Do Institutional Investors Raise Housing Prices?

Parsing the evidence

The current president has recently announced that he wants to ban large institutional investors from buying single-family homes, which he blames for increasing housing prices. As is often the case, it is not clear what legal authority he is basing this on, and I expect that nothing much will come of it. However, it does raise some interesting questions. Lots of people have blamed institutional investors for raising home prices, although economists have pushed back simply on the grounds that they own very little of the housing market. Are they right? Is ownership by institutional investors a contributor to the rising price of housing in America?

I believe the answer is no, not really. It appears to lower the cost of getting housing through lowering rents. The best work, when one takes a discerning look, is united on this point.

These investors, if you define investors as any entity which owns more than 100 homes (and, I believe, is not a builder), control only 2% of the market. Institutional investors control 0.5% of family homes, and Blackstone (which is particularly famous, if partly because everyone confuses it with Blackrock and vice versa) owns 0.06% of single family homes. Large institutional investors began entering the market in 2012, largely in the Sun Belt. Accordingly, most of the papers in the literature use data from the Greater Atlanta metropolitan area.

What makes it particularly interesting is that our usual methods of finding the effects are unsuitable. The go-to empirical method would be differences-in-differences, where we compare treated units, which are the homes which the institutional investors buy, with those that they do not buy. We assume that if they had the same trends in price before the event, they are both affected by the same data-generating process, and would have continued on the same path absent the intervention. This is generally supplemented by some reason to believe that the only event which mattered was the buying of homes – perhaps legal restrictions in some jurisdictions but not others, or the number of houses bought in a city is related to their exogenous pre-existing presence, or their expansion into different cities was unrelated to market conditions (after controlling for population and income), and many other possibilities. You may not believe these in all cases, but they are suitable for answering some questions.

However, differences-in-differences is obviously not suited for the effect of investors. One of the major tasks of investors is to predict which prices are going to rise. Unless the researcher can guess the exact channel through which they are making predictions, then pre-trends are pointless. They could be timing the market, or they could be causally increasing prices. You would consider it absurd to say that a hedge fund buying stocks is what was responsible for the stock price rising! This essentially rules out papers like Brian An (2023), which is just applying two-way fixed effects and hoping that’s enough. (Two-way fixed effects, by the way, is for each place, adjusting for the average over time; and then for each time period, adjusting each observation for the average in that period. The idea is to get rid of anything which affects a place evenly and everything which affects a time period evenly). Many papers are also taking the wrong outcome variable. They are concerned with homeownership, but that is not an economically interesting variable. The point of buying a house is so that you then receive a flow of housing from it, which can also be accomplished by renting. These have advantages and disadvantages, and moving from one to another is not strictly better.

We need to go one step deeper, and estimate a model. These models can be boiled down into two basic components. On the one hand, large investors may well have economies of scale. They might hire maintenance people who work for them, and thus have an adequately suited number of tasks, rather than having to contract out to workers who charge a premium for not being guaranteed work, or be better at managing properties with software, or perhaps be a more trustworthy borrower and pay less for capital. On the other hand, them owning lots of properties may increase their market power, and thus they price the homes above what is optimal. It is not ex ante obvious which effect prevails – you need to write down a model, and calibrate it to the data.

I think Joshua Coven (2025) has the best paper on this, with Barbieri and Dobbels (2025) also being excellent. They both agree that institutional investors lower rents, but only Coven has anything to say about home price. Zhichun Wang and Daojing Zhai (2025) is able to pinpoint the mechanisms by which this occurs. There is a disagreeing view from Sebastian Hanson (2024). It is a fine paper, but is, in my view, wrong. It is really examining something quite different, and so should not be cited in this context.

Coven first. He solves for the problem facing households over where to live, and over what type of housing to live in. The parameters he cares about are travel costs (directly measured), an aggregate supply response (borrowed from Baum-Snow and Han, 2024) which captures how much local builders respond to a price increase, and the price elasticity, which is identified with instrumental variables. Specifically, we’re finally able to use the original BLP characteristics instruments in a believable way. The idea is that the topographical features of an area say 8 miles away affect the demand for that area by being a competing option you could switch to, but not through affecting the enjoyment of the area directly. This is similar to what Berry-Levinsohn-Pakes did in their original 1995 paper, where they used the characteristics of other automobiles as an instrument. What’s funny to me is that approximately no one believes BLP 1995 as originally written, because the characteristics of competing projects are very obviously endogenously chosen in a reaction to the characteristics of other products in the market. Now we’ve wrapped back around to it working! (Sorry for nerding out, I just thought this was really cool).

Anyway, once we have the parameters, we can simulate what happens when three large institutional investors enter the market. Homeownership falls, because the investors convert to rental housing. Housing prices increase, while rents fall. The net impact is that the cost of housing fell.

There is also Felipe Barbieri and Gregory Dobbels. They find the same results, approximately speaking, as Coven, but is taking a different tack. Coven is focused on the extensive margin, with landlords setting quantity and taking price as given. Barbieri and Dobbels are instead fixing the housing supply, but having them set prices in a Nash-Bertrand game. The tradeoff to setting your price too high is that it means it must spend more time on the market, which is costly. Home prices increase, but rent prices decrease. (Hilariously, they refer to this as home-sellers benefitting. Coven refers to this as home-buyers losing!).

Why might this be happening? It’d be pretty cool if we could pinpoint that these returns to scale are coming from, and that’s what Wang and Zhai do. Most of the purchases are extremely concentrated, with 65% of units having more than 25 units within a one mile radius. This is not the scale at which market concentration really matters – people have lots of options for locating within a city, and can totally avoid a few blocks. What it does suggest is that maintenance is getting made way cheaper. However, their estimates are so small as to be economically trivial. I am also not sure how much anyone can really comment on dynamic discrete games, other than that they’re pretty darn cool, and they’re also totally opaque.

Sebastian Hanson (2024) comes from a finance background, and so speaks a slightly different dialect of economese than I am used to. The institutional investors have access to cheaper capital, and so enter to buy homes when lending constraints become tighter due to interest rates changing. Thus, the changes in housing prices are purely due to market timing – when an interest rate shock passes, housing prices tick back up to where they would have been in the absence of the interest rate shocks.

I think the two papers can be reconciled by considering the particulars of the model. Hanson’s model features no endogenous supply response. Housing supply is surely affected by price, so this is an enormous omission. Nor can it be argued that this is an inelastic market with harsh zoning restrictions, so the omission is especially glaring.

I do not think that the argument that it is only a small part of the market is a very good one. What matters is not national concentration, but local concentration. This point has been made again and again, but we really must internalize it. If you care about, for example, beer, going from one beer from a different company to three companies which sell nationally is a massive decline in the number of national competitors, but a massive increase in actual competition. In the markets where they enter, institutional landlords have had a positive and substantial effect.