Since the start of the financial crisis, the liabilities side of household balance

Transcription

1 current FEDERAL RESERVE BANK OF NEW YORK issues in Economics and Finance Voume 19, Number 2 F 213 F The Financia Crisis at the Kitchen Tabe: Trends in Househod Debt and Credit Meta Brown, Andrew Haughwout, Donghoon Lee, and Wibert van der Kaauw Since the onset of the financia crisis, househods have reduced their outstanding debt by about $1.3 triion. Whie part of this reduction stemmed from a historic increase in consumer defauts and ender charge-offs, particuary on mortgage debt, other factors were aso at pay. An anaysis of the New York Fed s Consumer Credit Pane a rich new data set on individua credit accounts reveas that househods activey reduced their obigations during this period by paying down their current debts and reducing new borrowing. These househod choices, aong with banks stricter ending standards, heped drive this deeveraging process. Since the start of the financia crisis, the iabiities side of househod baance sheets has been the subject of urgent interest among poicymakers and the media. Aggregate trends documented in the Federa Reserve System s Fow of Funds Accounts demonstrate a steep run-up in consumer debt from 1999 to 28, foowed by a pronounced decine through at east third-quarter 2. According to most views, the crisis began in the residentia mortgage market, as an increasingy arge number of borrowers, especiay in the nonprime segment, became deinquent on their mortgage payments. The increase in deinquencies and the enormous rise in residentia mortgage forecosures soon deveoped into a fu-bown financia crisis and ed to one of the sharpest market contractions in U.S. history. Whie many trends in the financia system payed a roe in these deveopments, househod behavior was ceary a fundamenta contributor. Since the financia crisis eased and economic growth resumed in the second haf of 29, many anaysts have pointed to consumer behavior as a crucia determinant of the vigor and sustainabiity of the economic recovery. 1 In this artice, we use a rich new data set on individua credit accounts to investigate the behavior underying these aggregate trends. Our data set the Federa Reserve Bank of New York (FRBNY) Consumer Credit Pane is an ongoing pane of quartery data on individua and househod debt (see box). It is created from a arge sampe of U.S. consumer credit reports provided to the New York Fed by Equifax, one of three nationa credit reporting agencies. The data aow for reiabe high-frequency measurement of trends in aggregate debt. Unike the Fow 1 See, for exampe, Federa Reserve Governor Eizabeth A. Duke s speech at Ohio Banker s Day, and Federa Reserve Governor Sarah Boom Raskin s speech at the New America Foundation,

2 CURRENT ISSUES IN ECONOMICS AND FINANCE v Voume 19, Number 2 The New York Fed s Consumer Credit Pane The FRBNY Consumer Credit Pane represents a 5 percent random sampe of U.S. individuas with credit fies as we as a of their househod members. a In a, the data set incudes anonymous credit fies on more than 15 percent of the popuation, or approximatey 37 miion individuas. We use information from the credit reports for those individuas for each quarter during the ast thirteen years, with current data through September 2. The samping expoits randomness in the ast two digits of individuas Socia Security numbers. b The procedure ensures that the Pane is dynamicay updated in each quarter to refect new entrants into credit markets. In addition, Equifax, the data provider, matches the primary individua s maiing address to a records in the data in order to capture information about other members of the primary individua s househod. These individuas are aso added to the sampe. This procedure enabes us to track individuas and househods consistenty over time, thus aowing us to study richer dynamics of consumer debt and reated poicy issues at both the individua and househod eve. Our credit report data incudes residentia ocation at the census bock eve and the individua s month and year of birth. The data aso contain detaied information on each individua mortgage oan, incuding origination date, origina baance, current baance, current (schedued) payment, current status (that is, current, thirty-days deinquent, and so on). Whie the mortgage information in the data set is very detaied and, we beieve, compementary to oan-eve information avaiabe from sources ike LoanPerformance and LPS (McDash), it differs in important ways from these sources. In particuar, the mortgage information does not indicate the seniority of individua mortgage oans. However, because the FRBNY Consumer Credit Pane data are coected at the borrower eve, they offer a different perspective on mortgage debt than is avaiabe in standard oan-eve data sets. In addition to information on debts secured by residentia rea estate, the data set incudes somewhat more aggregate data on individuas and househods other oans, such as credit cards, auto oans, and student oans. Here, the data incude the foowing: tota number of accounts of each type (for exampe, the tota number of bank-issued credit cards), credit imit on each type of account (for exampe, the combined credit imit on a credit cards), c tota baance on each type of account in each status (for exampe, the tota student