safe assets scarcity, liquidity and spreads

Safe Assets’ Scarcity, Liquidity and Spreads

Gabriella Chiesa

Università Bologna

FT September 2, 2010

Siemens, Europe's largest engineering group, is setting up its own bank. As well as broadening its sources of funding, it would allow the company to deposit cash at the Bundesbank, Germany's central bank. "Frankly, we'd be happy to get no interest rate just to know our cash is completely safe“, a person close to the company says.

• Bloomberg 27.02.012

Berkshire Hathaway disclosed its liquidity instruments: 80% of $10.8 billion on non-US Gov. Debt are from Germany, UK, Canada, Australia and the Netherlands.

Comment (Guy LeBas chief fixed income strategist):

"If a firm is looking at government debt as a source of potential liquidity, then it's extremely important to remain in these bulletproof nations"

Why don’t simply use bank deposits as a liquidity instrument?

• deposits are bank liabilities: their safety is constrained by bank income pledgeability and/or by deposit insurance:

• limited to "small" amounts: bk deposits are money/liquidity instruments for households.

• In Holmstrom and Tirole's parlance,

"small deposits”, by virtue of deposit insurance,

constitute outside liquidity

• Safety of large/non-insured deposits is constrained by the income flows that a bank can credibly pledge as guarantee for its non-insured liabilities -- the constraints that limit inside liquidity.

• The further constraint that limits the money/liquidity property of "large" deposits is the degree of information symmetry and trust required for their "transferability" -- key ingredient for a money/liquidity instrument.

Siemens and Berkshire Hathaway's are just an example of the so called "institutional cash pools”:

• large, centrally managed cash balances of global corporations and institutional investors.

• Institutional cash pools have become increasingly prominent since the 1990s as a by product of globalization: rise of large global corporations and centrally managed cash pools; Centralised liquidity management by investment funds; investment styles: ETFs require separately managed cash pools.

(Pozsar, 2011, and with specific reference to financial/banking corporations Bruno and Shin, 2011).

preferred habitat is not deposits, but insured

deposit alternatives:

• Government insured securities (government debt securities) and

• privately insured money market instruments, such as REPOs and asset backed commercial paper where collateral provides safety and substitutes for government guarantee

(Gorton, 2010; Stein, 2010; Krishnamurthy and Vissing-

Jorgensen, 2012; Singh and Stella, 2012)

To sum up:

the demand for money/liquidity instruments of corporations, large sector of the economy and most active player in the real investment process and in the money/credit market, is not satisfied by M2 types of money, but by

securities that meet the requirement of safety of principal and liquidity.

(by direct holdings, and/or indirectly via REPO arrangements)

The world's outstanding stock of safe assets has expanded steadily over the period 2001-2007 **

** Shadow banking (Securitization , Repo) has a demand side dimension:

cash pool demand for insured deposit alternatives exceeded the outstanding amount of Govern. guaranteed instruments.

And declined impressevely since 2007

FT 05.12.011: The decline of "safe" assets Presenting the unwanted mutant offspring of the most important

chart in the world*…

Two asset class lost the safety status:

(i) loan/mortgages-securitization

(ii) sovereign debt of the "peripheral" countries of the eurozone.

(i) are private claims on real investments (inside liquidity)

(ii) are securities issued by governments (outside liquidity)

• The evidence is then of a general decline of the amount of liquidity instruments available for satisfying the non-household sector's liquidity needs.

• Parallel to the observed decline of inside liquidity and the peripheral debt downgrading is the fall in the yields of the bulletproof nations' debt (in the eurozone, Germany) and the increase of the yields of peripheral countries' debt (e.g. Italian sovereign debt)

Why be concerned?

• firms and financial institutions ongoing entities whose project completion requires renewed injections of resources.

• Limited pledgeability of project outcomes constrains the amount of outside finance:

need of hoarding liquid/safe assets to cope with adverse shocks and/or to take future investment opportunities.

Rather than hoarding liquidity themselves, firms may secure credit lines:

contract with a bank for the right to draw specific

amounts of cash by a given date:

• liquidity to the firm is provided by the bk,

the burden of liquidity hoarding is on the bk:

• bk's liability associated with a credit line provision must be backed up by an amount of liquid assets sufficient to make the bank deliver on its promise.

