Consumer revolving credit and debt over the life-cycle and business cycle
Scott L. Fulford1 and Scott Schuh2
1Boston College [email protected]
2Federal Reserve Bank of Boston [email protected]
The views expressed in this paper are the author’s and do not necessarily reflect the official position of the Federal Reserve Bank of Boston or the Federal Reserve System.
FDIC Credit and debt Scott L. Fulford and Scott Schuh -1- Changing credit and changing debt
The relationship between consumer credit and debt On three different levels: In aggregate over business cycle On average over life-cycle For individuals Focus on credit cards Widely available and used Limits observable Start with descriptive and reduced from, then add insights from consumption theory to make sense of results Key distinction between credit cards as payments mechanism, and credit as a way to accumulate debt
Introduction Credit and debt Scott L. Fulford and Scott Schuh -2- Background
For most households, credit is larger than liquid assets, and is the marginal source for smoothing (Fulford (2015b) EER) Credit limits are much more volatile than income (Fulford (2015a) JME ) Household debts severely understudied (Zinman, forthcoming)
Introduction Credit and debt Scott L. Fulford and Scott Schuh -3- Background
For most households, credit is larger than liquid assets, and is the marginal source for smoothing (Fulford (2015b) EER) Credit limits are much more volatile than income (Fulford (2015a) JME ) Household debts severely understudied (Zinman, forthcoming) Other work has established increases in credit->increases in debts (Agarwal et al., 2015; Gross and Souleles, 2002) Limitations: Generally short term, for single credit accounts, and does not distinguish uses of credit cards for payments
Introduction Credit and debt Scott L. Fulford and Scott Schuh -3- Background
For most households, credit is larger than liquid assets, and is the marginal source for smoothing (Fulford (2015b) EER) Credit limits are much more volatile than income (Fulford (2015a) JME ) Household debts severely understudied (Zinman, forthcoming) Other work has established increases in credit->increases in debts (Agarwal et al., 2015; Gross and Souleles, 2002) Limitations: Generally short term, for single credit accounts, and does not distinguish uses of credit cards for payments Little empirical work on consumer credit over the life-cycle Some on mortgage debt over life-cycle, recent work on student loans
Introduction Credit and debt Scott L. Fulford and Scott Schuh -3- Data Sources
Equifax/CCP 5% sample of all credit accounts, with full history for an account (for computational reasons, actually use a smaller sample) Quarterly 1999-2015 Other than age, limited individual information beyond credit and debts Does not distinguish between those who revolve debt and those who pay full bill every month Survey of Consumer Payment Choice In field since 2008 Asks questions on consumer preferences and adoption of payment options Survey of Consumer Finances
Data description Credit and debt Scott L. Fulford and Scott Schuh -4- Credit and debts vary, use does not
1
14000 .9 .9
.8 .7 10000
.6 8000 .5
.4 utilization Credit .3 6000
Credit and debt ($ log ($ debt log scale) and Credit .2
5000 .1
0 4000 2000q1 2005q1 2010q1 2015q1 Date
Mean credit card limit Mean credit card debt Mean credit utilization (right axis)
Source: Author’s calculations from Equifax/CCP Business cycle Credit and debt Scott L. Fulford and Scott Schuh -5- Credit limits and debts increase with age
Credit Card Limit 100
50
20
10
5 Credit Card Debt
credit card limit or debt, thousands (if (if positive) thousands limit debt, card or credit 1
20 40 60 80 Age Credit utilization decreases slowly
.8 Credit Utilization
.6
.4
.2 limit Card / debt card Credit Credit
0 20 40 60 80 Age
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -7- Distribution of credit utilization
All ages Age 20-30 Age 30-40
2 5 2.5
4 2
1.5
3
1.5
1 Density Density Density
2 1
.5
1 .5
0 0 0 0 .2 .4 .6 .8 1 1.4 0 .2 .4 .6 .8 1 1.4 0 .2 .4 .6 .8 1 1.4 Credit card utilization ratio Credit card utilization ratio Credit card utilization ratio
