Finance professionals love Excel, its ability to implement models at short notice, allow changes to behaviour to be captured in real time and to fix problems on the fly. IT professionals hate it for pretty much the same reasons.
As a professional coder I can see the IT perspective. Excel is the ultimate in unstructured programmes. You have no source control, you can write self modifying code, the spreadsheet is the data repository and large spreadsheets can soon become unmanageable. There are very few spreadsheets around which do not contain a bug, whether it is hard coded values instead of formulae, ranges that don’t quite match up with expectations, or VBA code with data dependencies that no longer reflects the realities of the data on the worksheet.
And of course this can get expensive. In fact there is the EuSprg website that tracks errors in spreadsheets that have significant consequences. So why do we keep using Excel?
Looking at it from the point of view of a trader, he can see an opportunity in the market, he can rapidly check his idea with figures downloaded into Excel from his usual sources and he can quickly demonstrate whether the opportunity is viable in a few minutes work. The trade is executed, the bank or hedge fund has made money and his bonus looks healthier.
What would happen if he did not have Excel? Well he could request IT to make an application to calculate the same figures. They will mumble something about getting a BA to investigate sometime next month and he should have a working App in less than a year from now if resources can be found.
That doesn’t really work when the opportunity could close in the next couple of hours.
Alternatively he could deploy desk ‘Strats’ to build a python tool. That would be quicker but they would have to build a framework to access the correct data, understand the calculation required when the trader might not be fully sure what he wanted to do, implement the model and return the results. How does the trader know what the model is really doing? What if it does not give the answer he wants? How can he experiment with tweaks to the model?
Excel is here with all its faults, because it gives it users an ability to answer relatively complex questions at short notice. Unfortunately an innovative spreadsheet tends to outstay its welcome so what was once a special case becomes entrenched in the company’s systems because no one is able to transform the complexities of the worksheet into a controlled technology solution.
I have done my fair share of Excel bug tracking and deconstruction in my life in the Front Office of banks. In fact for a while my role was to increase the power of the Excel tools for traders and structurers by providing addins to extend Excel functionality.
Of course there is a great difference between prototyping an idea and capturing the details of a live transaction on a spreadsheet for the life of the transaction. This introduces a whole new set of problems. Who will manage it for its lifetime, how will useful information be extracted from it for reporting and risk management, what happens when the terms of the transaction changes.
Regulators hate it when trades get booked onto spreadsheets, but if the P&L associated with the trade is big enough, someone will sanction it. In the past, whole portfolios of complex derivative transactions would exist in spreadsheets with various Heath Robinson mechanisms to extract valuations and risks. At one institution I worked at, the limitation on the number of such trades was set by the space available to house the accountants needed to manage the spreadsheets.
I am rather ambivalent about Excel, it has certainly given me a career in migrating trades from Excel to more secure technology. My principle dislike is the way data can be embedded in the spreadsheet with no ability to manage and audit it. On the other hand it creates a very flexible User Interface for low volume, high variance complex products.
Maybe Excel is beginning to reach the end of its hold on banks? The recently announced SS1/23 document from the PRA which covers the validation of business models in UK regulated financial institutions notes that ‘Model use is defined here as using a model’s output as a basis for informing business decisions’. It also notes that for “end-user computing (EUCs) … Good risk management practices involve quantitative methods that support business decisions being tested for correct implementation and use”. Where EUC=Excel in most cases.
No Excel is not going to go away (I was told it would disappear from a trading floor in 2003 - still there and a component I built in Excel is now considered a ‘strategic’ element of their infrastructure) but it is going to get a whole load more scrutiny.
I believe I have some answers to the Excel love it or hate it debate. Go to www.persistables.com to see more.