Incorporang*Financial*Stability*in* Policy*Analysis ...1 Sum of all cross-border claims and locally...

Incorpora(ng  Financial  Stability  in  Policy  Analysis  

Garcia  Cicco  et  al.    Discussion  by  

Michael  B.  Devereux    

BIS-­‐CCA  Research  Network  Conference  Mexico  City  Jan  28/29,  2015  

Discussion  of  project  •  Document  describes  a  basic  OE  NK  model  with  FF  and  reports  results  of  es(ma(on  and  simula(on  of  similar  shocks  for    –  Chile,  Columbia,  Mexico  and  Peru  –  Adapts  NK-­‐FF  model  to  incorporate  commodity  effects  

•  Ambi(ous  and  highly  produc(ve  project  –  Document  reveals  a  huge  coordinated  effort  that  has  already  borne  fruit  

–  Substan(al  coherence  in  policy  frameworks  •  Reconcilia(on  of  models  needs  to  be  done  

–  Promises  to  offer  major  insights  into  monetary  and  macro-­‐pruden(al  policy  for  LAM  

Discussion  

•  Model  and  es(ma(on  too  detailed  to  get  into  the  nuts  and  bolts  

•  More  general  comments,  quibbles,  sugges(ons,  observa(ons  

Plan  of  discussion  

•  Discuss  the  basic  modeling  template  used  – Compare  with  other  FF  models  

•  Implica(ons  for  macro-­‐pruden(al  policy  •  Comment  on  assump(ons  about  financial  market  structure  

•  Comment  on  nature  of  capital  flows  •  Sugges(on  for  extending  the  model  to  a  richer  theory  of  interna(onal  financial  intermedia(on    

How  does  model  compare  with  other  FF  models?    

The current model (assuming only one intermediary)

Budget constraint of household

Ct + (1 +⌥t)Dt + Bt = WtHt + ⇤t + RDt Dt�1 + RtBt

⌥t = ⌥⇣QtKt+1

Nt

⌘�is external to household and ⌥

0() > 0.

Nt is intermediaries net worth.

Gives a spread on loan rate over the safe government rate

RDt = (1 +⌥t)Rt

Intermediaries borrow and lend at same rate

RKt+1 = RDt

Models with explicit intermediation structure (e.g.

Gertler Karadi)

Enforcement constraint Nt � QtKt+1

Equilibrium spread is

RKt+1 � RDt � �(

QtKt+1

Nt)

�0() > 0

Intermediaries choose deposits and investment subject to

enforcement constraint (spread is internalized).

Conclusion  in  simula(on  

•  Has  very  similar  aggregate  IRFs  

0 5 10 15 20 25 30 35 400

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8Spread

GKACMRBC

Calibrate  so  shock  to  spread  is  the  same  (here  a  TFP  shock)  

0 5 10 15 20 25 30 35 40−1.8

−1.6

−1.4

−1.2

−1

−0.8

−0.6

−0.4

−0.2

0Capital Stock

GKACMRBC

Both  models  have  a  financial  accelerator  which  looks  alike  

Conclusion  in  simula(on  

•  Has  very  similar  aggregate  IRFs  

•  Does  this  mean  it  has  same  implica(ons  for  macro-­‐pruden(al  policies?  

•  Perhaps  not?  – Contract  structure  in  financial  intermedia(on  may  be  very  important  for  response  to  pruden(al  regula(on  

With  externali(es  

Individual  return  

Aggregate  Return  

Investment  

Example:    with  explicit  balance  sheet  effects,  decisions  may  lead  to  more  risk-­‐taking  with  no-­‐loss  guarantees    

(may  show  up  only  at  higher  order  approx.  though)  

Aggregate  Distribu(on  

Individual  Distribu(on  

Guarantee  

Some  more  comments  on  the  model  

•  Domes(c  Households  have  full  access  to  domes(c  bank  deposits,  government  securi(es  and  foreign  bonds  

•  Is  this  reasonable  for  the  countries  involved?  •  Maybe  not  – Measures  of  financial  market  access  for  these  countries  are  far  lower  than  for  high-­‐income  countries  

– See  following  graph  

0  

10  

20  

30  

40  

50  

60  

70  

80  

90  

100  

Peru   Mexico   Columbia   Chile   High  Income  

*Global  Findex  Database,  World  Bank  

Adults  with  Account  at  Financial  Ins<tu<on  

Male  

Female  

How  much  difference  could  this  make?  

