Thursday, May 2, 2013

GDP Deflator Inflation ≠ Consumer Inflation

I keep hearing some people talking about the GDP deflator as if its a measure of consumer inflation. That it’s a measure of inflation is indisputable – that it measures domestic consumer inflation is not.

In that respect, the consumer price index – which directly measures price changes of consumption goods and services – is a much better measure.

But let’s see the differences:

GDP Deflator

Consumer Price Index

Measures all goods produced and consumed in an economy Measures only consumption goods
Based on a varying mix of goods and services Based on a fixed basket of goods and services
Imputed from the differences in valuation in current prices against valuation in prices in a base year (prices and volumes) Calculated based on changes in current prices against prices in a base year (prices alone)

The differences in calculation are quite substantial and meaningful – they’re not looking at the same thing. In fact, calculating the GDP deflator is fairly convoluted as these things go – first tabulate all goods and services in current prices, then measure the same goods and services in prices prevailing in the base year, then calculate the ratio of current price production to constant price production. This ratio (multiplied by 100) is the GDP Deflator index, from which growth rates (inflation) can be calculated.

The CPI on the other hand, is based on changes in prices only, as the volume of goods consumed by a “representative” (average) household is taken to be fixed. The index is just a weighted average of the changes in prices across the same basket of goods.

You can see from this that GDP Deflator inflation can’t be measured directly – it has to be derived from the underlying changes in both volumes and prices, and not prices alone as is done with the CPI. For this reason, it’s often referred to as the “implicit” GDP Deflator. The GDP Deflator is thus not really a price index as many have assumed, and shouldn’t be treated as one.

To get a better sense of the differences, here’s the chart of CPI and GDP Deflator inflation in Malaysia since 2005 (log annual changes):


(Don’t worry about the different base years in the chart above; they’re not relevant since the CPI I use is a spliced series).

You can see that the GDP deflator is much more volatile than the CPI, at least within the sample period used. And the reason for that is because the GDP deflator incorporates all goods and services produced, even those not consumed in Malaysia i.e. export prices.

Export prices for non-manufactured goods can be pretty volatile. For example (average monthly prices):


Hence, there’s a reason why tracking nominal GDP growth can be important – and why inflation under the GDP Deflator can actually be beneficial, not detrimental, to Malaysian incomes.

Increases in the prices of primary commodities that Malaysia produces and exports would inflate the current price valuation of Malaysian GDP relative to base year prices i.e. GDP Deflator inflation would be increasing.

But this has the effect of also increasing the terms of trade when demand (or supply) is inelastic, i.e. we can buy more imports with our export receipts. The net effect of changes in the terms of trade is in theory indeterminate, but empirically the income effect predominates over the substitution effect: we get a net gain in export receipts and consequently real incomes. The opposite is also true: slower GDP deflator growth in Malaysia is indicative of a loss in real incomes.

To illustrate what I mean, here’s the series for CPO export values and volumes, converted into indexes based on 2007=100:


Export production of palm oil (the blue line) hasn’t actually changed all that much since 2008 – there’s been some growth over the years, but nothing to get too excited about. But export values (the red line) have been all over the place, peaking in 2011 as high as double (index above 200) the income receipts reached in 2007. Higher export prices = higher GDP Deflator inflation = higher incomes.

The idea that the higher purchasing power from nominal income growth is fully offset by higher GDP Deflator inflation just doesn’t stand up to scrutiny. Neither does the idea that GDP Deflator inflation is a better measure of “true” consumer inflation than the CPI. It’s a different measure of inflation, and not one that always directly impinges on people’s wallets.

The better comparison for the GDP Deflator would be with the Producer Price Index, which tracks the costs of factory inputs. But the PPI doesn’t completely presage changes in the CPI either, as the pass-through from producer prices to consumer prices is incomplete.

While there’s certainly some basis to the argument that the CPI could be improved as a measure of consumer inflation (regional differences for one, price indices for different income levels for another), taking the GDP Deflator as a substitute is a non-starter. It’s too different, too lagging, too infrequent, and includes stuff that consumers don’t actually consume.


  1. I think that you're absolutely right when you said that we need more types of CPI's. As a layperson, there appears to be a large perceived discrepancy between the CPI numbers reported and the actual price movements experienced by the urban folk. Some suspect a conspiracy to deceive. I personally suspect incompetence, and that the CPI presently used is just badly designed.

    On one hand, the weighting seems to focused on essential goods, which are heavily subsidised, and on the other hand, asset price inflation and the attending chain effects on prices in Malaysia has been largely confined to urban areas. All this then skews the CPI towards the inflation experienced rural poor who spend most of their income on essentials, and makes it inaccurate for the average middle class urban family.

    Assuming that the reported CPI is a poor indicator of the inflation experienced by urban folk (who make up the bulk of the domestic economy), isn't it then correct to say that the government is wrong when they say inflation is low or under control? I'm curious to see what the data would show if consumer inflation was calculated and calibrated specifically for places like KL or Penang over the past 10 years.

    1. This comment has been removed by the author.

    2. @anon,

      The methodology used by DOS pretty much follows international best practice. I would therefore consider the charge of incompetence a trifle excessive.

      Rather than repeating myself, try here and here.

      You might also want to read this and this.

      The CPI weights are arrived at through the Household Expenditure survey, which covers more than 40,000 households.

      Separate CPI numbers are currently published for rural, urban, Sabah, Sarawak, and for households earning less than RM3000 a month. The difference in overall prices between urban and rural is about 5% (from base year 2010); between average and RM3k households, the difference is minor.

      BTW, property prices are NOT included in the CPI (cars however are).

  2. This comment has been removed by the author.