Karl Smith - Assistant Professor of Public Economics at UNC-CH & Blogger at Modeled Behavior 

Quickly tying together a bunch of threads: My general take is that neither GDP nor the Employment-Population Ratio is a stat we should care about for its own sake. By that same token, in-and-of-itself, I don't consider declines in unemployment from people leaving the workforce to better or worse than declines from people becoming unemployed.

There are lots of reasons, but fundamentally because these are a function of choices people make about their lives.

When lots of people chose to work, GDP will increase, the population-employment ratio will rise, and declines in unemployment will be dominated by folks becoming employed. When lots of people choose to retire, stay in school longer or stay home to raise a family the trends will reverse.

However, its not immediately clear that working is a better life choice than the other three options. Indeed, we generally consider the other three to be luxuries afforded by a wealthy society.

What matters is whether or not the labor market is functioning smoothly. If someone chooses to look for employment will they have hard time. If the answer to this question is no, and yet still few people look, then we have to conclude that from a macro-perspective things are going fine.

There may be micro level problems with inadequate skills, or oppressive gender roles or any other host of things. However, the macro-economy is doing its job.

With that in mind, I do want to point out a trend that could explain structural weakness in the economy from the production side.

Here is a look at employment in what we might think of as shopping center retail: clothing stores, department stores, general merchandise, books, music, etc compared to real retail sales

Despite the seeming acceleration in sales, this part of employment is rolling over.

Now look at the Employment Population Ratio and Unemployment rates for workers 16-19. Here I have re-normed unemployment so that I am graphing (60 = the unemployment rate) to match the early 90s. The implication is that if unemployment were zero then 60% of teenagers would be employed.

We see a large and growing gap between unemployment and the employment-population ratio. There are numerous micro explanations here.

One possibility, however, is that the relatively weak growth in shopping center employment relative to retail sales since 2000 and especially recently is driving down overall teen employment levels.

However, because teenagers are especially suited to shopping center employment they are dropping out of the labor force in response. That is, the End of Retail is causing a permanent shift in teenage employment because there are no substitutes for retail jobs.

This is a true structural downturn because it means that the production function is changing such that the productivity of teenage labor cannot meet the reservation wage.

When that happens a factor of production simply goes out of use. It also implies that for a time the economic gains from productivity enhancements will be muted. E-commerce means more efficient shopping but because we are not repurposing teenage labor but losing it completely, the measured gains are less than they otherwise would be.

On the other hand the non-measured gains to increased free time and -- one can dream -- increased school work are larger than we would have expected.