Methodology of industry statistics: averages, quantiles, and responses to atypical values
| Autor: | Martin Boďa, Vladimír Úradníček |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2020 |
| Předmět: |
finanční poměry
oříznutý průměr Central tendency Strategy and Management nesmyslné hodnoty Financial ratio non-sense values statistiky průmyslu quantile 0502 economics and business Statistics Economics Financial analysis Business and International Management power law in the tail financial ratios winsorized mean Winsorized mean winsorized průměr kvantil 05 social sciences Winsorizing Truncated mean zákon nepřímé úměrnosti industry statistics 050211 marketing Frequency distribution General Economics Econometrics and Finance 050203 business & management Quantile trimmed mean |
| Popis: | The paper notices troublesome aspects of compiling industry statistics for the purpose of inter-enterprise comparison in corporate financial analysis. Whilst making a caveat that this issue is unbeknownst to practitioners and underrated by theorists, the goal of the paper is two-fold. For one thing, the paper demonstrates that financial ratios are inclined to frequency distributions characteristic of power-law (fat) tails and their typical shape precludes a simple treatment. For the other, the paper explores different approaches to compiling industry statistics by considering trimming and winsorizing cleansing protocols, and by confronting trimmed, winsorized as well as quantile measures of central tendency. The issues are empirically illustrated on data for a great number of Slovak construction enterprises for two years, 2009 and 2018. The empirical distribution of eight financial ratios is studied for troublesome features such as asymmetry and power-law (fat) tails that hamper usefulness of traditional descriptive measures of location without considering different possibilities of handling atypical values (such as infinite and outlying values). The confrontation of diverse approaches suggests a plausible route to compiling industry statistics that consists in reporting a 25% trimmed mean alongside 25% and 75% quantiles, all applied to trimmed data (i.e. data after discarding infinite values). The paper also highlights the sorely unnoticed fact that the key ratio of financial analysis, return on equity, may easily attain non-sense values and these should be removed prior to compiling financial analysis; otherwise, industry statistics is biased upward regardless of what measure of central tendency is made use of. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|