AXIOMATICALLY SOUND POVERTY MEASUREMENT WITH SCARCE DATA AND PRICE DISPERSION

We derive a parametric formula of the Watts' poverty index for the bivariate lognormal distribution of price indices and nominal living standards. This enables us to analyze the contributions of price and nominal living standard distributions to poverty, to estimate poverty when only means and variances are known. We also derive a statistical inference framework. Using data from peasants in Rwanda in four quarters, we show that poverty estimates based on scarce information are generally not significantly different from nonparametric estimates based on full survey information.