Information Frictions in Real Estate Markets: Recent Evidence and Issues

The literature on the stochastic properties of price or return series is broad and has not been reviewed systematically since Gatzlaff and Tirtiroglu (1995). A selection of influential applications published subsequently includes Malpezzi (1999), Meen (2002), Capozza et al. (2004), Gallin (2008), and Schindler (2013). An updated survey is sorely needed. The search literature is the subject of an extensive review in Han and Strange (2015). Readers can refer to Duca et al. (2021) and Green (2008) for reviews on information effects on house price cycles and housing finance, respectively. Papers that examine information issues in land development include, for example, Konishi (2013), Fisher (2013), and Dong and Sing (2016). Reviews of behavioral real estate include Black et al. (2003), Seiler (2014), and Salzman and Zwinkels (2017).

Stigler (1961) is best known as the origin of search theory. It is seemingly the first published paper that applies marginal analysis to information, treating information acquisition as a transaction cost in a neoclassical framework (that ignores the non-rivalrous nature of information). While the most influential application of Stigler’s approach has been to job search, second place may belong to housing. Because the literature has developed as a distinct branch, separate from the standard topics in information economics that are our focus (i.e., adverse selection and moral hazard), we do not cover housing search in this review.

This result is formally developed in the Greenwald and Stiglitz (1986; 1988) theorem.

See also Grossman and Hart (1980a), Grossman (1981), and Milgrom and Roberts (1986).

For examples of the earlier works, see Yavas and Yang (1995), Chen and Rosenthal (1996), and Arnold (1999).

In practice, the listing price is not binding on the seller in the US. However, the seller may incur substantial costs if they fail to accept an offer at the price and terms stated in the listing agreement, because the broker’s fee is still earned and due in this case.

Houses are identical in Albrecht et al. (2016), and Carrillo (2012) assumes that the listing provides all relevant characteristics. While the seller does have private information about a property’s idiosyncratic features in Merlo et al. (2015), the behavior of buyers in response is not modeled.

The potential for vanishing omitted variable bias on the interaction term is an advantage of this design that does not appear to have been noted in the literature. See Nizalova and Murtazashvili (2016) for details.

Damodaran and Liu (1993) similarly use variation in sales price disclosure laws across states to characterize the information environment.

The research design in Garmaise and Moskowitz (2004) does not rule out alternative response channels. For example, an alternative explanation for the broker (market segmentation) result is that when information quality is low, brokers are more likely to draw on their professional trading network in acting as investors to reduce search frictions and/or draw a trading partner that is easy to work with. Literature in the section titled “Broker Trading Networks,” finds a heavy reliance on networks for commercial brokers in their professional activities. Those ties appear to be strong.

Regarding financing strategies, Garmaise and Moskowitz (2004) predict more seller financing, and less bank lending, when information quality is low for two reasons. First, such an environment increases the value of the informational advantage that sellers have over banks in lending. (Banks still have a tremendous liquidity advantage, of course.) Second, the Leland and Pyle (1977) separating signal model suggests that sellers in a market with asymmetric information will retain an equity interest in assets they are offering as a signal of product quality. Garmaise and Moskowitz contend that the security interest in a seller take-back mortgage is analogous to the equity stake. In contrast to their strong results for trading strategies, empirical tests in the paper mostly reject these hypotheses that investors use financial structure to mitigate asymmetric information.

The “low-rise” category refers to attached (row), semi-detached, and detached homes, and that high-rise properties comprise condominium or apartment units in multistory buildings. It may be that US buyers would perceive that what the authors call low-rises are more of a commodity product and high-rises are more differentiated. If so, it would be understandable if the perspective is reversed for cities in Asia, as maintained by the authors. To support their contention of greater heterogeneity, the authors document that low-rise properties constitute only 5% of the total housing stock in densely populated Singapore and transact at considerably higher prices during the study period.

The authors acknowledge the symmetric and asymmetric components may reverse in other contexts. Hong Kong is famous for its world-leading population density (prevalence of high-rise residences) and housing prices. Because fixing defects in the structural quality of high-rise buildings can be prohibitively costly, it is understandable that buyers there may be more concerned about latent defects in structural quality than locational attributes.

This involves taking the assessed value of the house, less an estimate of the depreciated replacement cost of the structure, following the cost approach from valuation. A critical assumption, which Clapp et al. question, is that values of land and structure remain additively separable after construction.