oan baance that is current, thirty-days deinquent, and so on). More genera information on the credit report incudes the foowing: indicators for whether the individua has a forecosure or bankruptcy within the ast twenty-four months, and ever, on the report; indicators for whether the individua has any accounts in coection and the amount of coection; a consumer score that is anaogous to the we-known FICO score. a See Avery et a. (23) for a detaied discussion of the contents, sources, and quaity of credit report data. b See Lee and van der Kaauw (2) for further detais about the sampe design and content of the FRBNY Consumer Credit Pane. c This fied is known as the high credit amount in the credit report data. It refers to either the credit imit (for credit cards, home equity ines of credit, and other revoving debt) or the highest baance (for mortgages, auto oans, and other instament debt). There are instances in which credit imits on revoving accounts are unreported, in which case the high credit variabe refects the historica high credit eve for the account. Avery et a. (23) and Hunt (22) point out that the reporting of credit imits in credit reports has improved consideraby in recent years. of Funds Accounts and reated industry-sourced aggregate data, the Pane data enabe us to observe debt accumuation, repayment, and deinquency at the individua consumer eve. We begin by using the Consumer Credit Pane to produce new measures of aggregate debt trends since Our evidence, ike evidence obtained from the Fow of Funds Accounts, shows that the eve of househod debt, after a sustained period of increase, began to decine in 28; moreover, aggregate deinquencies peaked at the end of Furthermore, whie debt continues 2 Whie we compare the Consumer Credit Pane to the Fow of Funds here, there are important conceptua differences between the two, incuding the fact that the Fow of Funds measures incude nonprofit organizations that are absent from the Consumer Credit Pane measures. See Lee and van der Kaauw (2). to decine, tota deinquencies have shown signs of stabiization in the Consumer Credit Pane in recent quarters. Next, we appy the Pane s consumer-eve data to address the caim that the decine in aggregate debt derived entirey from charge-offs defauts written off borrowers credit records by enders and not from a decine in consumers reiance on debt or an increase in their debt repayment. We decompose consumer debt changes into three components: charge-offs, housing transaction-derived changes, and more standard refinancing and repayment. We find that between 28 and 2, househods switched from borrowing to repaying, resuting in a neary $5 biion reduction in annua consumer cash fows from debt between the precrisis period and 2. 2

3 Chart 1 Tota Debt Baance and Its Composition Triions of doars 14 Other Student oan Credit card Auto oan Home equity ine of credit 8 Mortgage Source: Federa Reserve Bank of New York Consumer Credit Pane/Equifax. Note: Student oan data prior to 23 refect some deays in the reporting of student oans by servicers to credit bureaus. This coud ead to some undercounting of student oan totas in specific periods and impact other student oan-specific measurements. However, variabiity in student oan baances prior to 23 does not materiay affect the aggregate debt time series because the variabiity is sma reative to the tota baances. Other components of househod debt are unaffected. Consumer Debt during the Precrisis Period From first-quarter 1999 (when our data begin) through thirdquarter 28, we observe substantia increases in consumer indebtedness. On March 31, 1999, consumers owed about $4.6 triion to creditors. During the subsequent nine years, consumer indebtedness rose more than 17 percent, reaching $.7 triion at the end of third-quarter The driving force behind these changes was debt secured by residentia rea estate, which accounts for the great majority more than 7 percent in a periods of househod iabiities. Amounts owed on instament mortgages and home equity ines of credit (HELOCs) triped over this period from $3.3 triion to $ triion accounting for $6.7 triion of the tota $8 triion increase in consumer iabiities. Nonetheess, other forms of consumer debt aso rose sharpy, neary doubing from $1.4 triion to $2.7 triion. Many factors were responsibe for these increases, incuding rising popuations, incomes, stock and house prices, faing interest rates, and the democratization of credit. Indeed, whie consumer indebtedness the iabiities side of the househod baance sheet was rising sharpy, the Fow of Funds Accounts indicate that assets owned by the househod sector were growing as we, eaving consumers net weath (the difference between the vaue of assets owned and iabiities owed) to grow steadiy over the period. Like the Consumer Credit Pane, the Fow of Funds Accounts show an increase in consumer debt from 1999 through mid-28. Housing s share of overa debt is roughy 7 to 8 percent during the period in both series, and the reative contributions of housing and nonhousing debt to the consumer debt cimb are simiar in the Pane and Fow of Funds. Deinquency rates remained stabe from 1999 through 26 in the Consumer Credit Pane, with roughy 4 percent of tota outstanding debt thirty or more days past due (deinquent) and 2 percent of tota debt ninety or more