The Reserve Bank's decision not to cut interest rates was quick, unanimous and in line with the consensus among private economists. But what if it was wrong?

The Bank's decision was based on its inflation outlook, which holds that the consumer price index (CPI), which moderated to 6.4 percent in August, will fall back within the 3 percent-6 percent target band only in the second quarter of next year.

But what if targeted inflation was already in the target band ? would that have emboldened the Bank to make one more, final, rate cut? Another 50 basis point cut would take prime to 10 percent, the point that many economists believe should be the trough of the cutting cycle.

Since January this year, the Bank has been targeting inflation calculated on a new CPI basket in which the importance attached to food is lower, and that given to transport higher, than these values probably should be.

The problem seems to be that the new weighting (relative proportion of expenditure on different products) was introduced just as SA entered its worst recession in 17 years, but was based on consumption patterns of four years ago, when SA was at the height of a credit fuelled consumption boom.

A credit fuelled consumption boom

The current CPI basket formula was calculated according to a new, improved Income and Expenditure Survey (IES) conducted in 2005/2006. SA carries out a full IES every five years at a cost of R150-million-R200-million. It involves interviewing a representative sample of more than 21 000 households over a full 12 months.

Because consumers' buying patterns shift over time as they substitute alternative products for items whose prices are rising more rapidly than others, the CPI (which is a fixed-weight index) develops an upward bias over time. It is therefore international best practice to re-weight a CPI at least every five years. In the UK, new weightings are issued annually, and in the US every second year. Because of the cost, SA does it only every five years, as does Germany.

Patrick Kelly, executive manager for the CPI at Stats SA, says: "If we had unlimited funds, my best-case scenario would be to run an IES continually and issue new weights each year."

Some of the new weights generated by the 2005/2006 IES were so implausible that Stats SA spent months making adjustments. For instance, the new IES had a weighting of 11 percent for food, compared with 20 percent in the old IES, done in 2000. After consulting production and sales data, Stats SA bumped food up to 16 percent.

In a recession, the proportion of a household's income spent on food may rise as income shrinks. Other more discretionary spending is also cut back. Economists have been worried that SA's CPI basket may not reflect reality and, therefore, that monetary policy decisions may be inappropriate.

Mindful of this criticism, Kelly says, Stats SA conducted an exercise to recalculate the CPI weights based on current national accounts data, which contains information on household expenditure used to calculate GDP.

Continue on page 2...