Since the start of the most recent recession, the employment‐population rate for high school‐age youth (16 and 17 year olds) has fallen from around 23 percent to around 15 percent, and is now at its lowest level since the Bureau of Labor Statistics began collecting data (figure 1). However, teen employment has been falling at about this rate since the early 2000s. Taking a longer perspective, youth employment rates have been trending down since the early 1980s; during every subsequent recession, the teen employment rate has tumbled and never recovered to its pre‐recession level during the ensuing recovery.
...I conclude that labor demand explanations (stemming from increases in the supply of adult labor that is substitutable for teens) appears to explain more of the recent decline than do labor supply explanations (which are related to increased emphasis on schooling and college)
One story would be that as middle-class jobs disappear, older workers become fast-food cashiers, and teenagers no longer are employed. Smith gives a nod toward that story.
Smith's figure 8 does a nice job of describing the extent to which the actual youth employment rate (now 45 percent) falls short of what one would have expected based on cyclical factors allone (over 55 percent).
This is a classic case of a phenomenon where everyone will be tempted to offer their own explanation, based on anecdotes or a priori reasoning. I would caution you that Smith did actual empirical work, so it might be worth forming a judgment of the reliability of his results before jumping in with your less-substantiated theory. If nothing else, read the paper in order to make sure that he has not already tested your pet theory!