The Value of Awarded Design in Real Estate Asset Pricing by Minkoo

Home , 450 Lexington Avenue, 53W53, 750 Seventh Avenue, Citigroup Center, Lipstick Building, MetLife Building

The Value of Awarded Design in Real Estate Asset Pricing

Minkoo Kang

B.S., Interior Design, 2010

Hanyang University

Submitted to the Program in Real Estate Development in Conjunction with the Center for Real Estate in Partial Fulfillment of the Requirements for the Degree of Master of Science in Real Estate Development

at the

Massachusetts Institute of Technology

February, 2019

The author hereby grants to MIT permission to reproduce and to distribute publicly paper and electronic copies of this thesis document in whole or in part in any medium now known or hereafter created.

Signature of Author______Center for Real Estate January 15, 2019

Certified by______Dr. Andrea Chegut Research Scientist, Center for Real Estate Thesis Supervisor

Accepted by______Professor Dennis Frenchman Class of 1922 Professor of Urban Design and Planning, Director, Center for Real Estate School of Architecture and Planning

The Value of Awarded Design in Real Estate Asset Pricing

Minkoo Kang

Submitted to the Program in Real Estate Development in Conjunction with the Center for Real Estate on January 15, 2019 in Partial Fulfillment of the Requirements for the Degree of Master of Science in Real Estate Development

ABSTRACT

This study investigates the financial performance of awarded architectural design for buildings in Manhattan, New York. Awarded design is based on the achievement of the architect and/or the architecture firm receiving prestigious awards from the industry such as the Pritzker prize, AIA Architecture Firm Award, the Architectural Innovation Award of the Wall Street Journal to name just a few. To measure financial performance, I use several datasets, Real Capital Analytics, Compstak, Walkscore and NYC public data for New York City. To identify awarded design and compare it to non-awarded design, I employ a matched-pair analysis. I find 846 building transactions with 89 awarded design transactions that are matched geographically to 757 non- awarded design transactions within a quarter mile radius over the 2000 to 2017 period. The results of the multivariate hedonic analysis suggest that, compared with buildings in the quarter- mile neighborhood, office buildings designed by awarded architects and awarded architecture firms have a statistically and economically significant transaction premium of 23.1 percent, ceteris paribus, with a model that explains just under 90 percent of the variation in transaction price. Results of this analysis are intended as way for designers to have agency in the design build development practice and for developers and investors to understand the value of engaging in awarded design effects.

Thesis Supervisor: Dr. Andrea Chegut Title: Research Scientist for Center for Real Estate This page intentionally left blank

2 3 2 TABLE OF CONTENTS

I Obstacle & Intention 06 Setting the Architectural Stage 08

II Why (Still) New York: Big Apple, Big Data 10 Why Now: Current Climate of Design in the Built 14 Environment Money & Design: How Finance/Economics define 20 Design Stepping Stone 24

III Studying the Value of Design Methods 26 Data 38 Identifying Awarded Architects and Firms 40 Control Group Data 56 Descriptive Statistics 60 Methodology 66 Results: Awarded Designs and Transaction Prices 68 Results: Awarded Architects & Firms and 69 Transaction Prices Robustness: All Architects & Firms and 72 Transaction Prices Conclusion 74 Reference 76 Appendix 78 Acknowledgements 80

This page intentionally left blank

4 5 INVESTMENT & DESIGN

INVESTMENT

REAL ESTATE & DESIGN DEVELOPMENT & DESIGN

REAL ESTATE ARCHITECTURE ARCHITECTURE DEVELOPMENT & DESIGN

FROM SINGLE DISCIPLINE SET OF UNDERSTANDINGS TO A SHARED KNOWLEDGE PLATFORM

OBSTACLE INTENTION

In recent years, due to increased market education and growth in number of leading examples, investing in high quality design became a standard To resolve this issue, we find and measure the architectural legacy of the architects of New York City buildings and pair that with financial data. for the real estate market in New York City. Despite the growing interest, however, a limited number of studies and discussions have been gener- In addition to using the awards as means to identify architects who have won high status among their peers, this study intend to broaden the scope ated to help create a shared value surrounding the subject of design. In this study, we have identified that the difficulty of obtaining data related of understanding the value of design by adding substantially more driving factors in the real estate transaction pricing process by incorporating to design performance being one of the biggest hurdles in enabling further studies to disentangle the value of design. One principle problem is buyer and seller decisions for all of the transactions over time, specifying the type of awards, and adding information on architects to every build- knowing who designed a building, what is their architectural legacy, and can it be matched to a financial performance measure. ing in the sample data set. The added measures improved our model's ability to explain the variation in transaction price. We believe combining new measurements with the accumulated knowledge on design generated by architects will enable us to open up a substantial area for future research regarding the value of design, and moreover will help create agency for design in the realm of finance and economics.

6 7 SETTING THE ARCHITECTURAL STAGE

definitions A look into the future: a render of Manhattan. Photograph: CityRealty

design The terminology design is limited to the design of architecture, especially the design of commercial buildings.

built environment Man-made structures, features, and facilities viewed collectively as an environment in which people live and work. In this study, we are limiting the terminology to buildings and infrastructure in an urban setting.

design development In this research, the terminology “design development” is used to indicate a real estate development model that engages both the work of an architect and real estate development profession- al. It is a business model that typically requires the architect or a designer to invest in the equity New York 2020: A Sea of Design portion of the project to form a As part of its latest development report, real estate agency cityrealty has released a series of visualizations, illustrating the new york skyline in the partnership with a real estate de- year 2020. In summary, Cityrealty reported while fewer developers in 2016 are signing on to build sky-grazing towers, condominium prices are veloper. The partnership benefits still on an upward trajectory with anticipated sales totaling roughly $30 billion through 2019. The report added, the new ground breaking devel- from increased control in design. opments has largely concentrated on midtown in recent years, there is now set to be a new wave of construction in the financial district.

8 9 Why (Still) New York: Why (Still) New York: Big Apple, Big Data Big Apple, Big Data

60 1921 50 Government

Non-residential 1932 20

10 Residential

1999 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

Figure 1 - Annual Construction Spending in New York City, 2001-2020 (in billions)

2002

2018

left, image 1. New York City has historically been the center of modern ar- Note: Change in the skyline of New chitecture and the pinnacle of architectural aesthetics. Not York City from 1921. only is the city both a showcase and a testing bed of modern Source: https://www.theguardian. architectural innovations, but it is also a museum in its own com/cities/gallery/2018/oct/19/ right with an ever-growing collection of carefully preserved rising-high-the-evolving-skyline-new- architectural artifacts from the past. york-city-manhattan-in-pictures The reputation of the city as the capital of modern architecture 2022 figure 1. still remains intact thanks to the continued influx of skyline-al- Note: The graph shows the highest tering developments. The year 2018, marked the highest con- construction spending in 2018. struction spending New York has ever witnessed in its history. Source: Dodge Data & Analytics, NYS The highlight among the developments was the growth in the Department of Laber, public sector number of buildings designed by internationally renowned capital budgets. U.S. Census Bureau, architects. In 2014, more than 50 buildings designed by the Urbanomics so-called star-architects were to break ground in Manhattan alone. The ambitious developments are mostly upon com- pletion in 2019 and some are already available for sale in the market.

10 11 Why (Still) New York: Big Apple, Big Data Why (Still) New York: Big Apple, Big Data

Current Price Avg. Price/Ft2 based on 138 closings in the past 12 months

3000 3015 3097

2782 2787 2500 2517

2000 1969

1500

0 2013 2014 2015 2016 2017 2018

Figure 2 - Star-Architect Condo Price Change Summary During Past Five Years Shown by Median Price Image 2 - Previous Mayor of NYC, Michael Bloomberg's Tweet Regarding New York City Open Data Law

figure 2. Amongst various building types, especially pronounced was image 2. In addition to the rich history of the city and the current build- Note: Condominiums designed by the sales premium associated with the new supply of Man- Source: Twitter, https://twitter.com/ ing boom, the immense data on the built environment that is internationally renown architects marks hattan’s luxury condominiums designed by renown architects. mikebloomberg/status/ available today makes New York the ideal city for this research. the highest price per SF in New York 443753465488367617?lang=en According to the sales data provided by CityRealty, a New Analytics The continuing effort to open the city’s database encouraged City. York based real estate brokerage and consultant firm, as of and boosted researchers to better understand the built envi- Source: CityRealty Starchitect Condo June 2018, buyers of the condominiums designed by Pritzker ronment, as demonstrated in “Open Data Law”, NYC’s recent Index (https://www.cityrealty.com/ prize laureates paid an average $3,126 per square foot, higher Analytics endeavor to consolidate all public data into a single, easily-ac- nyc/building-indices/starchitect-con- than prices in indices such as The CityRealty 100 (at $2,477 cessible platform. The database and the advanced data pro- dos/building-list/112) per square foot), a sales data covering every sale in the past cessing technology available today is opening a new horizon 10 years for 100 of Manhattan’s most expensive condominium in understanding the price dynamics of the building industry in buildings. any given time ■

12 13 Why Now: Current Climate of Design in the Built Environment Why Now: Current Climate of Design in the Built Environment

Standard Design Standard Development Development Development Process Project Project Identification Identification

Project Aquisition Identification Design

Design Aquisition Aquisition Concept Design Phase Market Study Site Analysis Massing Study Concept Proposal Strength of engaging Design in the acquisition process: Concept Plan & Section Suggestive Views (Renderings) Due - Lowers risk by identifying project’s spatial feasibility Diligence - Improves financial underwriting by providing floor area Schematic Design (SD) Phase information with better accuracy Civil Site Plan - Creative solutions for unconventionaly shaped land plots Building Floor Plans Roof Plan Conceptual Details Structural & MEP Code Analysis Figure 4 - Standard Real Estat Development Process Compared with Design Development Process Permitting Design Development (DD) Phase Site Plans Building Elevations Site Sections Building and Wall Sections Typical Design Details Landscaping Plan Code Analysis Plan Details Architectural Floor Plans & Elevations Interior Elevations Reflected Ceiling Plans Schedules figure 3. In recent years, there was a surge of interest in the subject of Note: Diagram showing the relation- Construction Documentation (CD) Phase design from multiple areas of the building industry. Design has ship between the standard real estate Overall Detail Requirements for become one of the most important amenities for real estate Construction of the Project development process and its overlap developers. A growing number of design development com- with standard design process. panies are being established in the major gateway cities such as New York and San Francisco, with a mission to create bet- Construction figure 4. ter development through design and ultimately to differentiate Note: The design development model themselves from competitors. The two noticeable strategies allows the firm to closely manage for developers approaching design are either by working with design to maximize its potential value. famous architects or by creating an in-house design team to initiate and manage design in the closest manner. Hiring renown architects are not something new however, the latter Figure 3 - Standard Real Estat Development Process and Design Phase model is a newly growing trend in New York.

14 15 Why Now: Current Climate of Design in the Built Environment Why Now: Current Climate of Design in the Built Environment

Mortar Arch + Dev ALLOY Development LLC DDG Partners (2003) (2006) (2009) "Design is more than a feeling:

CEO Chairman CEO President it is a CEO-level priority for growth (Architecture) (Finance) (Finance) (Development) and long-term performance." Founder (Architecture & Development) President COO Chief Creative Officer (Architecture & Construction) (Development & (CCO, Architecture) Finance)

167 Eagle Street N/A One John Street 130,000sf 8 Octavia 71,000sf 91 Diamond Street N/A Dumbo Townhouses 18,000sf 400 Grove 36,070sf 873 Pacifc Street N/A 185 Plymouth Street 24,000sf 450 Hayes 80,000sf 585 Union Ave N/A Figure 5192 - Desgin Water Street Development Companies 19,600sf in New York 235 Valencia 20,000sf Image 3 - Mckinsey Quarterly Report, 2018, The Business Value of Design 356 Baltic Street N/A 459 West 18th Street 31,000sf 3550 South Ocean 90,000sf 167 Devoe Street N/A 450 Hudson Boulevard 39,518sf 41 Bond St 22,000sf 764 Metropolitan Ave N/A Glenmore Gardens 22,000sf 345 Meatpacking 8,000sf 245 Manhattan Ave N/A 12 Warren N/A 324 East 4th Street N/A XOCO 325 N/A Ironwood N/A 100 Franklin N/A 504 Humboldt Street N/A 180 East 88th N/A 139 Meserole Street N/A 532 West 20th N/A 292 Manhattan N/A The Standish N/A 100 Gold N/A 294 Ainsle Street N/A

figure 5. The real estate development industry is actively re-inventing image 3. Design is no longer a foreign concept for the investors as Note: The diagram shows the inter- their relationship with design. The two most well known design Source: https://www.mckinsey.com/ well. Investors have been witnessing the appreciation of de- disciplinary management structure of development companies in New York are Alloy Development business-functions/mckinsey-design/ sign from the market and how that translates into additional design development companies in New and ddg, established in 2006 and 2009 respectively, have our-insights/the-business-value-of- profit for their investment. Since 1982, few academic studies York City. architects as owners/partners of the business. Similar to the design from real estate finance and economics have attempted to role of creative directors in the fashion industry, they take uncover the investment premiums related with well-designed charge in managing the design from its inception to comple- buildings. There are some differences in the subject market, tion on behalf of the development company’s interest. building product, and the methods used to measure the value of design, however, the results unequivocally show on average 20% sales premium for well-designed buildings.

