27 April 2009

GDP, government spending and your tax dollars

How easy it is to commit your money to something when you are the President, no?

I had watched a bit of President Obama wanting to commit 3% of GDP to 'research and development' and really started to shake my head... and looked at Jonah Goldberg's post on it at NRO and responses by other readers, with one by a retired scientist, and then noting Rand Simberg's take on it. Mr. Simberg links to this AP article (which may or may not be transient) by Randolphe E. Schmid, thus getting this from the President:

"I believe it is not in our character, American character, to follow — but to lead. And it is time for us to lead once again. I am here today to set this goal: we will devote more than 3 percent of our gross domestic product to research and development," Obama said in a speech at the annual meeting of the National Academy of Sciences.

That 3 percent would amount to about $420 billion.

"We will not just meet but we will exceed the level achieved at the height of the space race," he said.

[..]

Obama said he plans to double the budget of key science agencies over a decade, including the National Science Foundation, Department of Energy Office of Science and the National Institutes of Standards and Technology.

He also announced the launch of the Advanced Research Projects Agency-Energy. It is a new Department of Energy organization modeled after the Defense Advanced Research Projects Agency, that led in development of the Internet, stealth aircraft and other technological breakthroughs.

And he said the Energy Department and the National Science Foundation will offer programs and scholarships to encourage American students to pursue careers in science, engineering and business related to clean energy.

And so we go around to the other side of things, as I was once IN an Advanced R&D office in the DoD.

I know smoke when I see it.

Lets start looking at the NSF FY'09 budget.

We have: Net Budget Authority: $5.594 billion

Civilian employment budget: $1.320 billion

Or about 24% of the budget goes to paying its workforce. This is the 'burdened' cost with all the perks (retirement in-kind payments, health coverage, communications, goods and services, etc.) of pay to the workforce.

Actual pay out of that? $0.154 billion.

Now this leaves out -

Internal education and HR: $0.890 billion

The auditing group: $0.064 billion

Inspector General: $0.013 billion

Now we are up to $2.287 billion for totally running the organization, including all perks and such.

About 41% to all of the government side of things: personnel, accountability, education, HR, etc.

That 41% is characterized as Operations & Maintenance, Employment (including perks and retention), plus Oversight.

For every $1 that goes into NSF exactly 59 cents gets spend on external work. Now the figures for Grants and such are high, and yet so much is going to the government, how can that be?

The answer: overhead cost.

The amount put up for Grants, subsidies, etc. is 'burdened' with the cost of the organization itself. Every dollar going to NSF, or any other government organization is a 'burdened' dollar, so that it can be spent for all the necessary upkeep to make it effective. And, believe it or not, 41% is damned good! I got to know some of the top players for spending funds and who was good and who wasn't and let me say that the above analysis is no knock on NSF. They are a model of government efficiency.

Now consider that Bell Labs (gone but not forgotten!), in 1982, had a budget of $1.6 billion (in 2009 about $3.53 billion using the BLS CPI adjuster) and employed 22,500 people (Source: Time Magazine 25 JAN 1982).

Number for NSF? Pretty constant around 1,100 (Source: Best Places to Work).

Now as NSF gets to spread the money around to teams, universities, sponsored projects, joint working relationships and so on, the numbers get a bit more difficult to work with. That said, although Bell Labs did a lot of that, there is no easy way to break out its budgetary numbers (although I've only taken a cursory look for same, not an in-depth one). That said, Bell Labs paid scientists DIRECTLY, not through intermediaries. Is it possible that all Grants, subsidies, etc. taken as an overall average have 22 or so people working on them?

I've helped run moderate sized R&D programs ($3-5 million Congressionally Directed Action) and I have problems remembering more than 10 people who worked on individual programs from the grantee/contractor side of things. Although their numbers are burdened, you pay that burdened cost to get the expertise, so those on the admin side for contractors/grantees are overhead on their part and carried as part of the budget. Now add in the student support, individual grants, etc. and the average number drops: to make up for all individuals getting support you need large groups getting large amounts of support, but in low numbers, to raise the average. And that just isn't your typical R&D project.

So, which organization leverages dollars better?

Bell Labs as it has a lower cost per employee ($157,000) while, even if you can get as many people leveraged by NSF for its total budget ($5.1 billion on total or $3 billion on the 59% leftover) (apples to apples) you get a higher cost per employee/contractor/grantee ($248,622)... and those would all be the equivalent of full time staff. Now, NSF may have oodles more people that it funds to bring that FTE via Grants, etc upwards. Yet if you work out the pay per staff at NSF ($140,000 average) and apply that to the 59% you get... 24,000 at government burdened cost. Now that is a bit higher than the industrial average... but $140,000 per person sounds right, given the scientific realm. Although Bell Labs did pay more, in inflation adjusted terms.

Of course to get them you have an annual cost of $2.287 billion to NSF staff (1,100).

Now depending on how many non-productive staff was at Bell Labs, if it approached 1:16 employees, I would be amazed as that is the goal of industrial work (based on the old German Army ratio of NCO's to enlisted/conscripts), which would get them about 1,400 support staff out of the 22,500 total. So a bit larger than NSF, but also on a lower budget. If you take the Bell Labs cost/employee ($157,000) and divide that into the overall NSF budget ($5.594 billion) you get - 35,700 people.

And NSF, if they get $140,000 equivalents plus their own staff gets you 25,100 people.

And Bell Labs still pays more.

Nasty, huh?

Same amount of money, Bell Labs employs more people at higher wages.

Government inefficiency is like that.

And NSF is a model of efficiency, really, there are only a couple of outfits better than them in entire government (across all the things done by the federal government).

Why am I not impressed by wanting government to spend more money on R&D?

You would be better just encouraging companies to do more (for every 1,000 employed only 71 are in R&D, via NSF 2004 report), but that is ALREADY 7.1% of the workforce.

The problem is, like Mr. Simberg points out, that it is asinine to put a percentage of GDP to R&D. The portion through government needs to be targeted to the base work that government has given to it: security, defense, etc. Anything else is most likely already paid for, and what happens is that pushing more government money into the system distorts the system. Wanting MORE government money just POLITICIZES research, which is just the OPPOSITE of what President Obama expressed:

In recent years, he said, “scientific integrity has been undermined and scientific research politicized in an effort to advance predetermined ideological agendas.”

He then drew chuckles, commenting: “I want to be sure that facts are driving scientific decisions, not the other way around,” Obama said.

The facts are: if there is money to be made in it, it is being done.

Some ENGINEERING projects desperately need funds, like the Polywell fusion folks (amongst many) as well as some of the Space Access groups... but those are not 'big ticket' items as the first needs a few millions and the second needs PRIZES awarded for PERFORMANCE, like the X-Prizes. Those are low-cost, big-payoff incentives that government has always done and done well. Set a goal and pay people to achieve it, let the competition have fun on it and guarantee contracts for SERVICE. That is how we created the airline industry - by prizes.

To de-politicize science you need to get the politicians out of it.

Just the opposite of what is being proposed.

No comments: