Inersae Ris Sharing and Morgage Loan Securiizaion Pu Liu Deparmen of Finance* Harold A. Dulan Chair Professor in Capial Formaion Rober E. Kennedy Chair Professor in Invesmen Sam M. Walon College of Business Universiy of Aransas Fayeeville, AR 72701 pliu@walon.uar.edu 479-575-6095 479-575-8407 FAX Rodrigo J. Hernandez Assisan Professor of Finance Radford Universiy rjhernand@radford.edu 540-831-6454 Yingying Shao Assisan Professor of Finance Deparmen of Finance College of Business and Economics Towson Universiy yshao@owson.edu 410-704-3839 This Draf: December 27, 2011 * Corresponding auhor. Do no cie wihou auhors permission. Commens are welcomed.
Inersae Ris Sharing and Morgage Loan Securiizaion Absrac This paper conribues o he coninuing debae on he impac of financial innovaions on he real economy. In paricular, we examine he role of bans morgage loan securiizaions in aggregae inersae ris sharing. Using daa for U. S. bans morgage loans securiizaions during 1989-2008, we idenify consumpion smoohing as an imporan channel hrough which loan securiizaion affecs he inersae ris sharing. The resuls in he paper sugges ha he posiive relaionship beween loan securiizaions and aggregae ris sharing enhances, raher han jeopardizes, financial sabiliy. Key Words: Securiizaion, Ris Sharing, Morgage Loans, Consumpion Smoohing JEL Classificaion: G18, G21 EFMA Classificaion: 560, 440, 510, 520
Inersae Ris Sharing and Morgage Loan Securiizaion 1. Inroducion Morgage loans are major funding sources for households in heir house purchases and morgage deb paymens accoun for abou 60 percen of oal household deb paymens 1. Bans issuing morgage loans ypically face significan riss arising from he flucuaion in housing value and he poenial defaul of morgage debs, and bans hisorically rely on acive managemen of morgage loan porfolios o manage he credi riss of hese loans. Securiizaion of morgage loans, as a newly developed financial ool o allow morgage loans o be raded naion-wide, provides anoher vehicle for bans o effecively manage he credi ris of he loan porfolios by diversifying he riss across sae boarders. There has been an ongoing debae over he role ha asse securiizaions have played in he economy. Before he 2007-2009 financial crisis, some policy maers and researchers argue ha securiizaions allow bans o disperse credi ris, reduce informaion asymmery, and herefore enhance financial sabiliy (Hill, 1997; Greenspan, 2005) 2, while oher sudies have suggesed poenial agency problems and disored incenives inroduced by 1 See he Consumer Expendiure Survey conduced by he Bureau of Labor Saisics (BLS) for deailed analysis. As par of he expendiure daa collecion, he BLS ass households o repor paymens on household deb, including morgage debs, vehicle loans, and oher consumer debs. 2 The former Federal Reserve Sysem Chairman Alan Greenspan (2005) remared ha Perhaps he mos significan developmen in financial mares over he pas en years has been he rapid developmen of credi derivaives. Moreover, his growh has been accompanied by significan produc innovaion, noably he developmen of synheic collaeralized deb obligaions (CDOs), As is generally acnowledged, he developmen of credi derivaives has conribued o he sabiliy of he baning sysem by allowing bans, especially he larges, sysemically imporan bans, o measure and manage heir credi riss more effecively (Remars by Chairman Alan Greenspan: Ris Transfer and Financial Sabiliy, o he Federal Reserve Ban of Chicago's Fory-Firs Annual Conference on Ban Srucure, Chicago, Illinois, May 5, 2005). 3
