Green Premium in the Tokyo Office Rent Market

Round 1

Reviewer 1 Report

An interesting article with a relatively simple analysis where the effect of a green premium is analyzed on the office market in Tokyo. The results presented are that there is a premium in the order of 6-7%. Greater effect for older medium-sized office properties and lower for newer and larger office properties. The main result is based on a simple OLS model with fixed time and area effects. To take endogeneity into account, the propensity score method has been used.

What I miss in the analysis is:

1) an updated literature study.

Many of the studies are up to 10 years old. Only one study newer than 2016. There should be both newer and more relevant studies to refer to.

2) would like to see a discussion on why we should expect a positive capitalization in rents. It is not completely obvious but depends a lot on what is included in the rents.

3) lacks a discussion of why we can expect an endogeneity. I guess it is not about reverse causality but maybe more omitted variables but it is not stated why. The cause of endogeneity also says something about PSM being a good tool for reducing the problem. Could one imagine a different approach? Instrument variables?

4) is it not a little surprising that  the green ratio is only 5.3%?

5) how representative is the sample, does it come from one company? Selection bias?

6) is RS results really something to present? Only 262 observations, what information does it give us?

7) low degree of explanation power, why?

8) would like to see all parameter estimates and not just the Green label dummy.

9) one could create interaction variables between green labels with the other included variables.

10) is there a risk of spatial dependence? Is it possible to test? 

Author Response

Once again, we thank the editor and the referees for their very valuable comments and suggestions for improvements!

Please find below our responses to the detailed points raised by the referee.

Referee 1, Comment 1: an updated literature study. Many of the studies are up to 10 years old. Only one study newer than 2016. There should be both newer and more relevant studies to refer to.

Response: Thanks for pointing this out. We have surveyed studies on green premiums for rent in recent years and added them to the following literature: Robinson et al. (2016) (Office, US), Ott and Hahn (2018) (Office, Germany), Szumilo and Fuerst (2017) (Office, US), Holtermans and Kok (2019) (Office, US).

We have also added the following to Section 1, lines 32–34: “In recent years, research results on the green premium of rent have been accumulated, and many studies have reported positive premiums [7–10].”

Referee 1, Comment 2: would like to see a discussion on why we should expect a positive capitalization in rents. It is not completely obvious but depends a lot on what is included in the rents.

Response: Thank you for this suggestion. Green real estate is expected to increase sales and lower costs for tenants by conserving energy consumption and making employees healthier. Tenants will be willing to pay more rent for green properties. Higher rents will also lead to higher transaction prices. If prices are expected to rise for developers and investors, then supply and investment in green real estate becomes a rational decision. The existence of a positive rent premium is an important issue in predicting whether the market will function on green values and whether the green real estate stock will increase. (Section 1, lines 24–31)

Referee 1, Comment 3: lacks a discussion of why we can expect an endogeneity. I guess it is not about reverse causality but maybe more omitted variables but it is not stated why. The cause of endogeneity also says something about PSM being a good tool for reducing the problem. Could one imagine a different approach? Instrument variables?

Response: Thank you for providing a valuable perspective for this research. In identifying the economic value of the green label, it is expected that the choice to obtain or not to obtain environmental certification will vary according to the characteristics of the property. This suggests that covariates will differ between the treatment and control groups in the sample, assuming that green labeling is an intervention in the effectiveness test. In particular, commercial real estate is highly heterogeneous, so estimating the premium requires careful control of the characteristics of the property. If no action is taken to estimate the effect, the estimate will include sample selection bias, which is one of the endogenous biases. Sample selection bias can be dealt with by using instrumental variables, difference-in-difference (DID), and propensity score. In DID, to control for the systematic difference between the certified and non-certified buildings, two observations are used for each building: one before certification and one after. DID allow controlling for unobserved effects, thereby mitigating a potential omitted variable bias present in many cross-sectional studies (Reichardt et al., 2012). Propensity scores form a sample with similar covariates, using the probability that an intervention will take place. To control more precisely for the variations in the measured and unmeasured characteristics of rated buildings and the nearby control buildings, we estimate propensity scores for all buildings in the rental sample and the sample of transacted buildings (Eichholtz et al., 2013). PSM is also expected to be comparable with other countries. (Section 1, lines 50–67)

Referee 1, Comment 4: is it not a little surprising that  the green ratio is only 5.3%?