16 17 Why Now: Current Climate of Design in the Built Environment Why Now: Current Climate of Design in the Built Environment

120K 6% 1/8/2019 Did Ariana Grande just drop $16M on a condo in Zaha Hadid's Chelsea building? | 6sqft

100K 5%

79,400 CELEBRITIES, CHELSEA 75,000 76,700 80K 73,400 4% 62,600 56,800 60K 48,900 3% 1/8/2019 Another Soaring Starchitect Tower Ascends Over Revitalized Downtown Manhattan 41,900 40K 36,500 2%

1% GET OUR NEWSLETTERS 20K 3,192 views | Sep 30, 2017, 08:39pm

0 0% Another Soaring Starchitect Tower MOST POPULAR ARTICLES 1993 1996 1999 2002 2005 2008 2011 2013 2015 Ascends Over Revitalized Downtown Manhattan Figure 6 - Average Compensation for All Architectural Staff Postitions over Time (in 2015 Dollars) Keith Flamer Contributor Writer, pop culture virtuoso and luxury lunatic

Figure 7 - Media Outlet on the Rise of Design

figure 6. Architects, on the other hand, have been riding the waves of figure 7. a pseudo-celebrity status. The popular terminology ‘star-ar- Note: According to the 2015 AIA change. Design is no longer the sole realm of architects. To- Note: The media often associates design chitect’ or ‘starchitectecture’ (Foster 2008; Barbas, Delautre, Compensation Survey, the average day, design is multi-faceted, highly specialized, and interdisci- with an emphasis on the celebrity and Oakman 2015; Ponzini 2016) well represents the general compensation for architectural staff plinary. So architectural work requires a creative manipulation status and the iconicity of the structure. public’s conception of the subject. The terminology associ- 27 positions is still recovering from the of specialized design developed by a socially diverse group of SHAatesRES design with an emphasis on the celebrity status and the https://www.6sqft.com/did-ariana-grande-just-drop-16m-on-a-condo-in-zaha-hadids-chelsea-building/ 1/7 Great Recession. The report, found that experts to deliver the job that once was done by architects. In- top image iconicity of the structure. While the implication of the terminol- average compensation for staff positions creasingly architectural work is distributed and dispersed, col- Sources: https://www.ft.com/content/ ogy is indeed one aspect of design, it also demonstrates the rose 3.5 percent since early 2013 (or laborative and entrepreneurial, knowledge-based and open- d064d57c-df01-11e6-86ac-f253d- limited outlook on the current climate of design. 1.75 percent per year). This growth sourced. (Peggy Deamer, 2014) b7791c6 https://www.forbes.com/sites/keithflamer/2017/09/30/another-soaring-starchitect-tower-ascends-over-revitalized-downtown-manhattan/#184535e74b33 1/11 is up from the Great Recession, during Design has been clearly a subject of growing interest from which annual compensation increased The growing diversity of the work, however, doesn’t seem to middle image different industries that define the built environment. However, an average of less than 1 percent, but be contributing to the growth of its market size. According to https://www.6sqft.com/did-ariana- design is represented and valued differently across the indus- moderate compared to the past two a survey done by Building Design Magazine, architects’ earn- grande-just-drop-16m-on-a-condo-in- try. As the findings show above, the subject has been adapted decades, when annual compensation in- ings have steadily deteriorated by 30% since 2008. While the zaha-hadids-chelsea-building/ and absorbed by different entities, each of them creating its creases ranged between 4 and 5 percent. result can be interpreted as an influence of the global financial own cluster of understanding and usage with very few over- Source: AIA Compensation Report 2015 crisis, previous periods show that fees rarely return to their bottom image laps. As the innovation theory proves, the innovation, in this pre-recession levels. (Charles Holland, 2014) https://www.forbes.com/sites/ context design, should be widely adopted in order to self-sus- keithflamer/2017/09/30/an- tain since the lack of agency of the subject often leads to de- On the other side of the spectrum, there is a widely accept- other-soaring-starchitect-tower-as- terioration (Everett Rogers, 1962). With New York experiencing ed notion that well-designed space adds more value to the cends-over-revitalized-downtown-man- its biggest building boom of the century, a collective effort built environment. Sadly, there isn’t much public discourse hattan/#184535e74b33 towards creating a common ground to nurture a more collab- beyond the point; the talk always seems to pivot between the orative future for design is imminent. As a contribution, our exorbitant price of real estate market and the fetishistic con- research focuses on identifying and bridging the gap between sumption of the character of few architects that has reached money and design ■

18 19 Money and Design: How Finance, Economics define Design Money and Design: How Finance, Economics define Design

$ : External Factors + Internal Characteristics

Due to the immense data collected on the US commercial real estate market, nowadays at any given time we can analyze : External Factors and understand the commercial real estate price dynamics : Internal Characteristics and can predict future trends with improved accuracy. Two of the main analysis techniques that are often used in this area are the Repeat Sales Index Method and the Hedonic Pricing The Hedonic Pricing Method is an asset Method. pricing method that starts from the premise that the price of a property is determined The Repeat Sales Index Method calculates changes in the both by internal characteristics and external sales price of the same piece of real estate over a specific factors that affect the property’s transaction period of time. By definition, this method can reflect the mar- price. ket conditions in any given period (Geltner & Fisher, 2007) and its strength lies on the ability to reflect the capital gains External Factors: Location, Transaction or depreciation in the market (Chegut, 2013). However, since Time, Building Age, Size, Parcel Area, LEED the methodology is based on available appraisal information Status, etc. and needs a set of properties of repeated sales, a significant amount of time is required to achieve a matured and reliable Internal Characteristics: Building Ame- dataset. Due to this drawback this methodology is not capable nities, Mechanical, Electrical, and Plumbing of capturing the innovations that are occurring in the market (MEP) Quality, Building Occupants Use, etc. and therefore is not considered to be an appropriate methodology for the purpose of this study.

- The Hedonic Pricing Method Figure 8 On the other hand, the Hedonic Pricing Method is a met- ro-level transaction based index. It is an asset pricing method that starts from the premise that the price of a property is determined both by internal characteristics of the property and the external factors that affect the property’s transaction price. Examples of external factors are location, time, age, area of the building, and the internal characteristics are building amenities, operation systems, LEED certification that are figure 8. offered by the building. Different from the Repeated Sales Note: Visual interpretation of the Index, Hedonic Pricing Methodology uses cross-sectional Hedonic Pricing Model data and does not require repeated observations of the same

20 21 Money and Design: How Finance, Economics define Design Money and Design: How Finance, Economics define Design

property. Due to this reason, the Hedonic approach is an ideal The biggest challenge behind design studies is the general methodology to measure the innovations and technologies conception in the subject’s elusiveness that subsequently lim- that are implemented in the current real estate marketplace. ited the measurement and availability of design related data. Surprisingly the most fundamental aspects of design are often Emporis is a real estate data mining company As one of the key aspects of innovation in the built environ- The New York City Department of Buildings disregarded from both private and public data sources that with specializing in high-rise and skyscrapers ment, the complex nature of design can be disentangled and are frequently used when studying the built environment. For measured by adopting the Hedonic Pricing Methodology. instance, the information on architects who are responsible Buildings in Manhattan 83,301 buildings The inherent strength of the Hedonic analysis is that it is an Enforces the city’s building codes and for the overall design quality of the building is often omitted. In Identified in Database (100%) extremely flexible approach that can yield credible results re- zoning regulations fact, Emporis and New York City Department of Buildings (DoB) garding a wide range of subject matters. For instance, ‘Loca- are the only two sources of data among all the private and Information on the 7256 buildings Issues building permits tion:’ one of the external characteristic that is commonly used public data providers that include such information. Even with Architect of building (9% coverage) in real estate asset pricing studies, can be further studied by The data base includes detail informa- these two data providers, the quality and accessibility of the Design Architect 146 buildings looking into related variables such as walk-scores; to mea- tion regarding over 1,000,000 new and data are highly limited. The database of Emporis only includes of building (0.17% coverage) sure the walkability of the neighborhood, or building visibility existing buildings. data on the high rise buildings and the coverage is less than scores; to measure the presence of the building compared 10% of the buildings in Manhattan. On the other hand, since Landscape Architect 125 buildings Currently the information on architects to the surrounding context. Since, the two examples, walk- one of the Department of Building’s function is to issue build- of building (0.15% coverage) are only available upon request and per ability and the iconicity, are often elements that is frequently ing permits, the information on architects should be available buildnig basis. discussed and valued in the design process of the building, for every building that is built, however, such information is Interior Designer 201 buildings the results of the hedonic analysis may support or guide the only available by request and by per building basis. of building (0.24% coverage) designer’s future work by providing numerical measurement on the performance related to the design decisions that were Using a compromised dataset can be particularly troublesome Figure 9-1 - Data Providers with made in the past. However there is limited data on design. Figure 9-2 - Data Providers with when analyzing the subject with the Hedonic Pricing Model Information on Architects Information on Architects since the result heavily relies on the quality and the quantity of So far we have identified largely two directions when measur- the data that is used. Design influences both the external and ing the value of design from the academic literature of real internal characteristics of the building but so far only a handful Note: Emporis and NYC DoB are the only two data estate finance and economics. The two approaches are either Note: Emporis and NYC DoB are the only two data of measurements have been looked at. A better understand- providers for information on architects, however, the relying on peer recognition; taking the award-winning build- providers for information on architects, however, the ing of the relationship between the quality of design and value quality and the accessibility of the data are highly ings and comparing with others to measure its transaction or quality and the accessibility of the data are highly is needed since it could enhance communication between limited. rental premium, or collaborating with a group of experts to ex- limited. the city, investors, developers, and architects, who frequently amine and evaluate the design of the building to understand argue with their own set of assumptions about the relationship the premium related to it. Given the extent of externalities and between the design, cost, and return ■ internalities generated by design, both approaches show limited ability in explaining the value of design beyond the recognition of its associated premium.

22 23 stepping stone Stepping Stone

“Design” is important in ways of affecting the current building industry, however, it is considered a difficult topic to discuss in the context of business decision making. Design still remains an elusive subject and has been studied very little in the context of economics and finance. The popular reason being the lack of consensus on the definition of “design” and its effect on hindering the measurement for its “value” (Vandell, 1989).

Data empowers agency, however, there has been an absence of valid measuring systems to value the contribution of design in the built environment. By utilizing the immense pool of data available today, our research aims to challenge this popular notion of design and to provide a missing link to help left, image 4. understand the fuller picture of the current ecosystem of the Michael Heizer, Slot Mass (section drawing), 1968-2017 built environment. Hopefully, this research can be used as a 18-ton rock and 2 steel earth liners depicted, stepping stone for future studies to ultimately help create an Courtesy of the artist and Gagosian Gallery agency for design in the discussion of finance and economics ■

24 25 studying the value of design methods Studying the Value of Design Methods

D.E.Hough &C.G.Kratz, 1982 Can “Good” Architecture Meet the Market Test?

Data Recognized by “official” Criteria 139 Commercial Office build- Conclusion Is $1.85 (or $1.64) per authority LDMK & CAIA ing rents in Chicago CBD in 1978 square foot truly the value of “good” 20 buildings out of 139 new architecture? If so, at an average rentable area of 844,000 square feet for post-1955 Chicago office buildings, the annual return to this attribute would be $1.6 million (or $1.4 million, using the $1.64 per square foot premium).

F.Fuerst, P.McAllister & C.B.Murray, 2010 Designer Buildings: Estimating the economic value of ‘signature’ architecture

Data CoStar US national database Criteria Pritzker Prize+AIA Gold Medal, Conclusion Compared with buildings for commercial office rental (16,932 499 buildings out of 16,932 in the same submarket, ODSAs have buildnigs observed) & sales (9,418 rents that are 5% - 7% higher than non- sales observed) in 682 submarket ODSAs and sell for prices 12% - 17% clusters higher. In other words, for the average structure, movement into the next The related academic papers that I have identified below are The premiums indicated in both of the studies were large higher design quintile will increase rents the attempts using the hedonic pricing method to under- enough to hint a strong relationship between the design qual- from $27.58/SF to $28.96SF. stand design in the built environment. Largely two different ity of the award-winning architects and the economic perfor- approaches were found when understanding and measuring mance of the building. However, the study leaves a few unan- the effects of design on the value of the building. One set of swered questions. First, the data does not integrate the cost papers examine the quality of design by associating it with the of providing good design, i.e. construction and operation cost architect’s achievement and the recognition of their peers by associated with the building’s iconic structure and additional K.D.Vandell & J.S.Lane, 1989 The Economics of Architecture and Urban Design looking at a sample of buildings designed by architects who fee charges from hiring award-winning architects. Second, the have won important architectural prizes (Hough and Kratz, data does not provide any indication on the different aspects 102 Commercial Office Survey done by panel of The coefficient for 1982; Fuerst, McAllister, and Murray, 2010; Cheshire and Der- of design and associates the value relying solely on the archi- Data Survey criteria Conclusion building Rents and Vacancy architects. Examined in 4 categories and DESIGN, although positive, is not sig- icks, 2014). The other approach chooses to conduct a survey tect’s representation; as the production of design becomes rates in Boston, 1979 - 1986 given overall rating. nificant. Consists with the notion that by a group of experts to grade the overall design quality of the more complex the delineation of the input of awarded archi- design does not necessarily have to sample buildings. The experts score building elements such tect on the design becomes questionable. Third, the award Decorativeness of Facade, cost more to the extent that “overin- as façade fenestration, building material, massing composi- criteria do not capture the current innovations of the industry Categories Color and texture of surface material, quality vestment” may contribute to negative tion, etc. of the sample buildings and the overall design score since the type of awards that are considered in the study only of surface material, massing marginal returns to design. of the building is derived by averaging the scores of each ele- represents the category of lifetime achievement awards which ment (Vandell and Lane, 1989; Nase, Berry, and Adair, 2016). is based on the architect’s work throughout their career with a threshold of at least 30 years of accumulated projects. In 1982, Hough and Kratz in one of the earliest and most often cited academic papers examining the economics of architec- Research done in 1990 by Vandell and Lane attempted to un- ture argued that commercial buildings in the central business cover few of the missing links of the previous study by includ- I.Nase, J.Berry & A.Adair, 2016 Impact of quality-led design on real estate value district (CBD) of Chicago that have won a Chicago AIA Jury ing the construction and operation costs into the framework award outperformed in rents per square foot as high as 23% of the regression analysis and by disentangling the multiple relative to the market for comparable buildings. dimensions of design into categories such as, the decorative- Data 424 Condominium Criteria Survey done by group of local ex- Conclusion Empirical findings indicate ness of the façade, color and texture of the surface material, units in Belfast city center, perts. 7 categories on a 5-grade Likert scale. that from the seven building quality Another similar study was done in 2010. Fuerst, McAllister, and quality of the surface material, and massing. 102 class-A 2000 - 2008 features initially investigated, the ones Murray conducted a national, rather than a city, level research commercial office buildings in Boston and Cambridge were Categories Facade material, facade iden- mostly valued by end users are those focusing on buildings designed by Pritzker prize and/or AIA evaluated by a group of architects accordingly and the results tity, quality of material used, fenestration, that are easily perceived visually. Gold medal winning architects in the USA. The results of the confirmed a strong influence of design on rents. Buildings massing, height in floors, building condition hedonic analysis also showed premiums that are 5%-7% that were rated in the top 20% for design quality were predict- higher in rents, and 17% higher sales prices in the buildings ed to extract almost 22% higher rents than those rated in the designed by the award-winning architects compared with oth- bottom 20%. In contrast, the data showed a weak relationship er buildings in the same submarket. between vacancy behavior and design quality. Finally, good Table 1 - Design Value Studies

26 27 studying the value of design methods studying the value of design methods

on Archit ation ects fo form r B In uil D din D gs A in d a ta b a se s m r i F

s t c M e t i

A h

T c

r C

0 1 2 3 4 5 Our Research Approach H

r w

w r

b F

I u

N i

I t

1. Examine the Quality of Design by 2. Scoring Process for Design Quality Variables b y

o m

Associating it with Architecture Awards ex) Massing, Fenestration, Material Quality, etc. i

r c

l i

r l

i u l i m i d s

i n n i g

Figure 10 - Typical Methods Used to Measure the Effects of Design on the Value of the Building Figure 11 - Our Reseacrh Approach

design was shown to cost more to produce on average, but appropriateness to the surroundings of a building’s material In line with the studies listed above, other studies find that figure 10. not necessarily in every case. The magnitude of the point esti- quality, fenestration, and massing. The other design features certain architectural styles (Asabere, Hachey, and Grubaugh, Note: The diagram illustrates the two mates of the rent, vacancy, and construction cost effects sug- namely façade material, façade identity, overall building condi- 1989), features of exterior of the building (Moorhouse & Smith, popular study methodologies used in gest that good design may not, in fact, be more profitable on tion, and floor height were found to be statistically non-signifi- 1994) and urban design features of the neighborhood (Song & measuring the value of design. average, but as with a lottery, may provide a small probability cant. Knaap, 2003) achieve rental and sales premium for the resi- of a high return to the developer. dential market. The more recent studies that are currently un- figure 11. The scoring methodology that is used in the two studies are dergoing in this area attempts to include daylight and views to Note: The diagram illustrates our A similar methodology was used to study the price premium based on the assumption that the experts’ opinions are close- understand the subject. These findings are informative for the research approach, focusing on different related with good design on the residential market by Nase, ly correlated with the judgment process in architectural design current research since they demonstrate how the market, in types of awards and incorporates infor- Berry, and Adair in 2016. The study combined the hedonic practice, therefore, the estimated numerical value incorpo- general, values design in the context of the built environment. mation on architects and architecture modeling approach paired with the spatiotemporal model to rates the value of design with less error. However, the method firms of every buliding in the data set. understand the impact of quality led design on the residential inherits certain limitations. For example, the subjectivity of Our research differs from these previous studies. It focuses market in Belfast, Ireland. The research took transaction data design may be amplified due to the small number of experts on different types of awards that can include a broader range of 424 condominium units and conducted a survey done by included in the group causing skewness or inconsistency in of architects, it includes architecture firms as well as indi- local experts. The survey included 7 categories on a 5-grade the resulting score data. In addition, the design elements in vidual architects and further elaborates on the estimation by scale such as façade material, façade identity, quality of mate- which the buildings are measured by are solely focusing on incorporating information on architects and architecture firms rial used, fenestration, massing, the height of floors, and over- the exterior of the building which leaves out the quality of the of every building in the data set. The data and methodology of all building condition. The empirical findings indicate that from interior space and the building’s inner spatial relationship that this research are detailed below ■ the 7 design characteristics examined, the ones most valued is important for understanding the user’s experience and the by the home buyers are those that are most visible, i.e. the performance of the building.

