Instrumental Variables

Marginal Revolution linked recently to this paper which claims that overuse of Instrumental Variables (IVs) in econometric studies has led to them being less useful. So far as I can tell, the argument given in the paper is exactly backwards... what am I missing?

For the uninitiated, an IV is a variable which is correlated with the thing we want to study but is unequivocally not caused by it. Let's say we want to study the effect of x on y, then an instrumental variable z is useful if its only effect on y is through its effect on x. To use the language of Pearl's causal graphs, this is equivalent to saying that every path in the causal graph from z to y passes through x.

For example, suppose we want to measure the effect of a change in the price of apple on the demand for apples. It may seem difficult to do this directly, as an increase in demand is likely to lead to an increase in supply, and so price is affected by demand. A potential instrumental variable is the weather. If the the weather is favourable for growing apples then more apples will end up being grown, and this is presumably independent of demand - the increase in apples will have a predictable increase in the supply (and therefore the price) and we can measure the effect on demand.

Now, the authors of this paper claim that:

A Tragedy of the Commons has led to overuse of instrumental variables anda depletion of the actual stock of valid instruments for all econometricians. Each time an instrumental variable is shown to work in one study, that result automatically generates a latent variable problem in every other study that has used or will use the same instrumental variables, or another correlated with it, in a similar context. We see no solution to this. Useful instrumental variables are, we fear, going the way of Atlantic Cod.

As I said, I think this is exactly backwards. It is not the fact that new papers are produced which use these instrumental variables in a new context which introduces the latency problem: it reveals a latency problem which already existed. The previous studies were already invalid. The new studies just reveal the fact.

E.g. Imagine that people buy more apples when they have high levels of vitamin D in their blood (because apples are a substitute for fish oil). Then you have to correct for the effect of the Sun on vitamin D levels when you're using the sun as a proxy for apple demand. The problem here though, is that the weather conditions already weren't a good IV for demand in apples. The fact that a new study appears to demonstrate this is not a "Tragedy of the Commons" in any meaningful sense - the study has not been made any worse, we've just found out that it was bad.

On the other hand, Alex Tabarrok, who is more of an economist than I am, appears to be taking the paper vaguely seriously, and not to have noticed this. As I said before: am I missing something?

Published with Blogger-droid v1.6.5

Super Saver Delivery - classic price discrimination?

I've just ordered a book from Amazon. I selected the Super Saver Delivery option, and Amazon informed me that I should expect this book before February 9th. Now, I'm not exactly a stranger to ordering books from Amazon. I spent in the region of £500 last year, and I'm very nearly 100% confident that I will receive this book before the end of the week (that's February 5 for those who are keeping track). Why does Amazon do this? The two delivery options "Super Saver Delivery" and "Standard Delivery" are, as far as I can tell, completely indistinguishable in terms of service. The only difference being that one pays something in the region of £2 per order for the latter, and nothing at all for the former.

I can think of several reasons why they would exaggerate times: it's a pleasant surprise when your book arrives before it is supposed to (but then why not claim that all deliveries will take "up to 2 weeks"; it also covers them in case delivery takes longer than they might expect (but, again, why not give an even more exaggerated time). I can also think of reasons why they might want to introduce a "second-class" delivery service, in order to justify charging for their "first class" service (ie, the normal delivery rates that one would expect them to charge anyway). Similar to the way that second class customers on trains are deliberately inconvenienced in order to make first class more attractive. However I don't think this is what's going on here.

I think Amazon is indulging in some good old-fashioned price discrimination. As I said earlier, I'm a pretty regular customer; I order something on the order of £50 worth of books a month. As such, I know how long it takes for my books to be delivered, and I'm not even vaguely tempted to press the "standard delivery" option when it comes to the payment page. I'm exactly the sort of person who is most price-sensitive when it comes to deliveries, and I'm in a position to know that I can get the same service more cheaply.

Contrast Dear Aunt Doris, who orders a book off Amazon once a year for her nephew's Christmas present. She needs to get her delivery when she wants it, and she has no idea that "standard delivery" is the same as "super saver delivery". I mean, why would it be? That would just be crazy, surely? So Dear Aunt Doris pays the extra £2 for her one book a year, because she's not the sort of person who gets the chance to notice that she's paying for exactly the same service she could have gotten for free.

This is almost textbook price discrimination. You charge a lower price for the same service to those customers who are most likely to go look elsewhere, and most likely to be in a position to know exactly how much the service is worth. It's fun noticing phenomena you've read about in books in real life. I vaguely wonder if this theory of Amazon delivery pricing is testable (well, it obviously is if you work for Amazon...) but either way, I'm fairly confident it explains what's going on.