•  Perhaps  a  lot  – Response  to  external  interest  rate  shocks  may  differ  (see  below)  

– Ability  to  ac(vely  use  sterilized  interven(on  – May  affect  the  dynamics  of  financial  accelerator?  

•  Also,  the  extent  of  `financial  inclusion’  may  affect  the  conduct  of  monetary  policy  – See  recent  BIS  wp  by  Mehrotra  and  Yetman  2015  

Interest  rate  shocks  for  Chile  and  Colombia  

•  Expansionary,  as  in  Mundell  Fleming  model  Figure 5: Impulse responses to a foreign interest rate shock: Chile.

10 20 30 40−5

0

5

10x 10−3 R!

! y

10 20 30 400

2

4

6

8x 10−3 R!

! c

10 20 30 40−0.02

−0.01

0

0.01

0.02R!

! i

10 20 30 40−0.5

0

0.5

1R!

! !c, a.b.p.

10 20 30 40−0.5

0

0.5

1R!

! R, a.b.p.

10 20 30 40−0.05

0

0.05

0.1

0.15R!

! q

10 20 30 400

0.02

0.04

0.06R!

! a

10 20 30 40−0.01

0

0.01

0.02R!

! l

10 20 30 40−0.05

0

0.05

0.1

0.15R!

! spr, a.b.p.

10 20 30 40−0.02

0

0.02

0.04R!

! levE

10 20 30 40−0.02

0

0.02

0.04R!

! levB

10 20 30 40−0.06

−0.04

−0.02

0R!

! rcap

10 20 30 40−0.01

0

0.01

0.02R!

! ltv

10 20 30 400

0.02

0.04

0.06

0.08R!

! R!

No MPPβcap=−50

βcap=−500

βltv=50

βltv=500

Note: This figure shows the impulse responses of selected variables to a 1% foreign interest rate shock in period 1. Theimpulse responses are computed at the posterior mean of the estimated parameters when the only policy variable at workis the short-term interest rate (“No MPP”) and when in conjunction to the policy rate capital requirements and LTVlimits are used (using one instrument at the time) according to the feedback rule INST

t

= INST+�(VARt

�VAR), whereINST

t

and VARt

are the alternative policy instruments and the variable to react, respectively, in logs. In this case, VARt

is the total credit spread.

[4] Henrıquez, C. (2008), “Capital Stock in Chile (1985-2005): Methodology and Results” (in Spanish),Studies in Economic Statistics 63, Central Bank of Chile.

[5] Medina, J. P., and C. Soto (2007), “The Chilean Business Cycles Through the Lens of a StochasticGeneral Equilibrium Model,” Central Bank of Chile Working Papers 457.

[6] Roberts, G. O., A. Gelman, and W. R. Gilks (1997), “Weak convergence and optimal scaling of randomwalk Metropolis algorithms,” Annals of Applied Probability, vol. 7(1), pp. 110–120.

11

Figure 4: Impulse responses to a negative commodity price shock: Colombia.

10 20 30 40-0.1

0

0.1

0.2

0.3pCo!

! y

10 20 30 40-0.7

-0.6

-0.5

-0.4

-0.3

-0.2pCo!

! c

10 20 30 40-1

-0.5

0

0.5

1pCo!

! i

10 20 30 402

4

6

8

10

12

14

16pCo!

! !c, a.b.p.

10 20 30 400

5

10

15

20pCo!

! R, a.b.p.

10 20 30 401

1.5

2

2.5

3

3.5pCo!

