Biases in Maximum Simulated Likelihood Estimation of Bivariate Models

This paper finds that the maximum simulated likelihood (MSL) estimator produces substantial biases when applied to the bivariate normal distribution. A specification of the random parameter bivariate normal model is considered, in which a direct comparison between the MSL and maximum likelihood (ML) estimators is feasible. The analysis shows that MSL produces biased results for the correlation parameter. This paper also finds that the MSL estimator is biased for the bivariate Poisson-lognormal model, developed by Munkin and Trivedi (1999. “Simulated Maximum Likelihood Estimation of Multivariate Mixed-Poisson Regression Models, with Application.” The Econometrics Journal 2: 29–48). A simulation study is conducted, which shows that MSL leads to serious inferential biases, especially large when variance parameters in the true data generating process are small. The MSL estimator produces biases in the estimated marginal effects, conditional means and probabilities of count outcomes.

A Kuhn–Tucker model for behaviour in dictator games

We consider a dictator game experiment in which dictators perform a sequence of giving tasks and taking tasks. The data are used to estimate the parameters of a Stone–Geary utility function over own-payoff and other’s payoff. The econometric model incorporates zero observations (e.g. zero-giving or zero-taking) by applying the Kuhn–Tucker theorem and treating zeros as corner solutions in the dictator’s constrained optimisation problem. The method of maximum simulated likelihood (MSL) is used for estimation. We find that selfishness is significantly lower in taking tasks than in giving tasks, and we attribute this difference to the “cold prickle of taking”.

Designed quadrature to approximate integrals in maximum simulated likelihood estimation

Maximum simulated likelihood estimation of mixed multinomial logit models requires evaluation of a multidimensional integral. Quasi-Monte Carlo (QMC) methods such as Halton sequences and modified Latin hypercube sampling are workhorse methods for integral approximation. Earlier studies explored the potential of sparse grid quadrature (SGQ), but SGQ suffers from negative weights. As an alternative to QMC and SGQ, we looked into the recently developed designed quadrature (DQ) method. DQ requires fewer nodes to get the same level of accuracy as of QMC and SGQ, is as easy to implement, ensures positivity of weights, and can be created on any general polynomial space. We benchmarked DQ against QMC in a Monte Carlo and an empirical study. DQ outperformed QMC in all considered scenarios, is practice-ready and has potential to become the workhorse method for integral approximation.

Estimation of dynamic models of recurrent events with censored data

Summary In this paper, we consider estimation of dynamic models of recurrent events (event histories) in continuous time using censored data. We develop maximum simulated likelihood estimators where missing data are integrated out using Monte Carlo and importance sampling methods. We allow for random effects and integrate out this unobserved heterogeneity using a quadrature rule. In Monte Carlo experiments, we find that maximum simulated likelihood estimation is practically feasible and performs better than both listwise deletion and auxiliary modelling of initial conditions. In an empirical application, we study ischaemic heart disease events for male Maoris in New Zealand.

Job Flexibility and Occupational Selection in Ghana: An Application of Maximum Simulated Likelihood†

In many African labour markets, the vast majority of self-employed workers are female. It is often hypothesised that self-employment enables workers to balance income-generation with caring for children and other domestic tasks and, since responsibility for these activities is divided unequally in the household, this effect is stronger for women than men. However, testing whether ‘job flexibility’ matters is difficult because variables that proxy for domestic obligations—such as the number of dependents in the household—may be endogenous to occupational choice. In this paper, we build a new estimator using maximum simulated likelihood that allows us to use selection on observables as a guide to selection on unobservables within the multinomial choice problem individuals face when deciding their occupation. We apply this approach to detailed cross-sectional data from Ghana. Our results show that having extra dependents in the household pushes women towards own account self-employment substantially more than men, even under more conservative assumptions about the extent of endogeneity.