Having examined the impact of computers on wages the logical next question to ask is what this means for the returns on capital. Here is the TL;DR version: higher returns in the short run and lower returns in the long run. We are already well into this short run period and we are beginning to see the forces that will shape the long run.

Let’s start with the short run. As we are substituting computers (which I mean broadly here including robots) for labor we are increasing the importance of capital as a factor of production. That in and of itself should increase the returns to capital. One place to see this direct effect is in the value creation in the information technology sector of the economy, i.e. the suppliers of hardware and software. For instance the NYSE Arca Computer Tech Index is up nearly 9x since 1995 for a CAGR of 12%.

This effect, however, is significantly amplified by the shift in the bargaining position of labor that I described previously. Unskilled labor has been pushed to its reservation price, skilled labor is receiving its marginal product and all the value creation is being split between top management and capital. The best way to observe this is by looking at the investment returns in companies such as Walmart and Amazon that epitomize this change in the retail industry (both up over 200x for Walmart since the early 80s and for Amazon since IPO).

Now one might legitimately say that by looking at stock market returns on companies or individual sectors that are performing well I am cherry picking. But even the very broad averages are up tremendously. For instance, the S&P 500 is up nearly 20x since 1980 for a CAGR of 9% compared to an annual growth rate of 4% for the previous 25 years.

In the long run, however, the impact of computers is highly likely to reduce the rate of return on capital. Why? Because capital is completely fungible and there is a lot of it globally. As information about investment opportunities flows more easily around the world, capital will compete driving down the rate of return.

Early stage investing is a great example of this. Pre-internet it was virtually impossible for a startup to explain to more than a few dozen people what it is doing in a cost efficient manner. Today a company might attract angel investors from all over the world based on their web presence or through a site like AngelList. Crowdfunding projects routinely now raise hundreds of thousands and even millions of dollars. Over time this will compress returns for startup investors at all levels, including traditional venture capital firms.

This raises the interesting question about what the long run expected rate of return should be. This depends to some degree on the risk of the investment opportunity, so the more precise question is what determines the long run risk free rate of return. The nominal risk free rate of return is bounded below by zero. Other than exceptional circumstances (such as the risk of forfeiture) nobody would agree to lend out money only to get less of it back in the future. The actual risk free rate — meaning no risk of repayment — is then rationally bounded below by the expected rate of inflation.

When lending at the expected rate of inflation though the real return is still zero (averaging over time). What force might result in a positive expected real risk free rate of return? Time preference. If a lot more people would like to consume more today than they can presently afford they will want to borrow from those who are more patient (and/or have accumulated more capital).

In upcoming posts I will have a lot more to say about both inflation and time preference as computer technology is having profound impacts there as well. I will argue that both are declining which eventually will push down the real return on capital.