days past due (severey deinquent). However, deinquency rose quicky during 27, reaching 6.7 percent by the end of the year and 8.5 percent by the peak of consumer debt in third-quarter 28. Severe deinquency cimbed to 3.6 percent by the end of 27 and 5.1 percent by third-quarter 28. Hence, the data revea both a precrisis period of credit expansion associated with very steady consumer debt performance and emerging evidence of repayment difficuties as eary as 27. Consumer Debt since the Financia Crisis Since the end of third-quarter 28, U.S. consumers have reduced their indebtedness by $1.4 triion, resuting in a decrease in the aggregate consumer debt baance from $.7 triion at its peak in third-quarter 28 to $.3 triion at the end of third-quarter 2. Chart 1 shows the tota debt observed on credit reports for the entire ife of the Pane, in the aggregate and broken down by oan type. Tota househod debt has decreased roughy percent since its peak. Mortgage-reated debt now accounts for 76 percent of tota debt, with the remainder comprising credit cards, auto oans, student oans, and other consumer debt. 4 3 By comparison, GDP rose about 58 percent over the period, from $9.1 triion to $14.4 triion. 4 Since the eariest iterations of the Consumer Credit Pane, a major change has occurred in the measurement of student oan indebtedness. For more on the changes we have made to our definitions of student oans, see page 2 3

4 CURRENT ISSUES IN ECONOMICS AND FINANCE v Voume 19, Number 2 Chart 2 Tota Debt Baance by Deinquency Status Percent Severey deinquent days ate 9 days ate 6 days ate 3 days ate Current Source: Federa Reserve Bank of New York Consumer Credit Pane/Equifax. Like the Consumer Credit Pane, the Fow of Funds Accounts indicate a substantia decine in outstanding househod debt from its 28 peak to the present. The magnitude of the decine in the Fow of Funds is ower, however, at sighty more than $96 biion. Both series indicate steep decines in debt (more than 3 percent) in 29: in the Fow of Funds, decines are moderate (.2 percent to 1.6 percent per year) starting in 2, whie the Pane indicates continued decines of 1.5 percent to 3.7 percent annuay through third-quarter 2. Lee and van der Kaauw (2) discuss some additiona differences in the measurements. Deinquency in the Consumer Credit Pane continued to cimb foowing the debt peak in third-quarter 28, reaching its maximum eve to date in fourth-quarter 29 (see Chart 2). At that point, deinquency had grown from its previousy stabe 4-to- 5 percent of outstanding debt to.9 percent. Severe deinquency peaked in first-quarter 2 at 8.7 percent of outstanding debt, despite having never reached 3 percent for the entire period. Put differenty, deinquency and severe deinquency rates roughy triped and quadruped, respectivey, over a period of three-and-a-haf years. It is interesting to see that the deterioration of househod debt started as eary as the end of 26 and acceerated from that point through fourth-quarter 29. Both measures of deinquency decined decisivey from fourth-quarter 29 to third-quarter 2, the most recent avaiabe quarter of data. For third-quarter 2, deinquency stands at 8.9 percent of outstanding debt and severe deinquency at 6.6 percent. These figures represent a decine of roughy 25 percent Footnote 4 continued of the 2:Q3 Quartery Report on Househod Debt and Credit, avaiabe at DistrictReport_Q32.pdf. Chart 3 Fow into Deinquency by Loan Type: Annua Rate for the Preceding Four Quarters Percent Mortgage 2 Credit card 3 4 HELOC 5 Source: Federa Reserve Bank of New York Consumer Credit Pane/Equifax. Note: HELOC is home equity ine of credit. Auto oan Other in each case. The downward trend in deinquency rates was broken ony once, with a one-quarter uptick in eary deinquencies in third-quarter 2. The fow into deinquency is a measure of debt baances that were previousy current but became newy deinquent in each quarter. Chart 3 shows that debt performance deteriorated across a debt types, but the deterioration of instament mortgage debt excuding HELOCs preceded that of other categories. Between fourth-quarter 25 and fourth-quarter 28, new instament mortgage deinquencies triped, from $98 biion to more than $3 biion. After that, the deterioration of mortgage

5 Chart 4 Number of New Forecosures and Bankruptcies Chart 5 Nonmortgage Debt Changes Other than Charge-Offs Thousands 1, 8 Bankruptcies Biions of doars 3 25 Auto oan + credit card + other Student oan 2 Forecosures Source: Federa Reserve Bank of New York Consumer Credit Pane/Equifax. debt sowed steadiy. New mortgage deinquencies reached a recent ow of $14 biion in third-quarter 2. 5 Recent improvements in the performance of mortgage and other househod debt are naturay refected in a decrease in forecosures and persona bankruptcies. 6 Chart 4 shows the quartery number of new forecosures and bankruptcies nationwide, with approximatey 2 miion peope experiencing a new forecosure and approximatey 2 miion experiencing a new bankruptcy in 29. Bankruptcies and especiay forecosures decined substantiay between second-quarter 2 and third-quarter 2. The second-quarter 2 to third-quarter 2 decine in bankruptcies, when annuaized, impies a decrease of 1,68,48 annua fiings. The second-quarter 2 to third-quarter 2 decine in forecosures, when annuaized, impies a decrease of 968,32 individua forecosure experiences. 