Essay 1 – Product-location and temporal aggregation effects on information distortion in the retail supply chain.
In the typical retail supply chain, replenishment decisions are generally made at a Weekly-SKU level. Yet extant literature measures information distortion such as bullwhip (e.g. Bray and Mendelson, 2012) and production smoothing (e.g. Cachon et al., 2007) rely primarily on aggregated data. Further, a detailed look at the general supply chain demand volatilities literature indicates that the most commonly used product-location and temporal aggregate levels are firm and monthly, respectively. Essay 1 utilizes three categories of grocery products at a weekly-SKU level, and measuresinformation distortion adherent to the current dominant “bullwhip” (Lee et al. 1997) perspective, and tests product-location and temporal aggregation’s effects. Essay 1 contributes to the demand volatilities stream of literature by demonstrating the importance of matching the level of analysis with the level of managerial decision-making, that simply aggregating data introduces unintended measurement errors that likely bias theoretical and managerial implications.
Essay 2 – Temporal aggregation’s effect on forecasting in the retail supply chain.
Replenishment and distribution decisions in the retail supply chain are often guided by statistical forecasting (McCarthy et al. 2001). Temporal aggregation can result in indiscriminant minimization of variance. Essay 2 extends essay 1 by using simulation method to test commonly-used forecasting methods such as moving average and autoregressive models to examine temporal aggregation’s effect on forecast error.
Essay 3 – Temporal aggregation and portfolio effect in the retail supply chain.
Theoretical development and managerial implementation of the well-known portfolio effect in inventory management are based on the notion that stochastic demand shocks of two different locations may be centralized to achieve a lower balanced shock at an aggregate level. Yet little is examined with regard to temporal aggregation’s effect. Our analytical models demonstrate that not only does temporal aggregation achieve the similar portfolio effect, but temporal aggregation fundamentally transforms a demand series’s distribution properties.