Download as pdf or txt
Download as pdf or txt
You are on page 1of 120

D EBT A N D F IS CA L

PO LICY I
U N D E R G R A D UAT E M AC R O E C O N O M I C S ,
OX F O R D 2 0 1 9 - 2 0 , R I C K VA N D E R P L O E G
OUTLINE OF THREE LECTURES
• A. Normative theories of public debt
– Ricardian equivalence
– Sustainability and dynamics of government debt
– Intertemporal tax smoothing
• B. Positive theories of public debt
– present in-office bias
– weak Minister of Finance: common-pool problem and debt bias
– partisan politics and strategic debt management
– signal political ability
– other debt distortions
– delayed stabilization: war of attrition/game of chicken
• C. Cost of debt and model of default risk and default
• D. Tackling debt bias and Fiscal Councils
• E. The balance sheet approach to government liabilities & assets and economic crises
A . N O R M AT I V E T H E O R I E S O F
PUBLIC DEBT

ROMER, SECTIONS 12.1-12.4


OUTLINE: PART A
• Chopping up flow and intertemporal government budget
constraints: solvency and no-Ponzi games
• Ricardian equivalence result
• Reasons why Ricardian equivalence might fail in practice
• Correcting for growth and inflation
• Sustainability of government debt
• Maastricht convergence norms
• Tax smoothing: minimizing PDV of tax distortions
• Keynesian stabilization policy
RICARDIAN
EQUIVALENCE
CONSUMER & GOVERNMENT FLOW BUDGET CONSTRAINTS
• For consumers/households the flow budget constraints are
d(B+E)/dt = r (B+E) +Y – T – C
with initial bond holdings and equity holdings B(0) and E(0)
• For the government the flow budget constraint is
dB = r B + G – T
with initial government debt B(0) given
• Integrating these budget constraints from the distant future
back to the present together with the no-Ponzi-games
condition gives the corresponding present-value budget
constraints
CONSUMER & GOVERNMENT PRESENT VALUE BUDGET
CONSTRAINTS
• Consumers present-value budget constraint is
PVt[C] £ financial assets + human wealth = At + PVt(Y – T)
with At = Et + Bt and PVt(Y)= òt¥ exp(-r(s-t))Y(s)ds
• Government’s present-value budget constraint is
PVt[T] ³ net public liabilities = Bt + PVt[G]
• Consumer present-value budget constraint thus becomes
PVt[C)] £ Et + PVt[Y – G]
• With logarithmic utility, Euler equation says that consumption
growth rate equals rate of interest r minus rate of time
impatience r > 0, so PVt[C] = òt¥ [exp(-rt) Ct exp((r-r)t)] = Ct/r
RICARDIAN EQUIVALENCE …
• Optimal aggregate consumption is thus
Ct = r (Et + PVt[Y – G])
• Hence, timing of taxes does not affect the household budget
constraint and does not affect consumption at all – it does
not matter whether public spending is financed by current
taxes or by debt (i.e., future taxes)
• Rational households anticipate that a tax cut must be
followed by a tax hike, so private consumption is unaffected!
• If they believe tax cut is followed by trimming size of public
sector, they will boost consumption (100% crowding in)
EXERCISE: CES INSTEAD OF LOGARITHMIC UTILITY

• With CES utility, u(C) = C1-1/h /(1-1/h) where h > 0 denotes


the elasticity of intertemporal substitution, consumption
grows at the rate h(r - r) – interpret this
• Show that the marginal propensity to consume out of total
wealth is now not r, but (1-h) r + h r and thus that
Ct = [(1-h) r + h r] (Et + PVt[Y – G])
• Interpret the effect of higher r on the MPC in terms of
intertemporal substitution and income effects
• Note that Ricardian equivalence result still holds
DOES RICARDIAN EQUIVALENCE HOLD?
• Imperfect capital markets: a tax hike implies that households
must borrow against the anticipation of lower taxes in the
future – what if they cannot offset current taxes by borrowing
or can do this only at a interest rate premium?
• Distorting taxes: tax rate might curb labour supply (driven by
intertemporal substitution effects)
• Birth of new generations of households: these may help to
shoulder the burden of future taxes, hence a temporary tax
cut may boost consumption (Weil, 1989, JPUBE
DOES RICARDIAN EQUIVALENCE HOLD? ..
• Death of households: you may not be around to enjoy the
future tax cuts or to shoulder the future tax hikes (if
households are not dynastic and no operational bequest
motive; annuity markets) – see Blanchard (1985, JPE)
• Ricardian equivalence less valid if financial markets are
underdeveloped
• Path of public spending is affected by costs and benefits - a
temporary tax cut financed by bond issue implies higher
taxes in future, so future primary spending G may need to be
lower and this will affect C
DOES RICARDIAN EQUIVALENCE HOLD? …
• Imperfect information: households may not realise that a
current tax cut must, for given time path of G, lead to a future
tax hike
• Imperfect insurance markets and the bird-in-hand problem
• Interest rate r and economic growth g are not exogenous
• Note that public spending hikes are fully crowded out: from a
present-value perspective any hike in public spending leads to
a corresponding fall in private consumption.
• Due to intertemporal consumption smoothing, private
consumption falls by the permanent increase in public spending
PUBLIC
DEBT
DYNAMICS
AND LONG-TERM
S U S TA I N A B I L I T Y
GOVERNMENT BUDGET IDENTITY
• Each period, the government must finance its expenditure
plans and pay interest on the government debt
• To finance its expenditure the government can tax, sell new
bonds or print money
• Government budget identity:
G + iB = T + ΔB + ΔH

gov’t exp. interest tax rev. new bonds new money

–We abstract from the possibility that the government can


borrow from the central bank (i.e. ΔH=0)
GOVERNMENT DEFICITS AND DEBT
• Actual government deficit = G + i B – T

expenditure revenue
• Primary government deficit = G – T
• Rearrange the budget identity
G + iB = T + ΔB

