BMSS 2007 Case study of debt sustainability - Spain

Document Sample
BMSS 2007 Case study of debt sustainability - Spain Powered By Docstoc
					           BMSS 2007
Case study of debt sustainability –
       Spain under Philip II


          Hans-Joachim Voth
                                               Key points

Philip II’s defaults universally seen as major disasters for
bankers, arising from irrational overlending and overborrowing.
We argue that large fluctuations in debt capacity and liquidity
made defaults an equilibrium outcome.
These variations were largely anticipated. Defaults were
always considered a possibility, even in the written lending
contracts.
We show that lending was largely sustainable. Defaults were
the results of temporary liquidity crises, and were quickly
followed by fresh lending.
                                               Key points

Philip II’s defaults universally seen as major disasters for
bankers, arising from irrational overlending and overborrowing.
We argue that large fluctuations in debt capacity and liquidity
made defaults an equilibrium outcome.
These variations were largely anticipated. Defaults were
always considered a possibility, even in the written lending
contracts.
We show that lending was largely sustainable. Defaults were
the results of temporary liquidity crises, and were quickly
followed by fresh lending.
                                      Background

Under Philip II (1556-98) Spain rules over the largest
colonial empire at the time. Exploits the vast silver
mines of Potosí.
Almost continuously at war (France, Dutch
Provinces, Ottoman Empire, England).
Borrows from a range of international bankers and
domestic bondholders to the tune of 60% of GDP.
Defaults 4 times.
                                         Outline

Data
Contract Structures and Conditionality
High Variance of Revenue and Expenditure Shocks
Sustainability
Conclusion
   Constructing a yearly revenue series

Data from separate revenue streams in Ulloa (1977).
Only contains revenues for which receipts exist (as
opposed to contracted revenues).
  This improves accuracy but increases missing data.

We use a lower-bound estimate to replace missing
data.
We use knowledge of the periodicity of revenue to
refine estimates.
                          Spain under Philip II
Ruled over one of the largest empires in history.
Incurred the largest level of debt in the Early
Modern world, rivaling contemporary standards.
Defaulted repeatedly on its debt, and yet managed
to maintain access to capital markets.
Philip’s reign is often referred to as the origin of
both modern sovereign lending instruments, as well
as that of serial defaults.
                                   Big questions

How could Early Modern monarchs credibly commit
to repay their debts?
What was the role of defaults?
Why did lending resume so quickly after the
bankruptcies?
Were bankers lending rationally? Was the fiscal
position of the Crown sustainable?
Spain’s fiscal position: the traditional view

 High costs of early modern warfare, and the
 constant wars of Philip II produce a perpetual gap
 between expenditures and revenues in Castile.
 Bankers were nonetheless lured to lend to the king
 by high interest rates and the illusion of security
 provided by silver revenues from the Indies.
 When the king inevitably defaulted bankers were
 ruined in successive waves, only to be replaced by a
 new generation that fell into the same trap as the
 previous one.
                                      Sound bites

“Expenditure continued to rise above all reasonable
limits while its income was declining, with a visible
drop in fiscal revenues.” (Braudel 1966).
“If by fiscal crisis we simply mean an unbridgeable
gap between unavoidable expenditure and
disposable income, then the Spanish Monarchy was
in an almost continuous fiscal crisis.” (Thompson
1994).
             Evaluating Spain’s fiscal position
We compile a comprehensive dataset on Spain’s fiscal position for the
second half of the sixteenth century, including new archival data.
We evaluate sustainability by:
    The traditional definition (stability of debt / GDP ratio).
    Forward-looking simulations analogous to Value-at-Risk.
    Policy rules approach (Bohn 1998).

