UK money growth in the long expansion: what can it tell us about the role of money ? Michael Mcleay Ryland Thomas The Causes and Consequences of the Long UK Expansion Conference 20 th September 2013
Background • Since the financial crisis began a lot of work has been done on the impact of credit and money on the economy • Impact of credit supply shocks – Gilchrist & Zakrajsek (2012), Bleaney, Mizen & Veleanu (2013) – BoE work on UK: Bell & Young (2011), Barnett & Thomas (2013) • Impact of money supply and demand shocks – Motivated by analysing QE which works via money supply eg Cobham and Kang (2012), Bridges and Thomas (2012) – Money supply vs demand shocks, Chadha et al (2008) • What happens if we apply this analysis to the long expansion period ?
Two underlying questions • What was the role of the money ‐ creating sector as a source of shocks ? – Monetary shocks versus standard macro shocks eg Chadha et al (2013) – Can we distinguish between different types of shock in the monetary sector; eg Competition, Wholesale Funding costs, Bank risk taking • What was the role of money in propagating shocks ? – Credit may have direct allocative effects on the economy – Does the money created by credit matter ? – Does it lead to monetary overhangs and hot potato effects ? – Are these persistent enough to show up in the data so money has incremental information about future activity/inflation ?
Reasons not to bother with money • What’s the right measure of money ? – Broad versus narrow ? – Retail versus wholesale ? – Divisia versus simple sum aggregates ? – Financial versus non ‐ financial sector ? • The poor experience of monetary targeting in the UK – Link between money and inflation only strong at long horizons/certain regimes – Short ‐ term movements in velocity unpredictable – Relationship depends on shocks hitting the economy and policy regime in place (eg Sargent and Surico (2010) – But does not mean there is no information in money • But many of these issues apply to other variables/concepts – Output gaps, marginal cost gaps, labour market gaps, natural rate of interest
We don’t have a standard model • Many textbook models have no explicit role for bank ‐ created money • A lot of work on incorporating banks into macromodels – no workhorse model that can easily be taken directly to the data – More progress on incorporating credit than money ? – Difficult to embody hot potato effects, difficult to track distribution of money holdings over time – Requires heterogeneous agents, sequential transactions in decentralised markets • So approach has to be empirical but informed by different theories
Our eclectic approach • What features do we implicitly want to capture ? – Credit may have allocative/supply effects – Post ‐ Keynesian view of endogenous money by created by credit (and other transactions) so focus on broad money. – Buffer ‐ stock ideas: money accepted but not necessarily demanded – Individual agent versus aggregate adjustment: hot potato effects • Use a cointegrated VAR approach – Allows us to incorporate monetary and other real disequilibria directly (M ‐ M*) – Can then carry out SVAR or SEM analysis depending on the focus • Look at aggregate and sectoral models – Fundamental shock identification easier to do at aggregate level: Part 1 – Propagation role of money easier to discern at sectoral level: Part 2
Money and credit in the long expansion M4(a) M4 Lending (b) Nominal GDP percentage change on a year earlier 40 30 20 10 0 ‐ 10 ‐ 20 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 • Nominal variables look relatively stable in long expansion period • But various pointers/puzzles point to impending financial crisis
Fact 1: Credit growth > Money growth Chart 2: Sterling counterparts to broad money over the long expansion £trillions 1.2 1 0.8 0.6 0.4 0.2 0 ‐ 0.2 ‐ 0.4 ‐ 0.6 1993 ‐ 1997 1998 ‐ 2002 2003 ‐ 2007 Net lending to the public sector Loan securitisations Net lending to intermediate 'other financial corporations' Net non deposit liabilities Net lending to non ‐ residents Foreign currency position M4 lending Broad money Credit grew faster than money for most of this period • • Build up of non ‐ deposit wholesale liabilities, securitisation and overseas funding
Fact 2: Money growth > nominal GDP Velocity of money Inflation, broad money growth and nominal GDP growth Index, 1993=100 Percentage Nominal GDP 200 Long expansion change on a year Broad money 180 14 earlier M4x CPI inflation 160 12 140 10 120 8 100 Divisia 80 6 60 4 40 Notes and Coin 2 20 0 0 1963 1973 1983 1993 2003 2013 1993 1998 2003 2007 All measures of money generally grew faster than nominal GDP over this period • • But there was no large pick up in CPI inflation despite double ‐ digit money growth
Fact 3: Corporate money more volatile Chart 5: Contributions to broad money growth by sector Household 4Q growthrate and PNFC 14 Non ‐ intermediate OFCs 12 10 Total 8 6 4 2 0 ‐ 2 ‐ 4 1998 2003 2007 • Most of the pick up in broad money growth was due to the corporate sector • Household sector money stable • Implications ?
