HW # 3 Economics of Crime 1. a. Suppose in Mexico there is a single drug kingpin who supplies all of the drugs. The kingpin’s marginal cost of drugs of 20 and the demand curve in the united states is 220-Q. Suppose 1 person overdoses for every 10 units of drugs that are consumed. What is the price and quantity of drugs, and the number of overdoses? b. Suppose the CIA and DEA assassinate the kingpin and their lieutenants start two new drug traffic organizations (DTOs). These two firms act as Cournot duopolies. What is the new market price and quantity sold? What is the new number of overdoses? c. Now suppose that the duopolies start a drug war with each other. For each 10 units they sell, 2 people are murdered. What is the total number of murders? What is the total number of deaths (murders and overdoses) and how does this compare to when a monopoly (single kingpin) was present? d. What does this tell you about the effectiveness of the war on drugs? 1 2. a. Define and give 2 examples of systemic violence and psycho-pharmacological violence? b. Define statistical and taste based discrimination. Why do economists care about distinguishing between them? c. Why would more guns potentially increase crime? Why could they potentially increase crime? 3. Suppose a state (OR) passes a law allowing for individuals to stand their ground You have the following data on homicides in OR and WA, a neighboring state which didn’t pass the marijuana law. The parentheses indicate the estimated standard error on usage (squaring the standard error generates the estimated variance). OR WA Before Law 5.3 7.0 (0.7) (0.8) After Law 8.2 7.2 (0.6) (0.9) a. If you were to compare OR’s homicides before and after the law was passed, what would be the estimated increase homicides potentially attributed to the change in the law? Is the change statistically different from zero (based on a 95 percent confidence interval)? b. What might be a shortcoming of the simple before/after comparison of OR’s law change? 2 c. Construct the difference-in-difference estimate using WA as neighboring state adjust for other potential underlying trends. Is the estimated effect of the law on homicides still statistically significant at the 95 percent level? What are we assuming about WA to make it valid comparison group? d. Write an econometric equation which has only dummy variables that one could use for estimating the difference-in-difference estimate you constructed in part c. 4. Consider now using information on two types of crimes. Underage drinking and selling alcohol to minors. a. If you consider running the regression crimei = β0 + β1 ∗ over21i + f(agei) + ui , discuss why using a the 21 drinking age can be thought of as a natural experiment of sorts. b. Utilize the data crime21.dta. Estimate regression discontinuity models with the crimes as the left hand side variables where you control for age relative to 21 ‘age’, a dummy variable for being over 21 ‘over21’, and the interaction between age and being over 21 ‘ageint’. I rescaled the age so it is already relative to turning 21 (otherwise you would need to do that). Estimate a similar same regression also including a squared term for age, and that squared term interacted with over21 (you need to create those variables. Based on those linear and quadratic regression discontinuity models, how does underage drinking (‘underage’) and selling alcohol to minors (‘minor’) change at 21. Assume age follows a quadratic pattern which might change in slope with legal access to alcohol at age 21? c. Using the ‘scatter’ command, type ‘scatter minor age, xline(0)’, and ‘scatter underage age, xline(0)’. Do the graphs reveal changes in criminality at age 21? 3 5. Utilize data from the kansas dust in the wind.dta You will be comparing crime rates in KS and OK. As we discussed in class, Kansas became wet (meaning alcohol could be consumed in public restaurants and bars) in 1986. I want you to analyze this change and estimate a difference in difference model. a. First, I want to set the data as a time series. Type ‘tsset KS year’. Now type ‘twoway (tsline crime if KS==1) (tsline crime if KS==0)’ to create a time series plot with both KS’s and OK’s time series. Based on this, does OK look like a reasonable control group (hint: think about the key identifying assumption for a difference in difference model). b. What is the mean of KS before the law change and after the law change (hint: use the ‘summarize’ command along with ‘if’, or for example ‘summarize crime if KS==1 & year<1986’). What about the mean of OK before and after the law change? Based on those means, what is the difference-in-difference estimate of the law change? c. Create a dummy variable for after the law change. Also create an interaction dummy between KS and the after dummy. Estimate a regression including the dummy variable for KS, a dummy variable after the law, and the interaction dummy. The interaction dummy is another to estimate the diff-in-diff. How does it compare to your answer in part b? 4

