ICFC TUTORIALS

DEMAND ANALYSIS USING POOLED TIME SERIES-CROSS SECTIONAL DATA
Topics Covered
· Types of data available for econometric modelling: time-series, cross-sectional and pooled time-series / cross-sectional or panel data
· The power of a pooled dataset: enhanced variation
· Econometric issues and techniques
· A case study using pooled data on the mobile phone market in the U.K. to estimate price elasticity

Chris Dineen is a Managing Consultant, Economic Analysis with Teligen, Total Research. (formerly Eurodata), a London based telecommunications consultancy specializing in pricing, competition and regulatory analysis. Before joining Eurodata in 1998, he worked for 11 years in the demand analysis and regulatory areas with Bell Canada. Mr. Dineen received his education in economics and econometrics at Concordia University in Montreal and University of Toronto. He is a member of the Planning Committee of the International Communications Forecasting Conference (ICFC).

SURVIVAL/HAZARD MODELS FOR TELECOMMUNICATIONS DEMAND APPLICATIONS
This tutorial will introduce methods known as survival or hazard models for modeling the occurrence of events. These models make use of timing information available in some data to create a causal explanation of risk as a function of selected covariates. The applications to telecommunications market demand include analysis and prediction of subscription to, or churn from, a network for residential and business customers. Examples will be presented from a detailed data set on long distance service subscribers.

R. B. Williamson is a telecommunications demand analyst with seventeen years experience in work with companies such as SBC Communications and US WEST. He is currently Worldwide Director for Telecoms research for Taylor Nelson Sofres, a global market research firm headquartered in London, and assists international clients with market demand modeling, pricing analysis, loyalty, and satisfaction studies for mobile and fixed networks.

CHALLENGES FOR FORECASTERS IN A DIGITAL AGE
The first portion of the tutorial will be a panel discussion that will have various companies describe how forecasting incorporates itself into the budget process. It will also include discussions that all will discuss on how to remain a viable part of the company as demand shifts from switched access lines to broadband services.

The second portion will describe a study done to aggregate wire centers (10 to 40) to a level that will allow them to be forecast accurately. The key is to keep the forecast at a fairly low number of aggregate wire centers in order to preserve the information to be obtained at small area geographies.

Ron Luginbill has been employed for 31 years in Detroit working mostly as a forecaster for the Baby Bell that owns the Michigan territory. Current duties is as an Area Manager- Local Area Forecasting AIT/SNET/National Local, which requires the supervision of 10 Local Area forecasters who forecast switched, unbundled loops, DSL and Non-switched services for approximately 150 wire centers each. Past positions include 6 years as a top down Administrative Access Line Forecaster, 20 as a forecaster of the Michigan Economy, 2 years Regulatory Manager, and 2 years in Corporate Planning. He earned a Master's Degree in Economics from Kent State University.

RECENT ADVANCES IN SURVEY-BASED ANALYSES OF BRAND MARKETS SHARE
Survey choice analysis -- which directly models customers' stated choices among brands and products -- has become the standard technique to forecast competitive impacts of brand-specific price and feature changes on each brand's share. Recently, techniques have become available to estimate Customer-level Choice Models, as presented by Kenneth Train at the 2000 ICFC Conference in Seattle.

Now, in 2002, we will present two Customer-level Choice Models techniques (Method of Simulated Likelihood and Hierarchical Bayes) to demonstrate how they avoid the pitfall of assuming the same price and feature preferences for customers whose preferences are actually very different.

The tutorial's Case Study approach will assess the ability of these models to:
" Estimate brand market share, competitive impacts and price or feature elasticities for specific market segments;
" Enable market segmentation based on individual purchase behavior;
" Help your company target customers with high purchase motivation;
" Empower your company to tailor product and marketing message to specific audiences.

John Colias offers almost two decades of experience with conjoint and choice-based optimization methods, new product development and forecasting, advanced segmentation tools, market share analysis, and brand value. His recent optimization and forecasting applications evaluated new consumer and B2B services for telecommunications, energy, high-tech, banking, and packaged good clients. As Vice President of M/A/R/C® Research, John manages the marketing science department where his team designs, implements, and interprets custom quantitative research. Prior to joining M/A/R/C, John was a Research Economist at BellSouth. John holds a doctorate in economics from The University of Texas at Austin with econometrics and modeling specializations.