ΔB = (G - T) + i B
change in debt = primary deficit + nominal interest on
outstanding debt
EXPRESS AS PERCENTAGES OF GDP
• Divide the government budget identity (i.e.
ΔB º G - T)+iB by nominal GDP or Py
actual deficit DB G - T iB
º = +
GDP Py Py Py

primary deficit G -T
ºd = = g -t
GDP Py

actual deficit
= d + ib
GDP
B
where b º
Py
SOME ALGEBRA
B Rearrange
bº Þ B º bPy
Py DB
Db = - p b - g yb
Py
Use the following approximation Recall
DB » PyDb + byDP + bPDy DB
= d + ib
Py
Divide by Py Obtain
DB bDPy bDyP DbPy Db = d + ib - p b - g y b
= + +
Py Py Py Py Db = d + (i - p - g y )b
DB Recall that the real interest rate
= bp + bg y + Db
Py is defined as r = i - p , thus

Db = d + (r - g y )b
CASE 1
(REAL INTEREST RATE > GROWTH RATE) - UNSTABLE
CASE 2 (REAL GROWTH RATE > REAL INTEREST RATE):
GREECE BEFORE 2007

- high GDP growth rate (5.5% in 2006)


- access to borrowing at low interest rates (barely above
German Bunds)
- b = 107% GDP
GREECE IN 2007-09

Government reacts to financial crisis and recession


with fiscal expansion:
• primary deficit increases from -2% to -10.4%
• b increases to 129% GDP
GREEK CRISIS, 2009-10

• Doubt about Greek statistics on


• public debt
• government budget deficit
• GDP growth rate
• Change in market sentiment leads to higher borrowing costs
PAINFUL REMEDY SINCE

Largest fiscal consolidation in developed world


over such short period

• 10.4% of GDP over 3 years


FISCAL CONSOLIDATIONS 2009-13
Greece

10
Ireland
Iceland
Fiscal consolidation (2009-2013)

8
United States
Spain
Portugal
6
United Kingdom

France
Italy
4

Australia
Estonia Belgium
Germany
KoreaNorway
NetherlandsIsrael
Austria
2

Canada
Finland
Japan
New Zealand
Sweden
Denmark
0

Switzerland

0 50 100 150 200


Debt/ GDP in 2009)
Source: IMF, WEO
DYNAMICS OF GOVERNMENT DEBT
• Fractions of nominal GDP: b = B/PY, t º T/PY and g = G/PY
• Let d = D/PY = g – t denote the primary government deficit
and g the real rate of economic growth in GDP or Y
• Assume r > g, so that debt stability means government must
run a primary surplus
• If r < g, growth of economy big enough to cover debt
repayments so that a primary deficit does not necessarily lead
to debt instability
• Government deficit includes interest payments: DB = D + i B
GROWTH-CORRECTED GOVERNMENT BUDGET
CONSTRAINT
• Growth-corrected government deficit as fraction of GDP is
Db = (r – g) b + d
with real interest rate defined as r = i – p
• The uncorrected deficit-GDP ratio: i b + g - t = Db + (g+p)b
• A non-zero risk of sovereign default pushes up the interest
rate and worsens the debt dynamics
• See section 14.3 of Carlin and Soskice
• Ignore monetary financing (seigniorage – “tax” on money
holdings) and treasury borrowing from central bank
SUSTAINABLE GOVERNMENT DEBT
• Sustainability requires that debt-GDP ratio does not rise ad
infinitum or at least does not grow faster than the growth-
corrected real interest rate r – g
• This is so if the no-Ponzi-games condition holds, since then
-pvt[d] ³ bt, where pvt[d]= òt¥ exp(-(r-g)(s-t))d(s)ds
• Hence, the present discounted value of primary government
surpluses must equal at least the outstanding debt
• Sustainable deficit corresponding to stable debt-GDP ratio is
d+rb=gb
NO PONZI-GAMES CONDITION: DERIVATION
• Integrate db/dt = (r - g) b + d backwards in time gives
òt¥ [db(s)/ds - exp(-(r-g)(s-t))] b(s)ds =
= [b(s) exp(-(r-g)(s-t))] t¥
= lims®¥ [b(s) exp(-(r-g)(s-t))] - b(t)
= òt¥ exp(-(r-g)(s-t))d(s)ds
• Hence, we get b = PV[- d] = PV[t - g] provided that the
no-Ponzi-games condition
lims®¥ [b(s) exp(-(r-g)(s-t))] = 0
holds (i.e. debt-GDP ratio rises less rapidly than r - g )
MAASTRICT TREATY: BUDGETARY NORMS
• The corresponding primary surplus is then
–d = t – g = (r – g) b which implies i b + d = (g + p) b
• The Maastricht/SGP norms target 3% for the deficit-GDP ratio
and 60% for the debt-GDP ratio.
• With trend inflation of 2% and trend real growth of 3%, one has g
= 5% and sustainable deficit corresponding to a 60% debt-GDP
ratio would indeed be 5% of 60% = 3% of GDP (i.e. g b)
• Other than that, these numbers were picked out of “thin air” and
bias against investment and the golden rule and tend to favour
selling public assets
LONG-RUN DEBT SUSTAINABILITY

• Goal: what is needed to keep the debt/GDP ratio from growing Db£ 0?

Db = d + (r - g y )b
for Db £ 0
-d

(r - g y )
debt primary surplus /GDP
£
GDP (r - g y )

• For long-run debt sustainability, with a given long-run r in excess of the


expected long-run growth rate gy, there must be a long-run primary surplus
if the debt ratio is to be constant
IS GREEK DEBT SUSTAINABLE?
debt primary surplus / GDP
£
GDP (r - g y )

Recall:
there must be a long-run primary surplus if the debt ratio is to be constant

Parameter values:
• average primary surplus of 4.1% of GDP (d) (last MoU)
• average GDP growth of 2.0% (gy) (IMF forecast 2017-22)

Interest rate on the 10-year government bond 6.1% (average May)

Maximum debt ratio = 4.1% / (6.1%-2.0%) = 100%

Hence, outstanding debt/GDP = 181% (2016 end)


Is unsustainable!
IS GREEK DEBT SUSTAINABLE?

debt primary surplus / GDP


£
GDP (r - g y )

Change assumptions

Parameter values:
• average primary surplus of 4.1% of GDP (d) (last MoU)
• average GDP growth of 2.0% (gy) (IMF forecast 2017-22)

Interest rate on the 10-year government bond 5.2% (average 2009)

Maximum debt ratio = 4.1% / (5.2%-2%) = 128.1%

Outstanding debt/GDP = 181%


is still unsustainable!
IS GREEK DEBT SUSTAINABLE?

