The Salary Curve (salaries versus experience) and its origins

The Salary Curve (salaries versus experience)

The figure above gives 1996 Total Salaries (base salaries plus bonuses) of All Engineers, All Degrees, from the Engineering Workforce Commission of the American Association of Engineering Societies. Curves for the 10, 25, 50, 75 and 90th percentiles are shown and identified in the legend. Subsequent plots will use the same color coding. The 10th percentile means that only 10% of salaries are below that curve and 90% of salaries are above that curve.

This curve is for all engineers, averaging over the significant differences in engineering salaries with discipline, degree, area of the country, and for managers and non-managers. For example, most surveys show that software engineers have lower salaries than telecommunications, automotive and electric/gas utility engineers. For 1995, the IEEE reports that software engineers had an average base salary of $43,150 compared to the average base salary for all engineers of $56,800, a ratio of 76%. Silicon Valley engineers command higher salaries than those in the MidWest. The references below give such breakdowns.

To confuse matters further, 55% of surveyed engineers reported getting a bonus in 1995 which significantly raised their total compensation. Compare the average base salary of $56,800 in 1995 to the average total salary of $71,900 in 1994!

Salary surveys can suffer from many biases:

"Average" salary numbers also suffer from an extreme sensitivity to the distribution of experience level. If software engineers averaged 10 years less experience than all engineers, that pay disparity would vanish under proper analysis!

Inflation is an insidious force that makes it difficult to compare our past salaries to current ones. To a first approximation over time periods of ~10 years, the salary curve scales directly with inflation. To put your previous salaries on the above salary curve, use Inflation Rates 1913 to 1996 With Conversions to 1996 Dollars.

Thus examine closely what is behind any salary surveys before you use them to compare your salary to it, and take them with a grain of salt!

My analysis here is confined to:

Note that these features of all salary curves for all engineers and scientists are generally true only for jobs that pay a premium for experience and recognize individual variation in performance. Many jobs do not have one or both of these features:

As you progress through these writings, please remember that I am only analyzing salaries, and that salaries are only one component of a job. I hope you are an engineer or scientist because you love doing what you do. A paycheck is in most cases a necessity, but it should not be the only reason for getting up in the morning. If it is, you need to change jobs.

What Determines The Absolute Value Of Median Salaries?

In a pure market economy, the pay for a job is determined by two related forces:

The pay for a job in a market economy is not determined by:

Even though intellectually it may seem "wrong" not to include some of the above factors, it makes us collectively economically poorer whenever pay deviates from the pay set by a purely market economy. However, in some cases our society has judged that the best society results from deviating in some cases from market pay, such as by establishing a "minimum wage" for all jobs, even at the cost of thereby decreasing the number of such jobs that are available by eliminating those jobs whose economic value is less than the minimum wage. The only area where there is such influence on salaries of engineers results from the subsidized education of engineers, designed to produce more engineers "which our society needs". This increased supply of engineers may then have a small effect in reducing salaries. (The demand for engineers caused by government work is of course a significant factor in overall demand for engineers, but here I consider that part of the overall market.)

Thus, for example, star baseball players get a lot of money because their effort generates tremendous revenues through television contracts for their clubs and through their advertising value in bringing in revenue for products they sponsor and because they are the top 100 or so out of over 250 million people. They have two things going for them: the economic value of their job is extremely high, and the supply of people that are the top 100 in that job is extremely low.

In general, engineers and scientists generate substantial economic value through their work and the supply of engineers and scientists is relatively low, since only a few percent of the population have the skills and desire to become engineers and are willing and able to undertake the training necessary for entry into the profession. (There are about 5 million scientists and engineers in the U.S.) Thus the median engineer salary of $60 - $70 K per year is roughly twice the median salary for all jobs in the U.S. of ~$30 K per year (see General Income Information). Engineers routinely get the highest starting salaries of any profession and some engineering fields are among the top five paying jobs.