Response: The percentage of green building is 5.3% less in the whole sample, but this can be attributed to two reasons. First, the huge market of Tokyo includes many small- and medium-size buildings. Owners of small- and medium-size buildings are not financially strong, and it is difficult for them to make additional investment in acquiring green label. Second, the percentage of green buildings is changing with time. It was only 0.064% in 2009 when the certification program was fully launched, but it has increased to 9.7% in 2019. (Section 2.1, lines 132–138)

Referee 1, Comment 5: how representative is the sample, does it come from one company? Selection bias?

Response:  The sample is based on data provided by Xymax, a leading property management company in Japan. Xymax regularly obtains information on contracted cases not only from itself but also from several major leasing agents. In the Japanese office leasing market, it is rare for tenants and owners to sign contracts directly without going through an agent, so we thought there would be no major bias in understanding the movements of the Tokyo office market. (Section 2.1, lines 81–85)

Referee 1, Comment 6: is RS results really something to present? Only 262 observations, what information does it give us?

Response: RS controls for covariates (other than age), which is an effective way to deal with endogeneity, but it is a very small sample and makes it difficult to discuss the overall market effect. We need a way to control for covariates while maintaining some sample size. PS Matching creates paired data with similar covariates in the regression space. Multiple methods are used to check the robustness of the analysis. (Section 2.3, lines 209–214)

Referee 1, Comment 7: low degree of explanation power, why?

Response: Both models have an adjusted R-squared of about 0.6. There are some unobserved variables such as tenant and owner attributes that could not be captured due to data limitations. In our estimation model, we consider these missing variables to be homogeneous. (Section 3.1, lines 280–283)

Referee 1, Comment 8: would like to see all parameter estimates and not just the Green label dummy.

Response: Variables other than the green label were added to Table 3 and Table 4.

In general, rents tend to be higher for smaller (newer) buildings, larger buildings, and buildings with more facilities. The Location Dummy and Time Dummy have been omitted because the tables are too large.

Table 4. Results of hedonic regression (Base model, Repeat sales, and PS Matching)

Table 5. Results of hedonic regression (propensity score clustering)

Referee 1, Comment 9: one could create interaction variables between green labels with the other included variables.

Response: Additional analysis was conducted, and the following was added to Section 3.2, lines 400–412. Even though the results are stratified by propensity score, there may be a certain influence between the age of the building and the environmental certification dummy. To adjust for this effect, we estimate equation (2) by adding the cross term of building age and green label as a robustness check. As a result, for Medium-Size Old buildings, the coefficient of the green label dummy is -0.010 (0.033), which is insignificant and the coefficient of the cross term of Age and green label dummy is 0.003 (0.001), which is positive and significant. For Large-New buildings, the coefficient of green label is 0.023 (0.009), which is significant, and the coefficient of cross term of Age and green label is 0.0002 (0.0006), which is not significant.  The results suggest that in medium-size and old buildings, the difference in depreciation over time increases with the age of the building due to the difference in maintenance, suggesting that the green label functions more as a differentiating factor. In the case of Large-New buildings, the green label does not function as a differentiating factor, so the same result was not obtained.

Referee 1, Comment 10: is there a risk of spatial dependence? Is it possible to test?

Response: There is room to consider the risk of spatial dependency. There is no significant difference in the mean of the Five wards dummy in Table 2 (Green Building: 0.831, Non-Green Building: 0.778), so we believe that the risk of spatial dependency is small. This is because our analysis focuses on a single city, Tokyo, so there is no geographical difference in terms of environmental regulations and certification items. For example, the risk of earthquakes is the same in the 23 wards of Tokyo. However, there may be differences in more detailed geographical conditions (susceptibility to flooding, strength of the ground). We have not included these in our analysis due to data limitations. (Section 4, lines 484–491)

Author Response File: Author Response.doc

Reviewer 2 Report

This study provides valuable empirical insights into the pricing mechanisms of eco-labels for commercial real estate, using  data from the Tokyo office market. The manuscript is a pleasure to read, it is very well organised and written in a concise, clear and transparent style. The authors are aware of the major pitfalls of this type of analysis, notably endogeneity, and mitigate them with a PSM analysis and other robustness checks.

My recommendation would be to accept this manuscript for publication, perhaps after another round of proofreading and a re-consideration of the title (perhaps better to include 'rent' or 'rental' in the title and drop the 'building' from 'office building market'.

I would also have been curious to see more details on the distribution of green premiums, e.g. by rent quantile or in various districts of Tokyo, for small vs. large or recent vs. old buildings. This would help illustrate the results a bit more but I do realise that it may be beyond the scope of this paper so not strictly necessary. The authors do provide a comparison betwee medium sized old and large new buildings but it is not entirely clear which factor (size or age) contributes to the differential outcomes.

Author Response

Once again, we thank the editor and the referees for their very valuable comments and suggestions for improvements!

