Boom and Gloom

• Published date: 01 October 2016
• Date: 01 October 2016
• DOI: http://doi.org/10.1111/jofi.12391

THE JOURNAL OF FINANCE •VOL. LXXI, NO. 5 •OCTOBER 2016

Boom and Gloom

PAUL POVEL, GIORGO SERTSIOS, REN´

ATA KOSOV ´

A, and PRAVEEN KUMAR∗

ABSTRACT

We study the performance of investments made at different points of an investment

cycle. We use a large data set covering hotels in the United States, with rich details

on their location, characteristics, and performance. We find that hotels built during

hotel construction booms underperform their peers. For hotels built during local

hotel construction booms, this underperformance persists for several decades. We

examine possible explanations for this long-lasting underperformance. The evidence

is consistent with information-based herding explanations.

HOW WELL DO investments perform, if they were made during a boom? And if

their performance is different, can strategic interactions between decision mak-

ers help explain the difference? Given the exuberance that has characterized

many booms, they have received substantial attention (see, e.g., Kindleberger

(1978), Greenspan (1996), Akerlof and Shiller (2009), or Glaeser (2013)). How-

ever, the type of data needed to study these questions has been lacking.

So far, evidence has been available only in aggregate form. For example,

IPOs and private equity (PE) and venture capital (VC) funds seem to per-

form less well if they raise funds during “hot” periods.1Data aggregated at

the firm or fund level, however, make it infeasible to study how strategic in-

teractions between projects in their markets relate to investment booms and

their performance. In a particular market, competition at the time of entry

may affect performance after investment booms. Also at the market level,

∗Paul Povel is with the University of Houston. Giorgo Sertsios is with Universidad de los Andes

(Chile). Ren´

ata Kosov´

a is with Imperial College London. Praveen Kumar is with the University

of Houston. We would like to thank two anonymous referees, an anonymous Associate Editor,and

the Editor (Ken Singleton) for very helpful comments. We are also indebted to Agnes DeFranco,

Mark Garmaise, Jerry Hoberg, Piotr Korczak, Gordon Phillips, Christophe Spaenjers, Marta Troya

Martinez, John Walsh, Ivo Welch;participants at the Adam Smith Corporate F inance Conference

(2014), the 1st Edinburgh Corporate Finance Conference (2014), the 2014 Summer Real Estate

Symposium, the 1st Finance, Organization and Markets Conference (2013), and the 11th Inter-

national Industrial Organization Conference (2013); and seminar participants at the University

of Houston, Universidad de los Andes, Universidad Adolfo Iba˜

nez, and Universidad de Chile. We

thank Smith Travel Research for providing us the data for this study, and the Cornell Hospitality

Research Center at Cornell’s School of Hotel Administration for helping us obtain the data. Sertsios

gratefully acknowledges the financial support of Fondecyt Iniciaci´

on (project/folio 11130073). We

have read the Journal of Finance’s disclosure policy and have no conflicts of interest to disclose.

1See Ritter (1991), Gompers and Lerner (2000), Gompers et al. (2008), Kaplan and Schoar

(2005), Kaplan and Str¨

omberg (2009), or Robinson and Sensoy (2013).

DOI: 10.1111/jofi.12391

2287

2288 The Journal of Finance R

informational spillovers may affect whether investments are made in the first

place. To explore whether such strategic interactions can help explain the

booms we observe, and the performance of investments made during such

booms, we need detailed data at the project/investment level in clearly delim-

ited markets.

In this paper, we address these issues. We use a unique proprietary data

set on the characteristics and performance of hotels in the United States. The

data are available at the hotel level and contain detailed information about

the properties as well as the economic and competitive environment in which

each hotel operates. The data also include the year of construction for virtually

all hotels built in the United States, and their location, allowing us to identify

whether having been built during aggregate (nationwide) and market-level

(county-level) investment booms has an impact on their operating performance.

