Papers

Question 1

Adar and Huberman (2000) - freeriding

Answer 1

Found that on student site Gnutella:
1. More than 70% of users were only downloading files and never u[loading new files
2. The top 1% of peers provided 47% of the answers

Question 2

Marwell and Ames (1981) - freeriding

Answer 2

In a one shot game where participants allocate tokens between cash ($1 to self) and the public good ($0.5 to all players), average investment in public goods was about 50%

Question 3

Isaac et al. (1985) - freeriding

Answer 3

In a one shot game where participants allocate tokens between cash ($1 to self) and the public good ($0.5 to all players), subjects contribute 50% in lab setting BUT public good contributions fall as the game is repeated (people are willing to cooperate at first but get upset and retaliate if others take advantage of them)

Question 4

Andreoni and Payne (2003) - crowding out

Answer 4

1. Government spending crowds out private donations through two channels: willingness to donate (decreased income from increased taxes) and fundraising (divert less resources to)
2. $1,000 increase in government grant leads to a $265 reduction in private fundraising expenditure for the arts organisation and a $54 reduction for social service organisations
3. Crowd-out could be non-trivial if fundraising is a powerful source of generating private contributions

Question 5

Andreoni and Payne (2011) - crowding out

Answer 5

1. $1 increase in government grant to a charity leads to $0.56 less private contributions
2. 70% of this reduction is due to the fundraising channel (most of crowding out occurs through fundraising channel)

Question 6

Falk (2005) - charitable giving

Answer 6

1. Investigated the relevance of reciprocity in charitable giving by including no, small or large gifts in solicitation letters
2. Number of people who donated was higher under the small (14%) and large gift (21%) conditions that the no gift condition (12%)
3. Including a gift significantly increased the frequency of donations (larger gift = higher increase in frequency)

Question 7

Dellavigna et al. (2012) - charitable giving

Answer 7

1. Door-to-door fund-raiser randomized experiment where groups given no advance warning, a flyer that informs of time of solicitation, and a flyer that also has a ‘do not disturb’ checkbox
2. If altruistic, frequency of opening door and, thus, frequency of donating should go up
3. If social pressure, both should decrease as people don;t open the door so don’t let social pressure effect them
4. Finding 1: flyers lower frequency of opening doors (social pressure outweighs altruism)
5. Finding 2: flyer with opt-out box decreased giving significantly

Question 8

Hoxby (2000) - Tiebout hypothesis

Answer 8

1. Compares cities where there are few large school districts (e.g. Miami) with those where there are many small school districts (e.g. Boston)
2. Hypothesis less homogeneity within communities and heterogeneity across communities in Miami
3. Finding: cities with few districts have less sorting across neighbourhoods (well established)
4. Finding: cities with many districts have higher test scores on average (contested)

Question 9

Rothstein (2007) - Tiebout hypothesis

Answer 9

Critiqued finding of Hoxby (2000) that cities with many districts have higher test scores:
1. Claimed overstated results (Hoxby refused to show data set)
2. Reverse causality problem (questioned instrumental validity of number of streams as instrument for number of schools)

Question 10

Rhode and Strumpf (2003) - Tiebout hypothesis

Answer 10

1. First paper to provide evidence against Tiebout theorem
2. Hypothesis: as mobility costs decline, Tiebout sorting becomes more efficient so tax and spending policies and preferences for public goods should vary more across jurisdictions
3. Findings: heterogeneity measures (CV and DG) and dissimilarity index have decreased between communities (rather than expected increase) - i.e. communities becoming more similar in tax and spending policies and in racial and residential competition

Question 11

Hines and Thaler (1995) - crowding out

Answer 11

1. The crowd-out of state spending by federal spending if low and often close to zero (“flypaper effect”)
2. Control group not well justified - states that get grants are probably the ones that like spending it the most
3. Recent studies show flypaper effect in the short-run but substantial crowd-out from block grants in the long-run

