Factors affecting risk mitigation revisited: the case of earthquake in Turkey

Abstract

This paper investigates the effects of earthquake risk perceptions and a number of socioeconomic variables on risk mitigation. The effects of perceived risk components (probability and severity) and perceived risk characteristics (dread, knowledge, controllability, and responsibility) are investigated jointly in a single model, together with degree of risk aversion and socioeconomic factors (income, gender, age, education, and number of children). Analyses are conducted separately in order to explain variability in five different forms of mitigation behaviors: (1) purchasing of compulsory earthquake insurance, (2) past mitigation behaviors, (3) monetary value of past mitigation behaviors, (4) intentions for future mitigation behaviors, and (5) monetary value of intended future mitigation behaviors. Data are obtained from homeowners in Istanbul, where a major earthquake is expected within 30 years. The results provide evidence that all three variable sets (risk components, risk characteristics, and socioeconomic variables) may act as significant determinants of mitigation and that the driving factors for mitigation may differ across forms of mitigation behaviors investigated. The study (1) provides partial explanation for the equivocal findings in prior research regarding the relationship between risk perceptions and mitigation behaviors, and (2) offers guidelines for policy‐makers in motivating communal earthquake risk mitigation.

Keywords:

• risk
• risk mitigation
• risk perception
• risk management
• earthquake risk
• hazard adjustment

Acknowledgments

This paper reports on research supported by grants from the Scientific and Technological Research Council of Turkey through Social Sciences and Humanities Research Group (SOBAG), Project number: 104K109.

Notes

1. Cruz, Steinberg, and Vetere‐Arellano (2006) explained further costs of the 1999 earthquake and Ewing, Kruse, and Ozdemir (2004) investigated its effects on some macroeconomic variables.

2. Five of the hazards are classified as ‘well‐defined’ hazards since death rates are known and the risks are relatively common; others are classified as ‘less‐defined’.

3. Istanbul is divided into three seismic zones by the Council of Ministers: Zone 1 being the lowest risk area and Zone 3 being the highest risk area. Factors such as building type and quality, ground characteristics, and distance to the estimated fault lines were used in determining different risk areas. Most citizens of Istanbul do have a general idea about the attributed territorial risk situation of their homes and work places. For details, see Istanbul Metropolitian Municipality’s website: www.ibb.gov.tr.

4. The English and Turkish versions of the questionnaire are available upon request.

5. The proportions of people who reported to (1) have adopted these ‘risk precautions’ vary from 0.4% for building a fire ladder to 14.8% for fixing furniture and utilities to the wall, and (2) have an intention to adopt the listed precautions range from 6.3% for building a fire ladder to 27.4% for fixing furniture and utilities to the wall. The reason that the adoption rates of the precautions are generally low could be due to the fact that most of them are perceived as not very effective and also highly costly and time‐consuming (Lindell, Arlikatti, and Prater 2009; Lindell and Prater 2002).

6. The perceived exposure in this study represents ‘public risk’ (see the discussion on public versus personal risk by Sjöberg (1999, 140)).

7. There appears to be a significant difference between the highest risk area and the lowest risk area. The proportion of respondents who seem to have undertaken precautions is higher in the highest risk area. However, the difference between the moderate risk area and the lowest risk area is nonsignificant.

8. There appears to be a significant difference between the highest risk area and the lowest risk area. However, the difference between the moderate risk area and the highest risk area is nonsignificant.

9. There appears to be a significant difference between the highest risk area and the lowest risk area. The proportion of respondents who seem to have intentions to undertake precautions is higher in the highest risk area. However, the difference between the moderate risk area and the lowest risk area is nonsignificant.

10. There appears to be a significant difference between the highest risk area and the lowest risk area. However, the difference between the moderate risk area and the highest risk area is nonsignificant.

11. There appears to be a significant difference between the moderate risk area and the lowest risk area. The proportion of respondents who seem to put higher monetary value for future precautions is higher in the moderate risk area. However, the difference between the highest risk area and the lowest risk area is nonsignificant.

12. There appears to be a significant difference between the moderate risk area and the lowest risk area. The proportion of respondents who seem to put higher monetary value for future precautions is higher in the moderate risk area. However, the difference between the highest risk area and the lowest risk area is nonsignificant.

13. Since the sample included in the analysis of monetary value of past precautions consists of homeowners who already indicated they do intend to undertake some mitigation action, the average level of dread in this subsample is, in fact, much higher than that in the overall sample (subsample mean is 9.47 as opposed to the overall sample mean of 7.9).