our own and may not necessarily reflect the view of the Federal Reserve Bank of New York or the Federal Reserve System Investors Appetite for MoneyLike Assets The MMF Industry after the 2014 ID: 799571
Download The PPT/PDF document "The views expressed in this presentation..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
The views expressed in this presentation are our own and may not necessarily reflect the view of the Federal Reserve Bank of New York or the Federal Reserve System
Investors’ Appetite for Money-Like Assets: TheMMF Industry after the 2014 Regulatory Reform
Marco Cipriani and,
Gabriele La
Spada
Slide2Hölmostrom, 2015
Dang et al, 2015; Lester et al. 2011 and 2012; Hanson et al 2015.
Which assets can be used as money?
(as a means of payment)Immune from adverse selectionInformation insensitiveMoney-like assets: debt-like securities with short maturity and/or low risk; debt on debtPublic money-like assets: TreasuriesPrivate money-like assets: Money Market Funds (MMFs) before 2016
Recent Theoretical Literature on Money
Slide3Recent empirical literature estimates the premium for money-likeness (convenience yield)
Krishnamurthy and Vissing
-Jorgensen (2012); Nagel (2017); van
Binsbergen et al. (2018)Convenience yield for Treasuries ranges between 20 and 40bpsMoney-likeness estimated indirectlyK-VJ: comparing yield on Treasuries with that of securities with a similar risk profileThe Premium for Money-like Assets
Slide4In 2014, a regulatory change (the
SEC MMF reform) affected the information sensitivity
of the shares issued by
a segment of the MMF industryQuasi-natural experimentDifference-in-differences designEstimates of the value attached to informational insensitivityPremium for money-likenessFirst empirical evidence directly supporting information-based theories of money
Our Work
Slide5Outline
Introduction
Some
institutional backgroundThe impact of the 2014 reformThe premium for money likenessIn progress
Slide6Economic Importance2017: $3 trillion AUM managed by
334 MMFs Three Types of MMFsPrime:
Treasuries, GSE debt, repos, CD, CP, ABCP, FRNs
Muni: mainly municipal securities in the form of FRNsIn this presentation: prime={prime,muni}Government:Treasuries, Agency debt, and repos backed by treasuries or Agency debt Two investor types (share classes): institutional and retailMoney Market Funds
Slide7Before 2014
investment in all MMFs redeemable at par (stable NAV)mandated repricing if NAV drops below 0.995 (breaking the buck)
no gates or
feesMMF shares: debt-like securitiesakin to uninsured bank depositsTypical example of privately-issued moneyUsed for payroll purposes by non-financial firmsRule 2a-7 before the 2014 reform
Slide8New SEC regulation approved in 2014
A reaction to the 2008 run on the industryAll prime MMFs: redemption gates
and
liquidity feesOptional if weekly liquid assets fall below 30%“Mandatory” if they fall below 10%Institutional share classes of prime MMFs: floating NAVGovernment MMFs unaffectedRegulation took effect in October 2016
The 2014 SEC Reform
Slide9Prime MMFs became
less “money-like” Institutional share classes to a greater extent than retail
Why? the reform made prime MMF shares less informationally insensitive:
retail investors must now consider the possibility that fund managers gate redemptions or introduce feesadditionally, institutional MMF shares are no longer a debt-like security: institutional investors have an incentive to acquire private information on the underlying MMF portfolio at all timesThe 2014 Reform and Money-likeness
Slide10Outline
Introduction
Some institutional background
Investors’ response to the regulationThe premium for money likenessIf time allows: intertemporal rate of substitution
Slide11Government vs. Prime: Total Net Assets (TNA)
Jan. 2015
Sep. 2017
Δ
Total
MMF
$3,057bn
$3,034
bn
-$23bn
Prime
MMF
$2,054bn
$796bn
-$1,258bn
Gov
Share
$1,003bn
$2,238
bn
+$1,235
bn
Gov Share32.8%73.8%+41 pp
New SEC rule
Slide12Within-Family Flows
Each dot is a fund family
Sample: Nov. 2015 – Oct.
2016OLS regression in level: slope 1, R2=0.92This relationship is absent before 2015
Slide13Retail vs. Institutional Investors
Retail
Institutional
Jan. 2015
Sep. 2017
Jan.
