the credit cycle and the business cycle in canada and the u.s.: two solitudes? pierre l. siklos wlu...

THE CREDIT CYCLE AND THE BUSINESS CYCLE IN CANADA AND THE U.S.:

TWO SOLITUDES?

Pierre L. Siklos WLU &Viessmann European Research

CentreBrady Lavender, Bank of Canada

“While most central banks have added a financial stability objective in recent

years, the monetary policy and financial stability wings of many of our

institutions have operated as two solitudes.” (Carney 2009)

2

The Role of Credit

• Credit availability is an essential part of MP effectiveness – Known for a long time (e.g., Roosa 1951) but ignored or

forgotten, until recently• Two channels of influence exist

– ‘price’ channel (i.e., interest rate)– ‘non-price’ or credit rationing channel (e.g., stemming from

asymmetric information problems)

3

Motivation & Background

• Current economic environment highlights the links between the real and financial sectors– Of special interest: the role of credit

• Credit has long been known to have a price and a non-price element– Price: interest rate– Non-price: ‘credit standards’

• Could the non-price element be “macro-economically” important?

4

The Questions Asked

• Do changing credit ‘conditions’ influence real economic outcomes?– The key source: survey of lending standards

• Do (monetary) policy rate shocks influence loan standards?• How does the picture change when ‘real time’ data are used?• If the U.S. and Canada are compared, then

– Are they indeed ‘two solitudes’?

5

Bottom Line: Previewing Some Results

• Tightening of standards is associated with a reduction of loans, output, and tighter monetary policy in BOTH the US and Canada

• Survey data contains some useful information that should be incorporated into macro-models, especially in light of the necessity to consider financial frictions

• U.S. and Canada appear to be ‘two solitudes’: impact of standards on loans considerably larger in US than in Canada

6

Related Literature• From Roosa (1951) to Fuerst (1994)

– Credit availability influences the effectiveness of MP

• Blanchard and Fischer (1989)– Credit rationing exists, so interest rates are not market clearing

• Stiglitz and Weiss (1981)– Interest rate changes create adverse selection (withdrawal of risk averse borrowers) and moral

hazard problems (incentives to engage in risky behavior): imperfect information in credit markets means they are not market clearing

• Schreft and Owens (1991)– Lending standards can change before cost of funds does. Therefore, non-price lending

standards represent an important link between MP and the financial sector• Measured via surveys

7

Non-Price Lending Standards and the Macro-economy

• Lown et.al. (2000)• Lown and Morgan (2006)• Swiston (2008) Beaton et. Al. (2009)

– 1% tightening leads to 2.5% reduction in loans, > 2% fall in investment– Tightening of standards leads to a fall in GDP (≈ 0.25-1%)– Tightening of MP leads to a tightening of standards (≈8%)– SLOS data precedes macro-data that would also be reflected in a fall in loans

• Murray (2012): “Stage 2…six key inputs to this meeting: …5. the Bank’s Senior Loan Officer Survey;” *

8

*“Monetary Policy Decision-Making at the Bank of Canada “, Remarks by Deputy Governor of the Bank of Canada Mortgage Brokers Association of B.C. Vancouver, British Columbia , 7 May 2012

Testing Strategy: Outline

ExtendedCore +Core

VAR/VECM

MacroMacro

Demand factors

Credit Credit

9

An Identification Problem?

• “Increases in loans and investment could be explained by easing lending standards or increasing loan demand. Thus, …it is essential to control for loan demand. There is an identification problem as changes in the price of loans, the going interest rate on loanable funds, reflects both demand and supply factors which operate simultaneously.”

10

Testing Strategy: Extension

ExtendedFAVARCore

VAR/VECM –CANADA

U.S. PCA

Macro

Credit

CAD VAR Core + Credit

11

Testing Strategy: Equations

12

0 1 1tt t y A A y ε

0 1 1 2 1tt tt y A A y A z ε

'0 1tt t y A y ε

* ' ' * '0 1 1 2 1tt tt y A A y A z ε

US UStt t y F e

1

1

( )ttt

tt

F FL

y y

Data & Stylized Facts

• SLOS U.S. (since 1970s) & Canada (since late 1990s) ~ ‘balance of opinion’– “Over the past three months, how have your bank’s credit standards

for approving loan applications for C&I loans or credit likes – excluding those to finance mergers and acquisitions – changed? 1) Tightened considerably 2) tightened somewhat 3) remained basically unchanged 4) eased somewhat 5) eased considerably” UNITED STATES

– "How have your institution’s general standards (i.e. your appetite for risk) and terms for approving credit changed in the past three months?” CANADA

• Canada has ‘price’ versus ‘non-price’ distinction but differences not informative

13

Figure 1a Senior Officer Loan Survey and Commercial Loans, 1972-2011: U.S.

