QE: What have we learned?, Lecture notes of International Finance and Trade

Download QE: What have we learned? and more Lecture notes International Finance and Trade in PDF only on Docsity!

QE: What have we learned?

Markus’ Academy, 3/24/

Arvind Krishnamurthy

Stanford University GSB, SIEPR, and NBER

What researchers and policymakers would like to know?

What is the impact of a given size of purchase/sale in a given

asset market in a given economic state on the macroeconomy?

 What are impacts on output? Distributional consequences?

International spillovers?

 Impact on inflation? Financial stability? Fiscal consequences?

 How do these impacts compare both in magnitude and extent to

conventional monetary policy?

Asset pricing

QE Event Studies

10 Year Treasury Yield (Left) and Trading Volume (Right)

Source: Krishnamurthy and Vissing-Jorgensen (2011)

Identification challenges

 Tight event windows ⇒ unlikely that economic news cause QE and asset market

reaction

 Identification challenge is around the channel(s) for QE

 “Conventional” broad channels:

 Signaling path of policy rate; signaling policy marker preferences

 Signaling news about economy

 “Unconventional” narrow channels:

 Impacts on liquidity premia (QE increases reserve balances)

 Impacts on risk premia (duration, credit, mortgage…)

 Impacts on safety/scarcity premia (QE changes supply of safe assets)

Difference-in-Difference (OIS vs. Gilt yield)

Source: Joyce, Lasaosa, Stevens and Tong (2011)

More “narrow” channel evidence

Source: D’Amico, English, Lopez-Salido and Nelson (2012)

MBS quantity evidence from DiMaggio, Kermani and Palmer (2015)

 If it is narrow channel mechanism, then MBS purchases should particularly spur conforming (not jumbo) mortgage originations, because Fed purchased conforming

Rodnyansky and Darmouni (2017): MBS QE and bank lending

 If it is narrow channel, then MBS not Treasury purchases should drive lending

 Banks hold different amounts of MBS and Treasuries in 2008Q1 (pre-QE)

 Spillovers to real estate lending, but less (none?) to C&I Lending

Asset Pricing Theory with Narrow Channels

 Any theory of QE must depart from a complete markets model and go

towards segmented markets

1. QE effects are “narrow” not “broad” --- they do not change the rep

agent’s SDF. Instead, they must be changing the SDF of significant

investors in the narrow market

2. Macro-calibration of rep agent SDF will get a demand curve that is

too elastic to be consistent with data

 Research needs to model the demand curves in the narrow market, and

map out what “narrow” means

Vayanos and Vila (2021)

 Model of the Treasury market yield curve delivering risk premia that are a function of supply

 Players:

 Preferred habitat investors (pension funds, insurance companies, bond mutual funds)  Yield curve arbitrageurs (hedge funds, bond dealers/bond trading desks)

 Arbitrageurs integrate the yield curve, demanding risk premia as compensation for interest rate shocks and future supply shocks:

𝜆𝜆𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟^ ∝ 𝛾𝛾𝛾𝛾𝑊𝑊, 𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤, 𝛾𝛾𝑤𝑤 = 𝑓𝑓(𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞 𝑜𝑜𝑓𝑓 𝑤𝑤𝑞𝑞𝑟𝑟𝑟𝑟)

 Risk premium on interest rate shocks give a way of thinking about a duration risk premium

 If arbitrageur risk aversion is high (e.g., balance sheet constraints) then risk premia are higher, and QE has a bigger impact  Duration local effects come from risk premia to future supply shocks

Duration Risk Premium and Spillovers

 Treasury yield also affected by safe asset demand effects.

 If 10-year preferred habitat investors (e.g., insurance company demanding 10 year safe bonds) increase their demand for 10-year bonds … the 10-year yield will fall.

 What is a pure duration risk-premium effect?

 Look at yield change on an asset not demanded by safe asset investors, but has duration risk, which the arbitrageur also prices

 E.g., non-investment grade corporate debt?

 And this is related to spillovers: what else does the arbitrageur pricing kernel price?

𝜆𝜆𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟^ ∝ 𝛾𝛾𝛾𝛾𝑊𝑊, 𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤, 𝛾𝛾𝑤𝑤 = 𝑓𝑓(𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞 𝑜𝑜𝑓𝑓 𝑤𝑤𝑞𝑞𝑟𝑟𝑟𝑟)

“Narrow” analysis from non-QE asset pricing research

 We can learn from understanding the impact of (---) buying 10-year

bonds, where (---) doesn’t have to be Fed

 Intermediary SDF, market segmentation, specialized demands

 Intermediary asset pricing (He and Krishnamurthy, 2013)

 Koijen and Yogo (2019) for equities

 Bretscher, Schmid, Sen and Sharma (2022) for corporate bonds

Macro effects of QE

QE Shock

Interest rate(s) in targeted market(s)

User cost of capital -> Investment

Household borrowing/saving rate -> consumption

Employment, Output

20

User cost of capital and firm investment

 Corporate expenditures will only respond to QE if QE affects the user cost of capital on the marginal unit of capital

 Suppose Google had two sources of capital  Cash (it has a lot…)  Corporate bond market

 The marginal source of capital is almost surely cash, where the user cost of capital is the nominal interest rate

 Corporate bond QE should be expected to have no effects on Google investment

 Evidence for the “no effect”: Acharya and Steffen (2020), Darmouni and Siani (2022)

Google Bond Yield and CDS; Fed Bond Purchase Program Announced 3/