Environmental and Natural Resource Economics
Homework Assignment:
Lead and the Environmental Kuznets Curve
Introduction
One of the key challenges with the topic of the Environmental Kuznets Curve is how to empirically estimate whether an EKC exists or not. This assessment attempts to delve more deeply into how to interpret empirical results that seek to estimate an EKC. The example used is from one of your course readings, a paper from Hilton and Levinson. These two authors have estimated equations to examine whether there appears to be an “Environmental Kuznets Curve (EKC)” between lead emissions and income for forty-eight countries over a twenty year time period from 1972 to 1992. These results are reported in the Journal of Environmental Economics (JEEM) in 1998 in Table II entitled, “Estimating the Environmental Kuznets Curve for Automotive Lead Emissions” (p. 134). (Tables for the T Distribution and the F Distribution to help answer the questions are provided as well).
(1) The first regression equation in Table II is estimated as quadratic in levels of income (G) with the dependent variable being total automotive lead emissions (in millions of grams). Regression coefficients, standard errors and R2 are shown below.
Lead = 25,149 – 1.58G + 0.0003G2 – 11,476D83 + 8.97D83G –
(9,435) (2.89) (0.0002) (3,623) (2.36)
0.0008D83G2
(0.0002)
+ 38.4(popdensity) – 308(year)
(14.5) (148)
where D83 represents 1 for the post-1983 period and zero for the pre-1983 period
R2 = 0.68
N = 528
F-Test of hypothesis that all time-interaction coefficients equal
to zero = 5.14
(a) What does the term “quadratic in levels of income” mean? What is the shape of the polynomial that is estimated in Equation One? What is the value of the leading coefficient of the income terms in the time period from 1983-1992, when taking into account that the equation has time interaction terms? Do these results seem consistent with the EKC theory?
(b) Which of the variables are statistically significant at a 5% significance level in the above Equation? (See the T Distribution Table provided).
(c) How do you undertake an F-test of the hypothesis that all time-interaction coefficients with respect to income are equal to zero? What are the results of the F-test of the hypothesis that all time-interaction coefficients with respect to income are equal to zero? How do you interpret these results at a 5% significance level? (See the F Distribution Table provided).
(2) Why do Hilton and Levinson use a variety of different function forms (e.g., quadratic in levels, cubic in levels, and quadratic in logs) to test whether there is an EKC for automotive lead emissions?
(3) What possible additional variables could you employ in the above equation (if any) to help determine if there is an EKC for lead over this time frame? Be specific in how you would add the variable (e.g., percentage of population with college degrees) to the regression equation.
(4) Lin and Liscow in a recent paper “Endogeneity in the Environmental Kuznet Curve: An Instrumental Variables Approach” (2012) (provided) addressed several possible endogeneity issues/problems with water pollutants and the EKC. To what degree would the same endogeneity issues arise with lead emissions and the EKC? How would you take their lessons for water pollutants and apply them when estimating an EKC for lead emissions?
Scoring
|
Rubric Area
|
Weight of Row
|
Unacceptable
|
Satisfactory
|
Exemplary
|
|
Environmental Kuznets Curve
|
5 points
|
(0 – 4.79 points)
Demonstrates an understanding and synthesis of some of the concepts about the Environmental Kuznets Curve.
|
(4.8 – 5.39)
Demonstrates an understanding and synthesis of most of the concepts about the Environmental Kuznets Curve.
|
(5.4 or greater)
Demonstrates an understanding and synthesis of all of the concepts about the Environmental Kuznets Curve.
|