Math PhD Bubble: What % Get Tenure?

[hamster143]

I'm curious, is there a "Ph.D. bubble" in math that is similar to the one in physics? Specifically, what percentage of graduating math Ph.D.s end up getting tenure?

 

[Newtime]

What do you mean by "bubble"?

 

[98whbf]

I think the OP might be referring to there being more people being awarded with Ph.D's every year than there are jobs opening.

 

[Vanadium 50]

I think the OP might be referring to there being more people being awarded with Ph.D's every year than there are jobs opening.

Yes, but that's hardly a "bubble". A model where every faculty member has 5-10 students who then all graduate to become faculty members with 5-10 students is clearly unsustainable. It's simply not realistic to expect that a PhD means a faculty job at a research university. Most PhD's move on to industry - which, IMO, is a healthy sign: it means that a PhD has value to other parts of society.

Saying there is a bubble in physics is like saying there's a bubble in major league baseball players. Not everyone gets to be one, but that's not the same thing.

 

[hamster143]

Yes, but that's hardly a "bubble". A model where every faculty member has 5-10 students who then all graduate to become faculty members with 5-10 students is clearly unsustainable. It's simply not realistic to expect that a PhD means a faculty job at a research university. Most PhD's move on to industry - which, IMO, is a healthy sign: it means that a PhD has value to other parts of society.

I can't agree with that. The whole point of getting a PhD is to do research for a living. There's no law that says that every professor in a research university must teach 5 students during his lifetime. Does Grigory Perelman read lectures and train postgrad students? How many postgrad students could call Albert Einstein their primary advisor?

There's no immediate reason to conclude that a PhD in physics or math has any value whatsoever outside academia. If you have a system that produces 100,000 virtuoso violin players a year, but only 20,000 of them find jobs playing violin for a living, and others move on to unrelated jobs in industry - is that healthy, and does that indicate that violin mastery has value to other parts of society?

 

[f95toli]

There's no immediate reason to conclude that a PhD in physics or math has any value whatsoever outside academia.

That is simply not correct. One of the main reasons why governments are willing to fund the training of so many PhDs (via funding agencies) is because they are needed in industry. Where I did my PhD the vast majority of the PhD students had no intention of ever going into academia. Most of them ended up working in the telecom industry, some went into finance etc.
Even if you do your PhD working in an "exotic" field you still pick up a lot of useful skills. One large company which had an R&D facility near my university were actively recruiting PhD students that had worked in string theory, not because they were interested in Branes etc but because they had found that they were really good at calculating various properties of radar systems.

 

[D H]

I can't agree with that. The whole point of getting a PhD is to do research for a living.

No way! The whole point of getting a PhD is to open doors to interesting, challenging, rewarding, and stable jobs.

There's no immediate reason to conclude that a PhD in physics or math has any value whatsoever outside academia.

That governments and industry provide the funding needed to make it possible to produce many times more PhDs than are needed by academia belies your claim.

 

[hamster143]

That is simply not correct. One of the main reasons why governments are willing to fund the training of so many PhDs (via funding agencies) is because they are needed in industry. Where I did my PhD the vast majority of the PhD students had no intention of ever going into academia. Most of them ended up working in the telecom industry, some went into finance etc.
Even if you do your PhD working in an "exotic" field you still pick up a lot of useful skills. One large company which had an R&D facility near my university were actively recruiting PhD students that had worked in string theory, not because they were interested in Branes etc but because they had found that they were really good at calculating various properties of radar systems.

I see this as a big problem. This is like training lots of people to play violin, because the ones that succeed will be proven to have agile fingers and keen sense of hearing and thus could be retrained to do jobs that require either (or both), even if they don't touch the violin for the rest of their natural lives. It makes no sense to study string theory and then go calculate properties of radar systems for a living - it's better to get a 6 month course on radiolocation followed by hands-on training. It makes no sense to study condensed matter and then end up working quant in an investment bank - it's better to get a B.S. in finance, once again followed by hands-on training.

If you invest 5 years into a Ph.D. and perhaps 4 more years into a post-doc, you should spend that on knowledge and skills that you will need for the 30 years that follow.

My question is, is the same in math as it is in physics.

That governments and industry provide the funding needed to make it possible to produce many times more PhDs than are needed by academia belies your claim.

