Perceiving Job Insecurity

A study in the Journal of Occupational & Environmental Medicine finds evidence linking perceived job insecurity in the Great Recession to poor health outcomes, even among workers who remain employed. The authors, Sarah Burgard, Lucie Kalousova, and Kristin Seefeldt, find that insecure workers--those that believe they are at risk of being laid off--are more likely to report poor self-rated health, symptoms of depression, and anxiety attacks. Sad, but not hugely surprising.

I originally intended to point to this study as yet more evidence of the harmful consequences of prolonged high unemployment. I intended, in particular, to write about how the anxiety and poor health consequences associated with the fear of losing a job must fall especially hard on people with low income. So I set out to gather a bit more data to back up that particular hypothesis, imagining it would be quick and simple task. Not quite.

Most of us are pretty aware of the unemployment rate in the U.S.--7.4% as of July 2013. But for people who do have a job, the more relevant statistic for their financial decision-making (and apparently also for their health) is the probability that they (and members of their household) will keep their job. This statistic is much harder to come by. How aware are workers of their risk of being laid off? How do you quantify job security?

The first place I looked was the Bureau of Labor Statistics, which provides data on layoffs and discharges. The monthly layoff and discharge rate for total nonfarm employment is around 1.3%. It peaked at 2% in early 2009 (see graph below). If we all believed we had a 1 or 2% chance of being laid off, we probably wouldn't be too stressed out about it. But the layoff and discharge rate does not directly translate into an individual worker's probability of losing a job, and it definitely does not translate into a worker's perceived probability of losing a job (the statistic most relevant for their health).

How can we get at people's perceived job insecurity? One way is to ask them. The Michigan Survey of Consumers asks survey participants, "During the next 5 years, what do you think the chances are that you (or your husband/wife) will lose a job that you wanted to keep?"

Broken down by income tercile, here is a graph of the mean responses. What initially surprised me the most is that the lowest income tercile has the lowest perceived job insecurity. In 2012, on average, people in the lowest income tercile reported a 17% chance of job loss, while people in the middle and upper terciles reported 19% and 20% chances, respectively.

Mean perceived chance of job loss by respondent or partner in next 5 years, by income tercile. Source: Carola Binder with data from Michigan Survey of Consumers. Moving-average filtered.

When you look at the distribution of responses, however, it becomes clear that you have to interpret the mean with a large grain of salt. Respondents are allowed to say any number from 0% to 100%. But they mostly just say one of two numbers: 0% or 50%. This is a common tendency across income levels, but especially among the lowest income tercile. In 2012, around 70% of respondents in the lowest income tercile chose 0% or 50% as their response. In the middle and upper income terciles, 58% and 47% of respondents chose those responses.

Percent of respondents who say that their chance of job loss in next 5 years is either 0% or 50%, by income tercile. Source: Carola Binder with data from Michigan Survey of Consumers. Moving-average filtered.

Prior to answering the question, survey takers are given this brief intro to help them understand probabilities: "Your answers can range from zero to one hundred, where zero means there is absolutely no chance, and one hundred means that it is absolutely certain. For example, when weather forecasters report the chance of rain, a number like 20 percent means 'a small chance', a number around 50 percent means 'a pretty even chance,' and a number like 80 percent means 'a very good chance.'" Nonetheless, most people seem to have tremendous difficulty quantifying their probability of job loss. Over half of people choose 0% or 50% as their response.

Whether or not you will lose your job can be represented by a bernoulli random variable. A bernoulli random variable is summarized by its mean (p). The Principle of Insufficient Reason, or Principle of Indifference, says that "if we are ignorant of the ways an event can occur (and therefore have no reason to believe that one way will occur preferentially compared to another), the event will occur equally likely in any way." This principle was discussed by Bernoulli, Laplace, and Poincare, among others. For a bernoulli variable, this principle says that if we are totally ignorant about its mean, our prior is that the mean is 0.5. This has a corresponding result in information theory: the entropy of a bernoulli distribution is maximized when p=0.5 (think of a 50% chance as being the "most uncertain.") If we have absolutely no information about how likely we are to lose our job, we might just guess that we have a 50% chance of losing it.

Keynes himself summarized the Principle of Indifference in his 1921 Treatise on Probability as follows:

"if there is no known reason for predicating of our subject one rather than another of several alternatives, then relatively to such knowledge the assertions of each of these alternatives have an equal probability" (pg. 52-53).

Keynes was one of many to critique this principle. His views on probability and uncertainty remain controversial, as does the Principle of Insufficient Reason. There is actually quite a large body of literature in statistics concerning "noninformative priors" that continues to study the fascinating and controversial issue of how to represent ignorance. There are also subfields of behavioral economics that study how people treat probability, particularly when it comes to low-probability events (like job loss, usually).

This post doesn't have a real conclusion, just some open questions. What do people do when they don't know their chances of having a job in the future? Do people "underplan" or "overplan" for the possibility of job loss? Would people be better off in general if they could estimate their probability of job loss more precisely? How would you readers estimate your own probability of losing a job in the next 5 years?

Jobless recoveries - mystery solved!!!

Bill Gavin and Menzie Chinn solve the mystery of the "jobless recoveries" we've been having since the early 90s:

In the earlier postwar recessions, the unemployment rate began to fall very quickly once the expansion began. By contrast, the unemployment rate continued to climb even after the recovery had begun for the 1990-91 and 2001 recessions. No one is predicting a rapid drop in the unemployment rate this time around, either...

Bill called my attention to the contribution of temporary layoffs to this changing behavior in the unemployment rate. He noted that the Social Security Amendments of 1958 explicitly exempted unemployment insurance from income taxation, and recalled a 1976 paper by Martin Feldstein which proposed that this gave firms a strong incentive to use temporary layoffs in response to a business downturn. By temporarily laying workers off rather than asking them to work shorter hours, the firm could deliver maximal after-tax compensation to its labor force, intending to hire those same workers back as soon as business improved. Temporary layoffs accounted for up to a quarter of those unemployed at the worst of the 1973-75 recession.

Bill believes that the key developments that changed this dynamic were the Revenue Act of 1978, which subjected unemployment benefits to partial taxation under the income tax law, and the Tax Reform Act of 1986, which made unemployment benefits taxable as ordinary income. Since the mid-1980s, the above graph shows that temporary layoffs have become a much less important feature of recessions...

If you subtract temporary layoffs from the number of unemployed, here's what the adjusted unemployment rate would look like. The earlier recessions look much more like the recent jobless recoveries.

Now THAT is good economics work. Simple, empirical, and predictive.

Perhaps we should consider once again exempting unemployment benefits from taxation.

You have been replaced by a machine

Here's an interesting blog post, which brings up the ominous possibility that our current huge level of unemployment is not "cyclical" (i.e., due to the financial crisis and big recession), but rather "structural" (i.e. due to jobs vanishing permanently). The basic idea is this: over the past 10 years, computers have become good enough to make whole classes of jobs (e.g. travel agents) obsolete. However, it's hard to fire a ton of people when times are good, so companies waited until there was a big recession to get rid of people whose jobs had been rendered obsolete by new technology. Now, those laid-off people can't get new jobs unless they learn an entirely new skill set. But that takes time and money, and many of those people are old already, and meanwhile technology is changing so fast that by the time they learn something new, that new thing might be made obsolete pretty quickly as well.

This is a strong argument for implementing a national-level government policy to help people retrain for new industries.

As a side note, at the bottom of the post is a link to a really excellent paper by my advisor, Miles Kimball, which shows that improvements in technology - which "neoclassical"/"RBC" economists claim are the cause of booms - are actually a cause of recessions. Booyah!!