Age 40-50 Age 50-60 Age 60-80
5 8
3
4
6
2 3
4 Density Density Density Density
2
1
2
1
0 0 0 0 .2 .4 .6 .8 1 1.4 0 .2 .4 .6 .8 1 1.4 0 .2 .4 .6 .8 1 1.4 Credit card utilization ratio Credit card utilization ratio Credit card utilization ratio
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -8- Credit utilization
Credit utilizationit = θt + θa + αi + βCredit utilizationi,t− 1 + eit
Credit utilizationt
Credit utilizationt−1 0.874*** 0.868*** 0.647*** (0.000876) (0.000892) (0.00131) Constant 0.0479*** (0.000461) Observations 347,642 347,642 347,642 R-squared 0.741 0.743 0.429 Fixed effects No No Yes Age and year effects No Yes Yes Number of accounts 10,451 Frac. Variance from FE 0.477
Credit utilizationit = Dit /Bit Dit : Credit card debt (including both revolving and non-revolving) Bit : Credit limit (if positive)
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -9- Credit utilization: non-parametric 1.1 1.1 1.1
1 1 1
.9 .9 .9
.8 .8 .8
.7 .7 .7
.6 .6 .6
.5 .5 .5
one year two years one quarter .4 .4 .4
.3 .3 .3 Credit utilization in Credit utilization in Credit utilization in Credit utilization
.2 .2 .2
.1 .1 .1
E[y|x] = 0.041 + 0.896x E[y|x] = 0.077 + 0.786x E[y|x] = 0.100 + 0.698x 0 0 0 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 1.1 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 1.1 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 1.1 Credit utilization today Credit utilization today Credit utilization today
.5 .5 .5
.4 .4 .4
.3 .3 .3
.2 .2 .2
.1 .1 .1
0 0 0
one year two years −.1 −.1 −.1 one quarter
−.2 −.2 −.2
−.3 −.3 −.3
Difference in utilization incredit Difference in utilization incredit Difference in utilization incredit Difference −.4 −.4 E[y|x] = 0.666x −.4 E[y|x] = 0.339x E[y|x] = 0.128x
−.5 −.5 −.5 −.5−.4−.3−.2−.1 0 .1 .2 .3 .4 .5 −.5−.4−.3−.2−.1 0 .1 .2 .3 .4 .5 −.5−.4−.3−.2−.1 0 .1 .2 .3 .4 .5 Difference in credit utilization Difference in credit utilization Difference in credit utilization from own mean from own mean from own mean
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -10- A little theory
Convenience users: Credit card debts are some fraction of consumption. Should behave like consumption:
Di,t = ωi,t Ci,t
Revolvers: Credit card debts accumulate (or are paid off). Should behave like negative assets:
Dit +1 = (1 + r )(Dit + Cit − Yit )
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -11- A little theory
Convenience users: Credit card debts are some fraction of consumption. Should behave like consumption:
Di,t = ωi,t Ci,t
Revolvers: Credit card debts accumulate (or are paid off). Should behave like negative assets:
Dit +1 = (1 + r )(Dit + Cit − Yit )
Different relationships between past debts and current debts for revolvers and convenience users. Use to separate statistically between them (EM algorithm)
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -11- Fraction Convenience
By age By utilization
.8 .8
.6
.6
.4
.4
Fraction convenience users Fraction convenience users .2
0 .2 20 40 60 80 0 .5 1 Age Credit Utilization
Fraction convenience: Estimated SCF Fraction convenience: Estimated SCF
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -12- Fraction revolving SCPC, conditional on having card
.7
.6
revolving .5
Fraction .4
.3 20 30 40 50 60 70 Age Group
Source: Author’s calculations from SCPC Life cycle Credit and debt Scott L. Fulford and Scott Schuh -13- Age and utilization
.6
.4 utilization
Credit .2 0
20 40 60 80 Age
Convenience users Revolvers Overall utilization
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -14- Credit utilization
Credit utilizationit = θt + θa + αi + βCredit utilizationi,t− 1 + eit
Credit utilizationt All All All Revolvers
Credit utilizationt−1 0.874*** 0.868*** 0.647*** 0.766*** (0.000876) (0.000892) (0.00131) (0.00125) Constant 0.0479*** (0.000461) Observations 347,642 347,642 347,642 R-squared 0.741 0.743 0.429 Fixed effects No No Yes Yes Age and year effects No Yes Yes Yes Number of accounts 10,451 Frac. Variance from FE 0.477
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -15- Credit into debt
Log Debtit = θi + θt + θa + αLog Limiti,t− 1 + βLog Debti,t− 1 + eit
