Socio-economic Models

Mark has a great comment on issues with interpreting what a dollar is worth. This idea is worth considering in Epidemiological studies as well. When trying to model socio-economic status, it can be tempting to treat income as a linear variable. Even worse, it can be tempting not to account for differences in costs of living. This reluctance to try and account for cost of living is an understandable temptation as the data on cost of living is extremely hard to generate and very context specific. How do you handle Manhattan (where a large house is unlikely to be even available, let alone affordable) when you compare it Duluth, Minnesota. Economics have some clever tricks, but they replace one source of error (incorrect standardization) with another one (modeling errors in the cost of living models).

But it is worth remembering that just because it is a clean number doesn't mean that you should just introduce it into the model "as is". Think long then model!

I'm not asking for money...

I'm just casually mentioning that much of the best reporting on the financial crisis has come from the good people at This American Life and that same crisis has hit TAL really hard.

It's not like I'm trying to guilt you into send a buck or two to their website.

When is a dollar not worth a dollar?

Joseph and I have been putting up a number of post on compensation recently (many inspired directly or indirectly by Felix Salmon), but there is something important that's been left out of the discussion both here and on the other blogs I've been reading on the subject:

Some dollars are worth more than others.

Having bounced back and forth from statistician to teacher to entrepreneur over the years, I've seen plenty of large fluctuations in income and I've developed a pretty good idea of what it costs to live comfortably as a single person in various cities.

In LA the cutoff is somewhere around thirty thousand. For that money you can get a decent apartment in pleasant neighborhood and still have enough income left over both to meet your basic needs and to get out and have some fun. (LA is one of the world's great cheap-eats towns)

If you go from mid-thirties to mid-nineties (which I did at one point), you will see a significant change in your lifestyle but it is nothing compared to the change you'll feel if you go from the mid thirties to the low twenties. Below thirty you start facing some ugly compromises. You may have to move to a rough part of town, Food becomes a larger part of your budget. Costs associated with work (wardrobe, transportation) remain annoyingly constant. Going out with friends becomes a great luxury (it's hard to convince the bartender to sell you two-thirds of a beer)

This discrepancy in lifestyle suggests the need for a different metric. Here's an example. To make the numbers come out even, let's talk about three incomes: 21K, 30K, 90K with our hypothetical Angeleno starting in the middle. In both absolute (60K) and relative terms (200%) the jump to the top tier dwarfs the jump to the bottom (9K/30%) but in impact on quality of life, the exact opposite holds.

What does all this mean? For one thing, it means that expected value and marginal changes are not the right tools to look at compensation, at least not in the way we normally use them. It means that the problem requires more of a piecewise approach.

Reputational Capital and Operational Definitions

The following obviously demands a longer post (or series of posts or articles or maybe a book or two) but I'll throw this out in the hope that someone else will do the work for me.

Rajiv Sethi has a characteristically strong piece on "Reputational Capital and Incentives in Organizations" which includes the following:

How, then, might a firm accomplish the subordination of short term goals to long term objectives in practice? There are two possibilities: one could hire individuals who are predisposed to behave in a principled manner even in the face of incentives not to do so, or one could design compensation schemes that adequately reward actions that preserve or enhance reputation. Economists, being fervent believers in the power of incentives, usually tend to favor the latter approach. But in this particular context, there are two possible problems with this. First, the contribution of any given transaction to the reputation of the firm is generally much more difficult to ascertain and quantify than any contribution to the firm's balance sheet. This makes it difficult to assign reward appropriately. Second, in order to serve as credible commitments to clients and customers, compensation schemes must be easily observable and not subject to renegotiation after the fact. This is seldom the case.

Which leads to today's essay question:

20 points

Are organizations biased toward properties that come with operational definitions? If so, can this bias lead to economically irrational behavior? (if you need extra space you can continue your answer on the back of the blog)

Delta -- "We love to fly; we just hate the passengers"

The airlines seem to be conducting an industry-wide experiment to determine just how badly you can treat customers and still sucker them into coming back.

From The Consumerist (h/t to Felix Salmon):

On Sunday, Andy emailed us from his seat on Delta Flight 2744 from Minneapolis to Washington, D.C., to let us know that he had no idea where his flight was going to land. The ticket he purchased said he was flying to Ronald Reagan National Airport, but Delta said it would all depend on whether they could beat their scheduled 10:19 arrival time and get there before the ten o'clock airport curfew--otherwise they'd have to land at Dulles. Strangely, they didn't mention this 10 p.m. curfew to Andy before he bought the ticket.

Pick a number, any number -- American Public Media edition

Marketplace host Kai Ryssdal closed with the following important point yesterday:

A very touchy way to look at the oil leak in the Gulf courtesy of a research report out of JPMorgan Chase.

Their analysis shows that for all the damage along the Gulf coast, the leak won't really hurt gross domestic product much. Might even increase it slightly in the short term.

Here's another take on what that report really tells you: That for all the attention paid to it by the economic press -- present company included -- GDP's a really blunt and unsubtle way to look at something as complicated as the American economy.

