The older an industry's firms, the fewer workers it loses. In Korea the link is strong.

Two industry columns from the pension register, plotted against each other, produce one of the tighter relationships in this data: the older an industry’s workplaces, the fewer of its workers walk out the door each month.

The relationship

Take the 105 industries with enough workplaces to measure reliably. For each, compute the average age of its firms and its monthly separation rate. Line them up and the correlation is -0.61 — strong, negative, and consistent. Old-firm industries are low-churn industries.

The endpoints make it concrete. Domestic banks average 33 years old and lose workers slowly. Construction subtrades average two to three years old and churn near the top of the range. The whole span from banking to earthmoving traces the line: as the typical firm gets younger, the typical job gets shorter.

Why age predicts stability

The link is not a coincidence of two lists happening to rhyme. Both are produced by the same thing — how the industry is built.

• Institutional industries — banks, credit unions, established manufacturers — are hard to enter and hard to exit. Their firms accumulate age because they persist, and a persistent firm holds its workers. Age and low churn are two readings of the same permanence.
• Project and open-entry industries — construction, cafes, facility services — form and dissolve constantly. Their firms are young because the model is short-lived, and short-lived firms shed workers fast. Youth and high churn are, again, the same fact seen twice.

So firm age works as a leading indicator of job stability. You can look at how old an industry’s workplaces are and predict, with a -0.61 pull, how likely its workers are to still be there next month.

The limit of -0.61

A correlation of -0.61 is strong but leaves room. About a third of the variation in churn lines up with firm age; the other two-thirds is everything else — how seasonal the work is, how tight the labour market is for that skill, how much of the “separation” is really workers moving up to larger firms in the same trade. Age explains a lot, not all. An industry can be young and stable, or old and churning, and sit off the line; the relationship is a tendency, not a law.

What we did not claim

Firm age here is enrolment age, censored at the pension system’s 1988 start, so the oldest industries are understated — which, if anything, compresses the relationship rather than inflating it. Separation is not layoff: some of the churn in young industries is workers being promoted out of small firms into big ones, not losing jobs. And this is one month, unadjusted for season. The direction and strength of the link are the claim; the mechanism behind it is an argument this file supports but does not settle.