2.2. TEORÍAS QUE SUSTENTAN LA INVESTIGACIÓN
2.2.2. Teoría de Gestión del Conocimiento de Gilbert Probst
The value of a statistical life (VSL) is generally estimated one of three ways. The first, called the hedonic wage approach, examines large datasets on wages and job risks (typically in
manufacturing jobs) to statistically identify the wage differentials that compensate a worker for higher on-the-job fatality risk. The second approach is a survey-based, stated-preference approach. The survey presents a product or program to respondents which will reduce their risk of dying by some specified amount (e.g. from 5/1000 to 1/1000) and elicits how much the respondent is willing to pay for the program. The VSL implied is the WTP divided by the risk reduction. (If mean WTP for a program that achieved a risk reduction of 4 in 1000 was $500, then the implied VSL would be $125,000.) The third approach examines expenditures on products that reduce risk of dying (i.e.
safer cars, bike helmets, etc.). There is now a large number of all three types of VSL studies in high- income countries, but relatively fewer in low-income countries. We examine four that are
particularly relevant for India.
Shanmugam (2001) uses the first approach – examining wage premiums for risky jobs – to estimate VSLs for adults in India. The paper uses survey data on wages of blue-collar workers in factories in one city in 1990 (Madras, in the state of Tamil Nadu). The mortality risk associated with a given factory was based on state government data on on-the-job fatalities, grouped according to an industry classification code. Shanmugam’s estimates ranged from Rs.10M – Rs.56M in 2001. Inflating these estimates to the same 2004 terms used throughout the dissertation19, the estimates range from Rs.11M – Rs.63M, or US$250,000 – US$1.4M.
Simon et al (1999) use a similar approach but do not limit their data to only one geographic area in India. The authors use nationwide data from the Occupational Wage Survey of the Indian Labor Bureau and data on job fatalities, again classified by industry classification code (average risk of dying on the job was approximately 15 per 100,000, compared with 8 per 100,000 in the U.S.). The regressions of fatality risk on wages, controlling for several personal and job characteristics, imply a VSL between Rs.17M – Rs.41M, or US$370,000 – US$920,000, in 2004 terms (Rs.6M – Rs.15M in the 1990 terms presented in Simon et al).
There are, however, concerns with using labor market studies for VSL estimates. The hedonic wage approach assumes that workers are aware of their on-the-job fatality risk. It observes the willingness-to-accept this higher risk for a higher wage, which may be much higher than an analogous willingness-to-pay to reduce risk (Hanemann 1991). It also applies only to working adults, typically to healthy male workers who take risky manufacturing jobs.
Bhattacharya et al. (2007) used the stated preference approach, asking 1200 commuters (pedestrians, cyclists and motorists) in New Delhi about their WTP for several different programs and
19 Inflated using data from IMF’s World Economic Outlook April 2007.
products that would reduce their risk of dying in a traffic-related fatality while commuting to work. They find that WTP increased with the risk reduction, income and existing road traffic fatality risk. Their preferred estimate of VSL is Rs.1.3M in 2007 Rupees, corresponding to about Rs.1.1M, or US$25,000 in 2004 terms.
All of the studies above were concerned with WTP to reduce mortality risk for working adults. Few studies asks parents their WTP to reduce their children’s risk of dying (or observe their WTP for risk-reducing products). Maskery et al (2007) presented parents in Matlab, Bangladesh with a generic nutritional health supplement that would reduce their youngest child’s risk of dying. They illustrated the risk reductions offered and average baseline risks facing children in the
respondents’ communities using illustrated risk ladders and extensive training on probabilities. The nutritional supplement offered either a 20% or 60% risk reduction from the stated baseline risk (5 in 1000 per year) over a period of five years. Mean WTP (for a one-month supply of the supplement) was US$1.5, or 2% of average monthly household income in the sample (US$75). Responses were not significantly different between the two levels of risk reduction. WTP was higher for young children (under 5) than for school-aged children (aged 5-17), implying a higher VSL for young children. Although the authors were not able to present age-specific mortality risks to respondents in the scenarios, they note that the actual baseline risk of dying for young children is three times higher than for older children. Using the baseline risk given in the scenario, VSLs for young children are US$120K – $320K and for older children are US$60K - $180K. If, however, respondents’ answers reflected reductions from the actual baseline risk facing the two groups of children, the VSL for young children and older children are US$24K - $75K and US$40K - $120K.
Given the available evidence, we choose the most conservative plausible estimates of VSL available. Given that average incomes in Tiljala are very similar to those in the Maskery et al (2007) study in Bangladesh (US$65 in Tiljala, US$75 in Matlab), we feel reasonably confident directly transferring estimates from that study. These are also the only available estimates of the value of mortality risk reductions for children in low income countries. Because the evidence seems weak that
parents have a different WTP to protect children of different ages, we use only one VSL estimate – US$25,000 – for both groups. Estimates for Indian adults in the literature above ranged from
US$25,000 to US$1.6M, although the populations in these studies were somewhat wealthier than our Tiljala population. To be conservative, we also use a VSL for adults of US$25,000.
Simply taking the most conservative VSL found in the literature is no guarantee that the true VSL might be still lower. The essential question is whether we believe Indian society is willing to spend $25,000 to prevent the loss of a life in a low-income slum in Kolkata. On the other hand, VSL might higher than $25,000, and indeed all of the evidence above indicates that it is. We therefore assume a lower bound of $20,000 and an upper bound of $50,000 in the uncertainty analysis.
4.6 Other parameters