I admit I made up the 1% figure, but I believe it is a reasonable estimate. Would you like to challenge the accuracy of the number? If anything, I am convinced that closer scrutiny would reveal it to be too high of an estimate. 2013 US GDP was 17 trillion dollars. Microsoft's 2013 revenue was 77 billion dollars, about 0.5% of GDP. Of course, Microsoft is not a pure software company; some portion of that revenue is hardware, services, and so on. There are other software companies, of course. I have never heard anyone reasonably justify a much higher cost than 1%.
It is true that the benefits of commercial software far exceed its costs, but from a public policy perspective that's completely irrelevant. The cost of producing this software is ~1% of the economy. If commercial software did not exist, the government or any public body could (provably) replicate the exact same benefits at the exact same cost. Government support for something on this scale is not a ridiculous idea; it's exactly how most basic research in science is actually done in the US. Hence 1% (or less, if I'm overestimating the cost) is the correct figure to use for policy prescrptions.