Please find below our responses to the detailed points raised by the referee 2.

Referee 2, Comment 1: My recommendation would be to accept this manuscript for publication, perhaps after another round of proofreading and a re-consideration of the title (perhaps better to include 'rent' or 'rental' in the title and drop the 'building' from 'office building market'.

Response: Thank you for pointing this out. The title will be changed to “Green Premium in the Tokyo Office Rent Market.”

Referee 2, Comment 2: I would also have been curious to see more details on the distribution of green premiums, e.g. by rent quantile or in various districts of Tokyo, for small vs. large or recent vs. old buildings. This would help illustrate the results a bit more but I do realize that it may be beyond the scope of this paper so not strictly necessary. The authors do provide a comparison between medium sized old and large new buildings but it is not entirely clear which factor (size or age) contributes to the differential outcomes.

Response: We are also interested in the distribution of the green premium. We believe that the size of the premium varies depending on the detailed location, rent level (building grade), tenant, developer, and owner attributes. However, due to the limitations of the data, we have not included it in this study. We would like to make this a future issue.

Author Response File: Author Response.doc

Reviewer 3 Report

see attached

Comments for author File: Comments.pdf

Author Response

Once again, we thank the editor and the referees for their very valuable comments and suggestions for improvements!

Please find below our responses to the detailed points raised by the referee 3.

Referee 3, Comment 1: More elaborations on the motivation and the research question should be provided in the Introduction Section, for example, why rental premium is required to be studied when price premium has been confirmed, why propensity score clustering is required, etc.

Response: Green real estate is expected to increase sales and lower costs for tenants by conserving energy consumption and making employees healthier. Tenants will be willing to pay more rent for green properties. Higher rents will also lead to higher transaction prices. If prices are expected to rise for developers and investors, then supply and investment in green real estate becomes a rational decision. The existence of a positive rent premium is an important issue in predicting whether the market will function on green values and whether the green real estate stock will increase. (Section 1, Lines 24–31)

Referee 3, Comment 2: The whole paper requires a detailed proofread, there are quite a lot of typos and grammatical issues. For example, theamount in line 58, assue in line 60, systemsis in line 68, … many of these errors can easily be identified with spelling checks.

Response: Thank you for pointing this out. We availed of editing and proofreading services of a professional editing company to correct the typographical and grammatical errors.

Referee 3, Comment 3: In office rental contracts (leases), there are common terms such as rent-free period, different tenure periods, management fee included, and renew options, etc., these terms can impose grave differences on the rents. It is suggested to analyse them or at least discuss the limitations of the dataset in controlling lease terms..

Response: We will now explain the scope of rent included in our analysis. First, rents include management fees. This is because tenants negotiate with agents and landlords for rents that include management fees. Second, the term of the contract is generally two years according to the business practice of office leases in Japan. However, there is an automatic renewal clause when the contract expires, and tenants can terminate the contract even within the contract period by giving a certain period of notice, so we believe that the length of the contract period has little impact on rents. Rent-free period is not taken into account due to data limitations, so it is possible that rents are higher in cases where a long rent-free period is set. Electricity and water are rarely included in the rent but are charged by the landlord according to the amount actually consumed. (Section 2.1, lines 86–95)

Referee 3, Comment 4: In explaining office rents, office grade is a key determinant. It is crucial to control this variable of office grade to ensure the premium has resulted from the green label.

Response: We also believe that office grade is a determinant of rent, as you have pointed out. Since there is no uniform standard in the market for office grade, I have not used it as a variable. The probit regression model (Equation 3) to estimate the propensity score evaluates good quality buildings. By dividing the sample into subsamples through propensity score clustering, we believe that we are able to control for grade.

Referee 3, Comment 5: It requires references for each of the green label systems in line 66: CASBEE, CASBEE for Real Estate, and DBJ Green Building Certification.

Response: Please refer to Appendix A for the history of each of the green label systems and the characteristics of the evaluation items. (Section 2.1, lines 112–113):

In 2001, the Ministry of Land, Infrastructure, Transport and Tourism (MLIT) launched a green building certification initiative called CASBEE. Like BREEAM and LEED, CASBEE uses a multifaceted approach to evaluate various aspects of sustainability, including (1) indoor environment, (2) quality of services, (3) outdoor environment of the site, (4) energy, (5) resources, and materials, and (6) off-site environment. CASBEE provides a comprehensive index called the Building Environment Efficiency (BEE), which is the ratio of the "environmental quality" (Q) value to the "environmental impact" (L) value. As of August 2021, 482 buildings have been certified. In 2012, a variation of CASBEE called CASBEE Real Estate was launched. It is a simplified version of CASBEE, designed to reduce the time and cost of the assessment. In addition, the MLIT has established the Building-Housing Energy-efficiency Labeling System (BELS) to certify the energy consumption performance (energy-saving performance) of buildings. Developers, sellers, and lessors are required to display energy-saving performance, and the evaluation and display of energy-saving performance, etc., is done with a five-point star mark.

