The mechanism I’m looking at is an increase in transferability of on the job, task- or product-specific knowledge. Its easier these days for workers to transfer their job-related knowledge from one job to the other. Increasing transferability increases the number of workers that change jobs (because costs of switching are decreased and because the present value of the knowledge gained in the new job increases) and this has two effects. First, it decreases inequality because workers can more easily move to jobs that have become more productive (maybe because of job-specific technological advances). Second, workers lose some knowledge when they switch jobs (knowledge isn’t perfectly transferable) and so it increases dispersion of wages at any given point in time. It turns out over the last couple decades the second effect dominates the first so with increased transferability of knowledge, we’ve seen increased wage inequality. (I’m currently working on quantifying this effect of transferability on inequality.)

This is a sectoral shifts mechanism similar to the one Kuznets proposed to derive his famous curve. I just have a parameter that makes it easier to move sectors (jobs in my model) and the high productivity sector (Kuznets’ cities) changes randomly over time.

Normatively, there’s nothing bad about increasing transferability of knowledge. Personally, I think its a good thing. It feels good to be an expert in something. So the wage inequality caused by it shouldn’t be a concern. The policy implication is that income equalizing policies will retard the efficient allocation of workers across jobs. You can have income transfers but they shouldn’t be correlated with income differences that have to do with transitory differences in job knowledge. My inner technocrat says we should subsidize job switches too, but I don’t think we know enough about the knowledge production function and I’m sure there are perverse side effects (like people switching jobs that shouldn’t).

Also, in the future the first effect may start to dominate the second effect. In this case increased transferability of knowledge will lead to decreasing inequality.