Pricing biophilia: what the evidence is worth

Key findings

• Office workers with the best window views and lighting quality used 53 hours of sick leave per year; those with the worst used 71. Ihab Elzeyadi measured this on payroll records, not questionnaires. View and lighting quality together explain 6.5 percent of sick-leave variance (Elzeyadi, 2011, USGBC Greenbuild). The effect is real, small and routinely overquoted as 10 percent.
• Green-certified offices carry rent premiums of 3 to 5 percent in peer-reviewed transaction studies (Eichholtz, Kok and Quigley, 2010, American Economic Review; Fuerst and McAllister, 2011, Real Estate Economics), 2.5 to 2.9 percent once panel models strip out building quality (Reichardt et al., 2012, Journal of Real Estate Research) and 7 to 11 percent in broker research (JLL, 2023).
• The most cited dollar figure in the field, Terrapin Bright Green’s 93 million dollars in annual US healthcare savings, contains a units error of a factor of one thousand. Its own cited source counts procedures in thousands; Terrapin read the number as absolute.
• No neutral statistic measures how far biophilic design has penetrated the market. Every percentage in circulation comes from a seller. Sadick, Kamardeen and Vu interviewed practitioners in the only peer-reviewed adoption research and describe uptake as limited (2023, Journal of Building Engineering).
• The narrow core of the evidence, window views, daylight quality without glare and certification premiums, is replicated across independent data types; the package sold as biophilic design is not. The confident label on this report covers the core, not the package.

What the primary studies actually show

The economic case for putting nature into buildings rests on a handful of studies that are cited far more often than they are read. I read them. The case survives, but it changes shape.

The Heschong Mahone Group analyzed test scores of 21,344 elementary students in three school districts for Pacific Gas and Electric and the California Board for Energy Efficiency (Heschong Mahone Group, 1999, Daylighting in Schools). The famous result, students progressing 20 percent faster in math and 26 percent faster in reading, describes one of the three districts: Capistrano in Orange County, the only one where 8,166 students could be tracked across a school year. And it describes the classrooms rated highest for overall daylight, not the ones with the biggest windows; window area alone came in at 15 and 23 percent. Seattle and Fort Collins delivered year-end scores only, 7 to 18 percent higher in the most daylit classrooms. The authors said plainly that the design was correlational and that the causal mechanism was beyond their scope.

What the compilations leave out is the replication history. The same group re-ran the Capistrano analysis in 2003 for the California Energy Commission and confirmed it: 21 percent faster learning as the central tendency across twelve models, with teacher assignment bias ruled out (Heschong Mahone Group, 2003, Reanalysis Report). Then they tried to replicate the finding in Fresno with more than 8,000 students, and the daylight effect vanished. The daylight rating had “the least explanatory power of the set of variables considered”. What survived Fresno was more precise: window view helped, glare and sun penetration hurt (Heschong Mahone Group, 2003, Windows and Classrooms). The authors suspected classroom acoustics had confounded the original daylight rating. Twenty years on, the defensible classroom claim is views and glare control, not glass area.

Ihab Elzeyadi matched two years of sick-leave records from the payroll department against the window views and lighting quality of 119 employees in a 1970s university building in Oregon (Elzeyadi, 2011, USGBC Greenbuild). Employees in the best conditions used 53 hours of sick leave a year, those in the worst 71, a difference of more than two working days. View quality and lighting quality carried equal weight, view alone explaining about 4.5 percent of the variance and both together 6.5 percent. The 10 percent that circulates includes a third variable, glazing area, and the claim that views were the strongest single predictor is not in the paper. One building, a cross-section, no causal design. The strength of the study is its outcome data: recorded absence from payroll files, not how people said they felt.

That distinction matters because the biggest biophilia numbers are feelings. Cary Cooper’s team surveyed 7,600 office workers in 16 countries for Human Spaces, a research program funded by Interface, the world’s largest carpet-tile maker (Human Spaces, 2015). Workers in offices with natural elements reported 15 percent higher wellbeing, 6 percent higher productivity and 15 percent higher creativity than workers in offices without them. These are self-reports in a cross-sectional survey. And for its sick-leave claims the report cites Elzeyadi, so the two sources do not count as independent evidence on absence.