7 5 Deinquencies and defauts in mortgage oans basicay drive the patterns in tota debt in Chart 2, given that mortgage-reated debt accounts for 7 to 8 percent of the aggregate debt baance. Athough mortgages are responsibe for much of the magnitude of consumer debt, the prevaence of mortgage debt among U.S. househods is either simiar to or substantiay ess than the prevaence of credit card debt, depending on the measure. Bucks et a. (29) find in the 27 Survey of Consumer Finances that 48.7 percent of househods hod home-secured debts; we observe 42.3 percent of September 27 Pane househods with homesecured debts. Bucks et a. show 46.1 percent of househods reporting credit card baances; we find that 76.1 percent of September 27 Pane househods have positive credit card baances on their credit reports. (This discrepancy between ender and borrower debt reporting is the subject of Zinman [29] and Brown et a. [2].) 6 The spike in the persona bankruptcy rate in 25 is due to the (much anticipated) change in bankruptcy aws that made fiing for bankruptcy more difficut after that year. See, for exampe, Morgan et a. (28). 7 It is important to reiterate that this measure of new forecosures is at the individua eve. It is the number of individuas with a forecosure newy added to their credit reports, as opposed to the number of mortgages or houses with a forecosure notice, a more commony reported figure. New forecosures are counted at the individua eve by utiizing an indicator to determine whether the person has experienced a forecosure start during the past twenty-four months. Thus an individua who sequentiay defauted on severa mortgages within a Source: Federa Reserve Bank of New York Consumer Credit Pane/Equifax. Note: Student oan data prior to 23 refect some deays in the reporting of student oans by servicers to credit bureaus. This coud ead to some undercounting of student oan totas in specific periods and impact other student oan-specific measurements. However, variabiity in student oan baances prior to 23 does not materiay affect the aggregate debt time series because the variabiity is sma reative to the tota baances. Other components of househod debt are unaffected. How Are Consumers Reducing Their Debts? The unusua decine in consumers use of credit that we have observed in the ast four years raises the question of its sources. At east three major mechanisms coud be at work: 1. decining consumer use of, and demand for, credit; 2. decining ender suppy of credit; and 3. an increasing amount of nonperforming debt written off by enders as a resut of the sudden increase in defaut rates. Since a arge increase in charge-offs occurred at the same time as the decine in debt, the third mechanism is a good pace to begin. If charge-offs expain the entire reduction in debt outstanding, then there is itte need to ook further in order to understand the roes payed by the other two mechanisms. So we now turn to the question: Is the reduction in consumer indebtedness mainy attributabe to defauts, or are consumers activey reducing their debts either vountariy or because credit has become very difficut to obtain? For nonmortgage debt, answering this question is reativey straightforward. Chart 5 shows the annua change in nonmortgage debt after stripping out charge-offs. We break nonmortgage debt into two components student oans and a other oans (incuding credit cards and auto oans). Unti 29, consumers were increasing both components of their Footnote 7 continued twenty-four-month period woud appear as a new forecosure ony once: when the first mortgage went into forecosure. Further, in the case of a joint mortgage account, two separate individuas coud experience the same forecosure. 5

6 CURRENT ISSUES IN ECONOMICS AND FINANCE v Voume 19, Number 2 Chart 6 Decomposition of Changes in Mortgage Baances Biions of doars 1,2 Change in mortgage debt 1, First-ien amortization, refinances, and junior-ien baance changes First- and second-ien charge-offs 3 4 Source: Federa Reserve Bank of New York Consumer Credit Pane/Equifax. nonmortgage debt obigations each year. In 29 and 2, net nonmortgage borrowing other than student oans became negative ($68 biion and $15 biion, respectivey), but student debt continued to grow. Since consumers had been borrowing an average of more than $2 biion per year between 2 and 27 using nonmortgage debt, this indeed ooks ike a change in consumers reiance on credit other than student debt. Mortgages are more compicated because after a charge-off and forecosure, there is typicay a house that can be resod, abeit often at a discounted price. For exampe, a borrower defauts on her $, mortgage and the ender repossesses her house. The ender then reses the house to a new buyer, who pays $8, for the property, making a 2 percent down-payment and financing the remaining $64, with a thirty-year mortgage. The amount charged off on the origina borrower s credit report in this case is $,, but the net change of mortgage indebtedness from this series of events is ony -$36, (=$64, minus $,). We do not have property-eve data on housing debt, and therefore cannot infer the amount of a charged-off mortgage baance recovered by the bank through resae of the asset or the resuting contribution to aggregate mortgage debt by the new owner. This