0:06
And before we ask this question, we should point out that n most jobs,
short-term incentives are not a big part of total compensation.
Rather, if we're looking at the pay mix as a whole, most people are getting
the majority of their pay through base pay, not through short-term incentives.
And so let's look at some examples.
If you think about nurses, or we think about perhaps office workers,
salaried office workers, these are jobs where
most people aren't getting the majority of their pay through short-term incentives.
But on the other hand, short-term incentives are prominent pieces
of total compensation package in some other jobs.
So for example, in a factory, a worker might be paid a piece rate
that is a certain commission for every unit that they produce.
Or in agricultural work,
people might be paid per basket or bushel of produce that they pick.
And in some jobs, we might see incentives, but
their use is relatively controversial, such as for teachers.
1:11
And so what makes these jobs that use short-term incentives
different from jobs that don't?
Or in other words, what is it that seems to make a job a good candidate for
short-term incentives?
And what is it that makes a job seemingly a bad candidate for short-term incentives?
And to answer those questions,
we want to look at kind of the typical problems of short-term incentives.
That is, these are the typical reasons why a job
might not actually be paid with using short-term incentives, and
instead might be paid more in terms of a base salary.
So the first problem is the alignment problem.
So what's the alignment problem?
The alignment problem is that the performance measures aren't very well
aligned to what it is that we want to get out of our workers.
And so kind of our lesson is, don't pay for A while expecting B.
So we might see this in examples like gaming, gaming an incentive plan.
There is a famous example from a company called Lincoln Electric.
And in that company, they used to pay typists by the keystroke.
What the found were that some of the most productive typists
were just spamming one key on their keyboard over their lunch break.
Actually, this wasn't a very productive activity.
2:26
The second problem is the measurement problem.
That is, you can't measure what it is that we want to measure.
And the lesson here is that you shouldn't pay
short-term incentives for things you can't easily measure.
And kind of the example from Lincoln Electric, again,
would be, we want those keystrokes to be high quality.
But at the same time,
we can't really measure easily what it is that makes a keystroke high quality.
2:53
The third problem is the metering problem, and that is that you can't define success.
Perhaps this is a job that you haven't had before in the past.
And so you can't tell the difference between a good output and a poor output.
In other words, we find it hard,
difficult to benchmark these jobs and to find success.
Kind of our lesson here is that we want to benchmark low short-term incentives, or
low target incentives for low performance, and high short-term incentives for
high performance.
And to do that, of course, we need to be able to define what low performance is.
And we need to be able to define what high performance is.
3:30
A fourth problem is the control problem.
So what's the control problem?
Well, that's where your success depends on factors that are outside of your control.
And our big lesson here is, don't use incentives for
measures that workers can't control.
Workers tend to be more risk averse than their firms.
And so when you're paying them incentives for
things that are outside of their control, you're pushing more risk on those workers.
And it's also potentially very demotivating
to get a cut in pay because of things that you never really had control over.
4:16
So the first example I want to get back to are paying teachers for
the test scores of their students.
So here, the alignment problem could be that we want teachers to teach both
technical skills and also soft skills like creativity and cooperativeness, and so on.
And that goes along with our measurement problem which is,
technical skills are easy to measure.
We can give, for example, a test at the end of the year to see,
to gauge students' learning on those technical skills.
But it would be very hard to administer a test that actually measures the students'
creativity or their ability to cooperate with others.
And so since it's difficult to measure those other things that we also want
teachers to be imparting to their students,
we end up having an alignment problem when we design these tests.
And that leads to things such as teachers teaching to the test, as opposed to
teaching both the skills that are on the test and also those other things,
those values that we want teachers to be imparting to their students.
5:18
We also have metering control problems as well.
So for example,
when we're giving an exam, we may not know what scores are grade appropriate.
So for example, if we have a new exam, and
it's being rolled out to perhaps a very small population,
then we might not know whether those scores are good scores or
poor scores, because we don't have any experience using that particular exam.
6:00
they have their teacher for a full year and
at the end of the year only 60% were reading at grade level, it could look bad.
But if I told you that before, when the students
had come into that person's class, 0% of them were reading at grade level.
Well in that case, even though 60% reading at grade level isn't what
the teacher wanted, it's still a very big improvement.
And so by that measure, according to the metering problem, well, maybe
the teacher actually did do a good job by helping some of those students catch up.
6:45
Let's take another example from factory piece rates.
So in a factory, if you're paying a piece rate, that is, a dollar value for
every unit that they produce, then quality might suffer.
That's because you're paying for quantity, but you're expecting both quantity and
also quality.
And so if we look at the measurement problem, we're saying that we can't, and
so kind of related is the measurement problem.
We may not be able to keep track of the numbers that workers produce.
And it might also be very difficult to measure the quality of the items that they
produce as well.
That's why, to solve the measurement problem,
we also might have to have some sort of a way of using serial numbers or
something to track the quality of the items as they come down the assembly line.
That way we can solve that alignment problem.
Now we get to metering control.
Our metering problem might be that for a specific task,
we can't define good performance.
So let's say that a worker perhaps gets a new piece of technology a new piece
of equipment, and
we find out that there's a 2% defect rate on the items that they're producing.
Now is that a good performance?
Well, we don't necessarily know, because we don't have experience with that item.
We don't have experience with that new piece of capital.
It could be the case that that 2% defect rate is very high or very low.
But we just don't know, because we don't have good data on that yet.
Also, we have the question of control.
Again, the control problem here would be that employees don't necessarily have
control over the workmanship of their inputs.
So for example, if we're paying workers for the quantity that they produce and
also for the quality of the items that they produce, it could be challenging.
Because perhaps the materials,
perhaps the steel that they're getting is defective or of low quality.
Or if they are assembling cars,
it could be that the suppliers are giving them poor engines to put into their cars,
and that's leading a lot of these cars to be returned to the manufacturer.
And again, these are for things that are outside of the worker's control,
because they're only getting those poor inputs from somewhere else.
8:54
And so kind of the big idea here is,
again, short-term incentives are only one element of the pay mix.
And is also, we can include multiple different short-term
incentives as part of our whole short-term incentive package.
So for example, a piece rate where we're just paying workers per unit they produce
might have quality concerns,
as workers allocate their effort more toward quantity of production.
And so one thing we can do is,
we can incorporate piece rates into a plan that also includes bonuses for quality.
9:39
there could be a problem that seems to be outside of a worker's control.
But after the incentive is introduced, it's discovered that worker could have
at least some ways of actually managing some of that risk.
So for example, there's a famous example of a company called Safelite Auto Glass.
When they introduced an incentive to their workers for
installing and repairing auto glass, they found that there
are some things that seemed to be outside of the worker's control.
So for example, whether the customer was actually present
after they called in the representative from Safelite Auto Glass.
10:19
But after the introduction of the incentive,
some of the most savvy workers began to phone ahead to make sure that the customer
was ready to have their car serviced.
And so this is just an example of how when introducing incentives,
you want to enable workers to fix problems as they arise themselves.
And again, this is,
in itself, a very productive, potentially very productive activity.