Monetary Polcy and the Redstrbuton Channel Adren Auclert MIT MFM Meetng, NYU Stern May 31, 2014 Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 1 / 16
Introducton Polcymakers and the redstrbuton channel Reductons n mortgage rates stmulate the economy substantally as households [can] reduce ther stream of mortgage payments. Ths can be partally offset by the fact that the ncome and spendng of net savers wll be negatvely affected by a declne n nterest rates. However, the offsettng effect s lkely to reman moderate. (Peter Praet, ECB Conference Speech, October 2013) Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 2 / 16
Introducton Redstrbuton channel Tradtonal vew: monetary polcy affects household spendng va a substtuton channel A redstrbuton channel s also actve when agents have correlated margnal propenstes to consume and balance sheet postons supported emprcally: Johnson, Parker, Souleles 2006; Baker 2013 Nomnal nterest rate changes mpact balance sheets by: redenomnatng nomnal wealth (future nflaton effect) redstrbutng between unhedged borrowers and savers (real nterest rate effect) Ths s expansonary f those agents whose balance sheets mprove when nterest rates fall have hgher MPCs Ths project: understand and quantfy the redstrbuton channel usng model and mcro data Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 3 / 16
Introducton Related lterature Dstrbutonal effects of nflaton: Doepke and Schneder (2006), Berrel (2013) Monetary polcy and nequalty: Mortgages: Cobon, Gorodnchenko, Kueng, Slva (2012) contract choce: Campbell and Cocco (2003) monetary polcy transmsson: Calza, Monacell, Stracca (2013), Garrga, Kydland, Sustek (2013), Interest rate rsk and portfolo choce: Merton (1971), Campbell and Vcera (2002) Fnancal nsttuton balance sheets: Brunnermeer-Sannkov (2012, 2013); Begenau, Pazzes, Schneder (2013) MPC heterogenety: measurement: Johnson et al (2006), Parker et al (2013), Man, Rao, Suf (2013), Baker (2013) aggregate demand effects: Man-Suf (2012), Farh-Wernng (2012, 2013), Kornek-Smsek (2013) Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 4 / 16
Introducton Outlne 1 Theoretcal framework Partal equlbrum General equlbrum 2 Data and next steps Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 5 / 16
Theoretcal framework Partal equlbrum No-uncertanty framework Consder an agent wth arbtrary preferences, facng no uncertanty, earnng real ncome {y t } and holdng long-term contracts. Solves max U ({c t }) s.t. P t c t = P t y t + ( t 1 B t ) + s 1 ( t Q t+s ) ( t 1 B t+s t B t+s ) +P t ( t 1 b t ) + s 1 ( t q t+s ) P t+s ( t 1 b t+s t b t+s ) Intal holdngs: Nomnal assets: { 1 B t+s } s 0 (deposts, long-term bonds, mortgages) Real assets: { 1 b t+s } s 0 (TIPS, prce-level adjusted loans) Intal real term structure: q t = ( 0 q t ) and expected prce level {P t } No arbtrage between nomnal and real bonds: Fsher equaton for nomnal term structure t Q t+s = ( t q t+s ) Pt P t+s Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 6 / 16
Theoretcal framework Partal equlbrum Present-value budget constrant Intertemporal budget constrant (usng a TVC/TC) ( q t (y t + ( 1 b t ) + t 0 q t c t = t 0 1B t P t )) W All fnancal assets wth same present-value are equvalent and can be restructured n each perod Intal balance sheets are arbtrary Example: ntal holdng of a sngle real lablty D P 0 n form of: Adjustable rate: 1 B 0 = D Fxed rate: 1 B t = M, t = 1... T, T t=0 Q tm = D Prce-level adjusted: 1 b t = m, t = 1... T, T t=0 q tm = D P 0 Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 7 / 16
Theoretcal framework Partal equlbrum Unexpected shock At t = 0 there s an unantcpated/uncontracted shock to monetary polcy, resultng n a change n 1 The prce level {P 0, P 1...} 2 The real term structure {q 0 = 1, q 1, q 2...} 3 The agent s real ncome sequence {y 0, y 1...} The Fsher equaton stll holds after the shock Consder the frst-order change n consumpton dc 0 and welfare du that results from ths change n the envronment Consumer theory MPC = c0 W, ɛh 0,t = ch 0 q t q t c 0 at ntal ({q t }, W, u) now balance sheets matter: for example, 1B t = 0 1 b t = c t y t t hedges aganst both the nflaton and the real nterest-rate shock Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 8 / 16