What I'm reading now

Once upon a time, when the Sparknotes message boards still existed* they had one called "what we're reading now" where people were supposed to write ongoing reviews of the books they were currently reading. What always surprised me what the number of people who were reading one book at a time! So, here's a list of the books that I currently haven't finished, but do intend to finish (it's probably incomplete, as it only includes ones that are by the side of my bed right now).

A Week in December - Sebastian Faulks

One of those "small world" type novels, small snatches of the lives of a variety of people living in London in late 2007. Fairly well-observed, and it's probably going somewhere. I can't imagine feeling as much for the characters in this as I did for the characters in Birdsong, but you can't expect anyone to write Birdsong twice. One serious gripe: the kids have accounts on YourPlace where they get "jabbed" by their friends who can post messages on their "doormats"; there is a virtual world called Parallax where the inhabitants refer to the real word as TL and trade in Vajos; there's a major bank about to collapse called "Allied National" and the lead singer of Girls from Behind is one of the leads on a reality show called It's Madness. It's slightly jarring living in this parallel world... even Al Qaeda has been given a different name. This make suspension of disbelief slightly harder, and I've pretty much no idea why you would choose to do it.

The News Where You Are - Catherine O'Flynn

Another novel set in the modern world (for some reason that's the only sort of novel I've been able to get into lately). To be honest, I've only read the first dozen or so pages. Has a fairly light touch, and I think it'll be an easy-going distraction. I doubt very much it will be any more than that, but I'm willing to be surprised.

Bridge The Silver Way - David Silver

A collection of bridge stories. My favourite books of bridge stories are the Chthonic books by Daniel Kleinmann. Unfortunately, there are no more of them for me to read, so I'm trying some others. These are readable, but not spectacularly good. I wouldn't recommend them.

The Time Paradox - Philip Zimbardo

This has been in the pile for a while. It's a book about time by the guy who did the Stanford Prison Experiment, and who has since done a lot of work on the psychology of time. It seems interesting, and seems to agree with . I want to know the contents of this book, but for some reason am struggling to actually read it. Will get there eventually.

The 4-Hour Body - Tim Ferris

Having read the 4-hour work week, agreed with most of it and subsequently ignored the vast majority of its advice, I couldn't wait to get my hands on this one. Fully of various reasonably well-researched tips and tricks about diet, exercise, sleep, etc. A lot of the things in this actually seem more actionable without going to too much effort than those in his previous book. I'll no doubt be playing with a few of them over the next few months. If any of them seem interesting/worthwhile/dangerous I'll probably blog about them at some point.

God is Not Great - Christopher Hitchens

Got this for Christmas, have been dipping in and out of it recently. Hitchens intellect is scary, he's a proper old-school Oxford academic (I just looked up to check that he actually did go to Oxford - I've no idea if I'd ever known that for certain before, it just feels as though he did). It's fun to read just for the sheer amount of stuff the man knows - his impressive takedown of the idea that Kosher food laws are in some way hygiene-related is the last chapter I read. Basically, I'm just basking in erudition. I'm sure I'll learn a few things too.

Priceless - William Poundstone

Yet another popular book about behavioural psychology. Extremely well-written, refers to several experiments and theories that I've never heard of, and gives the best intuitive explanation of Prospect Theory I've ever seen. I have actually finished this book, but I keep dipping into it because it was, simply, the best book I've read on one of my favourite subjects. Well-written, well-explained, well-researched, the chapters are a nice length to dip in and out of. I'm trying to think of something I didn't like about this book, but I'm struggling. I sort of which I'd decided to give the books in this post star ratings, so I could give this 6 out of 5.

Causality - Judea Pearl

After I wrote my post on Simpson's Paradox on Less Wrong, I decided I really should learn more about the theory of causality that I casually reference at the end. Partly because I kinda feel I should, but also because it just seems awesome. It's actual mathematics though, not exactly bedtime reading, although it hasn't been too tough so far.

Expert Political Judgement - Philip Tetlock

Tetlock's study is pretty famous. He basically concludes that "foxes" do better at predicting political events than "hedgehogs". To quote wikipedia: foxes "draw on a wide variety of experiences and believe the world cannot be boiled down to a single idea" hedgehogs "view the world through the lens of a single defining idea". I'm not sure how much more there is to get out of reading the whole book than what I just wrote in the paragraph above, but I've started so I'll finish. I've also promised Jess a summary - does that count? Or do I need to do it more detail?

So, there you go. There are no doubt others lying around that aren't in this pile next to my laptop. The pile is usually about this big. I can't understand how some people can read one book at once - what do you do when you're not in the mood for that one? (I sometimes even struggle to decide on one book to take on the Tube...). If I remember, this may become an occasional series.