The private sector has also established its own green building certification programs. Primarily promoted by financial institutions, the Development Bank of Japan and Sumitomo Mitsui Banking Corporation have launched their certification programs. For example, the DBJ Green Building Certificate, established in 2011, evaluates a wide range of sustainability indicators. It assesses (1) ecology, (2) amenities, (3) community, (4) risk management, and (5) partnership among stakeholders. The evaluation process is simple, and as of August 2021, 943 buildings (of which 413 are office buildings) have been certified.

The green labeling systems (CASBEE, CASBEE Real Estate, DBJ Green Building Certification) used for analysis in this study do not necessarily match the items they evaluate or in their evaluation methods compared to LEED and BREEAM. However, they all form a comprehensive environmental certification system, having many commonalities in a wide range of categories. The three green labeling systems used in our survey have not yet spread to the entire office stock in Tokyo, so the number of observations for each is small. This is difficult to see due to data limitations.

Referee 3, Comment 6: Summary statistics for the RS and the PSM models should be provided.

Response: Added Table 3 to Section 2.3.

Table 3. Summary statistics of the dataset (Repeat Sales and Propensity Score Matching)

Referee 3, Comment 7: It requires a robustness test to ascertain that the scores in the three green label systems can be interchangeably used as stated in footnote 2.

Response: To confirm that the three green label systems we used are interchangeable, we estimate the green premium for each green label system. The regression results are +0.061 (0.007) for DBJ Green Building Certification, 0.064 (0.018) for CASBEE, and 0.052 (0.097) for CASBEE for Real Estate (standard errors in parentheses). CASBEE is not significant due to the small sample size, but both are at a similar level of 5–7 %. Based on this result and the similarity of the evaluation items (see Appendix A), we believe that the three green labeling systems are interchangeable. (Section 3.1, lines 284–290)

Referee 3, Comment 8: Instead of using a dummy variable to represent green label, it is suggested to test on the green label scores as the market conceives a platinum and a gold certificates (in LEED) as two very different performance standards.

Response: In our study, we do not show the difference in the magnitude of the premium between high and low environmental performances. Since the green label is not yet widespread in the Japanese office leasing market, we focus on estimating the average effect of the green label. Each green label system shares the same overall environmental performance rating, but the number of grades and their criteria are different. Additional data development and analysis is needed to analyze the premium for high environmental performance, but due to data limitations, it is not included in our study. This will be the subject of future research. (Section 4, lines 506–512)

Referee 3, Comment 9: Is the dependent variable in rent or rent per sm? Table 2 shows rent per sm, but equation (2) shows contract rent. As there is a gross floor area variable as one of the independent variables, the dependent variable is better to use ‘rent’ instead of ‘rent per sm’.

Response: Thank you for pointing this out. We have corrected the wording in equation (2). We use the rent per square meter as the dependent variable. This is because the rent per unit area has a linear relationship with the attributes of the property. For example, the larger the size of the building, the higher the rent, and the older the building, the lower the rent. In addition, many studies, including the pioneering work of Eichiholtz et al. (2010) and Fuerst and McAllister (2011) have used rents per unit area. We also created explanatory variables similar to these so that we could compare them.

Referee 3, Comment 10: In office leases, lettable floor area is more commonly used instead of gross floor area, as it can exclude the lift lobby and other common areas.

Response: lettable floor area could not be used due to data availability limitations. Rent levels may differ depending on the size of the individual contracted parcels within the same building. In our study, we used the standard story area as an alternative variable, understanding that it is not perfect. We also use the gross building area as a variable that reflects the size of the building. The higher the grade of the building, the more extensive the common areas tend to be. From an investor's point of view, the leased floor area is important, but we thought that the total floor area was appropriate in evaluating the quality of the building.

Referee 3, Comment 11: It is an interesting idea to use repeat-sales method, but the BMN (1963) RS model cannot include building characteristic variables and location dummy, the “Yes” indicators in column (2) of Table 3 require detailed explanations.

Response: In our study, we use repeat sales to completely control for the effects of covariates such as location, duration of contract, equipment, etc. After creating a repeat sales sample, we use a hedonic approach with building characteristics as explanatory variables to estimate the effect of the green label. The pure effect of the green label can be expected to be extracted.