Liquidity provision is the key activity of banks:

the largest share of commercial and industrial loans are take-downs under loan commitments -- credit lines

The intimate relation between banks' liquidity provision and liquid assets holdings makes the availability of safe/liquid assets at the center of the credit/investment process.

This paper

• analyzes the interactions between the financial and the real sector in an environment where liquidity holdings is an input of the credit/investm process.

builds on Holmstrom and Tirole (1998, 2011), which provide a model framework in which liquidity conditions affect investm and asset prices, and incorporates a menu of outside liquidity instruments (sovereign bonds) that differ in terms of safety -- price fluctuations and collateral value.

→ The supply of liquidity is constrained in that income pledgeability limits inside liquidity, and not all sovereign debt is safe/liquid

• (i) pin down the determinants of liquidity/collateral premia and bond spreads, and with reference to the eurozone:

• (ii) the implications of the ECB policies on liquidity provision and credit, and

• (iii) the debt management policy that would increase welfare with no need for transfer payments.

Focus on the joint determination of firms'/banks' composition of liquid asset portfolios, real investment/credit-lines provision, liquidity/collateral premia and bond spreads:

i) The availability of liquid/safe assets is an important determinant of bond spreads, the tighter this is the greater bond spreads;

ii) social welfare can be improved by debt management: It amounts to insulate the safe part of the sovereign risky debt -- tranching the debt so as to create a security whose safeness is ensured by sufficient collateral (real assets and tax revenue). Such a policy increases welfare (aggregate invest/credit) and benefits the issuer (the I Gov) by reducing its cost of debt (with no cost for Eurosystem participants).

Literature

• Holmstrom-Tirole (1998, 2011). We incorporate a menu of outside liquidity instruments (sovereign bonds) that differ in terms of safety -- price fluctuations and collateral value. Focus on joint determination of firms'/banks' composition of liquid asset portfolios, real investment/credit-lines provision, liquidity/collateral premia and bond spreads

• The role of government debt in facilitating the intertemporal allocation of resources has been emphasized by several papers and relies on contractual frictions that limit the enforceability of claims on future income (a seminal paper is Woodford, 1990; Gorton-Ordonez, 2013).

• Bolton - Jeanne (2011) analyse the role for government debt securities as collateral for borrowing. The key assumption is the asynchronicity between resource availability and real investment opportunities (as in Woodford, 1990). They analyze various forms of fiscal integration that can mitigate the incentives to under supply safe debt and find that they reduce the welfare of the country that provides the "safe-haven" asset.

• We focus on debt management and find that insulating the safe part of the sovereign risky debt -- tranching the debt so as to create a security whose safeness is ensured by sufficient collateral (real assets and tax revenue) -- increases welfare and benefits the issuer by reducing its cost of debt.

Model three periods t=0,1,2. Agents are risk neutral and evaluate consumption streams according to U(c₀,c₁,c₂)=E(c₀+c₁+c₂) (A1) agents’ IMRS = 1 Agents’ endowments are sufficiently large to ensure that resource scarcity does not limit the investment scale, this will be constrained by contractual frictions (limited pledgeability) and safe assets' scarcity, not by resource scarcity.

Assets:

Storage technology (holding cash under the mattress) is prohibitively costly, purchasing power can be transferred into the future by investing in assets issued by firms/banks to be discussed below, and in sovereign-debt securities: G , I.

Gov. Bonds: G, I Outstanding amount: BG, BI

unit price at t= 0: qG, qI

• G safe: G’s unit value at date 1 is 1 for sure

• I risky: I unit value at date 1:

In eq. qi >1 , i=I, G (by A1)

*Liq. Premium: excess payment at t= 0 re to the expected value at t=1 : qi – 1 , i=G,I

a prob.p ;   b >a prob. 1- p

same expectedvalue asG :

ap+ b 1- p( ) =1 (A2)

• Securities can be used as collateral for borrowing. The fraction that can be raised per unit of collateral value is less than 1, the difference 0 < h < 1 is the ‘’haircut’’

Banks/Firms N firms/banks. A firm/bank i is endowed with Ai units of resources (capital)

All face the same investment opportunity: injection of resources at t=0 and t=1, return at t=2.