debt primary surplus / GDP


£
GDP (r - g y )

Change assumptions

Parameter values:
• average primary surplus of 4.1% of GDP (d) (last MoU)
• average GDP growth of 2.9% (gy)

Interest rate on the 10-year government bond 5.2% (average 2009)

Maximum debt ratio = 4.1% / (5.2%-2.9%) = 181%

Outstanding debt/GDP = 181%


INTERTEMPORAL TA X
SMOOTHING
R O B E RT B A R R O ( 1 9 7 9 , J P E )
HARBERGER WELFARE LOSS TRIANGLES
FROM DISTORTING TAXATION
• Welfare losses (CS + PS) are proportional to the square of
the income tax rate (and higher if wage elasticity is higher)
TAX SMOOTHING: BARRO (1979, AER)
• Ricardian equivalence is valid to first-order approximation
• Crucial idea is that government minimises present value of
distortions of raising tax revenue
• These distortions corresponds to the familiar Harberger
welfare loss triangles capturing losses in consumer and
producer surplus and are by approximation quadratic in the
tax rate t (assume marginal t is average t)
• So a tax sequence of 40% today and 40% tomorrow (ignoring
discounting) has a welfare loss of 0.32 and a tax sequence of
30% and 50% has a higher welfare loss of 0.34
TAX SMOOTHING ..
• This suggests it is efficient to smooth tax rates over time
• More formally, the government chooses a sequence of tax
rates to minimize the present discounted value of tax
distortions PVr(t2 Y)
subject to the present value budget constraint
pvt[g] + bt = pvt(t)
• This gives the optimal tax policy
tt = (r – g) (pvt[g] + bt)
(assuming that discount rate is same as real interest rate r)
TAX SMOOTHING ...
• Proof: taxes are smoothed if r = r, so t is constant over
time. Since the tax base Y is growing at the rate g, we must
use the growth-corrected interest rate r - g to evaluate
pvt(t) = t/(r - g)
Substitution into the PV-GBC gives the required result

• Optimal growth-corrected budget deficit is equal to gap


between actual and permanent primary public spending:
Dbt = (r – g) bt + gt – tt = gt – (r – g) pvt[g]
INTERPRETATION
• The tax rate is determined by servicing the outstanding public
debt and by the permanent component of public spending, i.e.
gP = (r – g) PVr-g(g), which amounts to the annuity value of
current and future public spending-GDP ratios
• No predictable changes in tax rate, so t follows random walk
• “If in doubt, .. smooth it out” – cf. LCH and PI hypotheses
• Exercise: if g is expected to rise at the rate q > 0, we get
gP = (r – g) g / (r – g – q) > g and thus there is a surplus as
Db = – q g / (r – g – q) < 0
• So save for anticipated hike in spending
INTERPRETATION..
• E.g. an expected increase in pension spending due to greying
of the population implies that the government must save for
these future commitments, since then the present value of
future spending exceeds current spending: (r – g) pvt[g] > gt
• A current war or deep recession necessitating temporary
public spending requires that the government must borrow
for these relatively high current commitments
• Countercyclical debt policy?
• Uncorrected deficit Db + (g + p) b may still be positive if the
growth-cum-inflation tax (g + p) b is high enough
INTERPRETATION...
• The deficit is determined by temporary increases in public
spending only; the tax rate is determined by permanent
increases in public spending
• Hence, no welfare case for a hike in taxes to pay off existing
public debt more rapidly
• Any unexpected additional public income should lead to
government saving if temporary and to lower taxes (or
possibly higher spending) if permanent
• Public investment projects with market rate of return can be
netted out of government budget constraint and does not
affect the tax rate. It should be financed by public borrowing
EMPIRICAL EVIDENCE BARRO (1979)
• Uses historical U.S. data 1917-1976 on inflation p with P the GNP
deflator, after-tax nominal rate on corporate bonds (1 - q) R as
proxy for expected inflation, Gbar and Ybar are normal/trend values
of real federal government spending and real GNP, and the stock of
nominal government debt outstanding at the end of the year B
• Hypotheses: growth rate of nominal public debt Dlog(B)
– outstanding debt and permanent spending have no effect
– constant = growth rate g if Y and G grow at same rate
– coefficient on inflation should be one (as it is nominal debt)
– coefficient on temporary spending = 1 if r = r, smaller if r > r
– downswing in the economic cycle should lead to debt expansion
EMPIRICAL RESULTS
• Empirics broadly agree with tax smoothing hypothesis
• Growth in nominal debt issue since WW2 can be
explained by small number of variables
• Outstanding debt and permanent or normal level of federal
spending play no role in explaining debt growth
• Inflation or proxy for expected inflation leads to one-for-
one growth in public debt
• If primary spending is higher than normal, debt expands.
• In recession the government runs up more debt
D EBT A N D F IS CA L
PO LICY II
U N D E R G R A D UAT E M AC R O E C O N O M I C S ,
OX F O R D 2 0 1 9 - 2 0 , R I C K VA N D E R P L O E G
OUTLINE OF THREE LECTURES
• A. Normative theories of public debt
– Ricardian equivalence
– Sustainability and dynamics of government debt
– Intertemporal tax smoothing
• B. Positive theories of public debt
– present in-office bias
– weak Minister of Finance: common-pool problem and debt bias
– partisan politics and strategic debt management
– signal political ability
– other debt distortions
– delayed stabilisation: war of attrition/game of chicken
• C. Cost of debt and model of default risk and default
• D. Tackling debt bias and Fiscal Councils
• E. The balance sheet approach to government liabilities & assets and economic crises
• Background: UK government debt 1694-2016
A . N O R M AT I V E T H E O R I E S O F
PUBLIC DEBT