Retail has two studies that get merged into one. Kathleen Wolf surveyed shoppers on what they would pay in business districts with and without mature trees and found stated premiums of 9 percent in small cities and 12 percent in large ones (Wolf, 2005, Journal of Forestry 103(8):396-400). Those are survey answers with mail response rates of 15 percent or less, not transactions. The 8.8 percent often cited alongside comes from her later strip-mall study and does not describe city centers at all. Lisa Heschong tracked monthly sales for an anonymous California retail chain across 73 outlets, 24 of them daylit through skylights, and found a chain-wide daylight effect on sales of 0 to 6 percent depending on model and period, with the strongest reading, 5.7 percent, during 2001, when the state’s power crisis forced all stores to half lighting power (Heschong, 2003, Daylight and Retail Sales, California Energy Commission). The 40 percent that circulates is her 1999 finding for a different chain; in the 2003 study it survives only as the upper bound for the most favorable stores. The transaction data show a small positive effect, and the authors say themselves it could dip below zero in another sample of stores.

The compilation problem

Terrapin Bright Green, a New York consultancy that sells biophilic design strategy, published The Economics of Biophilia in 2012. It is the most cited economic document in the field, and its headline numbers are where the case goes wrong.

The 93 million dollars in annual US healthcare savings is an extrapolation from the 1984 study in which Roger Ulrich showed that patients facing trees left hospital three quarters of a day earlier than patients facing a brick wall (Ulrich, 1984, Science 224(4647):420-421; covered in detail in TBR-01). Terrapin multiplied the shorter stay across 44,993 surgical procedures. Its own cited source, a CDC hospital survey, reports procedures in thousands: the table entry 44,993 means roughly 45 million procedures. Read consistently, Terrapin’s arithmetic yields about 93 billion dollars, a number nobody would have printed. The point is not which figure is right. A thousand-fold inconsistency sat in the field’s flagship number for over a decade, quoted by hundreds of publications, and apparently nobody checked.

The 2,000 dollars per office worker per year is labeled as what daylighting “can save”. The underlying figure, 2,074 dollars, is the total cost of absenteeism per US private-sector employee from Department of Labor statistics: 62.4 lost hours a year valued at average wages. It is the size of the whole problem, not the effect of any design intervention, and no published study licenses attributing the full amount to design.

The 247.5 million dollars in lost parental wages from New York school absences multiplies an attendance difference from a 1996 report by Innovative Design, an architecture firm describing its own school buildings (Nicklas and Bailey, 1996), with an EPA assumption of 75 dollars in lost wages per missed school day and 1.1 million New York students. An unreviewed observation scaled by an unrelated assumption is not evidence but marketing with citations.

Below the compilations sits a rawer layer. A “5 to 10 percent rent premium for biophilic offices” and “16 percent lower tenant turnover” both trace back to one unsourced blog post by RealVal, a valuation marketing site, which attributes the turnover figure to a 2022 JLL study that does not appear to exist. The claim that biophilic design has reached less than 5 percent of the 16.6 billion dollar US office furniture market (market size per Grand View Research, 2024) comes from VMC Group, a design practice selling biophilic services. The claim that only half of architects have foundational knowledge of biophilic design comes from a living-wall manufacturer’s online self-assessment with no sample description (GrowUp Greenwalls, 2026). None of these numbers survives a source check.

