Homework 2
Q1 Cigarette Smoking and Public Policy
5 Points
Since the release of the 1964 Surgeon General’s report, private and public health officials have waged an aggressive campaign against tobacco use. This campaign has included policies as diverse as restrictions on advertising, antismoking public service announcements, education programs, youth-access restrictions, and smoking bans in public places. In aggregate, these policies appear to have achieved some measure of success. Over the past 20 years, per capita cigarette consumption and the fraction of adult smokers have declined considerably.
One policy instrument that has received a considerable amount of attention in the last two decades is higher excise taxes on cigarettes. For instance, between 1990 and 2003, there were approximately 113 increases in state excise taxes on cigarettes. Over that time, the average state tax on cigarettes has increased from 23 to 73 cents/pack. The interest in trying to reduce smoking through higher taxes is spurred on in part by a large body of economic research showing that cigarette consumption falls after price or tax hikes.
In this homework assignment, you will use cross-section data for all 50 U.S. states to predict the proportion of the population that smokes cigarettes. In particular, we will investigate the claim that higher excise taxes on cigarettes are good predictor of cigarette smoking.
Data
Download the dataset ARE106_HW2_smoking.csv from Canvas and open it in R using the following command
D <- read.csv(“ARE106_HW2_smoking.csv”)
The variables in the dataset are:
- state: U.S. state
- smokers: Proportion of the population that smokes cigarettes (in 2007)
- extax: State excise tax per pack (in 2007)
- tax: Other cigarette taxes per pack (in 2007)
- funding: Tobacco control funding per capita (in 2007)
Models
Consider the linear regression models:
where each observation ii represents a state.
Q1.1
0.5 Points
Suppose the population models above are for smokers in all U.S. states in 2020. Do you think assumption CR1 holds? Explain.
Q1.2
0.5 Points
Suppose the errors in Model 1 are correlated across states that are adjacent to one another. Is OLS unbiased in this model? Is it BLUE? Explain.
Q1.3
0.5 Points
Suppose you were to estimate Model 1 using only states along the east coast. Would your estimate of beta_1β1 be unbiased? Explain.
Q1.4
1 Point
What is the mean proportion of the population that smoked cigarettes in 2007? [0.25 points]
What is the median? [0.25 points]
What does the difference (or similarity) between these two statistics tell you about the data? [0.25 points]
Verify your answer with a graph (upload a file of your graph below). [0.25 points]
Q1.5
0.5 Points
Before estimating the models, do you expect beta_1β1 to be positive or negative? Explain.
Q1.6
0.5 Points
Estimate regression Model 1.
What is the value of your estimate of beta_1β1? [0.25 points]
Interpret your estimate of beta_1β1. [0.25 points]
Q1.7
0.5 Points
Does your estimate of beta_1β1 in Model 1 suggest that excise taxes should be used to reduce cigarette smoking? Explain.
Q1.8
0.5 Points
Estimate Models 2 and 3. Does the inclusion of funding and tax in your regression affect your estimate of beta_1β1 from Model 1? Why or why not?