The role of debt in UK household spending decisions Phil Bunn Bank - - PowerPoint PPT Presentation

1

Phil Bunn Bank of England

Conference on the Causes and Consequences of the Long UK Expansion: 1992 to 2007 19 September 2013

The role of debt in UK household spending decisions

Outline

• Overview of developments in household balance sheets

– Trends in household balance sheets – How have these affected consumption

• Research on how debt affects consumption

– Should debt affect spending? – Methodology – Econometric results – Implications for aggregate consumption – Causality versus differing expectations

Developments in household balance sheets

Household debt to income

• The stock of debt more than trebled between 1992 and 2007
• Mortgage debt accounts for around 80% of total household debt

40 60 80 100 120 140 160 180 1987 1992 1997 2002 2007 2012 Total debt to income Mortgage debt to income Per cent of post-tax annual income

House prices and mortgage debt relative to income

40 60 80 100 120 140 60 80 100 120 140 160 180 1987 1991 1995 1999 2003 2007 2011 Index (1999 = 100) House price to income (rhs) Mortgage debt to income (lhs) Per cent of post-tax annual income

• Increases in mortgage debt appear largely related to rises in house prices

Real interest rates

• 2

2 4 6 1987 1992 1997 2002 2007 2012 Per cent Five-year real interest rates, five- years forward

• Falls in real interest rates are likely to have been a factor behind increases

in debt and house prices

• Looser credit conditions will also have played a role

Net acquisition of assets and liabilities

• Household acquired financial assets at a similar rate to liabilities
• 5

5 10 15 20 25 1987 1992 1997 2002 2007 2012 Per cent of post-tax income Acquisition of financial assets Acquisition of deposits Acquisition of financial liabilities

Household assets and liabilities

• Balance sheet positions also depend on changes in asset prices

50 100 150 200 250 300 350 400 450 500 1987 1992 1997 2002 2007 2012 Per cent of annualised post-tax income Financial liabilities Financial assets Residential buildings assets

Household capital gearing

5 10 15 20 25 30 35 40 45 50 1987 1992 1997 2002 2007 2012 Excluding housing assets Including housing assets Per cent

• Over the 1992 to 2007 period, debt did not increase substantially in relation

to the value of assets

Net wealth

• Net wealth rose significantly over the long expansion

50 100 150 200 250 300 350 400 450 1987 1991 1995 1999 2003 2007 2011 Nominal net financial wealth Nominal net wealth including housing Real net financial wealth Real net wealth including housing Indices (1992 = 100)

Debt Gross wealth

50 100 150 200 250 18-24 25-34 35-44 45-54 55-64 65+ 1995 2005 £, thousands 10 20 30 40 50 60 70 80 18-24 25-34 35-44 45-54 55-64 65+ 1995 2005 £, thousands

Changes in the distribution of balance sheets by age

• Distribution of balance sheets has changed significantly
• Younger households have become more indebted, whilst older households

have become wealthier

Annual consumption growth

• 6
• 4
• 2

2 4 6 8 10 12 1987 1992 1997 2002 2007 2012 Per cent 1956-2013 average 1999-2007 average 1992-1998 average

• Consumption grew at a similar rate between 1992 to 1998 (when D/Y

was flat) as it did between 1999 and 2007 (when D/Y was rising)

• No clear evidence that increases in debt led to a consumption boom

Household saving ratio

• 2

2 4 6 8 10 12 14 1987 1992 1997 2002 2007 2012 Recessions (a) Saving ratio (b) Per cent

• Saving ratio also declined at similar rate when debt to income was rising

to when it was flat

Housing equity withdrawal and consumption

• 6
• 4
• 2

2 4 6 8 10 86 88 90 92 94 96 98 100 102 1987 1992 1997 2002 2007 2012 Consumption (rhs) Housing equity withdrawal (lhs) Unsecured lending (lhs) Per cent of disposable income Per cent of disposable income

• There is some relationship between consumption and housing equity

withdrawal

Real house prices and consumption (annual growth)

• 20
• 15
• 10
• 5

5 10 15 20 25 30

• 10
• 5

5 10 15 1987 1992 1997 2002 2007 2012

Consumption (rhs) House prices(a) (lhs)

Per cent Per cent

• House prices and consumption are well correlated, but not clear whether

this reflects causality

Key points on household balance sheets

• Big increases in debt largely reflect higher house prices
• Falls in real interest rates and looser credit conditions are likely to have

been a factor behind those increases

• Household acquired assets at a broadly similar rate to liabilities
• There have been significant changes in the distribution of balance sheets
• No evidence of a debt fuelled consumption boom, but hard to tell from

aggregate data

• Increases in wealth are likely to have supported consumption

Research on how debt affects consumption

Saving ratios of different groups of households

• High debt mortgagors have typically had the lowest saving ratios, but

they have seen the biggest increases since the crisis

• 20
• 10

10 20 30 1992 1995 1998 2001 2004 2007 2010 Outright owners Renters Low debt mortgagors High debt mortgagors Total LCFS Per cent

Motivation

• High debt levels may make households more vulnerable to shocks
• Have indebted UK households made larger adjustments to their

spending in response to shocks associated with the financial crisis?

• Evidence from the US suggests debt has been important, Dynan

(2012) and Mian, Rao and Sufi (2013)

• Understanding how debt affected consumption before the crisis is

also important

Should debt affect household spending?