2015
Sep. 2017
Prime
$656
bn
$380
bn
$1,151
bn
$209
bn
Gov
$189
bn
$569
bn
$796
bn
$1,562
bn
Gov
Share
22.4%
59.9%
40.9%
88.2%
Institutional
Retail
+37.5
pps
+48pps
Slide142008 Run vs. SEC Reform
2008 Run
SEC Reform
Industry SizeShrankConstantFlows from Prime to GovernmentAcross familiesWithin families
To funds mainly specialized in Treasuries
To funds
mainly specialized in agency debt
Response
of Retail Investors
Hardly noticeable
Present (albeit weaker)
Consistent with the idea that
MMF investors wanted
to preserve the money-likeness of their
investment
Consistent with the idea that
MMF investors
responded to an increase in perceived risk and sought safe
assets
Flight to safety as opposed to preference for money-like
assets!
Slide15Outline
Introduction
Some institutional background
Investors’ response to the regulationThe premium for money likenessIf time allows: work in progress
Slide16Prime vs. Government Net Yields
Slide17Increase in the net yield of prime vs. government MMFs after the regulation
Spread: 8bps through Nov. 2015; jumps to 25bps in Oct. 2016; and has remained above 14 bps since.
Why? Investors are willing to pay a premium for money-like assets (government MMFs)
Spread for institutional share classes should increase even more (floating NAV)Confounding factors:Increase in risk taking by prime MMF managersDifference-in-difference approachThe Premium for Money-like Assets
Slide18The Diff-in-Diff Regression
Monthly regression at the family level: Jan 2015-Sept. 2017
i
family; j=share class (institutional or retail); t month; k fund typeyi,j,k,t:
average net yield weighted by share-class TNA
For each family in month t, there are four observations:
t
he weighted average net yield of its prime institutional funds
the
weighted average net yield of its prime
retail funds
the
weighted average net yield of its
government institutional
funds
the weighted average net yield of its
government retail
funds
The Diff-in-Diff Regression
Controls:αi,j,k= family X fund type X investor type FE
μ
j,k= investor type X month FErobust standard errors (heteroscedasticity, cross and autocorrelation)Two regulatory dummiesI(t≥Nov.2015): first conversions take placeI(t≥Oct.2016): reform goes into effectOne “Prime” and one “Inst.” (institutional) dummy
Slide20The Diff-in-Diff Regression
γ
1
+γ2 = premium for money-likeness paid by retail investorsγ3+γ4 = additional premium for money likeness paid by institutional investors
Slide21The Value of Fixed NAV: Net Yield
γ
1
+γ2 = premium for money likeness paid by retail investors=20 bpsγ3+γ4 = additional premium for money likeness paid by institutional investors=8 bpsestimates are comparable to Krishnamurthy and Vissing
-Jorgensen (2012)
and
Nagel (2017) for Treasuries
Slide22Robustness checks
Balanced panelLonger sample (from November 2010)
Saturated regression with MMF-type X time
FEPre-existing trendsTime-varying risk aversion (VIX)Monetary policy (EFFR)Risk taking by MMF managers
Slide23Robustness Check: Risk Taking
The reform may have impacted the relative risk taking of prime vs. government MMFs
Remaining prime MMF investors more risk tolerant
Prime fund managers invest in riskier asset classesFour proxies for fund-risk taking:Weighted average maturity (WAM)Weighted average CDS spreadPortfolio share of risky minus safe asset classesRisky asset classes: bank obligations
Safe asset: Treasuries, Agency debt and repos
Portfolio share of safe asset classes
Slide24Relative Risk-taking has not Increased
Relative risk taking by prime MMFs has not
increased
We re-estimate premium adding risk proxies as controls; results doe not change.
Slide25Outline
Introduction
Some institutional background
Investors’ response to the regulationThe premium for money likenessIf time allows: understanding the premium
Slide26Net Yield = Gross Yield - Fee
Therefore, the premium for money likeness observed after the reform must come either:From a change in relative fees
From an change in relative gross yield
unrelated to riskWhy unrelated to risk? because we have learned that the increase in the spread is not due to riskWe re-run our baseline regression using fees as the dependent variableUnderstanding the Premium
Slide27The Role of Fees
γ1+
γ
2 = the relative reduction of fees for retail investors is 10 bpsγ3+γ4 = the additional relative reduction in fees for institutional is 2 bpsoverall, slightly less than 50% of the premium for money likeness is paid through a reduction in feeswhat about the remainder?