As explained in Lefcoe (2004), implied warranties of habitability and workmanship have completely supplanted caveat emptor (let the buyer beware) as the governing principle for new home sales in the US. These implied warranties apply not only to buyers, but also to the “buyers’ buyers,” subject to time limits for filing an initial claim under state statutes of repose (p. 209).

For a description of quasi-experiments and hedonic property value methods, see Parmeter et al. (2013).

Of course, a more substantial body of work examines the capitalization of school quality. Most studies use school attendance zone or district boundaries to identify capitalization rates. See Black and Machin (2011) and Nguyen-Hoang and Yinger (2011) for detailed reviews. Given the focus of this review, our coverage is limited to recent papers with research designs based on information shocks.

In related research that relies on cross-sectional identification, Bakkensen and Barrage (2017) and Baldauf et al. (2020) find heterogeneous house price effects of local time-to-inundation projections between climate change believers and skeptics.

The exception is Cano-Urbina et al. (2019) who show that the oil spill causes a 4–8% decline in home prices (net of restitution payments) that persists until 2015. The finding of a price effect in the intermediate term could indicate a slow regional recovery or that buyers place greater emphasis on recent information/discount older information in updating subjective probabilities.

Like nuclear power accidents and marine oil spills, pipeline incidents, especially those involving property damage and human casualties, are low probability events with complex causal chains. Moreover, underground pipelines, unlike nuclear power plants and oil platforms, are relatively inconspicuous. It is likely, therefore, that many area residents may have been unaware that they owned property near a pipeline prior to the incident. And if they were aware of the pipeline’s location, they might not have known the type of fuel it transported.

Ambrose and Shen (2021) is part of a growing body of literature that examines the impact of fracking on house prices. See Mason et al. (2015) for a comprehensive discussion of the economics of shale gas development. Rather than the capitalization of net local benefits, our focus is on learning in an environment of imperfect information.

Hausman and Kellogg (2015) provide an example of the lack of information on the risks to property owners in areas with fracking activity. In estimating consumer and producer surplus changes as a result of shale gas development, the authors indicate they have no choice but to omit environmental externalities from their analysis. They conclude, “the data necessary to obtain an economic valuation do not exist” (p. 42).

Their results, for example, show that the sale price of homes that are dependent on well water are affected negatively, and homes on municipal water, positively, by proximity in a sample of 36 counties in Pennsylvania for 1995–2012. In contrast, Balthrop and Hawley (2017) find a consistently negative effect in their sample of home sales in Tarrant County (Dallas-Fort Worth metroplex), TX, the majority of which have access to city water. They show that a well within 3500 ft of the property reduces sale price by approximately 1.5–3% for 2005–2011.

Empirical results are consistent with expectations from behavioral economics. For example, Atreya et al. (2013) find a substantial drop in prices for houses located in a flood plain after the 1994 “flood of the century” in Albany, GA, and Bin and Landry (2013) show similar increases in the risk premium for flood zone properties in Pitt County, NC after the landfalls of Hurricanes Fran (1996) and Floyd (1999). In addition, Gallagher (2014) shows that flood insurance take-up increases immediately after a large flood in a nationwide panel of large regional floods. However, in all of these cases, the effects are temporary, with house prices returning to trend, or insurance take-up rates returning to baseline, respectively, in the short to medium term.

In addition to Hurricane Sandy, Keys and Mulder (2020) note that media coverage surrounding the releases of the Intergovernmental Panel on Climate Change’s AR5 report and the United States Global Change Research Program’s National Climate Assessment in 2014 also appears to have increased public awareness of climate risk.

According to a 2017 report, only 42% of FEMA flood maps had been validated as of December 2016 (Office of Inspector General, 2017). The share was similar (45%) in the fourth quarter of 2012 when Hurricane Sandy struck the Northeastern seaboard.

Simple shirking by agents does not appear to be driving the results. Inspired by Levitt and Syverson (2008), the authors find that when agents market their own higher-end properties, they are more likely strategically under disclose.

And Nowak and Smith (2017) and Shen and Ross (2021) are able to significantly increase the predictive power of the models when the tokens are chosen by machine learning rather than the econometrician. However, whether that information is beyond what a prospective buyer can glean from the photographs in the listing is an area for future research.

We also describe papers in the “Selected Theory” section for Part 1 that develop the consequences of asymmetric information between landlords and tenants on housing tenure choice.

The latter case is an example of the incomplete contracts problem. As described influentially by Hart and Holmstrom (1987), the problem arises when actions, quality, or states of the world are observable to the contracting parties, but not verifiable by outsiders, such as the courts.

Moving is costly and an evicted low-quality tenant simply relocates. Regulation can eliminate this negative externality on other landlords.

