new method are contrasted to those from the Pedroni (1995) cointegration tests and fully modified OLS and dynamic OLS esimators of the cointegrating vectors. Our overall results are the same across all approaches: The strong PPP hypothesis is rejected in favour of weak PPP with heterogenenous cointegrating vectors.

Development of health-related quality of life and symptoms of anxiety and depression among persons diagnosed with cancer during adolescence: a 10-year follow-up study

The main aim was to investigate the development of health‐related quality of life (HRQOL) and symptoms of anxiety and depression in a cohort diagnosed with cancer during adolescence from shortly after up to 10 years after diagnosis.

Estimating Nonlinear Structural Models: EMM and the Kenny–Judd Model

The estimation of nonlinear structural models is not trivial. One reason for this is that a closed form solution of the likelihood may not be feasible or does not exist. We propose to estimate nonlinear structural models using the efficient method of moments, as generating data according to the models is often very easy. A simulation study of the interaction model of Kenny–Judd shows promising results, for example, the bias of the parameter for the interaction effect is less for the efficient method of moments compared to quasi-maximum likelihood (QML) and characteristic function estimator (CFE) (Blom & Christoffersson 2001) and comparable to latent moderated structural equations (LMS) (Schermelleh-Engel, Klein, & Moosbrugger, 1998).

A Method to Generate Multivariate Data with the Desired Moments

We show how it is possible to generate multivariate data which has moments arbitrary close to the desired ones. They are generated as linear combinations of variables with known theoretical moments. It is shown how to derive the weights of the linear combinations in both the univariate and the multivariate setting. The use in bootstrapping is discussed and the method is exemplified with a Monte Carlo simulation where the importance of the ability of generating data with control of higher moments is shown.

Small-sample properties of some tests for unit root with data-based choice of the degree of augmentation

Abstract In the augmented Dickey–Fuller (ADF) regression one usually decides on the level of the “augmentation” prior to the performing of unit root test. This is a purely data-dependent method that uses either some information criteria or some sequential test of significance on parameter estimates. Contrary to earlier beliefs, our analyses reveal that the presence and/or absence of a drift and a time trend in the data generating process has a remarkable effect on the behaviour of the subsequent tests for unit root.