1 Draw a causal diagram, as in class. Include an outcome variable, a treatment variable, an observed confounding variable (control variable), an unobserved confounding variable, an instrumental variable, and a bad control variable.

2 If I want to estimate the (linear) causal relationship between the treatment variable and outcome variable under the assumption that there are no unobserved confounders, what linear regression model should I estimate?

Unobserved cofounders must be non-existent if treatment variable BI has a specific causal effect on outcome. To determine the causal relationship an Ordinary Least Squares regression is the model suitable to serve the purpose. This is provided there are nil unobserved confounders.

3 Suppose I am concerned that there are unobserved confounders that affect both the treatement and the outcome, and as such I cannot assume that the relationship between them is causal. How should I go about identifying the causal effect?

In order to identify the causal effect, there need to be control group to see is there any evidence of the causal effect. If there is a difference in between the control group and the expose group, then it will suggest that the changes is cause by the causal effect. Some example of causal effect includes the seasonality effect where an increase in wage in a particular season will influence the month following spending outlook. This is most prominent example from the paper is the income change at the year-end and how it influence the average spending spree over the year.

4 Describe the two properties that an instrumental variable must have. What is the exclusion restriction? What is the relationship between an instrumental variable and a natural experiment?

The error terms in the explanatory equation cannot be correlated with the the instrument.

The instrument must be correlated with the endogenous explanatory variables, When the treatment under consideration is unchanged, instrumental variables should not affect the outcome. ε (including unobserved exogenous factors) and the instrumental variable must be independent when the treatment under consideration is constant.

The exclusion restriction means the instrumental variable should not affect the outcome when the endogenous variable is held constant The instrument affects outcome only through its affect on the treatment would be the exclusion restriction.

In order To establish a specific causal relationship Instrumental variables are utilised to analyze natural experiments, which are assigned randomly

5 - Individual Questions

Hus Augustine – Paper: Kuhn, et al. 2011- The Effects of Lottery Prizes on Winners and Their Neighbors.

Describe the Natural Experiment.

Kuhn et all uses the winners and non-winners of the Dutch postcode lottery to analyze their consumer behavior after the win relative to the non-winners. The lottery would give a one of cash prize of 12,500 EUR, which is roughly equivalent to approximately 8 months of average income and 1 winner will receive a BMW motor vehicle. The objective of this paper is to test out Milton Friedman’s Permanent Income Hypothesis by exploiting the lottery winders and non-winners. Kuhn et al tests this theory by measuring their consumption patterns in terms of spending on nondurable and durable good and services.

What data do they use?

A weekly lottery in Denmark, allocated a lottery price to randomly selected postcode with 19 households on average. Participants receive 12500 EUR. Kuhn, Peter, Peter Kooreman, Adriaan Soetevent, and Arie Kapteyn used the data is gathered from this lottery known as the Dutch Postcode Lottery.

What is the natural experiment?

Kuhn, Peter, Peter Kooreman, Adriaan Soetevent, and Arie Kapteyn tested the notion of the permanent income hypothesis by using the unbiased distribution of wealth from the lottery. The Permanent Income hypothesis is the theory that agents do not change their consumption behavior on predictable or expected changes in income. According to Fuchs-Schundeln et al. Lottery wins are historical episodes that clearly identify unexpected large temporary shocks. Hence they can be used as natural experiments. Kuhn et al, sues the data from the Dutch postcode lottery to compare winners and non-winners.

What are their exclusion restrictions?

The exclusion restriction in this model is that the participation in the postcode lottery only affects consumption in the unexpected win of the lottery. Participation in the lottery alone has no affect on the entrant’s consumption.

Do you believe their exclusion restrictions?

Yes, the exclusion restriction is plausible as a normal economic agent would not make consumption decisions based on a lottery win event that is unlikely to be won. The demographic of the participants in entering into the lottery could be affect the out come of the model. In other words the reasons why one would enter into such lottery and the behavioral patterns of people that buy into the lottery could bias the result of the research.

What are their findings?

Kuhn, Peter, Peter Kooreman, Adriaan Soetevent, and Arie Kapteyn conclude that the Permanent Income Hypothesis largely holds. That is non-consumption of non-durables do not increase significantly amount lottery winners. Also, consumption of durables experiences somewhat of an increase.

+*****************************************

Zharif Zaini – Paper: Agrawal, S & Qian, W 2014, ‘Consumption and Debt Response to Unanticipated Income Shocks: Evidence from a Natural Experiment in Singapore’.

Describe the Natural Experiment.

This paper is written to study the policy experiment by the Singaporean government on the income shock that it has on the consumer. The two experiments that the researchers have used to examine the permanent income hypothesis is Singapore Growth Dividends in 2011 following government announcement of the Growth Dividend Program and the exact payout that was received 2 months following the announcement. The time gap in between allows the researcher to establish a control group that will indentify the existence of the causal effect. The natural experiment that was used to examine the fiscal multiplier is a distributed lag model using the announcement date of the Dividend Growth Program as the exogenous event and obtain the impulse response of credit card spending, debit card spending and credit card debt.

What data do they use?

. The data was acquired via credit card, debit card and bank checking account data from leading bank in Singapore from 2010 to 2012 and the sample population come from a random sample of all bank customers of leading Singaporean bank with 180,000 observation.

What is the natural experiment?

. The natural experience for this case is the announcement date of the Dividend Growth Program

What are their exclusion restrictions?

The exclusion restrictions for this case is the spending, debt and credit card usage

Do you believe their exclusion restrictions?

Yes

What are their findings?

The finding indicate that there is an increase in card spending by 8 cents per month for every dollar received, corresponding to a total increased of 80 cents in the 10 months period after the announcement. The trend continues for months after the accouchement and implementation and this satisfy the idea of permanent income hypothesis.

+*****************************************

Ali AlSawad -The Spending and Debt Response to Minimum Wage Hikes By Daniel Aaronson, Sumit Agarwal, and Eric French, 2012

The researcher use the “Minimum wage Hikes and Income increase” as an natural experiments to examine the permanent income hypothesis and estimate the magnitude, composition, distribution, and timing of the income, spending, and debt responses to minimum wage hikes among households with adult minimum wage workers. to examine the fiscal multiplier.

Data Used:

CEX interview survey (1982-2008); CPS (1980-2007) and SIPP (1983-2007) for data on income; proprietary dataset from national financial institution (1995-2008): credit card account and each credit card holder’s auto, home equity, mortgage, and credit card balance.

Natural Experiment:

Spending on nondurables and durables, including separate subcategories, and change in debt (total and subcategories).

Exclusion restrictions:

The majority of the spending response occurs in the form of durable goods and, in particular, new vehicles that are debt-financed. Consequently, the spending response is concentrated among a small number of households. And yes I believe of exclusion restrictions.

Finding:

Households where minimum wage labor is the source of at least 20 percent of household income spend 3.4 times the short-term increase in income, but mostly on cars.