securiizaion (Dahiya e al., 2003). Some sudies even sugges ha he recen 2007-2009 financial crisis was caused by morgage securiizaions for allowing asses of poor credi qualiy o spread o unsophisicaed and unproeced invesors, and evenually leading o he hisorical financial urmoil (Ban of Inernaional Selemen, 2008; Brunnermeier, 2009; Taylor, 2008). This paper conribues o he coninuing debae by examining he effec of he securiizaions of morgage loans made by US bans on aggregae ris sharing. We focus on morgage loans because morgage loans end o be made o local home buyers whose deb paymens heavily rely on heir wage compensaion generaed in he same communiy in which he ban operaes. Hence bans issuing morgage loans are exposed o local oupu ris ha is hard o be diversified away. If securiizaions, as designed for he purposes for morgage loans o be raded naionally, hey would faciliae ris diversificaion which in urn could increase credi supply and herefore have a posiive effec on real economy (Lousina and Srahan, 2009; Mian and Sufi, 2009; Demyany and Van Hermer, 2011; Keys e al., 2009 and 2010). Furhermore, he exan lieraure on ris sharing suggess he developmen in credi mares conribues o aggregae ris sharing by prevening reducions in consumpions. We herefore hypohesize ha he posiive impac of morgage securiizaions on credi supply furher conribues o ris sharing hrough he consumpion smoohing channel and hus we conduc he analyses by using he convenional ess in he ris sharing and consumpion smoohing lieraure. 4
Our main objecive is o examine wheher ban s securiizaion of morgage loans have a posiive impac on inersae ris sharing. We esimae he effec of loan securiizaions on inersae ris sharing by using he annual sae-level daa compiled from U.S. Ban Call Repors for he period 1989-2008. The resuls in our sudy sugges ha bans sales and securiizaion of morgage loans have a significanly posiive effec on ris sharing hrough consumpion smoohing. Moreover, he resuls reveal ha he impac of loan securiizaion on ris sharing is more pronounced in saes where he housing mare value is relaively low. The paper conribues o he exan lieraure on ris sharing. Very few papers have empirically esimaed he effec of financial innovaions, specifically, securiizaions, on real economy such as privae expendiure and we examine his relaionship in his paper. The paper is also closely relaed o he sudy of Demyany e al. (2007) which documens he impac of baning deregulaion on inersae ris sharing in he U.S. The sudy of Demyany e al. (2007) provides evidence showing ha he developmen of morgage-baced asses enhances personal income insurance, while our paper suggess an alernaive channel of ris sharing hrough consumpion smoohing. The res of he paper is organized as follows: In Secion 1 we discuss how he sales and securiizaions of morgage loans conribue o he inersae ris sharing hrough consumpion smooh channel. In Secion 2 we presen he empirical specificaion for 5
esing for ris sharing. Daa descripions are provided in Secion 3 and empirical resuls are presened in Secion 4. We conclude he paper in Secion 5. 2. Moivaion and lieraure review Before he 2007-2009 financial crisis, proponens of securiizaions mainly suppor securiizaion for is benefis in allowing bans o acively manage he credi ris and herefore in improving financial sabiliy. For insance, Neal (1996) wroe ha he developmen of mares for securiized asses and for loan sales has provided anoher mehod for managing credi ris. In he asse securiizaion approach, bonds or loans wih credi ris are pooled ogeher and sold o an ouside invesor... From an invesor s perspecive, purchasing par of he pacage is aracive because he diversificaion across many loans reduces he overall credi ris. In addiion, o he exen ha reurns from he pacage are no closely correlaed wih he invesor s oher holdings, diversificaion allows he invesor o reduce he credi ris of his overall porfolio (Economic Review, Federal Reserve Ban of Kansas Ciy, Second Quarer 1996, pp.18-19). Moreover, he former Federal Reserve Sysem Vice Chairman Donald Kohn (2007) poined ou ha he securiizaion of morgages and oher asses has been ransforming regulaed deposiory insiuions from holders of ineres rae and credi ris o originaors and disribuors of such ris There are good reasons o hin ha hese developmens have made he financial sysem more resilien o shocs originaing in he real economy and have made he economy less vulnerable o shocs ha sar in he financial sysem. (Speech a he Federal Reserve Ban of Alana's 2007 Financial Mares Conference, Sea Island, Georgia, May 16, 2007) However, he exan lieraure has also suggesed poenial agency problems and disored incenives inroduced by securiizaion. For example, Dahiya, Puri and Saunders (2003) show ha soc mare responded negaively o firms whose loans were sold by heir 6
