Allison's ode to the second moment: fifth issue


Welcome to the fifth edition of Allison’s Ode to the Second Moment, a newsletter that highlights the (mostly manmade) risks we face.

He created a monster!

Are 401(k)s a monster that destroyed American’s retirement security? Ted Benna, a man who claims he brought 401(k)s into the mainstream, has some regrets. Hmm…I am not sure we can totally blame him. First of all, countries all over the world moved to personal pension accounts around the same time. As special as American tax law is---it did not create the demand for personal pensions. What did? The fact that Defined Benefit (DB) plans are expensive and put so much risk on corporate balance sheets.

Besides DB plans aren't so great anyhow. At their peak less than a third of Americans had them (compared to more than 45% of American workers covered by a pension account through work today). And they expose participants to the tail risk of a big benefit cut if their employer goes bankrupt—or is just seriously underfunded and threatens to bankrupt the PBGC. Defined Contribution (DC) plans have their problems, but they aren’t inherently bad—just poorly executed. Are you really better off with a DB plan? In my opinion:

Good DC plan>Bad DC plan>Bad DB plan

What about a well-run DB plan? Well that depends if you ever plan on changing jobs. If you are wedded to your employer for life then:

Good DB plan>Good DC plan>Bad DC plan>Bad DB plan

But if like most of us you do change jobs, then:

Good DC plan>Good DB plan≥ bad DC plan>Bad DB

Defined benefit plans just don’t suit a dynamic labor market where people change jobs more. Rest easy, Ted.

But wait a second---I thought people aren’t changing jobs any more.

There was a time when Americans took risks, moved jobs, locations, and started businesses. Now not so much.

The popular explanation is housing costs created gated communities of rich, coastal elites who have unlimited opportunities and are leaving everyone else behind. There is some merit to that argument. But the truth is more complicated

First of all, the decline in labor market fluidity (changing jobs) isn’t so obvious. It is true there are fewer quits and hires; but that does not tell us the full story.

In the fluidity heyday, the 1980s, more than 25% of American men had been at their job for less than 1 year. Lots of people took jobs, discovered it was a bad fit, and left. That became less common by 2014, when just 19.5% of men have been at their job less than a year. Meanwhile tenure over 10 years is less common. How is that possible? In the past, people took several jobs that didn’t work out and then settled into long-term employment. Now people stay in jobs 2 to 8 years before moving on.

Maybe that’s worse for Americans, but it also could be better. There’s not much benefit to staying in a job for under a year. Technology might mean better matching and more time in jobs could mean better skill development.

Why people aren’t moving to higher wage areas is also a tough question. A new working paper argues the housing bust did not keep people from moving. Perhaps we should look at the total cost of living and skill-based wages to understand what is going on.

What if Big Data turns out to be racist?

Everyone is excited about big data that will tell us what products to buy, movies to watch, and if we are hardened criminals. Economists call this a gain in efficiency. But an investigation at ProPublica uncovers that algorithms used to predict recidivism predict black Americans are more likely to reoffend. That is not totally surprising because black Americans have higher recidivism rates. Now there might be some issues with specification and bias with these particular models. But it raises some important questions about how we use data.

Let’s say black Americans have higher rates of recidivism because they are victims of a racist justice system. Now suppose we estimate algorithms (using the racist data because it’s all we have) and use them to justify sending people to prison. More time in prison turns people into hardened criminals. Now we’ve created a vicious cycle that further entrenches institutional racism.

There is no great way to control for racist data. Like any new innovation, big data may have some terrible unintended consequences. Before we go all in, we need more discussion about its appropriate use.

RIP Washington Neo-liberal Consensus

Always a step behind the enlightened Davos attendee, the IMF admitted maybe neo-liberalism wasn’t such a good idea.

The thinking man these days says, “Capitalism is just great. But maybe we can tweak it to make it a little nicer.” Maybe the governments can favor certain nice-sounding industries, it can limit capital flows and trade so there no adverse effects (at least in the short-run—but this is a static Keynesian world. Who cares about the future?). Perhaps debt generates enough growth it pays for itself and interest rate risk can be ignored (unless it’s the financial sector where leverage is very bad).

Neo-liberalism open up markets and brought scores of people out of poverty—but that came at a cost of more volatility and income inequality within countries (which the IMF says absolutely hampers growth).
Now, I know this newsletter is about the importance of higher moments. But even I have to admit sometimes the level matters more.

Maybe more planning will deliver the upsides of capitalism and fewer downsides. I call it taking the destruction out of creative destruction. Planning had mixed results in the past, but we are smarter and better than ever before—we’ve got big data.

Until next time, pension geeks!