Pointlessness of job promises
With the elections over, attention will return to the more messy business of governance and contestation over policies and their implementation.
The state of the economy and the level of unemployment are key areas of concern and have been a major focus of campaigning. To the extent that this shows some responsiveness to voters, it is a good thing. However, too many grand claims and promises are being made.
At best they are not supported by serious evidence and at worst they are demonstrably misleading or simply without any substance.
I want to discuss two particular examples: the DA’s use of a Reserve Bank working paper to support claims about its proposed economic policies, and the exchange between DA and ANC representatives about which political party has the best record of creating jobs.
The Reserve Bank published a working paper last year titled, “Achieving higher growth and employment: policy options for South Africa”. The authors follow up on work by other researchers about factors that might be limiting growth and try to model the effects of policies that would lead to changes in five areas: labour intensity of growth (the extent to which growth creates jobs); low savings levels; a shortage of skilled workers; limited competition between firms; and the cost and inefficiency of “network industries” such as transport and telecommunications.
Before the elections, the DA claimed that this paper showed that policies like the DA’s “can increase economic growth to 8 percent, help the economy to create 6 million real jobs, and halve unemployment over the next 10 years”.
The Reserve Bank quickly denied that its paper had anything to do with DA policies. But what is an ordinary citizen to make of these claims?
A precedent for grand promises about job creation had been set after the 2009 elections when President Jacob Zuma promised to create 500 000 new jobs a year. And that in turn reflects a bad and unhelpful habit in post-1994 economic policy documents of aiming for particular growth or employment targets.
This doesn’t really make sense because predicting these variables is hard at the best of times, and influencing them is even harder. Economists are still debating how and why growth happened in developed countries, and there is no proven recipe for creating growth and jobs in developing countries. Further, our economy depends heavily on what is happening internationally. In that context, promising certain levels of growth or job creation is misleading.
Such problems can be seen in the DA-Reserve Bank case. The bank’s paper used a method known as “computable general equilibrium” modelling. The basic idea is to try to understand the economy by presenting links between different sectors in a single statistical table. This can then be refined to produce tables showing the inputs to, and outputs of, different economic sectors – including households. This model is then used to represent the economy and find the possible effects of changes to some variables as a result of a new policy.
It’s a nice idea, but there are three issues that undermine its practical value and that of similar economic impact analyses.
The first is that these methods are often used by researchers who already know what answer they want or are being paid to produce a particular result. For example, with big sporting events like the World Cup and the Olympics, bidding committees pay consultants to produce forecasts of the economic effect of the events.
In order to convince the public to support bids, many of these are tailored to yield large, positive estimates on growth and employment by making dubious assumptions. Typically, they ignore the fact that public money could yield better results by being spent on other things and that money spent on the event by consumers is diverted from other parts of the economy.
One reason that the promised economic benefits of the 2010 World Cup never materialised is that they were based on these kinds of flawed calculations.
The second, more fundamental, issue is that these modelling methods have limitations, which make it hard to have much confidence in their predictions. They require a great deal of information, which is often not available or may include significant errors. Even an accurate model does not necessarily tell us how one variable will change in response to another. Furthermore, to make the model less complicated researchers often make convenient assumptions that we know are wrong, like assuming there is no collusion between firms.
Finally, in order to model the effect of a particular policy, researchers will assume the effect of the policy on one sector then use the model to estimate the effect on the rest of the economy.
This reveals the third problem: it’s really not very useful to know what will happen to the economy if a policy works; what you first need to know is whether the policy will work.
So, on jobs, what matters is not whether the policies modelled in the Reserve Bank paper align with DA policies but the fact that the effectiveness of these policies is not known. The paper simply assumes the policies are effective in order to estimate the effects on the broader economy. To claim that this “proves” DA policies will create jobs and growth is therefore simply incorrect – it doesn’t really prove anything.
What is critically important, however, is that the choice of policies to model can reflect different ideological views of the world. For example, the Reserve Bank paper suggests that reducing barriers to importing skilled workers would increase skills in the economy and lead to growth that would create employment.
However, in theory skills could also be increased by discouraging skilled South Africans from emigrating by, for example, requiring them to pay back any public subsidy to their university education. It’s in such choices that political views become important, but it’s all too easy for that to be lost in seemingly complicated economic models.
Similarly, the argument about whether the ANC created more jobs in Gauteng than the DA did in the Western Cape is silly and dangerously misleading.
Employment figures from StatsSA are based on a sample survey of 30 000 households out of about 14 million, which means they are estimates. As a colleague pointed out to me, even the official employment figures reporting the creation of 653 000 jobs last year – mentioned by Zuma in his State of the Nation address – are sufficiently uncertain that it is possible no jobs were created or that double the number were.
The more localised the estimates the worse the uncertainty gets: so even when the estimates for the Western Cape and Gauteng differ by hundreds of thousands of jobs, this may not be accurately revealed by a small survey. There is so much uncertainty about the accuracy of the estimates that we can’t really make these comparisons with much confidence.
In addition, it is a dangerous game to play because it is usually impossible to tell whether province-level employment creation should be credited to provincial government or national government, or neither.
For example, should international tourism growth to Cape Town – and the associated job creation – be credited to DA management of the city, or to the ANC for management of the country and its infrastructure? Or just to exchange rate fluctuations? Engaging in misleading political point-scoring could, however, create a situation in which the national government deprioritises provinces governed by the opposition – a situation that is in no one’s long-term interests.
Politicians and policymakers should stop making claims and promises about job creation. The best they can do is try to choose the right policies, implement them well, monitor them carefully and revise them if they do not seem to be working. Debates about economic policy should be focused on the merits of policies, not grand promises and misleading claims.
l Muller is a senior researcher at the Development Policy Research Unit at UCT. This article is partly based on a previous piece published in the Alternative Information Development Centre’s Progressive Economics Bulletin (April 2014).