Actually no. The vast majority of non-trivial conventional software does on average have at least one bug hiding in it. Anything longer than about 900 lines of code has a bug expectation value of over unity.
Figure 3.1
Often it it lurks in the error recovery code from some rare event that never gets properly exercised during final testing. The first time it happens all hell breaks out.
The minimum is somewhere around 500 lines where the expectation value is about 1/2 so roughly around half the code of that length is bug free.
I am a fan of McCabes CCI which also gets a mention further down that page. It doesn't tell you where the bugs are but it can tell you if a routine is so complicated that bugs are likely to be lurking and allows you to compute the minimum number of test cases to traverse every path.
Word is the MicroSoft product that I truly hate with a vengeance. I have seen so many completely insane things happen with that rats nest the funniest being exponential growth of documents containing images that are edited with different versions many times in a corporate setting.
I have variously used supercalc, 123 and other spreadsheets for creating test data. They all have the advantage of being a very different way of doing things which helps avoid a lot of common fence post type errors.