! q

10 20 30 40-5.5

-5

-4.5

-4

-3.5

-3

-2.5

-2pCo!

! a

10 20 30 40-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1pCo!

! l

10 20 30 40-40

-30

-20

-10

0

10pCo!

! spr, a.b.p.

10 20 30 40-2

-1.5

-1

-0.5

0

0.5pCo!

! levE

10 20 30 40-1

-0.5

0

0.5

1

1.5

2pCo!

! levB

10 20 30 40-6

-5

-4

-3

-2

-1

0pCo!

! rcap

10 20 30 40-2

0

2

4

6

8pCo!

! ltv

10 20 30 40-120

-100

-80

-60

-40

-20

0pCo!

! pCo!

No MPP

"cap

=10

"cap

=100

"ltv

=-10

"ltv

=-100

Note: This figure shows the impulse responses of selected variables to a negative one-standard deviation

commodity price shock in period 1. The impulse responses are computed at the posterior mean of the

estimated parameters when the only policy variable at work is the short-term interest rate (“No MPP”)

and when in conjunction to the policy rate capital requirements and LTV limits are used (using one

instrument at the time) according to the feedback rule INSTt

= INST+ �(VARt

�VAR), where INSTt

and VARt

are the alternative policy instruments and the variable to react, respectively, in logs. In this

case, VARt

is real credit.

10

Solu(ons?  

•  Incorporate  working  capital,  as  in  Neumeyer-­‐Perri  (2004)?  

•  Can  easily  combine  into  a  model  with  financial  fric(ons    – Mendoza  2010,  etc.  

•  But  more  generally,  financial  linkages  may  be  much  more  complicated?  

Financial  linkages  

•  Banks  here  all  financed  with  domes(c  deposits  

•  Three  implica(ons  – No  direct  exposure  to  currency  risk  due  to  maturity  mismatch  

– No  direct  vulnerability  to  external  funding  shocks  (global  financial  cycle)  

– With  foreign  funding  of  banks,  gross  capital  flows  may  mager  

Restricted

3/10

home jurisdictions. Therefore, much of intraregional bank expansion, both in the

Southern Cone as well as in Central American countries, has tended to reflect

corporate banking activities. Against this background, trade was cited as an

important driver in the Latin American region as well. However, with only about a

quarter of trade flows being intraregional, compared to about half in Asia, the pace

of accompanying financial integration among Latin American economies has been

slower. The integration process has also been more concentrated, with Brazilian and

Colombian banks being among the most regionally active institutions, given the

rising regional presence of large corporates from both countries.

Among the other important drivers mentioned at both workshops was

macroeconomic and institutional convergence, with a broad trend towards better

institutions, more predictable policies and independent central banking facilitating

banks’ cross-border expansion. In Latin America, in particular, workshop participants

highlighted that South-South macroeconomic integration continued to be on the

rise, as illustrated by FDI flow patterns, and that this would tend to support financial

integration. In addition, participants at both workshops mentioned the (temporary,

in some cases) retrenchment of advanced economy banks from non-core EME

locations and business lines as a key catalyst for the most recent round of

intraregional bank expansion. Saturation of domestic markets and various “soft”

factors (such as geographical and cultural proximity) were also highlighted as

additional determinants of regional expansion.

Bank credit to Latin America and the Caribbean Graph 1B

International claims on the region1

USD bn

Share in international claims on the region

Per cent

1 Sum of all cross-border claims and locally extended claims in foreign currency.

2 Intraregional share is the sum of international claims

on the emerging Latin America and Caribbean region of regional banks (Brazil, Chile and Mexico) and Caribbean offshore banks (Panama)

divided by total international claims on the emerging Latin America and Caribbean region.

Source: BIS consolidated banking statistics (immediate borrower basis).