28 29 studying the value of design methods studying the value of design methods D. E. Hough & C. G. Kratz, 1982 D. E. Hough & C. G. Kratz, 1982

In 1982, Hough and Kratz in one of the earliest and most often cited academic papers examin- Can “Good” Architecture Meet the Market Test? ing the economics of architecture argued that commercial buildings in the central business district A considerable rent premium is paid for “good” new architecture but not for “good” old architec- (CBD) of Chicago that have won a Chicago AIA Jury award out performed in rents per square ture. Chicago AIA award increases the annual rent about $1.85/SF, however Land Mark status foot as high as 23% relative to the market for comparable buildings. decreases the annual rent about $0.81/SF. (D.E.Hough & C.G.Kratz, 1982)

-1 -0.5 0 0.5 1 1.5 2 ($/SF)

chicago aia award $ 1.85/SF ***

landmark status -$ 0.81/SF ***

Significance*** Coefficient % The standard Asterisks in a regression table error is our estimate of the standard indicate the level of the statistical deviation of the coefficient. significance of a regression coefficient.

*** p<0.01, ** p<0.05, * p<0.1

30 31 176 F Fuerst, P McAllister, C B Murray

studying the value of design methods studying the value of design methods F.Fuerst, P. McAllister, & C. B. Murray, 2010 F.Fuerst, P. McAllister, & C. B. Murray, 2010 Model 2 shows the results of the regression with separate dichotomous variables for ODSAs and offices designed by top-500 firms. In this model only the ODSAs are found to exert a positive and significant impact on rents. For offices designed by top- 500 firms, there is no significant impact on asking rent. In model 3 an attempt is made Another studyto examine done in whether 2010 by any Fuerst, signature-architect McAllister, premiumand Murray is due conducted to larger, corporate-stylea national, rather than What is the economic value of ‘signature’ architecture? city, level researchfirms. The focusing results on suggest buildings that designed this is the by case. Pritzker For ODSAsprize and/or where AIA the signatureGold medal win- Compared with buildings in the same submarket, Office Designed by Signature Architects have ning architectsarchitect in the is USA.not a The top-500 results firm, of there the hedonic is no statistically analysis also significant showed effect premiums on rent. that are rents that are 5% - 7% higher than Office Designed by Non-Signature Architects and sell for 5%-7% higherIn contrast, in rents, for and ODSAs 17% where higher the sales signature prices architect in the isbuildings also a top-500 designed firm, by the the effect award win- prices 12% - 17% higher. (F. Fuerst, P. McAllister & C. B. Murray, 2010) ning architectsis a 6%compared increase with in asking other rent.buildings Given in the the dominance same submarket. of SOM in this group, this finding can also be interpreted as an SOM effect.(5) Perhaps surprisingly, the coefficient for a top-500 firm is negativeöalbeit only significant at the 10% levelösuggesting that the appointment of a top-500 firm might have a negative impact on asking rent.

Table 3. Results from hedonic estimation of sale prices. Dependent variable is natural log of the sale price ($ per square foot).

Model 1 Model 2 Model 3

Constant 5.76*** 5.78*** 5.78*** ODSA a 0.12** 0.15*** Top 500 (non-ODSA) 0.23*** 0.23*** ODSA (non-top 500) 0.06 ODSA and top 500 0.17*** 0 (%) 2 4 6 8 10 12 14 16 Net lease 0.08*** 0.08*** 0.08*** Number of stories (ln) 0.17*** 0.16*** 0.16*** Size in square feet (ln) 0.18*** 0.19*** 0.19*** Transaction Price Site area in square feet (ln)À 0.08***À 0.08***À 0.08*** Premium Renovated within 5 years 0.07** 0.07** 0.07** 14.5% *** Renovated 5 ± 10 years ago À0.01 À0.01 À0.01 Renovated 10 years ago À0.03 À0.03 À0.03 > Rent Premium Age (ln), years À À À 0±2 0.17*** 0.17*** 0.17*** 6% ** 3 ± 6 À0.12***À 0.11***À 0.11*** 7 ± 10 0.36*** 0.35*** 0.35*** 11 ± 19 0.28*** 0.28*** 0.28*** 20 ± 23 0.12*** 0.12*** 0.12*** 24 ± 26 0.06** 0.06** 0.06** 27 ± 31 0.06** 0.06** 0.06** 32 ± 42 0.03 0.03 0.03 43 ± 62 À0.07** À0.07** À0.07** Occupancy rate À0.00 À0.00 À0.00 Class A 0.40*** 0.40*** 0.40*** Class B 0.07*** 0.07*** 0.07*** Poor market 0.08*** 0.08*** 0.08*** Strong market 0.06*** 0.06*** 0.06*** Very strong market 0.16*** 0.17*** 0.17*** 682 Submarket controls included

Adjusted R 2 0.37 0.38 0.38 F-test 7.96*** 7.98*** 7.97***

Number of observations 6 970 6 970 6 970 Significance*** Coefficient % The standard Asterisks in a regression table error is our estimate of the standard *** Significant at 1% level; ** significant at 5% level; * significant at 10% level. indicate the level of the statistical deviation of the coefficient. a ODSAÐoffice designed by signature architect. significance of a regression coefficient.

*** p<0.01, ** p<0.05, * p<0.1 (5) We also tested a model including SOM as a dependent variable. The result was similar to the coefficient for top-500 firms that are also signature architects in model 3.

32 33 studying the value of design methods studying the value of design methods K. D. Vandell & J. S. Lane, 1989 K. D. Vandell & J. S. Lane, 1989

A research done in 1990 by Vandell and Lane included the construction and operation costs into Does well designed buildings rent for more? the framework of the regression analysis and by disentangling the multiple dimensions of design For the average structure, 5.0% increase in rents with every increase of one in the design rating. into categories such as, the decorativeness of the façade, color and texture of the surface materi- In other words, movement into the next higher design quintile will increase rents from $27.58/SF al, quality of the surface material, and massing. 102 class-A commercial office buildings in Boston to $28.96SF. (K.D.Vandell & J.S.Lane, 1989) and Cambridge were evaluated by a group of architects accordingly and the results confirmed a strong influence of design on rents. Buildings that were rated in the top 20% for design quality were predicted to extract almost 22% higher rents than those rated in the bottom 20%.

Does good design result in lower vacancy? Insignificant, though consistently negative as expected and always with in the narrow range -.4003 to -.5127 in all specifications. This suggests that, at the mean, an increase of one quintile in -de sign quality would decrease the vacancy rate from 1.7% to 1.0%. (K.D.Vandell & J.S.Lane, 1989)

0 (%) 1 2 3 4 5 6

Rent Premium 5% **

Vacancy Rate 1.35%

*** p<0.01, ** p<0.05, * p<0.1

34 35 studying the value of design methods studying the value of design methods I.Nase, J.Berry & A.Adair, 2016 I.Nase, J.Berry & A.Adair, 2016

The study done by Nase, Berry, and Adair in 2016, combined the hedonic modeling approach What is the impact of quality led design for real estate value? paired with the spatiotemporal model to understand the impact of quality led design on the resi- Empirical findings indicate that from the seven building quality features initially investigated, the dential market in Belfast, Ireland. The research took transaction data of 424 condominium units ones mostly valued by end users are those that are easily perceived visually. (I.Nase, J.Berry & and conducted a survey done by local experts. The survey included 7 categories on a 5-grade A.Adair, 2016) scale such as façade material, façade identity, quality of material used, fenestration, massing, the height of floors, and overall building condition. The empirical findings indicate that from the 7 design characteristics examined, the ones most valued by the home buyers are those that are most visible, i.e. the appropriateness to the surroundings of a building’s material quality, fenestration, and massing. The other design features namely façade material, façade identity, overall building condition, and floor height were found to be statistically non-significant. JOURNAL OF PROPERTY RESEARCH 319

Table 2. hedonic and spatial model comparisons. 0 5 10 15 20 25 30 35 (%) Variable Model 1 Hedonic (OLS) Model 2 SEM (ML) Model 3 SAR (ML) Building condition constant 1.5332 (1.8328) 3.0714** (3.4862) −1.4176 (−1.6753) 2.14% age −0.1312* (−2.3588) −0.1644** (−3.0613) −0.1241* (−2.4979) area 0.9155* (12.6764) 0.8427** (12.3922) 0.8054** (12.1854) Garage 0.1761* (3.6562) 0.1930** (4.4050) 0.2111** (4.8876) Height Bedrooms −0.0108 (−0.4963) −0.0016 (−0.0774) 0.0083 (0.4216) 3.28% receproom 0.1357 (1.7292) 0.1741* (2.4100) 0.1682* (2.3957) floorno 0.0594** (3.8029) 0.0429* (2.8552) 0.0343* (2.3914) finishing 0.0823** (2.9538) 0.0529 (1.9259) 0.0432 (1.7071) Massing Identity 0.1172** (3.0372) 0.0817* (2.1308) 0.0447 (1.2421) 7.24% ** Materialqual 0.3195** (9.7516) 0.2394** (6.1798) 0.1749** (5.0190) fenestration 0.1102** (3.0104) 0.1102** (3.1965) 0.0909** (2.7652) Massing 0.0724** (3.5583) 0.0675** (3.4033) 0.0769** (4.2348) Finishing height 0.0328 (1.3987) 0.0140 (0.5887) 0.0079 (0.3748) 8.23% ** condition 0.0214 (0.6128) 0.0293 (0.8707) 0.0312 (1.0008) connect 0.0681** (2.9810) 0.0574* (2.5617) 0.0654** (3.2051) Fenestration Bpr 0.0102** (5.7598) 0.0077** (4.4278) 0.0067** (4.0656) attindex −0.0392** (−2.7200) −0.0408** (−2.8341) −0.0471** (−3.6494) 11.02% ** Dgreen −0.0898** (−4.5197) −0.0807** (−4.0771) −0.0882** (−4.9730) nearst 0.0220* (2.5806) 0.0201* (2.1370) 0.0289** (3.7711) Identity PWdist 0.0757** (7.6161) 0.0647** (6.4809) 0.0566** (6.1024) 11.72% ** yr2002 −0.0058 (−0.1151) 0.0577 (0.4684) 0.0527 (1.1633) yr2003 0.1260* (2.5254) 0.1721 (1.4025) 0.1846** (4.1258) yr2004 0.0556 (1.1846) 0.0969 (0.8203) 0.1056* (2.5103) Material quality yr2005 0.1744** (3.2972) 0.2234 (1.6761) 0.1878** (3.9737) 31.95% ** yr2006 0.4421** (8.7191) 0.5057** (4.1295) 0.3189** (6.4205) yr2007 0.6397** (11.9617) 0.6657** (5.3196) 0.4682** (8.5984) yr2008 0.4555** (9.0797) 0.4825** (3.9844) 0.3387** (6.9931) lambda (λ) 0.6581** (8.9581) rho (ρ) 0.4279** (7.5933) Significance*** Coefficient % The standard 2 2 sigma (σ ) 0.0210 0.0176 0.0167 Asterisks in a regression table error is our estimate of the standard R2 0.8494 0.8634 0.8706 indicate the level of the statistical deviation of the coefficient. log-likelihood 18.9413 218.4580 231.9330 significance of a regression N 373 373 373 coefficient.

notes: t values are in parentheses, * and ** denote 5% and 1% significance levels respectively. Multidirectional spatial *** p<0.01, ** p<0.05, * p<0.1 weight matrix M based on negative exponential distance with 1.2 km cut-off.

Table 3. Moran’s I tests for spatial autocorrelation in the ols residuals. M specification 1.9 km 1.8 km 1.7 km 1.6 km 1.5 km 1.4 km 1.3 km 1.2 km 36 37 Moran’s I 0.0404 0.0415 0.0383 0.0426 0.0581 0.0732 0.0789 0.0950 Moran’s I-statistic 16.7135 16.5283 14.9900 14.8962 16.1798 17.3529 15.7179 16.1742 Marginal probability (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) Mean −0.0279 −0.0280 −0.0281 −0.0285 −0.0289 −0.0291 −0.0298 −0.0306 sD 0.0041 0.0042 0.0044 0.0048 0.0054 0.0059 0.0069 0.0078

spatial autocorrelation in the residuals can be rejected for all specifications of M providing strong evidence for the need to control for spatial externalities through spatial econometric models (Table 3). The LM tests point towards the spatial lag (SAR) model for all different cut-off distances (Table 4).9 Additionally, in selecting the most appropriate M for our study we consider those that maximise the test statistics (Boots & Dufournaud, 1994). The results indicate that the matrix with the 1.2 km cut-off best describes the multidirectional spatial relation with the combined building effects in the data. To additionally support the choice of M, Bayesian posterior model probabilities are compared (LeSage & Pace, 2009).10 As the results indicate, the two candidates are the matrices with 1.3 and 1.2 km cut-off points. data Data

+ +

PRIMARY DATABASE ADDITIONAL DATABASE ADDITIONAL DATABASE

Variables in Use: Variables in Use: Variables in Use:

Price Building Class Walk Score assigned to Submarket every building in datrabase Transaction Year Built Year DATA ON ARCHITECTS Number of Floors + Building Area (SqFt) The information on architects Land Parcel Area were manually gathered through Renovation Year multiple sources such as, firm Buyer Type portfolio website, media articles, Seller Type and architectural magazines Lender Type

Figure 12 - Data Structure

The commercial building transaction database used in this Parcel Area, Renovation Year, Buyer Type, Seller Type, and In addition, we have included a Walk Score variable using the figure 12. empirical study was obtained from the Wide Data Project of Lender Type. This study uses RCA building transaction data as data provided by Walkscore.com. The Walk Score is a metric Note: The diagram illustrates the data MIT Real Estate Innovation Lab, which is a combination of a primary database. created to measure the walkability of the neighborhood with structure of this research. publicly available data from New York government entities, a score range from 0 to 100. Neighborhoods with access to Real Capital Analytics (RCA), and Compstak data. The inte- Compstak is a private commercial real estate data platform public transit, better commutes, and proximity to the people grated database provides fundamental hedonic variables that with offices in New York and Los Angeles. The data is crowd- and places, achieve higher scores. For this study, we have we will be using. sourced from verified and active professionals at commercial matched the address of individual buildings observed in the brokerages and appraisal firms and provides lease and sales building transaction sample dataset with the Walk Score pro- Real Capital Analytics (RCA) is a private data provider special- comparable data. Compstak database contains variables that vided from the website izing in property transaction data based in New York City. The include lease contract characteristics, tenant profile, and mar- database collects data from a network of independent sourc- ket variables to name a few. From this database, we included Finally, we have included the information on architects who es with particular emphasis on the building transaction data a variable which is Building Class. The Building Class variable designed the buildings in the integrated database. The infor- that includes financing details, prior transaction history, and is an important variable used to control for the overall quality mation was gathered using multiple sources that include, the true owner identification to complete profiles. From this data- of the buildings in the sample dataset and is a variable that architect’s web portfolio, architecture magazine, articles from base, we use variables including the transaction price for each is only available in the Compstak data. We have assigned the various publications, and Wikipedia ■ contract signed that becomes the dependent variable. The building identification number (BIN) for each transaction ob- variables used in this study are Price, Submarket, Transaction servation in RCA and matched with the Compstak data set for Year, Built Year, Number of Floors, Building Area (SqFt), Land better accuracy.

