HDRs defended

HDRs defended

Title : The quality of the external audit and the company's performance

Author : Riadh Manita (NEOMA Business School)

Thesis supervision : Pr. Mehdi Nekhili

Date of defence : 7 december 2020

Abstract : The central issue of this HDR is to identify the factors that can affect the quality of the external audit. Audit quality is generally defined as the combined probability that an auditor will detect anomalies, errors or omissions in the financial statements and disclose them to third parties. Despite research in this area, assessing and measuring audit quality remains complex. In this context, our work has made it possible to propose new assessment approaches, implement new empirical methodologies and propose metrics of the quality of the audit process. It has also enabled us to analyze and understand the ethical judgment of materiality (a key factor in audit quality), the audit quality deficiencies identified by the audit regulator (PCAOB) and the risks considered by major audit firms in their decision to accept new clients. Our research has also made it possible to analyze the auditor-auditor relationship and to identify the auditee's psychological reaction behaviors that may affect the quality of the audit. We also studied the impact of new digital technologies (Big data, Analytics, Artificial Intelligence, Blockchain) on the accounting profession and on the quality of the audit. Finally, our work focused on the governance of audit firms, decision-making processes and the determinants of career success in large audit firms. In a second line of research, we studied the impact of gender diversity in boards of directors and ownership structure on ESG and corporate performance.  Future projects will focus on the implications of new digital technologies on the regulation and regulation of the external audit market. They will also look at the imbalance of power between auditor and auditee and their effects on audit quality. Finally, they will look at the impact of gender diversity in audit firms on innovation, change management, recruitment policies and performance.

Title: Development of measurement models and dynamic structural models for the analysis of marketing decisions and strategic actions: The contribution of causal graphs

Author: Mouloud Tensaout

Thesis supervision: Prof. Nathalie Fleck

Date of defence: 13 November 2017

Summary: The research work defended in this document focuses on two axes: The validity of measurement models used in management sciences and the modelling and evaluation of marketing and strategic actions. In relation to the development of valid measures, the subject of the first axis, this involved a critical review on the development process and the use of reflective and formative measures in marketing. This review made it possible to identify the main debates raised in the literature on these measures. One of the recent controversies between two strands of research relates to the validity of the formative measures. Opponents of these measures have put forward epistemological and empirical arguments, particularly those relating to the instability of the parameters of these measures. The contribution was to show that formative measures are not affected by any instability when the formative measurement model is correctly specified and identified. Another recent question much discussed in the literature concerns the hypothesis of the superfluous use of formative measures in the social sciences. This hypothesis cannot be rejected. The second axis focuses on the development of models for evaluating marketing and strategic decisions as well as the process of validation of these models by confirmatory analyses or more generally by a structural equation model (SEM). The usual validation procedure for structural/causal relationships based on significance tests and estimated model fitting to the data comes up against the problem of observational equivalence, namely that rival models cannot be distinguished by means of indices and statistical fit tests. Several questions emerge in relation to the modelling of marketing and strategic decisions and the measurement scales of unobservable constructs (measurement models). What procedure should be implemented to develop valid indicators or measures to represent a latent variable (construct)? What procedures for specifying and validating a structural model encoding testable local constraints sufficiently robust to detect a bad specification? How do we identify these testable implications? Can we statistically distinguish between rival models with local testable implications?