Global financial cycles and risk premiums
33
Global financial cycles and risk premiums Òscar Jordà † Moritz Schularick ‡ Alan M. Taylor § Felix Ward ¶ Q Group, La Jolla, 2019 † Federal Reserve Bank of San Francisco; University of California, Davis ‡ University of Bonn; CEPR § University of California, Davis; NBER; CEPR ¶ University of Bonn Disclaimer: The views expressed herein are solely the responsibility of the authors and should not be interpreted as reflecting the views of the Federal Reserve Bank of San Francisco or the Board of Governors of the Federal Reserve System. 1/31
Transcript of Global financial cycles and risk premiums
Global financial cycles and risk premiums
Òscar Jordà†
Moritz Schularick‡
Alan M. Taylor§
Felix Ward¶
Q Group, La Jolla, 2019
†Federal Reserve Bank of San Francisco; University of California, Davis‡University of Bonn; CEPR
§University of California, Davis; NBER; CEPR¶University of Bonn
Disclaimer:The views expressed herein are solely the responsibility of the authors and should not be interpreted as reflecting the views of the
Federal Reserve Bank of San Francisco or the Board of Governors of the Federal Reserve System.
1/31
What we ask and do
Is there a global financial cycle? Has it grown stronger over time, andwhat are its drivers?
Long-run perspective: we study the synchronization of lending andasset prices in the global economy over the past 150 years
Focus on the international transmission of financial centermonetary policy
2/31
What we ask and do
Is there a global financial cycle? Has it grown stronger over time, andwhat are its drivers?
Long-run perspective: we study the synchronization of lending andasset prices in the global economy over the past 150 years
Focus on the international transmission of financial centermonetary policy
2/31
What we ask and do
Is there a global financial cycle? Has it grown stronger over time, andwhat are its drivers?
Long-run perspective: we study the synchronization of lending andasset prices in the global economy over the past 150 years
Focus on the international transmission of financial centermonetary policy
2/31
What we find
1 The global co-movement in credit, house prices, and equityprices has reached historical highs in the past three decades.
2 Particularly strong increase in equity price synchronizationsince the 1980s
3 The main driver of equity price co-movement is synchronizationof risk premiums, not dividends
4 Financial center monetary policy plays an important role forshifts in risk appetite
3/31
Where this fits in
Financial cycle and long-run asset marketsClaessens et al. (2011), Drehmann et al. (2012), Aikman et al.(2014) and Schuler et al. (2015); Jordà, Knoll, Kuvshinov, Taylorand Schularick (2018).
Global financial synchronizationBruno and Shin (2014); Cerutti et al. (2014); Obstfeld (2014); Rey(2015).
Monetary policy and risk premiumsDedola and Lombardo ()2012); Devereux and Yetman (2010);Fostel and Geanakoplos (2008); Krishnamurty and Muir (2017);Ward (2018); Miranda-Agrippina and Rey (2018); Kuvshinov (2018)
4/31
Data
1870-2016, annual, 17 countries:
Australia, Belgium, Canada, Denmark, Finland, France, Germany,Italy, Japan, Netherlands, Norway, Portugal, Spain, Sweden,Switzerland, U.K., U.S.
Credit, house prices, equity prices,GDP, consumption, investment and CPI-series
Jordà-Schularick-Taylor Macrohistory Database(Jordà, Schularick & Taylor 2016)http://www.macrohistory.net/data
Dividends and equity return premium
The Rate of Return on Everything, 1870–2015(Jordà, Knoll, Kuvshinov, Schularick & Taylor 2018)
5/31
www.macrohistory.net/data
6/31
The Global Financial Cycle, 1870-2016
7/31
Filtering and synchronicityBaxter-King band-pass filter for baseline (findings robust toother filters)
Use wide 2 to 32 year band
Report 15-year rolling-window Spearman bilateral correlations
Global measure is the average of the bilateral correlationcoefficients:
s̄t =∑
i∑
j<i si,jt
N ; N =n(n− 1)
2 . (1)
And a GDP-weighted equivalent:
s̄ωt =∑i
∑j<i
ωi,j,t si,jt , with (2)
ωi,j,t =(GDPi,t + GDPj,t)∑
i∑
j<i(GDPi,t + GDPj,t),
8/31
Business cycle synchronizationMore synchronization, especially after WW2
Cor
rela
tion
1885 1900 1915 1930 1945 1960 1975 1990 2005
−0.2
0.0
0.2
0.4
0.6
0.8
1.0
GDP Consumption Investment
9/31
Financial cycle synchronizationAsset prices growing more synchronized, especially equities
Cor
rela
tion
1885 1900 1915 1930 1945 1960 1975 1990 2005
−0.2
0.0
0.2
0.4
0.6
0.8
1.0
Total loans House prices Equity prices
10/31
What are the sources of rising equity price co-movement?
