Building an enduring statistics system not simply a game of marbles

WHEN it comes to systems of evidence in policy – statistics, technology and geography – are indivisible elements of the system. Business and leadership literature discusses how non-core elements can be a drain on business and the need for outsourcing.

A lean and mean institution is often needed to achieve business objectives. As a consequence of this new business intelligence many services got outsourced, including in government, where non-core services such as cleaning and security were parcelled out.

Many in the developing world have asked serious questions about this approach. In South Africa, where there are massive backlogs from our uneven development, these questions remain pertinent and a strike by private security later catapulted this question to the forefront. The jury on the wisdom of outsourcing aspects of services is still out.

Independence over methods, technology, human resources and finance, including procurement, is indeed an essential condition for ensuring production of numbers fit for use in a nation. Of course, the independence has to be managed with a tight accountability framework.

Wayne Smith, the former Chief Statistician of Canada, resigned his post publicly in September 2016, protesting what he argued was the federal government’s failure to protect Statistics Canada’s independence. Canada’s government had decided to introduce through centralisation shared computer services.

By so doing, because of the indivisibility of statistics, technology and geography, the federal government appropriated or forced Statistics Canada to outsource technology – a core function of statistics. It furthermore exposed private information enshrined to be protected by the UN Fundamental Principles for Official Statistics and in Statistics Law to hands and minds that are not entitled to such information. Wayne Smith would have none of it and he quit.

“I have made the best effort I can to have this situation remediated, but to no effect,” he said in a note to the National Statistical Council, which advised him. “I cannot lend my support to government initiatives that will purport to protect the independence of Statistics Canada when, in fact, that independence has never been more compromised.

“I do not wish to preside over the decline of what is still, but cannot remain in these circumstances, a world-leading statistical office.”

Technology

Earlier on in my career in statistics, I realised the importance of technology, and from a corner in Bophuthatswana I endeavoured as the then-director of statistics, with advice from Professor Kahimbaara, to modernise the institution.

This was through ensuring that we move from mainframe computers and using personal computers, acquiring Statistical Analysis Software (SAS) for our operations and training of staff.

But at the time the government had outsourced technology responsibility to consulting firm Ernst & Young. This company wanted to impose a federal government-like decision and approach on Bophuthatswana Statistics. Kahimbaara and I fought this decision.

I recall presenting our arguments both to Mogami, who was the permanent secretary, and Keikelame, who was minister of economy.

We won the battle and after procuring SAS we started the modernisation programme.

In 1995 when Mark Orkin, after his appointment in July, visited my office, he was amazed by the 30 staff members of then-North West who were working on PCs doing statistics stuff.

In 1994 Benny Mokaba arranged a visit for Professor Sangweni and Dr Skweyiya to visit my office. They

Independence over methods, technology, human resources and finance is an essential condition for ensuring production of numbers.

could not believe the advances we had made in geography in this corner of South Africa.

When it came to communicating results and arguments for statistics, Max Sisulu invited Kahimbaara and I to a portfolio committee meeting in Cape Town in 1994, where we presented on the function of statistics in the new South Africa.

The background

In 2003, Jairo Arrow, Piet Alberts, Ros Hirschowitz and I, as Statistician-General, were inspired by advances of the end-to-end production systems by SAS, including publishing. This was the way to go. But at the time I was facing numerous battles. These included trumped-up allegations that I appropriated funds for my private use.

Furthermore, there were allegations that I received funds from SAS. Allegations of this nature, when accompanied by adverse audit findings, a costly miscalculated consumer price index blunder, a high undercount in the census and economic statistics that were vacillating would bury one.

Thus it was very difficult to push for such a modernisation project offered by the SAS as the deputies in Information Technology, influenced largely by the cronies who were power thirsty, grew cold feet to proceed with SAS.

Little did they know that on the advice of Minister Trevor Manuel I had called for an investigation into the allegations against myself and requested Professor Sangweni, the head of Public Service, to run through all my affairs, including acquisition of property.

But when mud is thrown at you, no matter how hard you fight it at times, it creates patches where it will still stick.

So it turns out that in January 2006, a cut and paste job of a cell in a matrix of about 1 000 cells on manufacturing statistics, which was SAS generated pasted on a Word document, was transposed.

This led to my not-so-famous resignation, which Manuel and President Thabo Mbeki did not accept. The aim of the end-to-end SAS solution was to avert a cut and paste job.

But the Information Technology staff at the time was so scared of other forces at play that they could not implement in 2003 what the far-sighted leadership wanted. But when the birds came to roost they were nowhere to be seen. So Ros and I had to fall on our swords.

Building an enduring statistics system is not a game of marbles. As the institution enters a new phase, independence in methods, technology, budget and human resources with very strong accountability systems is the way to go.