1 Start: policy makers views on macro -finance 1. Framing: why study - - PowerPoint PPT Presentation

1

Start: policy makers’ views on macro-finance

• 1. Framing: why study financial cycles?
• Financial cycles as part of macro-financial linkages

 “Theories” on why and how financial cycles arise

• 2. How to Identify Financial (and Business) Cycles
• Different approaches for measuring
• Level vs. trend cycles: technical and policy differences
• 3. How to increase use of financial cycles in policy
• Link with forecasting and macro-financial linkages
• Financial stability analyses, vulnerabilities, and cycles

 Link with policies (macroprudential, e.g., CCyB)

• Macro-financial linkages tools at Federal Reserve Board

“I wasn’t used to thinking of the banking and financial sector as having such a critical role…Floating exchange rates, tightening budgets, liberalizing markets and worrying about wages were all according to the book…Tesobonos were not” Stanley Fischer, First Deputy MD, IMF, Euromoney, September 1997 “In the interest of full disclosure: This is a first pass by an economist who, until recently, thought of financial intermediation as an issue of relatively little importance for economic fluctuations…” Olivier Blanchard, Chief Economist, IMF, WP/09/80, April 2009

3

Before e the cris isis is

• Long shadow of two separate strands in literature

 Efficient financial markets. And RBC, new-Keynesian

• Limited evidence on supply-side channels

 “Finance is a veil”

Th The cr cris isis is highlighted many questions-old, new

• Financial markets less than efficient. Finance

matters in originating and propagating shocks

• Importance of macro-financial linkages

After r the cris isis is more evid idenc ence

• Empirical research on macro-financial linkages

 Much on demand-side; some on supply-side channels

• Still

l man any y (more) e) ques estions, tions, no unifie fied d frame amework work

4

 Part of question: “Why Macro-Financial Linkages”?

• Theories suggest financial cycles can arise, related

to, but also independent of business cycle

• Demand: financial accelerator mechanism
• Supply: also accelerator, and intra-financial system

 Mechanisms matter for measurement and policy!!

• Empirical approach for measuring will (thus) vary
• Demand side

 Overall/Sectoral, to identify accelerator effects

• Supply side

 Study (intra-) financial system, to get miss- pricing/allocation

Endogenous developments in financial markets propagat gate and amplify fy shocks ks in real economy



Corporate finance foundation

• Financial positions of agents (asset prices, debt,

wealth) affect access to finance → real outcomes



Variations built into D(S)GE models

1. External finance premium (Bernanke-Gertler, 1989) 2. Collateral prices (Kiyotaki and Moore, 1997) 3. Debt deflation (Irving Fisher, 1933), more recently 4. Open economy models (Krugman, 1999, etc.)

 Various other frictions (cash-flow, working

capital, trade finance, default probabilities, etc.)

 Frictions related to heterogeneity, project and

technological choices, productivity, governance

 Financial frictions operating via labor markets

• Demand: firms’ financing constraints affect hiring
• Supply: housing/mortgages

 Information, uncertainty

• More volatility, less precision in signals/credit standards

 ZLB/ELB can aggravate effects/externalities

7

 Net worth, asset prices decl

cline ine and collateral less valuable → final borrowers (corporations, households, sovereigns) have less access to finance, worsens real economy, adverse erse feedba backs cks – “Busts:” Asset prices drops, credit declines, external financing crunches with adverse real consequences

 Net worth, asset prices incr

crease ase and collateral more valuable → final borrowers have more access to finance, strengthens real economy, positiv itive e feedbac backs ks

• “Booms:” Asset price booms, credit increases, capital

inflows, lead to amplified positive real consequences

 Financial system (supply side) can be a source of shocks,

amplification and propagation. Several channels:

 Bank

nk lend endin ing g and d bank nk capital pital chan hannels els

• Some borrowers (households, SMEs) are bank dependent.

Traditional channels of transmission of monetary, regulatory

• policies. New work on unconventional policies, QE, etc.

