Thomas Holgersson

Testing for Multivariate Heteroscedasticity

In this article, we propose a testing technique for multivariate heteroscedasticity, which is expressed as a test of linear restrictions in a multivariate regression model. Four test statistics with known asymptotical null distributions are suggested, namely the Wald, Lagrange multiplier (LM), likelihood ratio (LR) and the multivariate Rao F-test. The critical values for the statistics are determined by their asymptotic null distributions, but bootstrapped critical values are also used. The size, power and robustness of the tests are examined in a Monte Carlo experiment. Our main finding is that all the tests limit their nominal sizes asymptotically, but some of them have superior small sample properties. These are the F, LM and bootstrapped versions of Wald and LR tests.

Testing for Multivariate Autocorrelation

This paper concerns the problem of assessing autocorrelation of multivariate (i.e. systemwise) models. It is well known that systemwise diagnostic tests for autocorrelation often suffers from poor small sample properties in the sense that the true size overstates the nominal size. The failure of keeping control of the size usually stems from the fact that the critical values (used to decide the rejection area) originate from the slowly converging asymptotic null distribution. Another drawback of existing tests is that the power may be rather low if the deviation from the null is not symmetrical over the marginal models. In this paper we consider four quite different test techniques for autocorrelation. These are (i) Pillais trace, (ii) Roys largest root, (iii) the maximum F-statistic and (iv) the maximum t2 test. We show how to obtain control of the size of the tests, and then examine the true (small sample) size and power properties by means of Monte Carlo simulations.

A simple multivariate test for asymmetry

Since many macroeconomic models are linear, it is not desirable to use them with an asymmetric dependent variable. In this article, we formulate a univariate test for symmetry, based on the third central moment and extend it to a multivariate test; the test does not require modelling and it is robust against serial correlation, Autoregressive Conditional Heteroscedasticity (ARCH) and nonnormality. In the empirical application of the test it is found that orthodox theory seem to be supported; consumption expenditure on durable goods is found to be symmetric while consumption expenditure on nondurable goods is asymmetric for the USA and the UK, with peaks being higher than troughs are deep. Also, the empirical importance of the choice between the univariate and the multivariate test for possibly correlated series is underscored; the results from the two approaches clearly differ. Given the widespread practice of using consumption expenditure on nondurable goods as the dependent variable in linear models for the USA and the UK, our results might be noteworthy.

Assessing Normality of High-Dimensional Data

The assumption of normality is crucial in many multivariate inference methods and may be even more important when the dimension of data is proportional to the sample size. It is therefore necessary that tests for multivariate non normality remain well behaved in such settings. In this article, we examine the properties of three common moment-based tests for non normality under increasing dimension asymptotics (IDA). It is demonstrated through Monte Carlo simulations that one of the tests is inconsistent under IDA and that one of them stands out as uniformly superior to the other two.

Three estimators of the Mahalanobis distance in high-dimensional data

This paper treats the problem of estimating the Mahalanobis distance for the purpose of detecting outliers in high-dimensional data. Three ridge-type estimators are proposed and risk functions for deciding an appropriate value of the ridge coefficient are developed. It is argued that one of the ridge estimator has particularly tractable properties, which is demonstrated through outlier analysis of real and simulated data.

A Note on a commonly used ridge regression Monte Carlo design

Ridge estimators are usually examined through Monte Carlo simulations since their properties are difficult to obtain analytically. In this paper we argue that a simulation design commonly used in the literature will give biased results of Monte Carlo simulations in favor of ridge regression over ordinary least square estimators. Specifically, it is argued that the properties of ridge estimators that are functions of p distinct regressor eigenvalues should not be evaluated through Monte Carlo designs using only two distinct eigenvalues.

Estimating mean-standard deviation ratios of financial data

This article treats the problem of linking the relation between excess return and risk of financial assets when the returns follow a factor structure. The authors propose three different estimators and their consistencies are established in cases when the number of assets in the cross-section (n) and the number of observations over time (T) are of comparable size. An empirical investigation is conducted on the Stockholm stock exchange market where the mean-standard deviation ratio is calculated for small- mid- and large cap segments, respectively.

Towards a multivariate innovation index

ABSTRACT This paper argues that traditional measures of innovation as a univariate phenomenon may not be dynamic enough to adequately describe the complex nature of innovation. Consequently, the purpose is to develop a multidimensional index of innovation that is able to reflect innovation enablers and outputs. The index may then be used (i) to assess and quantify temporal changes of innovation, (ii) to describe regional differences and similarities of innovation, and (iii) serve as exogenous variables to analyze the importance of innovation for other economic phenomena. Our index is defined in a four-dimensional space of orthogonal axes. An empirical case study is used for demonstration of the index, where 44 variables are collected for all municipalities in Sweden. The index spanning the four-dimensional innovation comprises size, accessibility, firm performance, and agglomeration. The proposed index offers a new way of defining and analyzing innovation and should have a wide range of important applications in a world where innovation is receiving a great deal of recognition.

Expected and unexpected values of individual Mahalanobis distances

ABSTRACT This paper derives first-order sampling moments of individual Mahalanobis distances (MDs) in cases when the dimension p of the variable is proportional to the sample size n. Asymptotic expected values when n, p → ∞ are derived under the assumption p/n → c, 0 ⩽ c < 1. It is shown that some types of standard estimators remain unbiased in this case, while others are asymptotically biased, a property that appears to be unnoticed in the literature. Second-order moments are also supplied to give some additional insight to the matter.

On regression modelling with dummy variables versus separate regressions per group: comment on Holgersson et al

On regression modelling with dummy variables versus separate regressions per group : comment on Holgersson et al.