Referee 3, Comment 12: Since the PSM is the core contribution of this paper, the PSM model requires elaborations and references on its validity in dealing with endogeneity bias. For example, are there clusters’ dummies included in the model? Are the building characteristic variables divided into the corresponding clusters? A model equation for the 2nd step of the PSM can help, i.e. line 140: “for each certified sample, we match the sample with 140 the nearest neighbor propensity score among the uncertified samples.”

Response: In identifying the economic value of the green label, it is expected that the choice to obtain or not to obtain environmental certification will vary according to the characteristics of the property. This suggests that covariates will differ between the treatment and control groups in the sample, assuming that green labeling is an intervention in the effectiveness test. In particular, commercial real estate is highly heterogeneous, so estimating the premium requires careful control of the characteristics of the property. If no action is taken to estimate the effect, the estimate will include sample selection bias, which is one of the endogenous biases. Sample selection bias can be dealt with by using instrumental variables, difference-in-difference (DID), and propensity score. In DID, to control for the systematic difference between the certified and non-certified buildings, two observations are used for each building: one before certification and one after. DID allows controlling for unobserved effects, thereby mitigating a potential omitted variable bias present in many cross-sectional studies (Reichardt et al., 2012).  Propensity scores form a sample with similar covariates, using the probability that an intervention will take place. To control more precisely for the variations in the measured and unmeasured characteristics of rated buildings and the nearby control buildings, we estimate propensity scores for all buildings in the rental sample and the sample of transacted buildings (Eichholtz et al., 2013).  PSM is also expected to be comparable with other countries. (Section 1, lines 50–67)

Referee 3, Comment 13: I think the paper does not address clearly on the differences between the RS model and PSM model. Since there can be many other building or neighbourhood quality variables other than size and age that are endogenous with the green label effect, the PSM model cannot  control. While the RS model is an approach that can also control the latent (unobserved) quality variables that are endogenous with the green label effect, as the same office unit is considered.

Response: RS controls for covariates (other than age), which is an effective way to deal with endogeneity, but it is a very small sample and makes it difficult to discuss the overall market effect. We need a way to control for covariates while maintaining some sample size. PS Matching creates paired data with similar covariates in the regression space. Multiple methods are used to check the robustness of the analysis. (Section 2.3, lines 209–214)

Referee 3, Comment 14: In lines 177-184, acquiring a green label is considered an expensive branding exercise, but it can help reduce utility bills and maintenance fee of the building in the long run. If data on the cost and benefit of acquiring a green label can be discussed, the interpretation of the results can have a more fruitful implication. i.e. whether the tenants willing to pay higher rents for green label offices are paying for energy savings, etc. or for corporate images?

Response: Thank you for your comments. We think your point is important, please refer to Section 3-3, lines 414–458, which summarizes the source of green premium:

Eichholtz et al. (2010) summarized the distinct ways in which investment in green building can lead to economic benefits such as (1) energy savings and waste reduction, (2) higher employee productivity, (3) a signal of social awareness and superior sense of social responsibility of the occupants, and (4) longer economic lives.

For (1) energy-saving, the premium can be estimated by evaluating the efficiency of energy and water consumption. In Japanese rent contracts, tenants primarily bear utility costs. Savings in utility costs will lead to profits for tenants. For (2) employee productivity, the premium can be estimated by evaluating the indoor environment, such as air and lighting, which plays a role in employee health and ease of work. This discussion could also be extended to an analysis of the quality of life in, and benefits of a green office building. Employee health brings benefits to companies through the prevention of employee turnover and the improvement of work efficiency. The green labels in our study include evaluation items related to indoor environment and service quality, suggesting that green offices may be evaluated from the aspect of employee productivity and health. However, it does not distinguish between energy savings and resilience, and so the direct benefits to employees are out of our scope. Certification systems that focus more on employee health, such as the WELL Building Standard, have emerged. In Japan, the CASBEE Wellness Office was launched in 2019. Through the development of these data, it may be possible to identify the benefits of employees at a rent premium in detail. The green label used in our study is a comprehensive index that includes not only energy but also equipment, operation, water, material, interior, site and surroundings, transport, waste, and containment (Appendix A). Therefore, the premium we estimate is likely to include economic benefits attributable to (1) energy-saving and (2) employee productivity. Yoshida et al. (2017) confirmed that the green premium disappears when energy and water consumption are added as explanatory variables. Their study concluded that a rent premium observed for green buildings is paid by tenants not for a brand associated with green building labels but for material benefits of green buildings with respect to lower costs of energy and water. In our study, it was difficult to collect energy and water data, especially for non-certified buildings, and as a result we could not identify only energy and water savings in our estimates.