lending bans. They also find ha a large porion of hese firms even filed for banrupcy a few years afer he loan sales, suggesing ha bans may have prior informaion abou he poenially poor performance and herefore sell hese loans o avoid fuure losses. Some sudies afer he recen 2007-2009 financial crisis sugges ha he process of asse securiizaion inroduces several layers of agency problems ha conribued o he financial crisis (Ashcraf and Schuermann, 2008). Duffie (2008) poins ou ha loan sales and securiizaions reduce bans incenive o monior and manage he credi ris of he securiized loans. The lac of incenive o monior leads o higher defaul rae in securiized loans han he un-securiized loans. Furhermore, Pisorsi e al. (2010) repor ha given ha a morgage loan becomes seriously delinquen, securiized loans end o experience a significanly higher foreclosure rae han similar loans held by bans. The resuls of hese sudies sugges ha he sale of loans and asses ends o be associaed wih higher level of foreclosure rae, poor performance of he firms, leading o a lower level of monioring on borrowers, lower level of ris managemen, and herefore conribues o financial insabiliy. While he above sudies have suggesed he poenial agency problem associaed wih asse securiizaion, oher sudies have idenified he benefis of securiizaions by examining he impac of securiizaion on bans lending behavior and credi supply. These sudies have documened ha securiizaions, when used by bans o acively managemen heir liquidiy and credi ris, have led o a higher level of credi supply (Lousina and Srahan, 2009; Mian and Sufi, 2009; Demyany and Van Hermer, 2011; Keys e al., 2009 and 2010). 7
In addiion, he lieraure on ris sharing has also suggesed ha increased borrowing and lending in he credi mares conribue o aggregae ris sharing from he smoohening of consumpions (Asdrubali e al., 1996). In his sream of research, sudies have found ha aggregae consumpion is posiively correlaed wih he availabiliy of household debs. For insance, Bacchea and Gerlach (1997) repor ha expeced growh in morgage and consumer credi is posiively correlaed wih he growh in non-durable goods and services expendiures, and McCarhy (1997) finds a significan lin beween availabiliy of credi and durable goods expendiures. Coulibaly and Li (2006) observe ha while households do no increase heir non-durable consumpion following he reiremen of heir morgage, hey do, however, increase durable goods consumpions, such as home furnishings and enerainmen equipmen. Based on above discussions, we conjecure ha he securiizaions of morgage loans conribue o inersae ris sharing hough a consumpion smoohening channel. When bans pool and securiize heir morgage loans, hey can ransfer some of he sae-specific oupu riss o he financial insiuions in oher saes. The reducion in credi ris allows bans o originae more loans or a a lower cos which may preven reducion in consumpions, and leading o smoohened consumpions. 2. Empirical mehodology 8
To es he impac of securiizaion of morgage loans on inersae consumpion ris sharing, we follow he mehodology developed by Asdrubali e al (1996) and used exensively in he lieraure. The consumpion ris sharing across saes is measured hrough a panel regression model in he following form: ln Consumpio (1) n lngsp Where ln Consumpion denoes he sae-specific growh raes of privae consumpion for sae in year, and ln GSP is he sae-specific growh raes of gross sae produc for sae in year, and All of he variables are measured in per capia erms. The growh raes of real per capia variables are calculaed as he firs differences of he naural logarihm of per capia-level values. The sae-specific variables are consruced using sae-level variables minus he mean across saes minus he mean across ime. Based on Asdrubali e al. (1996), if full ris sharing is achieved via consumpion smoohing, all saes should have idenical growh raes of consumpion because he consumpion does no co-move wih oupu; and a one-o-one co-movemen beween consumpion and oupu implies zero ris reducion hrough consumpion smoohing. Thus he coefficien in equaion (1) measures he uninsured idiosyncraic oupu ris, a value of =0 indicaes perfec insurance hrough consumpion smoohing, and a value of =1 indicaes zero insurance hrough consumpion smoohing. The objecive of his sudy is o examine wheher ban s morgage loan securiizaion conribues o he reducion of idiosyncraic oupu, we hus include an addiional variable 9
