Homework answers / question archive / PAPER-1 DOES SPENDING MORE ON TOBACCO CONTROL PROGRAMS MAKE ECONOMIC SENSE? AN INCREMENTAL BENEFIT-COST ANALYSIS USING PANEL DATA Students must be able to answer the following set of questions to be able to complete the project (Component 1; Paper 1)

PAPER-1 DOES SPENDING MORE ON TOBACCO CONTROL PROGRAMS MAKE ECONOMIC SENSE? AN INCREMENTAL BENEFIT-COST ANALYSIS USING PANEL DATA Students must be able to answer the following set of questions to be able to complete the project (Component 1; Paper 1)

Statistics

Share With

PAPER-1 DOES SPENDING MORE ON TOBACCO CONTROL PROGRAMS MAKE ECONOMIC SENSE? AN INCREMENTAL BENEFIT-COST ANALYSIS USING PANEL DATA Students must be able to answer the following set of questions to be able to complete the project (Component 1; Paper 1). Make sure your answers are focused around the suggested bullet points provided below. EXPLAIN AS CONVINCINGLY AS POSSIBLE, BUT IN YOUR OWN WORDs, NOT BY COPYING AND PASTING FROM THE TOBACCO PAPER. Submit the Report on Section 6 of the iLearn site created for Tobacco Project submition in one single PDF file. 1. Describe in your own words the important policy questions addressed in the paper. Why do you think there is an economic answer to the policy questions raised here? In the above discussion, motivate the reader by drawing attention to Figure -1 of the paper. • Figure-1 here. • States unusually low spending issues. • Economic benefit-cost analysis can address the issue. 2. There are two aspects that make the cigarette market distinctly different from other markets, namely (i) addictive product and (ii) oligopolistic. Explain these two concepts in the context of the policy issues raised in the paper. • Addictive product and the role of time in states’ performances to curb smoking. • Nature of oligopoly to be explained within context. 3. Explain how the paper has econometrically modeled the addictive nature of the product? • Elapsed time effect to be explained. 4. Explain how the paper has econometrically modeled the oligopolistic nature of the market? • Price endogeneity to be explained. 5. Econometrically, what feature of the market requires you to consider the panel data models, such as FE, RE? • Talk about unobserved factors typical of states that do not change over time but affect the model. 6. Discuss why tobacco prevention and control funding is regarded as “endogenous” in the models. What are the instruments used to correct for the endogeneity of this variable? • Discuss examples of unobserved factors that can, potentially, impact the amount of Control Funding dollars in states. • Discuss instrumental variables used/ 7. Estimate each of the three models, under price-based and tax-based specifications. Your estimation should correct for the possible heterscadasticity as is done in the paper. • Attach six STATA estimation output in Appendix. 8. Take a close look at the estimates of the error serial correlation (“rho” in the STATA output that you get when you run FE and RE models in STATA) in the case of the FE and the RE models. What can you infer from this serial correlation regarding (i) the performance of the OLS vis-à-vis FE/RE models, (ii) performance of the FE over the RE Models? NOTE THAT THE HIGHER THE VALUE OF THE SERIAL CORRELATION THE HIGHER THE VALUE OF LAMBDA, DISCUSSED IN THE PANEL DATA FIXED (FE) AND RANDOM EFFECT (RE) MODELS DISCUSSED IN THE HANDOUT POSTED ON ILEARN. • Discuss possible implications of high serial correlations. 9. Explain clearly if the variable representing “price of a substitute good” was not included in the model, (i) what kind of ECONOMETRIC problems you would have encountered? (ii) what kind of ECONOMIC problems you would have encountered? Discuss the own-price, cross-price and the overall price elasticity estimates focusing on the price-based FE model. • Discuss econometric problems of mis-specified models. • Discuss economic problem of incorrect price elasticity. 