Journalists (particularly financial journalists) are by necessity in the oversimplification business. The good ones (and the people at American Public Radio and NPR are very good) frequently remind themselves of this and that awareness brings an essential caution and humility to their work.

The bad ones get shows on CNBC.

Thoughts on defined contribution retirement plans

Via Felix Salmon comes this interesting tidbit on hidden fees in 401(k) plans. As long time readers know, I have passing interest in the idea of retirement plans and the differences between defined benefit and defined contribution plans. However, this sort of penalty really makes the plan less attractive.

From moneychimp we have an estimate of the real rate of return of the stock market:

Over the very long run, the stock market has had an inflation-adjusted annualized return rate of between six and seven percent.

So if the 401(k) has administrative costs in the 1-2% range and forces you to invest in mutual funds (a realistic option) which also have a 1-2% range before administrative costs then it is perfectly plausible that 50% of the returns are eaten up in administration costs. Now if the plan allowed a client to invest directly inin TIPS then real rates of return are in the 4.25% range (call it 4% after Vanguard levels of administrative costs).

But TIPS are debts owed by the US Treasury. Do you know what else has been lending money to the US government and has extremely low administrative costs (at least compared to these)? Social Security. Sure, the government could default on it but they could also default on treasury bonds, too.

The worst part of all of this is that, insofar as administrative costs are invisible and borne by employees, companies have no clear incentive to bargain for low plan administrative costs. Since employees are locked into the plan (as it is attached to employment), you have lack of liquidity (you need to change your job to change your 401(k) plan) and asymmetry of information. I was always under the impression that these were not ideal free market conditions.

None of this is to say that these plans are necessarily bad nor that any particular company has a questionable plan in place. But it is a call to think carefully about what the incentive and information structure look like. After all, if we get this wrong it has a fairly major impact on the future.

Minefields

Janet Stemwedel has a very well balanced post on the minefield that is work-life balance. It is an issue that I have grappled with as well and I have decided that there really is no optimal solution. I've had maid service before and would like to have it again. But I'd never thought through to the next step of who is liekly to be a amid and what the implications of that are.

But I suspect that the biggest insight that she has is how quick we are to judge the decisions of others.

My experience paying for skilled and loving childcare while I worked was that philosophers and career advisers do not lurk when it comes to such choices. They butt right in and tell you all the ways you are doing it wrong.

The whole article is a very good read and quite thought provoking. Definitely a blog worth checking out!

What I wish I knew before

I was reading a post from Andrew Gelman. Every now and again he gives statistical advice to reader who write in with questions. Today he had a question about a rare drug exposure in a large population.

This reminds me of one idea that I wish I'd known when I was first doing statistical in the financial world (before deciding to get a PhD). It was the idea of "off support inference" or of trying to extend inference beyond the bounds of the data given. This can happen when the few exposed participants are in a portion of the data where there are few controls and there is effect measure modification. Some fairly severe errors can result.

Which makes me think about to all of my risk modeling and makes me wonder if I worried enough about these issues at the time?

"Better a Krabby Patty than a damn"

OK, the jump from SpongeBob to Brave, New World may be a bit much, but the good people at Worthwhile Canadian Initiative certainly know their way around Bikini Bottom.

p.s. You can find an excellent radio adaptation narrated by Huxley himself here.

More on compensation -- compensation and pyramid schemes

As Joseph mentioned before, Ph.D.s don't take post-docs because they like being overworked and underpaid; they take the positions because they believe that the post-doc will lead to better things. This kind of implicit compensation isn't uncommon nor is it necessarily a bad thing. (I'm trying to think of how you would go about making future opportunities part of an explicit compensation package even if you wanted to).

Implicit compensation is, however, particularly prone to problems with asymmetry of information and fraud. Let's say you're recruiting garbage collectors and you start your pitch with a story about collector who found a fifty dollar bill in a trash can by the side of the road. You don't mention that in the past twenty years, this was the only case of cash in trash that you knew of. This would be asymmetry of information. Now let's say you changed the story and made the fifty into a thousand dollar bill. This would be fraud.

At least some of the post-docs in Joseph's post probably took those positions because of asymmetry of information. If they had known the actual odds of getting the careers they were hoping for many of them would have picked other jobs (usually ones with higher explicit compensation).

There's a common private sector version of this that occurs in companies that recently experienced rapid growth in size and or stock price. When you go to work for one of these companies, you will find yourself bombarded by stories of skyrocketing careers. You'll hear about salaries doubling and vice-presidents who started as interns five or six years earlier.

Some of these stories are probably apocryphal but even if every single one were true they would still be tremendously misleading because career advancement at a growth company is basically a pyramid scheme. The first generation of employees experience huge gains (though they also faced a high risk of having a company collapse under them). The second generation experiences large gains. After that things quickly level off, partially because growth inevitably levels off and partly because the company gets better at recruiting and particularly at luring in higher-level people.

