Tuesday, June 9, 2009

SDM thesis asks: Where did all the engineers go? - SDM Pulse, Summer 2009

By Dan Sturtevant, SDM ’07

Dan Sturtevant
SDM ’07
According to the National Science Foundation, the percentage of US students earning bachelor’s degrees in engineering is almost half what it was in 1985, even though engineering and computer science graduates have significantly higher lifetime earnings than those with any other type of degree.

This situation, which appears to violate the basic laws of labor-market supply and demand, has raised major concerns among US-based engineering and information technology firms, which fear the United States is not producing enough engineers to replace retiring workers and meet demands for future growth.

At Boeing Corporation, for example, a majority of the company’s engineers will be eligible to retire in the next 10 years. I therefore teamed up with Boeing to analyze the US technology labor market for my System Design and Management (SDM) Program master’s thesis, titled “America Disrupted: Dynamics of the Technology Capability Crisis.”

I began by creating a system dynamics model to represent the institutional forces present in the real world. For example, I took into account supply and demand forces that dictate the wages of engineers and others with strong quantitative and analytical skills; the rate at which K-12 students acquire knowledge or fall behind in science and math at various education levels; and the relationship between teacher wages, alternative opportunities outside the classroom, and the quality of education.

My model simulates such societal forces simultaneously, tying them together in an artificial world starting in 1940. My goal was to reproduce what has actually happened in the US economy, labor force, and school systems over the past 70 years and then project what might happen in the future if current policies are maintained or new policies are implemented. By doing this it was possible to gain insight into nonlinear causes of societal problems that cross institutional boundaries and operate over extremely long time ranges.

By going through a system dynamics modeling process, it was possible to identify feedback relationships that
could have caused normal market mechanisms to turn against themselves, leading to systemic failure. It allowed me to formulate and test a theory that high industry pay for science, technology, engineering, and mathematics (STEM) workers and low pay for STEM K-12 teachers directly causes long-term industry labor shortages that are self-perpetuating. This is because scarcity of STEM workers causes industry wages to rise as employers bid up the price of those skills in the short-term. Schools are left with fewer qualified and lower quality teachers as the best people choose to go into industry.

If you increase the wages of engineers in the short term, the market may give you a few more engineers, but you also pull people out of teaching, making the shortage even worse 10 years down the road. Already, the shortage of science/math teachers is estimated to be 200,000 nationally and rising.

My research revealed that teaching quality in STEM subjects has indeed declined significantly over the years.
While it is impossible to track the abstract notion of “teacher quality” over time, proxies that correlate with it
can be tracked to pick up on observable trends. Many researchers tracking these proxies have stated that
teacher quality has declined significantly since the 1950s, experiencing the steepest drop during the 1970s and 1980s. Observable drops in undergraduate GPA, class rank, selectivity of undergraduate institution, and scores on standardized tests including the SAT, GRE, and ACT have all occurred over the course of this time period.

The model I created was able to reproduce the time-dependent behavior of multiple seemingly disjointed
historical trends. These include the steep decline in overall K-12 teacher quality that happened in the 1970s and 1980s, the level of STEM teacher shortages nationally, a worsening balance of trade, and a rise and then fall of engineering graduates from university after 1985. All of these things are causally linked in the model. What is shown is that societal shifts that occurred in the 1950s through 1980s could have caused the unfortunate behavior seen from 1985 until the present day.

After reproducing history and seeing where the future might lie (the model predicts a worsening of the problem over time), various policy proposals to correct the situation were simulated. The purpose was to test their ability to move the system in a better direction. The model was found to exhibit “tipping” behavior: some reforms had negligible impact while others moved the system into a fundamentally better pattern of behavior.

One of the approaches I tried was simply to peg teacher pay to what STEM graduates were capable of earning in the outside market. This “architectural” difference made the discrimination against highly desired skills go away [see chart below].

Although some policy reforms like this were able to move the system past a “tipping point” and cause it to operate in a fundamentally better way, making such a transition in the real world would take considerable investment in education, and the benefits would not be fully felt for many years.

That said, it took a considerable amount of neglect for a long period of time to get us into the current situation. It will take a long time to get out of it.

The combination of poor student performance, increasing math illiteracy at a societal level, high STEM industry salaries, and STEM labor shortages will necessarily lead to increased outsourcing and a worsening balance of trade. The societal demand for the output of STEM labor will not simply go unmet. There’s bound to be a temptation to blame foreign competition, but if we fail to remain globally competitive in high-tech it will be because we destroyed ourselves. The problems we now face were entirely self-made.

STEM Teacher Wages as a Fixed Percentage of STEM Industry Wages
Tying STEM teacher pay to industry pay made the system more responsive and stable than fixed wage
increases that are unrelated to industry pay. The following set of simulations tests various teacher wage
policies that fix teacher wages to some percentage of industry pay, but not necessarily equal to industry
pay. Pay levels that are tested are wages of 50%, 75%, 100%, 125%, and 150% of STEM industry pay.

Higher wage ratios increase the number of qualified STEM educated teachers faster. System response to this policy also exhibits tipping point behavior.

No comments:

Post a Comment