MBS, which represens bans aciviies in securiizing morgage loans, and we also allow his variable vary by sae and over ime in he following regression equaion: ln Consumpion 0 ln GSP 1 MBS ln MBS 2 GSP (2) Where MBS measures he degree of securiizaion of morgage loan in sae in year. I is defined as he raio of he aggregae ousanding balance of securiized morgage loans of a sae in year o he aggregae amoun of oal ousanding morgage loans of he same sae in he same year. respecively. and are dummy variables measuring sae and ime fixed effecs, In his regression, he ey variable of ineres is MBS ln GSP, he ineracion erm beween oupu growh and bans securiizaion of morgage loans. The regression coefficien 1 measures he uninsured idiosyncraic oupu ris associaed wih a one-uni increase in morgage loan securiizaion. The regression coefficien 0 measures he average degree of uninsured idiosyncraic oupu ris wihou morgage loan securiizaion, and 0 + 1 measures he oal uninsured idiosyncraic oupu ris afer bans have engaged in securiizaion aciviies. Noe ha he regression coefficien 2 measures he conribuion of morgage loan securiizaion o he average consumpion growh, which is no of ineres of his sudy. We include i following he normal regression echnique ha includes he linear erm accompanied wih he ineracion erm. 10
3. Daa and sample saisics We collec annual sae-level daa for all of he 50 U.S. saes and Washingon D.C. from 1995 o 2008 from various sources. The daa abou bans loan issuance and securiizaion aciviies are compiled from U.S. Ban Call Repors. Macroeconomic daa are mainly obained from Bureau of Economic Analysis (BEA) and U.S. Census Bureau. The variables in he analysis are described as follows: Gross Sae Produc (GSP): GSP is divided by populaion in a given sae and deflaed by he consumer price index o obain real per capia sae gross domesic produc. Sae Personal Income (SPI): Similarly, we use BEA sae-level personal income per capia deflaed by consumer prices o obain real per capia personal income by sae. Housing Price Index: To measure he local housing mare, we use he Housing Price Index, obained from GeoFRED daabase supplied by Federal Reserve Ban of S. Louis. Based on he housing price index in 1988, we divide he saes ino wo equal-sized caegories (upper half and lower half) and refer hese wo caegories as high- housing value vs. low- housing value caegories. We pariion he sample in his way o capure he impac of he size hreshold in morgage loan securiizaion on bans securiizaion aciviies. A large par of morgage securiizaion is conduced hrough he governmen-sponsored enerprises (GSEs, i.e., Fannie Mae and Freddie Mac). However, by regulaion, he GSEs only buy morgages below a given size hreshold (he jumbo loan cuoff), herefore morgages below his hreshold are more liely o be securiized han hose above he 11
hreshold. Because he daa on individual morgage loans are no publically available, we use housing price index as a proxy of he average size of he morgage loans. The idea here is ha in low-value housing mares, he average size of morgages may o be smaller han ha in high-value housing mares, hence he securiizaions of morgage loans may be more acive, and he ris sharing conribuion of securiizaions would be prominen in low-housing value saes. Morgage loan Securiizaion (MBS): We collec annual daa on he ousanding balance of oal residenial morgage loans and securiized morgage loans from Call repor. We firs obain he daa for individual