ROMER, SECTIONS 12.1-12.4


United States
Public Debt,
percent of GDP,
1790-2014
SIMPLIFY: RECAP TAX SMOOTHING
• Keep matters really simple and get intuition and results quickly
• Assume zero growth and zero interest & discount rates
• Also assume 2 periods only and no defaults
• Flow and present-value government budget constraints:
B = G1 – T1 and G2 + B = T2 or G1 + G2 = T1 + T2
• Minimise present value of tax collection costs T12 + T22
subject to the government budget constraint
• This gives T1 = T2 = (G1 + G2)/2 º GP and B = G1 – GP
• Hence, debt is set to temporary spending component
• Taxes are smoothed over time to finance permanent spending
PRECAUTIONARY BUFFERS
• From microeconomics: households engage in precautionary saving if
they are prudent, i.e. if U’’’ > 0
• Let future tax revenue be uncertain: T2 + e
• If we replace Barro’s welfare loss function by general utility function
loss function L(T1) + L(T2), the government sets L’(T1) = E[L’(T2)]
• We thus have T1 = E[T2] if L is quadratic and L’” = 0 (as in Barro)
• However, if L’” > 0, we have L’(T1) = E[L’(T2)] > L’(E[T2]) and thus
the government sets T1 > E[T2] to be prudent (as L’’ > 0)
• The government accumulates precautionary buffers by setting
higher taxes and lower budget deficits, and thus saving more
• Even if G1 = G2, the government then saves and has B < 0
EXAMPLE
• Let L(Ti) = exp(Ti /R) where R > 0 is the risk tolerance parameter
• Assume that T2 is normally distributed with standard deviation s

• Hence, we have L’(T1) = exp(T1 /R)/R = E[L’(T2)] = E[exp(T2 /R)/R]

• Or T1 = R ln(exp(E[T2/R] + (s/R)2/2)) = E[T2] + R ln (s/R)2/2)) > E[T2]

• We thus see that current taxes are set higher than expected future taxes, and
more so if the future budget is more uncertain and risk tolerance R is low

• Note: This assumes that L’ > 0, L’’ > 0, L’’’ > 0


KEYNESIAN STABILISATION POLICY
• With temporary fall in output and tax revenue and rise in
unemployment benefits, tax smoothing suggests the
government should run up the deficit, not further tighten it
• If this is caused by temporary lack of effective demand,
Keynesian response is to boost aggregate demand by cutting
taxes and boosting consumer spending, by boosting public
spending, or by loosening monetary policy and thus boosting
investment and consumption
• Danger is that expansionary policy is hard to reverse
• If politicians follow pro-cyclical policies, worsening matters
KEYNESIAN STABILISATION POLICY..
• Effectiveness of fiscal policy is less in open economy due to
import leakage, but positive even if spending stimulus is
financed by tax increase – larger in developed countries
• With perfect capital mobility, fiscal expansion is ineffective
under a floating currency but powerful under a pegged
currency (Mundell-Fleming)
• Can become a negative effect in high-debt countries due to
anticipation of sovereign debt/currency crisis
• Automatic stabilisers soften the adverse effects of downturn
• See section 14.2 of Carlin and Soskice
MAASTRICHT NORMS: NO TAX SMOOTHING
• Maastricht norms demand that debt-GDP ratio had to go
towards b* = 60%, so taxes are hiked up and then fall
gradually as the debt-GDP ratio declines

Tax rate

Debt-GDP ratio, b
MAASTRICHT NORMS: FORMALLY
• Maastricht norms demand that debt-GDP ratio has to decline
towards b* = 60%
• Capture this by adding the cost of violating the debt-GDP target to
the welfare loss functions
• Hence, in case b > b*, choose sequence of tax rates to minimise the
intertemporal welfare loss criterion
pvt[t2 Y + y (b* – b)2] with y > 0
subject to the flow government budget constraint
Db = (r – g) b + g – t
• This gives falling (not smoothed) taxes from modified Euler equation
E[Dt] = – y (b* – b) < 0 if y > 0 instead of E[Dt] = 0 before
B. POSITIVE THEORIES OF
PUBLIC DEBT

ROMER, SECTIONS 12.5-12.7


ENDOGENOUS PUBLIC SPENDING
• Now allow for endogenous primary public spending too
• Minimise the welfare losses of tax distortions and falling short of
bliss or desired level of primary public spending G > 0, i.e.
T12 + a (G – G1)2 + T22 + a (G – G2)2 with a > 0
• Hence, the marginal tax rates and marginal value of public spending
have to equal the marginal value of public debt, i.e.
T1 = T2 = a(G – G1) = a (G – G2) so
T1 = T2 = G1 = G2 = a G/(1 + a) and no deficit, B = 0
Primary public spending and taxes are fully smoothed over time.
• Higher weight on public spending target a gives higher taxes
• If bliss levels change with time, taxes are smoothed but B ¹ 0
OUTLINE: PART B
• Use 2-period model to present different narratives and
theories of why debt management is politically distorted:
–1. Present in-office bias
–II. Weak Minister of Finance: dynamic common-pool
problems and debt bias
–III. Partisan politics and strategic debt management: try
to tie the hands of your political successor
–IV. Signal political ability
–V. Other reasons for debt distortions
–VI. Delayed stabilisation: war of attrition/game of chicken
WARNING

• Throughout the extensions of the theory of tax smoothing and


optimal determination of primary spending to allow for three types
of political distortions are illustrated with the 2-period toy model
with endogenous primary public spending and without loss of
generality assuming zero interest, discount and growth rates