The monetary sector as a source of shocks • Shocks emanating from banking sector v standard macroeconomic shocks – How important were they in the lead up to the crisis ? – Typical method is to add some balance sheet measure (eg quantity of credit) and a measure of spreads to standard macro time series to try and identify these separately • Different types of banking sector shocks – Can we use money in addition to credit to help us identify these – Use simple Monte ‐ Klein banking model to motivate this – Competition/cost of providing intermediation services – Wholesale funding costs/availability – Bank risk taking – In principle this also applies to QE and macroprudential policy
The monetary sector as a source of shocks Loan rates Credit Cost of ↓ ↑ intermediation Wholesale ↓ ↑ funding costs Bank risk taking ↓ ↑
The monetary sector as a source of shocks . (Monti ‐ Klein model) Deposit rate = Mark-down x [wholesale funding cost - intermediation] cost risk free+ risk premium Loan rate = Mark-up x [wholesale funding cost + intermediation + compensation for losses] cost expected + unexpected losses risk free+ risk premium
The monetary sector as a source of shocks Loan rates Credit Deposit rates Money M ‐ M* demand (M*) Cost of ↓ ↑ ↑ ↑ 0 intermediation Wholesale ↓ ↑ ↓ ↓ ↑ ? funding costs Bank risk taking ↓ ↑ 0 0 ↑
Estimating a Cointegrated SVAR Type of variable Data series Standard macro CPI inflation variables GDP y Bank Rate is Credit and Money Real Credit (M4Lx(ex)) m4lx ‐ p Real Broad money (M4x) m4x ‐ p Loan rate (Corporate bond yield) ib Deposit rate id Asset prices Long ‐ term govt. bond yield il Real exchange rate e Real equity prices pk • 10 variables VAR, over long sample period 1964Q1 to 2012Q4 – Allows us to compare long expansion period with earlier periods – But economy subject to structural change over this period
Data choices – credit spreads • Use corporate bond yield spread to proxy credit spreads: – Advantages: • Broad concept, likely to be related to overall credit conditions • Long time series back to 1960s – Disadvantages • not exclusively measure of bank loan rate spreads • only reflects household rates indirectly through bank funding costs
Long ‐ run relationships Description Relationship Money demand relationship m4x ‐ p = 0.5*y + pk + 8*(id ‐ is) + k 1 Term structure is = il + k 2 is = + k 3 Real interest rates Real equity price to GDP pk = y + k 4 Corporate bond spread ib = il + k 3 LR test of restrictions: 2 (24) = 42.004 [0.0129 ]* • Features – Can get theoretically appealing long ‐ run relationships but on the borderline of being rejected – No long ‐ run credit demand relationship – Credit spread is stationary, but deposit spread is non ‐ stationary
Estimate ‐ 5 cointegrating relationships • Given 5 long ‐ run relationships suggests there are 5 permanent shocks driving the stochastic trends in the data and 5 temporary shocks – King et al (1991), Mellander et al (1992), Robertson and Wickens (1994) • We use a mix of long ‐ run and timing restrictions to identify the shocks – Sign restrictions difficult to impose in a general way in this framework and can yield implausibly large contemporaneous effects from demand and policy shocks – But mix of LR/SR restrictions chosen appear to yield correct signs
Macroeconomic shocks Type of shock Permanent shocks Temporary shocks Aggregate supply Technology/TFP Cost push/mark up Aggregate demand World demand/preferences Domestic demand/confidence Monetary policy Core inflation/inflation target Deviations from rule • 6 Macro ‐ shocks split into permanent and temporary shocks – Nominal growth pinned down by inflation target shock – No long ‐ run real wage resistance so preference shock has no long ‐ run effect on output – Standard timing restrictions based on nominal rigidities
Recommend
More recommend