Median distances are greater, but still low, for professionally managed and class-A buildings at 16.3 and 65.6 miles, respectively.

For class-C properties, average distance discounts are as high as 22.1%, although such a large effect raises concerns about unobserved quality differences.

A material defect is a condition that would have a significant adverse effect on the value of the property. Some quintessential examples of defects necessitating disclosure include water infiltration, access easements, and zoning code violations.

We find no economic arguments in the literature in favor of sellers concealing known material defects from prospective buyers. On the contrary, Weinberger argues that requiring disclosure of material defects known to a seller by dint of occupation does not take away from the seller any legitimate bargaining advantages (Weinberger, 1996, p. 400). And following the logic of the Coase theorem, one could argue that the reassignment under common law of property rights in information from home sellers to buyers is Pareto neutral. However, the assumption of costless bargaining does not hold, because there is no realistic mechanism to obtain a binding commitment from a buyer not to engage in expensive exploratory litigation. Thus, it is important to compare alternative second-best institutional arrangements.

The phrase “federally-related” refers to loans originated by insured financial institutions and loans insured or guaranteed by Fannie Mae, Freddie Mac, Ginnie Mae, the Federal Housing Administration (FHA), and the Veteran’s Administration (VA).

See Gillingham et al. (2009) for a helpful description of common explanations and associated policy responses.

Two other long-running energy efficiency certification programs are the Building Research Establishment Environmental Assessment Method (BREAM), which rates commercial buildings in the UK on a scale of pass, good, very good, excellent, and outstanding; and the Energy Efficiency Rating (EER) which rates houses in the Australian Capital Territory on a scale of 0 to 6 stars.

However, Aydin et al. (2020) show that OLS estimates of energy efficiency capitalization, if anything, are biased downward in their study of residential properties in the Netherlands for 2008–2011.

For the agency costs of debt, which we do not cover, they add the potential costs associated with default.

Berle and Means are the authors of the 1932 book, The Modern Corporation and Private Property, which analyzes the evolution of the diffuse ownership structure of large corporations in the US and other common law countries. See Stigler and Friedland (1983) for a historical review.

Rather than incentive alignment, however, Jensen and Meckling (1976) is perhaps best known for expounding what is often called the “contractual view of the firm.” They describe the corporate firm as nothing more than a “nexus for contracting relationships,” and firm behavior as the ‘outcome from a process that brings the conflicting interests of individuals to equilibrium within this contractual framework’ (p. 311). The contractual view flows from the assumption that a complete contract between managers and shareholders is infeasible. Associated works mostly deal with how to efficiently allocate the residual control rights that remain unspecified by an incomplete contract. Seminal papers include Grossman and Hart (1986), Williamson (1985), Hart and Holmström (1987), Hart (1995) and Tirole (1999). We have relegated these seminal papers to a footnote because they belong more properly to the new institutional economics, rather than the economics of information. However, we acknowledge that it is difficult to separate the influence of informational and institutional works on the empirical literature we survey. That joint influence has been to focus attention on the consequences of particular forms of contractual relations and how those are affected by changes exogenous to the firm.

In addition, reducing the agency costs of free cash flow is a common explanation as to why firms might distribute dividends to shareholders while simultaneously raising additional external debt and equity capital.

In particular, Leland and Pyle (1977) show that firm value is positively related to the firm’s use of debt when the firm has information about an investment project that they cannot otherwise credibly convey. Heinkel (1982) obtains the same result with less restrictive assumptions about the riskiness of debt.

The pecking order is that firms prefer to finance investment through retained earnings to avoid costs associated with asymmetric information. If external funding sources are required, debt is preferred to equity.

The earliest studies of equity market microstructure associated bid-ask spreads with the costs of dealers maintaining inventories, e.g., see Garman (1976). Subsequent models added the assumption that dealers are risk-averse and not well diversified, e.g., Stoll (1978). Researchers have pointed out substantial limitations with both of these modeling approaches, motivating the need to examine the role of information.

Because their underlying assets are debt instruments, we exclude papers on mortgage REITs, consistent with our focus throughout the paper on equity investment. According to the FTSE-NAREIT Index, the 182 equity REITs in their index had a total market capitalization of $1.04 trillion as of October 2017. For comparison, the market capitalization of the 41 mortgage REITs was $66 billion.

