Explaining Market Data to Employees

By Jim Fox and Bruce Lawson, Fox Lawson & Associates, A Division of Gallagher Benefit Services, Inc.


Question: We just completed a market survey to determine how our pay compares to the market. When we started the survey, we thought that this would help answer questions that employees had, but instead, it has raised more questions than it has answered. Employees complain that the data is not correct and they know for a fact that others doing the same job in other similar organizations are paid more. We have tried to explain to them how we conducted the survey and the comparisons we made but they think that we “cooked the books” so that we would not have to raise their pay. What else can we say to them?

CompDoctorTM: Unless you found that they were significantly underpaid, employees may not believe any data that you present. After all, they have it on good authority that the guy in the neighboring town does not have nearly as much responsibility (volume of work, number of direct reports, budget, number of functions, number of committees, number of elected officials, miles of street, case load, or any other number of some things that you can count) as they do and the guy in the neighboring town gets paid more. This is nearly a universal truth. Unfortunately, if this were actually true, then the only person that is underpaid is the one that you are talking to at the time! Mathematically, that is impossible.

Nevertheless, you want to control the uproar and calm the masses before the lynch mob shows up at your door. While you will not be able to convince all of the people all of the time, you can convince most of the people most of the time with facts that stand on sound logic based on professional standards and processes that can be replicated. So, what is the logic and what are the standards. We will try to address the major ones below.

First, recognize that there is no such thing as “the market” for a specific job. The market fluctuates almost on a monthly basis, and varies based on how

you analyze the data. What that means is that the market is not one number that is fixed in time. Further, using various statistics, you can make this number move a bit. (No, I have not just given your critics ammunition to support their contention that you have “cooked the books.” I have just recognized what most people do not, and that is that the market is what you define it to be. Let us explain. There are several areas that need to be decided when you conduct a survey. They are:

1. The definition of the labor market 2. The benchmark, or comparable, jobs to be surveyed 3. The salary data that will be compared 4. How the salary data will be compiled. Let’s address reach one.


Definition of the Labor Market

Professional standards suggest that you survey the organizations that you compete with; in other words, the organizations that you may lose employees to and those that you may hire from. In this area, you need to be reasonable. You may have lost an employee recently to a high paying city because of family or life style changes. They may have gotten a similar job for more money at their new location. This comparison does not count since they did not leave your organization because of a better offer. This means that you should survey those organizations that likely will hire your employees.

The organizations you want to survey will differ by type of job. For example, a reasonable labor market for a nonexempt employee may be other employers within a 50-mile radius. Those in the professional, exempt job category may be organizations that are within a broader radius. Top-level jobs may be a broader regional area or even national. You need to be careful here because different parts of the country will have different levels of pay for the same job. If you are in central Missouri, for example, you can compare your top job to those in the New York City or Los Angeles area, but because the cost of labor is at

a different level in those areas, you may need to adjust any market data by a “cost of labor” index. In selecting comparable cites, you want to match as many organizational characteristics that define the character of the organization. For public sector organizations, this may mean comparisons on such things as per capita income, population, growth rate, crime rate, geographical size and proximity, population demographics, types of services offered, industry make up and so forth. While you will not be able to find another organization that matches perfectly, you want to identify those that are reasonably similar to your organization on several different key characteristics.

Finally, you want to identify about 20-30 organizations, if you can. The reason for this is that you need a sufficient number of comparable organizations so that the statistics you run are valid. Statistical standards suggest that you can make reasonably valid conclusions on the basis of 20 or more organizations. The Department of Labor says five. It really depends on how you use the data later, but a good rule of thumb is that more is better, but more than 20 may not be productive. Remember also that some organizations will simply not give you any data, so you need to plan on having data from fewer organizations than you originally identify as comparable.


Benchmark Job to Be Surveyed

Benchmark jobs are those that are commonly found in other organizations and defined in a similar fashion across organizations. They are commonly found in published surveys. They may also be called standard jobs.