What the market actually pays

The premiums that survive review are smaller and better documented. Piet Eichholtz, Nils Kok and John Quigley compared green-certified US office buildings against control buildings in the same locations and found rents about 3 percent higher, effective rents above 7 percent higher and selling prices about 16 percent higher (2010, American Economic Review). Franz Fuerst and Pat McAllister found rent premiums of 4 to 5 percent and sales premiums of 25 percent for LEED on CoStar data (2011, Real Estate Economics). Alexander Reichardt, Franz Fuerst, Nico Rottke and Joachim Zietz then ran panel models that control for unobserved building characteristics, and the rent premium settled at 2.5 percent for Energy Star and 2.9 percent for LEED (2012, Journal of Real Estate Research). Broker research reads higher: JLL puts green rent premiums at 7.1 percent for North American Class A stock, 9.9 percent in Asia and 11.6 percent in London (JLL Research, 2023). Avis Devine and Nils Kok found tenants stay longer in certified buildings, with lease renewal rates 5.6 percent higher in their Canadian sample (2015, Journal of Portfolio Management).

Views carry prices in housing too. Joke Luttik analyzed Dutch housing transactions and found premiums of 8 to 10 percent for water views and 6 to 12 percent for views of open landscape (2000, Landscape and Urban Planning). Earl Benson and colleagues graded ocean views in Bellingham, Washington and found everything from 8 percent for a partial glimpse to nearly 60 percent for an unobstructed view near the waterline (1998, Journal of Real Estate Finance and Economics).

Two caveats keep these numbers honest. First, certification is not biophilic design. LEED prices an energy label, not plants and views, and a clean price for the biophilic package itself does not exist. Second, the premiums may be inflated. Ben Dalton and Franz Fuerst pooled the premium literature in a meta-analysis and warn of selection effects, since certified buildings are newer and better located in ways the models miss, and of publication bias; they call the literature “disjointed and at least partly inconclusive” (2018, Routledge Handbook of Sustainable Real Estate). Fuerst stands on both sides of this ledger: he measured premiums and questions how clean they are. That is the honest state of the art, a real price signal of uncertain exact size.

Why adoption stays marginal

Two readings of this evidence collide. Terrapin, Interface and the vendors quoting them read it as settled science with large returns, waiting for the market to wake up. The skeptics read the foundation as soft. Yannick Joye and Agnes van den Berg examined the evolutionary story behind biophilia, the claim that humans are hardwired to recover around greenery, and concluded that “neither current empirical evidence nor conceptual arguments provide any strong support” for it (2011, Urban Forestry & Urban Greening). Kaitlyn Gillis and Birgitta Gatersleben reviewed the psychological literature on biophilic design and found solid evidence for nature exposure and daylight, sparse evidence for natural materials and little research on the combined patterns that biophilic design actually prescribes (2015, Buildings).

Position

Both camps overstate. The skeptics are right about the theory and the package, wrong to let that shadow the core findings, which need no evolutionary story to be true: views correlate with lower recorded absence, glare-free daylight with better learning outcomes, certification with rents. The industry is right that a real effect exists, wrong about its size and wrong to sell the broad package on the narrow core’s evidence. My reading of the adoption gap follows from that mismatch: the bankable findings are narrow while the product on sale is broad, and buyers sense it. Sadick, Kamardeen and Vu name budget restrictions, maintenance load and client influence as the barriers (2023, Journal of Building Engineering; Sadick and Kamardeen, 2024). Those are rational responses to an offer whose strongest numbers come from sellers. Add the split incentive, in which the developer pays for what the tenant harvests, and the absence of any code requirement, and marginal adoption is what an economist would predict. Closing the gap is translation work: moving the narrow findings into the spreadsheets where buildings are actually decided, and leaving the seller numbers out.

The sectors holding the strongest evidence, schools and hospitals, are the sectors with no market to price it. Office landlords capture a premium and act on it; the pricing studies show them already paying for views and labels. A school district captures no rent. Its gain shows up in absence statistics and learning outcomes that no budget line connects to window specification. Returns with a private owner get collected; public returns need a policy decision before anyone collects them.