Howe and Shilling (1988) argue that REIT value should be inversely related to leverage because REITs are not competing on an equal footing with firms that can deduct interest expenses. Jaffe (1991) dispute this argument, showing that under the assumption of equal tax rates for corporate and personal borrowing, the Modigliani and Miller (1958) irrelevance result obtains for nontaxpaying entities. And Shilling (1996) points out that with Jaffe’s model and differential taxes, you get 100% equity financing. When non-interest tax shields are incorporated, the model predicts that REITs should be 100% debt financed.

Wang et al. (1993) were among the first to point out that REITs often distribute more in dividends than required to maintain their tax exemption.

The story that the market disciplines managers into taking actions in the best interest of shareholders lest they be replaced after a takeover is well developed in Jensen (1988). Grossman and Hart (1980b) are often cited as providing the counterargument in theory. They raise the free-rider problem to dispute the notion that a corporation not run in the best interest of its shareholders is vulnerable to a hostile takeover.

As described by Einhorn et al. (2006), the typical provision gives the board discretion to waive the ownership limit if the suitor is not an individual under the tax code, that is, if the acquirer is another corporation. After acquisition, the shares can be resold in compliance with the 5/50 rule.

Edmans (2014) provide a review of the literature on large shareholder activism. As they explain, research examines how blockholders effect corporate governance through direction intervention (“voice”) and selling their shares (“exit”). They also describe ways blockholders may potentially worsen governance by pursuing corporate control for private benefit.

For a non-apples-to-apples comparison, in a random sample of 385 listed firms for 1995, Holderness (2009) find that 96% have blockholders (institutional and non-institutional) and in aggregate they own an average 39% of the common stock.

Older studies of REITs (Below et al., 1995a, 1996; Bhasin et al., 1997; Clayton and MacKinnon, 2000; Ghosh et al., 1996) do find that their bid-ask spreads are greater than those of traditional stocks, but the spread is decreasing over time in these studies until the relationship reverses in the more recent samples.

At the firm level, Tobin’s q is calculated by dividing the sum of market value of equity and book value of debt by the book value of assets, a proxy for replacement value.

The problem of nonadditive noise in scaled accounting measures has been pointed out recently by deHaan et al. (2019), but is not discussed in the REIT papers on ownership structure that we review.

For the view in economic theory, e.g., see Shleifer and Vishny (1986). In addition, results in the empirical literature find that institutional ownership is associated with greater transparency and increased value (Boone and White, 2015; Greenwood and Schor, 2009; Huddart, 1993).

In the US, the offer price is set just before trading begins. To adjust for market movements during the trading day, researchers subtract the contemporaneous return of a relevant index.

The magnitude of underpricing varies significantly over time: 7.2% for 1980–1989, 14.8% for 1990–1998, 64.6% for 1999–2000 during the dot-com boom, and 18.5% for 2001–2021 (Ritter, Jay, 2022).

For a discussion of potential sample selection bias in related studies, see Ahern (2009).

The magnitude of the discount is higher for industrial firms (∼3%) and lower for utilities (∼1%).

Myers (2001) argue that the price drop should not be interpreted as a transaction cost of issuance according to pecking order theory. Because they are adversely selected, firms that issue are simply worth less on average than firms that do not.

A limitation of the tax-risk tradeoff framework is that it cannot explain why the price reaction to an SEO would be negative on average. This observation parallels the commonly raised failure of tradeoff theory to explain the empirical regularity that profit and debt ratios are negatively related, e.g., (Rajan and Zingales, 1995).

What has been a stylized fact, the post-issuance slump, is perhaps best described now as a matter of active debate in the corporate finance literature. To begin with, there are acknowledge econometric challenges in comparing issuing (growth) versus non-issuing (value) firms that must be overcome with careful matching techniques. In addition, post issuance performance appears to be changing over time. For example, Fu and Huang (2016) confirm Ritter’s results for 1970–2000, but document that firms issuing equity for 2003–2012 do not experience long-run abnormal returns.

As explained by Ritter (2003), findings of relative issue size effects are likely understated due to simultaneity bias. The problem is that the size of the issue is endogenously determined based on feedback received from investors during the marketing process. That is, negatively received issues tend to get scaled down.

That is, the securities offering reform of 2005, and the amendments to forms S-3 and F-3 in 2008.

The idea is that greater information asymmetry increases the likelihood of a firm using an underwriter, but that certification costs are inversely related to firm quality. As expected based on these predictions from theory, Billett et al. (2019) find that the likelihood of an SEO relative to an ATM increases in the interaction of proxies for firm opacity and quality.

As discussed in detail in Brown et al. (2011), the typical firm may not experience a change in officers or directors in a given year nor in their ownership shares.

The instruments for insider ownership are the same as in Himmelberg et al. (1999): stock price volatility, firm size and its square.

