Thursday, March 19, 2015

Idea for a new economic indicator

This post was inspired by, but is not directly related to, this article.

When talking about whether the population as a whole is making economic gains or losses, people often talk about middle class vs. low income vs. high income, or they look at average or median incomes for the population as a whole and for various demographics.  Less often, but sometimes, they talk about the ratio of income to tuition to housing prices. (The Globe and Mail has a useful comparison tool.)

It occurs to me that another useful indicator would be to look at changes in income over time with people who bring various levels of education, skills and experience to the table.  For example, how has the income level of a person with an undergraduate degree and 10 years of work experienced evolved over the years?  What about a newly-minted Ph.D.?  What about a student working their way through college?  What about people who have been freelancing for 5 years?

It might be useful to get somewhat specific (Is the person with an undergrad degree and 10 years of work experience a translator or a teacher or a computer program?), but the data would cease to be comparable if you got too specific (I don't know how informative it would be to track the income of social media specialists or FORTRAN programmers over decades).

If the data is available, it would also be interesting to track negative factors.  How has the income of people who were laid off one year ago evolved?  (i.e. were they more or less likely to get new jobs within a year in previous decades?)  What's the situation of people who started a business within the past two years?  What about people who are involuntary entrepreneurs (i.e. they didn't want to start a business, but couldn't get hired)?

I think this would fill in some blanks, and it has the potential to draw attention to certain problems that may be hidden by the other, more commonly used indicators.

3 comments:

Lorraine said...

Here's a bit of econometrics which is easily technically feasible. Constraints on its possible implementation, if any, are legal (and of course political), not technological.

Let's say you're the IRS. (I'm assuming Revenue Canada is in a similar situation). You have a record of all the 990's (employer quarterly withholding tax statements filed in the United States). In short, you have a database table with the following columns (as well as a few others):

EIN (employed identification number)
SSN (social security number)
gross pay (amount of wages earned by holder of SSN in the employ of holder of EIN)

Without exposing SSN's, imagine some of the statistical queries that would be possible. In particular, I'm thinking of plotting Lorenz curves. In a Lorenz curve the X (horizontal) coordinate of a point on the curve represents an income percentile for the population of a country. The corresponding Y coordinate represents the percentage of the sum of all personal income that was earned by all the people at or below that percentile ranking. The curve will bow downward from a straight diagonal line. How far downward is a fairly objective indicator of the amount of income equality. How about plotting the Lorenz curve (and calculating the Lorenz Gini?) of each employer? This data would be catnip for the "vote with your dollars" types.

Note that, thanks to moonlighting, SSN's and EIN's (humans and businesses) is a many-to-many relation. So another data analysis project might be plotting a Lorenz curve of the total (possibly across multiple employers) earnings of each employee of a particular employer. It certainly wouldn't surprise me if maintaining a mostly part-time payroll is part of the business model of many companies. It would be interesting to see how many of these part timers nevertheless have a gainful income (if not benefits package) through moonlighting. If it's true the things many people say about employers in the retail/restaurant occupational ghetto jerking (mostly part time) people around on short notice concerning shifts, hours and scheduling, then moonlighting opportunities are out of reach of many and income potential is limited to a single part time job.

This methodology has a few serious flaws. It doesn't capture non-wage income, and I suppose there must be a few people who hold down a J.O.B. in spite being independently wealthy. More importantly, it doesn't capture offshored labor. I think ratio of a company's income to its total domestic payroll would be a good indicator, though, and again, IRS should in theory have the full dataset (and not need to settle for a random sample).

laura k said...

Studies are done periodically on the kind of data you're suggesting, by orgs like the Pew Foundation. But as a regular indicator, it would be very useful.

impudent strumpet said...

@Lorraine: Bonus points for explaining a graph understandably using just words! And that would also be a very informative indicator.

@Laura: I'm glad people do the studies, at least that means the info is available somewhere.