• Still, the algebra is less important than the insights, and not strictly
necessary, so focus on the assumptions that are needed to get the
insights for each of the three types of political distortions
I. PRESENT IN-OFFICE BIAS
• Politicians like popular things to happen now rather than in
future, so tend to procrastinate and postpone taxation.
• And want primary public spending upfront
• Capture this with a present in-office bias b > 1 (cf. Aguiar
and Amador, 2011, QJE)
• Minimise b [T12 + a (G – G1)2] + T22 + a (G – G2)2
• Hence, b T1 = T2 = ba (G – G1) = a (G – G2) = l
• Putting this into the government budget constraint:
T1 = 2a G/(1+a)(1+b) falls with present-in office bias b
PRESENT IN-OFFICE BIAS ..
• Note T1 = T2/b < T2 and G1 > G2
and also note that B = (b – 1) G / (1 + b) > 0 if b > 1
where G is bliss or target level of primary public spending
• So present in-office bias causes too much public debt, too
low level of taxes and too much primary public spending in
the present (and by necessity the opposite in the future)
• Present in-office bias is a bit ad hoc, but can be justified by
re-election risk or as proxy for strategic debt management
(see III on partisan politics and strategic debt below)
II. WEAK MINISTER OF FINANCE
• Minister of Finance faces many spending ministers and may not
have much power, so there is a dynamic common-pool problem
where spending of each minister is too high and there is a bias
towards too much debt
• Illustrate with 2 different types of spending ministers aiming for
G and H, respectively, and ignore present in-office bias, so b = 1
• To show this, first derive non-cooperative solution
• So government budget constraints are
B = G1 + H1 – T1 and G2 + H2 + B = T2
WEAK MINISTER OF FINANCE..
• Use Principle of Dynamic Programming
• Spending minister for G choses G2 to minimise
(G2 + H2 + B)2 + a (G – G2)2 and thus sets
T2 = G2 + H2 + B = a (G – G2) and minister for H sets
T2 = G2 + H2 + B = a (H – H2) (i.e. “MC” = “MB”)
So, higher cost of debt means less spending and more taxes
• If G = H, G2 = H2 = (aG – B)/(2+a) and T2 = a(2G + B)/(2+a)
• Upon substitution, we see that welfare loss from period one
onwards for G minister is T12 + a (G – G1)2 + T22 + a (G – G2)2
or (G1 + H1 – B)2 + a(G – G1)2 + a(1+a)(2G + B)2/(2+a)2
WEAK MINISTER OF FINANCE…
• Minimising welfare to go with respect to G1 and B yields
T1 = a (G – G1) and T1 = a(1+a) (2G + B)/(2+a)2 so
T1 = (1+a)T2/(2+a) < T2
• Hence, we find that G1 > G2 and H1 > H2, and thus B > 0
• The cooperative solution internalises the dynamic
common-pool externality by minimising the sum of welfare
losses, so requires 2(G2 + H2 + B) = a (G – G2) = a (H – H2)
• Exercise: this cooperative outcome gives
T1 = T2 = 2aG/(4 + a) and thus G1 = G2 and
H1 = H2, and B = 0 so no upward public debt bias!
WEAK MINISTER OF FINANCE….
• No deficit bias if spending ministers take account of each other
• Not having a proper Minister of Finance means more public
spending and less taxes now, and consequently higher deficits –
more details in sections 13.1 and 13.2 of Persson and Svensson
• Size of public sector also larger, since spending ministers do not
face a too low marginal cost of spending
• Benefit of any spending is concentrated, whereas cost is spread
more widely
• To overcome this debt and public spending bias, the median voter
elects an ultra-conservative Minister of Finance with additional
votes or with veto rights in the Council of Ministers (Rogoff, 1985)
III. PARTISAN POLITICS AND
STRATEGIC DEBT FORMATION
• Partisan politics: parties and their clientele differ in their
preference about the size of the public sector (big vs small
government) as in Persson and Svensson (1989, QJE) or
differ in their preference about the type of public good (say,
rail roads versus motorways) as in Alesina and Tabellini
(1990, RES) – see sections 13.3.1 and 13.3.2 of Persson and
Svensson textbook and chapter 12 of Romer
• Incumbent faces an exogenous probability p of being
removed from office
PARTISAN POLITICS AND
STRATEGIC DEBT FORMATION..
• Republican government leaves more debt to ensure that the rival
Democratic opposition does not spend too much (as it likes a
bigger public sector than Republicans) or spend too much on its
partisan public goods – idea is to tie hands of a political successor
• For strategic reasons the incumbent leaves more debt the bigger
the partisan cleavage and the bigger the probability of losing office
• Reagan and Trump (?) leave a lot of public debt to restrain future
Democrat spending
• Transfer resources from future to present by borrowing to ensure
your political rivals do not frustrate your own political ambitions
EXAMPLE OF PARTISAN POLITICS
• Republicans are in office and only like G (“military spending”).
Democrats are in opposition and only like H (“social
spending”) with probability of Democrats winning election p
• Use again Principle of Dynamic Programming
• If Republicans win office, taxes and spending in period 2 are
T2 = a (G + B)/(1+a) and G – G2 = T2 /a
• If Democrats win election, taxes and spending in period 2 are
T2 = a (H + B)/(1+a) and H – H2 = T2 /a
• Expected second-period utility for the incumbent is thus
E[U2] =(1 – p)[{a/(1+a)}(G+B)2]+p[{a/(1+a)}2(H+B)2 + aG2]
DERIVATION
• If Republicans win office,
T2 = a (G + B)/(1+a) and G – G2 = T2 /a
• If Democrats win election,
T2 = a (H + B)/(1+a) and H – H2 = T2 /a

