Home > Inflation > There’s No Such Thing as… Car Prices

There’s No Such Thing as… Car Prices

A nice reminder that price index data is a work of fiction statistical genius on which it is perfectly safe to base macroeconomic policy.  Reading through the ONS announcements today, this is how they describe the new methodology for determining car prices used in the national accounts:

4.2.2 New approach to be implemented in September 2014

For the new approach the list prices from Glass’s Guide are reduced to take into account discounts negotiated at the point of sale. The percentage discount applied for each model is partly offset by an uplift to account for point of sale accessories/optional extras purchased.

Discussions with industry experts including Glass’s and HM Revenue and Customs (HMRC) established that data on discount prices and amount spent on optional extras isn’t available. As an alternative method, “target price” information from What Car magazine (monthly publication) has been used to estimate a best achievable discount. The target price is a guide to a typical achievable discount based on a team of What Car mystery shoppers (people posing as customers) who haggle with dealers. A discount percentage for each model is calculated using the target price data. Since this represents the best discount available, the discount calculated has been reduced by 30% to take into account not all customers will achieve this price (i.e. not everyone will negotiate the optimum discount).

“We took the figures for discounts out of a magazine and knocked 30% off for good measure.”  Why 30%?  Don’t ask too many questions.

The discount applied is further reduced to account for point of sale accessories/optional extras purchased. Research suggested metallic paint is the most popular extra added. This option isn’t typically available by moving to an improved model in the range which other optional extras often are. Looking across a range of models the cost of metallic paint typically offset the original What Car discount price by approximately 35%. ONS have therefore further reduced the discount calculated by a further 35%. The What Car best achievable discount price has therefore been reduced by 65% overall.

“Then we knocked off another 35%”.  Let’s hope there are no spreadsheet errors.

In fact it’s even worse: this method is used to calculate the deflator used for car sales, which is applied to the survey data on the volume of car sales to produce current price (nominal) spending.  The ONS really does produce RGDP first and NGDP second for some (many?) sectors.  The more I learn about price index and national accounts methodology, the more attractive nominal wage targeting becomes!

Categories: Inflation
  1. SK
    May 29, 2014 at 18:12

    Feels like we are not in the 21st century..
    Do they save these calculations in stone?

    • May 30, 2014 at 08:12

      I was going to say that “car prices” must be an area in which the ONS have developed expertise since they have been around a long time. Then I started thinking about hedonics again.

  2. ChrisA
    May 30, 2014 at 06:12

    If the discount of 65% is applied consistently, then for changes in the index it will wash out. And it is changes we are interested in, not the absolute value of the index, yes? Of course this begs the question of why apply an arbitrary discount in the first place.

    But sadly, we are all losing our innocence on how key indicators are compiled. Take the LIBOR fiasco, I had been using LIBOR as a gold standard for default interest rates in various developing country contracts for many years, scorning their offers of any local benchmarks (which were generally only shamefacedly offered, since the respect for LIBOR was so high). Shows what I know. To broaden the point further, it is like when you read in the newspaper an article about an area where you know quite a bit – almost always when not wrong it is presented in a highly misleading way. But somehow I manage not to generalize that to other newspaper reporting (its in the FT, it must be right…..).

    I guess the learning from all this is to use the most basic, least distorted reported data you can for any major decisions (like setting monetary policy). Perhaps we can get NGDP figures that do this (certainly better than inflation) but maybe even simpler figures like changes in wage rates are the way to go.

    • May 30, 2014 at 08:10

      I don’t know if this methodology is applied to the CPI too, but for macro policy we care about the rate of change, and it would be extremely surprising to me if discounting was constant when there are large shocks to demand. But I’m no car industry expert.

      Not to mention that lovely piece of crony capitalism, the car scrappage scheme, in 2009!

      I completely agree with your last paragraph.

      • May 30, 2014 at 08:30

        It is worth noting that measuring average wages is not simple either, the Average Earnings Index had to replaced with the current AWE after methodological problems.

      • ChrisA
        May 31, 2014 at 03:41

        Question: does it matter if the variable you are targeting is somewhat inaccurately measured if expectations are doing most of the work? For instance say we are targeting NGDP at 5% , we achieve this target on the measured variable but actually the way we calculate it underestimates by 1%. This is just like actually having a target of 6%. Since the original 5% was somewhat arbitrary then so what? The important thing is that the economy has seen that there is a credible target and will have responded. So isn’t this whole thing about the difficulty in measuring NGDP a red herring?

  3. james in london
    May 30, 2014 at 13:02

    Tragical-comical. And I thought of that Stanley Fischer quote from Marcus Nunes at Historinhas showing the new Fed Reserve member airly dismissing NGDP:
    “There are those who support setting a nominal GDP target. I think that this is very impractical. The data that we receive on nominal GDP are very unstable. There are changes of whole percentage points between the various estimates of GDP. For this reason, I think that there is no reason to use nominal GDP as a target.”

    I suppose he is thinking also that “actual” anything (eg inflation) doesn’t matter much, just expections for the “actual” (eg inflation). Of course, insofar as expectations are based on trending the stated “actuals” it is the blind leading the blind.

    However, it is nominal wages that get in all the cracks, to borrow someone’s phrase, and actual and expected changes for them are what really matters.

  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: