Try as they might, treasurers may never be able to totally dispense with the slow, manual and error prone application that is the spreadsheet. But rather than simply blaming software applications for data errors, should we be giving more serious thought to how we can improve the way we use them?
The fact that spreadsheets can be prone to producing dangerous financial errors is no great secret. Yet the recent study by modelling experts F1F9, which attempted to quantify the associated risk still makes for quite startling reading. The report, which harvested data from international market research company YouGov, claims that errors in spreadsheets could be putting billions of pounds at risk. Almost one in five UK businesses was found to have suffered a financial loss originating from errors in a spreadsheet (a population which represents companies with a combined turnover £318bn), while 57% of the large companies surveyed said that bad spreadsheets had been a cause of wasted time. Errors are not exactly sporadic either. Around 16% of large companies also admitted finding inaccuracies in spreadsheets more than ten times in 2014.
Yet anyone who dreams that imminent technological advancements will soon see us consigning all our tabular worksheets to the bin is deceiving themselves. There is plenty of anecdotal evidence to support this assertion. Most treasurers, even those that work at large companies with sophisticated treasury software, will tell you the same thing. For better or for worse, the spreadsheet is very much here to stay. Take for instance a large multinational that is expanding into new markets. Up at the group treasury level, they will very likely have some state-of-art TMS or ERP system with more-or-less straight through processing (STP) for certain types of transaction. The cost of setting up new treasury operations in a country often forces treasuries to take a slightly different approach, however. “If you want to do everything on a centralised basis it can cost you a lot,” says Denis Eckanauer, Regional Treasurer for ABB China. “Sometimes you need to find the right balance, and that might mean still doing some work on Excel spreadsheets.”
Enrico Camerinelli, a Senior Analyst at Aite Group agrees that sometimes companies need to take a pragmatic approach. “It’s common sense that it can be very risky to run a business just using a spreadsheet. But on the other hand it is such a useful, flexible instrument that I think we shouldn’t fight against it. We should try to make it our friend.”
Perhaps the problem is less about the application itself, and more around how it is used. Camerinelli certainly thinks this is the case. The F1F9 report places the blame at the door of companies that are not investing sufficiently in training (not a surprising conclusion, given that modelling training is one of the services F1F9 offers). Apparently, in one third of the businesses surveyed none of the financial decision makers had ever received any “official external training”. But as Camerinelli points out, it can also be the other extreme too. Sometimes an error is caused not by a lack of proficiency in using spreadsheets but, somewhat counterintuitively, an over-proficiency. “It is those spreadsheets with multiple tables, and links and references to other files. That is what I think makes spreadsheets really dangerous.”
If experts can make mistakes that are just as dangerous as complete novices when it comes to using spreadsheets then it would follow that training, while certainly helpful, can by no means prevent every error. That might help to explain both why a large number of companies in the F1F9 survey, even despite the increased regulatory focus on data accuracy, have not made much investment in modelling training. It might also help to explain why ClusterSeven, a provider of spreadsheet checking software, now boasts one third of the world’s top 30 banks as clients.
So, while a cost-effective, flexible application that can replace the spreadsheet might still be some way off, the technology capable of addressing the spreadsheet’s propensity to produce errors is evidently not.