Importantly, investments in the hotel industry are long-lived and irreversible,

allowing us to study both the short-term and long-term impact of local and

aggregate investment booms on performance. In addition, the hotel industry is

particularly suitable to study the role of strategic interactions on investment

booms, as agency problems among decision makers are not a major concern:

most hotels are owned and operated by individuals or partnerships.2

We find that investments made during a boom perform significantly less

well. Consistent with earlier papers on “cohort effects” for IPOs and PE or

VC fund investments (cited above), we find that investments made during ag-

gregate (nationwide) booms underperform for a few years.3More importantly,

we find that, after controlling for aggregate booms, investments made dur-

ing local booms underperform for a long time: the effects are significant even

30 years after a hotel was built. This underperformance is economically signif-

icant. A one-standard-deviation increase in the number of hotels built in the

same county-year reduces a hotel’s performance by 3% to 5%.4Interestingly, we

find that the underperformance of a hotel built during a local boom is driven

by the number of hotels from different quality segments entering the same

geographic market at the same time.5

What can explain the underperformance of hotels built during local booms?

Credit conditions at the time of a hotel’s construction cannot do so, as aggre-

gate credit conditions have an effect on aggregate investment cycles, which we

control for. In addition, our results are robust to the inclusion of variables that

control for aggregate and local credit conditions.

2Institutional details about the hotel industry are described in Section II.

3This is also consistent with a “real options” view of investments (e.g., capital may be available

at a low cost, inducing investors to exercise investment options early). See Grenadier (1996).

4We use revenue per available room (RevPAR), the standard measure of performance in the

hotel industry. Details on this measure are described in Section II. The underperformance we find

reduces the NPV of a hotel project significantly. An example of NPV reduction is given in Section

V.C. Details are in the Internet Appendix in the online version of the article.

5There are six segments: luxury/upper-upscale, upscale, midscale with food and beverage, mid-

scale without food and beverage, economy,and independent. See Sections II and III below for more

details.

Boom and Gloom 2289

Other simple explanations based on local conditions are not sufficient either.

Consider changes in local demand. If more hotels are built because market

participants expect a surge in demand, hotels built during local investment

booms should not perform worse than otherwise equivalent hotels. Similarly,

the underperformance is not due to a worsening pool of available sites in a

specific market. We find that the underperformance is not less pronounced for

hotels in segments where site selection is a priori less relevant (e.g., economy

hotels).6

More promising explanations for the underperformance of hotels built during

local booms focus on the strategic interactions between market participants.

These interactions are strongest at the local market level, where hotels compete

directly and where information about future demand and the attractiveness of

possible investments is transmitted directly or indirectly.

Competition might explain the underperformance if developers neglect possi-

ble entry by competitors during boom periods. Such “competition neglect” may

occur if agents base their decisions on noisy but easily available information

(Veldkamp (2006), Hoberg and Phillips (2010)), if there are coordination fail-

ures (Carlsson and van Damme (1993)), or if agents are overconfident (Camerer

and Lovallo (1999), Simonsohn (2010), Greenwood and Hanson (2015)). If ho-

tels compete most strongly with hotels of a similar vintage, this could lead to

excessive entry and long-run underperformance for hotels built during a local

boom.

However, competition neglect cannot explain the underperformance we find,

since in our data underperformance is related to the number of hotels opened

in the same year and market but in a different quality segment. That does not

mean that competition is irrelevant. To the contrary, a hotel’s performance is

reduced if it competes with a larger number of hotels in the same segment,

irrespective of their vintage. That is, more competitors in the same quality

segment and market do reduce a hotel’s performance, but that is irrespective of

the year in which those hotels entered the market. Interestingly,but consistent

with earlier agglomeration studies, more hotels operating in different quality

segments, irrespective of their vintage, do not reduce a hotel’s performance.

Thus, it is unlikely that underperformance, which is driven by the number of

hotels opened in the same market and year but in different quality segments,

is caused by some sort of competition neglect across quality segments.

Alternative explanations based on strategic interactions focus on problems of

information acquisition and transmission. Hotels are long-lived investments,

so expectations about future demand are crucial for the decision of whether to

enter a particular market. If such information is hard to come by and noisy, it

may be tempting to observe other developers’ entry decisions and imitate them.

Following the literature (for an overview, see Hirshleifer and Teoh (2003)), we

use the term “herding” to describe any such imitation.

6In an earlier version of the paper we show that hotels built during the peak of a local boom

perform less well than hotels built slightly later. This is also inconsistent with a worsening pool of

available sites in a specific market.