Equity price determinants:dividends Drisk-free rate Requity risk premiums ERP
We can directly observe dividends and risk-free rates
The residual time-varying equity risk premium is a measure forinvestor’s risk appetite: the compensation required to bear equitymarket risk
We can study the global co-movements of all three components:dividends, risk-free rates, risk premiums
11/31
Co-movement of safe rates R within historical range
Cor
rela
tion
1885 1900 1915 1930 1945 1960 1975 1990 2005
−0.2
0.0
0.2
0.4
0.6
0.8
1.0
ST rates LT rates
12/31
Dividend D and esp. return premium ERP comovement up
Cor
rela
tion
1885 1900 1915 1930 1945 1960 1975 1990 2005
−0.2
0.0
0.2
0.4
0.6
0.8
1.0
Dividends Equity return premium
13/31
Decomposing equity prices
Calculate risk neutral investor’s price QRN, based on presentvalue of dividends D for safe rates R
QRNt =∞∑k=1
k−1∏j=1
R−1t+j
Dt+k
How much equity price co-movement is due to co-movement inrisk neutral investor’s price QRN vs. residual component “riskappetite” ρ?
Qt = QRNt ρt
→ Compare comovement in QRNt and Qt
14/31
Most equity price co-movement due to risk appetite ρ
Cor
rela
tion
1885 1900 1915 1930 1945 1960 1975 1990 2005
−0.2
0.0
0.2
0.4
0.6
0.8
1.0
Equity prices Risk neutral price
15/31
Takeaways so far
Real and financial cycle synchronization trending up in the past 150years
Surging equity price synchronization since 1980s due to globalrisk appetite synchronization
Where does global synchronization in risk appetite come from?
16/31
Global risk-taking
Synchronized risk-taking in international markets: Bekaert et al.,2013; Miranda-Agrippino and Rey, 2015.
Global reach of U.S. monetary policy (Canova, 2005; Kim, 2001).Ehrmann et al. (2011)
Mechanisms: Bruno and Shin (2014), Bacchetta and van Wincoop(2013), Ward (2018)
17/31
Financial center monetary policy and global equity prices
Calculate international response of equity prices Q and risk neutralinvestor’s price QRN to financial center short-term rate changes ∆Rc
Local projections: let y be equity prices, dividends or safe rates; Xicontrols; h = 0, ..., 4
∆h+1yi,t+h = αhi +
5∑k=1
βhk∆yi,t−k +5∑
k=0
γhk∆Rci,t−k +5∑
k=0
δhkXi,t−k + ui,t+h
Intuition: (1) estimate effect of center policy changes oninternational dividends and risk free rates, (2) this yields the pricechange for a risk-neutral investor; (3) compare with actual effect ofpolicy changes on equity prices.
Difference between QRNt and Qt : effect of policy on equity riskpremium.
18/31
The response of global equity markets
Figure: Pre-1914 vs. Post-1980 equity price responses
Notes: Cumulative impulse response functions to +1ppt increase in financial center in-terest rates. Risk-neutral – risk neutral price (QRN). Center rate – financial center (U.K.and/or U.S.) short-term risk-free rate own response. Confidence bands calculated onthe basis of Driscoll-Kraay standard errors.
⇒ Excess equity price responses are a new phenomenon!
19/31
Financial center has large impact on risk premiums today
Figure: Pre-1914 vs. Post-1980 equity price responses
-10.0
-5.0
0.0
5.0
10.0
%
0 1 2 3 4Years
Pre-1914
-10.0
-5.0
0.0
5.0
10.0
%
0 1 2 3 4Years
Post-1980
Equity price
Risk-neutral
-10.0
-5.0
0.0
5.0
10.0
15.0
%
0 1 2 3 4Years
-10.0
-5.0
0.0
5.0
10.0
15.0
%
0 1 2 3 4Years
Dividend
-0.5
0.0
0.5
1.0
1.5
ppt
0 1 2 3 4Years
-0.5
0.0
0.5
1.0
1.5
ppt
0 1 2 3 4Years
Risk-free rate
Center rate
Notes: Cumulative impulse response functions to +1ppt increase in financial center in-terest rates. Risk-neutral – risk neutral price (QRN). Center rate – financial center (U.K.and/or U.S.) short-term risk-free rate own response. Confidence bands calculated onthe basis of Driscoll-Kraay standard errors.
