All right audit learners.
We've talked a lot about auditing and let's get a little bit more crystal clear
now on how auditing affects the quality of financial reporting.
I like to ask my students, suppose you are an auditor and I asked you to conduct a study,
design a study, to test whether auditing works or doesn't work.
What are some of the things you could do?
Immediately you would realize there's a bit of a problem,
because you can't use the...for public companies
especially, you can't use a traditional experimental approach.
The traditional experimental approach is to use
random assignment to assign a treatment to
one condition and no treatment maybe a placebo treatment to another condition, randomly.
So you would have audited companies and
not audited companies and then you'd want to make some comparison
about how those two types of companies performance fared subsequent to the audit.
But if most, indeed, all public companies get an audit, that's going to be a problem.
So one thing you might realize is that boy we would have to get our hands on
some private company data to do a test where
there's maybe still not randomly assigning into an audit versus not an audit.
But at least we could see what types of companies
who don't need to get an audit choose to anyway and when they
choose to get an audit, do you see any benefits to those companies
as opposed to other companies, who on all other observable characteristics, look similar.
That's one approach. Okay, maybe you
would have gotten there, and maybe you were well ahead of my thinking there.
But suppose I said, well how would you test this in a public company setting?
Well maybe you would say we would have to get
into the auditors working papers because what we would really want
to know is what would earnings and assets and liabilities have been without an audit,
compared to what they were with an audit.
What you're really getting at then is, hey does the auditor,
when they come into their field work stage and they've plan their audit,
they've assessed risk at a preliminary level,
and they send auditors into the field doing some nuts and bolts,
control tests and substantive tests,
to look for evidence that the controls are working as designed,
or to look for, in the case of substantive test, misstatements, should they exist.
If misstatements do exist, it would be one thing if
about half of the time client management seems to have understated income,
assets, and overstated liabilities.
The other half of the time a client management seems to
have overstated income and overstated assets and understated liabilities.
So I'll talk to you about two studies,
two pretty cool studies that look at both of these methods.
The Kinney and Martin one is an older one,
it's clear from 1994.
One reason it's so old is that the audit firms of today, although
there's some real glimmer of hope, here are just a little bit,
quite a bit, more reluctant to share their auditing working papers.
So Bill Kinney's one of the most prolific audit scholars out
there and Roger Martin who is also very
good in his own right, one of his students at Virginia,
Kinney was at Texas out of Michigan State.
Michel Minnis is an Illini, he's a professor at the University of Chicago booth right now.
We should have tried to hire him back because he does have a Champaign-Urbana roots.
Michael Minnis looks at private company data.
Let me show you some really cool results and I'm trying to
inform you about the profession that many of you are about
to enter into, to inform you and equip you with
knowledge about the value that your profession is adding to society.
So when you look at the Kinney and Martin study they had an analysis
of nine separate data sets of audit-related adjustments.
So this would be either cases of known air,
so an auditor finds an error and says,
okay management here it is. This is a misstatement. Management you fix it.
Or it could be a case of an estimate that
management has that the auditors not believed to be reasonable.
Then there would be a conversation and they'll
be at either a meeting of the minds or the auditor might say,
I just think the number you have is just outside of
the bounds are reasonableness, we're going to require an adjustment.
The last kind would be when you have a sample, and we'll
talk more specifically about this much later in the course.
But when you do a sample, suppose you take a sample
of 100 items and you only find three misstatements.
You still have a fudge factor you have to allow for.
Three misstatements projected out of 100 on
your sample, suppose there are 500 in the population, you would then
multiply that by four to get 12 more, right?
That's a projection. You haven't found any misstatement but it's called a projected
error and it's very reasonable to ask for those to be adjusted as well.
What's really cool in this study is that overall audit-related adjustments show
an overwhelmingly negative effect on pre-audit net earnings and net assets.
The average aggregate adjustment in fact was by
two to eight times the minimum amount that would
be material misstatement at those companies.
Suppose you take back to some of their discussions about
materiality, one common base of
materiality would be five percent of net income before taxes.
Another common one would be one percent of total assets.
So you look at these company's balance sheet, you say one percent of total assets is
x or you look at the income statement and say five percent of that income taxes is y.
Take those numbers by two to eight times,
let's say four or fives in the middle and that would tell you what
the average downward adjustment on earnings and net assets was in these data sets.
So let me give you a little bit of background on some of the particular data sets that
they looked at, one from the early 80's, one from the early 90's.
These happen to both be Peat Marwick clients and if you know
your audit firm trivia Peat Marwick is a precursor to KPMG,
so it's one of the big four companies now.
Number of audits 130,
240, what do you find in examples like this?
Well they found that earnings on average
had been overstated before the auditor had come in.
There were 144 overstatements in
the sample number two from the early 80's, as opposed to the cases when there are
understatements and ratio about 1.4 times as often you're seeing an overstatement on
average and an understatement, 63 percent non-zero.
If you look at the sample seven from the early 90's,
you have about a 1.5 clip.
You do see some understatements,
but you see the point here is that auditing seems to be removing bias on average here.
Now, I want to talk to you a little bit about Michael Minnis' study next.
So the Kinney study says that auditing is not just removing noise from the system.
If the average audit adjustment had been zero, then you can make a stronger case that
auditing removing noise from the system, half
the time understated, half the time overstated.
It doesn't look like that at all.
It looks like auditing is doing something that is helping investors by getting
rid of some of management serving bias in upward bias and earnings and assets,
that's kinda what's being shown here in these data.
Now, if you put your thinking cap on, you can critique this finding, right?
You could say, well,
the only reason the auditors are finding adjustments is
because management knows that auditors need to find something.
So they may engage in own strategic behavior pumping up earnings and assets
to let the auditor's find something and go through
the motions to find something and adjust their books.
I can't rule that out that is an alternative explanation for their findings,
but nonetheless it is some of the only evidence in
public companies showing that auditors are removing bias, not just noise, from the system.