Honesty and Earnings Management | Knowledge Labs: In the Know - YouTube

We examine the influence of face-based judgments  of CFO and CEO honesty and earnings management  

for a large cross-section of  U.S. publicly traded firms  

and after we control for financial incentives and  opportunities to engage in earnings management,  

we find that both CFOs and CEOs that are  perceived to be less honest actually engage  

in higher levels of earnings management,  both in terms of accounting accruals as  

well as real forms of earnings management, such  as discretionary R&D costs and operating costs.  

We also find, interestingly, that the biggest  effect happens when both the CEO and the CFO are  

perceived to be less honest. So, in other words,  if both of them are perceived to be less honest,  

that results in the most amount of earnings  management. There's effectively an interaction  

effect between the CEO and the CFO. So, if you  go back about a hundred years ago, 150 years ago,  

many people believed that you could just tell  based on somebody's face whether they were evil  

or whether or not this was a person to be trusted  and due to some early research in psychology, that  

belief was found to be untrue. Now, it's important  to understand that much of this early research  

was looking at just one aspect of the face, so for  example the eyebrows, or perhaps looking at the  

person's lips, it didn't holistically consider the  entire face. In the past 10 years or so there's  

been a number of researchers, specifically at the  University of Toronto, that have done a lot of  

research in the lab looking at what we call 'thin  slicing,' which is taking very short snippets,  

either looking at pictures or let's say two  to five seconds versus a video of somebody  

talking and then from there making a link between  that, between perception, and outcome variables.  

So, I'll give you an example: there's a paper  from about 10 years ago by Rule and Ambady, 2010,  

where they found the perceptions of power that  strangers make from undergraduate yearbook photos  

actually predicted those individuals success  as managers years later. There's another study,  

about 14 years ago, that found that celebrities  tend to be narcissistic way before they become  

famous. In our paper, we only get results when we  actually get a number of people to rate a photo.  

This is a really, really important point  that some people have come back and said well  

people have lots of different opinions  about a certain individual, I might say  

person X looks honest and my friend might  say that person looks dishonest, right? So,  

there is a lot of variability in people's  perceptions of a given person and there's  

only significance in terms of findings in our  paper when we take the average of those scores.  

So, all the pictures in our study are rated about  20 times. We take an average of those scores as an  

estimate of a person's honesty. It's only then  that we get results so it's almost analogous,  

if you will, to the the concept you might have  heard: wisdom of the crowd. So, it's actually  

when the whole crowd assesses, it's that  average score that has meaning. Individuals  

vary significantly with their judgments. Another  more related or more interesting study that we're  

looking at is looking at environmental, social,  and governance. This is a huge thing right now  

that you see in the media around environmental  concerns, social welfare governance, and what we  

find there is even stronger results, namely those  individuals that are perceived to be less honest,  

CEOs and CFOs, they actually do a lot less in  terms of engaging in environmental programs and  

social benefits for their employees and for the  public at large. So, I think that's going to be  

a really, really interesting study and with that  study we're also going to significantly increase  

our sample. With our first study, we looked at  three years 2011 to 2013, but this new study we'll  

be extending it from 2000, the year 2000 to the  year 2020. So, it'll be a very long time period,  

I think we'll find some very interesting results  that have happened over that 20-year cycle.