data shortcoming prevents us from aocating percentages of the mortgage debt decine to: 1) a net charge-off and 2) changes in active borrowing and repayment. We can, however, investigate active borrowing and repayment behavior among mortgage hoders over the period of aggregate debt decine. We focus on this type of behavior by dividing the change in mortgage baances (Chart 6, soid red ine) into three components Changes in mortgage debt reated to housing transactions (Chart 6, dashed bue ine) incude payoffs of mortgages associated with the norma that is, outside of 8 Our approach to decomposing the change in mortgage debt is reated to the method adopted by Greenspan and Kennedy (28) First-ien originations pus norma (non-charge-off) payoffs 8 9 forecosure sae of a house from one owner to another and the opening of new first mortgages for the purpose of buying a home, whether it is for sae by the previous owner or a ender. As expected, this series has faen sharpy as the vaue of housing transactions has decined. In this cacuation, we excude the reduction in debt attributabe to charge-offs. 2. For convenience, we aso show the negative contribution of charge-offs to mortgage baances (Chart 6, soid orange ine). We see here cear evidence of the forecosure crisis, as charge-offs on mortgage debt tota around $1.3 triion from 27 through Our fina series (Chart 6, dashed green ine), combines cash-out refinances of first iens, changes in junior-ien baances incuding HELOCs, and reguar amortizations of first-ien baances. Whie first-ien amortization reduces baances at a fairy steady pace, the other components have decined sharpy since 27. We interpret this component of baance changes as indicative of consumer responses to economic and financia conditions. Whie consumers were, on average, extracting equity and increasing their mortgage debt unti 27, they have started to pay down debt since then. Between 2 and 27, consumers increased their rea estate indebtedness by an average of $135 biion per year. In 28, this series turned negative and reached -$241 biion in 2. Taken together, the mortgage and nonmortgage series reported here indicate a major change in consumer behavior other than deinquency and defaut. Whie a kinds of borrowing contributed an annua average of about $35 biion to consumers cash fows between 2 and 27, by 2 consumers reduced their cash fows by $138 biion to reduce this debt. This represents a neary $5 biion change in annua cash fow from debt in just three years. So did consumers reduce their use of debt? Yes. Hoding aside defauts, from 27 through 2, consumers reduced their debt at a pace not seen over the ast ten years. A remaining issue is whether this reduced reiance on debt is a resut of borrowers being forced to pay down debt as credit standards tightened, or a more vountary change in saving behavior. Is the Debt Reduction Vountary? A number of measures avaiabe in the FRBNY Consumer Credit Pane provide some insight into whether consumers debt paydown is vountary, or the product of tightened credit standards. First, the Pane incudes information on the stock of open credit accounts. Chart 7 shows a substantia decrease in the number of open accounts, especiay credit card accounts, since the second-quarter 28 peak. At the peak, the data refected neary 5 miion credit card accounts in tota. By third-quarter 2, 6

7 Chart 7 Tota Number of Accounts by Loan Type Chart 8 Tota Number of New and Cosed Accounts and Inquiries Miions Bank card Thousands Open within tweve months Cosed within tweve months Auto oan Mortgage HELOC Inquiry within six months 9 Source: Federa Reserve Bank of New York Consumer Credit Pane/Equifax. Note: HELOC is home equity ine of credit. that figure had dropped neary a quarter, to 378 miion accounts, a eve that has been reativey stabe over the ast two years. This represents a striking change over a short period in U.S. consumers use of revoving credit. This net oss of miion credit card accounts does not, in itsef, te us whether consumers are choosing to reduce the number of credit cards they hod and presumaby their avaiabe credit or whether enders are restricting new credit and terminating od borrowing reationships in the aftermath of the crisis, or both. The Consumer Credit Pane series on account openings and cosings and on credit report inquiries sheds some ight on the question. The bue ine in Chart 8 shows the tota number of instament and revoving accounts opened within tweve months for each quarter of the avaiabe years of data. The rate of new account openings is high and fat through the midde of the decade, but then begins a decine in eary 28 that continues through third-quarter 2. By then, quartery account openings had faen neary 4 percent, from a peak of roughy 25 miion in each quarter from third-quarter 25 to third-quarter 27, to 158 miion in third-quarter 2. Since then, the number of account openings has increased modesty, to 177 miion, but remains we beow its peak sustained eves. Of course, new account openings may sow either because consumers seek fewer new accounts or because enders deny more appications. Data on credit report inquiries hep us make a distinction here. In genera, the type of credit report inquiry registered by our inquiry variabe is triggered by consumer appications for credit. Therefore, hoding the credit quaity of appicants constant, if the number of inquiries were to remain stabe whie the number of new account openings fe significanty, it woud be reasonaby safe to infer that creditors standards had risen, and enders were responsibe for the decine in new accounts. Source: Federa Reserve Bank of New York Consumer Credit Pane/Equifax. If quaity-adjusted inquiries track new accounts, however, then consumers woud appear to have heped generate at east some of the decine in new accounts by decreasing their appications for credit. 