Nonsense regressions due to neglected time-varying means

Hassler, Uwe

doi:10.1007/s00362-003-0144-0

Cited by 20 publications

(13 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…For example, if X and y both are affected by the structural break, then the predictive coefficient on X could be biased as Hassler (2003) and Elliott (2005) show. However, suppose an econometrician ignores the nonstationary properties of y t and estimates a misspecified model that does not include the structural break.…”

Section: Shifting Means and Predictive Regressionsmentioning

confidence: 99%

Can Managers Successfully Time the Maturity Structure of their Debt Issues?

2004

View full text Add to dashboard Cite

Section: Shifting Means and Predictive Regressionsmentioning

confidence: 99%

Can Managers Successfully Time the Maturity Structure of their Debt Issues?

2004

View full text Add to dashboard Cite

“…Since then, much research on nonsense regressions has been carried out mainly in the field of econometrics; see, for example, Choi (1994), Haldrup (1994), Hassler (1996a, b), Marmol (1996, 1998) and Cappuccio and Lubian (1997). All these researchers assumed non‐stationary, trending time series models, whereas Hassler (2003) shows that nonsensical regressions may also arise because of simple mean shifts in time series without trends. It is well known from everyday practice that nonsensical correlation may also arise in cross‐section analyses owing to hidden variables that introduce heterogeneity into the sample.…”

Section: Introductionmentioning

confidence: 99%

Nonsensical and biased correlation due to pooling heterogeneous samples

Hassler

Thadewald

2003

J Royal Statistical Soc D

Self Cite

View full text Add to dashboard Cite

The case of two variables is considered, where the sample consists of two heterogeneous groups. The behaviour of the pooled sample correlation coefficient is studied. The heterogeneity of the two groups may be interpreted as a hidden qualitative variable. It is shown that, even if the correlation is the same within both groups, the pooled correlation coefficient may be severely biased owing to heterogeneity of other group-specific parameters. In the case of uncorrelatedness, nonsensical correlation may arise from pooled estimation. These and further results are obtained and can be quantified or forecast from an asymptotic formula for the pooled sample correlation coefficient, which is well reproduced in finite sample computer experiments and illustrated with empirical examples.

show abstract

“…cause and effect, the so-called spurious correlation or either nonsense correlation mentioned respectively by Granger and Newbold (2001), and Hassler (2003). There are two more features of this work that we believe justifies it, and brings contemporaneity to its application: they are the nonlinear treatment of the variables, and the use of 10 minutes intraday data.…”

Section: Linear and Nonlinear Association Measures With Intraday/highmentioning

confidence: 74%

Linear and Nonlinear Association Measures With Intraday/High Frequency Data for All Ibovespa Stocks

Block¹,

Ceretta²,

Costa³

2016

R. eletr. estrat. neg.

View full text Add to dashboard Cite

LINEAR AND NONLINEAR ASSOCIATION MEASURES WITH INTRADAY/HIGH FREQUENCY DATA FOR ALL IBOVESPA STOCKSAlexandre Costa -Alexander Souza Block -Paulo Sérgio Ceretta ABSTRACTWe analyze four types of association measurements, the Pearson's Correlation, the Coefficient of Determination, the Multidimensional Association and the Maximal Information Coefficient, the two first linear, and the other two, nonlinear. We utilize 10 minutes intraday data of all Ibovespa stocks, that account for the main stock market index in Brazil. Not much of the methods used in this paper have been seen in financial literature, hence the motivation for this study. The methodology is meaningful to traders, as some stocks are highly correlated to the main index, one can base a strategy when they are trading independently in a given day as this pattern should revert to the mean. One of the most important findings of this work is that treating data as nonlinear yielded stronger results.Keywords: Multidimensional Association; Maximal Information Coefficient; Coefficient of Determination; Ibovespa; Intraday. RESUMENAnalizamos cuatro tipos de medidas de asociación, la Correlación de Pearson, el Coeficiente de Determinación, la Asociación Multidimensional y el Coeficiente Máximo de Información, los dos primeros, lineal, y los otros dos, no lineal. Utilizamos 10 minutos de datos dentro de un día, de todas las acciones Ibovespa, que cuenta para el principal índice de acciones en el mercado de Brasil. En la literatura financiera no hay mucho sobre los métodos utilizados en este trabajo, de dónde viene la motivación para el estudio. La metodología es significativa para negociantes, así como algunas acciones son altamente correlacionadas con el principal índice, un puede ser la base estratégica cuando están operando independientemente en un dado día, como este estándar podría reverter la significación. Uno de los más importantes resultados de este trabajo es que el tratamiento de los dados como no lineal produjeron resultados más fuertes. Palabras-clave:

show abstract

Nonsense regressions due to neglected time-varying means

Abstract: Level shifts, deterministic seasonality, spurious correlation,

Cited by 20 publications

References 21 publications

Can Managers Successfully Time the Maturity Structure of their Debt Issues?

Can Managers Successfully Time the Maturity Structure of their Debt Issues?

Nonsensical and biased correlation due to pooling heterogeneous samples

Linear and Nonlinear Association Measures With Intraday/High Frequency Data for All Ibovespa Stocks

Contact Info

Product

Resources

About