Topic Area: Population

Geographic Area: India

Focal Question: Do declines in the rate of childhood mortality lead to decreased fertility?

Source:

(1) Atella, Vincenzo; Rosati, Furio Camillo; “Uncertainty about children's survival and fertility:

A test using indian microdata.” Journal of Population Economics, 2000, Vol. 13 Issue 2,

p263-278.

Reviewer: Neil Crimins ’02

Review:

Atella and Rosati with the help of the Human Development of India Survey data attempted to understand the impact of childhood mortality on the rate of fertility. In 1994, the survey was conducted in two stages sampling 34,398 households in 1,765 villages located in 195 districts in 16 states. The first survey instrument was designed for adult males with the intention of determining the income and economic factors affecting a household, while the second survey determined rates of literacy, education, health, morbidity, nutrition, and demographic parameters for adult females members of the household. With this information, the authors performed a regression analysis to show how many of the characteristics listed above were related to the number of births that a family would have (269).

Theory provides some suggestions about the expected nature of this relationship. The reality of childhood mortality impacts parent’s decision making in the following way: first, the average number of children a family expects to survive, and second, the risk of being a family that looses a child are both unknown. Over the course of one’s lifetime he or she will hopefully live over three stages of life. The middle stage of life is the income accumulating stage where an individual is independent. The first and last stages of life are considered the periods when one is not working and is therefore dependent on the resources they have accumulated or have access too. The middle stage of life is also when fertility choices are made. Part of the finances for the first and later stages of life come from the middle stage in what are called intergenerational transfers. The differences between the first and last stages of course rests in the notion that over the course of one’s life she has the option during her working years to put money away in the hopes that it will be there when she needs it, while children do not come with a nest egg attached. From the perspective of the middle stage there is both a recoverable cost in the form of the hope that when they are old that their children will take care of them, and an unrecoverable cost in the form of the nurturing, education, and invested parental time that hopefully provide an economic reward later in life that is unique to the first stage. Therefore, for parents children are one way of investing in your retirement, along with other methods like putting the money you would not have spent on that child in a bank where it may earn a rate of return. Unfortunately, many in the developed world do not have access to institutions that will pay them a rate of return or even institutions that will simply hold onto one’s assets for the future. Therefore, the options that exist for middle stage individuals are limited to ensure that they will be taken care of in old age (264-66).

A child as the sole option for ensuring one’s third stage of life is taken care of comes with an assessable risk of defaulting on his/her responsibility when the child reaches the middle stage of life and is income earning itself. In the absence of risk parents would have one child and give that child all of the resources it required, but with risk comes the necessity for more children as a means of spreading out the risk of being left without a safety net in old age. Combine this risk of default with the risk of premature death before a child reach’s the middle income earning stage and you leave parents faced with an even greater prospect of poverty in their final years. Given that mortality is unknown, and the risk of default is unknown, it is advantageous for a parent in order to maximize their lifetime utility to spread their available resources over a larger pool of children. In India where there is a high risk of mortality relative to a developed nation like France, the aversion that middle staged individuals in India have to risk may be low or high. For example, if an individual has a low degree of risk aversion they will have fewer children (i.e. an income like effect), and as a result there will be more resources allocated per child. The secondary effect or what might be viewed as a substitution effect will result in increased willingness to transfer part of his income to old age. Both of these effects are part of the notion that not only is the mean value of what you will consume in old age important, but also the variance of how much you will consume in old age. Furthermore, given that the poor of India have little access to banks and other financial markets for reserving funds for old age consumption, the increase in the variance results in increases in the number of children parents will decide to have. By increasing the variance of the survival rate, parents are less likely to put all of their eggs in one basket, but rather spread out their available resources to many children. Thus there exists a cyclical effect that occurs when parents have no other option than to have children to ensure during old age they will be financially comfortable (267-68).

With the theory in mind, the authors returned to the empirical evidence to see what influence mortality really does have on fertility and how that influence compares to the influence of other factors like income, literacy, education etc.. The econometric estimates were divided into two parts: first, parents who already have children must decide how many more children, or what size will their family ultimately be. In this group, the results for women considered to be in their childbearing years (between the ages of 18-49) found that mortality had no effect on the expected number of children. While the education of the husband, had no significant influence on fertility choices, the education of the wife had a significant negative effect. The more educated a wife is the fewer children she is expected to have. Religion also played a role. Muslims and other religions tended to show higher rates of fertility than Hindus with Christians having the lowest rate of fertility (275).

The effects of income and wage rates on the number of children were difficult to interpret and therefore required a deeper analysis. The authors used the size of cultivated land and a measure of poverty that classifies households into four groups as proxies for the income effects. The results show that increased land size has a positive effect, meaning the more land you have the more children you will have. But underneath is the reality that reverse correlation might skew the results since more children may be necessary to farm larger plots of land. The four groups classified for the poverty measurement showed that increased income leads to increased number of children until you get to very high levels of income where the elasticity of demand for children becomes very small and subsequently the variations in the number of children are very small. This type of relationship suggests that a nonlinear trend may exist. And finally, the major question the authors seek to ask is whether the survival rate has a negative effect on the number of children a family has while controlling for the prior characteristics. The results show that in the case of both the mean and variance of the survival rate at the village level there is a negative effect, which is consistent with what would have been expected. When the average survival rate declines, the average number of children a family will have declines. Furthermore, the spread, or the probability of having a family with a large number of children also declines, leaving society with a greater chance of having families with a reasonable amount of children. As mentioned earlier, parent’s decisions to have children where linked to how many children they already had. The second part of the model separates the two, and looks simply at how children parents will have regardless of how many children they already have. As a result the age coefficients of the wife and husband are larger, Christians are more likely to have children, and increased education for the husband positively influences the decision to have children. Finally, the mean and the variance are significant with a negative sign, which means that an increase in the variance of the survival rate of children will lower the expected number of children a family has and the likelihood of the family having children in the first place (275-76).

Policy implementers need to focus not only on improving the survival rate, but also on the limiting the variance in family size over time. It does not help to lower the average with out increasing the odds that families will have smaller families. The feedback effect may rest in the ability of policy to decrease mortality, which would then send a signal to families that there is less of a risk associated with loosing a child. The results could end up being the exact opposite of what is intended, where families decide they now are able to have more children because the risk of death is far less of a concern. In order to combat the problem of overpopulation effectively, policy needs to attack each front with the same level of intensity. It will useless if money and time are spent on policies that leave village and townships worse off then before. Both the issues of variance and of articulating an appropriate signal for family size need to be aggressively understood before any solution to the population crisis in the developing world can be revealed (277).