Green technology adoption: An empirical study of the Southern California garment cleaning industry

Abstract

Green technology adoption may be limited by a variety of factors including a lack of sufficient private incentives, long equipment replacement cycles, or lack of information regarding the green alternatives. In California, multiple policy tools including financial incentives, command-and-control regulation, and information and training were used to overcome these obstacles in order to reduce the use of the polluting technology used in traditional dry cleaning. Exploiting the changing regulatory environment, I evaluate the effectiveness of the different policy tools by estimating a dynamic, durable goods replacement model with entry and exit. Because the strict regulations affect future years only, identifying the discount rate is crucial, which I estimate to be 0.94. Using counterfactual simulations under alternative policy regimes, I find that the provision of information and training offered through demonstrations increased adoption of that technology (wet cleaning) by over 200 %. Price incentives (via fees and grants) would have been ineffective at achieving widespread adoption of green technologies but were effective at accelerating adoption when combined with a future ban on polluting technologies. Using this combination of policy instruments led to a net welfare gain of $71 million (NPV) in 2002 when the policies were implemented.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12

Notes

  1. 1.

    Thirteen states are members of the coalition.

  2. 2.

    In 2009, the EPA considered whether to compel dry cleaners to phase out perc nationwide (Washington Post, Wednesday, April 8, 2009, page A03. http://www.washingtonpost.com/wp-dyn/content/article/2009/04/07/AR2009040703748.html)

  3. 3.

    More detail can be found in the CAARB California Dry Cleaning Industry Technical Assessment Report. Operational costs follow the same trends as the machinery costs: Carbon dioxide cleaning has high capital and operational costs but wet cleaning actually has lower costs than traditional perc cleaning. However, this does not account for the extra training required to use the greener technologies or the fact that wet cleaning is more labor intensive than traditional dry cleaning.

  4. 4.

    The SCAQMD is the air pollution control agency for all of Orange County and the urban portions of Los Angeles, Riverside and San Bernardino counties. It has well over half of the cleaners in the state.

  5. 5.

    SCAQMD Rule 1421 was first passed on December 9, 1994, but the final version was updated December 6, 2002. It states that “On or after January 1, 2003, an owner or operator of a new facility may not operate a perchloroethylene dry cleaning system. On or after December 6, 2002, an owner or operator of an existing facility shall be allowed to operate its perchloroethylene dry cleaning system(s) until the end of its useful life and, upon replacement, shall be allowed to operate no more than one perchloroethylene dry cleaning system per facility until December 31, 2020.” The “useful life” of the equipment is something that came under rigorous debate between the SCAQMD and the California Cleaners Association(CCA), the SCAQMD wishing to set the useful life at ten years, the CCA at 20. A 15 year useful life was the compromise.

  6. 6.

    According to the CAARB website, “Demonstration sites ultimately will, in time, create the regional and statewide infrastructure necessary for the long-term diffusion of these technologies.”

  7. 7.

    According to the former president of the California Cleaners Association (CCA), the cleaners have been working with the California Air Quality Management Districts across the state to try to limit the financial costs of the phase-out but the many changes in the regulations were unanticipated by the cleaners, who had limited ability to alter the legislation. The cleaners are being forced to transition away from perc, and the CCA is now concerned that new legislation might force cleaners away from the use of hydrocarbon cleaning as well, which is anecdotally considered by many cleaners to be the most cost-effective technology after perc. According to the CCA, cleaners are already being forced out of business by the new regulations and more restrictions will make the problem considerably worse.

  8. 8.

    A total of 15,204 permit applications by 9,381 cleaners were filed with the SCAQMD between 1956 and 2009, resulting in 13,991 permits. In this analysis I treat cleaners after an ownership change as the same firm. Most of the cleaners that appear in the SCAQMD data have perc equipment but some of the cleaners have equipment that use hydrocarbon or other petroleum solvents.

  9. 9.

    There were 602 SCAQMD grants and 104 CAARB grants by the end of 2009.

  10. 10.

    The UEPI is a community oriented research and advocacy organization based at Occidental College in Los Angeles, CA. Their website allows consumers to search for green cleaners (which they define to be wet or carbon dioxide cleaners) that are within a given radius of a user-provided address. Their database included 166 wet and carbon dioxide cleaners in California.

  11. 11.

