How we calculate Children’s Vulnerability Index (CVI)

Children’s Vulnerability Index (CVI) is the criteria we apply to identify the families who need job urgently for the sake of their children’s uninterrupted growth. We basically employ Alkire-Foster (AF) approach as developed by Oxford Poverty and Human Development Initiative which has been widely used to measure Multidimensional Poverty Index (MPI). We do some modifications in this method though to meet our own needs. Our unit of analysis is household which is characterized by an index representing its vulnerability to support children’s uninterrupted growth and thus the need of a secure job for the family. We use many other criteria representing family’s deprivation but place a sizable emphasis on children’s growth potential in the family when calculating this index. Other measures include those as applied by United Nations Development Program (UNDP), such as health, education and living conditions, along with other job-related indicators as well to suit it with our own requirement.

The AF method measures poverty based on two criteria: average number of deprivations and the intensity of deprivations. We believe that children’s growth potential in a family depends both on poverty measures such as in AF method as well as children’s prevailing condition in the family, and thus we add one more dimension into the AF method: children’s vulnerability in the family. In the AF method, three parameters are in play to identify the family as poor: first cut-off point (z-vector), second cut-off point (k) and the weight vector (w). We also exercise with these three parameters but with some alterations to make them consistent with our goal. We construct z-vector from minimum values of each deprivation (unlike any value within the range in AF method) and we do this because we assume that all families we consider for negotiating jobs are poor families as our projects target only poor communities. Besides, this assumption is valid because our goal is to rank the households within a poor neighborhood to negotiate the jobs for those beginning with the most deprived ones rather than delineating a poverty line to identify who are poor and who are not poor in the community. As to the second cut-off point (k), we employ the similar approach as in first cut-off point and this is the minimum value of the counting vector (the vector consisting of the elements of weighted number of deprivations for each family). This is to demonstrate our union approach in selecting this parameter such that the family is regarded as deprived even if it lacks in at least one indicator. In other words, we do not censor any data, unlike in AF method, except the family with highest profile to use it as the base family to rank others. The weights, on the other hand, are assigned for each indicator accordingly to reflect how local people perceive poverty in their communities along with our informed judgment (such as we place higher weights on children’s growth potential in the family). In calculating intensity of deprivations, the second criteria in the AF method, which is reflected in g0-matrix and g1-matrix, we standardize the observations about the minimum value and the standard deviation of the series so as to capture how far a family lies from the lowest-standing neighbor in the community, unlike the approach adopted by AF method according to which absolute differences are used. Our approach encapsulate the property that when the households are disproportionately distributed in a certain indicator, this approach assigns standardized values for each household in consistent with the distribution of the families to reflect the household’s relative position in the community. For example, if a community has twenty households and only two of them are not able to send their children in school due to financial difficulty, these two households will have higher standardized values (which essentially contributes more in calculating Children’s Vulnerability Index and thus receive higher priority in job negotiation) than the situation when there is a equal number of households in the community with the similar financial condition and children’s situation. This feature is not captured in absolute difference approach and each deprived family receives the same standardized value no matter how many of them are in the community.

The children’s vulnerability in the family, on the other side, captures how much a household is poor as compared to others based on the likelihood of the family falling into the class that is mostly deprived in terms of children’s growth potential indicators. To capture this attribute of deprivation, we employ latent class analysis which generates multidimensional constructs of various deprivation indicators in certain classes of common features and estimate posterior probabilities of each household falling on each class. We first identify the class which is characterized by intensive deprivation in terms of children’s growth potential and then we estimate the posterior probabilities that the family will fall on this class. And these posterior probabilities are used to construct an index for each family reflecting how much likely the family posing threat to children’s growth in the future.

We finally combine all these three deprivation indices to construct a composite Children’s Vulnerability Index (CVI) reflecting a household’s relative standing in the community and thus the need of a job for the family to raise their children in secure financial environment.