 Leverage

erage and d liquidi quidity ty chan annels ls

• Like capital channel, but not just for banks. Plus indirect, asset

price effects through (common) exposures. And (complex) interactions among financial intermediaries and markets, through traded securities, etc. Newer area of research.

 All show how financial sector create, amplify shocks  Cros

• ss

s secti tiona nal, l, intercon erconnec ectio ions, s, SIFIs, FIs, TBTF TF

• Many externalities, market failures within financial system

leading to system risk. Interacts with time-series dimension

9

Financial Accelerator again: shocks to assets or equity affect bank → amplify, propagate shocks

• 1. Capital shocks
• 2. Asset price shocks
• 3. Liquidity shocks, fire sales
• 4. Leverage cycles

 Interact with “standard” FA → double whammy → General equilibrium → virtuous+vicious cycles

• Intra-system dynamics: not (ye

yet) well understood erstood

10

assets equity debt assets equity debt assets equity debt increase in value of securities Final balance sheet increase in equity Initial balance sheet After positive asset price shock new purchase of securities new borrowing

11

Demand mand + Su + Supply pply Sid ides es Combined

• mbined Mean

ans: s: Many linkages, no unified framework

Financia ncial exposur ures es (st stock cks and flows) ) betwe ween en se secto tors rs

Household sector Corporate sector External sector

Public sector Other financial intermediaries

Banking system Financial markets

 Two main (complementary) approaches in the

business and financial cycle literatures

• 1. “Classic cycle” approach (turning-point analysis)

 Identify significant turning points (peaks and troughs)

• 2. “Growth cycle” approach

 Detrend: define cycle relative to a trend

⇒ Business Cycles (BC) and Financial Cycles (FC)

13

 NBER-based algorithm to determine peaks, troughs Based

• n Burns, Mitchell (1946) and Bry, Boschan (1971) for GDP,

typically Quarterly frequency (BBQ)

• Popularized for business cycles by Harding and Pagan
• Can include Markov Switching (MS) models here (Hamilton, 89)

 Simple measures, as input for further analyses:

• Average duration/typical pattern of cycle: relevant for policy
• Explain duration/amplitude/slope by various macro-financial variables

and (monetary and fiscal) policy

• Analyze boosts and busts (“financial crisis”) as extremes FCs
• Analyze degree of BC/FC synchronization within, across countries

 Concordance index (Harding, Pagan 2002b): fraction of time any two series are in the same phase in their respective cycles  Amplification across FCs and across countries

14

 Hodrick-Prescott (HP) filter, with frequency range

• Idea: measure “excessive credit” (new debt contributing less

to growth (lower productivity, for assets, e.g., housing)

• Typical: one-sided HP filter with λ=400,000) for long FC

 E.g., Basel committee For credit-to-GDP:, as in CCyB  Can augment end points w. forecasts before de-trending

• Performs well as early warning indicator of future crises

 Latent Dynamic Factor, Frequency Domain

Analysis, Spectrum, Multivariate (MVF), etc.

• Latent factors, relate to many variables, including financial

 Use state-space, spectrum representation, Kalman filters, etc.  Detect commonalities, obtain unique cycle, w/o assumptions

• MVF variations much used (IMF, BIS, Norges bank, etc.)

 Can robustly add variables (i.e., output gap est. Borio et al (2013)), improve forecasting (Liu, Matheson, Romeu, 1998)

• Frequency Domain Analysis, Spectrum: more recent

15

 Claessens et al. (2011, 2012): analyze FC and BC

in 21 (+23) AM (+EM) countries; 1960-2007. Use turning point analysis

• FC (credit, house, equity prices): long, severe, highly

synchronized within countries

• Strong linkages between BC and FC, accentuate each other.

Recessions with financial disruptions longer and deeper

• Now in Claessens et al. 2016 for 75 countries 1960-2015

 Drehmann et al. (2012), Borio (2012): for 7 AMs.