Slide28Supply Effects
Half of the premium for money-likeness stems from a change in relative fees
The remainder? (Possibly)
higher relative rate paid by (institutional) prime MMF borrowers because of a reduction of a supply of funds from prime MMFsHypothesis: because of supply effects, rates on assets typically part of prime MMF portfolios have increased more than rates on assets typically part of government MMF portfolios
Slide29Relative Gross Yield and Market Rates
Slide30Outline
Introduction
Some institutional background
Investors’ response to the regulationThe premium for money likenessIf time allows: the value of information insensitivity
Slide31Information Sensitivity: Gates and Fees
Gates and fees are triggered when weekly liquid assets (WLA) follow below a threshold:
the
money-like premium should decrease with the level of WLA
Slide32If the reform changed prime MMF information sensitivity,
prime MMF investors should be trying to acquire more information relative to government MMF investors
SEC platform (EDGAR) allows us to extract information on the number of
queriesInformation Sensitivity: Gates and Fees
Slide33The SEC reform of the MMF industry offers a quasi-natural experimental allowing the estimate of the premium for money-likenessThe premium is linked to a change in the informational sensitivity of prime MMFs
The premium is estimated on private moneyThe premium is estimated to be around 20 bps for retail investors and 8 bps for institutional investors
The estimate is robust to controlling for changes in portfolio risk
Conclusion
Slide34The End
Thanks!
Slide35Form N-MFP: monthly regulatory filing with the SEC
November 2010-presentFor each fund: type, TNA, gross yield, weighted average maturity (WAM), securities-level informationFor each share class: redemptions, subscriptions, net yield
iMoney
Net:November 2010-Septmeber 2017: 90% TNA coverageFor each share class: institutional, retailNet yieldDATA
Slide36Within-Family Flows
--- Levels
Each dot is a funds’ family.
Sample: Nov. 2015 – Oct. 2016.
Slide37Within-Family Flows --- Regressions
Slide38Within-Family Flows --- Excluding Conversion
Slide39Conversions
Sample: Nov. 2015 – Oct.
2016
Slide40Fund Families
Sample: Nov. 2015 – Oct. 2016
Slide41Within-Family Flows --- Control
November 2014-October 2015
Slide42Two dots per fund family: agency and treasury
MMFs
Sample: Nov. 2015 – Oct.
2016Within-Family Flows: Treasuries vs. Agencyβ
=0.90
β
=0.10
Slide43Investors’ Risk Appetite: Agency vs. Treasury Funds
Agency
Treasury
Jan. 2015
Sep. 2017
Jan.
2015
Sep. 2017
Total
$508bn
$1,574
bn
$495bn
$664bn
Share
16.6%
51.9%
16.2%
21.9%
+35.3
pps
+5.7pps
New SEC rule
Slide442008 Run: Agency vs Treasury Funds
Slide452008 Run: Within-Family Flows
Each dot is a funds’
family
Sample: Aug. 2008 – Oct. 2008OLS regression in level: slope 0.51, R2=0.20
Slide462008 Run: Agency vs Treasury Funds
Two dots per fund family: agency and treasury MMFs;
Sample: Aug. 2008 – Oct.
2008.
Slide47Risk Regressions
Slide48Adding Risk-taking as a Control
We re-estimate premium adding risk proxies as controls; results doe not change.
Slide49Funds with Stable Investor Base
Restrict sample to share-classes with stable investors baseShare classes whose TNA did not change by more than 5% in absolute value between Nov 2015 and Oct 2016
Regressions run at the share-class level
Premium estimates are similar
Slide50Robustness Check: Time-Varying Trends
Slide51Robustness Check: Time-Varying Risk Aversion
Changes in investor risk aversion may have differentially impacted prime and government MMFs at the time of the reform
We proxy risk aversion with the VIX
We add as controls the interactions of (lagged) VIX with the Prime and Institutional dummies and with the regulatory dummies (triple and quadruple interactions)
Slide52Controlling for Time-Varying Risk Aversion
Slide53….and Monetary Policy
Slide542008 Reserve Primary Fund Run
Aug. 2008
Oct. 2008
Δ
Jan. 2011
Δ
Total
$3,478
bn
$3,501
bn
+23
bn
$2,684
bn
-$794
bn
Prime
$2,578
bn$2,099 bn-479 bn$1,903 bn-$675 bn
Government$900 bn
$1,402
bn
+502
bn
$781
bn
-$119
bn
Gov. Share25.9%
40.0%+14.1 pp
29.1%
+3.2 pp
Slide552008 Run: Retail vs. Institutional Investors
Institutional
Retail
Only institutional investors participated in the run
Slide56Why Institutional Investors Pay More?
More sophisticatedAttentive to changes in the characteristics of their investment
Institutional MMFs are subject to a floating NAV
Contrary to the policy debate around the rule, the regression results suggest that fees and gates have had a greater impact on prime MMFs than the floating NAV