Monday, July 14, 2014

Housing, finance, and the macroeconomy

That is the title of a new NBER working paper by Morris A. Davis and Stijn Van Nieuwerburgh, and I think it is going to be a chapter in something--the Macro Annual, or a handbook, or something. In my view, these two have done some of the best work on housing; see Davis' stuff here and Van Nieuwerburgh's stuff here.

The paper is really, really good.

It is a survey of the macro literature on housing. The sections of the paper are

  • stylized facts
  • housing and the business cycle
  • housing over the life cycle and in the portfolio
  • housing and asset pricing
  • the housing boom and bust and the Great Recession
  • housing policy

There's something for everyone (well, almost; I'll discuss below). Most of the sections describe a simple model that characterizes the literature on that topic, discussing the model's interesting implications and shortcomings. The paper covers a lot of ground, so it doesn't lend itself well to summarizing. Go read it if you want your mind blown.

Here's a slice:

Housing is not only an important asset in the portfolio, it also has several features that make it different from investments in financial assets. First, it is illiquid in the sense that changing the quantity of housing may take time and/or require incurring substantial transaction costs. Second, it is indivisible: A limited assortment of types and sizes are available for purchase at any time (including a minimum size). Third, home ownership and housing consumption are typically intimately linked. Most households own only one home and live in the house they own. Fourth, housing represents the main source of pledgeable capital against which households can borrow. Investment in housing is much more leveraged than investments in other financial assets and the value of owned housing limits the amount of leverage in households' portfolios. Fifth, housing is tied to a particular labor market: People usually live where they work. (24)

I wonder how many people realize just how weird housing is. In a previous post I wrote this:

For most people, an owned house is a massively concentrated, highly leveraged, totally undiversified bet on one asset class (real estate) in one geographical region. It's a long-term bet on the local labor market and natural environment. It may be a long-term bet on the owner's job match or occupation. The home purchase includes a bundle of local amenities--school district, voting district, neighbors, public administration, commute, etc.--and the new owner is making a bet about the outlook for that bundle as well. 

The issue of concentration and asset class is an obsession of mine. Say Davis and Van Nieuwerburgh:

Renters and owners choose substantially different portfolios of financial assets, highlighting that conclusions drawn about optimal portfolio allocations over the life-cycle from models that do not include a rental/own housing choice may be misleading. (36)

It drives me crazy that when I read books about personal investing they rarely (if ever) mention housing as part of the portfolio allocation problem. Another point I've made in previous posts is that buying a house is buying a lifetime stream of rental inventory. That is, rather than paying for housing services as they arrive, like renters do, owners buy a massive flow of services all at once. Tenure is a pretty complicated decision! From the paper's discussion of tenure models:

Although housing is risky, driving down demand, current housing is a hedge against future housing demand shocks since price changes of housing units in the same market are correlated. . . The hedging demand dominates its inherent risk. . . . When households expect to increase their holdings of housing in the future, they buy a bigger home today in response to an increase in house-price uncertainty. If, instead, households expect to down-size in the future, they reduce their holdings of housing today in response to an increase in house-price uncertainty. (35)

What's missing?

My dissertation focuses on a question that is not covered in this paper: housing as collateral for entrepreneurship. Early in grad school, I was looking through some firm dynamics data and noticed that young firms were hit particularly hard by the Great Recession. Then I noticed that both housing and young firm activity started collapsing in 2006, leading the NBER recession date by between 9 and 21 months. I hypothesized that the collapse in the value of housing collateral could lead to a decline in entrepreneurship via a collateral channel. There are now some empirical papers finding suggestive evidence that housing and young firm activity are related (and I have some related empirical work in progress). See a summary here, see also here. This topic has received a lot more attention recently.

In 2012 I did some informal interviews with a handful of bankers. They all told me that housing collateral is important for young firms, particularly brand new ones. The bankers use earnings history to make decisions about many small business loans, but new firms have no earnings history. They must have collateral, and for many entrepreneurs a house is the only collateral lying around. One banker told me that once house prices started falling, he shut down lending to new businesses entirely. Others said that they started significantly discounting the value of housing collateral and tightened loan-to-value ratios. In short, the anecdotal evidence suggested that housing collateral mattered a lot for lending to young businesses.

I built a DSGE model to investigate the topic. In the model, there is a corporate sector but households can engage in production if they want. People can own or rent houses, and owned houses can be used as collateral for any kind of borrowing (including capital rental for your business). In early versions of the model, I took the house price as exogenous. That route taught me the limitations of partial equilibrium reasoning: when house prices receive no feedback from housing investment decisions, things can get pretty wacky. If people think house prices are about to fall, everyone can sell their house (or eat it, if possible), rent, and wait for things to bottom out, relying on cash from the sale to secure ongoing borrowing.* More broadly, general equilibrium matters for thinking about entrepreneurship. The opportunity cost of starting a business is often earning a wage; so you can get more entrepreneurship (at the extensive margin, at least) by doing things that destroy the labor market (which is consistent with some evidence; see Robert Fairlie's stuff). So the supply-side financial frictions that affected large firms can have an ambiguous effect on entreprenership generally. Housing collateral, on the other hand, directly affects firms whose balance sheets are tied to households.

You might think that building a model where entrepreneurs need collateral and housing happens to be collateral is like assuming the result. But the model doesn't have to deliver the result that lower home values reduce entrepreneurship. People could respond to the lower house prices by holding more housing (which is what happens, e.g., if housing enters utility Cobb-Douglas style), or by holding more financial savings. But, quantitatively, these options aren't enough. In model experiments, a lower house price is associated with less entrepreneurial activity. This happens despite the fact that there is an unconstrained corporate sector that can make up for lost output from missing entrepreneurs, so wages are not decimated and aggregate demand need not fall dramatically. So I can isolate the effect of housing on entrepreneurship without confounding it with a bloodbath in the rest of the economy. The paper isn't totally finished, but I think the model is generating interesting results.

Who cares? It matters if recessions that are accompanied by (or preceded by) housing sector collapse are likely to also see a collapse in young firm activity, since young firm activity is large compared with net job flows. A labor market is likely to recover from shocks more quickly if firm dynamics are robust; we don't want to have to rely on old firms as a group to generate labor market recoveries, since the gross job creation of expanding old firms is typically matched by gross job destruction of shrinking ones.

So I think its role as business collateral is another reason to care about housing.


*My partial equilibrium version of the model did teach me things, though. An exogenous house price lends itself to the interpretation of price as a technology parameter. A low house price means you can convert a few consumption/capital goods into lots of housing. In this sense, falling house prices are somewhat similar to rising TFP! This generalizes to the endogenous house price case, more or less, particularly if aggregate housing supply is somewhat inelastic. A lot of people are inclined to see falling house prices as all doom and gloom, but it's actually really good for people buying houses, and like any technology shock it can have positive spillovers for other people as well (though probably not positive on net, for homeowners). The technology interpretation also helps the case that model results for entrepreneurship are robust, since the lower price is making people better off in other ways.

1 comment: