Search results

This chapter takes a closer look at outliers and extreme outliers identified in the data derived from a complete case treatment of missing values in the European and North American datasets and consistently observe significant negatively skewed distributions with high excess kurtosis across all industries. We then plot the density functions for return on assets (ROA) across different industries in the two datasets and find pervasive observations in the tails where negative returns and outlying observations constitute a frequent and recurring phenomenon. We analyze the persistency of outliers and find noticeable percentages of outlying over- and underperformers hovering around 3–6% dependent on industry context. We further analyze potential size effects associated with extreme negative skewness but do not find that (even sizeable) elimination of extreme values reduce the phenomenon. Finally, we analyze the percentage of firm observations that must be eliminated to reach at distributions that fulfill the characteristics of a normal distribution and reach at a substantial percentage of around 5–10% dependent on industry. To conclude, the often-assumed normally distributed performance outcomes are typically wrong and discards the substantial number of outliers in the samples.

The presence of respondents with apparently extreme sensitivities in choice data may have an important influence on model results, yet their role is rarely assessed or even explored. Irrespective of whether such outliers are due to genuine preference expressions, their presence suggests that specifications relying on preference heterogeneity may be more appropriate. In this paper, we compare the potential of discrete and continuous mixture distributions in identifying and accommodating extreme coefficient values. To test our methodology, we use five stated preference datasets (four simulated and one real). The real data were collected to estimate the existence value of rare and endangered fish species in Ireland.

In this chapter, we first examine the distribution characteristics of firm performance across different competitive industry contexts and periodic economic conditions of growth, recession, and recovery. There is mounting evidence that the contours of accounting-based economic returns consistently display (extreme) left-skewed leptokurtic distributions with negative risk-return relationships, which implies the existence of many negative performance outliers and some positive outliers. We note how negative skewness, excess kurtosis, and inverse risk-return relationships prevail in industries with more intense competition and in economic growth scenarios where more innovative initiatives compete. As the study of outliers typically is ignored in mainstream management studies, we extract a total of 23 extreme performers using a conventional winsorization technique that identifies 16 negative and 7 positive outliers. We study the performance trajectories of these firms over the full period and find that negative performers typically operate in capital-intensive innovative industries whereas positive performers operate in activities that cater to prevailing demand conditions and expand the business in a balanced manner. The firms that under- and over-perform as measured by the financial return ratio both constitute smaller firms compared to the total sample and show how relative movements in the ratio numerator and denominator affect the recorded return measure. However, the negative outliers generally use their public listing to access capital for investment in more risky development efforts that require a certain scale to succeed and thereby limits their flexibility. The positive outliers appear to expand their business activities in incremental responses to evolving market demands as a way to enhance maneuverability and secure competitive advantage by honing their unique firm-specific capabilities.

This chapter first analyzes how the data-cleaning process affects the share of missing values in the extracted European and North American datasets. It then moves on to examine how three different approaches to treat the issue of missing values, Complete Case, Multiple Imputation Chained Equations (MICE), and K-Nearest Neighbor (KNN) imputations affect the number of firms and their average lifespan in the datasets compared to the original sample and assessed across different SIC industry divisions. This is extended to consider implied effects on the distribution of a key performance indicator, return on assets (ROA), calculating skewness and kurtosis measures for each of the treatment methods and across industry contexts. This consistently shows highly negatively skewed distributions with high positive excess kurtosis across all the industries where the KNN imputation treatment creates results with distribution characteristics that are closest to the original untreated data. We further analyze the persistency of the (extreme) left-skewed tails measured in terms of the share of outliers and extreme outliers, which shows consistent and rather high percentages of outliers around 15% of the full sample and extreme outliers around 7.5% indicating pervasive skewness in the data. Of the three alternative approaches to deal with missing values, the KNN imputation treatment is found to be the method that generates final datasets that most closely resemble the original data even though the Complete Case approach remains the norm in mainstream studies. One consequence of this is that most empirical studies are likely to underestimate the prevalence of extreme negative performance outcomes.

This chapter suggests a robust Hausman and Taylor (1981), hereafter HT, estimator that deals with the possible presence of outliers. This entails two modifications of the classical HT estimator. The first modification uses the Bramati and Croux (2007) robust Within MS estimator instead of the Within estimator in the first stage of the HT estimator. The second modification uses the robust Wagenvoort and Waldmann (2002) two-stage generalized MS estimator instead of the 2SLS estimator in the second step of the HT estimator. Monte Carlo simulations show that, in the presence of vertical outliers or bad leverage points, the robust HT estimator yields large gains in MSE as compared to its classical Hausman–Taylor counterpart. We illustrate this robust version of the HT estimator using an empirical application.

In this paper, we follow the same logic as in Hausman (1978) to create a testing procedure that checks for the presence of outliers by comparing a regression estimator that is robust to outliers (S-estimator), with another that is more efficient but affected by them. Some simulations are presented to illustrate the good behavior of the test for both its size and its power.