• Expected second-period utility for the incumbent is thus


E[U2] = (1 – p) {a(G + B)/(1+a)}2 [1 + 1/a]
+ p {a (H + B)/(1+a)}2 [1 + a(G - 0)2] or
E[U2] =(1–p)[{a/(1+a)}(G+B)2] + p[{a/(1+a)}2(H+B)2 + aG2]
EXAMPLE…
• Incumbent Republicans in period 1 set G1 & B to minimise
(G1 – B)2 + a (G – G1)2 + E[U2]
• The first-order optimality condition for G1 is
(G1 – B) – a (G – G1) = 0
• The first-order optimality condition for B is
0 = – (G1 – B) +
(1 – p){a/(1+a)}(G + B) + p {a/(1+a)}2 (H + B)
• The primary spending gaps and tax in period 1 are thus
T1 = G1 – B = a (G – G1), so G – G1 = (G – B)/(1+a)
and T1 = (G - B) a /(1+a)
EXAMPLE....
• Upon substitution into the second first-order condition,
T1 = (G - B) a /(1 + a) =
(1 – p) {a/(1+a)} (G + B) + p {a/(1+a)}2 (H + B) or
G - B = (1 – p) (G + B) + p {a/(1+a)} (H + B)
• If p = 0, B = 0. In general, optimal public debt follows from
2B = p [G + B - a (H + B)/(1 + a)] > 0 if H = G
• If H = G, then B = p G / [2(1+a) - p] > 0
• In general, public debt bias is bigger if chance of being kicked
out of office p is high, bliss level G is high, bliss level H is
low, and weight given to primary spending target a is low
APPLICATION: MANAGING SOVEREIGN WEALTH
FUNDS IN OIL-RICH COUNTRIES
• Oil revenue is typically anticipated, temporary and volatile
• Anticipated oil revenue, so borrow in advance of windfall
• Put the temporary component in an intergenerational
sovereign wealth fund (SWF), so the oil dividend is equally
shared between the present and all future generations
• So accumulate financial assets (abroad, not at home) when oil
revenue is high and live off the interest of the fund once the
oil has ceased to flow
• If oil price is expected to stay low forever, then cannot dip
into the fund but must raise taxes or cut spending
POLITICAL DISTORTIONS IN MANAGING SOVEREIGN
WEALTH FUNDS
• If there is no fund but central bank is preventing the currency from
appreciating, foreign reserves at central bank act as a de facto fund
• Put precautionary saving buffers in a stabilisation fund
• Putting oil revenue in sovereign wealth funds may not work, since
the fund may get raided for political reasons
• Political incentive to put oil revenue in illiquid, partisan investment
projects (e.g. schools or bridges in region where your part of the
electorate is)
• EU: Maastricht criteria, no borrowing constraints, lead to
inefficient cuts in investment in partisan strategic debt model
(Peletier, Dur and Swank, 1997, AER)
IV. SIGNAL POLITICAL ABILITY
• Deficit can result from attempts by elected leaders to signal their
abilities to voters (Rogoff, 1990, AER)
• Voters have better information about taxes they pay and public services
they receive than about the overall fiscal position of the government
• Suppose politicians differ in their ability to provide public services
cheaply: they then have incentive to choose high spending and low taxes
to signal that they are especially able
• Depending on how quickly voters cotton on and learn about the
government’s true fiscal position, one gets either cycles in fiscal policy or
a persistent debt bias
• Or: politicians differ in their ability to cut future wasteful public spending
as indicated by low taxes today (tax smoothing!) – this also gives
strategic rationale for debt bias
V. OTHER REASONS FOR DEBT BIAS
• Empirical evidence: spending and debt issue correlate with number of coalition
partners in government and with number of spending ministers in government
• A good Finance Minister is also prudent by using downward-adjusted estimate
of next year’s GDP to make his or her budget
• This leads over time to the accumulation of precautionary buffers
• However, such buffers are vulnerable to being raided by political rivals
including spending ministers of Minister of Finance’s own political party
• So reasons for deficit bias are politicians discounting too heavily because they
may be voted out of office and exploiting future generations and not having a
strong enough Minister of Finance
• Other reason is inadequate information (e.g. forecasts too optimistic,
government accounts opaque)
VI. DELAYED STABILISATION
• Disagreement about how to divide burden of cutting deficit can cause
delay in fiscal reform as each group tries to get other groups to bear a
disproportionate burden (Alesina and Drazen, 1991, AER)
• Typically, in times of hyper-inflation: battle about whether to raise
taxes on labour or on capital or whether to raise taxes or cut
employment government and subsidies
• Each party tries to delay to get a better deal for itself: war of attrition
• Thus signalling it is costly for a party to accept reform (cf. strikes)
• The less costly it is for a group to accept the larger share, the sooner
it decides that benefits of conceding outweigh those of continued
delay
• This does not explain a debt bias, but explains why a deficit persists
APPLICATION TO DEVELOPMENT AID

• External aid in developing countries should not ease pain


of unsustainable situation as this would delay the necessary
stabilisation.
• Effective foreign aid should be conditioned on an
unpopular stabilisation taking place and should shrink over
time if the stabilisation programme is postponed to
increase the incentives for rivalling parties to give in early
• So this makes a case for conditional aid
DEFICITS IN OECD
• Roubini and Sachs: countries that ran large deficits after
first oil price shocks in 1973 were short-lived and had
multi-party coalitions
• Is the relationship causal? Unfavourable economic and
budget shocks can lead to both deficits and weak
governments?
• Less empirical evidence for strategic deficits than for
common-pool problems (Belgium, Italy) and persistence of
deficits (Latin America)
D EBT A N D F IS CA L
PO LICY III
U N D E R G R A D UAT E M AC R O E C O N O M I C S ,
OX F O R D 2 0 1 9 - 2 0 , R I C K VA N D E R P L O E G
OUTLINE OF THREE LECTURES
• A. Normative theories of public debt
– Ricardian equivalence
– Sustainability and dynamics of government debt
– Intertemporal tax smoothing
• B. Positive theories of public debt
– present in-office bias
– weak Minister of Finance: common-pool problem and debt bias
– partisan politics and strategic debt management
– signal political ability
– other debt distortions
– delayed stabilisation: war of attrition/game of chicken
• C. Cost of debt and model of default risk and default
• D. Tackling debt bias and Fiscal Councils
• E. The balance sheet approach to government liabilities & assets and economic crises
• Background: UK government debt 1694-2016
C . C O S T O F D E B T A N D D E F A U LT

RO MER , SEC TI ONS 1 2 . 8 - 1 2 . 1 0


COST OF SUSTAINABLE DEFICITS
• If fiscal actions still satisfy present value budget constraint:
• Inefficiencies due to deviations from tax smoothing are probably small
except in times of war (e.g. relying too much on tax increases instead of
debt issue)
• Welfare effects due to failure of Ricardian equivalence
• Debt requires future distorting taxes, which curb future labour supply,
output, consumption and welfare
• Future generations suffer due to lack of public goods/higher taxes
• Adverse redistributive adverse effects of deficits from present to future
generations, leading to lower capital stock, higher real interest rate and
depressed wages (“crowding out”) – this corresponds to a redistribution
from workers to capitalists whilst workers are typically poor
DOES GOVERNMENT DEBT CROWD OUT CAPITAL?
• Not if higher debt leads to an equal rise in private savings
– Ricardian Equivalence: government debt implies future tax
increases, so savings rise to pay for these
• If saving does not rise by enough
– because, for example, agents do not allow for the higher taxes
their children will pay when leaving bequests
• then long-run crowding out will occur to some degree: a lower
capital stock will be accompanied by higher real interest rates
• The extent of crowding out could be large if
– saving is largely to fund spending in retirement
– this saving is not very sensitive to movements in interest rates
THE COST OF DISTORTIONS
• House and Tesar (2015): estimated costs of 1% fiscal improvement in the primary
fiscal surplus (t – g), “the target primary balance”
• Dynamic neoclassical general equilibrium model incorporating:
– present-value government budget constraint
– MRS between consumption and leisure = after-tax real consumption wage
– distortions from tax wedges (labour, consumption, capital)
– labour and capital mobility
– non-traded versus traded goods
• Assumptions:
– ongoing access to capital markets (no default)
– no tax evasion
– no anticipation of policies and no time inconsistency of policy
6
LESSONS FROM DYNAMIC GE TAX MODEL
• In baseline calibrated to Greek economy all types of expenditure and tax
policies cut GDP in both short and long run: typically 1% to 2% drops of
2014 GDP for a 1%-point improvement in primary surplus as % of
GDP so it is costly to tighten the fiscal stance.
• Static revenue calculations grossly under-estimate these costs, so must
allow for endogenous adverse effects on capital & labour.
• Meeting debt repayment schedule is much more costly due to Greece
being small and integrated in larger European economy: not taking
account of labour and capital mobility results in big over-estimate of
future revenue
• Delaying the implementation of tax hike or spending cut mitigates short-
run fall in GDP, but requires greater hardship in the long run – see figures
in next two slides 7
ESTIMATED COSTS OF IMMEDIATE POLICY TO
IMPROVE PRIMARY SURPLUS BY 1%-POINT OF GDP

