Appendix and Glossary

• Pages: 371-421

371

APPENDIX AND GLOSSARY

A. Introduction

This Appendix is designed to be a convenient reference for the

econometric concepts described in the text. It begins with a brief

discussion of basic statistical concepts, regression analysis and the

precision of regression results. Following the primer, this chapter

provides a quick-reference glossary containing terms relevant to this

discussion as well as numerous more advanced concepts. The glossary

includes references to econometrics textbooks and articles providing

more in-depth discussion.

B. Correlation Between Two Variables

A correlation coefficient describes how strongly two variables are

related. It can range in value from -1, indicating a perfectly negative

relationship between two variables, to +1, indicating a perfectly positive

relationship between two variables. Positively correlated variables move

in the same direction while negatively correlated variables move in

opposite directions.

Consider the relationship between temperature and precipitation in

various locations. Figure 1 plots the relationship between the monthly

averages of high temperature and precipitation in Orlando, FL. Each data

point reflects the average high temperatures and precipitation for a

particular month—January, February, etc.—calculated over several years

of data. As indicated by the trend line in Figure 1, the two variables are

positively correlated: there is more precipitation during warmer months.

The correlation coefficient between temperature and precipitation in

Orlando is 0.84.

372 Econometrics

Figure 1. Orlando, FL Climate

Figure 2 plots the relationship between the monthly average high

temperature and monthly average precipitation in San Francisco, CA. In

this location the two climate measures are negatively correlated: there is

more precipitation during cooler months. The correlation coefficient

between temperature and precipitation in San Francisco is -0.90.

Figure 2. San Francisco, CA Climate

Appendix and Glossar y 373

Figure 3 plots the relationship between the monthly average high

temperature and monthly average precipitation in London, UK. Any

relationship is hard to discern: one could say that the two climate

measures are virtually uncorrelated in this location. The correlation

coefficient between the average high temperature and average

precipitation in London is 0.05.

Figure 3. London, UK Climate

C. Classic Linear Regression Analysis

Regression analysis refers to techniques used to evaluate and model

the relationships among several variables. The values of a dependent

variable are analyzed in relation to the values of one or more

independent, or explanatory, variables that might explain variation in the

dependent variable. The coefficient on each independent variable in the

regression model indicates the degree of correlation between that

variable and the dependent variable, assuming all other independent

variables are constant.

Regression analysis is frequently used in antitrust matters, for

purposes such as investigating whether a firm has engaged in predatory

pricing or unilateral anticompetitive conduct, determining whether the

number of firms in a market is negatively correlated to price when

controlling for relevant factors such as cost, or predicting the percentage