Return to search

The impact of measurement errors and multicollinearity on the study of the relationship between accounting information and market-determined risks

This study develops a linear measurement structural equation model to study the relationship between accounting risk measures and market determined risk measures. Use of this model may reduce the risk of potential loss of information and/or erratic invalid inferences. / Extant literature in this area of research has produced conventional regression models which give inconsistent and conflicting results about the relationship between accounting risk measures and market determined risk measures. This study differs from previous research in several respects. First, the model introduced here takes specifically measurement error and multicollinearity problems endemic to this area of research into consideration. Secondly, besides contemporaneous relationship, incremental information content of accounting risk measures is also examined. Finally, replications of more important previous studies are conducted to compare with earlier findings. / Overall, compared with traditional regression approach this study finds that the new model produces much more clearcut and interpretable evidence about the relationship between accounting risk measures and market determined risk measures. All the financial constructs proxied by the accounting risk measures have the expected signs indicated by theoretical and/or analytical results. It is found that most, if not all, of the information contained in the selected accounting risk measures is impounded in contemporaneous market determined risk measures. There is only barely discernible incremental information content of these accounting risk measures. / Finally, replications of previous studies indicate that the instrumental variables approach in this area of research is not a viable methodology. It produces inconsistent, erratic and hard-to-interpret results. / Source: Dissertation Abstracts International, Volume: 49-06, Section: A, page: 1510. / Major Professor: Kenneth Stanley Lorek. / Thesis (Ph.D.)--The Florida State University, 1988.

Identiferoai:union.ndltd.org:fsu.edu/oai:fsu.digital.flvc.org:fsu_76282
ContributorsWoan, Ronald Jen-Jin., Florida State University
Source SetsFlorida State University
LanguageEnglish
Detected LanguageEnglish
TypeText
Format280 p.
RightsOn campus use only.
RelationDissertation Abstracts International

Page generated in 0.0021 seconds