Structural Estimation of Callers' Delay Sensitivity in Call Centers

Abstract

W e model the decision-making process of callers in call centers as an optimal stopping problem. After each waiting period, a caller decides whether to abandon a call or continue to wait. The utility of a caller is modeled as a function of her waiting cost and reward for service. We use a random-coefficients model to capture the heterogeneity of the callers and estimate the cost and reward parameters of the callers using the data from individual calls made to an Israeli call center. We also conduct a series of counterfactual analyses that explore the effects of changes in service discipline on resulting waiting times and abandonment rates. Our analysis reveals that modeling endogenous caller behavior can be important when major changes (such as a change in service discipline) are implemented and that using a model with an exogenously specified abandonment distribution may be misleading.

OM Forum—Causal Inference Models in Operations Management

“…The causal link between queuing time and firm revenues is an important topic in queue management research. Recent well-cited studies on this topic include those by Allon et al (2011), Lu et al (2013), Akşin et al (2013), Batt and Terwiesch (2015), and Song et al (2015). Song et al (2015) use a DID model to show that a dedicated queuing system reduces waiting time relative to a pooled queuing system.…”

“…Interested readers can refer to Greene (2000) and Wooldridge (2008) for textbook-level discussions on such models. 3 A few examples of structural modeling papers in OM include those by Akşin et al (2013), Caro et al (2014), Golrezaei et al (2014), Kim et al (2014), Li et al (2014), and Hyndman and Parmeter (2015). 4 A third reason for the endogeneity bias is due to measurement error in the explanatory variables.…”

Waiting Patiently: An Empirical Study of Queue Abandonment in an Emergency Department

“…For example, Larson (1987) discusses such issues as perceived queue fairness and waiting before or after service initiation, both of which likely impact expected residual time. Janakiraman et al (2011) Aksin et al (2012) in terms of both setting and methodology. Our study setting is a semi-visible, multi-class queue (in the ED, the waiting room is visible but the clinical treatment area is not) as compared to an invisible multi-class queue.…”

“…Our study setting is a semi-visible, multi-class queue (in the ED, the waiting room is visible but the clinical treatment area is not) as compared to an invisible multi-class queue. In terms of methodology, to estimate the latent structural parameters, Aksin et al (2012) imposes strong structural assumptions (e.g. known common hazard function, linear waiting costs, past time is sunk, etc.).…”

“…Abstracting from the optimal stopping problem formulation of Aksin et al (2012), we assume that the abandonment decision is the result of a patient repeatedly evaluating the following personal utility function:…”