NLOGITs

NLOGITs, or Nested Logit Models, are a class of discrete choice models used in econometrics and related fields to analyze situations where individuals make choices from a set of alternatives that can be grouped into a hierarchy or nest. Unlike a standard Multinomial Logit (MNL) model, which assumes independence of irrelevant alternatives (IIA), NLOGITs allow for correlation between the unobserved utility components of alternatives within the same nest. This is particularly useful when choices are made sequentially or when certain alternatives are more similar to each other than to others.

The structure of a NLOGIT involves a tree-like representation of the choice set. At the highest level, an individual chooses among broad categories or nests. Once a nest is chosen, the individual then chooses among the specific alternatives within that selected nest. The probability of choosing a specific alternative is calculated as the product of the probability of choosing the nest containing that alternative and the conditional probability of choosing that alternative given that its nest has been selected.

The key advantage of NLOGITs over MNL is their ability to relax the IIA assumption. This is achieved by introducing parameters that capture the degree of similarity or correlation between alternatives within a nest. If these parameters are close to zero, the NLOGIT behaves similarly to an MNL. However, if the parameters are significantly different from zero, it indicates that alternatives within a nest are substitutes for each other, and ignoring this structure could lead to biased estimates. NLOGITs are commonly applied in areas such as transportation mode choice, product selection, and labor supply decisions.