38 39 identifying awarded architects and firms IDENTIFYING AWARDED ARCHITECTS AND FIRMS

Since the objective of this research is to understand the value years of experience based on their lifetime contribution on In this research, we have listed the award winners of all three of design through the contributions of renown architects and expanding the knowledge of the industry. These awards are categories from the year 1940 and each award group resulted architecture firms, defining the significant architects and firms considered the highest recognition and considered to be the in 147 architects for group 1, 32 architects for group 2, and 55 becomes the central issue for this research. most influential in the architecture industry. The awards are architecture firms for group 3. Within this result, we have iden- given annually or biannually. tified in total 18 awarded architects and firms who designed As the scale of the building grows and adds complexity to the an existing building in Manhattan at the time of the research. project, the business management aspect of the architecture The second group incorporates awards that are given to the and design industry have continuously evolved. To prevent contemporary and innovative architects. The group includes: Within 18 awarded architect/firms, 4 Firms have received the knowledge loss and to secure the design quality of the Cooper Hewitt National Design Award and the Wall Street more than two awards from three award categories and they company, architecture firms are creating a new management Journal Innovation Awards in Architecture. These awards are are Edward Larrabee Barnes, I. M. Pei & Partners, Kevin Roche model that is in between the typical master and apprentice given annually and a large part of the evaluation is based on John Dinkeloo and Associates, and Skidmore, Orwings & Mer- system and the hierarchical corporate management model the impact of the architect’s project on the year the award rill (SOM). Each group of architects and companies are identi- (Booth, 2006). is given. Due to this reason, the demographics of the past fied as variables such as Awarded Architects, Awarded Firms, winners of these awards tend to be younger than the lifetime and Awarded Architects and Firms respectively throughout To incorporate this recent trend, we have employed multiple achievement award laureates. this research. types of awards that each has a significant difference in their evaluation criteria, but nevertheless carries similar weight in The third award category is an attempt to recognize the col- We have identified 16 buildings designed by Awarded Archi- value among the industry. Largely three types of award groups laborative effort and the business management side of architects, 14 buildings designed by Awarded Firms, and 22 de- are considered in this research. tecture design by including the AIA Architecture Firm Award. signed by Awarded Architects and Firms. Overall 52 buildings The first group includes lifetime achievement awards that The AIA Architecture Firm Award is a unique award since it and 89 transaction observations were found in the treated evaluate the architect’s accumulated body of work throughout recognizes the architecture firm that has produced a nota- group data. In addition, information on the awarded architects their career. The group includes: the RIBA Royal Gold Med- ble architecture for at least a decade. The candidates of the and firms are assigned to each building in the data set ■ al, AIA Gold Medal, the Pritzker Prize, the UIA Gold Medal, award are any individual firms or successor firm or organiza- and the Golden Lion for Lifetime Achievement Award. These tion of architects whose home office is based in the US. awards are typically given to architects with more than 30

40 41 TABLE 2 - AWARD CRITERIA & AWARDED ARCHITECTS and FIRMS identifying awarded architects and firms

Awarded Firms: Architecture firms who won the AIA Architecture Firm Award Royal Gold Gold Gold Golden Pritzker Medal Medal Medal Lion Prize Davis Brody Bond '75 100 William Organization Organization Organization Organization Organization Five Manhattan West RIBA AIA UIA Venice Pritzker How often How often How often Biennale Foundation How often Annual Annual Triennial How often Gensler '00 233 Spring Street First awarded First awarded First awarded Biennial Annual 161 6th Avenue First awarded 1848 1907 1984 First awarded From 1950 From 1950 From 1950 2000 1978 From 1950 Kohn Pedersen Fox '90 745 7th Avenue 69 awarded 59 awarded 26 awarded From 1950 Five TIme Square 12 awarded 41 awarded One Vanderbilt 111 Murray Street Innovator National Architecture 1100 6th Avenue Awards Desgin Award Firm Award 441 8th Avenue Organization Organization Organization 10 Hudson Yards WSJ Cooper Hewitt AIA How often How often How often Murphy / Jahn 425 Lexington Avenue Annual Annual Annual 05 65 East 55th Street First awarded First awarded First awarded 2010 2000 1962 From 1950 From 1950 From 1950 Hugh Stubbins & Assoc. '67 Citi Group Center 7 awarded 26 awarded 55 awarded

Awarded Architects: Architects who won the lifetime achievement awards and/or innovation awards Awarded Architects & Firms: Architects/firms who won both the lifetime achievement award and AIA Architecture Firm Award

Aldo Rossi '90 557 Broadway Edward L. Barnes '07 '80 499 Park Avenue 555 Broadway 7 Byant Park

Alvar Alto '63 809 United Nations Plaza I.M.Pei '10 '79 '14 '83 JP Morgan Chase HQ Deutsche Bank HQ 31 West 52nd Street Jean Nouvel '01 '08 45-47 W 53rd St '03 '68 750 Seventh Avenue

César Pelli '95 Three World Financial Center Four World Financial Center Kevin Roche '93 '82 '74 125 West 55th Street 900 3rd Avenue Avenue of Americas Plaza NY Mercantile Exchange 787 Seventh Avenue

Norman Foster '83 '94 '99 610 Lexington Avenue Skidmore Orell & Merril '57 '83 '88 '96 12 West 57th Street 875 3rd Avenue 425 Park Avenue Paine Webber Building 919 T3rd Avenue Marine Midland Bank 28 Liberty Bertelsmann Building Fumihiko Maki '11 '93 '93 51 Astor Place 300 Madison Avenue 34-36 East 51st Street Mies van der Rohe Seagram Building '59 '60 One Manhattan West Two Manhattan West

Philip Johnson '78 '79 5 East 44th Street 450 Lexington Avenue Sony Plaza 461 5th Avenue Lipstick Building 510 5th Avenue Worldwide Plaza 830 3rd Avenue Walter Gropius '56 '59 200 Park Avenue

42 43 TABLE 3 - CATALOGuE OF TREATED BuLIDING TRANSACTIONS identifying awarded architects and firms

Architect Skidmore, Orwings & Skidmore, Orwings & Kevin Roche Merrill (SOM) Merrill (SOM) Property Name 875 Third Avenue Bertelsmann Bldg 31 West 52nd Street Sub-Market Midtown East Midtown West Midtown West Year Built 1982 1990 1986 No. Floors 29 FL 44 FL 30 FL Building Image Building Area 719,000 SF 1,058,287 SF 729,011 SF Building Class A A A

Skidmore, Orwings & Skidmore, Orwings & Skidmore, Orwings & Skidmore, Orwings & Merrill (SOM) Merrill (SOM) Merrill (SOM) Merrill (SOM) 830 Third Ave 510 5th Ave 461 Fifth Ave Marine Midland Bank Midtown East Midtown West Midtown East Downtown 1956 1954 1988 1967 13 FL 5 FL 26 FL 52 FL 144,000 SF 70,000 SF 200,000 SF 1,200,866 SF A B A A

Skidmore, Orwings & Skidmore, Orwings & I.M.Pei & Partners Skidmore, Orwings & Merrill (SOM) Merrill (SOM) Merrill (SOM) 300 Madison Ave PaineWebber Building 499 Park Ave 450 Lexington Midtown East Midtown West Midtown East Midtown East 1910 1960 1981 1992 16 FL 42 FL 28 FL 32 FL 490,560 SF 1,749,000 SF 292,966 SF 910,473 SF A A A A

Kevin Roche Skidmore, Orwings & Skidmore, Orwings & Kevin Roche Merrill (SOM) Merrill (SOM) JP Morgan Chase HQ 450 Lexington 450 Lexington Deutsche Bank HQ Downtown Midtown East Midtown East Downtown 1988 1992 1992 1988 47 FL 32 FL 32 FL 47 FL 1,612,000 SF 910,473 SF 910,473 SF 1,612,000 SF A A A A

44 45 TABLE 3 - CATALOGuE OF TREATED BuLIDING TRANSACTIONS identifying awarded architects and firms

Edward Larrabee Barnes Kevin Roche Kevin Roche Skidmore, Orwings & Merrill (SOM) 125 West 55th Street 750 Seventh Avenue 750 Seventh Avenue 875 Third Avenue Midtown West Midtown West Midtown West Midtown East 1989 1989 1989 1982 23 FL 36 FL 36 FL 29 FL 548,881 SF 591,169 SF 591,169 SF 719,000 SF A A A A

Skidmore, Orwings & Kevin Roche Skidmore, Orwings & Skidmore, Orwings & Merrill (SOM) Merrill (SOM) Merrill (SOM) Worldwide Plaza 31 West 52nd Street 450 Lexington Avenue PaineWebber Building Midtown West Midtown West Midtown East Midtown West 1989 1986 1992 1960 47 FL 30 FL 32 FL 39 FL 2,055,583 SF 729,011 SF 910,473 SF 1,749,000 SF A A A A

Skidmore, Orwings & Skidmore, Orwings & Edward Larrabee Barnes Skidmore, Orwings & Merrill (SOM) Merrill (SOM) Merrill (SOM) 510 Fifth Avenue 34-36 E 51st St 125 West 55th Street 28 Liberty Midtown West Midtown East Midtown West Downtown 1954 1922 1989 1963 5 FL 10 FL 23 FL 57 FL 61,159 SF 41,000 SF 548,881 SF 2,215,030 SF B NA A NA

Skidmore, Orwings & Skidmore, Orwings & I.M.Pei & Partners Skidmore, Orwings & Merrill (SOM) Merrill (SOM) Merrill (SOM) 919 Third 12 West 57th Street 499 Park Ave Worldwide Plaza Midtown East Midtown West Midtown East Midtown West 1970 1904 1981 1989 46 FL 11 FL 28 FL 47 FL 1,316,758 SF 84,000 SF 292,966 SF 2,055,583 SF A A A A

46 47 TABLE 3 - CATALOGuE OF TREATED BuLIDING TRANSACTIONS identifying awarded architects and firms

Pei Cobb Freed & Part- Hugh Stubbins & Associ- Kohn Pedersen Fox As- Gensler ners ates sociates (KPF) 7 Bryant Park 745 Seventh Avenue 233 Spring St Midtown West Citigroup Center Midtown West Downtown 2015 Midtown East 2001 1926 30 FL 1977 38 FL 10 FL 470,000 SF 59 FL 1,020,000 SF 249,148 SF A 1,800,000 SF A A A

Skidmore, Orwings & Kevin Roche Murphy/Jahn Kohn Pedersen Fox As- Merrill (SOM) sociates (KPF) PaineWebber Building 31 West 52nd Street 425 Lexington Ave Five Times Square Midtown West Midtown West Midtown East Midtown West 1960 1986 1987 2002 39 FL 30 FL 31 FL 39 FL 1,749,000 SF 729,011 SF 750,000 SF 1,101,779 SF A A A A

Edward Larrabee Barnes Skidmore, Orwings & Hugh Stubbins and Asso- Davis, Brody & Associ- Merrill (SOM) ciates ates 787 Seventh Avenue Two Manhattan West Citigroup Center Midtown West Midtown West Midtown East 100 William 1985 2020 1977 Downtown 51 FL 62 FL 59 FL 1972 1,706,007 SF 2,000,000 SF 1,800,000 SF 21 FL A NA A 357,000 SF A

Skidmore, Orwings & Kevin Roche Murphy/Jahn Kohn Pedersen Fox Associ- Merrill (SOM) ates (KPF) One Manhattan West Deutsche Bank HQ Park Avenue Tower future One Vanderbilt (partial) Midtown West Downtown Midtown East Midtown East 2019 1988 1986 1913 67 FL 47 FL 36 FL 17 FL 2,000,000 SF 1,612,000 SF 615,857 SF 160,482 SF NA A A B

48 49 TABLE 3 - CATALOGuE OF TREATED BuLIDING TRANSACTIONS identifying awarded architects and firms

Kohn Pedersen Fox As- Davis, Brody & Associates Kohn Pedersen Fox Associ- Ludwig Mies van der Rohe sociates (KPF) ates (KPF) 111 Murray Street Five Manhattan West Five Times Square Seagram Building Downtown Midtown West Midtown West Midtown East 1984 1969 2002 1958 10 FL 16 FL 39 FL 38 FL 145,525 SF 1750,000 SF 1,132,865 SF 820,000 SF NA A A A

Murphy/Jahn Davis, Brody & Associates Gensler Gensler

425 Lexington Ave 100 William One Soho Square One Soho Square Midtown East Downtown Downtown Downtown 1987 1972 1904 1926 31 FL 21 FL 15 FL 10 FL 750,000 SF 357,000 SF 450,000 SF 316,000 SF A A B A

Kohn Pedersen Fox Associ- Davis, Brody & Associates Kohn Pedersen Fox Associ- Kohn Pedersen Fox Associ- ates (KPF) ates (KPF) ates (KPF) Five Times Square Five Manhattan West HBO HBO Midtown West Midtown West Midtown West Midtown West 2002 1969 1906 1906 39 FL 16 FL 15 FL 15 FL 1,101,779 SF 1,750,000 SF 344,000 SF 344,000 SF A A A NA

Murphy/Jahn Hugh Stubbins & Associates Kohn Pedersen Fox Associ- Kohn Pedersen Fox Associ- ates (KPF) ates (KPF) Park Avenue Tower Citigroup Center 441 Ninth 10 Hudson Yards Midtown East Midtown East Midtown West Midtown West 1986 1977 1953 2016 36 FL 59 FL 8 FL 52 FL 619,631 SF 1,800,000 SF 423,000 SF 1,813,465 SF A A NA A

50 51 TABLE 3 - CATALOGuE OF TREATED BuLIDING TRANSACTIONS identifying awarded architects and firms

Philip Johnson / Alan Ritchie Philip Johnson Philip Johnson Foster + Partners Architects 5 E 44th St Lipstick Building Lipstick Building Shangri-La hotel project Midtown East Midtown East Midtown East Midtown East 1940 1986 1986 1926 6 FL 34 FL 34 FL 10 FL 15,726 SF 592,000 SF 592,000 SF 81,017 SF NA A A NA

Alvar Aalto Foster + Partners Walter Gropius Ludwig Mies van der Rohe

809 United Nations Plaza Fmr YMCA MetLife Building Seagram Building Midtown East Midtown East Midtown East Midtown East 1964 1926 1963 1958 11 FL 10 FL 58 FL 38 FL 100,000 SF 81,017 SF 2,840,000 SF 820,000 SF NA NA A A