Leave a Reply

Your email address will not be published. Required fields are marked *

HW # 3 Economics of Crime 1. a. Suppose in Mexico there is a single drug kingpin who supplies all of the drugs. The kingpin’s marginal cost of drugs of 20 and the demand curve in the united states is 220-Q. Suppose 1 person overdoses for every 10 units of drugs that are consumed. What is the price and quantity of drugs, and the number of overdoses? b. Suppose the CIA and DEA assassinate the kingpin and their lieutenants start two new drug traffic organizations (DTOs). These two firms act as Cournot duopolies. What is the new market price and quantity sold? What is the new number of overdoses? c. Now suppose that the duopolies start a drug war with each other. For each 10 units they sell, 2 people are murdered. What is the total number of murders? What is the total number of deaths (murders and overdoses) and how does this compare to when a monopoly (single kingpin) was present? d. What does this tell you about the effectiveness of the war on drugs? 1 2. a. Define and give 2 examples of systemic violence and psycho-pharmacological violence? b. Define statistical and taste based discrimination. Why do economists care about distinguishing between them? c. Why would more guns potentially increase crime? Why could they potentially increase crime? 3. Suppose a state (OR) passes a law allowing for individuals to stand their ground You have the following data on homicides in OR and WA, a neighboring state which didn’t pass the marijuana law. The parentheses indicate the estimated standard error on usage (squaring the standard error generates the estimated variance). OR WA Before Law 5.3 7.0 (0.7) (0.8) After Law 8.2 7.2 (0.6) (0.9) a. If you were to compare OR’s homicides before and after the law was passed, what would be the estimated increase homicides potentially attributed to the change in the law? Is the change statistically different from zero (based on a 95 percent confidence interval)? b. What might be a shortcoming of the simple before/after comparison of OR’s law change? 2 c. Construct the difference-in-difference estimate using WA as neighboring state adjust for other potential underlying trends. Is the estimated effect of the law on homicides still statistically significant at the 95 percent level? What are we assuming about WA to make it valid comparison group? d. Write an econometric equation which has only dummy variables that one could use for estimating the difference-in-difference estimate you constructed in part c. 4. Consider now using information on two types of crimes. Underage drinking and selling alcohol to minors. a. If you consider running the regression crimei = β0 + β1 ∗ over21i + f(agei) + ui , discuss why using a the 21 drinking age can be thought of as a natural experiment of sorts. b. Utilize the data crime21.dta. Estimate regression discontinuity models with the crimes as the left hand side variables where you control for age relative to 21 ‘age’, a dummy variable for being over 21 ‘over21’, and the interaction between age and being over 21 ‘ageint’. I rescaled the age so it is already relative to turning 21 (otherwise you would need to do that). Estimate a similar same regression also including a squared term for age, and that squared term interacted with over21 (you need to create those variables. Based on those linear and quadratic regression discontinuity models, how does underage drinking (‘underage’) and selling alcohol to minors (‘minor’) change at 21. Assume age follows a quadratic pattern which might change in slope with legal access to alcohol at age 21? c. Using the ‘scatter’ command, type ‘scatter minor age, xline(0)’, and ‘scatter underage age, xline(0)’. Do the graphs reveal changes in criminality at age 21? 3 5. Utilize data from the kansas dust in the wind.dta You will be comparing crime rates in KS and OK. As we discussed in class, Kansas became wet (meaning alcohol could be consumed in public restaurants and bars) in 1986. I want you to analyze this change and estimate a difference in difference model. a. First, I want to set the data as a time series. Type ‘tsset KS year’. Now type ‘twoway (tsline crime if KS==1) (tsline crime if KS==0)’ to create a time series plot with both KS’s and OK’s time series. Based on this, does OK look like a reasonable control group (hint: think about the key identifying assumption for a difference in difference model). b. What is the mean of KS before the law change and after the law change (hint: use the ‘summarize’ command along with ‘if’, or for example ‘summarize crime if KS==1 & year<1986’). What about the mean of OK before and after the law change? Based on those means, what is the difference-in-difference estimate of the law change? c. Create a dummy variable for after the law change. Also create an interaction dummy between KS and the after dummy. Estimate a regression including the dummy variable for KS, a dummy variable after the law, and the interaction dummy. The interaction dummy is another to estimate the diff-in-diff. How does it compare to your answer in part b? 4

Leave a Reply

Your email address will not be published. Required fields are marked *