Question 1

(a)

The equation's Edu_i is 0.06. This coefficient is the natural logarithm change in hourly wage for a one-unit change in years of schooling, holding all other parameters constant. Each year of study should increase hourly earnings by 6%. Education doesn't guarantee higher pay. Selection bias, when higher-paid people study more, can affect education and pay. To determine how schooling influences earnings, selection bias must be addressed.

(b)

We can test the null hypothesis that the coeficient on White ei is equal to 0.1 against the alternative hypothesis that is not equal to 0.1. The test statistic is calculated as follows:

t = (b - bθ) / SE

Where b is the estimated coefficient, which is 0.08, bθ is the hypothesized value which is 0.1 and SE is the standard error. The standard error can be calculated from the t-ratio.

SE = b / t-ratio

If the calculated t-value falls within the critical region determined by chosen significance level & degrees of freedom, we reject the null hypothesis. To calculate the required value, we need to know the standard error of the estimated coefficient.

SE = b / t-ratio = 0.08 / 2.43 = 0.033

Now, we can calculate the test statistic.

t = (b - bθ) / SE = (0.08 - 0.1) / 0.33 = -0.02 / 0.033 = -0.606

The degrees of freedom for this test are the number of observations minus the number of

parameters in the model, excluding the intercept. In this case, the number of observations is 50,000,

and the number of parameters is 5 (including the intercept). So, the degrees of freedom are:

df =50,000 -5 -1

df =49994

Using a t-distribution table or statistical software, we can find the critical t-value for the chosen

significance level (e.g., 0.05) and the degrees of freedom. For a two-tailed test at a significance

level of 0.05, the critical t-value is approximately 1.96. Since the calculated t-value (-0.606) is less

than the critical t-value (1.96), we fail to reject the null hypothesis that the coefficient on White e_i is

equal to 0.1.

(c)

The topic of gender discrimination in the labor market has been extensively researched and debated

in academic literature. The study examines gender discrimination in the Indian labor market using

a theoretical framework that incorporates both taste and statistical discrimination. The authors find

that gender gaps in wages reflect gross inequality and discrimination, which exists across various

locations, sectors, and types of work. They identify that women are more likely to be excluded

from the more prestigious and better-paying occupations, contributing to the wage gap. The study

concludes that gender discrimination is a significant factor in the labor market, particularly in the

Indian context (Gupta and Kothe, 2021). The main findings of this paper are:

• Gender gaps in wages reflect gross inequality and discrimination.

• Women are more likely to be excluded from prestigious and better-paying occupations.

• Gender discrimination is a significant factor in the Indian labor market.

This meta-analysis of field experiments investigates gender discrimination in the labor markets hiring process. The study combines the results of 19 correspondence tests from 12 different countries across the world. The findings show that in almost all experiments, there is evidence of discrimination against female applicants. The study also suggests that providing more information in job applications can reduce the level of discrimination. The authors conclude that gender

discrimination persists in the labor market, despite efforts to address it through policy and legislation (Sousa, Ricardo Spindola Diniz and de, 2019). The main findings of this paper are:

• In almost all experiments, there is evidence of discrimination against female applicants.

• Providing more information in job applications can reduce the level of discrimination.

• Gender discrimination persists in the labor market, despite efforts to address it through policy and legislation.

(d)

Scenario I

To investigate whether the return to education depends on the level of education, we need to modify the regression equation to include an interaction term between education and a measure of the level of education.

ln w_i = 3.45 + 0.06 Edu_i + 0.23 Exp_i + 0.08 White_i + 0.11 Male_i + (0.01 Edu_i × Edu Level_i) + ε_i

Here, Edu Level_i is a new variable that represents the level of education, such as high school, college, or graduate degree. The interaction term Edu_i × Edu Level_i captures the effect of education on wages at different levels of education. To test whether the return to education depends on the level of education, we would perform a test for the significance of the interaction term.

