This isn't really related to finance, but rather analytical methods generally so I think it is worthwhile. http://seekingalpha.com/article/220327-summer-employment-it-s-all-about-demand
The University of Chicago economist Casey Mulligan offered this graph to somehow demonstrate that employment is nearly entirely a function of the supply of workers available:
As someone who worked in retail in the summer for four years in a row, I can vouch for how this works and it is shown in the data as well. Retailers, probably the single largest employer of teenagers, start posting job openings around early June because during the summer, compared to January through May, sales pick up considerably. This is particularly true with the back to school season. The same traffic patterns hold for restaurants and movie theaters where summer business is better than late winter and early spring. Now, why does this translate into such a pick up in temporary employment for teenagers exclusively? That's because during most of the year the available labor pool that would fill these summer jobs is of a stable mix of age groups, but in the summer the marginal new workers are disproportionately teenagers. As such, for a set amount of hiring, teenagers are likely to be a larger proportion of that labor force.
Now, that doesn't squelch what Mr. Mulligan said yet. What does is the fact that his ratio of teenage employment to their next nearest age cohort is shifting down throughout the course of this recession. Sure, it holds the same pattern largely because the same seasonal sales patterns are there, just at lower levels. Now, what would we expect to see in terms of marginal new hiring with depressed activity? Let's say I am a store chain with 100 stores and normally I would hire 2,000 seasonal workers compared to an existing staff of about 20,000 (these are big stores for the sake of argument). I do this with my sales at an indexed level of 100. However, if my sales still follow the same pattern but are at 95 the next year, it doesn't make sense to add as many seasonal workers since my existing staff can probably deal with the lower sales volume. As such I might only hire 1,500. The next year is even worse and sales drop to a level of 90 so I cut seasonal workers to 1,000. All the while, I keep proportionately more of my experienced either full-time or consistent part-time workers since their productivity is much higher and they are less strained with the lower sales volume.
This is indeed what we have seen and Mulligan's graph demonstrates it quite well. The proportion of teenage workers to other workers has dropped with demand as there is less overall hiring of teenage workers, which is a function of what retailers have been seeing for sales. This applies to the other service industries too, but I chose to focus on retail. In a situation where demand is driving the decline, you would expect the most marginally attached workers to be affected the most. Indeed, that is what has happened.