Some of my favourite quotes

Here are a few of my favourite quotes I've collected over the years. Like all good quotes, I think they say something that can't easily be said by using more words. I'm sure there will be more when I remember them:

"If I'm selling to you, then I'll speak English. If you're selling to me dann muessen Sie Deutsch sprechen"

Willy Brandt

"If he knew, he's too evil to be Prime Minister, and if he didn't know, he's too stupid to be Prime Minister."

Nye Bevan on Anthony Eden, after Suez

"The Noah Principle: predicting rain doesn't count, building arks does."

Anon

"It's hard to make a man understand something when his livelihood depends on him not understanding it"

Upton Sinclair

Dry February

Over the past 24 hours I've taken the entirely arbitrary and capricious decision that I'm not going to drink any alcoholic drinks during February 2011. A long time ago someone asked me if I could stop drinking for a month, and my response ever since has been "yes, but I don't want to". Well, so much for that cliché, I'm not particularly sure I do want to, but I've decided to, and I'm now making a public commitment.

(I'll also be spending February getting back into the gym and the swimming pool properly for the first time in 6 months -the two are not entirely unrelated. You might also have observed that I've chosen the shortest month of the year for this particular experiment... coincidence? Well, I think so...)

Unfortunately, I don't think a public commitment is likely to actually stop me from drinking any alcohol at all in 2011 - I just don't mix with people who disapprove strongly enough of drinking alcohol (in fact, it's much more common for me to mix with people who disapprove of *not* drinking alcohol). Social opprobrium just won't do it.

So, I'm going to have my own personal little StickK contract (yes, for those in the know, that is a little bit like a "PIN number"). If I knowingly drink any alcoholic beverage between midnight tonight and midnight on February 28, 2011, I will donate £200 to... the Church of England.

I've been thinking for a while about which not-quite-worthy cause I could choose, and I think the C of E is probably a good compromise. If I say I'll give the money to the London Homeopathic Hospital (which was my first choice) then I'm pretty sure that I just won't do it even if I lose. On the other hand, while I pretty violently disagree with the C of E about just about everything, I'm at least willing to countenance the possibility that the marginal pound in their collection plates is a force for good.

So... why am I posting this? Partly because I want to get it out there. This is the sort of thing that isn't worth doing if you don't do it publicly, and partly (this is basically the same reason) because I want anyone who reads this and sees me in February to help enforce it. Seriously, you see me with a drink in my hand, you tell me there and then to get my chequebook out (I won't have it with me, but actually, I would guess it's possible to donate to the C of E online, so I might well be able to it instantly).

Why am I not taking a StickK contract out? Partly because I haven't quite figured out their business model - how do they make money if the users don't pay them any? Partly because I'm pretty sure one has no control over who the money goes to if you do lose. And partly because I want to see if it's necessary. If you have a blog, and if enough of your friends read it, does that replace the need for StickK?

Things they should teach in school part 2: Touch-typing

I mean, really, this is a gimme isn't it? You spend 6 hours a day for at least 10 years going to a building where they are supposed to be teaching you basic life skills, and they don't even teach you how to use the single piece of equipment that you're going to spend more of your time using than any other (disclaimer, I just made that statistic up, but I can't even think of another candidate, so I'm going to go ahead and assert it as fact).

It took me about 10 hours to learn to touch type, very possibly less, and it must save me several hours a week in time spent not looking at the keyboard and, more importantly, not having to correct mistakes that I didn't notice because I was looking at the keyboard instead of the screen.

I propose that teaching kids to touch type would be more useful than literally everything that they learn in school after the age of eleven. Anyone got any counter examples?

Young Person's Railcard Oyster Discount just got better!

My previous post on this question lamented the fact that lots of people don't seem to know that they can get discounts on travel on the London Underground network by loading their Young Persons Railcard onto their Oyster Card. Just take both railcard and Oyster card into any Tube station, and they'll do it for you. Well, now I'm lamenting it double, because things just got even better.

With absolutely no fanfare, TFL have extended the YPR discount to off-peak single fares, as well as the previous discount on the off-peak daily cap. See here for details. So now an off-peak zones 1-2 single will only cost you £1.25.

For those of you who are wondering, the off-peak daily cap applies as long as your first journey in the day came after 09.30. That is, you don't have to make all journeys in off-peak times in order for the off-peak cap to apply. For single fares, peak is 6.30-9.30 and 16.00-19.00. (For anyone who wants to stay out late, the "day" finishes at 04.30 the following morning).

For some reason, this seems to have been completely buried by TFL, I even had to edit the wikipedia page myself just now to contain the relevant information! My post, which comes from a relatively tiny blog is still in the top 10 hits on Google for "Young persons railcard oyster", for some reason. This information could be saving quite a lot of people quite a lot of money if they knew about it. Please tell your friends!