Managerial Decision Making

Managerial Decision Making

Situation Background

This scenario is about Burns Auto Corporation (BAC), an automobile dealership company with a major regional presence that is operating in a highly competitive environment and is contending with uncertainty regarding the accuracy of its sales forecasting (Scenario OneÐ'--Burns Auto Corporation: Read Me First). In order to align with the "turn and earn" approach soon to be implemented by auto manufacturers, BAC needs to reassess their current sales forecasting methods. They need to consider appropriate variables (e.g.: household income, interest rates, sticker prices, seasonal promotions, and gas mileage), collect sales figures from dealerships, and align data against other variables in order to build this model.

The leaders of BAC are presented with three (or possibly more) different statistical models in which to use to forecast sales (Week Four scenario, U of P). BAC already has a very good understanding of forecasting, to the point where theirs is almost a direct opposite problem than that of USA World Bank. USA World Bank did not know enough about statistically driven decisions whereas BAC knows enough to have brought in consultants who have presented three potentially good forecasting models in which to choose. Focusing in on the correct model is their immediate situation.

The forecasting models BAC are considering include correlation and regression analysis, time series analysis, sensitivity analysis, and game theory as methods to improve their future car sales decision making.

Issue Identification

BAC needs to ensure that they are identifying appropriate variables and they need to decide whether to use a top-down or an individual dealership approach in which to conduct data analysis. They also need to be certain they are not making a risky, expensive decision based on only two years of difficult inventory management and sales forecast problems.

BAC has employed the help of two different consultants to build forecasting models. The regression models Peter Reardon builds are designed to generate monthly sales forecasts. Correlation and regression analysis serve the purpose of quantifying relationships among variables. Regression analysis allows the analyst to determine, within a degree of probability, the magnitude of the impact (Correlation and Regression Analysis, U of P). Peter's model considers price indices as independent variables, unit sales as dependent variables, and promotions as dummy variables.

Peter also introduces the game theory and sensitivity analysis to show the impact of the competitive reaction to strategies. (Scenario OneÐ'--Burns Auto Corporation). All of the models he presents are primarily for a top-down analysis of sales.

John Peterson introduces an alternate methodology to sales forecasting: trend analysis, or time series analysis. Forecasts using time series analysis will assist BAC with the planning that is being undertaken by producing a model that better reflects monthly fluctuations in sales (Time Series Analysis, U of P). John's model considers time and monthly dummy variables as the independent variables and unit sales as the dependent variable.

John feels that the time series analysis is a good model for BAC because their sales have increased over time in a consistent way. The dummy variables in his model represent adjustments to monthly sales that occur on a month-to-month basis. His model is also adaptable to a dealership-by-dealership model.

The following table illustrates the differences between the regression and the time series/trend analysis models:

Comparison of Models: Burns Auto Corporation

Regression Model

Time Series/Trend Analysis Model

Unit sales = dependent variable

Unit sales = dependent variable

Price indices = independent variable

Time = independent variable

Promotion = dummy variable

Monthly dummy variables = independent variable

The two models are similar, but the regression model uses price indices as the independent variable, whereas the time series model uses time as the independent variable.

Stakeholder Perspectives

The key stakeholders in this scenario are the employees (including the consultants), the customers, and Thomas Burns. Whenever executives have to make strategic decisions that affect key stakeholders, competing interests can lead to different levels of cooperation (U of P, 2006). It is critical that BAC considers all stakeholders in any business decision, particularly one as important to the company's profitability as sales forecasts.

Opportunity Identification

With the identification of issues as they relate to the key stakeholders, BAC can realize several opportunities. One opportunity that presents is that BAC has the chance to build a new and flexible statistical model that will allow them to adapt quickly to changes in variables (Scenario OneÐ'--Burns Auto Corporation). A flexible forecasting model will facilitate deeper analysis of sales data.

Much of their trend analysis can come from running scenarios and evaluating past sales performance. This presents another opportunity to test their models prior to implementing them. BAC has allocated a $50,000 budget toward this project and their plan is to run in phases:

Phase IÐ'--Test Models = $20,000

Phase IIÐ'--Finalization and Implementation = $30,000

BAC has the opportunity to take the models presented to them and combine them with what they already know about forecasting to facilitate building a model unique to their own company.

Problem Definition

As stated earlier, BAC has a very good understanding of statistical modeling to the point where now they have at least three models in which to choose. Because the leaders of BAC are very knowledgeable in sales forecasting, much of the groundwork has already been done. The problem is that the leaders cannot agree on which model is the best choice to most accurately forecast sales. To realize their end-state, BAC needs to scrutinize the models available and choose the one(s) that will most