Arnold Kling  

Economics vs. Computer Science

PRINT
A Signaling Model of Consumer ... Thoughtful Stuff from Steven P...

Nick Schulz writes,


The number of new computer science majors today has fallen by half since 2000, according to the Higher Education Research Institute at UCLA. Merrilea Mayo, director of the Government-University-Industry Research Roundtable at the National Academies, says the drop-off was particularly pronounced among women.

Meanwhile, elite schools are reporting that the number of economics majors is exploding. For the 2003–2004 academic year, the number of economics degrees granted by U.S. colleges and universities increased 40 percent from five years previously.


This is not necessarily a bad thing. It could be that in the past, in order to develop a useful computer-based product, you needed to be a computer science major. However, as computers have become easier to use and more generic, you do not need to be a computer science major to develop important applications. Think of Google hiring Hal Varian and a team of economist/quants to try to optimize its marketing by making better use of its data.

Yes, we still need cutting-edge innovation in computer science, but that is a function of really top-flight people, well into the 99th percentile of the distribution of intelligence. College students who are between the 80th and the 99th percentile might do more good using the stuff than attempting to help out on the bleeding edge.

In general, the role of the diffusion and utilization of innovation tends to be under-stressed. In an email, Nick sent me a link to a paper by one of my favorite economists, Amar Bhide. The paper argues that venturesome consumers are an important component of the ecosystem of innovation. He suggests that this means that the United States will tend to benefit the most from innovation, regardless of where the innovation is first developed.

It could be that majoring in economics will turn students into the venturesome consumers that the computer industry needs. It might not need an army of less-than-stellar computer scientists.


Comments and Sharing





COMMENTS (19 to date)
Dennis Mangan writes:

Gee, you don't think that the massive numbers of immigrant computer engineers willing to work for $30K a year has anything to do with Americans not wanting to waste a college education, do you?

General Specific writes:

"This is not necessarily a bad thing. It could be that in the past, in order to develop a useful computer-based product, you needed to be a computer science major. However, as computers have become easier to use and more generic, you do not need to be a computer science major to develop important applications."

The above statement is a major oversimplification, if not outright wrong. Embedded systems--which make up the significant portion of computer applications--continue to require an understanding of computer science concepts and principles, both at the systems engineering, software engineering, and test engineering phases. Little to nothing has changed to reduce the need for software engineers, and in particular the complexity of the systems has increased the need for skilled professionals, many of which are hired from India because they are not available in the United States.

True, there are higher level applications--e.g. web sites and the likes--which can be assembled by people with less skills than in the past, not unlike the ability of someone with little or no computer science background to use Microsoft Word, in contrast to the complexity of older word processing languages (e.g. those that ran on Unix in years past).

But even web sites must be created with sound software engineering practices in terms of sizing, efficiency, reliability, etc.

The appropriate skills won't be acquired in an economics curriculum.

John Thacker writes:

Gee, you don't think that the massive numbers of immigrant computer engineers willing to work for $30K a year has anything to do with Americans not wanting to waste a college education, do you?

No, because immigrants perform all sorts of jobs, not just computer engineering. You'll find that sort of griping anywhere, but it has little to do with which majors are chosen.

I obtained my undergraduate degree in 2001; it was clear at the time that plenty of people who had little love or aptitude for computer science were majoring in the subject because it was the hot thing and the supposed sure path to riches. It's entirely unsurprising that after the tech bubble the numbers decreased.

Many of those people graduated with majors in CS and went on to consulting or Wall Street. Now they're graduating with majors in economics and going on to consulting and Wall Street.

Kimmitt writes:

I dunno; my dad graduated with a Physics degree and went into CS. My uncle graduated with a music degree and went into CS. I graduated with a Poli Sci degree and went into IT (and probably should have stayed there).

