Introduction: This article estimates the multidimensional poverty index using a method to obtain weightings associated with the indicators used. Generally in the literature, to calculate this type of compound matters, arbitrary values are assigned to those weightings. This way of operating therefore leaves aside the subjectivities present that could arise in computations when arbitrarily establishing the weights, thus avoiding biases of estimation. Methodology: The methodology used is known as the partial least squares regression in path modeling (plspm). The model estimates the weightings under a system of equations that makes it possible to objectively observe the relative importance of the indicators associated with poverty. Results: The fundamental contribution of this article is that, in contrast to other studies that use multidimensional approaches, it is able to systematically estimate the weightings that each dimension provides in building the poverty indicator. A second contribution from this study is that the weightings are obtained using an innovative methodology that has been little explored in areas of economics. Conclusions: The empirical application of the proposed methodology makes it possible to elucidate the great differences among two Latin American countries. This shows the market contrasts among these two countries despite their location in relatively nearby regions.