Philip Johnson Cesar Pelli & Associates Philip Johnson / Alan Ritchie Foster + Partners Three World Financial Center Architects Sony Plaza Downtown 5 E 44th St 425 Park Midtown East 1986 Midtown East Midtown East 1984 52 FL 1940 1957 36 FL 2,100,000 SF 6 FL 31 FL 855,000 SF A 15,726 SF 567,340 SF A NA A

Ludwig Mies van der Rohe Cesar Pelli & Associates Fumihiko Maki Philip Johnson Cooper Union Engineering Seagram Building 900 Third Ave Dev Site Lipstick Building Midtown East Midtown East Midtown South Midtown East 1958 1984 1960 1986 38 FL 36 FL 9 FL 34 FL 820,000 SF 595,105 SF 158,816 SF 592,000 SF A A A A

52 53 TABLE 3 - CATALOGuE OF TREATED BuLIDING TRANSACTIONS identifying awarded architects and firms

Aldo Rossi Aldo Rossi Philip Johnson Fumihiko Maki

Scholastic 555 Broadway Sony Plaza 51 Astor Place Midtown South Midtown South Midtown East Midtown South 1999 1900 1984 2013 10 FL 12 FL 36 FL 13 FL 112,500 SF 216,000 SF 855,000 SF 400,000 SF NA B A A

Cesar Pelli & Associates Cesar Pelli & Associates

900 Third Ave Four World Financial Center Midtown East Downtown 1984 1986 36 FL 34 FL 595,105 SF 2,084,079 SF A A

Foster + Partners Cesar Pelli & Associates New York Mercantile Ex- 425 Park Avenue change Midtown East Downtown 1957 1997 31 FL 17 FL 567,340 SF 502,000 SF A A

Ludwig Mies van der Rohe Ateliers Jean Nouvel

Seagram Building 53W53 Midtown East Midtown West 1958 2000 38 FL 5 FL 820,000 SF 28,291 SF A NA

54 55 control group data CONTROL GROUP DATA

0.25 Mile Radius 0.25 Mile Radius

Building Designed by Identify All other Office Buildings in Awarded Architect / Firm Create 0.25 Mile Radius Database within Radius Identification

RESULT: 56 AWARDED DESIGN BUILDINGS 89 AWARDED BUILDING TRANSACTIONS 833 OFFICE BUILDINGS 846 TOTAL TRANSACTION OBSERVATIONS

Building Designed by Create 0.25 Mile Radius Awarded Architect / Firm Identification

Figure 13 - Control Group Data Process Diagram Figure 14-1 - Control Group Data Process, GIS

In order to understand the effect of the awarded architects to capture all the commercial buildings that intersect with figure 13. and firms on the transaction price, we matched each of the the integrated database. In this way, we created 4 clusters of Note: The diagram illustrates how the awarded buildings in this sample to nearby commercial build- nearby office buildings. Each small cluster—0.2 square miles— data was filtered to construct the control ings in the similar location using the Geographic Information contains one awarded building and at least one non-awarded group data set. System (GIS). Out of 2,399 office building transactions iden- nearby building. In addition, we have collected the information tified in the integrated database constructed combining RCA on architects for all buildings in the control group data set ■ figure 14. and Compstak, 52 buildings were designed by awarded archi- Note: The series of maps show the lo- tects/firms. Based on the latitude and longitude of each treat- cation of the treated buildings, quarter ed building we created a one quarter mile radius buffer zone mile radius, and the filtered data points.

56 57 control group data control group data

Identify All other Office Buildings in Database within Radius figure 14. Note: The series of maps show the location of the treated buildings, quarter Figure 14-2 - Control Group Data Process, GIS mile radius, and the filtered data points.

58 59 descriptive statistics DESCRIPTIVE STATISTICS

TOTAL TRANSACTION OBSERVATIONS BUILDING TRANSACTION PRICE (AWARDED DESIGNS)

N Min Max Mean SD

Transaction Price 89 $ 4.4 M $ 2.98 B $ 607.8 M $ 618.9 M

Log (Price) 89 6.64 9.47 8.53 0.55

846 Price per SF 89 $ 109 $ 1,951 $ 731 $ 404

Log (PSF) 89 2.04 3.29 2.79 0.26

BUILDING TRANSACTION PRICE (NON-AWARDED DESIGNS)

N Min Max Mean SD INTEREST VARIABLES (NUMBER OF BUILDING TRANSACTIONS)

Transaction Price 757 $ 1 M $ 3.4 B $ 178.8 M $ 339.7 M

Awarded Log (Price) 757 6 9.53 7.79 0.65 Design Price per SF 757 $ 30.8 $ 6,800 $ 630.6 $ 580.1 Awarded Awarded Architects Log (PSF) 757 1.49 3.83 2.68 0.32 89 Architects & Firms 27 38 MARKET CHARACTERISTICS

757 24 Midtown W

350 326 90 84 Midtown E 80 300 254 74 Non-Awarded Awarded 70 68 66 Design Firms 250 61 59 DowntownDowntown 60 50 200 186 49 50 44 40 41 40 40 43 150 40 34 Figure 15-1 - Descriptive Statistics 33 Midtown S 30 Note: The graphs represent the statistics of each variable used in the regression analysis. 100 58 Upper E 20 50 11 9 22 10

0 0 The transaction sample data set contains 489 commercial information on the architects and firms are assigned to each '00 '01 '02 '03 '04 '05 '06 '07 '08 '09 '10 '11 '12 '13 '14 '15 '16 '17 office buildings with 846 transaction observations. Among building in the data set. Out of 318 architects and firms in total, which 52 buildings were designed by awarded architects/ 194 were non-awarded architects, 20 were awarded architects Sub Market Transaction Year firms (treated group) and 437 buildings were designed by and 104 architects were unidentifiable ■ non-awarded architects (control group). Related transactions Figure 15-2 - Descriptive Statistics were observed 89 and 757 times respectively. In addition, Note: The graphs represent the statistics of each variable used in the regression analysis.

60 61 descriptive statistics descriptive statistics

BUILDING CHARACTERISTICS (AWARDED DESIGNS) BUILDING CHARACTERISTICS (NON-AWARDED DESIGNS)

BUILDING CLASS NUMBER OF RENOVATED BULIDINGS BUILDING CLASS NUMBER OF RENOVATED BULIDINGS

Total : 89 Total : 89 Total : 757 Total : 757

Unknown

Renovated Unknown 14 Class A Not Renovated Class B 5 27 229 223 341 416 62 70 58 Renovated 247 Not Renovated Class A Class C

Class B

N Min Max Mean SD N Min Max Mean SD

Age 89 0 118 46.27 28.33 Age 757 7 208 86.3 26.64

Number of Floors 89 5 67 30.39 15.71 Number of Floors 757 2 77 18.48 13.21

Area SqFt 89 15,726 2,84 M 855,757 666,291 Area SqFt 757 2,365 2.63 M 331,081 455,666

Land Area SqFt 89 2,700 150,718 43,574 32,104 Land Area SqFt 757 800 130,680 16,220 19,074

Walk Score 89 94 100 99.16 0.95 Walk Score 757 92 100 99.28 0.96

Figure 15-3 - Descriptive Statistics Figure 15-4 - Descriptive Statistics Note: The graphs represent the statistics of each variable used in the regression analysis. Note: The graphs represent the statistics of each variable used in the regression analysis.

62 63 descriptive statistics descriptive statistics

TRANSACTION CHARACTERISTICS (AWARDED DESIGN) TRANSACTION CHARACTERISTICS (NON-AWARDED DESIGN)

Total : 89 Total : 757

BUYER TYPE SELLER TYPE LENDER TYPE BUYER TYPE SELLER TYPE LENDER TYPE

20 Corporate 33 Corporate 124 CMBS 34 Fund 31 Fund 14 Government 27 Institutional 14 Government 1 Corporate 6 Corporate 22 CMBS 33 Institutional 1 Government 39 Offshore 8 Institutional 2 Fund 41 Offshore 2 Government 33 Financial 14 Offshore 8 Institutional 248 Private 330 Private 2 Government 35 Insurance 9 Offshore 76 Intl. Bank 2 Financial 19 Private 4 Insurance 15 Private 15 Intl. Bank

84 Ntl. Bank

4 REIT 5 REIT 8 Ntl. Bank 5 REOC 42 Unknown 2 Pension Fund 9 Private 37 Unknown 1 Local Bank 37 Unknown 77 Local Bank

33 REIT 315 Unknown 1 REOC 2 Retailer 303 Unknown 20 Private 2 REOC 4 Retailer 249 Unknown

Figure 15-5 - Descriptive Statistics Figure 15-6 - Descriptive Statistics Note: The graphs represent the statistics of each variable used in the regression analysis. Note: The graphs represent the statistics of each variable used in the regression analysis.

64 65 System (GIS). Based on the latitude and longitude of each treated building we created a one quarter mile buffer zone to capture all the commercial buildings that intersect with the integrated database. As a result, we created a sample dataset that includes 833 buildings and 846 transaction observations.

(Include GIS map sequence to help understand the process)

methodology METHODOLOGY 4 DESCRIPTIVE STATISTICS

The transaction sample data set contains 846 observations within which 89 transactions are buildings designed by either Awarded Architects or Awarded Firms (treated group) and 757 transacted buildings designed by non-awarded architects and firms (control group). The observed transactions start from 2000 and ends at 2017.