bans and hen aggregae o he sae level. We measure bans aciviies of morgage loan securiizaions in a given sae for a given year by using he raio of he aggregae ousanding balance of securiized loans o he aggregae amoun of oal ousanding morgage loans in ha sae for ha year. [Inser Figure 1 abou here] Figure 1 displays he average growh raes of GSP and SPI across saes for each year from 1989 o 2008. The figure reveals a lead-lag relaionship beween he GSP and he SPI growh raes. For example, he declines in GSP growh rae in year 2000, 2002, and 2007 are followed by lagged reducions in he personal income growh rae soon afer, reflecing he effec of inersae income insurance. In Figure 2 we show he hea maps for he average GSP growh rae and SPI growh rae for he sudy period 1989-2008. The maps also sugges a close relaionship beween he GSP and SPI growh raes. 12
[Inser Figure 2 abou here] [Inser Table 1 abou here] Table 1 presens he mean raios of securiized morgage loans o oal morgage loans, he average GSP growh raes, and he average SPI growh raes for each sae and Washingon DC from 1989 o 2008. I is eviden ha bans manage heir credi ris in morgage loans acively in recen years. The average raio of securiized morgage loans across sae over ime is abou en percen. In addiion, he daa reveal ha he use of securiizaion varies significanly across saes. Among he 50 saes and Washing D.C in he sample, all he saes have bans securiizing morgage loans wih various exens. Figure 3 plos he ime average of bans morgage loan securiizaion aciviies across saes. I shows ha bans morgage loan securiizaion aciviies are volaile wih a significan drop in he early 90s, reflecing he dramaic impac of he economic recession and he credi crunch; he securiizaion aciviies reached o he pea in years around 2002, a ime when he bubble sared o form in he housing mare. Figure 4 presens he hea map of he average morgage loan securiizaion for each sae based on he ime average over he sample period. Similar o he paern revealed in Table 1, he degree of loan securiizaion varies significanly across saes. [Inser Figure 3 abou here] [Inser Figure 4 abou here] 13
4. Empirical resuls To examine wheher bans managemen of credi ris in morgage loans hrough securiizaion conribued o inersae ris sharing hrough smoohing personal consumpion, we esimae he regression Equaion (2). Follow he mehodology used by Demyani, Osergaard, and Sorensen (2007), we use a wo-sep GLS: we firs run a pooled OLS regression o esimae he variance of he error erms based on he residuals; we hen run he second regression, weighing he sae-level variables by he esimaed sandard error. Table 3 presens he resuls of he second sep of regression. In reporing he resuls, we muliply he esimaed value by 100, and refer o as he percenage of ris shared. [Inser Table 2 abou here] The resuls of Table 2 sugges ha loan securiizaion significanly conribues o consumpion smoohing. The average impac of idiosyncraic oupu ris on consumpion, as measured by 0, is abou 19% wihou he securiizaion of morgage loans. This esimae is boh economically and saisically significan. The regression coefficien for ey variable in he sudy, he ineracion erm MBS ln GSP, 1 (=- 14%) has he expeced negaive sign and is saisically significan. The resuls show ha wih he securiizaion of morgage loans, he degree of he shoc o he consumpion due o idiosyncraic oupu ris reduces o 5% (19% -14%), a significan improvemen in smoohing he consumpion. [Inser Table 3 abou here] 14
