27
Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program Meeting November 4 2010, Cambridge MA Jonathan A. Parker Kellogg School of Management, Northwestern University

Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

Embed Size (px)

Citation preview

Page 1: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

Discussion of “The Effects of Fiscal Stimulus: Evidence from the

2009 Cash for Clunkers Program”by

Atif Mian and Amir Sufi

NBER Public Economics Program MeetingNovember 4 2010, Cambridge MA

Jonathan A. ParkerKellogg School of Management, Northwestern University

Page 2: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

Outline

0. Summary1. Kick the tires2. Is this consistent with what we know?3. Did CARS work?

Page 3: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

0. Summary of paper

Page 4: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

0. Summary of paper

1. Big effect on impact– Reasonably precise estimation– CARS caused 360,000 purchases about half of

CARS purchases

2. Reversed after a few months

– Big standard errors: 95% confidence interval can’t reject no reversal

Page 5: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

0. Summary of paper

1. Big effect on impact– Reasonably precise estimation– CARS caused 360,000 purchases, about half of

CARS purchases

2. Reversed after a few months– Big standard errors on cumulative

3. There is a shred of evidence on the next direct step in the multiplier and it is big (needs work)

– Auto employment is 0.6% higher in the year following CARS

Page 6: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

1. Kicking the tires

1. Rules complex, measurement of clunkers not• Ratio of clunkers to cars averages either 0.5 or 0.17

2. Scaling purchases and clunkers a concern•

3. Estimation of total effect assumes no effect in bottom decile of

4. Heterogeneity in responses rural/culture

Page 7: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

Rural and Northern

Southern and small/urban

Page 8: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

1. Kicking the tires

1. Rules complex, measurement of clunkers not• Ratio of clunkers to cars averages either 0.5 or 0.17

2. Scaling purchases and clunkers a concern•

3. Estimation of total effect assumes no effect in bottom decile of

4. Heterogeneity in responses rural/culture• If Southern and rural states dislike green cars and like

light trucks for doing farm work, they have lots of clunkers and low responsiveness to the program

Page 9: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

1. Kicking the tires

1. Rules complex, measurement of clunkers not• Ratio of clunkers to cars averages either 0.5 or 0.17

2. Scaling purchases and clunkers a concern•

3. Estimation of total effect assumes no effect in bottom decile of

4. Heterogeneity in responses rural/culture5. What causes the identifying variation?

– eg: Florida, Nevada, Arizona vs S. Dakota

Page 10: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

Farm states booming U-rates: N and S. Dakota lowest

Subprime sand states . .U-rates: Nevada highest and Florida 4th highest

Page 11: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

1. Kicking the tires

1. Rules complex, measurement of clunkers not• Ratio of clunkers to cars averages either 0.5 or 0.17

2. Scaling purchases and clunkers a concern•

3. Estimation of total effect assumes no effect in bottom decile of

4. Heterogeneity in responses rural/culture5. What causes the identifying variation?

– If “controls” imperfect (or endogenous), need a theory of what generates identifying variation

Page 12: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

2. External validity (related evidence)

1. Incidence: Probably on consumer– Busse, Silva-Risso, and Zettelmeyer (AER 2006)

Customers get 70-90% of dealer rebates that the customer knows about (but only 30-40% of those they do not)

2. Reversal

Page 13: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

1. Incidence: Probably on consumer2. Reversal

– Busse, Simester, Zettelmeyer (MS 2010): “Best Deal You’ll Ever Get”

2. External validity (related evidence)

Page 14: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

1. Incidence: Probably on consumer2. Reversal

– Busse, Simester, Zettelmeyer (MS 2010): reversal– Parker, Souleles, Johnson, and McLelland (2010):

no reversal   Non-durable spending

Total spending Spending on new cars

 

Fraction of ESP spent during three months of arrival

0.254 0.757 0.342 (0.110) (0.360) (0.221)

 

Fraction of ESP spent during subsequent three months of arrival

0.157 0.479 0.315 (0.178) (0.568) (0.350)

   

Cumulative fraction of ESP spent over bopth periods

0.411 1.236 0.657(0.273) (0.892) (0.551)

2. External validity (related evidence)

Page 15: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

3. Did CARS work?Complaint: The paper’s tone is that CARS did not work.

“Our findings do not warrant the claim that all forms of fiscal stimulus fail to boost long-run economic output.”