Maybe it's because we're in a slump and the piece of paper is the thing -- and, I can tell you from experience, an Econ BA is far, far easier to acquire than a CS BA.

mjh writes:

I wonder how much open source plays a role in this. Linus Torvalds did more by writing his own OS and sharing it with anyone who had the chops to understand it, than my degree in Comp Sci did for me.

Maybe people optimize their situation by
1. Learning comp sci in their spare time and getting passing grades through peer acceptance of their code, and
2. Dedicating their degree to something else where an open forum isn't likely to provide the credentials they need

I don't know, of course, I'm just speculating. I'm pretty sure that if I were an undergraduate now, I would probably at least consider the above.

Josh writes:

Modern day computer programming and IT (which are what non-PhD CSers will do) should require no more skill or education that being an auto-mechanic. If cars were just being invented today, I have no doubt that auto-mechanics would mostly have college degrees in it. But as we see, a degree is totally unnecessary for them. Which is the same reason we feel fine outsourcing our computer programming to low-cost Indians, and why college degrees in this area are dropping.

General Specific writes:

The software engineering departments of cutting edge companies here in the US are largely housed with people having computer science degrees, with a few having mathematics, physics, chemistry, or--less commonly--linguistics, cognitive science, or another field involving programming. Few if any economists, who tend to inhabit the accounting side of the house (and to their misfortune, stock options in technology tend to be dispersed on the technical, not the business side of the house, except for the upper echelon of business folks).

Most foreign students hired in computer science received their degree at a US university, having studied computer science.

The idea that a bunch of hackers are creating most of he software that is used on a day-to-day basis is not true.

The analogy with auto mechanics largely holds for the IT service function, not the bulk of the software engineers producing software.

Dezakin writes:

Gee, you don't think that the massive numbers of immigrant computer engineers willing to work for $30K a year has anything to do with Americans not wanting to waste a college education, do you?
Working in this industry, if you're willing to work for $30k a year its because you haven't got the aptitude to work on a McDonalds register. Skilled immigrants aren't depressing IT wages in the least, and anyone pushed out because of the bubble popping really shouldn't have been hired to begin with.

Dan Weber writes:

Let me say upfront that I have a Master's Degree in EECS from MIT. To address several points in turn:

- The influx of "$30K programmers" scared me at first, but the threat never became reality. I haven't really seen a $30K programmer in person, either. My career has done fine in the past few years.

- There's a hell of a lot of computer work that is suited for a mechanic-level education.

- There's a hell of a lot of computer work that is suited for a PhD-level education.

- There is no lump of labor in computer science -- there will be no point that we've written all the software that's needed. Even if "Indian programmers" were doing a bunch of stuff at cut-rate prices, those skilled in the trade will be able to make a very good living. You could substitute "open source" in the above scare-quotes with minimal change.

That is, the fact that unskilled people can build some amazing things doesn't mean that there isn't tremendous opportunity for the highly-skilled individuals.

General Specific writes:

Dan Weber wrote: "That is, the fact that unskilled people can build some amazing things doesn't mean that there isn't tremendous opportunity for the highly-skilled individuals."

Exactly. Much work in computer science is problem solving of the most intense order (e.g. trying to find a subtle timing issue in 500,000 lines of code running on embedded processor communicating with three DSP chips and five peripheral devices, etc etc), all of which are communicating with fifteeen other devices on other boards. Very complex stuff and it takes a bright problem solver to set up the conditions in which the problem can be found--let along the person who can actually create a program that will not introduce the problem in the first place.

Brad Hutchings writes:

Arnold, I don't think any computer science degree holders are qualified to respond to your post... except maybe me. The reason fewer CS types are needed is that there are systems that hide the hard CS algorithms and data structures from application developers. With object oriented programming and relational databases mature paradigms, there isn't much sciency stuff that needs to be done to build today's applications. When stuff just works, you don't need a CS education to put it together. The value of the CS education is knowing how to make things work when they don't. It's funny, but most of the tools I learned with in coursework back in the day didn't actually work as advertised. I figured it was an intentionally cruel joke on the students who couldn't drive outside the lines to get stuff done, but it turns out the professors never actually test drove the lab assignments.