Constraints on further regionalisation. Private sector participants at both

workshops stressed that achieving meaningful growth in foreign markets can be

very challenging. One of the reasons why participants thought regional integration

would likely be slower, and more selective on a country-by-country basis, than

anticipated in the CGFS report was the risk of uneven regulatory implementation

0

100

200

300

400

500

02 03 04 05 06 07 08 09 10 11 12 13 14Intraregional banks2

US banksUK banksOther reporting banks

Swiss banksOther euro area banksSpanish banks

0

6

12

18

24

30

02 03 04 05 06 07 08 09 10 11 12 13 14Intraregional banks2

US banksUK banks

Swiss banksOther euro area banksSpanish banks

Chile:  Gross  Inflows/Ouilows  (IFS)  

-­‐0.1  

-­‐0.05  

0  

0.05  

0.1  

0.15  

0.2  

0.25  

0.3  

2005Q1   2006Q3   2008Q1   2009Q3   2011Q1   2012Q3  

Inflows  

Ouilows  

Model  with  explicit  role  for  gross  flows,  foreign  intermediaries    

•  Devereux  and  Lombardo,  2015    •  Global  banks  –  funding  for  emerging  market  banks  – Balance  sheet  constraints  at  both  levels  

•  US  policy  shocks  affect  BS  of  global  banks,  reduce  lending  to  EME  banks,  shrinking  in  BS  for  both  – Coordinated  increase  in  spreads  

•  Similar  to  Bruno  and  Shin  2014  

GDP  response:    With  financial  fric(ons  at  both  levels,  policy  (ghtening  in  US  causes  a  global  downturn    Without  FF,  does  not  

−0.0698

−0.0506

−0.0314

−0.0122

0.007GDP C

−0.1091

−0.0743

−0.0395

−0.0048

0.03GDP E

−0.0196

−0.0078

0.004

0.0158

0.0276Real Exchange Rate

−0.0936

−0.0677

−0.0418

−0.016

0.0099Banks Total Assets C

−0.1058

−0.0697

−0.0336

0.0025

0.0386Banks Debt E

−0.1598

−0.1098

−0.0597

−0.0097

0.0403Households NFA

BenchmarkNo FFPegLCD

Policy Shock C

−0.0698

−0.0506

−0.0314

−0.0122

0.007GDP C

−0.1091

−0.0743

−0.0395

−0.0048

0.03GDP E

−0.0196

−0.0078

0.004

0.0158

0.0276Real Exchange Rate

−0.0936

−0.0677

−0.0418

−0.016

0.0099Banks Total Assets C

−0.1058

−0.0697

−0.0336

0.0025

0.0386Banks Debt E

−0.1598

−0.1098

−0.0597

−0.0097

0.0403Households NFA

BenchmarkNo FFPegLCD

Policy Shock C

Contrac(on  in  Capital  inflows  and  ouilows    Fall  in  capital  inflows  to  EM  banks        Fall  in  capital  ouilows  to  US      US  dollar  denomina(on  makes  a  difference  