10. Use the tax-based regressions to replicate Figure 2 in the paper and draw attention to this figure to explain clearly the “contemporaneous effects”, “elapsed time effects”, and the overall effects of the control funding programs on cigarette demand. Use estimated effect sizes in the RE model to make your interpretations easily understandable. • Figure 2 here. • Discuss why contemporaneous effect is positive (cigarette demand going up), but elapsed time effect is negative and how it impacts cigarette demand over time. 11. Use the tax-based regressions to replicate Figure 3 in the paper and provide a detailed comment on the time path of the projected per-capita cigarette demand and the robustness of the three different models. • Figure 3 here. • Comment on Figure 3. 12. Carry out the benefit-cost analysis of the CDC recommended funding presented in Section 7 following exactly the procedure described therin. You will have to use EXCEL extensively to carry out this part of the project. What I need from you are: (i) reproduction of Tables 4 and 5; (ii) an analysis of the results. • Reproduce your own Table 4 here. • Reproduce your own Table 5 here. • Analyze the results in a convincing way. APPENDIX • Attach a copy of the Benefit-Cost Simulation Calculations from EXCEL Sheet 3. • Attach all the STATA estimations.
Paper 1 Coding Instructions in STATA If you have created logarithms of all the required variables, you are ready to start the Step 1 below. Make sure that in your data set you haven’t yet created (i) the lag variable for “totfundreal” and (ii) the interaction variable “t * totfundreal”. If you did, please drop that lagged variable and then go to Step 1 below. Step 1: Next command is to create a variable that is a lagged variable for “totfundreal”. gen totfundreal1=totfundreal[_n-1] The above command will generate one missing value which is the first row (year 1991 for Alabama) of the 850 rows you have on your data set. The above command will also create the 1991 (i.e., t = 0) value for each state that is the 2007 value of the previous state, which is UNACCEPTABLE. Please confirm this by looking at the newly created column at the end of the data set. Check this by clicking the “Data Browser”. replace totfundreal1=totfundreal if t==0 The above command will replace the first value (i.e., t = 0) of totfundreal1 for each state by the corresponding value of the original variable “totfundreal”. gen timefund = t*totfundreal The above command will create the interaction variable, as discussed in class. The following command is the first stage of the 2SLS technique regress totfundreal airscore youthscore totfundreal1 Now predict in the following command to get the predicted values of “totfundreal”. predict totfundrealhat Now create the new predicted value of the time interaction variable using the above predicted values using the following command gen timefundhat=t* totfundrealhat In the followingOLS regression, you will replace the following two engogenous variables: totfundreal to be replaced by totfundrealhat timefund to be replaced by timefundhat 2SLS –Second State Estimation of the Six Econometric Models Price-Based Model: Pooled Model: regress lqpc d2 d3 d4 d5 d6 d7 d8 d9 d10 d11 d12 d13 d14 d15 d16 d17 lpreal lpricesub lpopul lpcpdireal totfundrealhat timefundhat pcgr pcnt1524 pcnt25 unemprate, robust Panel Data Fixed Effects Model (FE): xtreg lqpc d2 d3 d4 d5 d6 d7 d8 d9 d10 d11 d12 d13 d14 d15 d16 d17 lpreal lpricesub lpopul lpcpdireal totfundrealhat timefundhat pcgr pcnt1524 pcnt25 unemprate, fe robust Panel Data Random Effect Model (RE) xtreg lqpc d2 d3 d4 d5 d6 d7 d8 d9 d10 d11 d12 d13 d14 d15 d16 d17 lpreal lpricesub lpopul lpcpdireal totfundrealhat timefundhat pcgr pcnt1524 pcnt25 unemprate, re robust Tax-based Model: Pooled Model: regress lqpc d2 d3 d4 d5 d6 d7 d8 d9 d10 d11 d12 d13 d14 d15 d16 d17 ltottaxreal ltaxsub lpopul lpcpdireal totfundrealhat timefundhat pcgr pcnt1524 pcnt25 unemprate, robust Panel Data Fixed Effects Model (FE): xtreg lqpc d2 d3 d4 d5 d6 d7 d8 d9 d10 d11 d12 d13 d14 d15 d16 d17 ltottaxreal ltaxsub lpopul lpcpdireal totfundrealhat timefundhat pcgr pcnt1524 pcnt25 unemprate, fe robust Panel Data Random Effect Model (RE) xtreg lqpc d2 d3 d4 d5 d6 d7 d8 d9 d10 d11 d12 d13 d14 d15 d16 d17 ltottaxreal ltaxsub lpopul lpcpdireal totfundrealhat timefundhat pcgr pcnt1524 pcnt25 unemprate, re robust