General Specific writes:

"there are systems that hide the hard CS algorithms and data structures from application developers. With object oriented programming and relational databases mature paradigms, there isn't much sciency stuff that needs to be done to build today's applications."

The above statement is misleading. Most software applications are and always have been embedded. Automobiles, ipods, telephones, etc. Higher level applications--particularly simple ones--can be put together with high level tools. But real-time embedded systems--whether satellite systems or automobile braking and control systems--almost always push the ability of memory and hardware speed--and are always subject to subtle conditions that require deep thinking skills at the requirements, design, and test stages. In addition, the software has to take into account the peculiar nature of the ASICS that are often designed in tandem with the standard off the shelf components.

There is no cut and paste, building block approach to software for most systems--even today. Complexity is the rule--not simplicity.

Lord writes:

Looking at the consumer reviews of ATSC HD DVRs, they are all too venturesome. With much hardware being developed abroad, there isn't a lot of need for them here.

General Specific writes:

Department of labor job outlook for software engineers:

Computer software engineers are projected to be one of the fastest-growing occupations from 2004 to 2014. Rapid employment growth in the computer systems design and related services industry, which employs the greatest number of computer software engineers, should result in very good opportunities for those college graduates with at least a bachelor’s degree in computer engineering or computer science and practical experience working with computers. Employers will continue to seek computer professionals with strong programming, systems analysis, interpersonal, and business skills. With the software industry beginning to mature, however, and with routine software engineering work being increasingly outsourced overseas, job growth will not be as rapid as during the previous decade.

Employment of computer software engineers is expected to increase much faster than the average for all occupations, as businesses and other organizations adopt and integrate new technologies and seek to maximize the efficiency of their computer systems.

Lord writes:

Employment of computer software engineers is expected to increase much faster than the average for all occupations

This is yet to be the case as it hasn't from 2004-2007 and in fact was flat over most of this period and only reached average rates recently. Still time though as things have started to pick up.

General Specific writes:

Lord: I take these numbers and projections with an extremely large grain of salt. But my main point is that there are nicely paying--and intellectually challenging and satisfying--jobs out there for software engineers and people with technical and thinking skills that allow them to solve problems in a software environment.

Lord writes:

Yes, for these job projections to work out there would have to be substantial real wage increases in the field but there is little indication the work is really that valuable and all industry is focused on is lowering costs. I doubt we will see it grow much above trend.

Yuecel E. writes:

IT is the key for successful business. Just think about recent developments such as Service Oriented Architecture, which enable the connection of the various divisions of an organization to communicate with each other in an efficient way by using middleware technologies such as database management systems.

From my point of view, the shift to economic majors is the result of the excessive offer in computer science majors in the job market, and as we all know from economics, if supply is bigger than demand, the excess of supply will result in a depletion of the offered product.

I am pretty sure, that in a few years, we will see that the Higher Education Research Institute will also observe the same behaviour among economic degrees.

Finally I would like to add, that “Computer Science” is a generalized name for all the skills and knowledge related with computers, e.g. network administration, database management, programming, etc. Without appropriate computer science knowledge, it is hard to contribute to innovation in business. Of course, that does not mean, that you have to be a certified programmer, system administrator or database manager, but general knowledge of IT combined with economic and business knowledge is the key for innovation and successful business.

Sean writes:

To claim that computer science can be picked up on the side is no different than claiming that economics can be picked up on the side. Anything can be picked up on the side, including theoretical physics. However, the depth of understanding and the skill level usually suffers.

This represents a fundamental confusion between programming and computer science. The same difference exists between mechanics and automotive engineers. The former can pick up the skills to fix cars but not build them while the latter requires advanced degrees. Programming and computer science are no different.

People wonder why a lot of software is full of bugs or is difficult to use. This confusion is the reason.

Comments for this entry have been closed
Return to top