We nex examine wheher he consumpion smoohing effec of morgage loan securiizaion is condiional o he housing mare. We separae he saes ino wo caegories based on he value of local housing mare ( high-housing value vs. low-housing value saes) as described in Secion 2 of he paper. We hen esimae he relaionship beween consumpion smoohing and securiizaion separaely for he wo sub-groups and repor he resuls in Table 3. For saes where housing values are low (i.e. high- housing value saes), i is liely ha more morgage loans are below he jumbo loan cuoff and can be sold o he GSEs, so bans securiizaion aciviies may be more acive han in he saes where he average size of morgage loans is larger, as indicaed by he high housing price. Therefore, i is liely ha he benefi of securiizaion in smoohing personal consumpion in low-housing value saes would be larger han in high-housing value saes. In Panel A of Table 3 we presen he resuls of Equaion (2) for he high-housing value saes, and Panel B shows he resuls for he low-housing value saes. The resuls are consisen wih he expecaions. While boh high-housing value and low-housing value saes show reducion in oupu ris by morgage loan securiizaion, only he low-housing value saes have a significan regression coefficien associaed wih MBS ln GSP. Moreover, he average income insurance wihou loan securiizaion, represened by regression coefficien, is higher in saes wih high value housing mare (63%) han in 0 saes wih low value housing mare (18%), a resul consisen wih he hypohesis ha beer ris sharing hrough consumpion smoohing. 15
5. Conclusions In his paper, we invesigae he benefis of ban s aciviies in loan securiizaion from he borrower s perspecive. Specifically, we examine wheher he securiizaion of major ype of loans, morgage loans, conribues o he consumer s ris sharing across saes. We find ha he securiizaion of morgage loans helps reduce he idiosyncraic oupu ris o consumers by smoohing heir consumpions. There has been an ongoing debae on wheher derivaive securiies in general and securiizaions in paricular enhance he growh of economy or cause financial insabiliy. This paper maes conribuions o he lieraure in providing he evidence o he debae on he pros and cons of securiizaions o he economy. To he bes of our nowledge, his is he firs sudy ha examines he relaionship beween loan securiizaions and real economy and he resuls in he sudy sugges ha bans loan securiizaion faciliaes aggregae ris sharing hrough consumpion smoohing. The resuls in his paper hence sugges ha he posiive relaionship beween loan securiizaions and aggregae ris sharing enhances, raher han jeopardizes, financial sabiliy. 16
REFERENCES Aerlof, G. (1970). The Mare for Lemons: Qualiaive Uncerainy and he Mare Mechanism, Quarerly Journal of Economics, 89, pp. 488-500. Asdrubali, P., Sorensen, B. and Yosha, O. (1996). Channels of inersae ris sharing: Unies Saes 1963-1990, Quarerly Journal of Economics, 111, pp.1081-1110. Ashcraf, A. and Schuermann, T. (2008). Undersanding he securiizaion of subprime morgage credi, Saff Repor No. 318, Federal Reserve Ban of New Yor, available a hp://www.newyorfed.org/research/saff_repors/sr318.pdf Ban for Inernaional Selemens, (2008). 78h Annual Repor, Basel: BIS. Bacchea,P. and Gerlach,S. (1997). Consumpion and credi consrains: Inernaional evidence, Journal of Moneary Economics, 40, pp. 207-238. Berger, A. and Udell,G. (1995). Relaionship lending and lines of credi in small firm finance, Journal of Business, 68, pp.351-381. Brunnermeier, M.(2009). Deciphering he liquidiy and credi crunch 2007-2008, Journal of Economic Perspecives, 23, pp.77-100. Coulibaly B. and Li,G. (2006). Do homeowners increase consumpion afer he las morgage paymen? An alernaive es of he permanen income hypohesis. The Review of Economics and Saisics, 88, pp.10-10. Dahiya, S., Puri, M. and Saunders, A.(2003). Ban borrowers and loan sales: New evidence on he uniqueness of ban loans, Journal of Business, 76, pp.563-82. Demyany, Y., Osergaard, C.and Sorensen, B. (2007). U.S. baning deregulaion, small business, and inersae insurance of personal income, Journal of Finance, 62, pp. 2763-2801. Demyany, Y., Van Hemer, O., (2011). Undersanding he subprime morgage crisis. Review of Financial Sudies, 24, 1848-1880. 17