    This identification of green cleaners depends on the diary description variable and provides me with information for cleaners who purchased green equipment before the incentives were available. I also use these data to determine whether firms exited the market prior to 1996, assuming that firms that inactivated a permit with no new permit left the market. If a firm is included in the data and has not purchased equipment since before 1983 or if the firm is currently classified as inactive and has no recorded action including exit since 1990, I assume it is inactive; in these cases, I use the last SCAQMD inspection date as the exit date. I remove superfluous actions when a firm inactivates and actives a piece of equipment in the same or adjacent years, presuming it is the same piece of equipment. I also remove a few firms from the data which do not do anything for a span of at least thirty years, and I therefore assume I have missing exit actions. There are 22 of these firms.

  12. 12.

    I keep track of the age of their most current equipment only.

  13. 13.

    All but one of these demonstration sites are cleaners who chose to become demonstration sites and received a CAARB incentive to purchase wet or carbon dioxide cleaning equipment and an additional incentive to become a demonstration site.

  14. 14.

    Other than themselves if they became a demonstration site.

  15. 15.

    I limit the analysis to the years 1999-2009 for a variety of reasons. First, using data in the early 1990s would be problematic since I do not have complete facility action data before 1996. Second, I want to restrict the analysis to use data after firms are aware of the issues surrounding perc cleaning since the period follows the CAARB’s formal classification of perc as a toxic air contaminant in 1991, the 1993 passing of the CAARB ATCM, and the SCAQMD’s passing of the original version of Rule 1421 in 1994. Finally, CAARB Rule 1421 forbids the use of transfer or vented perc machines after 1998, and as a result, I see a relatively large number of firms (388) exit the market in that year. Since I cannot identify which perc machines are of these types, I cannot control for this strict command-and-control regulation because I cannot determine if firms that bought new equipment or exited the market did so because of the mandate. In the analysis, I do still assume a stationary equilibrium in 1999, which is reasonable considering that firms can convert their vented machines to closed loop machines.

  16. 16.

    There are more purchase actions than I would expect when the age of the existing equipment is low. Some of this may be a result of assuming new permits are for new equipment, when in fact the permitted equipment may be used. However, since according to garment cleaning facility survey results in the 2006 CAARB technical assessment report, 89 % of machines are bought new and 96 % of owners said they would buy a new machine in the future, these errors should be kept to a minimum and not have a discernible impact on the results.

  17. 17.

    These data were purchased from R.L. Polk and Company.

  18. 18.

    Since I impose the condition of conditional independence of the unobservables, I implicitly assume that all dependence of the unobservables are accounted for through the observable state vector. Although I acknowledge that there may be serial correlation in the error terms, I believe that most correlation will be accounted for with the inclusion of firm heterogeneity.

  19. 19.

    After 2003, firms with perc equipment older than 15 years old have to discontinue use of the equipment. I do not always observe these firms immediately exit the market or buy new equipment, so I allow for this small minority of firms to remain in the market with profits equal to an estimated intercept. These firms may be outsourcing the cleaning or actually be inactive, waiting to purchase new equipment in a later year, and I do not attempt to distinguish between the two.

  20. 20.

    I use known average equipment costs for K e , as reported in the 2006 CAARB report shown in Table 1. The equipment cost data excluding grants are therefore time invariant. This is not an issue since the size of the grants greatly exceeds the actual changes in equipment costs over the period of study.

  21. 21.

    I assume that potential entrants already have the information and training that is acquired by visiting a demonstration site i.e. w i t = 1 since it is likely they are well informed regarding the use of the new technologies.

  22. 22.

    Alternatively, I could estimate the γ coefficients jointly when estimating the model of equipment choice; however, this is not possible when including firm heterogeneity. Since cleaners only visit a demonstration once, I effectively do not have a panel of observations for the decision to visit and so cannot identify heterogeneity in the decision to attend a demonstration. Because my estimation method for the profit parameters includes heterogeneity in all of the parameters, it is necessary to estimate the homogenous demonstration site visitation parameters in a first stage.

  23. 23.

    The reason for using multiple sets of draws is that there is a chance that a draw in the tail of the sampling distribution explains the actions by all firms with non-negligible probability. If the same set of draws is used for all firms, the optimization procedure might converge to a very tight distribution around that single draw. This is because the use of Bayes’ rule means that the weight on each draw is the ratio of the estimated distribution and sampling distribution as shown in Eq. 21, and so if there is a very small probability of the draw, this value can become very large. I normalize the weights to sum to 1 as done in Hartmann (2006) which also helps to alleviate this issue.

