Thursday, June 17, 2010

Measuring Inflation

The MTUC, among others is calling for a more “realistic” measurement of inflation (excerpt):

MTUC calls for a more realistic CPI benchmark

PETALING JAYA: The working class is not able to cope with the rising cost of living because the percentage of wage increase is based on an unrealistic Consumer Price Index (CPI).

Malaysian Trade Union Congress (MTUC) secretary-general G. Rajasekaran said most employers kept wage increases to a minimum 3% based on the unrealistic CPI figure furnished by the authorities.

“If this trend continues how can Malaysia achieve a high income nation status (by increasing per capita income from RM23,100 to RM45,500) within a decade?’’ he asked.

Rajasekaran said this in response to public complaints over the rising cost of goods and essential items.

The Statistics Department announced last month that the CPI had increased from 111.5 to 113.2.

Rajasekaran said the CPI figure given by the authorities did not reflect the actual situation on the ground, adding that information should not be suppressed because the Government needed realistic data when formulating policies.

“It is high time the Government commissioned an independent body to gauge the CPI. They must be truthful. They must capture the actual rise in cost of living,’’ he added.

Rajasekaran also said it would be a “living hell” if the Government revoked subsidies without putting in place a proper social security net.

Advanced countries like United States still provided subsidies to farmers to keep the price of produce low.

“Advanced countries also have social security scheme in aid of unemployed folk. We do not have such scheme,’’ he said.

Rajasekaran also said most people resorted to taking loans from Ah Long to make ends meet, adding that it was a risk they were forced to take because times were tough.

There’s no doubt that many have felt the pinch of rising prices over the last decade, as higher fuel and food prices have effectively reduced take-home pay. But saying that the CPI is “unrealistic” means that many don’t understand how and why the CPI is constructed, nor do they understand what inflation actually is and how it relates to income.

The problem as I see it has a number of dimensions, the first being that the Consumer Price Index, like all price indices, is a composite number. You are trying to convey in one metric the level of prices across a whole range of very different goods and services, across the whole country. Just by looking at that construction, you can see how people’s experience of inflation can differ against its composite measurement – someone in KL would have a very different view of how prices are moving, then someone in Pekan or Kluang for instance.

Also, price increases in one category, can and often are counterbalanced by price decreases in another. That happens even within each category, as price increases in vegetables are somewhat ameliorated by price controls on rice, sugar and other staples.

A further problem is that the weights used in each category of the CPI is based on the expenditure pattern of the average household, and these weights are periodically adjusted based on changes in that pattern. That means if your household expenditure priorities differ substantially from the “average”, you’re going to have a very different viewpoint on inflation. On an empirical note, we probably have here the main reason why there’s so much suspicion over the CPI measurement – most lower income households spend a greater than average portion of their income on food and transportation, which is where most of the price increases over the past five years has come from (click on the pic for a larger version: CPI components, 2005=100):


The result is a rise in the CPI that doesn’t seem to match what people are experiencing on the ground – particularly since people only notice the effect of price increases of some goods, even as prices are falling or are stagnant for other goods.

On that basis, it’s hard to say that the CPI’s measurement of inflation is “unrealistic”. Creating a more “realistic” CPI won’t do anything more than just spending money on something that isn’t actually broken.

The last and most fundamental reason is that many people still don’t understand what inflation actually is. The textbook definition is “a rise in the general level of prices”, which means that one-off price increases don’t actually count much (e.g. if we introduce GST, the petrol price hikes in 2008) as they only induce a one-time increase in the price-level. This has the effect of reducing real incomes, but without increasing the rate of inflation.

So what to do? There’s nothing fundamentally wrong with the CPI as it stands, and the current measurement is appropriate enough for policy purposes, if not for matching wages to cost of living increases.

Rather than revising the CPI entirely, I’d refine it in two ways:

  1. Publish on a geographical basis. We already do this for Peninsular Malaysia, Sabah, and Sarawak, but we should also have separate CPI measurements for each metropolitan area as well – Klang Valley, Penang, and JB.
  2. Use alternate weights to model the inflation experience for different income levels. Rural consumption patterns differ from urban (greater weightage on transportation for the latter, for example), and blue-collar workers have different spending priorities than white-collar workers. Using alternate weights can help account for these differences.