Use pre-specified frequency-based filter method at which FC is assumed to operate

• FC best captured by house prices and credit, not by equity
• Peaks strongly associated with systemic financial crises
• Duration and amplitude of FC increased since 1985 (from 11

to 20 years)

16

 “Fundamentals-based” models

• Dynamic-panel estimates with country-specific intercepts.

e.g., Debt=f(GDP, real rate, country-specifics, lagged debt)

• “Long-run equilibrium” models. e.g., house price =

f(income, interest rates, credit, equity prices, demographics), + affordability, risk-taking, standards

 Intra-financial sector models with time/cycle

• Regime-shifting models; Joint Probability of Default (JPoD);

Systemic contingent claims analysis (SCCA); Distance-to- Default; Expected Default Probabilities (EDF); SRISK; DIP; CoVaR: Diebold-Yilmaz, etc. etc.  Some market price, some accounting-based. Most use interdependency, “jointness” of distress probability

17

 Is it the level or growth that worries? Both matter  Level-based

 Simple, but not country-specific, pros and cons  Still: Parametric or Non-Parametric? What selection criteria?

 Trend-based

 HP, other uni- or multivariate frequency filters? Multivariate approaches: what to include?  How to deal with end-of-period, new data? Short- vs. long- term trend?

 Frequency of updating, use market prices yes/no?  All use (some) real quantities, so what price?

⇒ FC (and BC) thus vary!!

AUS CAN JPN NOR SWE CHE GBR USA FRA DEU GRC ITA NLD PRT ESP BRA CHI IND IDN MEX POL RUS ZAF THA TUR

• 6
• 5
• 4
• 3
• 2
• 1

1 2

• 30
• 25
• 20
• 15
• 10
• 5

5 10 15 20 Economic gap Financial gap

Note: The "economicgap"is the average between the output gap and the deviation of the 2015 inflation forecast from target. The "financial gap" is the average between the credit

• to-GDP gap and the property-price
• gap. Data are reported in Table 1.

(Percent, latest available data) Countries facing possible trade-off between economic stimulus and

19

Economic and financial gaps based on: credit growth and house prices vs. equity prices and bond yields

Source: Sandri et al. (2014)

AUS CAN JPN NOR SWE CHE GBR USA FRA DEU ITA NLD ESP BRA CHI IND IDN MEX POL RUS TUR

• 4
• 3
• 2
• 1

1 2

• 2.5
• 2.0
• 1.5
• 1.0
• 0.5

0.0 0.5 1.0 1.5 Economic gap Financial gap

Note: The "economicgap"is the average between the output gap and the deviation of the 2015 inflation forecast from target. The "financial gap" is the average between the gap measures of equity required returns and sovereign term premiums (refer to GFSR, Oct 2014). Data are reported in Table 1.

Countries facing possible trade-off between economic stimulus and financial stability

Risky to rely on pure data (-mining

 Correlations can be positive due to other factors

• Fundamentals (shocks) and Policy actions (fiscal, monetary)
• International, via risk-sharing and otherwise

 Without real connection between two economies

• Second best: e.g., more markets does not mean necessarily

higher welfare

 Heterogeneity: cyclical positions, structural

features, variations in financial systems

 No formal benchmark—how to define excess?  Financial cycles are very low-frequency: what about

short-term movements? Ignore? ⇒Need to be able to tell story of the “why and what”

20

A.

• A. Co

Consi sisten stency cy bet etween een BC a C and FC C in foreca casts sts

 Evaluate how likely

ly a s sce cenari ario

• is, to judge a

forecast (but not necessarily as a forecasting tool)

 Generate distributi

ributions

• ns for GDP, investment,

consumption, employment

• Take variables for say 100+ countries, long period
• Do by country groups, say All, AE, EM, and LIC

 Then condition each real on financial variable(s)

• Real private sector credit growth, current account balance, real

house price growth, and possible output gap potential

 Get likelihood of a real scenario given “financial cycle”

22

 Stock and Watson type GDP “predictions”

• Include financial variables: stock returns, capital

flows, interest rates, valuation measures, commodity prices, risk factors, others

 Gilchrist, Yankov, and Zakrajsek (2009), etc.