Source: House and Tesar (2015) 8


BENEFITS & COSTS OF DELAYING FISCAL TIGHTENING

Source: House and Tesar (2015) 9


PUBLIC DEBT AND LOW INTEREST RATES?
• Safe interest rate has been below growth rate of GDP for some time
(see next slide), so fiscal cost of rolling debt over may be zero
• Even without fiscal costs, debt might crowd out capital but effect is
small as risk-adjusted rate is likely to be small if safe rate small
• The MPK is (due to mismeasurement or rents) lower than risky rate so
less crowding out of private capital as this depends on the MPK
• Multiple equilibria and risk of default lead to high risk premium on debt
• See Olivier Blanchard (2019, Presidential Address, AER):
https://www.aeaweb.org/aea/2019conference/program/pdf/14020_paper
_etZgfbDr.pdf and his lecture https://piie.com/commentary/speeches-
papers/public-debt-and-low-interest-rates
COST OF UNSUSTAINABLE DEFICITS
• If fiscal actions no longer satisfy present value budget constraint:
• Government default Forced or due to government choice because
combination of high taxes & low spending has become too unpopular
• Inflation An alternative to debt financing is money financing – in a recession
this may be helpful, but too much money in medium to long term can increase
inflation and can lead to hyper-inflation
• Sudden stops are associated with large costs and disrupt capital markets
• “Some unpleasant monetarist arithmetic”: a botched attempt to cut deficits and
get rid of inflation may eventually lead to even higher inflation due to the
higher debt service that has been run up (Sargent and Wallace, 1981, FRB of
Minneapolis Quarterly Review)
• Note that, unless default is forced, these are medium- to long-term problems
(e.g. at present interest rates are not so large)
WHEN WILL A GOVERNMENT DEFAULT ON DEBT?
• Strategic: Political cost of high taxes/low spending exceed political costs from debt
holders plus costs of being shut out and unable to borrow again for some time -
political cost from debt holders smaller if large proportion of debt is owned overseas.
• Forced: Even when public debt is sustainable, if government cannot roll over debt,
default is forced if borrowing is in foreign currency, or monetization occurs if debt is
issued in own currency
• Even a small probability of default requires a risk premium on debt, which by raising the
cost of debt, makes default more likely (so vicious circle)
• Borrowing in own currency significantly curbs (largely eliminates?) default risk as
market need not worry about what rest of market thinks – helps to understand why
the debt crises following the recession has been located in Eurozone
– http://mainlymacro.blogspot.co.uk/2014/12/government-debt-financial-markets-
and.html
– http://www.voxeu.org/article/panic-driven-austerity-eurozone-and-its-implications
MODEL OF DEBT CRISES AND DEFAULT
• What can make investors unwilling to buy debt at given interest rate? Is
such a crisis likely to occur unexpectedly?
• Default is “all or nothing”: if potential tax revenue T is not enough to
pay principal and interest R B with r = R – 1, government repudiates all
• The cumulative density function of uncertain tax revenue T is F(.)
• Debt pays interest factor R with probability 1 – p and zero else
• Investors are risk-neutral, so safe RS is independent of R and B
• Equilibrium condition: (1 – p) R = RS so p = (R – RS)/R
• R approaches infinity as default probability approaches one
• Default if and only if T < R D, so p = F(R D)
• Hence, probability of default increases in principal plus debt liabilities
MODEL OF DEBT CRISES AND DEFAULT ..
• Multiple equilibria in probability of default
• At A the probability of default is low and interest premium small
• At B default much more likely, and interest rate premium high
• Finally, there is equilibrium where default is certain (p = 1) and investors refuse
to hold government debt whatever the interest factor
• First and third equilibria are stable, but B is unstable (check it!)
• Small changes in fundamentals can lead to very different outcomes
• Default when it occurs may always be quite unexpected
• Default more likely if government borrows more, the safe interest factor rises
and distribution in potential tax revenue shifts downwards
• Each of these increases p, and makes it more likely that full default is the only
equilibrium
• See section 12.10 of Romer
D . TA C K L I N G D E B T B I A S A N D
FISCAL COUNCILS

