Question 1

alt

Question 2

Under the assumption that there are no unobserved cofounders, the relationship between policy and unobserved confounders will equal 0. This implies that the Beta 1 of our policy variable is an unbiased estimate and causal impact.

Therefore our linear regression model should look like this; Outcome = B0 + B1*policy + XB + Ei

Question 3

To identify a causal effect it would require a set of natural experiments that identifies exogenous variations in both the treatment variable and the outcome variable. After that, we should compare the effects of treatments and outcomes

Question 4

Instrumental variables (IV) are used to establish causal relationships when controlled experiments are not feasible or when correlation between the explanatory variables and error term is suspected. An IV has two properties:

The exclusion restriction demonstrates that the instrumental variable is not correlated with any other determinants of the dependant variable, only the treatment variable. Instrument, Z, should not affect the outcome, Y, when X is held constant.

Natural experiments are historical events that provide observable, quasi-random variation to the treatment variable with a plausible identifying assumption. In macroeconomics, due to the difficulty in implementing standard field experiments on the entire population, a natural experiment is the most feasible comparison we can make to an instrumental variable in order to examine the impact of policy to the outcome observed.

Question 5

Lydon Gui - Permanant Income Hypothesis

Paycheque Receipt and the Timing of Consumption by Stephens, Melvin, J. (2006, 07).

Summary

Stephens examines the consumption response to monthly paycheck receipts. Since the amount and arrival date of paychecks are known in advance, the receipt of a paycheck does not coincide with the receipt of new information.

What data do they use?

Using data from the UK’s Family Expenditure Survey, this paper finds that household consumption is excessively sensitive to paycheck receipt. The results cannot be explained by any underlying monthly expenditure fluctuations common to all households. The presence of liquidity constraints as measured by wealth and age can account for the excess sensitivity results.

What is the natural experiment?

It would be the regular monthly paychecks receipts and if regular household income increases or decreases as a result of this. Its PIH rejects the null at Beta = 0.

What are their exclusion restrictions?

Pay frequency information is not collected for self-employed workers so they are excluded from the analysis. Some monthly paid households are excluded if one of their diary weeks does not fall within one of the four groupings described in the next section.

Do you believe their exclusion restrictions?

Given the amount of reliability and precision Stephen’s wanted to put in his findings, I believe it was entirely reasonable as it typically involves frequencies higher than the monthly one and analyse whether consumption is stable or decreases over the payment cycle.

What are their findings?

Under Permanent Income Hypothesis, instant consumption increases by 5 percent in week when households receive monthly paychecks household consumption should not respond to paycheck arrival.

Aidan Hanna - Fiscal Multiplier

The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program by Mian & Sufi (2012)

What data do they use?

Their data set includes 957 U.S. metropolitan or micropolitan statistical areas (CBSAs) for which we have a number of economic outcomes available from January 2004 through June 2010. Data on monthly purchases of new cars and new light trucks are from R.L. Polk, a company specializing in automotive intelligence which closely tracks the aggregate census retail sales data on new motor vehicles.

What is the natural experiment?

The natural experiment this paper used was Obama’s 2009 Cars Allowance Rebate System (CARS) or otherwise known as ‘cash for clunkers’, where the authors evaluated the impact it had on short-medium term auto purchases. The program is representative of a large number of fiscal stimulus programs that attempt to induce consumption or investment with cash subsidies.

What are their exclusion restrictions?

Many spikes in auto purchases are due to incentive programs by car dealers and manufacturers, and it is therefore impossible to assess the impact of CARS using aggregate data alone. The authors approach is to form counterfactual outcomes based on cross-sectional variation across U.S. cities in their exposure to the CARS program.

Do you believe their exclusion restrictions?

The advantages of this approach are clear. By forming a control group that we can follow over time, we can more realistically assess the counterfactual outcome for cities with a large number of clunkers. The control group is allowed to respond to aggregate shocks such as the March 2010 incentive.

What are their findings?

Their findings show evidence of a large effect of CARS that very quickly reverses. The research shows a slightly larger drop in employment growth for high CARS exposure cities, with a sharp relative increase in the second quarter of 2009. After the implementation of CARS, there was a slow-down in job losses in both the high and low CARS exposure cities. There is no evidence of a dramatic structural break in employment growth in high and low CARS exposure cities. Household defaults and declines in house prices were much more severe in low CARS exposure counties prior to CARS. Once again, there does not appear to be any major structural break in the patterns after the CARS program. If anything, there is some evidence that the increase in household defaults moderated more in high CARS exposure cities.