Email this article	Comment on this Article	Print this article

Is there room for improving th

Calculating survey weights for

However, in South Africa, we seem to have become very rich very quickly as the amount consumers spend on food relative to other items has declined by over 32%.

This is significant, as food prices are one of the major drivers behind the year-long rise in inflation. A word of caution, though – some of the details in the new CPIX weights could leave economists a little drunk or at least smoking their cigars.

Since 2000, food prices have risen by 13,5% more than the overall CPIX - yet in relative terms food expenditure has declined by some 32,3%!

Some of this may be explained by the fact that many South Africans have become wealthier; and therefore relative expenditure on food has declined.

However, there are some items in the new CPIX basket whose relative weight does not make sense. Perhaps I should provide some background first.

The 2006 Expenditure Survey was conducted in 24 000 households - down from the 36 000 households used in the 2000 Expenditure Survey. This may prove to be a significant flaw in the survey as only 0.19% of households are now surveyed – down from 0.33% in 2000.

However, this time round the survey took more time and people were required to keep records of all of their transactions.

Households were required to keep track of everything they bought at the shops, if they went to the movies or tipped the car guard.

This should have helped the survey to become more representative - but one wonders if the participants were diligent in following through with the task.

Be that as it may, the households in the survey certainly must have provided StatsSA with a wealth of information on which to compile the weightings of the CPIX. The problem, however, is that some logic was lost in the process of calculating the new weights.

The devil is in the details

Firstly, I’d like to address some relative weights that do not make sense:

Households, although smaller today than ever before (down from 4.72 people to 3.6 people in a household), spend more on funerals than they do on electricity. This is despite electrified households growing by nearly 70% between the survey years.

Electricity comprises 1.96 % of all household expenditure, whereas funeral services comprise 3.01%. While the 3.6 people in a household can each die only once - the fact that they have an 87% chance of using electricity on a daily basis makes a good case that something is wrong here.

Simply put, a funeral – priced at 15 years of household electricity consumption - works out to be fairly expensive in relative terms.

Many South Africans have also lit up despite the health warnings on cigarette packs, the Department of Health clampdowns on public smoking and those pesky tax increases.

The relative weight of cigarettes alone is 18.4% more than electricity. If one includes cigars, smoking costs consumers 30% more than electricity.

South Africans spend nearly twice as much on alcohol than they do on electricity. Smoking and drinking combined really costs consumers 3 times more than their electricity! No wonder funeral services lighten the wallet that much!

Another oddity is the fact that we spend more on bread than on electricity - which to me does not make sense either.

Expenditure on maize is no longer distinguishable and now just falls under “Other Bread and Cereal Products”. Thus, it would appear that our staple food is no longer maize but rather bread and chicken.

Moreover, logic dictates that the number of products on the shelves in stores has increased in the last while due to the increased wealth of South Africans – but the CPI will now measure fewer products. They say this is an international trend in a knowledge-age – is it?

History has some weighty issues

However, it is the weighting changes over time that makes the least amount of sense.

Take electricity for example. Household electricity consumption has grown over 98%, while nominal GDP rose by just 88%.

One would subsequently have thought that the weight of electricity would have increased, yet the weight of electricity fell from 3.43% to 1.96% - a decline of nearly 43% in relative terms.

While petrol sales increased by 115% between 2000 and 2006, the weighting of petrol declined from 4.72% to 4.07% - a decline of nearly 14%. Again, compared with both nominal GDP and household expenditure, this does not make sense.

Moreover, the spending on cars rose by 117.9% of household expenditure – while running costs (which includes petrol, tyres etc.) decreased by nearly 26%!

The history of car prices and car sales doesn’t quite support the huge increase given to the weighting of motor vehicles.

New vehicle sales did increase by 109% over the period but RMTC implies that second hand car sales did not increase that quite that much.

Petrol should have shown an increase in historical terms and perhaps in relative terms with such growth in expenditure on motor vehicles.

One would certainly not expect the decrease in the weighting of expenditure on motor vehicles that has taken place.

Health spending is a more difficult subject - but the fact is that many people visit doctors and hospitals every day.

Nearly 40% of the population makes use of the private sector, while less than 20% of the population are on medical aid. That makes the decline of 70.5% in direct health spending seem unrealistic.

While I have no direct evidence of health spending in a nation with a high AIDS infection rate and a high TB infection rate - this kind of drop makes no sense.

Perhaps the minister of health should be made aware that private health spending has actually declined by 70.5% in relative terms and she should subsequently leave the private health care sector alone (I left out the insurance component in both cases).

Another oddity is the fact that the weighting of insurance has increased by 307% in relative terms - which seems very high.

Most South Africans would say that they are spending more but the number of insured people has not increased that dramatically.

Would insurance premium increases explain this inflated weighting? If so, surely this would mean a shocking rise in inflation in January next year.

The count of Medical Aid members has grown by less than 10% between 2000 and 2006. Short-term insurers have been complaining about people “saving” by not keeping insurance payments up.

On a more humorous note, South Africans have been understandably been driven to drink – with relative spending on alcohol showing an increase of more than 100%.

In addition, relative spending on hotels and restaurants has shown an increase of 336%. It’s no wonder then that we have such an accident rate!

Another inflationary warning here is that the restaurant bill usually seems inflated - so that may just add 247 basis points to the CPIX on its own.

If this trend continues, Stats SA may only need to conduct inflation surveys in the hotel bar in 2025!

Hopefully by that time I’ll be Statistician General – at least then I’ll be able to reserve a stool at the bar!