  24. 24.

    I use a normal distribution for the sampling distribution, using rounded estimates from a previous estimation with ρ = 0.9 as the mean of the sampling distribution, and the identity matrix multiplied by a scale factor of four as the covariance matrix to ensure sufficient coverage of the parameter space.

  25. 25.

    The test statistic is twice the difference in the log likelihoods at their maximums, equal to 112, minus penalty terms for the number of parameters, which is the same under both sets of assumptions.

  26. 26.

    Although in theory I can estimate some of the off-diagonal elements of the covariance matrix, data limitations prevent this in practice.

  27. 27.

    It is tough to directly compare the likelihoods in the homogenous and non-homogenous models due to the normalization of the weights and the constraint on the minimum variance.

  28. 28.

    https://www.youtube.com/watchv=i2z_ySLcMHk

  29. 29.

    29 In this model, the demonstration site parameters are estimated concurrently with the utility parameters assuming homogenous firms.

  30. 30.

    To further assess the endogeneity concern, I estimated the base model with heterogeneity using fit values of the distances to demonstration sites variables, regressing the distances on exogenous instruments. For instruments, I used a dummy variable indicating if the zip code is a low-income, cancer risk area where there was extra effort to get cleaners to become demonstration sites, as well as the number of total dry cleaners within a five kilometer radius, since most demonstration sites are cleaners and a high number of nearby cleaners increases the probability of one becoming a demonstration site. Although one could anticipate cleaners being reluctant to share the information with their competitors, I have found that cleaners actually welcome the chance to share their knowledge by speaking at educational meetings, and they are compensated for becoming a demonstration site. The results using the fitted distance variable are similar to those using the actual distances for the other coefficient values.

  31. 31.

    In the absence of the regulations, I set the extra cost of perc in 2007 to be zero, in addition to removing the phase-out and regulation on purchasing perc post-2007.

  32. 32.

    www.epa.gov/chemfact/f_perchl.txt

  33. 33.

    I calculate this using the expectations of policymakers assuming they anticipated the regulatory changes that occurred.

  34. 34.

    The permit data do enable me to determine how many cleaners do in fact use multiple pieces of equipment since I do not observe discontinuation of use.

References

  1. Ackerberg, D.A. (2009). A new use of importance sampling to reduce computational burden in simulation estimation. Quantitative Marketing and Economics, 7, 343–376.

    Article  Google Scholar 

  2. Bajari, P., Hong, H., & Ryan, S. (2010). Identification and estimation of discrete games of complete information. Econometrica, 78(5), 1529–1568.

    Article  Google Scholar 

  3. Baker, L.C. (2001). Managed care and technology adoption in health care: evidence from magnetic resonance imaging. Journal of Health Economics, 20(3), 395–421.

    Article  Google Scholar 

  4. Bass, F.M. (1969). A new product growth model for consumer durables. Management Science, 15, 215–227.

    Article  Google Scholar 

  5. Bollinger, B., & Gillingham, K. (2012). Peer effects in the diffusion of solar photovoltaic panels. Marketing Science, 31(6), 900–912.

    Article  Google Scholar 

  6. Chay, K.Y., & Greenstone, M. (2005). Does air quality matter? Evidence from the housing market. Journal of political economy, 113(2), 376–424.

    Article  Google Scholar 

  7. Doraszelski, U. (2001). The net present value method versus the option value of waiting: A note on Farzin, Huisman and Kort (1998). Journal of Economic Dynamics and Control, 25(25), 1109–1115.

    Article  Google Scholar 

  8. Engers, M., Hartmann, M., & Stern, S. (2009). Are lemons really hot potatoes? International Journal of Industrial Organization, 27(2), 250–263.

    Article  Google Scholar 

  9. Farzin, Y.H., Huisman, K.J.M., & Kort, P.M. (1998). Optimal timing of technology adoption. Journal of Economic Dynamics and Control, 22(5), 779–799.

    Article  Google Scholar 

  10. Fong, M., Chowdhury, H.R., Houghton, M., Komlenic, M., & Villalobos, S. (2006). California dry cleaning industry technical assessment report, Technical report, State of California Air Resources Board.

  11. Goettler, R.L., & Clay, K. (2011). Tariff choice with consumer learning and switching costs. Journal of Marketing Research, 48(4), 633–652.

    Article  Google Scholar 

  12. Gordon, B. (2009). A dynamic model of consumer replacement cycles in the PC processor industry. Marketing Science, 28(5), 846–867.

