As HuskyDog pointed out above, both these studies are using results from military screening I.Q. tests. A few decades back, afaik, military careers were held in higher esteem.. Perhaps more people these days just don't care as much about trying to look fantastic to the military, so when taking the tests, they just don't put in as much effort as they used to.
Oh, and another thing: it could also be that any trend towards lower expediency, even if accuracy improved by some factor, would appear as a I.Q. score decline. I bet people are gradually forgetting and/or not being informed that I.Q. tests are timed severely, and are taking more time to answer each question, and/or wasting more time on a tricky problem instead of making a best guess and quickly moving onto the next.. Too bad the paper is behind a paywall. I'd be very curious to see if only the final score was used, or if its components (accuracy / time) were also analysed for trends.
Anyways, until these scores are compared with another, properly administered, test dataset that is not within the context of military force, I'd prefer to remain a sceptic of these results.