The salary of engineers increased substantially during from 1957 to 1996 due to the increased demand coming from the space race of the 1960s, the ongoing technology boom in our society, the 1980s Reagan defense buildup, the commercial aerospace boom furthered by the 1980s deregulation, and exploding telecommunications needs. Thus in 1957, the average engineering salary of $6,695 per year was only 1.41 times the average salary of $4,753 per year (Statistical Abstract of the United States 1958), as compared to today's ratio of over 2. Thus we are currently living in the "Golden Age" for engineering salaries - enjoy it while we have it! If the factor of 1.41 still applied, median engineering salaries would be ~$40 K, instead of $60 - $70 K per year.

Special Case - Scientists Who Get Time To Do Their Own Research

In general, the salaries of scientists whose jobs allow them to do their own research, such as many faculty positions and some national lab positions, are lower than the salaries of scientists who must spend 100% of their time doing a job specified entirely by their employer. Two examples:

The reasons for these lower salaries should be obvious by now: research positions produce less direct economic benefit and the demand for these jobs is very high, which is the opposite of the situation for star baseball players! Although it may be the case that research in the end indirectly leads to economic benefits, no employer usually benefits from this with current income that can be used to pay the salary checks of such researchers. (The money has to come from somewhere for those paychecks....) An employer faced with the decision of hiring either a full time scientist or engineer for a job or a similar person who spends only 50% of his time on that job, will always pick the 100% person unless there is an absolute necessity to employ someone who does research or if the 50% scientist or engineer is at least twice as productive as the 100% person. Thus the supply of such jobs is low. And with huge demand for such positions (stories of 200 applications for a single job abound), classic supply and demand theory demands that the pay for such positions be significantly less.

The common appellation of "slave labor" to describe the employment of graduate students and postdocs is not accurate - those people (I was one of them!) have chosen to accept those positions rather than being in the regular labor force, for a variety of reasons. At any time, most of those people could get significantly higher-paying jobs, but chose not to do so because money is, after all, not everything.

Nobel-prize winners and similar stars are an important exception to this rule of lower salaries. Because their prestige results in definite economic benefits to the university that pays them, they command significantly higher salaries than engineers in industry, resulting primarily from competition between universities in attracting and keeping them. Most of us don't have to worry about being in that situation!

The Shape of the Salary Curve

The salary curve shows that engineers are paid more as they accumulate experience. This must be the case for two reasons:

Do not make the mistake of thinking of the salary curve as people of one generation versus people of another generation. For example, some people think that the salary of older engineers has been determined primarily by them enjoying special conditions during their career than experienced by younger engineers presently. Think of the same person simply advancing in their career with time, with their salaries being continuously determined by the marketplace.

The salary curve can indeed be temporarily distorted by temporary non-equilibrium conditions, such as happened during the early 1980s. At that time, high inflation caused a rapid adjustment of beginning salaries, but the salary of already-employed engineers took 3-4 years to adjust. Supervisors at JPL were routinely hiring fresh-outs at higher salaries than the supervisors were making at the time! Similar changes can occur when demand changes dramatically, usually caused by periods of recession or boom times in the overall economy. However, in the end, the overall salary curve will adjust to properly compensate for the value of experience.

The Magnitude Of The Increase Of Salaries With Experience

On average, very senior engineers are paid almost twice that salary of a "fresh-out" engineer. That increase of salaries with experience is the result of the very dynamic process of at least the following factors:

All the factors except the first one are quite complicated, but I believe that they are in fact much less important than the first one, which I will now discuss in detail.

When I hire programmers, software engineers and scientists to do a given task that we have obtained money to perform, I, like most other managers, go through the following steps. I break the task into its component tasks (a work breakdown structure). I then decide how many people, with which specialties and experience levels, I need to hire for the component tasks.

Some jobs usually require very experienced people, such as a system architect for a major software task. But many jobs can be filled by either experienced people or by junior people. It is the judgment of the marketplace that for most positions, in the aggregate, that you get the same results if you hire one very experienced person or two junior people.

If that equivalence were not obtained, in the aggregate, employers would change their hiring behavior. For example, if salaries for experienced people were to increase through administrative fiat at a particular company to be three times the salaries of fresh-outs, that company would find that the demand for experienced engineers, from the hiring supervisors at that company, would drop dramatically. Were I a supervisor in those circumstances, I personally would hire very few senior engineers, preferring to have three junior engineers on the job rather than a single senior engineer, assuming that all the engineers have the same capabilities except for experience, except for those positions where a senior engineer is required.

