A Student Guide to Managerial Economics at Dickinson College*
January 2001
Stephen Erfle
Associate Professor and Chair
International Studies
Dickinson College
Carlisle, PA 17013
* Comments are greatly appreciated. Your comments will help make this class more useful and enjoyable to future IB&M majors. If you prefer to provide your comments anonymously, just email them to the IS/IB&M secretary – she can take your name off, and give them to me.
Managerial economics teaches you how to use the economic tools you began learning in introductory microeconomics in order to make managerial decisions. This guide provides an overview of how the various parts of managerial economics tie together. I hope this guide will provide a roadmap for the course. Certain parts of the course are admittedly difficult, but they will be less difficult, or at least more manageable, if you understand why I am teaching each topic that is being covered in class and in lab. Before I begin this discussion, I would like to briefly explain why I think this is an integral part of the International Business and Management (IB&M) curriculum.
Background and overview
During the 1994-95 academic year, I took a sabbatical in which I worked as an economist for Seagram Classics Wine Company. Although I am trained as an industrial organization / regulation economist, I became interested in how managers might more effectively use economics for decision-making purposes, i.e., I got interested in managerial economics. When I returned to Dickinson College in the fall of 1995 I had revised my vision of how I could best teach economics to students. Basically I had come to believe that economics might best be understood by placing greater emphasis on practical applications, and less on theoretical development. Too often, economists focus on theoretical modeling at the expense of empirical relevance.
Upon returning from sabbatical, I became actively involved with a group of faculty who were attempting to create a business major at Dickinson. The faculty of the College approved the resulting major, IB&M, during the fall of 1996. As you are undoubtedly aware, one of the core courses in this major is managerial economics. In my view, managerial economics is most appropriately taught using the same toolkit that business managers rely on – today, that means use Excel as the platform to teach the course.
My initial vision for this course was to teach a variety of topics covered in managerial economics textbooks using Excel as the teaching platform. I wanted to use this class to simulate what students will face working in a business setting. In such settings, quantitative issues are typically examined via Excel. (I found this out first hand at Seagram – I had never used Excel (or other spreadsheet programs) prior to starting work there – learning Excel was a byproduct of one of my first tasks that year.) I began searching for managerial economics texts that were integrated with spreadsheet programs and found that such texts did not exist (despite the interest that publishers had in the idea of an integrated text)[1]. Those texts that were integrated with software programs use specialized programs (written by the authors), rather than the general-purpose programs (such as Excel) found in the business world. To my mind, specialized programs that are tied to textbooks provide little benefit as they are never encountered in the real world. If a student learns how to do a present value problem using a general-purpose program they will have learned much more than they would by doing that same problem using a specialized software package that will never be seen again. As a result, I began to develop Excel based lab assignments that would supplement materials from a managerial economics text.
I have taught four iterations of this course since its inception in the spring of 1998. I have sacrificed breadth of coverage for depth – this has been necessitated, in large part, because IB&M majors are not required to take either calculus or statistics as part of the major. As a result, basic calculus and statistical concepts that are useful tools for managerial decision-making must be taught from the ground up.
As it now stands, the course is broadly divided into three parts. The first examines the concept of maximization, the second provides an introduction to statistical analysis with the goal of understanding basic regression modeling, and the third essentially combines the ideas presented in the first two parts of the course. By the end of the semester, you should be able to answer complicated but realistic questions using the tools you have at your disposal. Excel is used throughout the course, both in lab, for homework, as well as in most classes. I now turn to a discussion of the content of lectures and labs during each of these parts of the course. The following pages relate directly to the outline you received with your syllabus for the semester.
Weeks 1-4: The Concept of Optimization and an introduction to Excel
During the first third of the semester, we examine basic constrained optimization in class, and at the same time learn the rudiments of Excel in lab. You are not required to have a background in calculus or Excel prior to this class; you need to learn the basics of each before we can proceed. I believe that this is best accomplished by learning each of these tools independently, rather than trying to learn both at once. In doing so, I am following the economist’s maxim of trying to understand problems by varying one thing at a time.[2] As a result, we will learn the basics of optimization with pencil and paper in class, rather than try to learn optimization with Excel (although Excel certainly could be useful in this regard). The discussion of class content and lab content will also be done sequentially in this section.