(Add Descriptive Statistics) i Commercial Office Buildings

logPi Logarithm of the Transaction Price 5 METHODOLOGY α System (GIS). Based on the latitude and longitude of each treated building we created a oneConstant We use the MIT Real Estate Innovation Lab NYC Wide Data database to estimate a semi-log quarter mile buffer zone to capture all the commercial buildings that intersect with theβ integratedEstimated Coefficients for equationdatabase. relating As a result, the transaction we created price a sample to the datasethedonic that characteristics includes 833 of buildings a building: and 846 Hedonic Characteristics transaction observations. = + + + Xi Vector of Hedonic Char- acteristics(e.g. Location, (Include GIS map sequence to help understand the process) Transaction Time, Building 𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥 𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 𝜶𝜶𝜶𝜶 𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝑷𝑷𝑷𝑷 𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝑷𝑷𝑷𝑷 𝜺𝜺𝜺𝜺𝑷𝑷𝑷𝑷 The dependent variable is the logarithm of the transaction price in commercial office Features, and Transaction SystemSystem ( (GISGIS)). .Based Based on on the the latitude latitude and and longitude longitude of of each eachFeatures) treated treated building building we we created created a a one one buildings . is a vector of hedonic characteristics (e.g., location and time, building features, System (GIS). Based on the latitude and longitude of each SystemtreatedSystemSystem building(GIS ((GISGIS). Based) ))we... BasedBasedBased created on the ononon a thequarterthelatitudethequarter one latitudelatitudelatitude mile andmile andandlongitudeandbuff buff er longitudelongitudeerlongitude zone zone 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃of to eachto ofcapture ofofcapture eacheach eachtreated treatedalltreatedtreatedall the buildingthe commercial commercial buildingbuildingbuilding we created wewewe buildings buildings createdcreatedcreated a one that athataa oneoneone intersect intersect with with the the integrated integrated quarter mile buffer zone toand capture transaction all the commercial features) buildings for buildingsthat intersect ,with and the integrated database.database.is a vector As As a a ofresult, result, dummy we we created created variables a a sample sample with dataset datasetδ a value that that includes includes Estimatedof 1 833 833 Coefficients buildings buildings and and for 846846 System (GIS). Based on the latitude𝑃𝑃𝑃𝑃 and𝑋𝑋𝑋𝑋𝑃𝑃𝑃𝑃 longitude of eachquarterquarterquarterquarter treatedSystem mile milemilemile building buff ( GIS buffbuffbuffer )zoneer.er er weBased zonezonezone createdto capture toontoto capturecapturecapturethe a onelatitudeall the allallall commercial the thetheand commercialcommercialcommercial longitude buildings of buildingsbuildings buildingseach that treated intersect thatthatthat building intersectintersectintersect with wethe withwithwith created integrated thethethe integratedintegratedintegrated a one database.quarter mile As abuff result,er zone weif4 created tobuildingDESCRIPTIVE capture a sample all isthe designeddataset commercial STATISTICS that includesby buildingsdatabase. Awardeddatabase.database.database. quarter833 that Asbuildings intersect AsAsmileaAsArchitects, result, aaa buff result,result,result, and erwithwe 846zone wecreatedwewe the transactionAwarded createdtransactioncreated createdto integrated capture a sample aaa samplesampleobservationssample observationsall Firms, datasetthe commercial datasetdatasetdataset or that. .Awarded thatincludesthatthat buildings includesincludesincludes 833 Architects that buildings 833833833 intersect buildingsbuildingsbuildings and and with846 andandand Dummy the 846846846 integrated Variable transactiondatabase. observationsAs a result, .we created a sample dataset thattransaction includestransactiontransactiontransactionSystemdatabase. 833 observations(GIS observationsbuildingsobservationsobservations )As𝑃𝑃𝑃𝑃. Based a result, .onand 𝑔𝑔𝑔𝑔 the𝑃𝑃𝑃𝑃 ...we 846 latitude created and a longitudesample dataset of each thattreated includes building 833 we buildingscreated a oneand 846 Firms and 0 otherwise. is a quarterconstant, mile buff er zoneand to capture are estimated all the commercial coefficients buildings that and intersect withisgi an the errorintegrated transaction observations. transaction observations(Include(Include. GIS GIS map map sequence sequence to to help help understand understand the the process) process)Vector of Dummy Variables 𝑃𝑃𝑃𝑃 database. As a result, we created a sample dataset that includes 833 buildings and 846 (Include GIS map sequenceTheterm. to transactionhelp understand sample the process) data(Include set GIS contains map sequence 846 observations to help understand within the process) which 89 transactions are(e.g. Value of 1 if buliding i is (Include(Include(Includetransaction GISGISGIS observations mapmapmap sequencesequencesequence. tototo helphelphelp understandunderstandunderstand thethethe process)process)process) 𝛼𝛼𝛼𝛼 𝛽𝛽𝛽𝛽 𝛿𝛿𝛿𝛿 𝜀𝜀𝜀𝜀𝑃𝑃𝑃𝑃 designed by awarded archi- (Include GIS map sequencebuildings to help designed understand bythe eitherprocess) Awarded (Include GIS Architects map sequence or Awardedto help understand Firms the (treated process) group) and 757 tects or firms) transactedUsing the logarithm buildings ofdesigned transaction (Include by non price GIS- awardedmap instead sequence arcof to thehelphitects understandtransaction and thefirms process) pric (controle per squaregroup). foot The may 44 DESCRIPTIVE DESCRIPTIVE STATISTICS STATISTICS observedcause measurement transactions error start for from the 2000size andand constructionends at 2017. cost may vary over the sizeεi of the Error Term 4 DESCRIPTIVE STATISTICS 4 DESCRIPTIVE444 DESCRIPTIVEDESCRIPTIVE STATISTICS STATISTICSSTATISTICS building. However, we have conducted the sameTheThe hedonic transaction transaction analysis sample sample data data using set set contains containsthe transaction 846 846 observations observations price within within which which 89 89 transactions transactions are are 4 DESCRIPTIVE STATISTICS 4 DESCRIPTIVE STATISTICS The transaction sample data set contains 846 observationsThe within transaction4 DESCRIPTIVEwhich 89 sampletransactions STATISTICS data buildings setbuildingsare contains designed designed 846 observations by by either either Awarded Awarded within which Architects Architects 89 transactions or or Awarded Awarded are Firms Firms (treated (treated group) group) and and 757 757 (Addper sf Descriptive for every model Statistics) and a TheThe substantial transactiontransaction samplepricingsample datadata diffe setsetrence containscontains was 846846 not observationsobservations found. In withinwithin addition, whichwhich 8989 looking transactionstransactions at areare buildings designed by either Awarded Architects or Awardedbuildings Firms The (treated designed transaction group) by sampleeither and 757Awardedtransactedtransacted data set Architectscontains buildings buildings 846 Figureor designed designed Awarded observations 16 - by Semi-Logby Firms non non within- -awarded(treatedawarded Equation which group)arc arc 89hitectshitects transactions and and and757 firms firms are (control (control group). group). The The The transaction samplethe data functional set contains form 846 observations of the results,buildingsbuildingsbuildings withinThe transaction thewhich designeddesigneddesigned coefficients 89 sample transactions bybyby eithereither eitherdata setand AwardedAwardedAwarded arecontains the ArchitectsstandardArchitects846Architects observations ororor errors AwardedAwardedAwarded within were which FirmsFirmsFirms 89comparable. (treated(treated(treatedtransactions group)group)group) are andandand 757757757 transacted buildings designed by non-awarded architects andtransacted firmsbuildings (control buildings designed group). designed The by either observedobservedNote: by non TheAwarded -equation awardedtransactions transactions Architectsrelated arc thehitects start transactionstart or from fromAwardedand price 2000 firms2000 to the Firms and (controland hedonic ends ends (treated characteristics group). at at 2017. 2017. group) The of a andbuliding 757 buildings designed by either Awarded Architects or Awardedtransactedtransactedtransactedbuildings Firms (treated designed buildingsbuildingsbuildings group) by designeddesigned designedeither and Awarded 757 bybyby nonnonnon Architects---awardedawardedawarded or Awarded arcarcarchitectshitectshitects Firms andandand (treated firmsfirmsfirms (controlgroup)(control(control and group).group).group). 757 TheTheThe observed transactions startThis from could 2000 and mean ends the at 2017. sample observed datatransacted transactions set contains buildings start designed fromrelatively 2000 by and non similar ends-awarded at scale 2017. architects of buildings and firms (controltherefore group). the The transacted buildings designed by non-awarded architects andobservedobservedobservedtransacted firms (control transactionstransactionstransactions buildings group). designed startstartstart The fromfrom fromby non 200020002000-awarded andandand endsendsarcendshitects atatat 2017.2017.2017. and firms (control group). The observed transactions start(Add(Add fromDescriptive Descriptive 2000 and Statistics) Statistics) ends at 2017. observed transactions start5difference METHODOLOGY from 2000 in and using ends transaction at 2017. observed price transactionsand transaction start from 2000 price and per ends sfat 2017.diminishes. The regression (Add Descriptive Statistics) (Add Descriptive Statistics) analysis using the transaction(Add(Add(Add price DescriptiveDescriptiveDescriptive per sf Statistics)Statistics)Statistics) can be found in the Appendix section. (Add(Add Descriptive Descriptive Statistics) Statistics) (Add Descriptive Statistics) We use the MIT Real EstateWe use Innovation the MIT Real Lab Estate NYC Innovation55 METHODOLOGY METHODOLOGY Wide DataLab NYC database Wide Data to estimateUsing the logarithm a semi -oflog transaction price instead of the trans- 5 METHODOLOGY database5 METHODOLOGY to estimate a semi-log equation relating the transaction price per square foot may cause measurement error equation relating the transactionaction555 METHODOLOGYMETHODOLOGY5 price METHODOLOGY price to the to hedonic the hedonic characteristicsWeWe use use characteristicsthe the MIT ofMIT a Realbuilding:Real Estate Estate of Innovation Innovation a building:for the Lab Lab size NYC NYC and Wide Wide construction Data Data database database cost to mayto estimate estimate vary over a a semi semi the- -logsizelog of 5 METHODOLOGY We5 METHODOLOGY use the MIT Real Estate Innovation Lab NYC Wide DataWe database use the toMIT estimate Real Estate a semi equationInnovationequation-log relating relating Lab NYC the the Widetransaction transaction Data databaseprice pricethe to tobuilding. the theto hedonicestimate hedonic However, characteristics acharacteristics semi we- loghave conducted of of a a building: building: the same hedonic equation relating the transaction6 RESULTS price to the: AWARDED hedonic characteristicsequation DESIGNSWeWeWe We useuse use relatingof use the the athe building:the MIT MITANDMIT the MIT= RealReal transaction RealTRANSACTION Estate Estate+Estate Estate Innovation InnovationInnovationprice +Innovation to Labthe +Lab Lab hedonicNYC PRICESLab NYCNYC Wide NYC characteristics Wide WideData Wide database DataData Dataanalysis databasedatabase database ofto aestimate building:using to toto estimatetheestimateaestimate semi transaction-log aa semisemisemi price---logloglog per sf for every model and equation relating the transaction price to the hedonic characteristics of a building: We use the MIT Real Estate Innovation Lab NYC Wide Dataequationequationequation database relatingrelatingrelating to estimate thethethe transactiontransactiontransaction a semi- log pricepriceprice tototo thethethe hedonichedonichedonic characteristicscharacteristicscharacteristicsa substantial== +ofofof+ a aa building:building:pricingbuilding:++ difference ++ was not found. In addition, equation relating the transaction price to the hedonic characteristics of a building: = + + + equation relating the transaction price to the hedonic characteristics of a building: == ++ ++ ++ looking at the functional form of the results, the coefficients 𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥 𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 𝜶𝜶𝜶𝜶 𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝑷𝑷𝑷𝑷 𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝑷𝑷𝑷𝑷 == 𝜺𝜺𝜺𝜺+𝑷𝑷𝑷𝑷+ ++ 𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥++𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 𝜶𝜶𝜶𝜶𝜶𝜶𝜶𝜶 𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 𝜺𝜺𝜺𝜺𝜺𝜺𝜺𝜺𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 The dependent variable isThe the dependent logarithm variable of the is thetransactionTheThe logarithm dependent dependent of price variable thevariable= transaction + is isin the the +commercial logarithm logarithm and+ theof of the thestandard office transaction transaction errors price price were comparable.in in commercial commercial This office office could mean 𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥 𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 𝜶𝜶𝜶𝜶 𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝑷𝑷𝑷𝑷 𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝑷𝑷𝑷𝑷 𝜺𝜺𝜺𝜺𝑷𝑷𝑷𝑷 𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥 𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 𝜶𝜶𝜶𝜶 𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝑷𝑷𝑷𝑷 𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝑷𝑷𝑷𝑷 𝜺𝜺𝜺𝜺𝑷𝑷𝑷𝑷 The dependent variable isbuildings the logarithm . of =the is transactiona+ vector+ priceofThe +hedonic dependent The in dependent commercial characteristics variable variable office is theis buildingsthebuildings logarithm logarithm (e.g.,𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥 𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑖 . .of 𝑋of location the 𝜶𝜶𝜶𝜶is the isis a a transaction 𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷vector vectovector𝑷𝑷𝑷𝑷 andr 𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹of ooff 𝑷𝑷𝑷𝑷hedonic hehedonicprice price time,-𝜺𝜺𝜺𝜺𝑷𝑷𝑷𝑷 the incharacteristics characteristics buildingincommercial sample commercial data features,office (e.g., (e.g.,officeset contains location location relativelyand and time, time, similarbuilding building scale features, features, of build - 𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 𝜶𝜶𝜶𝜶𝜶𝜶𝜶𝜶 𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 𝜺𝜺𝜺𝜺𝜺𝜺𝜺𝜺𝜺𝜺𝜺𝜺𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 buildings . is a vector of hedonic characteristics (e.g., locationbuildingsTheThebuildingsThe anddependentdependent .dependent time, .is abuilding is vectorvariablevariable a vectorvariable features,of isofhedonicisand and hedonic theisthe thetransaction transactionlogarithm logarithm logarithmcharacteristics characteristics𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥 𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 ofoffeatures) features) of thethe 𝜶𝜶𝜶𝜶the transaction𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 transaction(e.g.,(e.g., transaction𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷 for for 𝑷𝑷𝑷𝑷location locationbuildings buildings𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹 price𝑷𝑷𝑷𝑷price andprice and 𝜺𝜺𝜺𝜺 time,, 𝑷𝑷𝑷𝑷 ,and andtime, in inin inbuilding commercial commercialcommercial commercialbuilding is is a a features,vector vector features, officeofficeofficeof of dummy dummy variables variables with with a a value value of of 1 1 and transaction features)donic for buildings 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃characteristics , and (e.g., location is a and vector time,𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑋𝑋𝑋𝑋𝑋𝑋𝑋𝑋𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 buildingof dummy features, variables ings therefore with a valuethe difference of 1 in using transaction price and andThe transaction dependent features) variable for is thebuildings logarithm𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥𝐥 ,𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 and of 𝜶𝜶𝜶𝜶the is transactiona𝜷𝜷𝜷𝜷𝜷𝜷𝜷𝜷 vector𝑷𝑷𝑷𝑷 𝜹𝜹𝜹𝜹𝜹𝜹𝜹𝜹 ofand 𝑷𝑷𝑷𝑷 dummybuildingsbuildingspricebuildings transactionand𝜺𝜺𝜺𝜺buildings𝑷𝑷𝑷𝑷 transaction variables in . .. commercial features). isisis features) awithaais vector vectorvectora vectora for valueoffice for ifofbuildingsofofif building buildings building hedonicofhedonic hedonic of hedonic 1 , , characteristicsis andcharacteristics characteristicsisand characteristicsdesigned designed isis aa vector vectorby by Awarded (e.g.,Awarded(e.g.,(e.g., (e.g.,of of dummy dummy locationlocation𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃location location𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 Architects, Architects, variables variables andand andand time, time,time,withtime,Awarded Awarded with a buildingbuildingbuildingvalue a valueFirms, Firms, of 1 features,features,features, of or or1 Awarded Awarded Architects Architects and and 𝑃𝑃𝑃𝑃 𝑋𝑋𝑋𝑋𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 𝑋𝑋𝑋𝑋𝑃𝑃𝑃𝑃 and transaction𝑃𝑃𝑃𝑃 𝑋𝑋𝑋𝑋features)𝑃𝑃𝑃𝑃 for buildings 𝑖 , and 𝑔 is a vector of transaction𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 price per sf diminishes. The regression analysis if buildingsbuilding is. designed is a vector byif Awarded buildingof hedonic Architects, characteristics is designed Awarded (e.g., byFirms,if Awardedbuildinglocationandandand ifor and building transactiontransactiontransactionAwarded𝑃𝑃𝑃𝑃 transactionand 𝑋𝑋𝑋𝑋is𝑃𝑃𝑃𝑃 designedArchitects, time,is Architectsdesigned features)features)features) building features) by byAwarded and FirmsAwarded forAwardedforFirmsforfeatures, for buildingsbuildingsbuildings buildingsand andArchitects, Architects, 0 0 otherwise. Firms,otherwise. ,,, andandand, and AwardedAwarded or is isis is aisAwardedaais aFirms, vector vectora vectorFirms, avector constant, constant, or ofoforof ofAwarded dummy dummy AwardeddummyArchitectsdummy 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 and and Architects variables variablesvariablesvariables Architects are are andestimated estimated and withwith and aaa valuevaluecoefficients coefficients ofof 11 and and is is an an error error 𝑃𝑃𝑃𝑃 𝑔𝑔𝑔𝑔𝑃𝑃𝑃𝑃 dummy variables with a value of 1 if building𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑖𝑔𝑔𝑔𝑔 is𝑃𝑃𝑃𝑃 designed by using the transaction price per sf can be found in the Appen- Firmsand transactionand 0 otherwise. features) Firmsis fora constant, buildings and 0 otherwise.,and and are is estimated a vector Firmsis if ififaof coefficients buildingbuilding buildingFirms constant, dummyifand building𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 and0 𝑃𝑃𝑃𝑃 otherwise.𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 variables 𝑋𝑋𝑋𝑋is0𝑋𝑋𝑋𝑋isis𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃and otherwise.𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 designed 𝑃𝑃𝑃𝑃designeddesigned𝑋𝑋𝑋𝑋is 𝑃𝑃𝑃𝑃 designed andis with an is𝑔𝑔𝑔𝑔 by by𝑃𝑃𝑃𝑃 byaiserror term. a term.by constant,a Awarded AwardedareAwardedvalue constant, Awarded estimated of𝑃𝑃𝑃𝑃 1 Architects,Architects, Architects, Architects,and and𝑔𝑔𝑔𝑔 𝑃𝑃𝑃𝑃 are coefficients AwardedAwardedestimatedAwardedestimated Awarded coefficients Firms,Firms,Firms, coefficientsFirms, and oror oror AwardedAwarded AwardedandAwarded isand an is an ArchitectsiserrorArchitectsArchitects error an error andandandand 𝑃𝑃𝑃𝑃 Awardedterm. Architects, 𝑃𝑃𝑃𝑃 Awarded Firms, or Awarded Architects𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼 dix section𝛽𝛽𝛽𝛽𝛽𝛽𝛽𝛽 𝛿𝛿𝛿𝛿■𝛿𝛿𝛿𝛿 𝜀𝜀𝜀𝜀𝜀𝜀𝜀𝜀𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 term.if building 𝑃𝑃𝑃𝑃 𝑋𝑋𝑋𝑋is𝑃𝑃𝑃𝑃 designed by Awarded 𝑃𝑃𝑃𝑃Architects, Awardedterm. Firms,FirmsFirms Firms orandand𝑃𝑃𝑃𝑃 Awarded and 00 otherwise.otherwise. 0 otherwise. Architects isis is ais aaand constant,constant,aconstant, constant, 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 andandand and𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃𝑔𝑔𝑔𝑔𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 areareare are estimatedestimatedestimated estimated coefficientscoefficients coefficientscoefficients andandand isisis ananan errorerrorerrorerror 𝛼𝛼𝛼𝛼term. 𝛽𝛽𝛽𝛽 𝛿𝛿𝛿𝛿 and Firms and 0 otherwise.𝜀𝜀𝜀𝜀𝑃𝑃𝑃𝑃 𝛼𝛼𝛼𝛼 𝑖 𝑖 is a constant𝛽𝛽𝛽𝛽, 𝑖 𝑖 and𝛿𝛿𝛿𝛿 𝑖 𝑖 are es- 𝜀𝜀𝜀𝜀𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 𝑔𝑔𝑔𝑔𝑃𝑃𝑃𝑃 term. 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 𝛼𝛼𝛼𝛼 UsingUsing the the logarithm logarithm𝛽𝛽𝛽𝛽 𝛿𝛿𝛿𝛿 of of transaction transaction price price instead instead of of the 𝜀𝜀𝜀𝜀the𝑃𝑃𝑃𝑃 transaction transaction pric pricee per per square square foot foot may may Firms and 0 otherwise. is a constant, and are estimated𝛼𝛼𝛼𝛼timatedterm.term.term.Using coefficients the logarithm 𝛽𝛽𝛽𝛽andand of 𝑖 transaction 𝑖 isis𝛿𝛿𝛿𝛿 anan erroerror pricer term. instead of the transaction price per𝜀𝜀𝜀𝜀 square𝑃𝑃𝑃𝑃 foot may This page intentionally left blank Using the logarithm𝑃𝑃𝑃𝑃 of transaction price instead of the transaction price per square foot𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼 maycausecause𝛼𝛼𝛼𝛼 measurement measurement𝛽𝛽𝛽𝛽𝛽𝛽𝛽𝛽𝛽𝛽𝛽𝛽 error error𝛿𝛿𝛿𝛿𝛿𝛿𝛿𝛿𝛿𝛿𝛿𝛿𝛿𝛿𝛿𝛿 for for the the size size and and construction construction cost 𝜀𝜀𝜀𝜀cost𝜀𝜀𝜀𝜀𝜀𝜀𝜀𝜀𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 may may vary vary over over the the size size of of the the term. Using cause the logarithm measurement of transaction error for the price size andinstead construction of the transaction cost may vary pric overe per the square size of thefoot may building.building. However, However, we we have have conducted conducted the the same same hedonic hedonic analysis analysis using using the the transaction transaction price price cause measurement errorUsing𝛼𝛼𝛼𝛼 for the thesize andlogarithm construction𝛽𝛽𝛽𝛽 𝛿𝛿𝛿𝛿 of transactioncostcause mayUsingUsingbuilding. Using varymeasurement thethe overprice the logarithmHowever,logarithm thelogarithm instead size error we ofof of𝜀𝜀𝜀𝜀 have 𝑃𝑃𝑃𝑃 transactionoftransaction forthe transaction of the conducted the size transactionpriceandprice theprice construction sameinsteadinstead instead hedonic ofof pricof the thecost the analysis e transactiontransaction transactionmayper using varysquare over the pricpric pric transaction eethe ee foot perperper size squaresquare squaremay of price the foot footfoot maymaymay perper sf sf for for every every model model and and a a substantial substantial pricing pricing diffe differencerence was was not not found. found. In In addition, addition, looking looking at at building.Using the However, logarithm we ofhave transaction conducted price the insteadsame hedonic of the transactionanalysisbuilding.causecausecausepercause using sf However, measurementmeasurementmeasurement forpric measurement everythee per transaction modelwe square have error errorerrorand error footconducted aprice for forforsubstantial for may thethethe the sizesizesize thesize pricing andand andsame and constructionconstruction constructiondiffe construction hedonicrence was analysis costcostcost 66notcost found. maymaymay mayusing varyvary varyInvary theaddition, over overover overtransaction thethe thelooking sizesizesize price at ofofof thethethe 67 per sf for every model andcause a substantial measurement pricing difference error was for not the found. size In andaddition, construction lookingthethe functional functional at cost form form may of of the the vary results, results, over the the thecoefficients coefficients size of and and the the the standard standard errors errors were were comparable. comparable. cause measurement error for the size and construction costperbuilding.building.building. sf maythe building.for functional everyvary However,However,However, overHowever, model form the of weandwewe sizethe we have havehavea results, havesubstantialof the conductedconductedconducted conducted the coefficients pricing thethethe the samediffesamesame sameandrence the hedonichedonichedonic hedonic standard was analysisnotanalysisanalysis analysis errors found. were usingusing usingusing In comparable.addition, thethethe transactiontransactiontransaction looking at pricepriceprice the functional form of the building.results, the However,coefficients and we the have standard conductedThis errors could weremean the comparable.the samesampleThisThis datahedonic could could set contains mean mean analysis the therelatively sample sample using similar data data scale theset set contains containsoftransaction buildings relatively relatively therefore price similar similar the scale scale of of buildings buildings therefore therefore the the building. However, we have conducted the same hedonicthe peranalysisperper functional per sfsfsf for forforsf using for everyeveryevery form every the modelmodelofmodel modeltransactionthe results, andandand and aaa substantialsubstantial asubstantial theprice substantial coefficients pricingpricingpricing pricing and diffediffediffe diffetherencerence rencestandardrence waswaswas was errors notnot notnot found.found. found.found. were Incomparable.InIn addition,addition,addition, lookinglooking looking at atat This could mean the sample data set contains relatively similar scaledifference of buildings in using transactionthereforedifferencedifference theprice and in in transactionusing using transaction transaction price per price pricesf diminishes. and and transaction transaction The regression price price per per sf sf diminishes. diminishes. The The regression regression per sf for every model andper a sf substantial for every pricing model diffe andrenceThis a thewasthethe substantialcould the functionalfunctional functionalnot functional meanfound. formtheformform In pricingform sampleaddition, ofofof theofthethe the results,datadifferesults,results, looking results, setrence thethecontainsthe atthe coefficientscoefficientscoefficients coefficientswas relatively not andandandfound. and similar thethethe the standardstandardstandard standardscaleIn addition, of errorsbuildings errorserrorserrors were werelookingwere therefore comparable.comparable.comparable. at the difference in using transaction price and transaction price perdifference sf analysisdiminishes. in using The the transaction transactionregressionanalysisanalysis price price and perusing using transactionsf can the the be transaction transaction found price in theper price price Appendix sf diminishes.per per sf sfsection. can can be beThe found found regression in in the the Appendix Appendix section. section. the functional form of thethe results, functional the coefficients form of and the the results, standardThisThisThis couldcould thecould errors meanmean coefficients mean were thethe the comparable.samplesample sample and datadata data the set set set containscontainsstandard contains relativelyrelatively relatively errors similarsimilar similar were scalescale scale comparable. of ofof buildings buildingsbuildings therefore therefore thethe analysis using the transaction price per sf can be found in theanalysis Appendix difference using section. the in transactionusing transaction price per price sf canand betransaction found in theprice Appendix per sf diminishes. section. The regression This could mean the sample data set contains relatively similardifferencedifferencedifference scale inininof using usingusingbuildings transactiontransactiontransaction therefore pricepriceprice the andandand transactiontransactiontransaction pricepriceprice perperper sfsfsf diminishes.diminishes.diminishes. TheTheThe regressionregressionregression This could mean the sample dataanalysis set using contains the transaction relatively price similar per sf can scale be found of buildingsin the Appendix therefore section. the difference in using transaction price and transaction price analysisperanalysisanalysis sf diminishes. usingusingusing thethethe The transactiontransactiontransaction regression pricepriceprice perperper sfsfsf cancancan bebebe foundfoundfound ininin thethethe AppendixAppendixAppendix section.section.section. difference in using transaction price and transaction price per sf diminishes. The regression analysis using the transaction price per sf can be found in the 6 RESULTSAppendix: section.AWARDED DESIGNS AND TRANSACTION PRICES 66 RESULTS RESULTS:: AWARDED AWARDED DESIGNS DESIGNS AND AND TRANSACTION TRANSACTION PRICES PRICES analysis using the transaction price per sf can be found in the Appendix section. 6 RESULTS: AWARDED DESIGNS AND TRANSACTION6 PRICES RESULTS : AWARDED DESIGNS AND TRANSACTION PRICES