Not all jobs are standard jobs. For example, accountant, HR analyst, budget Officer, building inspector, and programmer are typically benchmark jobs. On the other hand, coordinator, project manager, accountant III, or airport plumber may not be good candidates because they are either too generic, or designed too specifically.




If you have a formal job evaluation system like a point factor system, professional standards suggest that you only need to survey about one-third of the job titles in your organization. However, if you do not have a formal job evaluation system for determining pay grades, then you need to survey at least 50 percent of the job titles.

Just because you did not survey a particular job, does not mean that you cannot arrive at a reasonable level of pay that is both market competitive and internally fair. The reason you should have a formal job evaluation method is so that you can estimate with reasonable accuracy and precision what you should pay for a job when you have not collected, or cannot collect, market data.

Finally, when you compare your job to one in the market, you want to match about 70 percent or more of the duties and responsibilities in the other organization. You will never get a 100 percent match, so trying to obtain that objective is not reasonable or practical.

While you have probably heard this: “my job is so unique, that I don’t know that you can compare me to any other.” This is either a code for “I am overpaid already” or a statement that confirms that it should not be used as a benchmark, because the employee has already said that there are no comparables. Let your job evaluation system determine the correct internal pay in this case.


Salary Data Compared

Typically it is wise to collect the following data:

1. The minimum salary paid
2. The maximum salary paid
3. The actual average of the employees’ pay

You may also want to collect benefits information, hours worked, sick leave, any longevity pay, if they pay on a step plan or open range, etc.

Since you will be collecting these data from the HR department, these are the official rates of pay, regardless of what your employees state.

Again, I remember a client employee that claimed that another organization paid a specific job about $2,500 more per year. They swore by it, and had documentation. Come to find out, they were comparing the wrong level of job in the job series. It pays to verify.


Salary Data Compiled

From the data collected, you will want to do a couple of things. Eliminate the outliers. These are data points that are about two standard deviations from the average for the job. You want to use this test to make sure that the job matches are correct and, if they are and the data are still exceptionally high or low, you will want to eliminate it because there is something different about the data that you cannot determine. Since such data are extremely different from the rest of the data, by excluding it, you will have increased the reliability and validity of the remaining data.

Once this is done you want to calculate the average, and the median. The median is the middle number of the data points for each job and represents the most stable number. If the distribution is skewed in any way (lots of low salaries or lots of high salaries) the median will not be as affected as the mean.

There are other statistics that you can calculate as well, such as the weighted mean. This number is an average weighted by the number of incumbents in the class. It will differ from the straight mean or unweighted mean based on the distribution of the employees’ pay.

To the left is a chart that might cause you some concern. In this chart are the results of a survey in the central part of the U.S. for a senior property appraiser. Table 1 shows how the salary numbers can move around a bit.

These data show that there are some very high numbers, (probably from one or two employers) that skew the distribution to the high side. But when you factor in the number of employees, there is good number of lower paid employees. If I were an employee, I would want to select the weighted average including the outliers. You, on the other hand, should select the median excluding outliers because it is the best and most stable data.

Let’s look at another set of data to confirm that the median excluding outliers is the best data (Table 2). Here is data from the property appraiser, a lower level job, but in the same job family.

Here the weighted average including outliers is lower, while the median excluding outliers remains a good choice. So the guideline here is that you need to pick one comparison number and stick with it. Our money is on the median excluding outliers, since it is the more stable number and represents the number where one half are paid more and one half are paid less. What could be fairer than that?


The Comp Doctor™ is the team of Jim Fox and Bruce Lawson of Fox Lawson & Associates, a Division of Gallagher Benefit Services, Inc., a compensation, benefits and human resources consulting firm that specializes in assisting governments in fixing their compensation, benefits and classification systems. You may find them on the Web at www.foxlawson.com. If you have a question, you would like to have them answer, please write to them at james_fox@foxlawson.com or bruce_lawson@foxlawson.com. They will try to include it in the next issue of Comp Doctor™.