• t=0: i chooses investment size Ii which is also the amount of resources it invests at date 0.

• t= 1: reinvestment: 𝜃 per unit of investm. defined at t=0.

• t=2: return per unit of investment: y

• Limited pledgeability constrains outside financing:

pledgeable per-unit return: r < y

non-pledgeable R = y – r

(haircut creditors apply in extending loans backed by real investm.)

• Positive NPV : y > 1+ 𝜃 (A3)

• reinv. at t=1: 𝜃 > r (A4)

reinv. requires liquidity holdings:

S = 𝜃 - r

i is either a firm that faces a constant-to-scale real investment opportunity which requires one unit of resources at t= 0, and 𝜃 at t= 1, per unit of investment

The firm will be able to complete investment if its liquidity holdings at t= 1 does not fall below SI,

S = 𝜃 - r

i is a bank facing a continuum of borrowers/firms. Borrower has an investment opportunity that requires one unit at t= 0 and, with prob. λ, σ additional units at t= 1. Bk at t= 0 chooses the size of credit-lines portfolio I. Credit line allows bk's borrower to withdraw one unit at t= 0 and σ at t= 1. By pooling borrowers' liquidity needs, at t= 1 bk will face withdrawals of total amount λσI Reinterpreting 𝜃 as 𝜃= λ σ, then under the maintained assumption that income pledgeable to outsiders is r per unit, bk will satisfy borrowers' liq needs (credit lines withdrawals) iff its liq holdings at t= 1 does not fall below SI. If i is a bk, then I is the amount of credit extended by i.

• Liquidity needs:

• We assume:

This will imply strictly positive liq premia

S I iå

)5(1

AS

ASBB

iIG

Liquidity Demand and Inv. Choice

• Liquidity must be planned in advance: at t=0, i chooses its liq.assets portfolio: LG, LI

• The value of I gov. bond holdings at t=1 will depend on the state realization:

State Liquidity held Liquidity needed

α α LI + LG

SI

β βLI + LG SI

The % of the bond portfolio revenue that is pledgeable is 1-h, accordingly, for investors to be willing to supply funds to bank/firm i the following participation constraint must hold:

I+ qILI+qGLG - A < (1-h) [p(αLI+LG-SI)+

+(1-p)(βLI+LG-SI)] (PC)

The bank/firm expected profits:

non-pledgeable income:

RI + h [p(αLI+LG-SI)+(1-p)(βLI+LG-SI)]

• These profits will obtain provided the firm/bank holds sufficient liquidity so as to meet date 1 reinv. needs (credit lines withdrawals):

αLI+ LG > SI (LC)

We rule out gov. bond short-selling:

LI > 0 , LG > 0 (NNC)

The firm/bank optimization problem amounts to choose the investment/ credit-lines-portfolio size I, and its gov. bond portfolio (LI, LG) so as to max profits st the investor part.constr. (PC), the liq./reinv. constr. (LC) and the no-short-selling constr. (NNC):

Choose (I, LI, LG) so as to :

Max {RI + h [p(αLI+LG-SI)+(1-p)(βLI+LG-SI)]}

st

I+ qILI+qGLG - A < (1-h) [p(αLI+LG-SI)+

+(1-p)(βLI+LG-SI)] (PC)

αLI+ LG > SI (LC)

LI > 0 , LG > 0 (NNC)

State Idle Liquidity

α 0

β βLI+LG- SI ≡

(β-α)LI

At an optimum, investors’ part. Constr. (PC) and the liq constr (LC) bind: αLI + LG = SI . If not, then i would benefit by reducing the bond portfolio size: it would free up h units of capital per unit of portfolio reduction and thereby expand real investm. and profits. Hence at an optimum:

• Using I bonds for liq purposes entails holding (β-α) units of date 1 idle liq with prob 1-p, and 0 units with residual probab., per unit of I bond. The certain equiv. of this lottery is (1- α)

• Observation 1: I bond holdings entails a financial investment that yields (1 – α) units of date 1 idle liq. per bond.

