I'm not saying ShadowStats is wrong, or isn't at least onto something worth considering, but it's important to understand that ShadowStats methods are questionable at best. From their explanation on their methodology:
"The ShadowStats number—a broad unemployment measure more in line with common experience—is my estimate. The approximation of the ShadowStats “long-term discouraged worker” category—those otherwise largely defined out of statistical existence in 1994—reflects proprietary modeling based on a variety of private and public surveying over the last two-plus decades. Beyond using the BLS U.6 estimate as an underlying monthly base, I have not found a way of accounting fully for the current unemployment circumstance and common experience using just the monthly headline data from the BLS." (http://www.shadowstats.com/article/c810x.pdf , Pg8)
So, take their data with a grain of salt since their data is admitedly a "proprietary modeled estimate," and without that model, it's data, their survey methodology, etc.. there is no way to verify the legitimacy of their chart. As such, I wouldn't use them as evidence of anything. I do however think it's worth discussing the decline in the Labor Force Participation Rate, which is publicly available data from the BLS.