6 RESULTS: AWARDED DESIGNS AND TRANSACTION PRICES 666 RESULTSRESULTS ::: AWARDEDAWARDEDAWARDED DESIGNSDESIGNSDESIGNS ANDANDAND TRANSACTIONTRANSACTIONTRANSACTION PRICESPRICESPRICES 6 RESULTS: AWARDED DESIGNS AND TRANSACTION PRICES

6 RESULTS: AWARDED DESIGNS AND TRANSACTION PRICES

RESULTS: AWARDED DESIGNS RESULTS: AWARDED ARCHITECTS & FIRMS AND TRANSACTION PRICES AND TRANSACTION PRICES

Regression Fixed Effects (1) (2) (3) Log (Price) Model 1 Model 2 Model 3 Regression Fixed Effects (1) (2) (3) Log (Price) Model 1 Model 2 Model 3 Awarded Architects 1.124*** 0.089 0.177* [0.293] [0.095] [0.102] Awarded Designs 1.530*** 0.171*** 0.231*** Awarded Firms 1.670*** 0.248** 0.321*** [0.144] [0.059] [0.059] [0.227] [0.113] [0.100] Awarded Architects & Firms 1.729*** 0.182*** 0.209*** Constant 17.721*** 6.757** 8.525*** [0.186] [0.062] [0.069] [0.239] [2.625] [2.960] Constant 17.704*** 6.836*** 8.494*** Location & Transaction Time FE YES YES YES [0.243] [2.635] [2.962] Building Features FE NO YES YES Transaction Features FE NO NO YES Location & Transaction Time FE YES YES YES Building Features FE NO YES YES Observations 846 846 846 Transaction Features FE NO NO YES R-squared 0.229 0.899 0.906 F Adj R-Squared 0.21 0.90 0.90 Observations 846 846 846 R-squared 0.232 0.900 0.906 Robust standard errors in brackets F Adj R-Squared 0.21 0.90 0.90 *** p<0.01, ** p<0.05, * p<0.1 Robust standard errors in brackets *** p<0.01, ** p<0.05, * p<0.1

Table 4 - Awarded Design, Base Case No Award Table 5 - Awarded Architects and Awarded Firms, Base Case No Award

Awarded Designs Buildings that are designed by awarded In addition, in terms of location, relative to Midtown West, Awarded Architects Buildings that are designed by archi- score, in addition to the fixed effects of Column (1). Column (3) architects and/or firms properties in Downtown are valued 30% less, properties in tects who won the lifetime achievement award controls for the transaction features by including buyer type, Midtown East are traded with a premium of 9%, and proper- seller type, and lender type, in addition to the fixed effects of Table 3 shows the regression analysis results for the inte- ties in the Upper East Side are transacted with a premium of Awarded Firms Buildings that are designed by architecture Column (1) and Column (2). grated transaction database. Having the logarithm of the 44.7%. Class A buildings compared to other building classes firms who won the AIA Architecture Firm Award transaction price as a dependent variable and relating it to are valued with a premium of 37%, and finally larger and taller Ceteris paribus, the regression results show all three treated a set of hedonic characteristics. The results explain 90.6% buildings transacted with a significant premium. All analysis Awarded Architects & Firms Buildings that are designed by categories, Awarded Architects, Awarded Firms, and Awarded of the variation in the logarithm of transaction price with an has also been modeled for the log of the price per square foot. architects/firms who won both the lifetime achievement award Architects and Firms show 17.7%, 32.1%, and 20.9% transac- adjusted R-squared of 90%. Column (1) to (3) measures the Results are statistically similar and available upon request ■ and AIA Architecture Firm Award tion price premium respectively compared to non-awarded different fixed effects. Column (1) controls for the location and buildings with a positive and significant coefficient. Coeffi- transaction time. Column (2) controls for building features by We further broke down the “Awarded Designs” variable to cients for each variable used in the regression analysis can be adding variables such as building age, number of floors, build- “Awarded Architects”, “Awarded Firms”, and “Awarded Archi- found in the Appendix. ing area, land parcel area, building class, renovation, and walk tects and Firms”. Table 4 shows the regression analysis results score, in addition to the fixed effects of Column (1). Column (3) for the integrated transaction database. Having the logarithm Keeping constant the observable characteristics, the result controls for the transaction features by including buyer type, of the transaction price as a dependent variable and relating it of the regression suggests Awarded Firms achieve a higher seller type, and lender type, in addition to the fixed effects of to a set of hedonic characteristics. The results explain 90.6% sales premium compared to Awarded Architects and Awarded Column (1) and Column (2). of the variation in the logarithm of transaction price with an Architects and Firms. Given the significant positive coefficient adjusted R-squared of 90%. Column (1) to (3) measures the of building age and building size, this may be related to the The regression result shows, ceteris paribus, Awarded De- different fixed effects. Column (1) to (3) measures the different fact that the buildings designed by Awarded Firms are on av- signs are transacted with a 23.1% premium compared to fixed effects. Column (1) controls for the location and transac- erage bigger and more recently built compared to the other non-awarded buildings with a positive and significant coef- tion time. Column (2) controls for building features by adding award categories ■ ficient. Coefficients for each variable used in the regression variables such as building age, number of floors, building analysis can be found in the Appendix section. area, land parcel area, building class, renovation, and walk

68 69 results: awarded designs and transaction prices awarded architects & firms and transaction prices

35 32.1%*** (0.100)

23.1% *** 25 (0.059) 20.9% *** (0.069) 17.7%* 20 (0.102)

0 (%)

Awarded Awarded Awarded Awarded Designs Architects Firms Architects & Firms

Table 4 Result Table 5 Result

Significance*** Coefficient % NOTE: Asterisks in a regression table indicate The Regression Coefficient tells about The regression model controls for transaction features (buyer type, seller the level of the statistical significance the change in the value of dependent location and transaction time, building type, and lender type) of a regression coefficient. variable corresponding to the unit features (age, number of floors, build- change in the independent variable. ing area, land parcel area, building *** p<0.01, ** p<0.05, * p<0.1 class, renovation, and walk score), and

Figure 17 - Regression Analysis Result Comparison

Note: Ceteris paribus, the hedonic analysis result shown in Figure 15 indicates that buildings designed by awarded architects/firms are transacted with a 23.1% premium relating to buildings that are designed by non-awarded architects. We further specified the study by looking into different type of awards with three categorical variables, Awarded Architects, Awarded Firms, and Awarded Architects & Firms. The result suggests that ceteris paribus, Awarded Architects, Awarded Firms, and Awarded Archi- tects and Firms show 17.7%, 32.1%, and 20.9% transaction price premium respectively compared to non-awarded buildings. This page intentionally left blank

70 71 ROBUSTNESS: all architects & firms and transaction prices ROBUSTNESS: ALL ARCHITECTS & FIRMS AND TRANSACTION PRICES

As a robustness check, we specified the Awarded Design cat- nificantly different from zero. However, these designers have egory by expanding the categorical variable to include data on small samples. individual architects/firms to ensure that no one designer was driving the result of one building. Looking into the architects and the associated buildings that have a significant positive coefficient, a number of overlap- We have studied the regression analysis result on the relative ping influencing factors for the transaction premium can be transaction premiums associated with awarded architects found. To name a few, the buildings that are showing a trans- using the non-awarded architects as a base case. The result action premium have achieved a landmark status or became explained 90.7% of the variation in the logarithm of transac- a recognizable building for a certain industry, the most innovation price with an adjusted R-squared of 90%. The regression tive construction technology was deployed at the time of con- model used the same methodology of three fixed effect mod- struction, rich historical and cultural background, high-quality els that have been previously described and the regression interior, and amenities such as a celebrated restaurants or result represents the coefficients and significance of each cafes are in place. In addition, a number of observations were awarded architects based on the results of the model (3) likely to be valued higher due to the potential development which controls for location and transaction time, building fea- value associated with the land or as a reflection of the major tures, and transaction features. renovation of the building.