We conclude that Spain’s fiscal position was sustainable throughout
Philip’s reign.
We further find that Philip conducted a “virtuous” fiscal policy,
vigorously increasing his primary surplus in response to debt
increases.
We argue that “defaults” are simple reschedulings, driven by liquidity
shortfalls and the king’s bargaining position with respect to the cities.
                              Historical background

Philip II (r. 1556 - 1598) ruled over the Spanish Empire at the
peak of its extension and influence.
Drove Spain into major military enterprises with little financial
gain:
   The War of the Holy League / Battle of Lepanto (1570-73).
   The Eighty Years War / Dutch Revolt (1566 - 1648).
   The Invincible Armada (1588).
   There was only one year of complete peace in the entire reign.
Ran up debt on the order of 50% of GDP.
Defaulted three times: 1557-60, 1577, 1596.
                                                                                                                        Revenue
 16



 14



 12



 10



  8



  6



  4



  2



  0
  55


         57


                59


                       61


                              63


                                     65


                                            67


                                                   69


                                                          71


                                                                 73


                                                                        75


                                                                                77


                                                                                       79


                                                                                              81


                                                                                                     83


                                                                                                            85


                                                                                                                   87


                                                                                                                          89


                                                                                                                                 91


                                                                                                                                        93


                                                                                                                                               95
15


       15


              15


                     15


                            15


                                   15


                                          15


                                                 15


                                                        15


                                                               15


                                                                      15


                                                                              15


                                                                                     15


                                                                                            15


                                                                                                   15


                                                                                                          15


                                                                                                                 15


                                                                                                                        15


                                                                                                                               15


                                                                                                                                      15


                                                                                                                                             15
                                                                             Year

 Sales tax      Customs (internal and external)          Monopolies          Direct taxes    Millones      Church revenues        Confiscations
                                                                                                             Revenue
 16



 14



 12



 10



     8



     6



     4



     2



     0
 55

          57

                59

                       61

                             63

                                    65

                                          67

                                                69

                                                      71

                                                             73

                                                                   75

                                                                          77

                                                                                79

                                                                                      81

                                                                                            83

                                                                                                  85

                                                                                                        87

                                                                                                              89

                                                                                                                    91

                                                                                                                          93

                                                                                                                                95
15

         15

               15

                     15

                            15

                                  15

                                         15

                                               15

                                                     15

                                                           15

                                                                  15

                                                                        15

                                                                               15

                                                                                     15

                                                                                           15

                                                                                                 15

                                                                                                       15

                                                                                                             15

                                                                                                                   15

                                                                                                                         15

                                                                                                                               15
                                                                        Year

                    Sales tax                             Customs (internal and external) Monopolies
                    Direct taxes                          Millones                        Church revenues
                    Confiscations                         Indies
                                             Revenue

Most revenue streams were very stable as a result of
tax farming and contracting with the cities.
Almost all the volatility in the series is the result of
silver revenue from the Indies.
Two large increases negotiated between king and
Cortes, both in sales and excise taxes.
   1576, increasing the alcabalas.
   1591, introducing the millones.
                        Castilian Debt Instruments

Asientos
   Short-term loans contracted with bankers.
   Often involved international transfers.
   They were quite expensive. Inflated exchange rates, lifetime pensions,
   and other concessions were used to circumvent usury laws.

Juros
   Perpetual or lifetime bonds.
   Their service was guaranteed by a specific tax stream.
   Often used as collateral for asientos.

Asientos and juros gradually acquired many features of modern
debt instruments.
                     A new series of asientos

Historical scholarship has so far focused on the
asientos of Charles V (r. 1516 - 1556).
We compile a new series of asientos for Philip II’s
reign from the holdings at the Archive of Simancas.
So far we have recorded the date, identity of the
banker(s), and amount of the loan.
We are expanding the database with the terms,
conditions and cost calculations for each of 450+
asientos underwritten during Philip’s reign.
                                          Asiento with Juan Curiel de la Torre
                                                           December 15, 1567
Lends 200,000 flemish ecús of principal, to be delivered in Anvers on May 1, 1568. The principal
has a gold content equivalent to 192,080 ducats.
The capital is repaid in two equal payments of 100,000 ducats in Castile. This conversion implies
an exchange premium of 4.12%.