Sometimes the salaries of senior engineers at a particular company do indeed reach values above their "marketplace" value. This increase can happen not through any thought-out administrative fiat, but just as a result of the accumulation of raises throughout their lifetime, with no "reality checks" at any point. What happens to these people? If they are lucky, they are in a situation where they can't be fired and they continue to be employed, reaping the benefits of a salary higher than market. However, I have known more than a handful of such cases where the person is laid-off, or cannot find new employment after their current project ended. Most of these people then retire, even though they would rather keep on working.

There are three recent examples illustrating this phenomena:

Although I, like most scientists and engineers, am quite certain that I should be rewarded with above-average raises each year because, like probably at least 70% (99%?) of scientists and engineers, I am certain I am well above-average, I have to keep in mind that there is someone just like me available for hire, with my same capabilities except for the number of years of experience. If my salary gets out-of-whack compared to the salary of a younger version of myself, there's little doubt what will happen eventually: my boss will fire me and hire the younger version of me!

The Distribution Of Salaries At Constant Experience Levels

Some people have better athletic ability in a particular sport than others; some people are better mathematicians than others; and some people are better engineers, programmers and scientists than others. If the better ability of a person in a given job produces more economic output, the marketplace will reward that.

All managers know that the ratio between the best and worst performers is at least a factor of 2-3, and in specific cases for specific problems can easily reach a factor of 10. I refer here to tasks where those performers each have the skills to do the job, and the end product will be of acceptable quality from both performers. Frederick P. Brooks, Jr. quotes a study by Sackman, Erikson and Grant that measured performances of a group of experienced programmers:

      "Within just this group the ratios between the best and worst performances averaged about 10:1 on productivity measurements and an amazing 5:1 on program speed and space measurements!"
     The Mythical Man-Month, Addison-Wesley, 1982, p. 30.

What does the salary curve say the relative economic value is between such engineers? The ratio between the 90% salary and the 10% salary peaks at about 1.9:1.0 for experienced engineers. Why isn't this ratio closer to 5-10?

There are at least two factors that make the ratio closer to 2 than to 10:

  1. Engineers don't spend 100% of their time being as productive as measured in those experiments. The long-time columnist for the L.A. Times, Jack Smith, was fond of quoting that "reasonable" employers should only expect their employees to work hard about 50% of the time, with the rest of the time spent around the water-cooler, for example. In fact, perhaps no more than 10-20% of one's time is spent at the intensity measure by those productivity measurements. If one assumes that during that other 80-90% of the time the productivity ratio is 1.5, for example, the overall productivity ratio is then (2.4-3.2):1.

  2. In many cases, there is value in having multiple people with the same total cost compared to a single person. Travel and meeting with customers, etc. is obviously enhanced, and tasks can be accomplished simultaneously rather than sequentially. It is hard to imagine preferring to employ 1 super-engineer compared to 5-10 still competent engineers. I conjecture that this second factor, on top of the first factor, produces the observed marketplace ratio.

The ratio of the 90% salary for an engineer with 33 years experience to the 10% salary for a fresh-out engineer is 3.3:1. Thus the factors of experience and competence combine so that the marketplace judges that the experienced person in the top 10% is worth 3.3 fresh-outs, no matter what their GPA was.

How accurately are people placed onto a given percentile on the salary curve? Mistakes certainly occur, but are infrequent. See "Can You Get A 20% Raise By Changing Jobs?". If you are not as high up on the salary curve of your organization as you think you ought to be, you should talk to your supervisor to explore ways to increase your value to your organization.

Salary Curves on the Web

Here are some salary curves I have found on the web for scientists and engineers. In some cases, the links below will take you to the most recent salary surveys, later than the date quoted below.

These links from 1997 no longer work. If anyone knows the new location, please let me know.

Go To:

Copyright © 1997-1999 by Tom Chester.
Permission is freely granted to reproduce any or all of this page as long as credit is given to me at this source:
Comments and feedback: Tom Chester
Last update: 12 November 1997, with a single update to salary curves on the web on 13 August 1999.