Date 1 idle liquidity is partially pledgeable -- the amount of outside financing that can be raised per unit is 1-h, the residual fraction (haircut) h is financed with inside capital

→ I bond holdings then entails an unwarranted financial investment that subtracts capital to real investment undertaking.

• Substituting (LC) into the investor participation constraint (which at an optimum holds with equality) and using the identity (1-p)(β -α)≡(1-α) gives the size of investment I:

)1(]11[

1

hqqZ

Sq

ZLAI

IG

G

I

• a positive haircut h>0, implies that at eq. Z<0: using I bonds for liq purposes lowers real investm.

*I bond entails investm in date 1 idle liquidity ( (1- α) per unit).

Outside finance that can be raised per unit of date 1 idle liq is 1-h : the firm/bank invests h(1- α) units of its own capital per unit of I bond holdings: Real investm. falls accordingly.

hqqqZ

Sq

ZLAI

I

IG

G

I

1)1(

1

Let QI (qG) : ∂π(qI = QI,..)/ ∂LI = 0

QI (qG) = [1+ α(qG-1)]- h(1- α)

qI < QI (qG) → ∂π/ ∂LI > 0 (LI= SI/α , LG=0)

qI > QI (qG) → ∂π/ ∂LI < 0 (LI= 0, LG=SI)

qI = QI (qG) → ∂π/ ∂LI = 0 (LI>0, LG=SI-LI >0)

qI LI=0 LG=SI qI=QI(qG) (LI>0,LG>0)

1 LG = 0 , LI=SI/α

Q qG

Q= qG : QI(qG)=1

LI > 0 , if qI < QI(qG)

• The threshold 𝑄𝐼(𝑞𝐺) is decreasing in I bond volatility (increasing in α) and in non-pledgeability/haircut h :

• QI(qG) < qG (by 𝜌 > 1 and qG >1)

→ If the I bonds are used as liq instruments, then necessarily they sell at at a discount w.r.t. the safe G bonds.

Equilibrium: Aggregate Investment, Bond Prices and Spreads

At an equilibrium, necessarily:

• the G bonds are used for liquidity purpose, LG > 0

qI > QI(qG)

• the liq premium on the G bond is strictly positive – the amount of G bonds outstanding is absorbed entirely by the demand for liquidity : ∑LG = BG .

q*I

qI=QI(qG) (LI>0,LG>0)

1 LI = 0, LG=SI

Q q*G

q*G : ∑LG = BG ; ∑LG = S∑I - ∑αLI

Q= qG : QI(qG)=1

Q f(I vol, h)

ΣLI qG* s* qI

* Inv. Credit

Return on capital r*

Q < X

BI

X s > 0 > 1 (BG +αBI)/S

rmax

X< Q < 𝑋 < BI

Q Betw 1 between betw

Q > 𝑋

0 𝑋 𝑠 =𝑋 −1

1 BG/S rmin

𝑋= q*G: ∑LG = S∑I - ∑αBI ; 𝑋 = q*G : ∑LG = BG

𝑠 = 1 − 𝛼 𝑞∗

𝐺− 1 1 −ℎ𝑆

𝑅+ ℎ 1 −

𝑆+1

𝑅

• Credit, real inv. and return on capital are increasing in the stock of liquid assets, the reverse for liquidity premia and bond spreads.

• volat. affects bond spread, effect larger when liq is tight than when liq is abundant:

how the asset behaves when liquidity is abundant is less relevant than how it behaves when liquidity is tight (H-T 96)

• An increase in vol.

i) a substitution away from I bond;

ii) depletion in the stock of assets eligible as liq instruments: increase in the liq premium qG

*-1.

• As vol increases, eq shifts from the first row to the second row and for sufficiently high vol, I bonds lose the status of liquid asset (third row).

• Parallel effects are produced by an increase in non-pledg/haircut h: as h increases the opportunity cost of the unwarranted investment in date 1 idle liquidity that volatile assets entail, increases, and eventually the I bonds are excluded from asset holdings for liquidity purposes.

