Excel is great for data entry and reporting, but as a modeling tool it may not always be the best choice. Here is a great example:
It is noted that the Excel bug was just one of the issues found with the Reinhart and Rogoff paper. It is not often that rather technical data problems reach newspaper headlines but in this case the underlying paper was very influential in guiding economic policy towards austerity.
I am fairly regularly involved in conversion of spreadsheet models to modeling language implementations. That often (almost always) reveals errors in the spreadsheet formulas. Here is an interesting paper on spreadsheet errors: http://panko.shidler.hawaii.edu/ssr/Mypapers/whatknow.htm.