9 It is possibe, of course, that consumers are not appying for credit because they beieve that ending standards have tightened. In addition, the decine in inquiries is ikey to refect a reduction in credit card offers to consumers. Our data by themseves thus cannot absoutey distinguish whether suppy or demand has decined. Aggregate inquiries data do, however, offer a partia picture of consumer actions reating to account openings. The credit report inquiries series in Chart 8 (green ine) tracks the new account series quite cosey. As the tweve-month rate of new account openings fas by more than a third from its 25-7 pateau of around 25 miion to a ow of 158 miion in third-quarter 2, the rate of inquiries quite simiary drops from a pateau of around 24 miion inquiries per six-month period from 25-7 to a second-quarter 2 ow of 15 miion inquiries, before bouncing back sighty and stabiizing at around 165 miion inquiries very near the tweve-month account openings eve during the ast two years of data. The avaiabe evidence suggests that fewer appications for credit from borrowers contributed to the decine in new account openings. The drop in new account openings is ony haf of the picture for credit account transitions. The orange ine in Chart 8 shows the number of credit account cosings in the past tweve months for each quarter in the Pane. Account cosings have risen since 1999, athough not steadiy. From third-quarter 28 to thirdquarter 29, cosings underwent a sudden, steep increase from 226 miion to a peak of 376 miion. They have since moderated to 9 One caveat is that not a credit inquiries go to a credit bureaus, and our data come from a singe bureau. If our coverage of the reporting market is reativey stabe over time, then our concusions shoud be reiabe. However, if there are arge shifts in enders preferred credit reporting firms over the Pane, then these coud appear as changes in inquiry rates in our data and coud generate spurious credit report inquiry trends unreated to consumers actua appication choices. 7

8 CURRENT ISSUES IN ECONOMICS AND FINANCE v Voume 19, Number 2 Chart 9 Credits Limits and Baances for Credit Cards and HELOC Accounts Triions HELOC baance 1. Credit card baance Credit card imit HELOC imit Source: Federa Reserve Bank of New York Consumer Credit Pane/Equifax. Note: HELOC is home equity ine of credit. 185 miion, sighty beow their 24-8 eves. So at a time when new account openings were quite depressed, due at east in part to a drop in consumer appications for credit, accounts were aso being cosed in record numbers. As with account openings, we cannot infer from the account cosing rate whether borrowers or enders were primariy responsibe for cosing existing accounts. However some enders incuding Citibank, Bank of America, Advanta, and Chase Bank reportedy cosed arge numbers of accounts in 29, particuary troubed and inactive accounts. These resuts are consistent with recent survey evidence indicating that 13 percent of consumers had a credit card account cosed by their bank during 29 (Chakrabarti et a. 2). Chart 9 documents a steep, nine-quarter decrease in borrowing imits on credit card accounts from third-quarter 28 to thirdquarter 2, foowed by a recent stabiization and an ongoing (if ess pronounced) borrowing imit decrease for home equity revoving accounts. For credit cards, tota credit imits decreased by 28 percent over the nine quarters, pushing up the utiization rate (baance divided by credit imit) by 4 percentage points, despite a decine in credit card baances during the period. Credit card utiization rates have faen since 2, as consumers have paid down baances faster than imits have decined. At the same time, the reativey modest HELOC imit decrease, in the face of neary fat HELOC baances, ed to a 5 percentage point increase in utiization that has persisted through third-quarter 2. Consumers sedom request reductions in their credit imit. Further, in quartery Federa Reserve Senior Loan Officer Opinion Surveys (Board of See two American Banker artices by Kate Fitzgerad: Bi Due for 9 Account Cosings, March 8, 2, and Issuers Found to Cut Inactive Accounts, September 14, 2. Some portion of the decine in aggregate credit imits derives from the cosing of accounts Governors of the Federa Reserve System 28-), arge fractions of oan officers reported owering credit imits for existing consumer accounts from Apri 28 through eary 2. The data thus suggest that both borrowers and enders have acted to curtai consumers existing credit in the face of growing deinquency rates and broader financia market uncertainty. Concusion The FRBNY Consumer Credit Pane provides a unique ook into househod borrowing and debt payment behavior. Overa, the Pane suggests that whie tightened ending standards have payed a major roe