    Article  Google Scholar 

  13. Gowrisankaran, G., & Stavins, J. (2004). Network externalities and technology adoption: Lessons from electronic payments. The RAND Journal of Economics, 35 (2), 260–276.

    Article  Google Scholar 

  14. Guyton, K.Z., Hogan, K.A., Scott, C.S., Cooper, G.S., Bale, A.S., Kopylev, L., Stanley B.J., Makris, S.L., Glenn, B., Subramaniam, R.P., Gwinn, M.R., Dzubow, R.C., & Chiu1, W.A. (2014). Human health effects of tetrachloroethylene: Key findings and scientific issues. Environmental Health Perspectives, 122(4). 122(4):.

  15. Hannan, T.H., & McDowell, J.M. (1984). Market concentration and the diffusion of new technology in the banking industry. The Review of Economics and Statistics, 66, 686–691.

    Article  Google Scholar 

  16. Hartmann, W.R. (2006). Intertemporal effects of consumption and their implications for demand elasticity estimates. Quantitative Marketing and Economics, 4, 1570–7156.

    Article  Google Scholar 

  17. Hartmann, W.R., & Nair, H. (2010). Retail competition and the dynamics of demand for tied goods. Marketing Science, 29(2), 366–386.

    Article  Google Scholar 

  18. Jensen, R. (1982). Adoption and diffusion of an innovation of uncertain profitability. Journal of Economic Theory, 27(1), 182–193.

    Article  Google Scholar 

  19. Mahajan, V., Muller, E., & Bass, F.M. (1990). New product diffusion models in marketing: A review and directions for research. The Journal of Marketing, 54(1), 1–26.

    Article  Google Scholar 

  20. Mulligan, J.G. (2003). Market segmentation and the diffusion of quality-enhancing innovations: The case of downhill skiing. Review of Economics and Statistics, 85(3), 493–501.

    Article  Google Scholar 

  21. Nair, H. (2007). Intertemporal price discrimination with forward-looking consumers: Application to the US market for console video-games. Quantitative Marketing and Economics, 5(3), 239–292.

    Article  Google Scholar 

  22. Narayanan, S., & Nair, H. (2013). Estimating causal installed-base effects: A bias-correction approach. Journal of Marketing Research, 50(1), 70–94.

    Article  Google Scholar 

  23. Rust, J. (1987). Optimal replacement of gmc bus engines: An empirical model of Harold Zurcher. Econometrica, 55(5), 999–1033.

    Article  Google Scholar 

  24. Shriver, S. (2015). Network effects in alternative fuel adoption: Empirical analysis of the market for ethanol. Marketing Science, 34(1), 78–97.

    Article  Google Scholar 

  25. Singh, H.B., Salas, L., & Stiles, R.E. (1982). Distribution of selected gaseous organic mutagens and suspect carcinogens in ambient air. Environmental Science and Technology, 16, 872–880.

    Article  Google Scholar 

  26. Smith, L. (2012). Dynamics and equilibrium in a structural model of wide-body commercial aircraft markets. Journal of Applied Econometrics, 27(1), 1–33.

    Article  Google Scholar 

  27. Smith, V.K., & Huang, J.-C. (1995). Can markets value air quality? a meta-analysis of hedonic property value models. Journal of Political Economy, 103(1), 209–227.

    Article  Google Scholar 

  28. Tucker, C. (2008). Identifying formal and informal influence in technology adoption with network externalities. Management Science, 55(12), 2024–2039.

    Article  Google Scholar 

Download references

Acknowledgments

I would like to thank the Environmental Protection Agency for their support of this research. I would also like to thank Harikesh Nair, Sridhar Narayanan, and Peter Reiss for their helpful comments as well as two anonymous referees for their insightful feedback. I would especially like to thank Wesley Hartmann for his invaluable support and guidance throughout. All remaining mistakes are my own.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Bryan Bollinger.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Bollinger, B. Green technology adoption: An empirical study of the Southern California garment cleaning industry. Quant Mark Econ 13, 319–358 (2015). https://doi.org/10.1007/s11129-015-9163-0

Download citation

Keywords

  • Dynamic programming
  • Discrete choice
  • Technology adoption
  • Information provision
  • Importance sampling
  • Regulated markets
  • Environmental policy
  • Green technology

JEL Classification

  • D21
  • L16
  • L20
  • L51
  • 033
  • C61