Larry Summers and Brad DeLong have even more than usually stimulating thoughts about secular stagnation and extremely low interest rates. I notice a strong focus on non residential fixed capital investment.
disclaimer: I owe debts I can never repay to both of them, so this post might even be civil.
Summers responds to Marc Andreesen on secular stagnation. The post is rather brilliant (no surprise there). In particular, I very much liked and strongly agree with (the parts I understand) of
the evolution of productivity growth is not essential to the argument about whether equilibrium real interest rates have declined. This decline could have occurred for many other reasons – including increased foreign saving, lower priced capital goods, slower labor force growth, increased demand for safe assets, more burdens on financial intermediation, and lower rates of inflation.
I will discuss why I like Summers’s list of possible causes of secular stagnation after the jump, but here I just note that it is appropriately long. A model adicted economist would look at one possible explanation and assume away all the others. In fact the point (if any) of this post is to add another explanation — lower demand for housing. I note that the second and third explanations focus on non-residential fixed capital investment.
This passage also clearly focuses on non-residential investment
Marc and I agree that we are headed into a period of soft real interest rates, where there will be more money available than great deals. This may, as he suggests, not be all bad; as it will make it easier for risky ideas to get funded.
Brad too doesn’t mention houses. For example, here is his first concern about problems with low interest rates
safe interest rates expected to be very low for a long time artificially boost the value of long-duration assets–so capital is misallocated and we wind up with a capital structure that has in it too many long-duration relatively-safe projects that make at best very small contributions to societal well-being.
Emphasis mine and he doesn’t mean housing project.
Yet the latest bubble was in housing and the components of aggregate demand which have been strangely low during the long disappointing phase of the recovery (which phase might be over by now) were G and residential investment.
Standard macroeconomic models don’t include a housing sector. Compared to most economists DeLong and Summers are not at all model addicted. Yet even they ignore housing when discussing investment in general and long run growth and such.
I can think of two more explanations. First lower population growth causes much lower housing investment — houses last a long time so a whole lot of gross housing investment is expanding the housing stock not replacing depreciated and torn down houses. So far, I can see why houses factories and office parks are lumped together — the issues are basically the same.
Second and again, maybe there has been either bubbles or extremely low real interest rates since the productivity slowdown (starting 1973). That is maybe the housing bubble has lasted for decades and the generally recognised housing bubble post 2000 was just more extreme. People have long believed that houses are good investments because the relative price of housing trends up. Shiller has long argued that they have been mistaken. If he is right, demand has been high because of irrational expectations of price appreciation for a very long time. The argument that, during bubbles, asset prices must shoot up and become clearly crazy so the bubble can’t last is based on the standard assumption that investors know the full history of prices. This is true of stocks but definitely not true of houses. Again if Shiller is right, the current low forecast of house price appreciation and current low demand for housing might be the new normal.
Finally US housing demand has included the demand for huger and huger houses. The trend will slow if the desire for more and more floor space is satiated. This is an issue related to preferences and quite different from any discussed by DeLong or Summers.
There is a similar issue related to consumption and savings. The suspicion that inequality leads to secular stagnation is based on the idea that the super rich are satiated — that they can’t possibly consume a large fraction of their huge incomes basically because they don’t have the time. Standard assumptions about utility functions are made so that there can be unbounded growth at a constant rate with a contstant balanced growth real interest rate. It is perfectly possible to write a model in which growth stops because consumers are satiated.
Finally, back to housing
Some of the explanations on Summers’s list of possible causes of secular stagnation are obvious. For example “increased foreign savings” clearly can cause low US demand by causing low US net exports. In fact during the period when the US seemed to need bubbles to keep aggregate demand high enough for normal unemployment (at least the past 20 years) the US has always run huge trade deficits. US absorption (that is demand) has been greater than US production. The problem of low demand was imported. Slower labor force growth clearly reduces investment for a given required return on capital — it is the Alvin Hansen’s original explanation of possible secular stagnation. Higher demand for safe assets will make the requires safe real rate lower for the risky rate consistent with adequate investment. Increased costs of intermediation imply a lower safe rate paid to savers for the same cost to investing firms.
One is not obvious or conventional and very interesting.
I can understand the role of lower priced capital goods (an old topic for DeLong and Summers). If firms want the same normal ratio of capital to labor and capital is cheap compared to consumption, the value of investment in terms of consumption goods will be low so value of investment will be low compared to the value of savings. If extremely low interest rates are required for a higher than the old normal ratio, then they will be needed to make I=S. I think this story doesn’t work if the production function is Cobb-Douglas and that it requires an elasticity of substitution of capital and labor less than one, but I personally am sure that elasticity is less than one.
I don’t understand why inflation would cause a lower natural real rate of interest.