in the decining iabiities of the househod sector, consumer-initiated reductions in debt have contributed as we. This concusion raises the further question of why househods chose to reduce their debt, especiay during a period in which many saw their incomes stagnate or drop. Here our data are ess informative, but we can combine them with other data sets to form some expanations. Two merit specia attention. The first is that the decine in consumer debt is temporay correated with the very rapid rise in unempoyment rates in the second haf of 28. Rising unempoyment affects consumer debt decisions in two potentiay offsetting ways. First, househods may choose to buid their precautionary savings (or increase their avaiabe credit) to insure their cash fows against a job oss. This action woud tend to reduce debt baances outstanding. Second, househods that do experience a job oss may use their credit accounts to smooth their consumption, eading to more borrowing. Undoubtedy, the data in Charts 5 and 6 refect the net effects of both types of househod behavior. The overa decrease in debt woud seem to suggest that the precautionary motive dominates. However, concurrent changes in asset vaues and other factors confound such concusions. As the economic outook improves, reductions in unempoyment risk shoud temper the demand for precautionary saving by househods and ower borrowing to smooth consumption during unempoyment spes, with an uncertain net effect. A second possibe source of decine in consumer debt comes from the other (asset) side of househod baance sheets. Here we focus on mortgage debt, the major form of coateraized househod debt. An important consequence of the initia increase and subsequent drop in average housing prices for househods is the dramatic fa in home equity. According to the Federa Reserve s Fow of Funds Accounts, tota equity of homeowners rose with the increase in home prices. However, it did so at a much ower rate, with homeowners equity share in their homes actuay staying reativey constant unti the end of 26. On average, for each 1 percent rise in home prices, homeowners increased their mortgage debt by 1 percent (through higher baances on first mortgages, cash-out refinances, second mortgages, and home A more detaied discussion of the potentia infuences of recent events on househod saving can be found in Chakrabarti et a. (2). 8

9 equity ines of credit), so that their equity share in their homes actuay remained constant. When home prices began to fa in 27, owners equity in househod rea estate began to fa rapidy, from amost $13.5 triion in first-quarter 26 to a itte ess than $5.3 triion in first-quarter 29 a decine in tota home equity of more than 6 percent. At the end of 29, owners equity was estimated at $6.3 triion, sti more than 5 percent beow its 26 peak. Given that the recent decine in housing prices is unprecedented since the Great Depression, we have itte evidence on the effect of such arge decines in housing weath on the demand for debt. However, if a arge decine in net worth can in fact be expected to increase the margina vaue of net saving, then the drop in owner s equity may have induced significant net saving via mortgage debt reduction, among other methods. In ight of (modest) recent improvements in credit avaiabiity, an important question is how much further consumers vountary actions wi ower aggregate debt before they begin to spend again. This question is important for the economic outook. Whie househod debt pay-down has heped improve househod baance sheets, it has aso ikey contributed to sow consumption growth since the beginning of the recession. Thus, the trajectory for consumer indebtedness has important impications for consumption and economic growth going forward. We wi continue to monitor these important trends in our data, and make key information avaiabe on our website. 13 References Avery, Robert B., Pau S. Caem, and Genn B. Canner. 23. An Overview of Consumer Data and Credit Reporting. Federa Reserve Buetin, February: Board of Governors of the Federa Reserve System Senior Loan Officer Opinion Survey on Bank Lending Practices. Apri. Avaiabe at Brown, Meta, Andrew Haughwout, Donghoon Lee, and Wibert van der Kaauw. 2. Do We Know What We Owe? A Comparison of Lender- and Borrower- Reported Debt. Federa Reserve Bank of New York Staff Reports, no. 523, October. Bucks, Brian K., Arthur B. Kennicke, Traci L. Mach, and Kevin B. Moore. 29. Changes in U.S. Famiy Finances from 24 to 27: Evidence from the Survey of Consumer Finances. Federa Reserve Buetin, February: Chakrabarti, Rajashri, Donghoon Lee, Wibert van der Kaauw, and Basit Zafar. 2. Househod Debt and Saving during the 27 Recession. Federa Reserve Bank of New York Staff Reports, no. 482, January. Federa Reserve Bank of New York. 2. Quartery Report on Househod Debt and Credit. May. Avaiabe at _economy/househodcredit/districtreport_q.pdf. Federa Reserve Statistica Reease, G.19, Consumer Credit. June 2. Avaiabe at Greenspan, Aan, and James Kennedy. 28. Sources and Uses of Equity Extracted from Homes. Oxford Review of Economic Poicy 24, no.1 (spring): -44. Hunt, Robert M. 22. The Deveopment and Reguation of Consumer Credit Reporting in America. Federa Reserve Bank of Phiadephia Working Paper no. 2-21, November. Lee, Donghoon, and Wibert van der Kaauw. 