Their use of the word significantly in this context does not refer to statistical significance. The percentages you are referring too are just basic metrics based on the last years data, not something that was t-tested. The longer time period changes the starting index, it doesn't really make this type of analysis more or or less statistically significant. As the author alluded, this could mean that economic mechanisms close this gap over time when it appears, or it could be a recent effect. Your criticism isn't really valid.