But the stories of amazing career paths become part of the corporate mythology and take years to fade. In the case of Wal-Mart, new employees were still hearing about millionaire cashiers more than a decade after the jump in stock prices that created them.

More on compensation -- tenure edition

[for Joseph's earlier posts on compensation see here and here.]

We often hear that the labor market will be fine if we just let market forces fix the problem. The assumption is that workers will consider the available jobs, weigh the requirements of each job against the compensation and rationally choose the optimal position. But the assumption behind that assumption is that employers will honor contracts, particularly regarding compensation.

This compensation can take a number of forms (salary, bonuses, pensions, benefits, stock options, etc.) and can be paid out over a variety of schedules. The only thing that matters is that the agreed-upon compensation is paid if the employee completes the agreed-upon duties.

Guarantees of job security are often part of compensation. Think of football coaches and CEOs. Failing to honor these guarantees is no different in either principle or practice from taking a delivery then failing to pay the agreed-upon price.

With that in mind, check out the latest from Colorado:

While other states have tried to modify tenure, Colorado's law was the boldest education reform in recent memory, according to Kate Walsh, the president of the Washington-based National Council on Teacher Quality, which promotes changing the way teachers are recruited and retained, including holding tenured teachers accountable with annual reviews.

Walsh thinks Colorado is now at the head of the pack in the second round of the Obama administration's Race to the Top competition, a $4.35 billion pot of stimulus money designed to prod just such changes.

"If I was a betting woman, I would absolutely put Colorado in first place," she said.

The new law requires teachers to be evaluated annually, with at least half of their rating based on whether their students progressed during the school year. Beginning teachers will have to show they've boosted student achievement for three straight years to earn tenure.

Teachers could lose tenure if their students don't show progress for two consecutive years. That won't be a possibility until 2015, however, because lawmakers slowed down the process under political pressure from the teachers' union. Teachers can appeal dismissal all the way to the state Supreme Court, and school districts have the burden of proving why they should be terminated.

Just a friendly reminder, if you are going to do business with Colorado, insist on being paid first and make sure to count the money. The state isn't that big on honoring its contracts.

Compensation

This is a huge topic and I am an amateur. But Mark's comment on implied compensation got me thinking that this topic is worth pursuing further. In a lot of the academy, you are investing effort now in the hope of future rewards. As a general strategy, I suspect that this approach is likely more successful than a purely short term strategy.

However, post-doctoral fellowships are an example of a job with a lot of inherent variablity. Now, I was fortunate enough to have had an exceptionally good post-doctoral experience with strong mentors that I highly respect. But I recognize that not all experiences are equal.

In a sense, the strongest anaology to this that I see in general society is that of pensions. it may seem to be an odd link, but bear with me. Pensions also promise a positive improvement in future welfare in future for sacrifices now (in terms of lower compensation, for example). Even the defined contribution form of pension (the 401(k), for example, or the RRSP in Canada) requires that the contributor have trust that these funds will be available in the future. Outside of market risk, you have issues like political risk or people who steal from pension funds (which are often hard to detect as the moeny is supposed to be inaccesible for 40 years and constantly watcing it isn't an ideal strategy insofar as it drives future behavior).

So how do you handle this level of uncertainty about future rewards? Part of it is the social constract that ensures that all parties do their best to keep these sorts of obligations. But it is unclear what you do when the implied compensation is difficult for the employer to fulfill. For example, if the academic job market is rendered flat due to fundign cuts, even the best mentor may have trouble placing a mentee in a good positions. Similiarly, Megan McArdle has opined on the lack of decent choices with under-funded public pensions.

I lack good ideas about ways to handle these sorts of problems but would be open to suggestions!

Funding Post-Docs

There was a neat post over at Prof-like Substance's blog about NSF funding. He wonders if the Canadian system is all that it is cracked up to be -- awards of $30,000 per year for 5 years. At these rates, a post-doc needs to come with their own funding or else one isn't possible.

I have two reactions.

One, salary support in Canada tends to be more robust. Putting the grant with the post-doctoral fellow gives them flexibility and more power in the relationship (the supervisor needs to give positive benefits to the post-doc to get them to come to a specific lab).

Two, why does every new lab need a post-doc? If the career of a professor spans 30 years and the average post-doc serves for 5 years then we are either expecing a robust growth in the number of scientific positions or we expect most post-docs not to become professors.

Now there could be good reasons to be a post-doctoral fellow that do not involve a career in academia afterwards. But the most common reason is a desire to be mentored towards a career. If the post-doc is a requirment then they are really poorly paid technicians.

Is this a good thing?

Journal Boycotts

I'm not going to wade in and argue if one side is correct or not but the boycott of Nature Publishing Group by the University of California seems like it will ahve a tough time working as planned. The NIH open access initiative will help but the costs to getting papers by inter-library loan isn't going to be pretty. And, for some fields, NPG has a bunch of key journals. You can't know all of the literature but it won't be a pleasant experience to have trouble tracking down key papers.

Heck, jsut ask me about finding the original paper on pooled logistic regression. It was seriously not fun!