The instrumental variable approach are included because of concern by the authors about the potential endogeneity of trading volume, which is included in the regressions to control for market effects.

The CGQ index combines about 60 governance elements into a decile-based score that indicates a company’s relative governance risk across four categories: 1) board structure, 2) compensation, 3) shareholder rights and takeover defenses, and 4) audit and risk oversight. Other governance indexes that researchers might consider include the Accounting and Governance Risk (AGR) Index from Audit Integrity, the CG Rating from Governance Metrics International (GMI), the Corporate Library’s (TCL) Governance Rating, and the S&P’s Corporate Governance Score (CGS).

The exception is a significant effect for a subsample of REITs with low payout ratios.

Results in Lecomte and Ooi (2013) represent a partial exception. They report that among the 21 REITs listed on the Singapore exchange, those with higher corporate governance scores tend to earn greater risk-adjusted returns, but do not outperform operationally, for 2002–2008. The scoring framework was developed by the authors for the Asia Pacific Real Estate Association (APREA) based on 27 external and internal governance factors.

For example, not all countries represented in the sample are in the Euro area and so exchange rate risk likely affects bid-ask spreads and is correlated with risk disclosure metrics.

Seminal papers on the diversification discount include Lang and Stulz (1994), Berger and Ofek (1995), and Laeven and Levine (2007).

Amihud and Lev (1981) and Denis et al. (1997) are representative of works that support the agency cost view. For broad discussions of potential channels, see Rajan et al. (2000) and Campa and Kedia (2002).

For these alternative findings, see Whited (2001) for investment opportunities, Hoechle et al. (2012) for firm heterogeneity, and Campa and Kedia (2002) for the diversification decision.

More to the point, none of the general reviews on corporate finance and governance listed in Table 8 even discuss REITs, other than to say they are commonly excluded from samples in the studies they cover.

The literature on the competitiveness of the brokerage industry is beyond the scope of this article. Interested readers can turn to Barwick and Wong (2019) for a recent, critical assessment of the evidence.

In addition, resetting the listing with a new agent may reduce a seller’s bargaining power. For example, Daneshvary and Clauretie (2013) report that changing an agent before expiration increases marketing time by 2.3 months and reduces transaction price by 2.1%.

Assuming the seller and buyer are represented by different brokers, the usual case, the commission fee, say 6%, is split among four parties. The two brokers retain their overrides and pass the remainder, typically 1.5% to 2.5% each to the affiliated agents who handle the transaction according to the terms of their respective broker-salesperson agreements.

In a net listing agreement, the broker earns as commission all funds received above the seller’s desired net proceeds, as agreed upon in the contract. With kinked commissions, brokers earn greater commission rates upon crossing agreed upon sale price thresholds.

Both papers assume homogeneous agents and properties. In addition, the model in Fisher and Yavas (2010) allows owners to list properties without paying a commission.

Hendel et al. (2009) is a noteworthy exception. The authors compare home sales advertised on an online FSBO platform with brokered sales from the local multiple listing service in Madison, Wisconsin for 19,982,005. Madison is an interesting market because, at 21% during the study period, the FSBO share of sales there is much higher than the national average. Controlling for sample selection, they find that sale prices (pre-commission) are not statistically different, but an FSBO home takes longer to sell and has a lower probability of a sale.

Hayunga and Munneke (2021) categorize non-agent buyers and sellers as individuals, companies (banks, builders, and joint ventures), estates, or government entities.

In the section on “Broker Trading Networks”, we discuss related findings in Han and Miller (2015) that experienced agents specialize in listing, while junior agents specialize in selling.

Finding a null effect, not just a reduction, is not surprising as other features of the market in Singapore favor transparency. In particular, most sales are of flats in relatively homogeneous buildings. It is also common practice for sellers to obtain an appraisal report prior to listing and then use the information in price negotiations.

While not trivial, the cost of an appraisal is relatively small compared with the potential agency costs that have been reported in the empirical literature. Furthermore, it is unlikely that an appraisal ordered by a seller pre-listing will be subject to the same well-documented bias as a lender assignment. For example, Dotzour (1988) examines corporate relocation appraisals, which do not suffer from the same conflicts of interest as lender assignments because they have no contract price for the appraiser to reference. Estimates show that a sample of 500 appraisals have a mean error of -0.06% relative to the subsequent sale price, a value that is not statistically different than 0.

The only work on incentive issues in commercial brokerage, for example, is a study on contract design by D’Lima (2019). This is clearly an area where more research is needed.