Duffie, D. (2008) Innovaions in credi ris ransfer: Implicaions for financial sabiliy, BIS Woring Paper 255. hp://www.bis.org/publ/wor255.pdf Goron, G. and Souleles,N.(2006). Special purpose vehicles and Securiizaion, in (Rene Sulz and M. Carey, eds), The Riss of Financial Insiuions, pp. 549 97, Chicago: Universiy of Chicago Press. Greenspan, A. (2005). Ris ransfer and financial sabiliy, Remars by Chairman o he Federal Reserve Ban of Chicago's Fory-firs Annual Conference on Ban Srucure, Chicago, Illinois, (via saellie) on May 5, 2005. Hill, C. (1997). Securiizaion: A low-cos sweeener for lemons, Journal of Applied Corporae Finance, 10, pp. 64-71. Keys, B., Muherjee, T., Seru, A., Vig, V. (2010). Did securiizaion lead o lax screening: evidence from subprime loans, Quarerly Journal of Economics 125, 307 362. Keys, B., Muherjee, T., Seru, A., Vig, V. (2009). Financial regulaion and securiizaion: evidence from subprime loans, Journal of Moneary Economics 56, 700 720. Kohn, D. (2007). Financial sabiliy and policy issues, speech a he Federal Reserve Ban of Alana's 2007 Financial Mares Conference, Sea Island, Georgia. Lousina, E., Srahan, P. (2009). Securiizaion and he declining impac of ban finance on loan supply: evidence from morgage originaions, Journal of Finance, 64, 861-889. McCarhy, J.(1997). Deb, delinquencies and consumer spending, Curren Issues in Economicsand Finance, Federal Reserve Ban of New Yor. Mian, A., Sufi, A. (2009). The consequences of morgage credi expansion: evidence from he U.S. morgage defaul crisis, Quarerly Journal of Economics, 124, pp.1449 1496. Neal, R. (1996). Credi derivaives: New financial insrumens for conrolling credi ris, Economic Review, Federal Reserve Ban of Kansas Ciy, Second Quarer, 15-27. Pisorsi, T., Seru, A. and Vig, V. (2010). Securiizaion and disressed loan renegoiaion: Evidence from he subprime morgage crisis, Journal of Financial Economics, 97, pp.369-397. 18
Shin, H., (2009). Securiizaion and financial sabiliy, The Economic Journal, 119, pp. 309-332 Taylor, J. (2009). The financial Crisis and he policy responses: An empirical analysis of wha wen wrong, Sanford Universiy Hoover Insiuion, Roc Cener Woring Paper No. 30. 19
Table 1: Summary Saisics: This able summarizes bans average securiizaion of morgage loans and small business loans across saes using daa from Call Repor. The daa for morgage loans are from 1989 o 2008, while he earlies daa for small business loan are from 1995. For each ype of loan, we aggregae he oal morgage loans and loan securiizaions a he sae level by summing he amoun of loans repored by individual bans in he sae. The percenage of securiized loans in each sae is measured as he raio of securiized loans o he oal loans originaed in ha sae. We also repor he mean GSP growh rae and privae consumpion growh rae for each sae during 1989-2008. Securiized morgage loans as a % of oal GSP Growh (%) SPI Growh (%) Sae morgage loans Alabama 6.17 1.35 4.35 Alasa 0.75-0.12 3.77 Arizona 0.54 1.18 3.93 Aransas 39.38 1.44 4.52 California 41.70 0.97 3.99 Colorado 1.57 1.84 4.47 Connecicu 1.38 1.31 4.19 Delaware 43.25 2.00 3.57 Disric of Columbia 0.48 2.58 5.37 Florida 3.70 1.20 3.89 Georgia 46.97 0.89 3.89 Hawaii 6.85 0.71 3.84 Idaho 0.35 1.08 4.29 Illinois 1.72 1.26 4.04 Indiana 1.57 1.13 3.84 Iowa 2.53 1.86 4.31 Kansas 2.42 1.56 4.34 Kenucy 0.35 1.14 4.17 Louisiana 10.95 2.05 5.05 Maine 2.92 0.98 4.11 Maryland 4.07 1.22 4.21 Massachuses 8.16 1.28 4.43 Michigan 28.19 0.67 3.47 Minnesoa 1.18 1.61 4.37 Mississippi 0.87 1.34 4.68 Missouri 0.53 0.94 4.06 Monana 1.04 1.70 4.56 Nebrasa 1.76 1.78 4.46 ( o be coninued) 20
Securiized morgage loans as a % of oal GSP Growh (%) SPI Growh (%) Sae morgage loans Nevada 38.30 0.94 4.03 New Hampshire 5.68 1.20 4.04 New Jersey 1.33 1.08 4.17 New Mexico 13.37 1.89 4.58 New Yor 28.72 1.44 4.14 Norh Carolina 11.80 1.23 4.02 Norh Daoa 0.63 2.98 5.37 Ohio 12.48 1.47 4.10 Olahoma 4.63 1.33 3.99 Oregon 0.49 0.78 4.12 Pennsylvania 9.01 2.53 5.11 Rhode Island 0.97 1.29 4.21 Souh Carolina 1.01 1.88 4.47 Souh Daoa 43.15 1.86 4.36 Tennessee 41.39 1.03 4.27 Texas 9.18 1.51 4.28 Uah 0.44 1.50 4.45 Vermon 2.04 1.57 4.52 Virginia 3.80 1.39 4.14 Washingon 7.60 2.24 5.72 Wes Virginia 1.16 1.47 4.10 Wisconsin 5.42 1.33 3.99 Wyoming 1.14 0.78 4.12 Average 9.90 21