Log Debtt All All All Revolvers [1] [2] [3] [4]
Log Debtt−1 0.505*** 0.758*** 0.604*** 0.859*** (0.00157) (0.00119) (0.00132) (0.000594)
Log Credit Limitt−1 0.414*** 0.134*** 0.313*** 0.132*** (0.00262) (0.00148) (0.00243) (0.000784) Observations 296,369 296,369 361,280 361,280 R-squared 0.432 0.667 0.610 0.927 Accounts 10,028 10,718 10,718 Fixed effects Yes No Yes Yes Zero included No No Yes No Age effects Yes Yes Yes Yes Long term credit impact 0.862 0.875 0.879 0.990 Credit salience σ 0.443 0.665 0.530 0.754
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -16- Changing with utilization and age for Revolvers
Effect of 1% change in limit Effect of 1% change in debt
1 .3
.25 .9
.2
.8
.15
.7 .1 Average marginal debt effect log of
Average marginal effect credit logof limit .05
.6 20 30 40 50 60 70 20 30 40 50 60 70 Age Age
Utilitzation 0.1 0.4 0.7 Utilitzation 0.1 0.4 0.7 0.8 0.9 1 0.8 0.9 1
Source: Author’s calculations from Equifax/CCP
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -17- Review of results
Big fluctuations in credit and debt over life-cycle and business cycle Yet credit utilization for credit cards very stable Stability comes from individuals returning quickly to mean utilization after shocks to debt and credit
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -18- Review of results
Big fluctuations in credit and debt over life-cycle and business cycle Yet credit utilization for credit cards very stable Stability comes from individuals returning quickly to mean utilization after shocks to debt and credit Individual stability from two sources: 1 Convenience users: Use credit cards as payments, around 10% of available credit. 2 Revolvers: Using 50-60% of available credit. Quickly return to their “target” utilization following increase in limit or debts. For Revolvers: All of increase in limit leads to additional debt eventually! Fraction Revolver falls slowly from around 70% in 20s to around 30% in 70s
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -18- References I
Agarwal, Sumit, Souphala Chomsisengphet, Neale Mahoney, and Johannes Stroebel. 2015. “Do Banks Pass Through Credit Expansions? The Marginal Profitability of Consumer Lending During the Great Recession.” Tech. rep., SSRN. Fulford, Scott L. 2015a. “How important is variability in consumer credit limits?” Journal of Monetary Economics 72:42 – 63. ———. 2015b. “The surprisingly low importance of income uncertainty for precaution.” European Economic Review 79:151–171. Gross, David B. and Nicholas S. Souleles. 2002. “Do Liquidity Constraints and Interest Rates Matter for Consumer Behavior? Evidence from Credit Card Data.” Quarterly Journal of Economics 117 (1):149–185. Zinman, Jonathan. forthcoming. “Household Debt: Facts, Puzzles, Theories, and Policies.” Annual Review of Economics .
Life cycle Credit and debt Scott L. Fulford and Scott Schuh -19- Credit ratings increase with age 800
750 700 650 Equifax Credit Rating to Rating (280 Credit 850) Equifax 600 20 40 60 80 Age
Appendix Credit and debt Scott L. Fulford and Scott Schuh -20- Credit rating inequality large
99 800 90 75
700 Median
600 25
10 500 percentiles rating credit Equifax
1 400
20 40 60 80 Age
Appendix Credit and debt Scott L. Fulford and Scott Schuh -21- Mortgage debts
Have mortgage debt? Log mortgage debt (if positive)
.6 12
11.5 .4 11
.2 Have positive mortgage debt 10.5 log mortgagelog debt (if positive) 10
0 20 40 60 80 20 40 60 80 Age Age
Source: Author’s calculations from Equifax/CCP
Appendix Credit and debt Scott L. Fulford and Scott Schuh -22- Student loans
Have student loan debt? Log student loan debt (if positive)
10 .5
9
.4
8
.3
7
.2
6
.1 Have student loan or personal finance debt log student loan/personal finance debt (if positive)
5
0 20 40 60 80 20 40 60 80 Age Age
Source: Author’s calculations from Equifax/CCP
Appendix Credit and debt Scott L. Fulford and Scott Schuh -23- Auto loans
Have auto loan? Log auto loan debt (if positive)
10 .4
9.5
.3
9 .2 Have auto debt
log autodebt (if positive)
.1 8.5
0 8 20 40 60 80 20 40 60 80 Age Age
Source: Author’s calculations from Equifax/CCP
Appendix Credit and debt Scott L. Fulford and Scott Schuh -24-