14

-40

-20

0

20

40

60

80

100

1975 1980 1985 1990 1995 2000 2005 2010

Net tightening of standards Commercial loan growth

Perc

ent (

annu

al)

No standards reported1984- 90

Figure 1b Senior Officer Loan Survey and Commercial Loans, 1999-2011: Canada

15

-40

-20

0

20

40

60

80

100

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Balance of Overall credit standards Growth in business credit

SLO

S in

dex

Annual precent change

Data & Stylized Facts

• SLOS U.S. (since 1970s) & Canada (since late 1990s) ~ ‘balance of opinion’

– “Over the past three months, how have your bank’s credit standards for approving loan applications for C&I loans or credit likes – excluding those to finance mergers and acquisitions – changed? 1) Tightened considerably 2) tightened somewhat 3) remained basically unchanged 4) eased somewhat 5) eased considerably”

– "How have your institution’s general standards (i.e. your appetite for risk) and terms for approving credit changed in the past three months?”

• Canada has ‘price’ versus ‘non-price’ distinction but differences not informative

16

Price and Non-Price Survey Indicators: SLOS for Canada

17

-60

-40

-20

0

20

40

60

80

100

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Differential SLOS: non-priceSLOS: pricing

+ m

eans

net

tigh

teni

ng o

f sta

ndar

ds -

mea

ns n

et lo

osen

ing

of s

tand

ards

Other Series & Samples

• Real GDP, GDP Deflator, Commodity prices, Policy Rate– Defines basic macro model

• Add: Loans, SLOS– Defines ‘core’ or ‘benchmark’ model

• Add: expected real GDP growth, term spread, FCI*– Defines extended model– * acts as a quasi F since it “measures risk, liquidity, and leverage in money

markets and equity markets as well as in the traditional and ‘shadow’ banking systems”

• US: 1972:1-2011:1 (disjointed: 1984-1990 omitted), CAD: 1999:2-2011:1

18

Core Series US

19

8.4

8.6

8.8

9.0

9.2

9.4

9.6

1975 1980 1985 1990 1995 2000 2005 2010

US rea l GDP

3.2

3.6

4.0

4.4

4.8

1975 1980 1985 1990 1995 2000 2005 2010

GDP deflator

1

2

3

4

5

1975 1980 1985 1990 1995 2000 2005 2010

Oi l Price

0

4

8

12

16

20

1975 1980 1985 1990 1995 2000 2005 2010

US federal funds rate

7.0

7.5

8.0

8.5

9.0

9.5

1975 1980 1985 1990 1995 2000 2005 2010

Commercial loans

-40

-20

0

20

40

60

80

100

1975 1980 1985 1990 1995 2000 2005 2010

SLOS

Loga

rithm

of t

he le

vels

Loga

rithm

of t

he le

vels (i

ndex

)

Loga

rithm

of t

he le

vels

Perc

ent

Loga

rithm

of t

he le

vels (B

illio

ns o

f $)

+ in

dica

tes ne

t tig

hten

ing

of sta

ndar

ds

- in

dica

tes ne

t loo

seni

ng o

f sta

ndar

ds

Core Series Canada

20

13.80

13.85

13.90

13.95

14.00

14.05

14.10

14.15

2000 2002 2004 2006 2008 2010

rea l GDP

4.52

4.56

4.60

4.64

4.68

4.72

4.76

4.80

4.84

2000 2002 2004 2006 2008 2010

GDP deflator

5.4

5.6

5.8

6.0

6.2

6.4

6.6

6.8

2000 2002 2004 2006 2008 2010

Commodity prices

0

1

2

3

4

5

6

2000 2002 2004 2006 2008 2010

Overnight Ra te

13.5

13.6

13.7

13.8

13.9

14.0

14.1

2000 2002 2004 2006 2008 2010

Business credit

-40

-20

0

20

40

60

80

2000 2002 2004 2006 2008 2010

SLOS

Loga

rithm

of t

he le

vels

Loga

rithm

of t

he le

vels

Loga

rithm

of t

he le

vels

Loga

rithm

of t

he le

vels

Perc

ent

+ si

gnifi

es n

et ti

ghte

ning

of c

redi

t sta

ndar

ds

- si

gnifi

es n

et lo

osen

ing

of c

redi

t sta

ndar

ds

FCI

21

-4

-3

-2

-1

0

1

2

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

BOC_FCI

+ =

impr

ovin

g fin

anci

al c

ondi

tions

/ - =

det

erio

ratin

g fin

anci

al c

ondi

tions

USA Canada

Other Series & Samples• Real GDP, GDP Deflator, Commodity prices

– Defines basic macro model• Add: Loans, SLOS

– Defines ‘core’ or ‘benchmark’ model• Add: expected real GDP growth, term spread, FCI*