20/31
The response of global house prices
-6.0
-4.0
-2.0
0.0
2.0%
0 1 2 3 4Years
-3.0
-2.0
-1.0
0.0
1.0
%
0 1 2 3 4Years
0.0
0.5
1.0
1.5
ppt
0 1 2 3 4Years
Full Sample
-6.0
-4.0
-2.0
0.0
2.0
%
0 1 2 3 4Years
House price Risk-neutral
-3.0
-2.0
-1.0
0.0
1.0%
0 1 2 3 4Years
Housing rent
0.0
0.5
1.0
1.5
ppt
0 1 2 3 4Years
Risk-free rate Center rate
Post-1980
⇒ No excess house price response, i.e. risk premiums not correlated. Kuvshinov(2018): Time Varying Risk Puzzle. No co-movement of discount rates between equityand housing in long-run sample. No joint pricing kernel...
21/31
Can countries decouple from the effects offinancial center policy with a floating exchangerate regime?
22/31
Do exchange rate regimes make a difference?
Use peg/float distinction from Ilzetzki et al. (2017) and estimate localprojections for floating/fixed:
∆hyi,t+h = αhi +
L∑k=1
βhk∆yi,t−k +L∑
k=0
γhk∆Rct−k
+
L∑k=0
δh∆Rct−k × floati,t +L∑
k=0
ϕhkXi,t−k + ui,t−1+h, h = 1, ...,H ,
(3)
where αi are country-fixed effects, ∆hyi,t+h are h-year changes thedependent variable and ui,t+h are error terms.
23/31
Yes for the full sample
-6.0
-4.0
-2.0
0.0
2.0
4.0
%
0 1 2 3 4Years
Pegs
-6.0
-4.0
-2.0
0.0
2.0
4.0
%
0 1 2 3 4Years
Floats
Equity price
All years.
24/31
Much less today
-20.0
-10.0
0.0
10.0
20.0
%
0 1 2 3 4Years
Pegs
-20.0
-10.0
0.0
10.0
20.0
%
0 1 2 3 4Years
Floats
Equity price
Post WW2.
25/31
Growing spilloversEquity price response, full sample:
Equity price response, Post-1945:
⇒ Peg-float distinction blurs in the second half of the sample26/31
Policy Shocks
27/31
Instrumenting policy changes
LP instrumental variable approach:
∆hyi,t+h = αhi +
5∑k=1
βhk∆yi,t−k +5∑
k=0
γhk∆R̂ci,t−k +5∑
k=0
δhkXi,t−k + ui,t+h,
(4)
R̂ct is prediction from a first-stage regression of the effective federalfunds rate Rc on the high-frequency instruments:
Rct = θ0 + θ1FF1t + θ3FF3t + θ6ED6t + θ9ED9+ θ12ED12+ ϵt , (5)
where FF1, FF3, ED6, ED9 and ED12 are Gertler-Karadi high-frequencyinstruments.
28/31
Robustness: high frequency instrument
-40.0
-20.0
0.0
20.0
%
0 1 2 3 4Years
Post-1990: US rate changes
-40.0
-20.0
0.0
20.0
%
0 1 2 3 4Years
Post-1990: HF-Instrument
Equity price Risk-neutral
⇒ OLS results robust to using Gertler-Karadi IV
29/31
Risk premium effects in bond markets-0.20.20.61.01.41.82.2
ppts
0 12 24 36Months after shock
Safe rates
-0.20.20.61.01.41.82.2
ppts
0 12 24 36Months after shock
Risky ratesPre-1914
Pegs Floats
-0.20.20.61.01.41.82.2
ppts
0 12 24 36Months after shock
Safe rates
-0.20.20.61.01.41.82.2
ppts
0 12 24 36Months after shock
Risky ratesPost-Bretton Woods
Pegs Floats
Source: Felix Ward 2017. Global risk taking, exchange rates, andmonetary policy.
30/31
Conclusions and policy implications
ConclusionsFinancial cycle synchronization up for past 150 years
Increasing equity price synchronization since 1980 due toglobal risk appetite synchronization
U.S. monetary policy a trigger of risk appetite across globalequity markets
Policy implicationsRisk-appetite spillovers can confound monetary policy andfinancial stability targets
Potential role for macropru and capital account policies
Potential benefits to international economic policy coordination
31/31
