The economic modelling unit is the brains behind the Reserve Bank’s inflation targeting policy
David Le Page
To each age, its anathema. Idolatry, witchcraft and the plague have been succeeded by inflation and so, deep within the glass monolith of the Reserve Bank and many floors up, we have an economic modelling unit. Theirs is the task of making economic predictions more than a year in advance to help beat down the inexorable growth of wages and prices.
For just as those who shoot fast-moving, fluffy things must aim somewhere ahead of their target, so the Reserve Bank must wield its principal weapon – interest rates – a year to 18 months ahead of time to allow the effects to percolate through the economy.
Predicting the future is a dicey business at the best of times but the unit, run by Daleen Smal, which is the brains behind the bank’s new inflation targeting policy, is constructing concoctions of statistics, coefficients and equations that should produce results resembling what will happen in the real world.
Of course, there will always be flies in the ointment, like Opec, the international oil cartel which has forced up the price of crude oil 100% over the last year. Or floods in the Northern Province which whack agricultural production. These factors cannot be predicted by the models.
What can be simulated, however, is the more regular behaviour of the South African economy. To do this, the Reserve Bank team uses several different models simultaneously. This is a departure from previous practice – a single, large model was used from the late 1980s. Now it’s possible to use whichever model seems to best capture reality at any particular time.
One might expect that the bigger the model – and the more factors it takes into account – the more accurate it would be. But in practice it seems that the larger the model, the more guesstimation – and potential error – is needed in constructing it.
Modelling seems rather like baking a formidable cake. You take a whole lot of ingredients – 200 sets of different economic data, like interest rates, employment rates, tax rates, government spending, trade data, consumer price and production price index figures (CPI and PPI), inflation and wage statistics.
You decide on how much you’re going to use of each ingredient, then throw them into a bowl – a computer – and bake by applying different equations (no less than 400 in one model), then look at what comes out. How light or heavy is it, how hard the crust, how sweet? Did it collapse in the middle? Did it burn up and explode? Or, did it rise gently into an even and nutritious product, with only the mildest inflationary flavour, leaving growth intact and employment uncharred?
Of course, complicating the job is the fact that figures such as the CPI do not remain unchanged, but undergo constant revision by Statistics South Africa long after everyone’s got upset about them.
But it’s not just figures that go into building an economic model.
There are also basic assumptions, such as if average wages increase by 2%, people will spend 2% more. Sometimes, the increase in consumption might not correspond directly to wage increases. Adjustments have to be made for whatever pattern proves closest to reality.
Another typical assumption would be that if the oil price increases, the balance of payments will deteriorate, oil being an import.
The models are far from complete, and it’ll take four or five years to assess just how well they’re standing up. But adjustments will be needed constantly. For example, when South African Airways buys Boeings in bulk for R4,3-billion, the costs knock monthly national trade statistics way off average.
This demonstrates the importance of intervention from the forecasters; the numbers cannot be blindly entered into the model and judgement calls are frequently necessary. This demands considerable experience and insight, and is the part of the process most vulnerable to criticism from those who consider economics something of a black art with few claims on being a science.
Indeed, concedes Smal, it comes down to being “one part science, one part art, and one part luck”. But the uncertainties are of impeccable lineage. The new models have been constructed in collaboration with modelling experts in the central banks of Australia, Canada, Israel, Sweden and the United Kingdom. Inflation targeting has driven monetary policy in all these countries for some time now.
Nor will the results of the modelling be applied without reservation to determining interest rates. They are just one of the factors considered in monthly monetary policy committee meetings, which advise Reserve Bank Governor Tito Mboweni.
Inflation targeting is unlikely to work solely because a central bank is pulling its strings in a particular direction. At best, it works because everyone comes to the party. The government has got to say: “Okay, we’re only letting those prices we control increase 5% more this year.” Unions, anticipating lower inflation, may mode-rate their salary demands. Businesses, expecting lower input costs, can plan lower pricing.
If, for example, the government decided overnight that it was going to give all public servants – and not just Cabinet ministers, as suggested – 12,9% pay rises, this would put considerable upwards pressure on inflation. It might well mean that interest rates have to be pushed higher in order to compensate. In other words, mortgage holders would be subsidising that pay rise through payments to the bank as well as to the Receiver.
While the Ministry of Finance is firm about the Reserve Bank taking primary responsi- bility for inflation targeting, its economists are busy dissecting the relationship between government spending and, in particular, its pricing policies for goods such as health and energy.
“It would be counterproductive if we agreed on prices that went way outside [the inflation target],” says Pippa Green of the Department of Finance. “We should struggle to contain prices where possible.”
Another important government contribution could be its rumoured decision to end Telkom’s licence earlier than anticipated, increasing competition and bringing down prices.
l Experiment with a “virtual economy” for the United Kingdom on the Net at ve.ifs.org.uk