Moreover, the green label we used favors a long-life structure. As a result, certified buildings have a smaller depreciation value and a larger premium. Our study segmented the market appropriately through PS clustering and observed a more significant premium for older buildings. This result shows the economic benefit of (4) longer economic lives. The result of more significant premiums for older buildings is consistent with Yoshida and Sugiura (2015) for Tokyo condominiums.

In contrast, the (3) signal of a sense of social responsibility of the occupants was not included in the evaluation items of the green label we used. As a result, we could not explicitly verify it. However, green buildings may attract companies with higher social responsibility and are included in our estimate of the premium. To clarify the economic benefits in this way, we need to add variables that reflect corporate social responsibility, such as credit score, to the model.

Referee 3, Comment 15: There are some literatures that have found green label effect on office rents, for example, Reichardt, Fuerst, Rottke, Zietz (2012) had made use of DID to deal with the endogenous issue, and their model can estimate temporal changes of the green label effect, which has been mentioned as a limitation in this manuscript. Their method should be reviewed and discussed to highlight the advantages of PSM versus DID.

Response: Thanks for the suggestion. I have included my response to this comment in my response to your comment 12. Please refer to it. Also, in addition, I have added the following in Section 2.3, lines 201–203.

RS is an approach similar to the DID mentioned in Section 1. Due to data limitations, we did not focus on temporal changes in premium size in our study.

Referee 3, Comment 16: Also, Eichholtz, Kok & Quigley (2010) “The Economics of Green Building” has also made use of a panel model to find economic premiums of green labels in rent and asset values. More importantly, they also confirm that the attributes rated for both thermal efficiency and sustainability contribute to premiums in rents and asset values. In other words, they identify the causes of the premium empirically, besides just identifying the premium.

Response: Thank you for your important comment. In our study, we have organized the four ways of economic value of green premium in Section 3.3 (response to your comment 14). However, due to data constraints, it is not enough to estimate premiums for each of the four ways on the economic value of green premiums and to make a deep discussion. This will be a subject for future research.

Author Response File: Author Response.doc

Round 2

Reviewer 1 Report

Well done! 

I am very pleased with the changes that have been made to the article. Good work!

Author Response

Thank you very much for your comment. Your comments were interesting and informative, and made our paper much easier to understand.

Furthermore, thank you for pointing out the quality of the English. We have checked the tenses, spelling, and grammar again. For English editing, we used Editage, a translation service provided by CACTUS (https://cactusglobal.com/brands/editage/  ). Editing Certificate is attached.

Author Response File: Author Response.doc

Reviewer 3 Report

See attached

Comments for author File: Comments.pdf

Author Response

Response by Onishi, Deng and Shimizu to the Referee 3

Manuscript ID: sustainability-1381861

Title: Green Premium in the Tokyo Office Rent Market

Sustainability

Once again, we thank the editor and the referees for their very valuable comments and suggestions for improvements!

Please find below our responses to the detailed points raised by the referee 3. (Please see the attachment.) 

Comment: the motivation of using the propensity score clustering is important but it is not well explained in the introduction section.

Response: To match the samples with and without green labels, PSM discards many of the samples that do not have green labels. It is assumed that many of the samples that will be discarded are samples of older, small to mid-sized buildings that have little investment capacity and are difficult to invest in to obtain a green label. The submarkets that these samples comprise of account for a large portion of Tokyo's office stock, and have a different market structure than the prime buildings, making the spread of green labels a challenge. Thus, PSM may not be able to investigate the effect of green labeling in submarkets that consist of older, small and medium-sized buildings with few green buildings. To address this issue, we use propensity score-based clustering to divide the sample into prime and affordable buildings, and test the effectiveness of green labels in each category. (Added in Section 1, Line 65-75)

Comment: The current pdf still shows a lot of typos, for example in page 7, there seem to have many changes without crossing out the previous ones (as underlined below) probably because the revised pdf file is not in track change mode, it does not cross out the deleted parts:

“+0.066 (0.005), indicatinged that it is positive and significant (the value in parentheses represents the standard error). This indicatesed that contact rents in certified buildings are 6.6% higher than non not-certified buildings. Column 2 of Table 4 shows the results for the repeat sales sample. The estimated result iswas 6.3%. The standard error iswas 0.034, larger than that of the base model, and iswas not significant at the 5% level. When we controlled for the building dummy instead of the characteristics and location dummy, the estimate for green premium iswas +0.0148 and its standard error was +0.0337, which iswas not significant. In addition, the sample size is was small (262) because of the repeat-sales condition.”