     The first payment in October 1567 (before delivery).
     The second payment “at the fair of May 1568” (to be held on or after May 1570).
The second 100,000 ducats bear simple 9% interest yearly. This interest is partially paid from a
lifetime juro which yields 3,000 ducats annually (21,000 ducats of face value). This is a
guarantee on the payment of 1/3 of the interest.
The banker is allowed to export a total of 200,000 ducats from Castile.
The loan is collateralized with a juro yielding 7,000 ducats per year. The banker gets to choose
the tax stream and the yield of the perpetuity to a maximum of 7.14% (98,000 ducats face value
- almost full collateralization).
The banker cannot collect interest on the collateral juro while the loan is in good standing. He
can collect the interest or sell it if the loan goes into default.
The banker is allowed to sell the collateral while the loan is in good standing; he must return a
juro of equal value and quality if the loan is paid in full.
                                         Cost of the asiento
The Crown makes an up-front payment of 100,000 ducats.
It receives 192,080 ducats six months later in Flanders.
Assuming the banker enjoys 7.14% interest on the 100,000 ducats for the first
six months, the net loan is worth 85,456 ducats.
3,000 ducats are paid on January first of each year towards interest.
Assuming the loan and additional interest are repaid in May 1570, the cost of
financing to the Crown is 29.44%.
The banker is further allowed to:
    Trade with the collateral (the costs of returning a lower-quality juro
    would be borne by other bondholders).
    Export 4.12% more bullion than needed (profits foregone by the Crown).
                             Short-term borrowing
10




 8




 6




 4




 2




 0
 1520   1530   1540   1550   1560   1570   1580   1590   1600
                             Year
                                                       Juros

Defined by:
   Term (lifetime or perpetual).
   Yield.
   Revenue stream providing the interest payments.
Issued in the name of a specific person.
Could be traded with the consent of the king, who was all too
willing to issue licenses in exchange for fees, loans and other
concessions.
In the second half of the sixteenth century, they gradually
became on of the main investment vehicles of the rentier
classes.
                          Service of long-term debt
100.00%                                                                                 5.0


                                                                                        4.5
90.00%

                                                                                        4.0
80.00%
                                                                                        3.5

70.00%




                                                                                              Millions of ducats
                                                                                        3.0


60.00%                                                                                  2.5


                                                                                        2.0
50.00%

                                                                                        1.5
40.00%
                                                                                        1.0

30.00%
                                                                                        0.5


20.00%                                                                                  0.0
       55

       57

       59

       61

               63

                       65

                       67

                       69

                       71

                       73

                       75

                       77

                       79

                       81

                       83

                       85

                       87

                                                                           89

                                                                                   91

                                                                                   93

                                                                                   95
    15

    15

    15

    15

            15

                    15

                    15

                    15

                    15

                    15

                    15

                    15

                    15

                    15

                    15

                    15

                    15

                                                                        15

                                                                                15

                                                                                15

                                                                                15
                                               Year

                    Debt service as % of revenue      Yearly debt service
                         The 1557-60 default

Philip suspended payments on asientos underwritten
by the Fugger and Welser families.
A tentative settlement brokered by Genoese
bankers resulted in a second suspension in 1560.
Eventually the Fuggers were granted control of
Crown revenues and monopolies.
Lending did not resume in full until 1566, with
Genoese bankers providing the bulk of the funds.
                              The 1575 bankruptcy

Triggered by the inability of the king to obtain higher revenues
from the Cortes.
Suspension of payments on all asientos and collateral juros.
The bankers formed a consortium representing around 70% of
outstanding debt to negotiate with the king.
The deteriorating military situation in the Netherlands
convinced the Cortes to grant a large increase in sales taxes.
A settlement - medio general - was reached with most bankers
in 1577. Lending resumed immediately.
                         The 1577 medio general

Rescheduled 14.6 million ducats of debt (two years’ worth of
revenue).
Debt collateralized by juros suffered a 30% write-off.
Uncollateralized debt suffered a 58% write-off.
On average, the king agreed to repay 62 cents on the dollar.
The consortium of bankers agreed to provide a common loan
for 5 million ducats over the following year.
Some bankers managed to obtain better deals on the side.
                                     “Snapshot” data



       Revenue        Revenue
Year                                  Juros      Juros Service
       (Artola)    (our estimate)

1560   4,192,237     3,102,002      19,000,000     1,468,000

1565   5,600,000     4,196,908      25,000,000     1,861,000

1577   8,700,000     9,142,759      36,000,000     2,730,000

1598   9,731,408        n/a         68,000,000     4,634,000
                    Traditional sustainability

 The debt is said to be sustainable if the primary
 surplus is large enough to keep the debt to GDP
 ratio constant given the interest rate and the
 growth rate of the economy.

