Empirical Household Finance. Theresa Kuchler (NYU Stern)

Transcription

1 Empirical Household Finance Theresa Kuchler (NYU Stern)

2 Overview Three classes: 1. Questions and topics on household finance 2. Recent work: Online data sources 3. Recent work: Administrative data sources 1

3 Overview Today s Lecture Administrative and regulatory data Guest: Johannes Stroebel Regulating Consumer Financial Products: Evidence from Credit Cards Asymmetric Information and the Targeting Problem of Bank Mediated Stimulatory Policy 2

4 Administrative and Regulatory Data Recall: Short-comings of most existing household finance datasets Limited information on full household balances (esp. in US data) Often low frequency information, cross-section Sample selection 3

5 Administrative and Regulatory Data Recall: Short-comings of most existing household finance datasets Limited information on full household balances (esp. in US data) Often low frequency information, cross-section Sample selection New datasources: Online data Information provided by households (via online services) More comprehensive view of household balance sheet High frequency data collection 3

6 Administrative and Regulatory Data Recall: Short-comings of most existing household finance datasets Limited information on full household balances (esp. in US data) Often low frequency information, cross-section Sample selection New datasources: Online data Information provided by households (via online services) More comprehensive view of household balance sheet High frequency data collection But: Sample Selection Very limited information on firm side Administrative /regulatory data 3

7 Administrative and Regulatory Data Collected by government agency as part of administration /regulatory oversight Consistent data collection over time Participation required by law Minimize sample selection 4

8 Administrative and Regulatory Data Collected by government agency as part of administration /regulatory oversight Consistent data collection over time Participation required by law Minimize sample selection Data collected from firms Information on supply side and demand as perceived by firms Limited information on household positions with other firms 4

9 Administrative and Regulatory Data in the US Credit Cards 5

10 Administrative and Regulatory Data in the US Credit Cards Credit Card Panel Monthly panel of all credit card accounts held by regulated banks (most) Access by the OCC, CFPB 5

11 Administrative and Regulatory Data in the US Credit Cards Credit Card Panel Monthly panel of all credit card accounts held by regulated banks (most) Access by the OCC, CFPB Consumer Credit Panel Monthly panel of credit bureau records for subset of US population Access by the Fed system, CFPB 5

12 Administrative and Regulatory Data in the US Credit Cards Credit Card Panel Monthly panel of all credit card accounts held by regulated banks (most) Access by the OCC, CFPB Consumer Credit Panel Other Mortgages Monthly panel of credit bureau records for subset of US population Access by the Fed system, CFPB IRS (tax records) 5

13 Today s Work Johannes Stroebel (joint with Sumit Agarwal, Souphala Chomsisengphet and Neale Mahoney) Regulating Consumer Financial Products: Evidence from Credit Cards Asymmetric Information and the Targeting Problem of Bank Mediated Stimulatory Policy 6

14 Methodology: Diff-in-Diff Goal: Want to know effect of X on outcome y 7

15 Methodology: Diff-in-Diff Goal: Want to know effect of X on outcome y Need: (Quasi-)Randomly change X for subset of consumers Differences in outcome y due to change in X 7

16 Methodology: Diff-in-Diff Goal: Want to know effect of X on outcome y Need: (Quasi-)Randomly change X for subset of consumers Differences in outcome y due to change in X (Quasi-)random changes? Policy changes, eligibility criteria,... 7

17 Methodology: Diff-in-Diff Goal: Want to know effect of X on outcome y Need: (Quasi-)Randomly change X for subset of consumers Differences in outcome y due to change in X (Quasi-)random changes? Policy changes, eligibility criteria,... before after Control group Y c,before Y c,after Treatment group Y t,before Y t,after 7

18 Methodology: Diff-in-Diff Goal: Want to know effect of X on outcome y Need: (Quasi-)Randomly change X for subset of consumers Differences in outcome y due to change in X (Quasi-)random changes? Policy changes, eligibility criteria,... before after Control group Y c,before Y c,after Treatment group Y t,before Y t,after T reatment effect = (Y t,after Y t,before ) (Y c,after Y c,before ) 7

19 Methodology: Diff-in-Diff Goal: Want to know effect of X on outcome y Need: (Quasi-)Randomly change X for subset of consumers Differences in outcome y due to change in X (Quasi-)random changes? Policy changes, eligibility criteria,... before after Control group Y c,before Y c,after Treatment group Y t,before Y t,after T reatment effect = (Y t,after Y t,before ) (Y c,after Y c,before ) where Y t,after Y t,before : Change in outcome for treatment group Y c,after Y c,before : Change in outcome for control group What would have happened if no treatment 7

20 Regulating Consumer Financial Products: Evidence from Credit Cards 8

21 Regulation Consumer Financial Products - Discussion Why interesting? CARD Act Large policy change inherently interesting 9

22 Regulation Consumer Financial Products - Discussion Why interesting? CARD Act Large policy change inherently interesting Economic Insights? Competitive structure of industry and borrower behavior Implications for future regulation 9

23 Methodology: Regression Discontinuity 10

24 Methodology: Regression Discontinuity Treatment Forcing Variable 10

25 Methodology: Regression Discontinuity Treatment Question: Effect of tutoring on subsequent scores? Intensive Tutoring Twice weekly Tutoring Once weekly No Tutoring Forcing Variable Test Score 10

26 Methodology: Regression Discontinuity Treatment Intensive Tutoring Twice weekly Question: Effect of tutoring on subsequent scores? Idea: Students around cut-off very similar Tutoring Once weekly No Tutoring Forcing Variable Test Score 10

27 Methodology: Regression Discontinuity Requirements: 11

28 Methodology: Regression Discontinuity Requirements: Substantial change in treatment around discontinuity 11

29 Methodology: Regression Discontinuity Requirements: Substantial change in treatment around discontinuity Other variables trend smoothly through cut-off Note: do not need to be the same, but smooth trend 11

30 Methodology: Regression Discontinuity Requirements: Substantial change in treatment around discontinuity Other variables trend smoothly through cut-off Note: do not need to be the same, but smooth trend No bunching around discontinuity Otherwise manipulation of being just above /below cutoff 11

31 Methodology: Regression Discontinuity Requirements: Substantial change in treatment around discontinuity Other variables trend smoothly through cut-off Note: do not need to be the same, but smooth trend No bunching around discontinuity Otherwise manipulation of being just above /below cutoff Common discontinuities Eligibility formulas Cut-off rules 11

32 Asymmetric Information and the Targeting Problem of Bank Mediated Stimulatory Policy 12

33 Discussion Great example of how to present regression discontinuity results 13

34 Discussion Great example of how to present regression discontinuity results Exploration of heterogeneity 13

35 Discussion Great example of how to present regression discontinuity results Exploration of heterogeneity Multiple experiments Often the case with big data See also Einav, Kuchler, Levin, Sundaresan (2015): Assessing Sale Strategies in Online Markets using Matched Listings, AEJ:Micro 13

36 Questions? 14