There’s no way to perfectly match what people are experiencing on the ground with what the CPI is saying, no matter how you fiddle with it – suiting the CPI to lower income groups’ inflation experience means it will be completely out of whack for everyone else. The refinements suggested here do no more nor less than match best practice elsewhere, and more happily, won’t be that hard to put in place. Price data is already collated geographically, and the household expenditure surveys already capture differences in expenditure patterns between households at different income levels.


  1. Good job - (Not a cooment but a note to you)
    I have been trying to reach you via email. Is there any possibility of inviting you for a brown bag seminar at Faculty of Economics, UM? Please, reply to
    Deputy Dean (Undergraduate)
    Faculty of Economic, UM

  2. CPI like PCI are all composite numbers.

    For more effective economic and targetted policies, like you say, the govt must get state by state figures. I mean the cost of living in KL and Selangor is way, way above the national average, especially food prices!!

    And just like the KLSE Index, too the govt must re-look at the CPI basket and revise it regularly, when necessary.

    Plain 'ol common sense and yet...?

    we are all of 1 race, the Human Race

  3. Yes, i think you are right regarding CPI and Inflation. I am one who don't really understand them. My family have never really taught me much about them though we have been running businesses since Kingdom of Kedah was established. Lots to catch up.

    My understanding on economics is way primitive.

  4. I agree with the MTUC whole heartedly. Malaysia's CPI number is a joke. International Fisher theory postulates a link between inflation and the currency exchange rate. Why is it that this theory seems not to work in the Malaysia case?

    The CPI number is subject to discretion as to the choice of goods, the location and the channel and the weight.

    In the USA, how they measure the CPI can only be described as HONEST and TRANSPARENT. Prices of 80,000 goods are polled and the weighting clearly. Furthermore, the CPI is generally overstated because of substitution bias. For the benefit of readers, it means that in 1980 under the category of Electronic Typing Device - the US will measure the price of a Typewritter. In 2010, they may measure the price of a HP Notebook. A HP Notebook has more functions than a Typewritter, hence if the price is higher its due to the substitution with a more advanced instrument.

    In Malaysia, which prices are being measured? Are the same people polled or a different group? If a high weighting is given the commodity linked prices i.e. rice,sugar and cooking oil, of course CPI is understated as these numbers are controlled. Yet people spend money on Baby powder, on milk, on napkins and on Proton Sagas, on private healthcare etc.

    The CPI number given by the Department of Statistics is condemned privately by most investors. The number is a joke, not comparable across nations, the result of price controls, not transparent in its measurement and generally designed to make the economy look good for political purposes.

    Why not BNM float an inflation linked note offering 3% real return.I'm sure it will be oversubscribed by all the international money managers in the known world?

    Yeah rite.

  5. Wenger, the interest parity condition is a terrible predictor of exchange rate behaviour. The most influential paper on this, and it's still relevant today, is Meese & Rogoff (1983) (warning pdf link). Here's some lovely insight by Rogoff on how the paper came about that you might want to read as well (also pdf link).

    Statistically, exchange rates are non-stationary processes, while interest rates are almost always stationary processes, which means that interest rates cannot have an impact on the level of exchange rates, only its rate of change.

    If you prefer a more normative explanation, exchange rates are relative, not absolute, prices - so it depends on the counterparty you're comparing with. Since the CPI is also an imperfect measure of inflation (whether it's fairly measured or not), there's always a degree of uncertainty around even that approach.

    With respect to your comments, the problems with Malaysia's CPI is less the construction of the CPI itself (technical details here - sorry another pdf link), but rather, as you point out, the presence of price controls. But we're not exactly alone on that score either as many developing and developed countries also have a finger on the scales in various ways.

    I'd prefer an imprefect measure that's at least published and relatively transparent (accurate or not), than not having one at all - try finding CPI and interest rate series for some of the Middle East nations, even the rich ones.