– Defines extended model– * acts as a quasi F since it “measures risk, liquidity, and leverage in money markets and

equity markets as well as in the traditional and ‘shadow’ banking systems” • US: 1972:1-2011:1 (disjointed: 1984-1990 omitted), CAD: 1999:2-2011:1

– US: begin with 1972-1990 (Lown & Morgan)** + 1990-2011 & 1972-2011 (disjointed)• ** No data for 1968-1971 & vintage of data differs

22

An Important Addition: real-time dataVintages: U.S. Real GDP Significance Vintages: U.S. Potential

Output2000 December Just before P 2000 July

2002 March Just after T 2002 February

2007 September Just before P – PRE-CRISIS 2007 August

2009 September Just after T – FIN CRISIS 2009 August

Vintages: CAN real GDP Significance From Bank of Canada

2002 March See US

2007 Q3 See US

2007 Q4 Peak CAD bus cycle

2009 Q3 BoC interest rate comm.

23

An Important Addition:real-time data

24

-10

-8

-6

-4

-2

0

2

4

6

1975 1980 1985 1990 1995 2000 2005 2010

2000 Q4 vintage 2002 Q1 vintage2007Q3 vintage 2009 Q3 vintage

-5

-4

-3

-2

-1

0

1

2

3

4

1975 1980 1985 1990 1995 2000 2005 2010

2000 Q4 vintage 2002 Q1 vintage2007 Q3 vintage 2009 Q3 vintage

100 tim

es (log

rea

l GDP - log

pot

entia

l rea

l GDP)

Congressional Budget Office Estimates

H-P Filter estimates

US

Output Gaps in Real Time: A Closer Look for the US I

25

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2000 2002 2004 2006 2008 2010 2012 2014

CBO 2000 prices - 2009 vintage CBO 2005 prices - August 2011 vintage

US Potential Output- CBO Estimate

Nor

mal

ized

Output Gaps in Real Time: A Closer Look for the US II

26

-4

-3

-2

-1

0

1

2

3

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

January 2008 December 2008December 2009

Perc

ent

-12

-10

-8

-6

-4

-2

0

2

4

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

July 2009 August 2009

-10

-8

-6

-4

-2

0

2

4

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

June 2010 December 2010March 2011 August 2011

Perc

ent

Output gaps - US

An Important Addition:real-time data

27

Canada

-5

-4

-3

-2

-1

0

1

2

3

2000 2002 2004 2006 2008 2010

GAP_200201 GAP_200703GAP_200704 GAP_200903

-.3

-.2

-.1

.0

.1

.2

2000 2002 2004 2006 2008 2010

GAPHP_200201 GAPHP_200703GAPHP_200704 GAPHP_200903

Percen

tBank of Canada

H-P Filter

Canadian Real Time Data: A Specific Vintage, 2011 Q1

28

1,260,000

1,280,000

1,300,000

1,320,000

1,340,000

1,360,000

1,380,000

I II III IV I II III IV I II III IV I II III IV I

2007 2008 2009 2010 2011

Potential real GDP 2008Q1 Potential real GDP 2009Q1Porential real GDP 2010Q1 Potential real GDP 2011Q1realized real GDP 2011Q1

Mill

ions

$Real ized and Potential Output - CANADA

VAR/VECM Issues I

• Lag length?– AIC, HQ, SC...but parsimony wherever results are robust

• Series transformations?– All in log levels EXCEPT: SLOS, Spread, forecasted growth rate– Real GDP, GDP Deflator, loans ~ I(1)– SLOS, Comm. Prices, Spread, Policy rate, FCI ~ I(0)

• S.E. via MC

29

VAR/VECM Issues II

• What CI?– {policy rate – Loans}, {policy rate-SLOS}, {real GDP-

Loans}• Does the ordering matter?

– [MACRO, CREDIT]: Core– [MACRO, DEMAND IDENTIFIERS, CREDIT]

• Conventional IRFs & VDs + GIRFs

30

Empirical Results, US: Summary

• 1972-2000– Real GDP falls when standards are raised– Loans decline when standards are tightened– No impact of loans to standards

• Identification problem solved? Not entirely is EXTENDED model is considered

• 1972-2011– Results are similar...