One of the core concepts of microeconomics is marginal analysis. As you recall, a firm maximizes profits by setting marginal revenue equal to marginal cost. More generally, any economic actor increases their welfare by undertaking an activity so long as the marginal benefit of that activity exceeds the marginal cost of that activity. Welfare is maximized at the point where the marginal benefit equals the marginal cost of the activity. In introductory microeconomics you learned that marginal means “change in” – or more specifically slope. Marginal revenue is the slope of total revenue; marginal utility is the slope of total utility; and so on. You may also be aware (even if you have not had calculus) that calculus is the study of change in functions. In particular, a slope is simply a derivative – and this is the reason we must understand some basics from calculus.[3] We must first examine functions before examining slopes of functions.
You may have noticed that I said optimize at first, then I used examples in which the firm or the individual tried to maximize. Optimize is simply a more general goal, because sometimes we want to minimize rather than maximize (for example, economists often talk about minimizing cost). Luckily, marginal analysis allows us to analyze both situations using the same tool – the derivative. If you remind yourself that a derivative is simply another way to talk about slope this should help you avoid being confused.
Initially we examine derivatives in the context of functions of one variable (univariate functions). This allows us to analyze some simple economic situations, but most economic decision-makers face more than one decision variable. For example, a firm wishing to maximize profits would want to set not only the profit maximizing level of output (or more accurately the appropriate price/quantity pair), but also the profit maximizing level of non-price decision variables such as advertising. This is easily modeled using multivariate functions.
Oftentimes economic decision-makers face constraints on their actions that must be taken into consideration when deciding the course of action to follow. To take a simple example, in introductory microeconomics you assumed that a rational consumer maximizes utility. But that consumer faces a critical constraint on their ability to maximize utility – they must stay within their budget. Such problems are called constrained optimization problems. There are a variety of methods of dealing with constrained optimization, one of which we learn in this course. The rest of section applies these basic tools in order to help you gain familiarity with their use.
The first three labs provide an introduction to Excel.[4] This is accomplished by attacking successively more complicated present value problems using Excel.[5] The first lab provides a break-even analysis of an investment project discussed in the text, the revival of the Broadway musical Showboat. The second lab extends the Showboat analysis by examining the more realistic case of facing a probability distribution of when the show will shutdown. This lab introduces you to linking worksheets and also highlights Excel’s goal-seek function. The third lab is more free-form. I provide you with various assumptions about wages and college costs and ask you to answer the question: “should you go to college?” The goal is to have you create your own architecture to answer the question. . The first lab midterm tests basic Excel techniques by requiring you to design a structure to answer a present value question that is given as a word problem.
Weeks 5-9: Introduction to Regression Analysis
From this point forward, lectures and labs are much more integrated – rarely does a lecture occur which does not include at least some material based on Excel, and some lectures are entirely Excel based. As noted above, this section of the course introduces you to an important statistical technique used to estimate a variety of economic relationships – regression analysis. The goal of this course is to provide you with a user perspective for applying quantitative analysis – Excel provides the perfect platform to that end. Certainly there are better econometric software packages in the market, and it is more convenient to use such packages if one is attempting to undertake advanced econometric estimation, but none of these packages are widely available. By contrast, Excel is ubiquitous.
In the first part of the course we learned, for example, how a manager might maximize profits given demand and cost equations. The second part of the course tackles the question of how to obtain those equations. A manager might simply assume they know how marginal costs of production, or price they can charge for their product, varies with output, but such assumptions will likely lead to less than optimal performance. Managers make better decisions if their decisions have some empirical basis in fact. Of course, it is not always easy to obtain empirical data (and we will largely ignore the question of how to obtain such data in this course). Instead we will focus our attention on how we can use empirical data to model economic behavior.
We therefore begin by discussing some basic statistical concepts in week 5. In order to show you the value understanding regression analysis, I will provide you with the results of a regression analysis based on my own research into pricing in the residential real estate market. This presentation will not be technical in nature – rather, it is provided as a non-technical introduction to the topic. From this I hope to provide you with a flavor for the kinds of questions you can examine by using regression analysis. The paper on which this is based is available on CourseInfo, but it is not required reading for the course (and you will not be tested on anything that comes out of this lecture).