Governments provide non-directed science funding that is available to all qualified applicants to spend more or less as they please. NSF budget is X, the share of this budget that is allocated to physics is Y, and X and Y are determined by the state of national economy, by the whims of politicians, and by historical trends. Unlike, say, in medicine, where all med students are pretty much guaranteed employment in their primary field of study (because there's a centralized cartel called AMA that keeps a tight leash on the number of spots in med schools nationwide), no one ever bothers to count PhD's or tenured professors.

A further proof is that physics students have to be paid to work on their PhD's. if PhD had significant weight in the industry, we'd have people piling up into physics programs and taking on sizable student debt, like they do in medicine and law. Instead we have to provide free tuition, pay stipend, and even so we end up filling half the slots in physics grad programs with Chinese nationals (for whom the PhD carries an added non-monetary benefit - path towards US citizenship).

 

[f95toli]

I see this as a big problem. This is like training lots of people to play violin, because the ones that succeed will be proven to have agile fingers and keen sense of hearing and thus could be retrained to do jobs that require either (or both), even if they don't touch the violin for the rest of their natural lives. It makes no sense to study string theory and then go calculate properties of radar systems for a living - it's better to get a 6 month course on radiolocation followed by hands-on training.

The difference is that when you work as a PhD student you learn how to do research which is a skill in itself. Moreover, you also learn (and demonstrate the ability) how to plan your time etc for a long-term project; something you are usually unable to do in industry if you work as e.g. an engineer since R&D projects are usually much shorter then that.
Also, if one were to follow your line of reasoning to its logical conclucsion everyone that does a PhD in a certain field should then continue in that field. It is not at all unusual for people to move to a complettely different field of reseach after a PhD even if they stay in academia. In the research group I belong to there are people who did PhDs working neutron scattering, astrophysics, theoretical optics, semiconductor physics etc. I even used to have a colleague who had PhD in chemistry (who was orignally hired as a chemist, but "drifted" over to physics by working on carbon nanotubes).
We all work more or less on the same things. Some of them also work in two very different fields at the same time; e.g. our in-house expert on STM also works on optical experiments (frequency references).

Governments provide non-directed science funding that is available to all qualified applicants to spend more or less as they please. NSF budget is X, the share of this budget that is allocated to physics is Y, and X and Y are determined by the state of national economy, by the whims of politicians, and by historical trends.

This is not always true. The systems differ from country to country, but when you write a grant applications you usually have to specificy if you plan to hire a PhD student.
It is also not uncommon for governements to direct money specifically to the training of PhDs. This is what happened in my case, there were a lot of PhD students starting at the roughly same time as me; the reason being that the funding for students had gone up that year as the result of the telecom companies complaining about the lack of PhDs.

A further proof is that physics students have to be paid to work on their PhD's. if PhD had significant weight in the industry, we'd have people piling up into physics programs and taking on sizable student debt

I think the main reason for this is that the starting salaries for people with PhDs when they go into industry is usually not that much higher than for someone with an MSc. But having a PhD can certainly "pay off" later on in you career, most of the people I meet who e.g. manage R&D projects in industry are PhDs. But yes, it is true that for you as an individual it might not make financial sense to do a PhD, but that does not change the fact that society as a whole (not to mention companies) can benefit.

 

[f95toli]

It makes no sense to study string theory and then go calculate properties of radar systems for a living - it's better to get a 6 month course on radiolocation followed by hands-on training.

I guess I should address this point more specifially. I don't know exactly what these guys are working on (military radar, most of it is classified) but I am quite sure it is something that involves a LOT of that math. There is no way someone with just a BS would be able to do their job. Presumably the same place also hire people with e.g. a PhD in math, although I imagine some knowledge of electromagnetics and physics in general might come in handy in this case.

I think the same thing applies to quants. The reason why people with PhDs in physics or math end up in that job is that we are the only ones that know enough about math and modelling. There is no way someone who has only graduated from busines school have learned enough math.

 

[twofish-quant]

I can't agree with that. The whole point of getting a PhD is to do research for a living. There's no law that says that every professor in a research university must teach 5 students during his lifetime.

Yes there is. If you don't have teaching and research assistants, you don't get funding. If you don't get funding then your salary doesn't get paid. Both the people that you mention (Einstein and Perelman) are funded by systems that are very, very different from the system that most university professors get funded.

I wouldn't say it's a "bubble" as much as a "ponzi scheme."

There's no immediate reason to conclude that a PhD in physics or math has any value whatsoever outside academia.