• Negative impact on real GDP larger• Negative impact on loans smaller• Could weakening standards and greater importance of financial sector have played a

role?

31

U.S. Real Time Data

• Results largely unchanged!– ...but a TIGHTENING of standards leads to Loans

INCREASE for only for 2009Q3 and it seems PERMANENT

• Results insensitive to ordering

32

IRFs: US, 1999-2011*

33

-.02

-.01

.00

.01

.02

1 2 3 4 5 6 7 8 9 10 11 12

Resp

onse

of l

og re

al G

DP to

SLO

S

-.08

-.04

.00

.04

.08

1 2 3 4 5 6 7 8 9 10 11 12

Resp

onse

of l

og L

oans

to S

LOS

-8

-4

0

4

8

12

1 2 3 4 5 6 7 8 9 10 11 12

Resp

onse

of S

LOS

to lo

g Lo

ans

QuartersQuarters Quarters

FIGURE 3: EXTENDED VAR

IRFs: 2009 Q3 Vintage

34

-.02

-.01

.00

.01

.02

1 2 3 4 5 6 7 8 9 10 11 12

Resp

onse

of l

og r

eal G

DP

to S

LOS

-.08

-.04

.00

.04

.08

1 2 3 4 5 6 7 8 9 10 11 12

Resp

onse

of l

og L

oans

to

SLO

S

-8

-4

0

4

8

12

1 2 3 4 5 6 7 8 9 10 11 12

Resp

onse

of S

LOS

to lo

g Lo

ans

Quarters QuartersQuarters

FIGUR 2c 1st COLUMN, EXTENDED VAR

Empirical Results, Canada: Summary

• Results reminiscent of ones for the U.S.: Recall that the GFC did not materially affect Canada’s financial sector but we did have a drop in GDP

• Negative impact on GDP from a tightening of standards disappears in the extended model

• Small (but stat sig) response of loans from STANDARDS• Real time data makes little difference to the results

– Real & financial link is NOT dependent on crisis but applies equally to US and CAD data

35

US Factor Scores for CAD FAVAR

36

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2000 2002 2004 2006 2008 2010

Macro conditions Macro conditions 2009Q3

-2

-1

0

1

2

3

4

2000 2002 2004 2006 2008 2010

Financial conditions 2009Q3 Financial conditions

Fact

or sco

res

Also see FIGURE 5

Macro US Factor Scores

37

-6

-5

-4

-3

-2

-1

0

1

2

1975 1980 1985 1990 1995 2000 2005 2010

1972-2011 1989-20111999-2011 1999-2011 (2009Q3 vintage)

Fact

or S

core

s

Financial US Factor Scores

38

-2

-1

0

1

2

3

4

5

1975 1980 1985 1990 1995 2000 2005 2010

1972-2011 1989-20111999-2011 1999-2011 (vintage 2009Q3)

Fact

or S

core

s

FAVAR for CANADA

39

-.0004

-.0002

.0000

.0002

.0004

1 2 3 4 5 6 7 8 9 10 11 12

Resp

onse

of l

og r

eal G

DP

to S

LOS

-.008

-.004

.000

.004

.008

1 2 3 4 5 6 7 8 9 10 11 12Resp

onse

of l

og B

usin

ess

Cred

ot t

o SL

OS

-8

-4

0

4

8

12

1 2 3 4 5 6 7 8 9 10 11 12Resp

onse

of S

LOS

to lo

g Bu

sine

ss C

redi

t

Scaling changed

Variance Decompositions I

• Confirms the significant explanatory power of SLOS on real GDP– But SLOS explains more of the VAR of ALL vars than for US data…do standards

matter more in Canada?• Shocks of Loans to FFR but NOT vice-versa• Shocks of FFR to SLOS more ‘important’ than any other variable

– But term spread matters not for the US but is sig for Canada• Real time data AMPLIFIES results based on revised data

– e.g., SLOS explanatory power for FFR 25 times larger, and SLOS VD for Loans 3 times larger

• SLOS less important in extended model but Loans become more important

40

VD: An Illustration

41

Conclusions

• Empirical exploration of links between lending standards, credit, and macro activity in CANADA and spillovers from the US– Credit shock play a bigger role in US than in Canada….two solitudes– Real time data considerations matter…shock has bigger –ve impact on US real GDP

in 2009Q3 vintage– MP was more effective in Canada in impacting loans & standards– SLOS is a proxy for ‘frictions’ in financial markets and should be considered as a

candidate for inclusion in empirical macro models• EXTENSIONS?

– More VECM work? Longer sample needed– Other types of VAR? may require better theoretical underpinnings

42