Ceteris paribus, the results mainly have a low statistical sig- However, the result should be considered with extreme cau- nificance, but some with a cautionary threshold indicate that tion. There is a probability of measurement error due to the compared to buildings designed by non-awarded architects. small number of differentiating samples related to each archi- The architect Ludwig Mies van der Rohe achieved the most tect. In other words, the result does not suggest performance transaction premium of 60.6% with a p-value less than 0.001. measurements for individual architecture firms since the re- Followed by KPF (51.4%), and Hugh Stubbins and Associates sults are based on a small differentiating sample. On average (27.4%). Positively awarded designers represent one-third of 4.9 and 2.5 transaction observations were found per architect ■ the transaction sample, the others are not statistically sig- This page intentionally left blank

72 73 conclusion Conclusion

left, image 5 Rem Koolhaas & Madelon Vriesendorp, 1972, The CIty of the Captive Globe, Delirious New York, page 294-295

This study expanded the conversation ed skillfulness or profitability of individ- who have won both the lifetime achieve- small number of observations related integrating the related cost associated As a first step, a creative approach in on the value of design by investigating ual architects/firms. In essence, this ment award/innovation award and AIA to each architect and firms. The result with design and identifying other poten- gathering a new set of data related to the transaction prices of commercial study is an endeavor to recognize the Architecture Firm Award. The result sug- does not suggest any measurements tial omitted variables: for example, the the design performance is needed. For buildings associated with award-win- authorship of each building and study gests that ceteris paribus, Awarded Ar- related to the architecture firm’s perfor- premium associated with the awarded example, the design-related elements ning architects/firms and their building their impact on the price dynamics of chitects, Awarded Firms, and Awarded mance since the results are based on a architects may be influenced by the that we found in the buildings that have design in Manhattan over the 2000 to commercial office buildings in Manhat- Architects and Firms show 17.7%, 32.1%, small differentiating sample. On average larger SF of a building, higher construc- gained significant transaction premium, 2017 period. tan, New York. and 20.9% transaction price premium 4.9 transaction observations were found tion budget, better interior quality, or such as iconicity, relevance in a certain respectively compared to non-awarded per architect. other endogenous factors. industry, adopting the most innovative In line with the relative studies using the When controlling for location and buildings. construction technology, rich cultural awards as means to identify architects transaction time, building features and In general, this study has focused on a In recent years, due to increased and historical background, high-quality who have won high status among their transaction features, the result of the Moreover, we have expanded the cat- quite precise subject that is to under- market education and growth in the interior, can be studied as new data peers, this study intended to broaden hedonic analysis suggests that build- egorical variable to study the relative stand and acknowledge the architect’s number of leading examples, invest- points in the future studies. We believe the scope of understanding design ings designed by awarded architects/ transaction premiums associated influence on the commercial office ing in high-quality design became a combining the new measurements with by adding substantially more controls firms are transacted with a 23.1% pre- with awarded architects using the building’s transaction value. In other standard for the real estate market in the accumulated knowledge on design in the model: incorporating buyer and mium than buildings that are designed non-awarded architects as a base words, it was a mere attempt to figura- New York. Despite the growing interest, generated by architects will enable us seller decisions for all of the transaction by non-awarded architects. We further case. Ceteris paribus, the result shows tively understand the value associated however, a limited number of studies to open up a substantial area for future over time, specifying the type of awards, specified the study by looking into dif- a significant transaction premium of with the designers of the built environ- and discussions have been generated research regarding the value of design, and adding information on architects to ferent type of awards with three cate- 60.6% on the buildings designed by ment. The results indeed show a signifi- to help create a shared better valued and moreover will help create an agency every building in the sample data set. gorical variables, Awarded Architects: Ludwig Mies van der Rohe relative to cant premium associated with the archi- surrounding the subject of design. In for design in the realm of finance and The added variables improved the over- who have won lifetime achievement the buildings designed by non-awarded tects, however, it only illustrates a small this study, we have identified that the economics ■ all fit of the model as well as explaining awards and/or contemporary innovation architects, followed by KPF (51.4%), and part of the relationship of design and difficulty of obtaining data related to the variation in price. It is important to awards, Awarded Firms: companies that Hugh Stubbins and Associates (27.4%). value that is based on price rather than design performance being one of the bear in mind that this empirical study have received the AIA Architecture Firm However, the suggestive results should the performance of the building design. biggest hurdles in enabling further stud- does not intend to measure the weight- Award, and Awarded Architects & Firms: be treated with extra caution due to the Further improvement can be made by ies to disentangle the value of design.

74 75 references References

Asabere P. K., Hachey G. & Grubaugh S., 1989, Architecture, historic zoning, and the value of homes, Journal of Real Nase I., Berry J. & Adair A., 2016, Impact of quality-led design on real estate value, Journal of Property Research, 33:4, Estate Finance and Economics, 2(3), pp. 181–195. page 309–331

Booth R., 2006, Architect Foster brands it like Beckham to grab a fee, The Sunday Times, 22 October, page 42 Fuerst F., McAllister P. & Murray C. B., 2011, Designer Buildings: Estimating the economic value of 'signature' architecture, Journal of Environment & Planning, 43, page 166–184 Charles H., 2014, Architect will crit for coffee: The economies of architecture,Yale School of Architecture Journal Perspecta 427: Money, MIT Press, page 156–161 Vandell K. D. & Lane J. S., 1989, The Economics of Architecture and Urban Design, AREUEA Journal, 17, No. 2

CityRealty, 2018, The CityRealty 100 index, Retrieved from, https://www.cityrealty.com/nyc/building-indices/the-cityrealty-100/1

CityRealty, 2018, Star-architect condo index, Retrieved from, https://www.cityrealty.com/nyc/building-indices/starchitect-condos/building-list/112

Coxe W.. Hartung N. F., Hochberg H. H., Lewis B. J. Mattox R. F. and Piven P. A. 1986. Charting Your Course: Master Strategies for Organizing and Managing Architectural firms.Architectural Technology, May/June, 52–58.

Deamer P., 2014, WORK, Yale School of Architecture Journal Perspecta 427: Money, MIT Press, page 28–40

Fein S., 2007, Starchitect branding and the cost of “effortless living” or another episode in the continuing quest for social status through design, Harvard Design Magazine, 26, retrieved from http://www.gsd.harvard.edu/hdm.

Foster H., 2008, Image Building. In Architecture Between Spectacle and Use, Williamstown, Massachusetts: Sterling and Fran- cine Clark Art Institute, page 164–179

Frampton K., 2005, The work of architecture in the age of commodification,Harvard Design Magazine, 23, page 1–5

Ponzini D., (2014) The Values of Starchitecture: Commodification of Architectural Design in Contemporary Cities, Organizational Aesthetics: Vol. 3: Iss. 1, 10–18.

Ponzini D., Nastasi M. 2011. Starchitecture. Scenes, Actors and Spectacles in Contemporary Cities. Turin: Allemandi.

Rogers E. M., 1962, Diffusion of Innovations, New York, Free Press of Glencoe

Song Y. & Knaap, G, 2003, New urbanism and housing values: A disaggregate assessment, Journal of Urban Economics, page 218–238.

Hough D. E. & Kratz C. G., 1982, Can 'Good' Architecture Meet the Market Test?, Journal of Urban Economics, 14, page 40–54.

76 77 [0.115] Transaction 2017 1.221*** appendix [0.116] APPENDIX Age -0.006** Regression Results: Awarded design (base case no award) [0.003] Regression Results: Awarded Architects & Firms (base case no award) Age Squared 0.000** [0.000] Regression Fixed Effects (1) (2) (3) Number Floors 0.008*** Regression Fixed Effects (1) (2) (3) [0.002] Log (Price) Model 1 Model 2 Model 3 Log(SqFt) 0.724*** Log (Price) Model 1 Model 2 Model 3 [0.039] Awarded Designs 1.530*** 0.171*** 0.231*** Log(Land SqFt) 0.021 Awarded Architects 1.124*** 0.089 0.177* [0.042] [0.144] [0.059] [0.059] Class A 0.369*** [0.293] [0.095] [0.102] [0.087] Awarded Firms 1.670*** 0.248** 0.321*** Constant 17.721*** 6.757** 8.525*** Class B 0.050 [0.227] [0.113] [0.100] [0.060] [0.239] [2.625] [2.960] Class C -0.210** Awarded Architects & Firms 1.729*** 0.182*** 0.209*** [0.086] [0.186] [0.062] [0.069] Location & Transaction Time FE YES YES YES Renovated -0.004 Building Features FE NO YES YES [0.039] Constant 17.704*** 6.836*** 8.494*** Walk Score -0.001 Transaction Features FE NO NO YES [0.029] [0.243] [2.635] [2.962] BT Corp -0.042 Observations 846 846 846 [0.111] Location & Transaction Time FE YES YES YES BT Fund 0.159* R-squared 0.229 0.899 0.906 [0.091] Building Features FE NO YES YES F Adj R-Squared 0.21 0.90 0.90 BT Gov't -0.089 Transaction Features FE NO NO YES [0.086] BT Inst 0.125* Observations 846 846 846 Robust standard errors in brackets [0.070] *** p<0.01, ** p<0.05, * p<0.1 BT Offshore 0.105 R-squared 0.232 0.900 0.906 [0.082] F Adj R-Squared 0.21 0.90 0.90 BT Private -0.106** [0.044] BT REIT 0.086 Robust standard errors in brackets [0.090] *** p<0.01, ** p<0.05, * p<0.1 Regression Results: Awarded design (base case no award) BT REOC -0.653** Regression Results: Awarded Architects & Firms (base case no award) [0.281] BT Retailer -0.233* Treated Regression (3) Treated Regression (3) [0.112] Variable (3) Variable [0.138](3) Variable (3) Variable (3) Log (Price) Model 3 ST Corp 0.006 Log (Price) Model 3 Transaction 2015 1.267***[0.112] [0.115] [0.101] Transaction 2015 1.267***[0.101] ST Corp 0.007 Awarded Designs 0.231*** Transaction 2017 1.221*** ST Gov't 0.229*** Awarded Architects 0.177* Transaction 2016 1.378***[0.101] [0.101] [0.059] [0.116] [0.083] [0.102] Transaction 2016 1.378***[0.115] ST Gov't 0.228*** ST Inst 0.146 Awarded Firms 0.321*** Transaction 2017 1.201***[0.115] [0.083] Downtown -0.299*** Age -0.006** [0.132] [0.100] Transaction 2017 1.201***[0.113] ST Inst 0.142 [0.053] [0.003] ST Offshore 0.096 Awarded Architects & Firms 0.209*** [0.113] [0.129] Midtown East 0.090** Age Squared 0.000** [0.075] [0.069] Age -0.006** ST Offshore 0.102 [0.045] [0.000] ST Private 0.054 Age -[0.003]0.006** [0.075] Midtown South 0.068 Number Floors 0.008*** [0.094] Downtown -0.300*** Age Squared 0.000**[0.003] ST Private 0.058 [0.085] [0.002] ST REIT 0.164** [0.053] Age Squared 0.000**[0.000] [0.094] Upper East Side 0.447*** Log(SqFt) 0.724*** [0.066] Midtown East 0.093** Number Floors 0.008***[0.000] ST REIT 0.160** [0.125] [0.039] ST REOC 0.487** [0.045] Number Floors 0.008***[0.002] [0.064] Transaction 2001 0.046 Log(Land SqFt) 0.021 [0.237] Midtown South 0.074 Log(SqFt) 0.723***[0.002] ST REOC 0.484** [0.112] [0.042] ST retailer 0.306 [0.086] Log(SqFt) 0.723***[0.039] [0.238] Transaction 2002 0.179 Class A 0.369*** [0.253] Upper East Side 0.447*** Log(LandSqFt) [0.039]0.021 ST retailer 0.308 [0.128] [0.087] LT CMBS 0.080 [0.125] Log(LandSqFt) [0.042]0.021 [0.250] Transaction 2003 0.124 Class B 0.050 [0.054] Transaction 2001 0.042 Class A 0.365***[0.042] LT CMBS 0.081 [0.116] [0.060] LT Financial -0.107 [0.111] Class A 0.365***[0.087] [0.054] Transaction 2004 0.268*** Class C -0.210** [0.083] Transaction 2002 0.176 Class B [0.087]0.050 LT Financial -0.107 [0.088] [0.086] LT Government Agency -0.159 [0.128] Class B [0.060]0.050 [0.083] Transaction 2005 0.495*** Renovated -0.004 [0.131] Transaction 2003 0.123 Class C -0.211**[0.060] LT Government Agency -0.158 [0.099] [0.039] LT Insurance -0.010 [0.116] Class C -[0.086]0.211** [0.131] Transaction 2006 0.727*** Walk Score -0.001 [0.062] Transaction 2004 0.266*** Renovated [0.086]-0.002 LT Insurance -0.011 [0.097] [0.029] LT International Bank 0.095 [0.088] Renovated [0.039]-0.002 [0.063] Transaction 2007 0.979*** BT Corp -0.042 [0.061] Transaction 2005 0.495*** Walk Score [0.039]-0.000 LT International Bank 0.094 [0.095] [0.111] LT National Bank 0.014 [0.099] Walk Score [0.030]-0.000 [0.061] Transaction 2008 1.033*** BT Fund 0.159* [0.064] Transaction 2006 0.724*** BT Corp [0.030]-0.042 LT National Bank 0.014 [0.108] [0.091] LT Pension Fund 0.596** [0.097] BT Corp [0.111]-0.042 [0.064] Transaction 2009 0.688*** BT Gov't -0.089 [0.240] Transaction 2007 0.975*** BT Fund [0.111]0.160* LT Pension Fund 0.604** [0.172] [0.086] LT Private 0.071 [0.095] BT Fund [0.091]0.160* [0.240] Transaction 2010 0.518*** BT Inst 0.125* [0.075] Transaction 2008 1.030*** BT Gov't [0.091]-0.088 LT Private 0.071 [0.134] [0.070] LT Regional/Local Bank -0.136* [0.108] BT Gov't [0.086]-0.088 [0.074] Transaction 2011 0.739*** BT Offshore 0.105 [0.074] Transaction 2009 0.686*** BT Inst [0.086]0.124* LT Regional/Local Bank -0.136* [0.095] [0.082] [0.171] BT Inst [0.069]0.124* [0.074] Transaction 2012 0.864*** BT Private -0.106** Constant 8.525*** Transaction 2010 0.516*** BT Offshore [0.069]0.113 [0.105] [0.044] [2.960] [0.134] BT Offshore [0.082]0.113 Constant 8.494*** Transaction 2013 1.037*** BT REIT 0.086 Transaction 2011 0.736*** BT Private -0.107**[0.082] [2.962] [0.120] [0.090] Observations 846 [0.096] BT Private -[0.044]0.107** Transaction 2014 1.206*** BT REOC -0.653** R-squared 0.906 Transaction 2012 0.861*** BT REIT [0.044]0.080 Observations 846 [0.112] [0.281] F Adj R-Squared 0.90 [0.105] BT REIT [0.090]0.080 R-squared 0.906 Transaction 2015 1.269*** BT Retailer -0.233* Transaction 2013 1.034*** BT REOC -0.633**[0.090] F Adj R-Squared 0.90 [0.120] BT REOC -[0.310]0.633** [0.101] [0.138] Robust standard errors in brackets Transaction 2014 1.197*** BT Retailer [0.310]-0.226 Transaction 2016 1.390*** ST Corp 0.006 *** p<0.01, ** p<0.05, * p<0.1 Robust standard errors in brackets [0.101] [0.112] BT Retailer [0.138]-0.226 *** p<0.01, ** p<0.05, * p<0.1 ST Gov't 0.229*** Transaction 2015 1.267*** [0.138] [0.083] [0.101] Transaction 2016 1.378*** 78 [0.115] 79 Transaction 2017 1.201*** [0.113]

Age -0.006** [0.003] Age Squared 0.000** [0.000] Number Floors 0.008*** [0.002] Log(SqFt) 0.723*** [0.039] Log(LandSqFt) 0.021 [0.042] Class A 0.365*** [0.087] Class B 0.050 [0.060] Class C -0.211** [0.086] Renovated -0.002 [0.039] Walk Score -0.000 [0.030] BT Corp -0.042 [0.111] BT Fund 0.160* [0.091] BT Gov't -0.088 [0.086] BT Inst 0.124* [0.069] BT Offshore 0.113 [0.082] BT Private -0.107** [0.044] BT REIT 0.080 [0.090] BT REOC -0.633** [0.310] BT Retailer -0.226 [0.138] acknowledgments

I'm eternally grateful for the optimism, encouragement, and knowledge of my brilliant advisor, Dr. Andrea Chegut.

I would also like to express gratitude to my talented colleagues at MIT Real Estate Innovation Lab and MSRED.

And finally, as ever, thanks and love to my family.

This page intentionally left blank

80 81