B E E T S M A A N D D E B R U N , VOX B O O K , 2 0 1 8
TACKLING DEBT BIAS
• Strong minister of finance; not too many parties
• Fiscal rules such as Maastricht/SGP criteria: lots of critiques
• Why does Europe control debt at federal level? The fear is that a
union may encourage national fiscal expansions with harmful spill-over
effects
• Prudent fiscal rule: t = tP ³ gP + (rP – gP) b (reflects tax smoothing)
• Fiscal councils: (semi-)independent bodies to make sure fiscal policy is
sustainable over the long term; guard against deficit bias by providing
independent forecasts and calling government to account when
needed; CPB in Netherlands, CBO in the US, and OBR in the UK
• Too often fiscal watchdogs turn into lapdogs: Beetsma and Debrun
FISCAL COUNCILS
• Emerged after global financial crises due to success of independent
inflation-targeting central banks during 1990s and 2000s and fiscal
rules proved insufficient to guarantee prudent management of public
finances before 2007/8
• Fiscal Councils help to boost credibility of fiscal consolidation
packages and act as commitment device for successive governments
when consolidation is spread over time
• Mandates: to produce forecasts for growth, the output gap and the
pubic finances on which decisions confirming to their targets must be
made (OBR); positive policy analysis (CBO and CPB); normative
recommendations (Swedish Fiscal Council ..)
• Still far cry from the independence and mandates of ECB or BoE
• And some fiscal councils do not seem to have much bite at all
E. BALANCE SHEET APPROACH TO
GOVERNMENT LIABILITIES &
ASSETS AND ECONOMIC CRISES
PRACTICAL CONSIDERATIONS
• Public debt less is less problematic if countries have assets
(e.g. oil in the ground, infrastructure, equity holdings, etc.)
• Public debt is more fragile if the maturity structure is very
short, since then the ministry of finance runs the risk of
not being able to roll over the public debt
• Public debt denominated in foreign currency is risky if the
local currency depreciates in value
• So countries with marketable assets and with longer-
maturity debt that is issued in their own currency are less
at risk from having an unsustainable public debt
TYPES OF GOVERNMENT DEBT
Maturity structure:
• Treasury bills pay a nominal interest rate i and is short-run government debt
• Long-run government debt such as a consol promises to pay £1 forever, so
the price of this government paper is £1/i
• Return on long-run debt is the long-run interest rate plus expected capital
gains; according to efficiency theory this should equal short-run interest rate
• If the long interest rate exceeds the short interest rate (upward sloping term
structure), the price of long bonds is expected to fall over time
• If it is below the short rate (downward sloping term structure), the price of
long bond is expected to rise of time
• If short and long bonds are perfect substitutes, we can just aggregate them
Indexation of debt:
• Bond may be indexed to the rate of inflation (indexed bonds), so the cost of
higher inflation is borne by the government – disincentive to create inflation
THE “BALANCE SHEET” APPROACH
} Traditional models of crises focus on flow variables (CA, fiscal balance)
} New understanding: a crisis happens if there is a plunge in the (domestic
and foreign) demand for domestic-currency- denominated financial
assets
} Scale of capital outflows may be huge in comparison with observed
flow fundamentals before crisis
} Crisis may originate in official sector, but also in liabilities between
residents
} Problems in one sector may spill over to other sectors triggering a
balance-of-payments crisis
} Leading indicator: financial fragility between sectors, as measured by the
composition of their balance sheets 24
FOUR RISKS (BALANCE SHEET MISMATCHES)
} Maturity mismatches: between long-term assets and short-term
liabilities
} Typical example: banking sector, which provides maturity transformation
} May arise in domestic as well as foreign currency
} Can lead to a crisis if market refuses to roll over short term debt
} Currency mismatches: assets and liabilities denominated in different
currencies
} Change in exchange rate leads to capital loss
} Typical case: currency substitution, liability dollarization in emerging
markets
} Even if one sector (banks) hedges this risk (by lending in dollars) that
just transfers the risk to other sectors in the economy (corporate) 25
TWO MORE
} Capital structure mismatches: excessive reliance on debt
over equity
} Debt is not contingent Þ greater exposure to revenue
shocks
} Sometimes provoked by regulatory or tax treatment
unfavourable to FDI (Korea, Thailand, Russia)
} Undercapitalised banks and other financial institutions
} Solvency problems: assets < liabilities
} Maybe because of excessive leverage or investment in low-
yield assets
26
FINANCIAL INTERLINKAGES BETWEEN SECTORS

27
CONTINUED
• Sometimes government tries to alleviate problems by directly taking on
liabilities from private sectors, or by providing guarantees against capital
losses
– Fiscal bailouts of banks, currency swaps by central bank
– However, this does not eliminate the risks, just transfers them to the
official sector (1st generation)
• Crises triggered by balance sheet problems lead to severe economic
downturns
– ↑D for foreign assets forces the economy to generate more resources
to purchase them (CA reversal), usually via sharp contraction in
imports
– Capital losses Þ negative wealth effect + credit crunch Þ AD
contraction (consumption and investment)
– Result: “output overshooting” – greater output drops than
ultimately allowed for by improvement in competitiveness through
devaluation
– But also fast recoveries – so “V-shape” recoveries 28
“V-SHAPE” RECOVERIES
15

10

5
Hong Kong
% GDP growth (per capita)

Indonesia

0 South Korea
t-3 t-2 t-1 t t+1 t+2 t+3 Malaysia
Thailand
Brasil
-5
Russia
Argentina
Uruguay
-10

-15

-20
Source: World Bank, World Development Indicators 29
BACKGROUND SLIDES ON
UK GOVERNMENT DEBT 1694-2016
UK GOVERNMENT DEBT 1694-2016
• See Martin Ellison, Oxford & Andrew Scott, LBS (2020, AEJ: Macro) and
Martin’s webpage: http://users.ox.ac.uk/~exet2581/data/data.html
• Most of the time market value of debt and its par value (the value to be
paid at redemption) track each other closely, but in 2016 the market value
of debt was 35% higher than the par value (last time this occurred was 300
years ago) – bull market for gilts means debt is not so cheap as it looks
• UK government has issued bonds across the whole spectrum of maturities;
since WW2 the UK has shifted to short-run bonds (< 7 years) but since
1990 the UK has shifted towards bonds with maturity > 15 years
• Over 3 centuries average maturity has declined continuously but recently
maturity spectrum has been filled out
UK DEBT: ELLISON AND SCOTT (2020) CONTD..
• Market value of public debt changes due to (i) cost of funding, (ii) inflation, (iii)
real GDP growth, and (iv) borrowing, i.e. fiscal deficits – see next slide
• High real return of 2.5% (i.e. 4.4% minus 1.9% inflation) per year over 3 centuries
• Growth corrected real return was 0.7%, i.e. 4.4% minus 1.8% real GDP growth
• Over 17th, 18th and 19th century debt has been sustainable due to primary
surpluses, but during 20th century saw a switch to running primary deficits
• Inflation and weak returns in 20th century helped to keep UK debt sustainable;
compared to US, the UK financed WW2 debt via inflation and devaluations
• Since financial crisis, UK debt has been pushed up in equal amounts by running
deficits as well as by rise in the price of bonds, especially long bonds
• Long bonds helped contain debt during war but pushed up value of debt in
aftermath of financial crises
DECOMPOSITION OF DEBT DYNAMICS

39

You might also like