But there *is*. One reason I think that Ph.D.'s need to be encouraged to look at life outside the university is that the world becomes a much nicer place, once you realize that academia is not the only option.

 

[twofish-quant]

It makes no sense to study string theory and then go calculate properties of radar systems for a living - it's better to get a 6 month course on radiolocation followed by hands-on training. It makes no sense to study condensed matter and then end up working quant in an investment bank - it's better to get a B.S. in finance, once again followed by hands-on training.

No it's not. People with bachelors in finance just don't have the mathematical or research skills to do quant work. One thing that makes the Ph.D. different from the bachelors is that you are *creating* knowledge. I have no clue what mathematical techniques are going to be used in finance next year, neither does anyone else. Ph.D.'s are used to situation where there isn't a textbook. Most people with bachelors are not.

The fun thing about working in an investment bank is that it's like graduate school. Only with more money.

If you invest 5 years into a Ph.D. and perhaps 4 more years into a post-doc, you should spend that on knowledge and skills that you will need for the 30 years that follow.

Which is why jobs in investment banks or medical imaging or whatever are perfect. I'm basically writing the same sorts of programs that I am in Wall Street that I did in graduate school. Only with more money.

 

[kote]

The situation with philosophy PhDs can shed some light on this I think. Philosophy programs graduate PhDs in proportions similar to those of physics programs. Of the 50% who don't drop out, most never get jobs in academia. If they get jobs in academia the pay is equivalent to undergrad engineering pay. If they don't get jobs in academia the degree itself is worthless.

There is an institutional problem in academia that is causing an oversupply of PhDs. PhD students are basically free labor. Schools want as many as possible. Professors are relatively expensive labor. Schools want as few as possible. The demand for graduate degrees is inflated for a number of reasons, one of which is the fact that students' advisors for the most part have only ever known academia and going that route worked out for them. The supply of graduate positions is driven by how cost effective they are for the institution. Notice at no point has the good of society or the student been mentioned. It's simple economics and it's uniform across specializations that are and are not useful in the work force.

 

[twofish-quant]

Also, if one were to follow your line of reasoning to its logical conclucsion everyone that does a PhD in a certain field should then continue in that field. It is not at all unusual for people to move to a complettely different field of reseach after a PhD even if they stay in academia.

Also the fields aren't quite is different as they might seem. There are only so many ways of writing a parabolic partial differential equation, so if you get good at simulating parabolic PDE
s with neutrino diffusion, and a bank needs someone to simulate a parabolic PDE involving stock option pricing...

 

[twofish-quant]

People with master's degrees generally do know enough math to do modeling, however where people with master's degrees can run into trouble is if the rules change if it turns out that you need to teach yourself some math that no one has ever taught you.

 

[twofish-quant]

There is an institutional problem in academia that is causing an oversupply of PhDs.

I really don't think that society has an oversupply of Ph.D.'s. There is a huge problem in that Ph.D.'s are being taught that the "normal" thing for a Ph.D. to do is to get a job in academia, but that's something different.

If you look at unemployment rates for Ph.D.'s. I don't know of any physics/math Ph.D. that has had any problem looking for a job, once they looked outside the academy.

Notice at no point has the good of society or the student been mentioned. It's simple economics and it's uniform across specializations that are and are not useful in the work force.

The problem is that if the solution is "make society less educated" its probably not going to cause good things to happen. The only real solution I can see is to make Ph.D. students and their advisors more aware of life outside the academia.

What I think is going to be really interesting is that if you have more Ph.D.'s outside the academia than inside, this will likely change the power structures within the academia.

 

[kote]

I really don't think that society has an oversupply of Ph.D.'s. There is a huge problem in that Ph.D.'s are being taught that the "normal" thing for a Ph.D. to do is to get a job in academia, but that's something different.

It's a relative thing. Only a few hundred people even apply to philosophy phd programs each year, so you don't have a ton of unemployed philosophers wandering around everywhere. I was just pointing out that the same factors that are driving the definite oversupply in philosophy are driving the numbers in other majors. It's an institutional business decision that is only in the interest of the school. In many majors this works out for the students too, but not in all.

There is no different path for philosophy phds. It's academia or trying to write a novel. Yet state branch college X that placed one student in a community college in the past decade isn't going to scrap their program because they are underfunded and desperate for TAs and status. The reason I blame the schools and not the students in these cases is the wholly inadequate advising of undergraduates when it comes to choice of major or career path.