2. An Introduction to the FRBNY Consumer Credit Pane. Federa Reserve Bank of New York Staff Reports, no. 479, November. Morgan, Donad, Benjamin Iverson, and Matthew Botsch. 28. Seismic Effects of the Bankruptcy Reform. Federa Reserve Bank of New York Staff Reports, no. 358, November. Suivan, James X. 28. Borrowing during Unempoyment: Unsecured Debt as a Safety Net. Journa of Human Resources 43 no. 2 (spring): Whitehouse, Mark. 2. Defauts Account for Most of Pared Down Debt, Wa Street Journa Rea Time Economics Bog, September economics/2/9/18/number-of-the-week-defauts-account-for-most-of -pared-down-debt/; ast accessed October 1, 2. Zinman, Jonathan. 29. Where Is the Missing Credit Card Debt? Cues and Impications. Review of Income and Weath 55, no. 2 (June): See nationaindicators.htm ( Househod Credit Conditions ) for quartery reeases of many of the key data series and occasiona suppementa reports. About the Authors Meta Brown and Donghoon Lee are senior economists, Andrew Haughwout a vice president, and Wibert van der Kaauw a senior vice president in the Microeconomic Studies Function of the Federa Reserve Bank of New York. Current Issues in Economics and Finance is pubished by the Research and Statistics Group of the Federa Reserve Bank of New York. Linda Godberg and Thomas Kitgaard are the editors of the series. Editoria Staff: Vaerie LaPorte, Mike De Mott, Michee Baier, Karen Carter, Anna Snider Production: Jane Urry, Jessica Iannuzzi, David Rosenberg Back issues of Current Issues are avaiabe at The views expressed in this artice are those of the authors and do not necessariy refect the position of the Federa Reserve Bank of New York or the Federa Reserve System. 9

10 CURRENT ISSUES IN ECONOMICS AND FINANCE v Voume 19, Number 2 RELATED READINGS FROM THE FEDERAL RESERVE BANK OF NEW YORK Pubications of the Research and Statistics Group Avaiabe at _annuas/index.htm An Introduction to the FRBNY Consumer Credit Pane Donghoon Lee and Wibert van der Kaauw Staff Reports, no. 479, November 2 In this paper, the authors introduce the FRBNY Consumer Credit Pane, a new ongitudina database with detaied information on consumer debt and credit. The pane uses a unique sampe design and information derived from consumer credit reports to track individuas and househods access to and use of credit at a quartery frequency. In any given quarter ranging from the first quarter of 1999 to the present, the pane can be used to compute nationay representative estimates of the eves and changes in various aspects of individua and househod iabiities. In addition to describing the sampe design, the use of sampe weights, and the credit report information incuded in the database, the authors provide some comparisons of popuation statistics and consumer debt estimates derived from the pane with those based on data from the American Community Survey and the Fow of Funds Accounts of the United States. Househod Debt and Credit Report Avaiabe at The Federa Reserve Bank of New York s Househod Debt and Credit Report provides a quartery snapshot of househod trends in borrowing and indebtedness, incuding data about mortgages, student oans, credit cards, auto oans, and deinquencies. The report aims to hep community groups, sma businesses, state and oca governments, and the pubic to better understand, monitor, and respond to trends in borrowing and indebtedness at the househod eve. The Liberty Street Economics Bog Avaiabe at Just Reeased: Press Briefing on Househod Debt and Credit Meta Brown, Andrew Haughwout, Donghoon Lee, Joee Scay, and Wibert van der Kaauw Liberty Street Economics bog, February 28, 213 Just Reeased: Has Househod Deeveraging Continued? Andrew Haughwout, Donghoon Lee, Joee Scay, and Wibert van der Kaauw Liberty Street Economics bog, August 29, 2 The boggers use newy avaiabe 2 and 2 FRBNY Consumer Credit Pane data to update their anaysis on househod deeveraging (presented in a March 21, 2, Liberty Street Economics bog post). Grading Student Loans Meta Brown, Andrew Haughwout, Donghoon Lee, Maricar Mabutas, and Wibert van der Kaauw Liberty Street Economics bog, March 5, 2 Brown, Haughwout, Lee, Mabutas, and van der Kaauw examine the overa student oan debt market as of third-quarter 2, and find it ikey that deinquency rates for these oans are understated. Fip This House : Investor Specuation and the Housing Bubbe Andrew Haughwout, Donghoon Lee, Joseph Tracy, and Wibert van der Kaauw Liberty Street Economics bog, December 5, 2 During 24-6, rea estate investors used financia everage (mortgage credit) to purchase mutipe residentia properties, which ikey heped push house prices up. But when prices turned down in eary 26, these investors defauted in arge numbers and contributed to the intensity of the housing bust. Have Consumers Been Deeveraging? Meta Brown, Andrew Haughwout, Donghoon Lee, and Wibert van der Kaauw Liberty Street Economics bog, March 21, 2 Since its peak in summer 28, U.S. consumers indebtedness has faen by more than a triion doars. The boggers demonstrate that a significant part of the debt reduction was produced by consumers borrowing ess and paying off debt more quicky a process often caed deeveraging. The resuts of the New York Fed s atest Quartery Report on Househod Debt and Credit with specia emphasis on househod borrowing and student oans are presented.