The readers’ editor on… a painful lesson for the High Pay Centre and the Guardian

http://www.theguardian.com/commentisfree/2014/aug/31/painful-lesson-high-pay-centre-guardian

Version 0 of 1.

It is often hard enough to get the sums right when relying on your own calculations as a journalist. When you are relying on the workings of others, there is a whole other layer of risk. On 18 August the Guardian carried a page lead with the headline “Pay gap widens at biggest UK companies”.

The first paragraph said: “The bosses of Britain’s 100 biggest listed companies are earning on average 143 times more than their staff, according to data that exposes the growing imbalance between how the nation’s workforce and its business leaders are rewarded.”

The data came from a study published by the High Pay Centre, an “independent non-party thinktank established to monitor pay at the top of the income distribution and set out a road map towards better business and economic success”.

The story went on to say that the pay gap is widest at Randgold Resources, and that its boss’s salary of £4.4m was “nearly 1,500 times that of his average employee, many of whom work in the company’s African mines”.

The figures from the story, including the one that stated bosses on average are earning 143 times more than their staff, were also used in an online poll. However, that statistic and others relating to two other companies were wrong. The problem lies in the way the data was represented.

This was explained in a correction published on 21 August: “In an article about how much more bosses earn on average than their staff there were a number of errors as a result of inaccurate data supplied by the High Pay Centre, on which the story was based (Pay gap widens at biggest UK companies, 18 August, page 17). Mark Bristow, the chief executive of Randgold Resources, was paid £4.4m last year, which is 270 times more than the £16,412 earned by the mining group’s average employee, not 1,500 times as we had it. The mistake arose because the High Pay Centre stated that Bristow was paid 1,498 times more than the average salary of £2,968 per employee. In its calculations, the High Pay Centre wrongly included the 12,128 contract staff employed by Randgold, leading to a lower average pay per employee. In fact, Randgold employs only 2,979 people directly.”

The problems were complicated, and I couldn’t resolve them in 24 hours so took the article down, pending investigation, although I realised a little late that I had left the poll online with the incorrect ratio.

As I investigated further with a view to restoring the article with the correct figures – it is an important subject that deserves coverage – it appeared that only the Randgold figures were wrong, and the High Pay Centre readjusted the figures on its website to reflect a new ratio of 131 times the average salary.

However, the reporter who wrote the story realised there were further problems with another company’s figures, therefore we went back to the High Pay Centre to ask them to confirm that all the figures of the top 10 companies were now correct.

The High Pay Centre’s problem is that analysing 100 FTSE companies’ pay data in a way that allows like-for-like comparisons is too big a job for a small not-for-profit company, so the thinktank commissioned a highly respected external institution to do the original work for them.

As I asked the High Pay Centre to confirm that the data was now completely accurate, the deputy head went through the top 10 companies and found a further error with a third company. Once again this altered the ratio, this time to 130 times average salary. But neither the Guardian nor the thinktank had the resources to check the other 90 companies’ data, and so the article and the poll have been permanently deleted, with a footnote on the page making clear that both have been taken down because we were unable to verify all of the data.

It is a painful lesson for both the High Pay Centre and the Guardian, which also pulled a Comment article based on the erroneous figures just before it was published. Deborah Hargreaves, the director of the thinktank and a former Guardian and FT journalist said: “This is the first year that companies have had to publish a single figure for their CEO pay. In the past this had to be added up from a number of different tables in the annual report and the resulting figure was often disputed … So it was relatively straightforward for us to compile a list of CEO pay using company figures. The government has issued guidelines for how these figures should be compiled so we know they are at least comparable.

“One of the other requirements companies were supposed to meet under new pay rules that came into force last October was to show how they had compared CEO pay to pay and conditions in the rest of the workforce. This was introduced instead of forcing them to publish their pay ratio, which is strongly resisted by the business sector.

“However, we felt that this requirement was not being adhered to and in March we, along with investors and charitable trusts wrote to the regulator complaining about this.

“This is why we felt it was important to compile pay ratio figures ourselves. In order to save us some time and help with the data collection, we bought the employee data from PIRC [Pensions and Investment Research Consultants Ltd, an independent governance and shareholder consultancy], who collated it from annual reports. We then put it together with our figures for CEO pay to calculate the ratios.

“However, the guidelines are not that strict for how companies are required to publish the data for their annual pay bill and number of employees. Disclosure is sometimes not clear. This is where the problems arose. Some companies include contractors and so-called associates in the overall number for their employees.

“But pay data for these additional people is not included in the pay bill. So when working out averages, the data is flawed. Once we started checking all the employee figures we realised where the problems had been caused. To be fair to PIRC, the employee data is not clear in some annual reports and needs careful checking. I think it is vital that we continue to work on this data. However, we have decided that we need to do our checking in-house … It is an important lesson in the use of data – it needs to be checked and checked again.”