• In a simple life-cycle model, households borrow or save to

smooth their consumption

• Debt has no effect on spending decisions
• But assumptions of the simple model may not hold

– Households’ ability to borrow may change – Households are not certain about their lifetime incomes

• Debt could affect consumption temporarily, if households ability

to bring forward consumption changes

• High debt levels may make households respond differently to

shocks

Research design

• Household level consumption equation including a mortgage

debt to income variable

• Coefficient on the leverage variable is allowed to vary over time
• Life-cycle type model, similar to that used in housing and

consumption literature (Attanasio et al (2009)), but including debt variable

• Does not prove whether there is a causal relationship between

debt and consumption

Research Design

it it i t it it 1 it

e X cohort year year Y D Y D β c      

' 5 ' 4 ' 3 ' 2

β β β * / β /

log of real non-housing consumption for household, , at time, ratio of outstanding mortgage debt to household disposable income time dummies cohort dummies, captures a pseudo fixed effect Controls

i t

it

c cohort Y D / year X

vector

We estimate the following equation:

Data

• Living Costs and Food Survey (formerly FES)
• Repeated cross section of UK households (household level)
• 1992 – 2011
• 5,300 per year
• Use non-housing consumption
• Secured debt data: level of outstanding mortgage debt

Results

• Households with high debt to income ratios tend to consume

more

• On average, household with a debt to income ratio of 3 rather

than 2 have consumed about 2% more

• Impact of debt on consumption does vary over time

• Debt had a larger effect on consumption in the early 2000s and a smaller

effect after 2007

• 5
• 4
• 3
• 2
• 1

1 2 3 4 5 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Significant at 5% level Debt & year interaction terms (β2) Per cent

Regression results

Impact of a 1 unit increase in debt to income ratio

• n consumption, relative to 2007

• Overall effect of debt on consumption is always positive
• 4
• 2

2 4 6 19921994 199619982000 2002 20042006 2008 2010 2007 impact Impact relative to 2007 Total Per cent

Regression results

Impact of a 1 unit increase in debt to income ratio

• n consumption

Regression results

Durable Non-durable

• Durable consumption effect seems to have been stronger pre-crisis
• Similar effects on durable and non-durable consumption after the crisis
• 6
• 4
• 2

2 4 6 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Significant at 5% level Debt and year interactions (β2) Per cent

• 6
• 4
• 2

2 4 6 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Significant at 5% level Debt and year interactions (β2) Per cent

Impact of debt on aggregate consumption

90 110 130 150 170 190 1992 1995 1998 2001 2004 2007 2010 National Accounts data Excluding the estimated impact of debt Indices (1992 = 100) 2 4 6 8 10 12 14 1992 1995 1998 2001 2004 2007 2010

National Accounts data Excluding the estimated impact of debt

Per cent

Non-housing consumption Saving ratio

• Increases in debt were largely matched by an accumulation of assets, but

some evidence that debt supported consumption before the crisis

Impact of debt on aggregate consumption

1 2 3 4 5 6 2007 2008 2009 2010 2011 Estimated impact of debt Other factors National Accounts Total LCFS total Percentage point change from 2007

Contributions to the rise in the saving ratio from 2007

• Changes in how debt affects consumption have weighed on consumption

growth since 2007 and have pushed up saving ratio

Causality versus differing expectations

• Our results do not prove that there is a causal link between debt

and consumption.

• High debt households may consume more because they expect

faster future income or house price growth

• Difficult to distinguish between:

i. High debt levels encouraging households to spend before the crisis and leading them to make larger revisions afterwards

• ii. Some households taking on high debt because they became

increasingly optimistic about future income prior to the crisis and then making larger revisions afterwards

Causality versus differing expectations

• 2012 Bank of England/NMG survey is potentially helpful
• Cross sectional survey of 4000 households, asking questions

about household finances

• Includes questions about whether households have cut spending

due to debt concerns and about income shocks

• Limited by only asking these questions in 2012 and by asking

about income shocks over the past year

NMG survey results

Reduced spending in response to debt concerns Yes No Median debt to income ratio 2.4 1.7 Proportion that experienced a negative income shock that they expect to persist(a) 32% 12% Proportion who are think that a fall in income is quite likely over the next year(b) 30% 15%

• Household who have cut spending due to debt concerns are much more

likely to have had an income shock or be uncertain about income

• Suggests debt may have amplified the impact of income/uncertainty shocks

NMG survey results

• Impact of uncertainty seems bigger for high debt households, but income

shocks have a similar impact on high and low debt

20 40 60 80 100 <2 2+ Fall in income not likely Fall in income quite likely Mortgage debt to income ratio Percentage of households that cut spending due to debt concerns 20 40 60 80 100 <2 2+ No income shock Income shock Mortgage debt to income ratio Percentage of households that cut spending due to debt concerns

Mortgagors Debt to income<2 Debt to income>2 Non-mortgagors Proportion that experienced a negative income shock that they expect to persist(a) 21% 21% 20% Proportion who are think that a fall in income is quite likely

• ver the next year(b)

22% 21% 23%

NMG survey results

• No evidence that high debt households are more likely to have had an

income shock

• But question only refers to period between 2011 and 2012

Conclusion

• Increases in debt 1992-2007 largely matched by accumulation of assets
• But debt may have provided some modest support to consumption in the

decade before the crisis

• Since 2007, the extent to which high debt households consume more

has fallen back, weighing on consumption growth

• Hard to distinguish between there being a causal relationship between

debt and consumption and high debt households having different expectations

• But we find some evidence that high debt levels may have amplified

shocks to income since 2007