Example: Nov 9-011 h ↑↑ spreads exploded

ECB Policy

Lending Facility:

• eligible collateral: bk loans, sov. debt

• haircuts: lower than mkt ones

- implicit subsidy to sector that can access the facility (i.e. banks)

- positive externalities on non-banks, via market prices, as well as reduction of bond spreads.

• Deposit Facility A further safe/liquid asset (available only to bks). The rate at which deposits are rewarded defines the date 0 price of one unit of liq. at the future date 1: qDF =1/(1+iDF) This asset is a perfect substitute for the G bonds*, relevant if : qDF < qG

* In which case bks find it profitable to substitute bond holdings with deposits at the central bank: Liquid assets' availability increases and with it credit and aggregate investment, while liquidity premia and bond spread shrink. *July 11-012 Deposit Fac. rate was lowered to 0 (qDF ↑), REPO rate on G bonds lowered to 0/negative from that date

• Deposit Facility

Credit expansion as well as bond spreads :

deposit facility rate rather than cuts

(possibly the transferability of these claims -- ECB debt certificates )

deposit-facility rate and/or ECB debt certif.:

availability of safe assets:

cost of holding liquidity

• Drawback: mkt yields of gov. debt

• a policy that does not have these drawbacks and does not involve the subsidization which underlines the effectiveness of the ECB's lending policies

Debt Management

• It amounts to insulate the safe part of the sovereign risky debt :

• tranching the debt so as to create a security whose safeness is ensured by sufficient collateral (real assets and tax revenue).

Such a policy increases welfare and benefits the issuer by reducing its cost of debt

Debt Management: Secured Debt

The I bond can be viewed as a bundle of two securities: a safe one that pays α for sure, and a risky one that pays 0 with prob. p, and β - α with the residual prob. 1-p.

For a liquidity seeking institution the safe component is highly valuable, the risky component constitutes an unwarranted financial investment that subtracts resources to real investment undertaking.

• Unbundling the security package improves welfare.

It amounts to tranching the debt so as to create a security whose safeness is ensured by sufficient collateral (real assets and tax revenue)

The former I debt is replaced by

• αBI safe securities : each pays one unit with strict priority

• BI “risky” securities that make holders residual claimants, i.e. pay (β – α) with prob 1-p, and 0 with prob. p. These are tailored for “buy and hold” investors and priced qr =(1-p)(β - α), i.e.

qr = (1 - α)

• The safe security is a liq. instrument, perfect substitute of G bond

• The stock of safe/non-volatile assets raises to BG+αBI

safe security price: qs = qG

IG

G

G BBSq

ASq

1 : 

The revenue per unit of former package (former I bond) increases to qr+αqG :

qu = 1 + α l

l = qG – 1

unambiguously greater than the revenue per unit of the former security bundle.

• The spread between the liq instruments is nil

• Aggregate invest/credit

I  =

BG +aBI

Så

• A debt management policy that insulates the safe part of I sovereign debt benefits the issuer, reduces cost of debt, and produces positive externalities: it expands aggregate investment/credit whenever condition (C₂) fails to hold, and lowers sovereign G's cost of debt if (C₂) holds.

C2 = condition for former eq: ΣLI=BI (1st row of the Table)

If C2 holds, unbundling has no effect on aggr. investm/credit, however the price for liq. q*G ↑

Results

• i) credit, real investment and return on capital are increasing in the amount of liquid assets, the reverse holds for liquidity premia and bond spreads

• ii) bond spread is largely driven by the liquidity premium. Bond volatility impacts on bond spread, the tighter liquidity, the greater the effect.

• iii) the share of safe/liquid assets is constant (in line with the empirical evidence provided by Gorton, Lewellen and Metrick, 2012). • A corollary of i): Ricardian Equivalence breaks down

Scope for debt management:

• tranching the debt so as to create a security whose safeness is ensured by sufficient collateral (real assets and tax revenue) increases welfare and lowers the cost of debt:

• European Safe Bonds, where safety relies on tranching and collateral, are welfare improving. Interestingly, they benefit the bullet-proof nations too.