The Opportunity Cost of Pro-Environmental Activities: Spending Time to Promote the Environment

Abstract

This study examines how working hours and the opportunity cost of time affect an individual’s decision to participate in a time-consuming pro-environmental activity—namely, the purchase of refillable products—while using data from the Japanese General Social Survey 2002. By purchasing refillable products, an individual can reduce the amount of waste that needs to be disposed; however, she or he must spend time refilling bottles (i.e., a pro-environmental activity). Our empirical results demonstrate that individuals who work long hours and those with high wage rates tend to purchase nonrefillable products. Therefore, both time constraints and the opportunity cost of time are important factors in the decision to undertake pro-environmental activities.

Appendices

Appendix A. Estimation of Reservation Wage

From Eq. 5, we obtain

$$ \omega_{i} = \gamma_{0} + \varvec{\gamma^{\prime}}_{1} \varvec{\rm Z}_{1,i} + \varepsilon_{1,i} . $$

(5)

Using the latent variable, Eq. 7 can be expressed as

$$ z^{*} = \varvec{\delta^{\prime}Z}_{i} + \varepsilon_{i} . $$

Assuming that the error terms were correlated with

$$ \varepsilon_{1} ,\varepsilon \sim N\left[ {0,0,\sigma_{\varepsilon 1}^{2} ,\sigma_{\varepsilon }^{2} ,\rho } \right], $$

we estimated the following log-likelihood function:

$$ {\text{logL}} = \sum\nolimits_{I = 1} {{ \log }\left[ {\frac{{{ \exp }\left( { - \left( {1/2} \right)\varepsilon_{1,i}^{2} /\sigma_{\varepsilon 1}^{2} } \right)}}{{\sigma \sqrt {2\pi } }}\Upphi \left( {\frac{{\rho \varepsilon_{1,i} /\sigma_{\varepsilon 1} + \varvec{\delta^{\prime}Z}_{\varvec{i}} }}{{\sqrt {1 - \rho^{2} } }}} \right)} \right]} + \sum\nolimits_{I = 0} {{ \log }\Upphi \left( { - \varvec{\delta^{\prime}Z}_{i} } \right)} , $$

(A1)

where I is the index. The first term is applied to the workers and the second to the non-workers.

Appendix B. Calculation of Hourly Wage

Japanese General Social Survey 2002 includes data pertaining to both annual labor income and total number of working hours in the previous week. We multiplied the total number of working hours in the previous week by 40, in order to obtain the annual number of working hours. Thereafter, we approximated the hourly wage by dividing the annual income by the total number of working hours in a year. However, the approximate hourly wage may include a measurement error. Therefore, as a robustness check, we also estimated the annual labor income. We did not consider the number of working hours; instead, we estimated the annual income that respondents or spouses expect to earn in the labor market. The results of the annual income model are based on this approach and are presented in Table 4.