 

[kote]

It's also interesting to take a look at a non-academic focused environment - med school. In academia you have an oversupply of cheap labor keeping salaries low and competition high. In the medical profession where students aren't that valuable to the school and training them is expensive, you have an artificial undersupply of graduates that keeps salaries and employment high.

Even looking at business school, assistant professors of accounting make over twice the salary that new philosophy professors make. Students actually realize that they have other options when they are considering business school and you don't get the oversupply that you do in other fields. What students going to top philosophy programs, for example, don't realize, is that on average they are just as competitive on the job market as students applying to business schools - even more so if you go by GPA and GRE scores.

Edit: For the record I'm not a bitter philosophy grad student or anything, but I considered it and did my research . I'm not as familiar with other concentrations. I know philosophy and physics are similar in many ways, so I think it works.

 

[98whbf]

As a philosophy major with plans on graduate school and no desire to enter the business world this thread makes me very uneasy.

 

[kote]

As a philosophy major with plans on graduate school and no desire to enter the business world this thread makes me very uneasy.

http://www.philosophicalgourmet.com/perspective.asp
http://leiterreports.typepad.com/blog/advice_for_academic_job_seekers/
http://aidan.mcglynn.googlepages.com/adviceforwannabephilosophers
http://lemmingsblog.blogspot.com/2007/01/should-i-become-professional.html

 

[98whbf]

Thanks, I'd read the first one before but the others are new to me.

 

[twofish-quant]

As a philosophy major with plans on graduate school and no desire to enter the business world this thread makes me very uneasy.

I think it's sort of a shame, since one of the purposes of philosophy is to answer the questions "why?" What should be the goal of academia? How should academia interact with the business world? How does academia interact with the business world?

I've found that reading philosophy was pretty critical in my staying somewhat sane. One thing that I figured out was that academia is a business, and when people start talking about "job markets" then you by definition are dealing with the world of business.

 

[hamster143]

Yes there is. If you don't have teaching and research assistants, you don't get funding. If you don't get funding then your salary doesn't get paid. Both the people that you mention (Einstein and Perelman) are funded by systems that are very, very different from the system that most university professors get funded.

I wouldn't say it's a "bubble" as much as a "ponzi scheme."

Right, the system is set up in such a way that one side effect is overproduction of Ph.D.s. Since Ph.D.s don't really bring any money into the system (NSF & other federal funding excluded), it is perfectly possible to produce fewer Ph.D.s and employ more research professors on the same budget. Since an experienced full-time researcher is probably several times more productive than a typical Ph.D. who spends most his time acquiring knowledge that the researcher already possesses, overall quality of physics research would go up.

Ponzi scheme is a good description.

No it's not. People with bachelors in finance just don't have the mathematical or research skills to do quant work. One thing that makes the Ph.D. different from the bachelors is that you are *creating* knowledge. I have no clue what mathematical techniques are going to be used in finance next year, neither does anyone else. Ph.D.'s are used to situation where there isn't a textbook. Most people with bachelors are not.

It's still a massive misallocation of resources to have people study things they won't need in real life. Science has gotten a lot more complex in the last century and people need to know a lot more just to do basic research. Which is why real science nowadays is done by people in their 40's and 50's, and people like Witten (58) or Georgi (62) are still very much on top. A string theory Ph.D. would spend 2 years just learning basic tools of the trade up to and including QFT, and then 2 more years getting a partial picture of the more recent developments, which leaves him one year to write a thesis, i.e. "create knowledge". And then he gets a job in an investment bank, learns a lot of new stuff, and forgets all about QFT and string theory.

Someone who gets a Ph.D. in finance instead, studying financial tools and models and trying to create knowledge there, would probably become a better quant in less time than someone whose brain is loaded with conformal field theory and other esoteric stuff with narrow scope of application.

<off-topic>
Out of curiosity, is there still demand for quants on Wall Street nowadays? I'd have expected that field to shrink significantly.

 

[twofish-quant]

Right, the system is set up in such a way that one side effect is overproduction of Ph.D.s. Since Ph.D.s don't really bring any money into the system (NSF & other federal funding excluded), it is perfectly possible to produce fewer Ph.D.s and employ more research professors on the same budget.

Also there is student tuition which requires massive number of teaching assistants. I really don't think its possible to have fewer Ph.D.'s and more research professors since you end up needing a lot of graduate students to do grunt work. A lot of astrophysics research (and for that matter financial programming) involves staring at one hundred thousand lines of source code trying to figure out why energy doesn't balance. For this type of work, you need lots of highly skilled warm bodies.