Theoretcal framework Partal equlbrum Unexpected one-tme shock At t = 0 there s an unantcpated/uncontracted shock to monetary polcy, resultng n a one-tme change n 1 The prce level {P 0, P 1...} dp = Pdπ 2 The real term structure {q 0 = 1, q 1, q 2...} dq = qdr 3 The agent s real ncome sequence {y 0, y 1...} dy 0 The Fsher equaton stll holds after the shock Consder the frst-order change n consumpton dc 0 and welfare du that results from ths change n the envronment Consumer theory MPC = c0 W, ɛh 0,t = ch 0 q t q t c 0 at ntal ({q t }, W, u) Assumng separable utlty, β t u (c t ) t 0 σ (c) u (c) cu (c) ɛ h 0,0 = σ (c 0 ) (1 MPC) Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 9 / 16
Theoretcal framework Partal equlbrum Consumpton and welfare response Impulse response to the shock To frst order, the date-0 consumpton and welfare change n response to the shock are dc 0 = MPCdΩ + c 0 du = U c0 dω t 0 ɛ h dq t 0,t q t where dω = dw t 0 ctdqt, the net-of-consumpton wealth change, s gven by dω = ( ) 1B t dpt Q t P 0 P t t 0 }{{} + t 0 Revaluaton of NNP ( q t (y t + 1B t P t ) + ( 1b t) c t ) dqt q t } {{ } Real revaluaton of future flows + t 0 (q ty t) dyt y t }{{} Real ncome change Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 10 / 16
Theoretcal framework Partal equlbrum One-tme change One-tme unexpected change For a one-tme change: dc 0 = MPCdΩ σ (c 0) (1 MPC) c 0dr where dω comprses net nomnal poston (NNP) and unhedged rate exposure (URE): dω = ( ) ( ( ) ) 1B t 1B 0 q t dπ + y 0 + + ( 1b 0) c 0 dr + dy 0 P t P 0 t 0 }{{}}{{} URE NNP Ths formula contnues to apply under forms of market ncompleteness: Unnsurable dosyncratc rsk MPC = c0 y 0 (consumpton response to one-tme transtory ncome shock) (In some cases) borrowng constrants MPC = 1 Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 11 / 16
Theoretcal framework General equlbrum General equlbrum Consder general equlbrum wth heterogenous agents = 1... I Assume no net supply of fnancal assets/no government, and market clearng at every date: ( t 1B ) ( t = 0 1b ) t = 0 Ct = ct = yt = Y t Hence URE = 0 NNP = 0 dy0 = dc 0 dc Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 12 / 16
Theoretcal framework General equlbrum GE aggregate demand response Aggregate demand response to a one-tme shock dc = ( ) y Y MPC dc }{{} Aggregate ncome channel ( ) + Cov MPC, dy y dc C }{{} Earnngs heterogenety channel + Cov (MPC, URE ) ( σ 1 MPC ) c }{{} dr Interest rsk exposure channel }{{} Substtuton channel ( Cov MPC, NNP ) dπ } {{ } Fsher/Pgou channel Flexble-prce model: take redstrbuton as gven, set dc = 0, solve for dr Keynesan model: take dr as gven, solve for dc. Example: New-Keynesan dc = σcdr In ether, the covarance terms quantfy the redstrbuton channel Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 13 / 16
Data Correlatons n the populaton Italan Survey of Household Income and Wealth 2010: URE Non-fnancal Income - Consumpton + Deposts - D (debt payment for fxed-rate and debt lablty for adjustable-rate debt) MPC vs URE n populaton Average MPC n centle 30 40 50 60 70 0 20 40 60 80 100 Percentles of URE Usng these numbers, the redstrbuton and substtuton channels have the same sgn, and same magntude f the EIS s around 0.025-0.05 Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 14 / 16
Data Cross-sector and cross-regon redstrbuton Issues: 1 The household-sector URE s not mean-0 Counterparts are banks, government. Ther MPC? 2 There s substantal cross-regonal heterogenety n household UREs Next steps: model ths cross-sector and cross-regonal heterogenety Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 15 / 16
Data Redstrbuton channel? Growth n per capta real GDP (2008 to 2012, per cent) 10 5 0 5 10 West Vrgna Oregon Texas Vermont Lousana Indana Nebraska New Hampshre Iowa New York Kentucky Massachusetts Alaska Maryland Mnnesota Wsconsn Mchgan Illnos Montana Oho Rhode Pennsylvana ManeIsland Tennessee Delaware Kansas Utah South Dakota Washngton Vrgna Arkansas Oklahoma North Carolna Alabama South New Carolna Mexco Hawa Msssspp Mssour Colorado New Jersey Georga Calforna Connectcut Idaho Wyomng Florda Arzona Nevada 5 6 7 8 9 10 11 Share of nterest n personal ncome n 2007 (per cent) Source BEA. Sample ncludes all U.S. states except D.C. and North Dakota Thank you! Adren Auclert (MIT) Redstrbuton Channel May 31, 2014 16 / 16