 But allow for heterogeneity across countries

• EMs: Stock returns, term spread, default spreads,

portfolio flows help predict output growth. Global factors (corporate spreads, commodity prices) also

• US: Term spread, stock returns, commodity prices

lack any predictive power; increase in portfolio flows indicate future declines (Banegas, 2016)

GDP Q-growth and FS indicator modeled using factor-augmented VAR (FAVAR) or FA-quantile regressions

→ Models exist for each sector, but multiple overlaps → While still lots of judgme

gment nt needed, provide “FCs”

→And link with macroprudential policies (for example, CCyB)

28

Publ blic ic sec ector tor:

Public debt, external debt, FX exposure, short-term debt (including in FX)

Household seholds: s:

Debt (credit growth): especially housing, debt burden; FX exposure; speculative activity

Corporati

• rations:
• ns:

Debt (credit growth), debt burden; FX mismatches

Financial ancial interm rmedia diarie ries: s:

Exposures (FX, housing), debt accumulation (credit); funding; risk attitudes

Asset et Mar arket ket Ri Risk:

Real estate risk, equity asset prices, market misalignments

External rnal sect ctor:

• r:

Current account balance/GDP; EBA External debt/GDP

 High external liabilities:

• Sovereign debt
• Banking sector ST liabilities

→ FX risk to households

 Low FX reserve coverage  Banking sector

• Concentrated exposure to real

estate

• Had parent banks: FX “lender of

last resort”

 Lax supervision  Flattered public finances  Never-ending convergence

story

 ?...  High external liabilities:

• Sovereign debt
• Corporate sector debt

 Low FX reserve coverage  Banking sector

• Concentrated exposure to real

estate

• Mostly domestic banks: no FX

“lender of last resort”

 Lax supervision  Flattered public finances  Never-ending “growth

miracle”

 ?...

29

and/or and

And not much new or exotic either in recent credit/property booms and busts

• Confidence underpinned by new growth and financing paradigm (US,

Eastern Europe, Ireland, Spain. Earlier: Japan, East Asia)

– Rising debt and leverage across sectors

• Lax bank supervision (Ireland, Iceland, US. Earlier: Chile, Scandinavia)

– High exposures to real estate, high leverage – Declining lending standards, focus on collateral not risk, predatory lending

+ Increased competition in banking (Ireland, Iceland, euro area. Earlier: Chile, Mexico 1990s, Russia)

• Funding availability & low cost

– Wholesale (Ireland, Iceland, Northern Rock), cheap HQ funding (EE) – Mismatches: FX (Eastern Europe. Earlier: Asia), indexing (Iceland), duration (EE) – Inflated credit ratings (strong fiscal Ireland and Iceland, growth in most EE). Earlier: many emerging markets

→ New trigger/twist: money market runs, shadow banking, run on repos, spill-over of subprime/structured finance shock to non-US banks

1.

Historical: flow of fund cross-check

• Check that the real, overall forecast is consistent with the financial flows of

funds patterns (e.g., bank capitalization and credit growth). Less today

2.

Sub-sector, granular forecasts

• Make forecasts consistent with relevant financial sector developments. E.g.,

housing starts to be consistent with patterns in supply of housing finance, house prices,... Corporate sector investment to be consistent with overall external financing conditions…. Etc.

3.

Model-based forecasts

• Various DSGE-type + other models to assess overall impact of financial

conditions on real and other forecasts. And do up/down scenarios (“alt sim”)

• Examples: FRBNY DSGE and FRB models. Tend to be similar, but financial

intermediation is still hard to model, often have to resort to spreads

4.

Financial stability analyses

• Quarterly internal financial stability reports: assesses vulnerabilities,

spillovers to the real economy, domestic, international. Various tools

• Ad-hoc analyses, as risks/vulnerabilities appear