Question 12

Hoxby (1999) - fiscal federalism

Answer 12

1. 1976: Supreme court rules that disparities in school spending above a threshold were unconstitutional
2. Wealthy districts forced to give all their tax revenue above the threshold to the common pool to fund poor districts
3. Local government had no incentive to raise taxes causing taxes and school funding fell in rich districts and property taxes were no longer able to fund schools adequately (deterioration of California public schools)
4. 15% reduction in spending/person in California vs. increase in spending in states with low tax prices (e.g. NJ and Pennsylvania)
5. Conclusion: extreme penalisation leads to quality decrease in public good

Question 13

Schonholzer (WP) - Tiebout hypothesis

Answer 13

1. Question: do people pay premium to live in certain neighbourhoods due to the quality of public goods or due to peers who co-reside?
2. Boundaries between Santiago City and Cupertino School overlap - exploit as access same city services but different schools
3. House price premium of $20,600 - $2,300 (13%) of this associated change is due to local government services

Question 14

Weitzman (1974) - pollution policy

Answer 14

1. When there is uncertainty about MB and/or MC, price and quantity policies may no longer be equivalent
2. In context of pollution abatement, when the MD curve is relatively flatter than the MC curve, always better to use price regulation
3. Vice versa for relatively steeper MD curve

Question 15

Greenstone (2003) - pollution policy

Answer 15

1. Investigate impact of Clean Air Act (1970) on TSPs pollution
2. Difference-in-difference between attainment and nonattainment counties (nonattainment needed to take action as over polluting)
3. Result: sudden massive reduction in TSPs pollution in nonattainment counties then tapers out
4. Result: infant mortality rate decreased relative to attainment counties but then reverted to previous difference

Question 16

Barreca et al. (2016) - pollution policy

Answer 16

1. Mortality effect of an extremely hot day declined by c.75% between 1931-1960-2004
2. Adoption of residential AC from 1960 explains the decline
3. Worldwide adoption of AC will speed up the rate of climate change (how to discount cost?)

Question 17

Guethin et al. (2022) - voting

Answer 17

1. Complete divergence between the effects of income and education on the vote
2. Higher educated voters less likely to vote for left until 1991 from which point, they were more likely to vote left than the less educated
3. Top income voters consistently less likely to vote for left than lower income counterparts
4. Inference: individuals may lie at different points of the voting spectrum on different issues, so appealing to one end of the spectrum may be vote-maximising

Question 18

Washington (2008) - voting

Answer 18

1. Compares legislators who have daughters to those with the same family size who have sons
2. Finds that daughters increase a congressman’s propensity to vote liberally (particularly on reproductive rights issues)
3. Supports notion that personal ideology matters: politicians are responding to their own experience, not just to the demands of voters (reflect personal preference rather than median voter in constituency)

Question 19

Lee et al. (2004) - voting

Answer 19

1. Looked at close elections for US representatives (constituency virtually the same whether a candidate gets 49.9% or 50.1%)
2. Median voter implies that a Democratic representative elected with 50.1% should vote similarly to a Republican representative elected with 50.1% of the votes
3. In reality, closely elected representatives vote very differently depending on party (voting on party lines not median voter issues)

Question 20

Fox et al. (2022) - commodity tax

Answer 20

1. In the US until 2018, out-of-state sellers with no physical presence in a state had no legal duty to collect the taxes owed
2. When sellers failed to collect the tax, consumers were obligated to self-declare their untaxed purchases and remit the tax themselves
3. 2018 Supreme Court judgement allowed states to transfer responsibility to remit taxes to the seller
4. Judgement raised tax revenues by 7.9% and increased burden of the tax was fully passed on to consumers
5. Supports idea that buyer is more likely to pay attention to the tax if they have to pay it (tax is more salient) - greater quantity sold when seller remits the tax