Where the research stands

Hung and Zhong each reviewed the health evidence and both extend the case (Hung, 2021; Zhong, 2022). Adoption research is younger and thinner: the Australian interview studies by Sadick and Kamardeen are the first systematic look at why practitioners do not specify what the evidence supports. What is missing: a hedonic study that prices biophilic interventions as such, separate from certification; a replication of the classroom findings outside California with Fresno’s acoustics lesson designed in; and any neutral adoption statistic. ASID’s trend reports list biophilia among wellness trends but publish no penetration figure, and no statistical agency tracks the category. Until those exist, every market-size claim in this space should be read as advertising.

Implications

For institutional real estate investors

The bankable spread sits on certification and view quality: 2.5 to 5 percent on rents in the academic studies, more if you trust broker research, plus measurably better tenant retention (Devine and Kok, 2015). Underwrite that, and treat package claims like a 5 to 10 percent biophilic premium as unpriced upside at best. Dalton and Fuerst’s caveat belongs in the model: assume part of any measured premium is selection. In a flight-to-quality market the refurbishment play with the best risk profile is view access, glare control and daylight quality, the features the primary studies actually support.

For public building owners in schools and healthcare

The strongest evidence has been public for 26 years and was publicly funded. Fresno sharpened it: buy views and glare control, not glass area. The translation lever is administrative. Absence already sits in your budget lines and window specification sits in capital planning; the two never meet at renovation time. A funding program that conditions money on view and glare criteria would move more than another study, because the market will not do this for you. Nobody in it captures your returns.

For office landlords

JLL counted 207 million square feet of US office leasing in 2025, up 5.2 percent on 2024, with asking rents stagnant overall while trophy and Class A rents grew (JLL, US Office Market Dynamics, Q4 2025). Savills reads Europe similarly: take-up stable year on year, vacancy at 9.0 percent, prime rents up 3.4 percent and prime vacancy it describes, anecdotally, at 2 to 3 percent (Savills, European Office Leasing, Q4 2025). Demand is concentrating on the best space. Verified nature features, meaning views, daylight quality and outdoor access, are among the few differentiators with primary evidence behind them. Specify them from the studies rather than from vendor decks and they hold up in a negotiation with a tenant’s analyst.

For retail districts and city associations

Kathleen Wolf’s street-tree numbers are advocacy-grade, not forecast-grade: stated willingness to pay, 9 to 12 percent by city size, from low-response mail surveys. Lisa Heschong’s skylight study is transaction-grade and small, 0 to 6 percent chain-wide. Quote each as what it is and the case survives a hostile analyst; quote 12 or 40 percent as measured till results and the whole argument becomes attackable.

Stop doing

Stop quoting Terrapin’s dollar figures; the arithmetic behind the 93 million is broken and the 2,000 dollars is mislabeled labor statistics. Stop citing the 5 to 10 percent biophilic premium, the 16 percent turnover figure, the sub-5 percent penetration and the 50 percent architect-knowledge number; all four come from sellers and none survives a source check. And stop commissioning surveys of how people feel about plants. The field has enough of those. The missing data is transactional: rents, absence records, sales.

Methodology

I checked every number in this report against its primary source in June 2026: the original PDFs of Heschong Mahone Group 1999 and all three 2003 follow-ups including Daylight and Retail Sales, Elzeyadi 2011, both Wolf 2005 papers, Terrapin 2012 including its appendix calculations, Human Spaces 2015, the pricing studies (Eichholtz, Kok and Quigley 2010; Fuerst and McAllister 2011; Reichardt et al. 2012 in the working-paper text; Devine and Kok 2015; Luttik 2000; Benson et al. 1998) and the market reports (JLL Q4 2025 in the original PDF, Savills Q4 2025, JLL Research 2023). 29 sources are named in the text. Vendor claims that failed sourcing are named as such rather than silently dropped. Limits: the schools, sick-leave and retail studies are correlational, and the retail authors themselves flag sign uncertainty; Reichardt’s figures are quoted from the working-paper version of the published article; Savills marks its prime-vacancy figure as anecdotal; no neutral adoption statistic exists, so every penetration statement here stays qualitative. The cross-study synthesis and the position are mine.