Response:  There was a bug in the changelog function, and I'm sorry it was hard to read. It has been check and fixed again. For English editing, we used Editage, a translation service provided by CACTUS (https://cactusglobal.com/brands/editage/ ). 

Comment: There seems to have a standard of office grade in Japan as reported by big international real estate consultants such as CBRE or Cushman and Wakefield, see https://www.retalkasia.com/news/2019/02/05/tokyo-office-vacancy-rate-falls-below-1-across-all-grades-rents-rise/1549334417, https://www.cushmanwakefield.com/en/japan/insights/japan- marketbeat/office-marketbeat-report. It is important to align the empirical models with the practices in Japan to avoid misspecifying the determinants. A robustness test on the association between the Propensity Score Clustering and the Office Grading can be included.

Response: The sample split by propensity score clustering has similar results to the grading used by market participants. In Tokyo, office buildings are graded according to specifications such as size and newness; the CBRE [28] defines the Grade A category as one which has a total area of at least 33,000 square meters and an age of less than 15 years. The Grade A minus category is defined as one with a total area of at least 23,000 square meters, and which also complies with the new earthquake resistance standards (built after 1981). The Grade B category is defined as one with a total area of 6600 square meters or more, and which also complies with the new earthquake resistance standards. Comparing the CBRE grades with our propensity score clustering results, large-sized new buildings are classified as 86% of the grade A sample, 64% of the grade A minus sample, 26% of the grade B sample, and 6% of the no grade sample. Medium-sized old buildings are classified as 9% of the Grade A sample, 18% of the Grade A minus sample, 33% of the Grade B sample, and 16% of the No Grade sample. The higher the grade, the higher the propensity score clustering tends to be. (Added in Section 3.2, Line 344-357)

28. CBRE. https://www.cbre.co.jp/th-th/research-reports/office-marketview (accessed on 13 October 2021)

Comment: I mean including the references of the international standards, not just the descriptions of the history. Referencing of the sources of the information is IMPORTANT. For example, if your referred CASBEE is the same as the CASBEE (Comprehensive Assessment Systems for Built Environment Efficiency) at https://www.ibec.or.jp/CASBEE/english/beeE.htm, then the assessment criteria mentioned at the website seem to be different from the authors’ descriptions in the Appendix A. The webpage mentions 4 assessment fields instead of six as follows:“CASBEE covers the following four assessment fields: (1) Energy efficiency (2) Resource efficiency (3) Local environment (4) Indoor environment.”

Response: We add the sources to Appendix A and the references. In addition, CASBEE covers the four fields you pointed out, but when assessing, it uses 80 assessment items reorganized into six categories (three for Q environmental quality and three for L environmental impact). IBEC, which operates CASBEE, explains that these four fields are largely the same as the target fields for the existing assessment tools described above in Japan and abroad, but they do not necessarily represent the same concepts, so it is difficult to deal with them on the same basis. Therefore, the assessment categories contained within these four fields had to be examined and reorganized. As a result, the assessment categories were classified as shown in Figure 5 into BEE numerator Q (Built environment quality) and BEE denominator L (Built environment load). Q is further divided into three items for assessment: Q1 Indoor environment, Q2 Quality of service and Q3 Outdoor environment. Similarly, L is divided into L1 Energy, L2 Resources and Materials and L3 Off-site Environment [37].

37. Institute for Building Environment and Energy Conservation（IBEC）. https://www.ibec.or.jp/CASBEE/english/beeE.htm (accessed on 13 October 2021)

Comment: from Table 3 (PSM), it seems to show a positive collinearity between gross building area and green label. It is important to carry out coefficient diagnostics such as Variance Inflation Factor (VIF) in the regression models to test the multicollinearity between gross building area effect and the green label effect.

Response: The VIF of the green label dummy is 1.210 (in Full Sample), which is small and did not exceed 10. There is no strong collinearity between the green label and the covariates including total floor space.

Comments: In Eichiholtz et al. (2010), they included Building Size variable only, but in this manuscript, it includes both Gross Building Area and Standard Story Area and the signs of their log effects on log unit rent are opposite and significant. More importantly, this manuscript also include No. of Stories as a variable, but mathematically Standard Story Area times No. of Stories equals Gross Building Area. In other words, the three variables are highly multicollinear, the reason why they are estimable is probably due to the logarithm form in areas but linear form in No. of Stories. It is suggested to remove the log in the two area variables to make it a log-linear hedonic model (as stated in Equation 2) and check whether the three are exact collinear.