  *    (r − g )
ps = −          d
       (1 + g )
 We perform the analysis using Artola’s data, which
 implies lower growth rates, biasing the results
 against sustainability.
                        Calculating sustainability
The sustainability equation requires data for GDP growth and
interest rates.
We substitute GDP growth with revenue growth:
   We do not have (good, high-frequency) GDP estimates.
   Even if we had them, it is not clear what they would mean in a
   pre-modern context.
   The sustainability equation assumes a stable fiscal system. Castile
   was steadily increasing taxation and experiencing natural
   resource windfalls, which caused revenues to grow faster than
   economic activity.
   What really matters for debt sustainability are the revenues the
   state is able to draw on to service the debt.
                Estimating the interest rate

Use the debt-difference equation:

                    (rt − gt )
dt − dt−1 = − pst +            dt−1
                    (1+ gt )

This requires calculating the primary surplus.
            Calculating the primary surplus

The primary surplus is defined as revenue minus non-interest
expenditure.
We do not have expenditure data, but we have debt service
data.
We calculate the annual fiscal deficit for each period as the
annualized increase in the debt stock.
We subtract the yearly debt service from the fiscal deficit to
obtain the annual primary surplus.
As a by-product, we obtain estimates of non-interest
expenditures.
                               Primary surplus, debt growth
                                        and revenue growth


            Average Yearly Fiscal      Average Yearly   Average Yearly Debt
 Period
              Deficit (∆ debt)
                       ∆                  Revenue             Service

1560-1565         -1,200,000             4,896,119           1,664,500
1565-1577          -916,667              7,150,000           2,295,500
1577-1598         -1,523,810             9,215,704           3,682,000
1560-1598         -1,289,474             7,995,010           2,978,697




               Average Yearly          Average Yearly
                                                         Annualized Debt      Annualized Revenue
 Period        Primary Surplus          Expenditures
                                                          Growth Rate            Growth Rate
            (Debt service - deficit)   (revenue - ps)

1560-1565           464,500              4,431,619            5.64%                 9.21%
1565-1577          1,378,833             5,771,167            3.09%                 3.74%
1577-1598          2,158,190             7,057,514            3.07%                 0.53%
1560-1598          1,689,224             6,305,786            3.41%                 2.24%
                                   Traditional debt sustainability




            Debt as % of       Annualized                                         Primary surplus
             Revenue at                                                                             Actual Primary
                            change in debt as Revenue Growth   Implied interest      needed to
 Period     beginning of                                                                            Surplus as % of
                              % of Revenue       Rate (g)         rate (r)         stabilize debt
                                                                                                     Revenue (ps)
            period (dt-1)        (∆dt)
                                   ∆                                                   (ps*)

1560-1565     453.22%           -1.36%           9.21%             11.17%             8.13%             9.49%


1565-1577     446.43%           -2.72%           3.74%              7.59%             16.56%            19.28%


1577-1598     413.79%           13.57%           0.53%              9.52%             36.99%            23.42%


1560-1598     453.22%            6.46%           2.24%              8.46%             27.59%            21.13%
                Forward-looking sustainability

Traditional debt sustainability implies an ex-post calculation between
end-points.
Debt sustainability calculated like this in modern-day data often
offers too sanguine an assessment
Rationality only requires that bankers had an ex-ante expectation of
sustainability.
We use an analogy with Value-at-Risk calculations to simulate
potential revenue paths given the data available in 1577, and
estimate what fraction of them would have been sustainable.
We use our yearly series. Their higher growth rates imply a higher
volatility than that of Artola’s data. In this framework, higher
volatility biases the results against finding sustainability.
                                           Speed limits

