Welcome back, today we're talking about salary structures and internal benchmarks.

And so in our previous video we looked at this world in which you can actually

observe all of the different benchmarks for

all of the different grades that you have, but that's kind of an ideal circumstance.

Often times we won't be able to observe the benchmarks for

all of our different grades.

So for example, we might be able to adequately benchmark an engineer and

a senior engineer, but we won't be able to adequately benchmark also the lead

engineer, and the principle engineer, and so on.

And so what are we supposed to do?

Well so let's take an example where we are able to benchmark two of these jobs in our

hierarchy, but we're not able to benchmark those middle engineers.

So, and we're going to see what we're supposed to do.

So the first thing we're going to do, is we're going to establish control rates.

That is, for those jobs that we actually are able to benchmark,

we're going to essentially calibrate our whole organizational hierarchy and

our pay hierarchy.

And from these control rates, we're going to fit a pay policy line.

That is, we're going to take the data that we have available and we're going to

fit a line, and then we're going to fill in the gaps using that pay policy line.

And that pay policy line is essentially a method for

mapping our pay grades and our job grades to median pays.

So let's take an example from a slightly bigger pay structure.

So suppose we have 7 different levels, and

here we're just going to put our midpoints for our salary structure.

So suppose that we are able to find good benchmarks for levels 1, 2, 4 and 7.

But we don't have very good benchmarks for 3, 5 and 6.

So again, our first step is to establish our control rates,

which are going to be the rates for those jobs that we can benchmark.

And then, we're going to fit our pay policy line.

And so one method for doing this is called regression.

Regression is just simply a method for

fitting a best fit line through our available data.

So you're going to use regression to fit a line through those

jobs that we can actually benchmark, and

then we're going to fill in the midpoints from that regression line.

And this is an example of what it might look like.

Those predicted values from that regression would establish the midpoints

for the structure for those specific grades, and then we can add and subtract

maybe 10 or 15% from those different grades to establish our full structure.