“. . .any argument that the CARS program had [a positive aggregate effect] must be consistent with (a) the sharp relative reduction in auto purchases we find in high CARS exposure cities after the program, and (b) the lack of any discernable relative impact on employment, house prices, or household defaults in high versus low CARS exposure cities.”

Page 16: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

3. Did CARS work?

2007-III

2007-IV

2008-I

2008-II

2008-III

2008-IV

2009-I

2009-II

2009-III

2009-IV

2010-I

2010-II

2010-III

-8.0%

-6.0%

-4.0%

-2.0%

0.0%

2.0%

4.0%

6.0%Annualized Real Growth Rates

GDP PCE

Page 18: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

3. Did CARS work?

2007-III

2007-IV

2008-I

2008-II

2008-III

2008-IV

2009-I

2009-II

2009-III

2009-IV

2010-I

2010-II

2010-III

-8.0%

-6.0%

-4.0%

-2.0%

0.0%

2.0%

4.0%

6.0%Annualized Real Growth Rates

GDP PCE CARS

Page 19: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

1. Big bang for buck: Cost $2.85 caused $38 billion in demand– Mian-Sufi: CARS cost $2.85 billion and caused

360,000 car purchases– If average new car cost $25,000 then estimates

imply that CARS increased “demand” directly by $38 billion

– Implies CARS increased PCE directly by 0.4% in 2009 Q3 and decreased PCE directly by 0.2% in 2009 Q4 and 2010 Q1 for only 2.5 billion

Page 20: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

3. Did CARS work?

2007-III

2007-IV

2008-I

2008-II

2008-III

2008-IV

2009-I

2009-II

2009-III

2009-IV

2010-I

2010-II

2010-III

-8.0%

-6.0%

-4.0%

-2.0%

0.0%

2.0%

4.0%

6.0%Quarterly Growth Rates

GDP PCE CARS

Page 21: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

1. Big bang for buck: Cost $2.85 caused $38 billion in demand

– Aside: roughly half of purchases (360K/677K) would have occurred anyway

– Could have other benefits: greener world

2. The multiplier can be huge at the ZLB!– Or less any price effects (crowding out)

Page 22: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program
Page 23: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

1. Big bang for buck: Cost $2.85 caused $38 billion in demand

– Aside: roughly half of purchases (360K/677K) would have occurred anyway

– Could have other benefits: greener world

2. The multiplier can be huge at the ZLB!– Or less any price effects (crowding out)

3. But lots in ERRA besides CARS going on, no?

Page 24: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

Source: Figure 10 in Cogan and Taylor (2010) , NBEER WP 16505

“Considering both the federal and the state and local sector, we find that the effects of ARRA on purchases to have been remarkably small for the first six quarters of the program despite the large overall size of ARRA.”

Cogan and Taylor (2010)

Page 25: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

“Considering both the federal and the state and local sector, we find that the effects of ARRA on purchases to have been remarkably small for the first six quarters of the program despite the large overall size of ARRA. . . . Basic economic theory implies that temporary increases in transfer payments have a much smaller impact than government purchases.”

Not for CARS . . . CARS is about temporary price variation for an asset – intertemporal substitution strong

Cogan and Taylor (2010)

Page 26: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

1. Big bang for buck: Cost $2.85 caused $38 billion in demand

– Aside: roughly half of purchases (360K/677K) would have occurred anyway

– Could have other benefits: greener world

2. The multiplier can be huge at the ZLB!– Or less any price effects (crowding out)

3. Not lots in ERRA besides CARS going on, so CARS potentially useful– Temporary low price is a good way to generate

demand (like investment tax credits)– But, reversal does take demand from the future

Page 27: Discussion of “The Effects of Fiscal Stimulus: Evidence from the 2009 Cash for Clunkers Program” by Atif Mian and Amir Sufi NBER Public Economics Program

Mian-Sufi: CARS caused a large amount of spending reversed in 6 months to a year

Does not imply CARS did not significantly help stabilize the economy

Quite plausible that CARS was highly successful and also plausible that it was not worth the expenditure