Iterating the debt difference equation forward implies that the
debt must be (less than or) equal to the present value of
future primary surpluses.
      ∞
dt = ∑δ i pst +i
     i=1


We know the average primary surpluses that Philip II had at his
disposal; we use them to calculate a “speed limit” - how much
debt could he have accumulated without violating the
sustainability condition.
               Simulation of revenue paths

We calibrate the simulation to the 1560-1577
period:
  Average revenue growth: 4.9%
  Standard deviation of revenue: 12.4%

We then simulate revenue paths for the 1578-1598
period.
In 72% of cases, the debt / revenue ratio stays
below the 453% speed limit calculated for 1565-77.
 Simulation of revenue paths




actual
                                Dynamic simulation

The previous calculation assumes the speed limit stays
constant for the full period.
Better assumption: agents updated the speed limit as fiscal
news poured in.
Results change very little - we find that in 71.5% of the
simulated years the debt / revenue ratio would have been
sustainable.
This underestimates true sustainability, as we do not rule out
clearly unsustainable paths in which lending would have been
curtailed before the end of the period.
               Dynamically simulated paths
                spee dl       debt_rev

 5.5061




2.8965 8
           1                             22
                          n
                       An unlikely path
              speedl       debt_rev

13.0033




 2.3023
          1                           22
                       n
       Sustainability as a function of the
                          primary surplus


                                          Percentage of years with a
    Historical Period   Primary Surplus   debt to revenue ratio above
                                                 sustainability
Hypothetical                 10.0                    71.0
Average 1560-77              16.4                    38.0
Average 1566-77              19.3                    28.5
Average 1577-98              23.4                    28.8
Hypothetical                 26.0                    15.0
Hypothetical                 30.0                    9.0
Hypothetical                 35.0                    6.0
                       The policy rules approach

Traditional sustainability analysis asks if there is a unit root in
the debt/GDP ratio. If the answer is yes, we conclude that the
path is not sustainable
Bohn (1998) shows that the time-series properties make this
procedure problematic
   Unit root tests have low power and are usually inconclusive on
   series of debt and primary surplus.
   When the interest rate is below the growth rate persistent
   primary deficits do not necessary violate the government’s
   intertemporal budget constraint.
Key idea by Bohn: Sustainability requires that negative fiscal
news be matched by increases in the primary surplus.
  Estimating the yearly primary surplus

We only have four observations for the total stock of debt (1560,
1565, 1577, 1598).
We do have yearly data for short-term borrowing through asientos.
Asientos were either repaid or converted into juros.
We estimate the repayment to conversion ratio in each of the three
periods.
We assume the ratio was constant throughout these sub-periods to
estimate an yearly series of outstanding debt.
We combine this series with our yearly revenue figures to estimate
the yearly primary surplus.
The more mistakes we make in estimating the yearly debt series, the
smaller the fiscal response coefficient will be.
                                                          1568 = 100




                                         50



                                     0
                                              100
                                                    150
                                                              200
                                                                       250
                                                                             300
                                                                                   350
                                                                                         400
                              15
                                68
                              15
                                69
                              15
                                70
                              15
                                71
                              15
                                72
                              15
                                73
                              15
                                74
                              15
                                75
                              15
                                76
                              15
                                77
                              15
                                78
                              15




Debt
                                79
                              15
                                80
                              15
                                81
                              15
                                82
                              15
                                83
                              15
                                84
                              15
                                85
                              15
                                86
                              15
                                87
                              15
                                88
                              15
                                89
                              15
                                90
Primary Surplus (3-year MA)
                              15
                                91
                              15
                                92
                              15
                                93
                              15
                                94
                              15
                                95
                              15
                                96
                              15
                                97
                                                                                               Debt and primary surplus
                  Inconclusive unit root tests



