Software Is Eating the American Dream

Friday, May 22, 2015

Software is eating the world, and maybe people’s jobs and wages too.

I’ve been thinking a lot recently about the distribution of wealth in the US. There’s a general feeling that the rich keep getting richer and the middle class is stagnating, while the poor are completely out of options to move up. An oft-cited statistic contrasts the growth in total wealth vs the stagnation of wages in the past decade, unlike prior economic growth.

Let’s split people into three categories. Bucket one: people who make no income whatsoever. Bucket two: people who make income primarily in the form of wages (regardless of how much). Bucket three: people who make income, but not primarily in the form of wages (essentially, investors).

In general, bucket three has had a great amount of growth in the past decade. The people in this bucket derive income from return on investment - they sink money into a resource and reap the benefits from the output of the resource. Elementary macroeconomics defines two classes of resources: labor (specifically, people in bucket two) and capital (everything else).

I think the big takeaway, the big difference between recent economic growth and prior economic growth, is that this time around the growth has almost completely been fueled by return on investment in capital instead of labor. Why? Because we’ve been hard at work making all sorts of technology (capital investment) that replace people (labor). This is a huge difference from before, where capital investment instead usually worked to improve the productivity of labor, or encouraged investment in more labor.

We can even split this up into two classes of prior innovation: Things like mechanical tools that make workers more productive, or innovations like the assembly line that make labor cheaper. Tools lead to the creation of a skilled labor class, and commensurate wage increases (giving bucket two more capital, potentially allowing them to move to bucket three). Innovations like the assembly line, while reducing the value of the individual laborer, lead to an increase in demand for labor as labor became so much cheaper (enabling bucket one folks to move to bucket two).

But the kind of technology we are creating today (specifically, automation) does neither. An algorithm to automate trading does not create a skilled labor class in the field of finance. A self driving truck does not increase demand for labor, skilled or unskilled. And it’s really difficult to figure out what to do in this situation when it comes to dealing with its impact on wealth distribution.

For people in bucket three, investing in this technology is a no-brainer: if you can get more from a relatively cheap one time investment in software, why continue to pay for labor? But the solution can’t be to force people in bucket three to invest in labor, unless we want to trade total economic growth for higher wages or more jobs. I don’t think that’s the direction we want to head in.

This leaves people in bucket two in a rut. While their jobs are safe from automation for now, it doesn’t open up pathways to further skilled labor (with the possible exception of developing more automation). The usual cure here of “further education for higher wages” doesn’t apply if there’s no increase in demand for skilled labor. People in bucket one are completely lost, as the entry level jobs get even more and more automated and demand for cheap labor contracts (or, more realistically, temporarily moves and eventually contracts).

I don’t really know what’s the way to correct this situation. Should we encourage everyone to become software developers? Should the government encourage more investment back into technologies that increase investment in labor? How would it even do so? What would that look like? One possible solution is straight-up wealth redistribution - we don’t have to worry about what bucket we’re in (beyond questions of self worth) if every bucket has access to a “comfortable” lifestyle. But that requires a ton of buy in from the people in bucket three, which is unlikely to happen. So where do we go?