Scenario II

To investigate whether the return to education is higher for white individuals than non-white individuals, we need to modify the regression equation to include an interaction term between education and a dummy variable for race. This interaction term will allow us to examine whether the relationship between education and wages varies across different racial groups. The modified regression equation would be:

ln w_i = 3.45 + 0.06 Edu_i + 0.23 Exp_i + 0.08 White_i + 0.11 Male_i + (0.02 Edu_i × White_i) + ε_i

Here, White_i is a dummy variable that takes the value 1 if the individual is white and 0 otherwise. The interaction term Edu_i × White_i captures the effect of education on wages for white individuals compared to non-white individuals.

Scenario III

To find the wage elasticity with respect to education, we need to modify the regression equation to include the logarithm of education as an independent variable. This will allow us to estimate the effect of education on wages in terms of the percentage change in wages for a one-unit change in education. The modified regression equation would be:

ln w_i = 3.45 + 0.06 lnEdu_i + .23 Exp_i + 0.08 White_i + 0.11 Male_i + ε_i

To estimate the wage elasticity, we would simply multiply the coefficient on lnEdu_i by 100 to convert it to a percentage change. For example, if the coefficient on lnEdu_i is 0.06, then a one-unit change in education would result in a 6% change in wages.

Tests to be carried out.

For scenarios i and ii, we would perform tests for the significance of the interaction terms to determine whether the relationships between education and wages vary across different levels of education or racial groups. These tests would involve calculating the t-statistics for the interaction terms and comparing them to critical t-values at the desired significance level. For scenario iii, we would not need to perform any additional tests, as the wage elasticity with respect to education is directly estimated from the regression coefficient on lnEdu_i.

Question 2

To estimate the effect of living in city A compared to city B on the relative demand for electric vehicles (EVs) in 2010, we can use the data provided. The difference in EV sales between the two cities in 2010 is:

65,918(City A) - 44,823(City B) = 21,905

This difference represents the additional EVs sold in city A compared to city B in 2010. To normalize this difference for the total number of petrol vehicles sold in each city, we can calculate the ratio of EVs to petrol vehicles for each city:

City A = 65,918

156,125 = 0.42

City B = 44,823

127,121 = 0.35

The difference of the ratios is:

0.42 - 0.35 = 0.07

City A had 0.07 EV demand growth in 2010 compared to city B. This gap assumes that just the city of residency impacts EV demand, and that demographics, income, and environmental consciousness are the same in both cities. Assuming the 2009 city A EV experiment did not affect 2010 sales,

(b)

Statistics suggest electric car subsidies fail. City A's 2009 EV experiment may explain its 2010 EV sales gap with B. The difference doesn't mean city A's EV incentives increased sales.

Analysis assumes only residency city affects EV demand and both cities have similar demographics, income, and environmental consciousness. Without charging infrastructure, public knowledge, tax refunds, or road toll exemptions, EV demand may shift. The poll ignores city economies and fuel car prices, which may affect EV sales. Monitor errors may overstate subsidies' EV demand impact. Subsidies' impact on EV demand is unclear. It only lists city-level EV sales, which may not include subsidies. City EV trials may increase demand without subsidies.

Difference in EV sales do not indicate that subsidies raised city A EV demand. Due to erroneous assumptions and unregulated EV demand determinants, subsidies are hard to link to sales growth.

Neither unsubsidized city nor control group is assessed. Since subsidies ignore other EV demand variables, they are hard to distinguish. City A EV demand may have increased due to fuel car prices, economic conditions, and environmental awareness, making subsidies difficult to identify.

Before 2009 subsidies, the study excluded city A EV demand. While subsidies may have raised demand, there is no benchmark. This study does not compare EV demand in city A with city B,

(c)

which did not get subsidies. No direct comparison between the two cities makes it hard to identify if city A's demand increase is due to subsidies or a trend.