A string theory Ph.D. would spend 2 years just learning basic tools of the trade up to and including QFT, and then 2 more years getting a partial picture of the more recent developments, which leaves him one year to write a thesis, i.e. "create knowledge". And then he gets a job in an investment bank, learns a lot of new stuff, and forgets all about QFT and string theory.

Most physicists aren't string theorists. Also QFT and string theory are like riding a bicycle, you don't forget how it works once you learn it, and presumably if you can do QFT, then someone can give you a book on stochastic differential equations which you can read over the weekend.

Someone who gets a Ph.D. in finance instead, studying financial tools and models and trying to create knowledge there, would probably become a better quant in less time than someone whose brain is loaded with conformal field theory and other esoteric stuff with narrow scope of application.

Conformal field theory really has less a narrow scope of application than you think.

Also with some exceptions, the quantitative research divisions of most investment banks just don't hire finance Ph.D.'s because they don't have either the mathematical or the computationu canal experience to run the systems. If you can handle QFT, then presumably you can handle stochastic differential equations and mortgage default models.

Also you are in a better position to get a position as a quant nowadays if you do something very heavily numerical. Condensed matter and astrophysics is more likely to land you a job at a IB than string theory. If you've done computational fluid dynamics, this is a big, big positive. The problem with most finance Ph.D.'s is that very few of them have ever programmed a supercomputer, which is a problem when it comes to running the grid that's at the heart of every investment bank or major hedge fund.

There are IB jobs for which do fit what finance Ph.D.'s do (portfolio management and algorithmic trading), there are also a set of jobs in finance that happen to hit skills that physics Ph.d.'s have. One other thing is that physics and math Ph.D.'s are willing to work relatively cheap. You give a physics/math Ph.D. a $160K job, and they will kiss your feet whereas most finance Ph.D.'s from big name schools will laugh at such small piddling sums.

Out of curiosity, is there still the job market for quants on Wall Street nowadays? I'd have expected that field to shrink significantly.

This year has been bad, but hiring is picked up enormously in the last two months. Personally, I think there is going to be a huge number of openings because you now have regulators that are totally paranoid about risk management and counterparty default.

 

[hamster143]

Also there is student tuition which requires massive number of teaching assistants. I really don't think its possible to have fewer Ph.D.'s and more research professors since you end up needing a lot of graduate students to do grunt work. A lot of astrophysics research (and for that matter financial programming) involves staring at one hundred thousand lines of source code trying to figure out why energy doesn't balance. For this type of work, you need lots of highly skilled warm bodies.

I doubt that many Ph.D.s do teaching assistance. For starters, you can't be a good TA unless you're a native English speaker (the best you can do is grade homework). When I was in grad school (a top 40 university), less than half of all students in my class were native speakers. A shift towards 1:1 student-teacher ratio would involve more grunt work for professors, but that's all for the better.

Also with some exceptions, the quantitative research divisions of most investment banks just don't hire finance Ph.D.'s because they don't have either the mathematical or the computationu canal experience to run the systems. If you can handle QFT, then presumably you can handle stochastic differential equations and mortgage default models... The problem with most finance Ph.D.'s is that very few of them have ever programmed a supercomputer, which is a problem when it comes to running the grid that's at the heart of every investment bank or major hedge fund.

That's an interesting perspective, though a lot of it seems to reflect deficiencies in finance Ph.D. training rather than the inherent value of physics Ph.D.

One other thing is that physics and math Ph.D.'s are willing to work relatively cheap. You give a physics/math Ph.D. a $160K job, and they will kiss your feet whereas most finance Ph.D.'s from big name schools will laugh at such small piddling sums.

Physics/math Ph.D.'s do tend to be a little out of touch with the real world. Eventually they'll realize that $160K in Manhattan for a married couple with children buys worse quality of life than, say, 100K in Charlotte...

 

[twofish-quant]

I doubt that many Ph.D.s do teaching assistance. For starters, you can't be a good TA unless you're a native English speaker (the best you can do is grade homework).

So grade homework. It has to be done. The problem with academia is that people are so concerned with goddamn stars that no one sees the grunt work that has to be done to run a department.

A shift towards 1:1 student-teacher ratio would involve more grunt work for professors, but that's all for the better.