       Variable   Augmented Dickey-Fuller      KPSS

primary surplus          -0.3 (-2.4)        0.457 (0.46)

debt                     0.98 (-2.4)        0.48 (0.46)

ps / revenue            -0.95 (-2.8)        0.27 (0.46)

debt / revenue          -0.66 (-2.5)        0.18 (0.46)
                                      Estimation
Basic relationship:
st = ρ • dt + ε t
We estimate linear, log and quadratic relationships.
We also estimate the cointegrating vector in a VEC
model.
Bohn’s specification includes variables that capture
the cyclical components of revenue and
expenditure. He shows that omitting them (as we
do) introduces attenuation bias.
                                      Estimation results
                                                    2       2
 Model of dt     Constant     dt     (dt -da )^2   R /Chi       N

Linear            -0.66     0.076                   0.39        31
                  (0.76)    (3.58)
                   [0.2]     [3.3]

Log               -5.77     1.744                   0.34        31
                   (3.2)     (3.7)
                  [2.99]    [3.47]

Quadratic         -0.84     0.084     -0.001        0.41        31
                  (1.1)      (5.1)     (0.7)
                  [0.5]     [4.99]    [1.94]

Log, scaled by     -1.6     0.17                    0.001       31
revenue           (1.11)    (0.2)
                   [1.1]    [0.2]

Cointegrating     -1.12     0.147                   175.1       31
Vector from                 (13.2)
VEC
                   Debt vs. primary surplus
8


                                                                                    1597


6

                                                                  1592              1596


                                      1584
4
                                     1585
                          1577 1580 1583
                                      1586 1588                     1593
                          1578 1581
                                                              1591         1594
                          1576
                                                           1590
                                                    1589
2        1568
                                     1582
                                             1587
           1570               1579                                                1595
            1571
          1569 1573
                       1574
                1572
         1567             1575
0
    20                           40                                      60                80
                                                    debt
Response of the primary surplus to debt

All specifications show a positive, significant and
economically important response of primary surplus
to increases in debt.
The log specification implies that primary surplus
grew by 174% with each doubling of debt.
The linear and quadratic coefficients, 0.076 and
0.084, are higher than those Bohn found for the US
between 1916 and 1995 (0.028 to 0.054).
                                                   Conclusions

Philip II’s finances were probably sustainable throughout most
of his reign.
   The Crown ran a constant primary surplus.
   It was very successful at increasing revenues throughout the
   period.
From a fiscal sustainability point of view, the behavior of
bankers was rational.
   Traditional sustainability calculations support sustainability until
   1577.
   A forward-looking analysis shows that in 1577 at least 80% of
   potential debt/revenue paths would have been sustainable.
   The policy rules approach implies that Spanish fiscal policy under
   Philip was at least as “virtuous” as that of the US in 1916-1995.
                                              Further research

Commitment
    How could Philip credibly commit to repay the large loans bankers were
    extending him?
Banker profitability
    Did each default really ruin scores of bankers and replace them with a
    fresh crop?
Defaults as equilibrium outcomes
    If not, can we explain Philip’s defaults as equilibrium outcomes in a
    context of market incompleteness?
Default as a negotiation device
    Philip II uses defaults to push through higher taxes
    Bankers effectively asked to live with short-term turmoil and partial
    write-offs in exchange for a more sustainable long-term revenue basis
       The Sustainable Debts of Philip II
Revenues, Expenditures and Primary Surplus in Habsburg Spain, 1560-1598




                                               Mauricio Drelichman
                                     University of British Columbia and CIAR

                                                 Hans-Joachim Voth
                                 ICREA / Universitat Pompeu Fabra and CEPR




                                                           UBC DIET lunch
                                                         January 31, 2007
                                         Sustainable debt / revenue ratios
                                 1600%

                                                             g=9%                                 g=7%
                                 1400%
                                                                                                                       g=5%

                                 1200%
sustainable debt/revenue ratio




                                 1000%
                                                                                                                                  Actual debt/
                                              1560-66                                                                             revenue ratios
                                 800%
                                                                                                                                        1598
                                                                                                                           g=3%
                                                                                                         1566-75
                                 600%


                                                                                                                                        1577
                                 400%                                                                                      g=1%
                                                                                                                                        1560
                                                                                                                           g=0%


                                 200%
                                                                1577-98


                                   0%
                                         11             10                9                   8                    7   6
                                                                              interest rate