Identificação manipulação de resultados

: o modelo de McNichols (2002)

Abstract

The quality of information issue has assumed a key role in the accounting and finance literature. The literature distinguishes various forms of earnings management. Earning management can be made via real decisions or via accruals. Moreira (2013) defines manipulation by real decisions as "discretionary choice of the transaction occurrence of time in order to produce the desired effect on results”. This thesis will examine the particular case of an accruals model, that is the model of McNichols (2002), which was estimated based on a sample of 8,930 American companies for the period between 1988 and 2014, with a total of 35,670 observations. Based on discretionary accruals estimated by the McNichols model (2002), 4 manipulation indicators were proposed, and then compared with cases of effective manipulation detected by the SEC (corresponding to 1,256 US companies between 1990 and 2013). The question that this thesis seeks to respond is whether the McNichols model (2002) is able to earnings management. The proposed indicators were prone to type II errors. In 32% of our sample, none of the proposed indicators was able to detect actual manipulations. However, one of the calculated indicators (based in absolute value ratio of discretionary accruals and working capital), correctly detected 64% of manipulations identified by the SEC.

Date of Award	30 Nov 2016
Original language	Portuguese
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Ricardo Ribeiro (Supervisor)