Table 2: Morgage Loan Securiizaion and Ris Sharing This able presens he resuls of he regression: lnconsumpio n 0 lngsp 1 MBS lngsp 2MBS, where ln Consumpion denoes he sae-specific growh raes of privae consumpion for sae in year and ln GSP is he sae-specific growh raes of gross sae produc for sae in year. ln Consumpion and ln GSP are measured in per capia erms. The growh raes of real per capia variables are calculaed as he firs differences of he naural log of per capia-level values. MBS measures he degree of morgage loan securiizaion of sae in year and is defined as he raio of he aggregae ousanding balance of securiized morgage loans in a sae o he aggregae amoun of oal ousanding morgage loans in ha sae. dummy variables measuring sae and ime fixed effecs, respecively. and are Variables 0 Expeced Sign Regression Coefficien -Saisics p-value ln GSP ( ) + 18.66 2.17 0.0301 MBS ln GSP ( ) 1 - -14.06-1.92 0.0559 MBS - -0.46-1.58 0.1142 Sae Dummies Year Dummies Yes Yes N 1020 R 2 0.35 22
Table 3: Morgage Loan Securiizaion and Ris Sharing: subsamples This able presens he resuls of he regression using wo subsamples based on he housing mare: lnconsumpio n 0 lngsp 1 MBS lngsp 2MBS, where ln Consumpion denoes he sae-specific growh raes of privae consumpion for sae in year and ln GSP is he sae-specific growh raes of gross sae produc for sae in year. ln Consumpion and ln GSP are measured in per capia erms. The growh raes of real per capia variables are calculaed as he firs differences of he naural log of per capia-level values. MBS measures he degree of morgage loan securiizaion of sae in year and is defined as he raio of he aggregae ousanding balance of securiized morgage loans in a sae o he aggregae amoun of oal ousanding morgage loans in ha sae. and are dummy variables measuring sae and ime fixed effecs, respecively. We spli he saes ino wo equal-sized caegories based on he housing mare: High (Low)-Housing value are saes in which housing mares are more (less) valued, as defined in Secion III of he paper. The degree of value of local housing mare in a given sae is based on Housing Price Index (HPI) daa obained from GeoFRED daa base from Federal Reserve Ban of S. Louis. Variables Regression Coefficien -Saisics p-value Panel A: High-housing Value Saes ln GSP ( ) 63.17 3.45 0.0007 MBS 0 ln GSP ( ) 1-15.39-1.10 0.2723 MBS -0.484-1.05 0.2965 Sae Dummies Year Dummies Yes Yes N 520 R 2 0.32 Panel B: Low-housing Value Saes ln GSP ( ) 18.65 2.17 0.0301 MBS 0 ln GSP ( ) 1-14.06-1.92 0.0559 MBS -0.463-1.58 0.1142 Sae Dummies Year Dummies Yes Yes N 500 R 2 0.46 23
Figure 1: The Growh raes of GSP and SPI over Time This figure shows he average growh rae of GSP and SPI across he 50 saes and Washingon D.C. during 1990-2008. Growh (%) 7 6 5 4 3 2 1 0-1 GSP SPI -2 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Year 24
Figure 2: The Growh rae of GSP and SPI across Saes The following wo figures show he average growh raes of GSP and SPI respecively over he period from 1990 o 2008 in each of he 50 saes and Washingon D.C., respecively. Panel A: GSP Growh (%) -0.12-0.97 0.98-1.20 1.20 1.33 1.34 1.54 1.55 1.86 1.86 2.97 Panel B: Sae Personal Income (SPI) Growh (%) 3.47 3.89 3.92 4.06 4.10 4.19 4.21 4.36 4.38 4.52 4.54 5.72 25
Figure 3: Ban s Loan Securiizaion over Time This figure shows bans morgage securiizaion aciviies for he 50 saes and Washingon D.C. during 1989-2008. The oal issued loans and securiized loans are aggregaed a he sae level by summing he amoun of loans repored by individual bans in each sae. The percenage of securiized loans in a given sae is measured by he raio of oal securiized morgage loans o he oal originaed morgage loans in ha sae. Percenage (%) 30 20 10 0 1980 1990 2000 2010 year 26
Figure 4: Ban s Loan Securiizaion across Saes This graph shows average bans morgage loan securiizaion during 1989-2008 in each of he 50 saes and Washingon D.C. The oal loans and securiized loans are aggregaed a he sae level by summing he amoun of loans repored by individual bans in each sae. The percenage of loan securiizaion in a given sae is measured by he raio of oal securiized morgage loans o he oal morgage loans in ha sae. 0.35-0.63 0.75-1.33 1.38-2.53 2.92-6.85 7.59-13.37 28.19-46.97 27