Measuring Inflation

Listener 20 January, 2001.

Keywords: Statistics;

Statistics New Zealand (SNZ) releases its December quarter 2000 estimate of the Consumers Price Index (CPI) this week. The increase is expected to be a high and, if it is a dull news day, there will be much angst. In fact we have known for some time that consumer prices are increasing more quickly than usual, mainly because the fall in the New Zealand exchange rate will push up the domestic prices of imports which flow on into consumer prices. There is the puzzle of how quickly they will flow through, and the degree to which the New Zealand sellers can and will absorb the import price hikes. There is also an argument about whether the exchange rate will stay down. Everyone expects some recovery – providing the world financial markets remain stable – but some of the predicted New Zealand exchange rate levels, seem fantastical. They would take the pressure of domestic inflation, but the export sector would suffer grievously again.

Behind all this is the nation’s assumption that we have a good measure of inflation in the CPI. It is a measure of the respect for the competence and integrity of SNZ that the premise is so broadly accepted. As a result, however, we tend to use the index uncritically – especially as if it is THE only measure of inflation. It covers only two-thirds of the nation’s expenditure, and since the prices of the other third – investment and government spending – sometimes track quite differently, how can the CPI really be said to reflect the overall level of prices of the economy? This says nothing about SNZ, but it does say we often use a measure for purposes quite different from that for which it is constructed.

There are complaints about the CPI’s allowances for changes in the quality of items. For instance, how to compare the price of the computer on which this column compared to the first one I used, which was much slower and less flexible. Complaints about quality change are often from people who have not read the technical papers but have a political agenda (usually to undermine the integrity of the CPI). There was a bad case in a recent US senate committee report, which did not mention that their CPI is already adjusted down by 2 percent p.a. for assessed quality changes.

Another complaint is that the CPI does not represent the precise spending of a particular group of people. Absolutely true, but does it matter? As a rule – not, because there are so many items in the CPI regimen whose prices move roughly together, that differences in between-group spending do not have a great effect. The one exception is housing, because householders face quite different outlays and price changes, depending on whether they own their house with or without a mortgage or rent it (and who is the landlord). Some years ago the retired demanded their own consumer price index. When they got one, they were astonished that prices for the over-65s were rising more slowly than for others. Statisticians were not surprised, because they knew the retired tended to live in housing (owner occupied without mortgage and subsidised rental) whose prices rise less than average.

A particular issue is the treatment of household debt servicing in the CPI. In its 1997 revision SNZ decided to exclude interest from the index. This means if interest rates rise (or fall) then the CPI will still remain the same, rather than rise (or fall) too. This is quite a different issue from the Reserve Bank inflation target which also excludes interest rates (and should, because the RBNZ is using the interest rate as an operational instrument). It happens that the RBNZ targets on the CPI without interest rates (and sometimes without some other things). In my view it would be better if it were to target on the price level for the entire economy, and not just two thirds of it.

It is unclear why SNZ excluded interest rates, although the cases for and against are finely balanced. An important, perhaps decisive, consideration was that the exclusion followed overseas practice. Even so, debt servicing is an important component of many household’s spending, so that the CPI probably no longer reflects household spending patterns exactly. SNZ are preparing another index which corresponds more closely to actual household outlays including interest payments, and that may be more use for assessing whether incomes are keeping up with prices.

However, whatever its strengths and weaknesses, this week all eyes will be on the official CPI which excludes interest. Fortunately interest rates are not changing very much at the moment, so the two indexes will track along together for a while, both showing a big change because of last year’s exchange rate fall.

While the CPI is a well crafted, like all measurement tools, it is best used for the purposes for which it is designed. If it used for another purpose then it is not the craftsmen’s fault if things go wrong. If you want to use it for another purpose, it is better to understand its construction.

Footnote for Listener 25 September 1999


The new construction of the Consumer Price Index represents a new purpose for the index. Historically (the official CPI goes back to 1914) its main purpose was to assess the changes in prices that income recipients (workers and beneficiaries) experienced. Because most consumers paid interest, the CPI included the cost of interest, as well as the cost of the goods and services they bought. From this month, interest costs are dropped out of the regimen, because the CPI is used for monetary policy targeting and because overseas financiers need an index which is similar to the ones in international use (which usually exclude interest costs).

The new CPI may not be the best measure for considering wages or benefit changes. It remains possible to calculate one with interest costs in it. My expectation is that we will end up with two: one for the financial community, one for income assessment. They will not usually differ that much.