Monday, June 21, 2010

In Which I Do More Copying and Pasting

Because this is new to me (I was one of those who thought is was just a "posh-sounding" thing), and I ought to look into it a little more.

Basis points - an admonition

Quite why I am telling people things that any decent financial journalists' stylebook ought to have I don't know, but I have seen so much grievious abuse of the humble basis point over the last month or so ...

1) A basis point is a measure of differences between two interest rates (that's what a "basis" is; also futures and forward rates but those are basically the same thing). It is not merely a posh-sounding way of saying a hundredth of a per cent. Talking about unemployment rates, GDP growth or whatever in basis points doesn't make you look cool, it makes you look like you don't know what you are talking about (I once saw someone write that the rate of income tax was 4000 basis points. Puke.) There is a partial and occasional exception for inflation rates, as of course the inflation rate is the basis between an indexed gilt and an ordinary gilt, but usually it is not correct to talk about inflation in bp either.

2) A basis point is a measure of differences between interest rates. The current 3m dollar LIBOR is 0.54%, it's not 54bp.

3) Given that the basis point has been invented, use it. If one interest rate is 0.6% and another is 0.48%, the spread between them is 12bp. It is not "a 25% difference".

It's not difficult. The phrase "basis points" really just means "and now, since I am talking about interest rates, I am warning you that I propose to talk about adding and subtracting them rather than multiplying and compounding them, and also I am multiplying the quantities by 100 to make them easier to deal with". Bloomberg, Reuters, the FT and WSJ are all regular offenders with respect to this poor little unit.

The above was posted by the management at D-squared Digest.

Feel free to dispute the definition, of course.


Anonymous said...

Not disputing the definition, but surely "percentage points" is just as accurate? Admittedly though "point five percentage point rise" doesn't exactly dance off the tongue...

Brendan said...

I suppose you could argue it's just as accurate, but I would say that depends on an assumption that everyone knows the context of what's being talked about. Here's why. If you say an interest rate has gone from 2.50% to 3.00%, you could say that was a difference of 0.50% (instead of a difference of 50 basis points), and there would be no ambiguity there. However, most people, especially financial news types, would tend to put it as something like "the rate increased 0.50%" or even "the rate increased by half a percent." (Assuming, I harrumph, that they didn't say "five percent," "fifty percent," "point oh five percent," etc., but let's not get started on the larger problem of rampant innumeracy in Teh Media Writ Large, or I'll really start thrashing in the tall grass.)

In most cases, saying something went up or down by some percentage is multiplicative, not additive; e.g., "Sale! 30% off storewide!," in which all prices are reduced by a factor of three-tenths (multiplied by 0.7), or "I got a 15% raise," meaning your new salary is now 1.15 times your previous salary. To put it another way, and returning to the first pair of numbers, a change from 2.50 things to 3.00 things would more usually be said to be an increase of 20%. Think of it as an amount of pocket change, say -- you acquired an eleventh and twelfth quarter to go with your existing ten. Or maybe even more to the point, as, say, an increase in some crime rate: an uptick of 2.50 robberies per 1000 people to 3.00 per thousand would often be reported (especially by those with an axe to grind) as "a 20% increase in crime."

One more way to look at it: I feel this uneasiness when talking political horserace talk. For example, it has always struck me as somewhat inaccurate to say "Obama beat McCain by 7%," when their respective shares of the popular vote were 53% and 46%, even though pretty much everyone correctly intuits what is meant by that. Note that if one team had scored 53 goals and another scored 46 over some length of time, we would say the first scored 15% more goals.

Now, I am not about to say that Obama beat McCain by 700 basis points, especially on the Internet, lest Daniel Davies hear about it and come beat me up. But however at odds with convention it would be, it would be more accurate, to my mind, than saying he beat him by 7%.

Well, I'm not saying this very well, no matter how much I type (someone snuck decaf into my coffee cup!!!1!) so maybe I'd best break off. But do you see the distinction between talking about percentage changes in multiplicative terms (much more common) versus additive terms (usually only done when we're talking about things that are already percentages of some sort)? And why it was a Good Thing therefore for someone to have invented the term basis points, and why I applaud D-squared's insistence that the term, once invented, be used properly?

dsquared said...

"Percentage points" is excellent and almost always correct in contexts other than interest rates. It's a useful rule of thumb that when someone says, for example "unemployment rose by half a percentage point" that they're usually likely to know what they're talking about.

Similarly "Obama beat McCain by seven percentage points" clears up the ambiguity without bringing the humble basis point into the question.

In all honesty, I think would sometimes in normal conversation say "rose by half a per cent", mainly out of laziness, but never "nought point five per cent".