The lab for the week is a statistics lab that provides you the opportunity to examine statistical data using very real world data – your own scores and the scores of previous classes on the first in-class and lab midterm. This allows you to use Excel to calculate mean, standard deviation, correlation, and z scores and to use these concepts to answer questions such as which class had the hardest exam? I also use this lab to introduce univariate regression by asking you to predict your lab performance as a function of your in-class performance. Not only does this provide a nice introduction to regression, and it allows you to explain what it means to have a negative (or positive) residual. It also allows you to understand my grading system that is based on z scores.
In week 6 we begin a series of lectures that deal with regression analysis. These individual lectures build on each other and are interrelated. The labs build on the lectures, and the lectures build on labs, so it is useful to have a feel for what is happening in both lab and lecture. The class by class description below may help tie these topics together into a more cohesive whole.
Regression lecture 1) The standard exposition of regression analysis proceeds from univariate to multivariate regression for good reason – two-dimensional graphs are more readily understood than are higher dimensional graphs. It is easy to visualize how an independent variable x relates to the dependent variable y, because one can readily see that some lines fit the data better than others. The method of least squares can be heuristically thought of as giving you the answer to the question what is the best fit line? I utilize the small data set of nine (x,y) observations that Mansfield describes on p. 141 to make his case for the method of least squares. You should read the material on least squares on pp. 144-5, but read to get the big picture, do not let yourself get bogged down in the mathematical detail. The reason you do not need to be concerned with all summation signs is that Excel does all of those calculations for you, in a matter of moments. Mansfield’s example has a best fit line of approximately y=2.5 + .5x and this is easily seen on a graph before running the regression. I will then run the regression and show that our visual best guess is mirrored by the results of the regression (both from the line fit plot, and from the estimated coefficients themselves). I will also hand out a copy of Excel’s regression output so that you can take notes directly on the output regarding its various parts. I will also use this regression to explain the concept of total variation in the dependent variable, and how this is broken down into variation explained by the regression, and unexplained variation – the coefficients are chosen so that unexplained variation is as small as possible. (This is called an analysis of variance (ANOVA).) Finally, I relate these concepts to R2.
Regression lecture 2) In this lecture move to multiple regression and discuss the interpretation of coefficients for each independent variable as simply slopes in different directions. With two independent variables, we are trying to find the best fit plane, and Excel can do this just as easily as it did one variable to find the best fit line. From here you should able to make the leap to how a set of n independent variables x1, …, xn relate to the dependent variable y without having to try to visualize the best fitting n-dimensional hyperplane. To do this, just keep in mind that the steepness of slope in a given direction is easy to conceptualize – it is simply rise (in y) over run (in xi if we are looking in the “ith ” direction). I also spend some time discussing the difference between an empirical estimate of a relationship, and the “true” underlying relationship (this distinction is discussed on pp. 141-3).
I then move to Excel in order to give you a preview of regression lab 1. I also begin to talk about other parts of the regression output such as the standard error of coefficients, and standard error of estimate (unfortunately, both are simply titled “Standard Error” in Excel’s output, so the difference must be explicitly pointed out in order to avoid confusion).
Regression lab 1) You learn how to estimate a demand relation. This is a hands-on introduction to multivariate regression analysis that allows you to grapple with dependent versus independent variables, and to interpret coefficient size and significance.
Regression lecture 3) Simulations using Excel allow me to examine a wide array of issues that must be understood in order to correctly use regression analysis. It is important to understand the limitations and underlying assumptions involved in using regression analysis, as well as what the regression output is actually telling us. In this lecture, I use a simulation model to get at the difference between magnitude and significance. Specifically I introduce error to either the slope or intercept term and show how this affects the resulting regression. This allows for an extended discussion of what we mean when we say a coefficient is significantly different from zero, as well as how we might interpret the 95% confidence interval for the coefficient. I also use this lecture to begin discussion of residual analysis, a topic that is central to the next four lectures. Two issues involving residuals are examined in this lecture: nonlinearities, and heteroscedasticity. The student presentation for the day acts as a springboard for the nonlinearities discussion; the simulation model allows an introduction to heteroscedasticity.