Fuel car costs, economic conditions, and environmental consciousness may alter EV demand, but the prediction ignores them. EV demand in city A may have changed, making subsidies hard to trace. A two-year analysis may miss subsidies' long-term EV demand implications. After incentives stopped, City A's EV demand may have increased or plateaued. The study lacks a control group, pre-subsidy data, and city B comparison, ignoring other EV demand drivers. These limits make linking subsidies to city A's EV demand growth problematic.

(d)

The two tables lack data to assess how subsidies effect electric vehicle demand. The tables illustrate electric car manufacturing, government subsidies, and market trends research. Insufficient data exist to determine subsidy demand.

We could use regression analysis to assess the incentives' impact on electric vehicle demand if the table data were merged. Regression analysis finds correlations between variables. Electric car demand and subsidies are variables.

Table 1: Subsidy Data

• Subsidy Amount (SA)

• Electric Vehicle Sales (EVS)

Table 2: Demand Data

• Electric Vehicle Demand (EVD)

• Petrol Vehicle Demand (PVD)

To assess the influence of subsidies on electric car demand, we must integrate these statistics by correlating subsidy amount to demand. Create a new variable to reflect the subsidy per vehicle and

use it in a regression analysis to assess the subsidy's effect on electric vehicle demand.

Subsidy Amount (SA)

Electric Vehicle Demand (EVD)

Regression Analysis

Estimated Effect of Subsidies on Demand

(e)

Part (d) data don't show EV subsidies increased demand. Part (d) predicts subsidy-driven EV demand using regression. This model overlooks fuel car pricing, economic conditions, and environmental consciousness that may affect EV demand.

According to this idea, subsidies increased EV sales. The regression model revealed government subsidies boosted EV market share 0.5% every 1%. This suggests subsidies increased EV demand by boosting their market share over combustion vehicles. This simplistic model ignores EV demand. Thus, the results estimate the subsidy policy's impact on EV demand, not a correlation.

(f)

Control groups and unsubsidized cities are excluded. Since subsidies ignore other EV demand variables, they are hard to distinguish. Pre-2009 city A EV subsidies are excluded. While subsidies may have raised demand, there is no benchmark. Part (d) does not compare subsidized and unsubsidized EV demand. No direct comparison between the two cities makes it hard to identify if city A's demand increase is due to subsidies or a trend. EV demand is affected by fuel car prices, economic conditions, and environmental consciousness, which Part (d) ignores. EV demand in

city A may have changed, making subsidies hard to trace. Item (d)'s two-year method may overlook subsidies' long-term EV demand effects. City A EV demand may have increased or stabilized when incentives expired.

Question 3

(a)

Income diminishes marginal utility, as shown by the concave projected labor supply curve. Note the negative coefficient of the squared wage factor (β_2) in the regression equation. The concavity of the graph illustrates that higher wages affect labor supply less.

(b)

ε_w = ∂h/∂w × w/h = (β_1 + 2β_2 w)/h

This elasticity depends on the wage level w, which is a function of the taxi driver's after-tax wage and other income. The elasticity is not constant and changes as the wage level changes.

(c)

The coefficients, average after-tax wage, and other income are used to determine labor supply elasticity to wage for a cab driver with kids in component (b). Children positively affect hours worked (β_4). Children make cabbies work harder. The constant part in the elasticity calculation shows that having kids does not change the taxi driver's labor supply elasticity to wage.

(d)

Tax increases lower taxi drivers' post-tax compensation. The regression equation (β_1) indicates that wage increases lead to more hours worked due to the negative wage component coefficient.

(e)

The model ignores economic factors, other jobs, and taxi drivers' education and skills, which could affect income and hours worked. These factors may explain wage-induced labor supply shifts. Work hours may affect wage, making it endogenous. A few taxi drivers receive extra for working overtime. Salary estimates on hours worked may be affected by not employing instrumental factors or panel data analysis.

(f)

This study uses only employed workers' data. This reduces the estimates' usefulness for measuring unemployed people's work preferences. Unemployed people may want different employment and react differently to income changes. Therefore, this study's estimates should not be used to generalize unemployed people's choices.