Response: We are aware of the controversy regarding the functional form of the hedonic function. In the full sample, for the log-linear hedonic model with the gross building area (GBA) and standard floor area (SFA) as the true numbers, the VIF is 4.904 for the stories, 3.262 for the SFA, and 7.639 for the GBA, which does not show strong collinearity. On the other hand, in the model we used (with logarithmic GBA and SFA), the VIFs are 4.961 for the stories, 13.445 for the SFA, and 23.717 for the GBA. The VIFs of the GBA and SFA are above 10, indicating collinearity. Here, we are not interested in the coefficients of the covariates, but in the coefficients and standard errors of the green labels. For the full sample, the coefficient of the green label is 0.078 (0.006) when the GBA and SFA are taken as true numbers, and 0.066 (0.005) when they are taken as logarithms. No difference is found in the standard errors, and whether the SFA and GBA are true or logarithmic does not seem to have a significant effect. On the other hand, the distributions of both GBA and SFA are extremely skewed toward small values. Based on these considerations, we adopt a model in which GBA and SFA are logarithmic.

Comments: If you are using BMN (1963) Repeat Sales Model, it does not include Location Dummies, please replace the ‘YES’ indicators by ‘NO’ on Location dummy. Else, it is better to provide the Equation of the Repeat Sales model and cite clearly the reference whether it is the BMN (1963) one.

Response: Our model differs from BMN in that we also include time and area dummies. We performed repeat sales sampling and then estimated the model using equation (2), which we added to Section 3.1, Line 256-258. Column 2 of Table 4 shows the results of the OLS estimation in Equation (2) after repeat sales sampling, using the same variables as in the base model.

Comments: A model equation for the 2nd step of the PSM is required, i.e. “for each certified sample, we match the sample with 140 the nearest neighbor propensity score among the uncertified samples.”

Response: We used equation (2) to estimate after sampling with PSM. We have added the following to Section 3.1, Line 263-266. Column 3 of Table 4 shows the results of the OLS estimation using Equation (2) with the same variables as in the base model, after we match the sample with 140, the nearest neighbor propensity score among the uncertified samples, for each certified sample by PSM.

Author Response File: Author Response.doc

Round 3

Reviewer 3 Report

Only two comments remain, but they require major changes. For more details on which responses, please refer to the attached.

1. The high VIFs indicate that the three variables are highly collinear due to the fact that GBA = SFA x No. of Storey. Since the green label effect can be dependent of the GBA or SFA effects, a correct specification of them is essential in estimating the green label effect. It is required to delete either one of the three variables and convert into either a semi-log or a log-log model rather than a hybrid without good reasons.
2. Please clarify the following definition of the sampling method in line 189, does it mean the same contracted tenant before and after the green label?

Line 189: “there were contracted cases both before and after the acquisition of the green label in the same building”

Also, if that is the case, it is suggested to rename it to “same-tenant sampling” or “repeat-contract sampling” to avoid confusion.

Comments for author File: Comments.pdf

Author Response

Once again, we thank the editor and the referees for their very valuable comments and suggestions for improvements! Please find below our responses to the detailed points raised by the referee 3. Please see also the attachment.

Further comments: The high VIFs indicate that the three variables are highly collinear due to the fact that GBA = SFA x No. of Storey. Since the green label effect can be dependent of the GBA or SFA effects, a correct specification of them is essential in estimating the green label effect. It is required to delete either one of the three variables and convert into either a semi-log or a log-log model rather than a hybrid without good reasons.

Response: We agree with your point, and since the collinearity of GBA (logarithmic), SFA (logarithmic), and Stories is confirmed, I changed the model for Equation (2) and Equation (3) to use only GBA (logarithmic) among the three variables. In Equation (2) (full sample), the VIF of GBA is 1.270, and no collinearity is observed. We have revised the description of the base model in Section 2.1, the analysis results in the text, and Tables 3, 4, and 5. Thank you for your advice.

Further comments: 1. Please clarify the following definition of the sampling method in line 189, does it mean the same contracted tenant before and after the green label? Line 189: “there were contracted cases both before and after the acquisition of the green label in the same building”

2. If that is the case, it is suggested to rename it to “same-tenant sampling” or “repeat-contract sampling” to avoid confusion.

Response: Thank you for pointing this out. Our analysis is based on new contracts, so it is not necessarily the same tenant before and after earning the green label. We don't have tenant IDs, so we can't confirm this, but it is almost certainly a different tenant. In order to clarify the definition, we will revise it as follows in line 189;

In this approach, we extract paired cases where new contracts occurred in the same building both before and after the acquisition of the green label (It is considered that tenants before and after acquisition of the green label are not nearly the same).

Author Response File: Author Response.doc

Round 4

Reviewer 3 Report

NA