Impact of Environmental Investments on Corporate Financial Performance: Decomposing Valuation and Cash Flow Effects

Market-based measures of value and performance are of particular importance because they are a better reflection of shareholder wealth (Mackey et al., 2007).

Also, REIT regulations in the U.S. require qualifying firms to pay out at least 90 percent of taxable corporate income as dividends. This pay-out policy results in a close correlation between REIT dividends and cash flows available for distribution. Therefore, the valuation of REITs is closely related to the present value of future corporate cash flows.

Similarly, there may also be a negative impact on corporate image associated with pursuit of dirty projects. In the presence of these possible offsetting market pricing effects, a corporation would be less likely to risk the negative impact of a brown discount, particularly given the persistent nature of corporate image.

Common measures of corporate value rely on the ratio of the market value of the firm’s assets relative to their depreciated book value. Depreciated book value does not update through time and is thus unable to account for changes in cash flow and risk characteristics of sustainable properties. The traditional market-to-book ratio thus conflates cash flow and risk effects of sustainable investment with corporate reputation effects.

We use the S&P 500 index instead of a REIT index. While the discussion of which index is more appropriate is beyond the scope of this project, in the beta regressions, we control for the information captured in the real estate index through the weighted real estate index for the locations in which a REIT operates. Therefore, our beta regressions capture both the broader stock market conditions and the local commercial real estate market returns.

Our results are robust to using three-mile radius, though the identification test results weaken. Our instrument is scaled by a radius surrounding each property. One alternative could be that electric vehicle charging stations can also be normalized by population density. The data do not allow us to scale by population density for each property location. However, in the Online Appendix, we also use population density at the census block level as a control variable in the first stage. The local population density variable does not have any significant impact on Green Share and the relationship between local electric vehicle charging stations and Green Share as presented in Table OA1. The second-stage results are also robust to the inclusion of local population density as an instrument.

The results are robust if our sample starts from 2000 when comprehensive certification data is first available. However, our local economic and demographic controls limit our sample to start in 2006.

These MSAs (Minneapolis/St. Paul, Manhattan, Houston, Atlanta, San Francisco, and Chicago) have the largest local environmentally-certified building shares, based on calculations using CBRE data and from the Eichholtz et al. (2019) dataset.

See http://​www.​usgbc.​org/​leed and http://​www.​energystar.​gov.

The certification data we possess end in 2014. While we believe that the certification effects shall continue, there are some considerations for the period after 2014. Environmental-sustainability premiums can remain significant in the property markets as long as there is excess demand for such properties. However, as the fraction of environmentally-sustainable properties increases in the commercial real estate markets, environmentally-sustainable investments will turn into a norm in the market and the associated premium will shrink. On the other hand, environmentally unsustainable buildings can be “punished.” Holtermans and Kok (2019) and Clayton et al. (2021) show that environmental sustainability premium remains significant in the office markets recently, so we expect that corporate-level benefits remain significant accordingly.

Since betas are estimated using CAPM, we alternatively weight standard errors by the variance of betas from the CAPM regressions. The results, presented in the Online Appendix Table OA4, are robust.

In unreported analysis, we also evaluate residential and industrial portfolios. Green Shares are very low for these property types and the sample size of the analysis is also limited. We do not find any significant results, potentially due to these sample restrictions.

Only the variable of interest results are included to conserve space, yet the model is specified as in the previous analysis and full results are available upon request. This will hold true for all analyses henceforth.