Economic development, environmental justice, and pro-environmental behavior

Abstract

Are a country’s environmental attitudes linked to its level of economic development? In recent decades, rapid industrialization and the use of cheaper but older production technologies have reduced environmental quality in less developed countries (LDCs). Moreover, these countries have been disproportionally affected by global pollution in that they suffer the effects while having emitted less than industrialized countries. To what extent are people in LDCs ready to make sacrifices to improve environmental conditions? International Social Survey Program 2010 data reveal that people in LDCs are less supportive of international agreements forcing their country to take necessary environmental measures than are citizens in the developed world. Moreover, they are more likely to think that wealthier countries should make more effort to protect the environment, and are less willing to make personal economic sacrifices or change their consumption behavior to accommodate environmental concerns. These results hold even after controlling for post-materialist values, political ideology, personal income, and several other demographic variables.

Keywords:

• political development
• environmental opinions
• international agreements

Notes

1. The Tjernström and Tietenberg (2008) study is one exception, but it considers public opinion about international efforts as an independent variable. They investigate whether individuals’ support for their country’s involvement in an international environmental agreement affects their attitudes toward greenhouse effects. Unlike their study, we focus on whether the level of economic development affects citizens’ views of international agreements, environmental justice, and pro-environmental behavior.

2. See Dalton and Rohrschneider (2015) and the project Web site (http://www.issp.org) for more details on the ISSP. Additional information is available from ISSP Research Group (2012).

3. The reliability of the scale is a Cronbach’s α of 0.78. We also constructed another index using a principal component factor analysis with promax oblique rotation. The analysis showed that these four questions represent one underlying factor, with an eigenvalue of 2.41, that explains approximately 60% of the total variance of the four variables.

4. One may wonder to what extent this index reflects environmental concerns of the poor. Some scholars suggest that LDC citizens would not care much about increases in taxes or prices as such measures are designed to decrease environmental hazards such as climate change and ozone depletion. LDC citizens, however, would be more interested in protecting their livelihood rather than the quality of life (Martínez-Alier 1995, Muradian et al. 2003). Nevertheless, we use the above index in our analysis, as the ISSP Survey does not include any other questions to identify LDC citizens’ specific concerns for protection of their livelihood.

5. The principal component factor analysis with promax oblique rotation showed that the responses to these six questions represented one underlying factor, with an eigenvalue of 2.79, which explains approximately 47% of the combined variance of the six variables.

6. These are Austria, Belgium, Canada, Croatia, Czech Republic, Denmark, Finland, Germany, Israel, Japan, New Zealand, Norway, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, South Korea, Taiwan, the United Kingdom, and the US, which are coded as 0. Low-income countries include Argentina, Bulgaria, Chile, Latvia, Mexico, Philippines, Russia, South Africa, and Turkey, which are coded as 1.

7. The ISSP survey included a question on respondents’ income level, but Lübker (2004) posits that using subjective social position has advantages over family income. For example, a graduate student is probably in the bottom income groups in many countries although she/he is socioeconomically advantaged because of her/his high human capital. A self-positioning question will capture his/her future earnings expectations and human capital as well as his/her current positioning. Lübker (2004, p. 119) also claims that even if someone ‘who mistakenly believes him/herself to be better-off than most,’ based on ‘this error of judgment,’ he/she will also miscalculate his/her cost and benefits and hence support environmental policies.

8. These other individual attributes include gender (women = 1, men = 0), age (in years), left-wing political affiliation (left = 1, otherwise = 0), cohabiting with child/children (1, otherwise = 0), and residency (urban = 1, rural = 0).

9. The correlation (eta) between developed/developing nation and aggregate environmental opinions are as follows: international agreements (–0.045), poorer countries make less effort (0.117), willing to pay for environment (–0.147), and willing to change behavior (–0.214).

10. We used multiple imputed survey data in our analysis. A major drawback of survey data occurs when missing values reduce the total number of observations to only a fraction of total available respondents due to list-wise deletion. When all the relevant variables are included in the analyses of the above-described dependent variables, the total number of cases that remained varied between 56% and 86% of all respondents. In order to avoid what King et al. (2001, p. 49) call ‘a loss of valuable information at best and severe selection bias at worst,’ we wanted to avoid list-wise deletion or mean replacement and conducted multiple imputation (MI) for the missing values. MI starts with incomplete data and creates m > 1 complete data sets ‘by replacing the missing values by plausible data values. These plausible values are drawn from a distribution specifically modeled for each missing entry’ (Van Buuren 2012, p. 16). These multiple imputed data sets only differ for the missing observations and the observed values all remain the same. We used Honaker et al. (2007) Amelia II and produced 10 imputed data sets, which were later combined into a single Stata 12 data file for further analysis. As expected, our analyses with the original non-imputed data set with missing observations resulted in significance shifts for only four coefficients. One involved social position (interaction with low-income country group), two living with children, and another with religious service attendance (interaction with low-income country group). Yet, these differences did not lead to complete overhaul of our conclusions, and we stick to our imputed data-set results.

11. The parallel lines assumption of the ordered logit models was tested using Brant tests. The results revealed that none of our models violated the parallel lines assumption (Brant 1990).

Additional information

Funding

This work was supported by the Turkish Academy of Sciences (TUBA) in the framework of the Young Scientist Award Program (GEBIP).