You're talking about something that's economically unsustainable at this point.

That's an interesting perspective, though a lot of it seems to reflect deficiencies in finance Ph.D. training rather than the inherent value of physics Ph.D.

My perspective comes from the fact that I work in an investment bank. There are about one hundred or so physics and math Ph.D's working here. I don't know of any finance Ph.D.'s, but rumor has it that they exist somewhere. It all makes sense if you see a physics degree as a degree in numerical modelling.

Physics/math Ph.D.'s do tend to be a little out of touch with the real world. Eventually they'll realize that $160K in Manhattan for a married couple with children buys worse quality of life than, say, 100K in Charlotte...

Well no one I know with $160K has a house in Manhattan, everyone lives in either New Jersey or Queens. Also you can work in Manhattan and fly to Charlotte on the weekends. Also $160K is something of a basic salary for a quant. There is a long tail for how much money you can make. If you have a physics Ph.D. there's probably a 1 in 50 chance you'll end up making $1M/year in finance, but that's higher odds than in any other field.

 

[hamster143]

Well no one I know with $160K has a house in Manhattan, everyone lives in either New Jersey or Queens. Also you can work in Manhattan and fly to Charlotte on the weekends.

As far as I know, there are basically no good public schools in Manhattan, which is why people end up commuting to Manhattan from NJ and White Plains and other remote places, 40 minutes one way ... whereas you could live in Charlotte or Portland or some other backwater city, have a job as an engineer that pays 100k, and live in a big house on a half acre lot within a 15 minute drive from work.

If you're going to live in Charlotte, rent a studio in Manhattan, and fly back and forth four times a month, that might cost more than the difference between 100k and 160k after-tax. Ditto sending kids to a private school in NY ...

Also $160K is something of a basic salary for a quant. There is a long tail for how much money you can make. If you have a physics Ph.D. there's probably a 1 in 50 chance you'll end up making $1M/year in finance, but that's higher odds than in any other field.

Hmm, maybe I should send around some resumes ...

 

[twofish-quant]

As far as I know, there are basically no good public schools in Manhattan, which is why people end up commuting to Manhattan from NJ and White Plains and other remote places, 40 minutes one way ...

Manhattan has some decent public schools, but you are not going to be living in Manhattan with $160K/year. Upstate is also pretty much out of reach.

If you're going to live in Charlotte, rent a studio in Manhattan, and fly back and forth four times a month, that might cost more than the difference between 100k and 160k after-tax.

Rent the studio in some very low rent part of NYC ($1000). Plane tickets are $1200/month, with a 40% marginal tax rate, that works out to about $37000. You're still $20K ahead. At that point it's a quality of life/career advancement issue.

whereas you could live in Charlotte or Portland or some other backwater city, have a job as an engineer that pays 100k, and live in a big house on a half acre lot within a 15 minute drive from work.

You could, but you'll pretty quickly hit the glass ceiling, at which point you get the depressing "is this all there is to the world" feeling. There is a glass ceiling in finance, but it's much, much higher than in other fields. Charlotte and Portland are nice quiet, pleasant cities, and if your goal in life is to lead a nice, quiet pleasant, stress-free life, then you probably don't want to end up anywhere near NYC or banking.

Both physics departments and investment banks are massive exploitative bureaucratic hierarchies in which the people at the top leach off the work of the people at the bottom leaving only crumbs for the people doing the grunt work that keeps the system going. The difference is that the amounts of money in finance is so much larger, than you can be ruthlessly exploited and still end up leaving a pretty nice life.

 

[Locrian]

Most PhD's move on to industry - which, IMO, is a healthy sign: it means that a PhD has value to other parts of society.

I don't think you can actually back this statement up. All you can say is that most don't get into academia, and some move to industry.

Nobody really knows the proportion of the people who get a PhD that move to industry.

 

[kote]

I don't think you can actually back this statement up. All you can say is that most don't get into academia, and some move to industry.

Nobody really knows the proportion of the people who get a PhD that move to industry.

http://www.aip.org/statistics/trends/highlite/emp3/table3.htm

60% of the potentially permanent positions taken in 05 / 06 were in industry if I'm reading this correctly.

You could also count the number of professors and compare to the number of phds.

Also many schools will have data like this from the princeton site: http://gradschool.princeton.edu/about/docs/ratestable/tablea/PHY.pdf - 2 phds out of 91 made it directly to research universities in the past 5 reported graduating classes from princeton.