Question 21

Benzarti et al. (2020) - commodity tax

Answer 21

1. Hairdressers in Finland got a VAT cut of 14 percentage points in Jan 2007 that was repealed in Jan 2012
2. DD analysis of prices of hairdressers (treatment) and beauty salons (control - tax cut didn’t apply)
3. Find tax decreases are only 50% passed on to consumers (producers pocket tax cut as customers are inattentive to taxes)
4. Find tax increases are almost fully passed on to consumers (producers can justify the price increases to consumers)
5. Contraction to theory (price does not revert to previous level when VAT repealed - increase much higher than decrease)

Question 22

Chetty et al. (2009) - randomized field experiment

Answer 22

1. DD: compare shopping behaviour for treated products vs. control products in treated store, before and after new tags are implemented
2. DDD: repeat analysis for treated products vs. control products in control stores
3. DD treated estimate = -2.14 and DD control estimate = 0.06 (non negative effect on demand in control stores - if treated store did not do experiment would have evolved in the same way)
4. DDD estimate = -2.2

Question 23

Chetty et al. (2009) - policy experiment

Answer 23

1. Use changes in beer excise and sales taxes across states
2. Excise tax is salient because included in posted price (sales tax is not)
3. Increased tax decreases demand irrespective of whether sales or excise tax BUT larger fall for excise tax (more responsive)

Question 24

Chetty et al. (2009) - key finding

Answer 24

1. Salience matters (posting sales tax reduces demand for those goods AND beer consumption is elastic to excise tax rate but not to sales tax rate)
2. If tax is salient to consumers, they are less elastic, and hence more likely to bear the tax burden

Question 25

Eissa (2002) - income tax

Answer 25

1. Looked at Tax Reform Act of 1986 (lowered top marginal tax rate on income from 50% to 28%)
2. Compared married women with spouse at the 99th percentile of the income distribution to those with spouse in the 75th percentile
3. Increase in total labour supply of married women at the top of the income distribution implies an elasticity of total labour supply wrt the after-tax wage of 0.8 and participation elasticity of 0.4

Question 26

Justify Eissa (2002)’s choice of group

Answer 26

Chose married woman because
1. Labour supply response of married women to change in net wage if larger than any other group
2. Based on secondary earner model (married women condition how many house of labour to supply on how many hours husband is already putting in)

Question 27

Justify Eissa (2002)’s control group

Answer 27

Use spouse in 75th percentile as a control group for spouse in 99th percentile because:
1. There are not many behavioural differences between the 78th and 99th percentile
2. If had used women pre- and post-policy, results would have been biased because trend of married women increasingly entering the labour market (external to policy)

Question 28

Eissa and Leibman (1996) - income tax

Answer 28

1. Look at increase in EITC in 1987
2. DD approach: compare single mothers (treated) to single women without kids (control)
3. Extensive margin: positive impact on participation of lone mothers (1.4-307ppts increase)
4. Intensive margin: no effect on hours of work
5. Consistent with theory that income and substitution effect are of opposite sign at the intensive margin but both positive at extensive

Question 29

Kleven et al. (2013) - income tax

Answer 29

1. Bosman ruling in 1995 removed restrictions on the free movement of labour (e.g. 3-Player Rule and Transfer-Fee Rule) - liberalisation of European football market
2. In-migration of foreign players: pre-Bosman inelastic to top earnings tax rate on foreign players vs. post-Bosman elastic (higher top tax rate = lower fraction of foreign players)
3. Out-migration of domestic players: increase in elasticity post-Bosman (higher tax rate elicits greater decrease in fraction playing home)

Question 30

Abadie et al. (2010) - income tax

Answer 30

1. “Beckham Law” in Spain in 2005 - imposes a flat tax rate of 24% (vs. progressive tax with top rate of 43%) for all foreign workers moving to Spain after Jan 2004
2. Synthetic control approach: largest weight on Italy (small positive weights on England France and Portugal)
3. Post-Beckham have higher share of top-quality foreign players than than synthetic control (trends remain similar for lower-quality players)
4. Increase in share of eligible foreign players relative to control but no increase in non-eligible