Economics is not physics. You're citing a paper that essentially says that measuring the economic effects of public research is hard. It is. But that neither proves nor disproves the matter, and it isn't going to change. It isn't a matter of having to do a better study, it's a matter of there being no good way to do a study.
This is quite common in economics, but it's especially bad here, because the benefits of research don't inherently result in transfers of currency, and the benefit is actually often to eliminate them. The existence of Linux means that businesses can make broadband routers without having a large internal software team, which means vendors have lower costs and more competition, which means you can get a broadband router for <$100 instead of $1000. That is clearly a useful benefit, but the specifics of it are going to depend on what people do with the surplus $900, which will change over time and cultures and populations in ways that have no relation to the research. There is not going to be any good way to measure that and any measurement is not going to be stable or generally applicable.
But the difficulty in measurement doesn't get you out of having to draw a conclusion, because you still have a decision to make and need something to base it on. So when data is not forthcoming, we use analysis. And the analysis clearly points to research producing large returns, because the product is one that is produced once and can thereafter be used and reproduced arbitrarily many times at negligible cost. It's the classic scenario that justifies governments building roads and infrastructure with tax dollars.
The caveat is that the research has to be useful. It is obviously of no benefit to allocate a crate of cash labeled "research funding" and then transfer it to cronies or apparatchiks. But then the fight isn't over whether we should fund research, it's over how to make sure the money goes to the right place.
Here's an article from Nature looking at the evidence:
http://www.nature.com/news/2010/100609/full/465682a.html