−0.0698

−0.0506

−0.0314

−0.0122

0.007GDP C

−0.1091

−0.0743

−0.0395

−0.0048

0.03GDP E

−0.0196

−0.0078

0.004

0.0158

0.0276Real Exchange Rate

−0.0936

−0.0677

−0.0418

−0.016

0.0099Banks Total Assets C

−0.1058

−0.0697

−0.0336

0.0025

0.0386Banks Debt E

−0.1598

−0.1098

−0.0597

−0.0097

0.0403Households NFA

BenchmarkNo FFPegLCD

Policy Shock C

−0.0698

−0.0506

−0.0314

−0.0122

0.007GDP C

−0.1091

−0.0743

−0.0395

−0.0048

0.03GDP E

−0.0196

−0.0078

0.004

0.0158

0.0276Real Exchange Rate

−0.0936

−0.0677

−0.0418

−0.016

0.0099Banks Total Assets C

−0.1058

−0.0697

−0.0336

0.0025

0.0386Banks Debt E

−0.1598

−0.1098

−0.0597

−0.0097

0.0403Households NFA

BenchmarkNo FFPegLCD

Policy Shock C

−0.1186

−0.0854

−0.0521

−0.0189

0.0144Asset Price C

−0.4151

−0.3011

−0.1872

−0.0732

0.0408Asset Price E

−0.0821

0.0542

0.1905

0.3268

0.4631Leverage C

0.0249

0.3458

0.6668

0.9878

1.3087Leverage E

−2.3456

3.0807

8.5071

13.9334

19.3597x 10−3 spread C

0

0.0205

0.0409

0.0614

0.0818spread E

BenchmarkNo FFPegLCD

Policy Shock C

−0.1186

−0.0854

−0.0521

−0.0189

0.0144Asset Price C

−0.4151

−0.3011

−0.1872

−0.0732

0.0408Asset Price E

−0.0821

0.0542

0.1905

0.3268

0.4631Leverage C

0.0249

0.3458

0.6668

0.9878

1.3087Leverage E

−2.3456

3.0807

8.5071

13.9334

19.3597x 10−3 spread C

0

0.0205

0.0409

0.0614

0.0818spread E

BenchmarkNo FFPegLCD

Policy Shock C

You  also  get  a  common  increase  in  spreads,  in  both  core  and  periphery  

Implica(ons  

•  Poriolio  posi(ons  and  gross  flows  may  be  important  in  environment  of  financial  fric(ons  

Implica(ons  for  policy?    

•  Op(mal  coopera(ve  monetary  policy  can  sharply  reduce  co-­‐movement  of  financial  shocks  

−4.0562

−2.9368

−1.8175

−0.6981

0.4212x 10−3 GDP−E

−1.5881

−0.6241

0.3399

1.3039

2.2679x 10−3 GDP−C

−9.18

−6.0836

−2.9872

0.1092

3.2056x 10−4 RER

−4.1758

−2.9414

−1.7071

−0.4727

0.7616x 10−3 Bank Assets C

5 10 15 20 25 30 35 40−4.1126

−2.9983

−1.8839

−0.7696

0.3447x 10−3 Bankdebt E

5 10 15 20 25 30 35 40−10.1023

−7.7237

−5.3451

−2.9666

−0.588x 10−3 Household NFA−E

Financial Shock C: Solid=Benchmark; Dashed=Ramsey

−4.0562

−2.9368

−1.8175

−0.6981

0.4212x 10−3 GDP−E

−1.5881

−0.6241

0.3399

1.3039

2.2679x 10−3 GDP−C

−9.18

−6.0836

−2.9872

0.1092

3.2056x 10−4 RER

−4.1758

−2.9414

−1.7071

−0.4727

0.7616x 10−3 Bank Assets C

5 10 15 20 25 30 35 40−4.1126

−2.9983

−1.8839

−0.7696

0.3447x 10−3 Bankdebt E

5 10 15 20 25 30 35 40−10.1023

−7.7237

−5.3451

−2.9666

−0.588x 10−3 Household NFA−E

Financial Shock C: Solid=Benchmark; Dashed=Ramsey

Benchmark  vs.  Ramsey  policy  

Other  issues  

•  Fiscal  side  is  minimized  –  Interac(on  between  commodity  cycle  and  fiscal  balances  may  be  important  for  some  countries  

– How  pro-­‐cyclical  is  fiscal  policy?  •  Need  to  incorporate  commodity  sector  more  explicitly  into  consump(on  and  investment  sectors?  – There  may  be  important  distribu(onal  effects  of  commodity  shocks  •  across  sectors  and  regions  

Big  picture  issues  

•  Allowing  for  risk,  non-­‐lineari(es,  crises,  sudden  stops  – May  be  very  important  for  evalua(on  of  macro-­‐pruden(al  policy  

– More  easily  done  in  small  models  – But  some  results  with  full  mul(-­‐country  DSGE  models    •  Devereux  and  Yu,  2015  

Conclusions  

•  All  these  sugges(ons  are  more  for  general  background  

•  Current  project  has  made  major  strides  •  Look  forward  to  seeing  future  developments