Possible Adaptation of Debt Service in Relation to Commodity Price Volatility

1
Possible Adaptation of Debt Service in Relation
to Commodity Price Volatility

• Christopher L. Gilbert
• Alexandra Tabova
• UNECA , Dakar, Senegal
• 17-18 November 2003

2
Part I

• Structure

3
Objectives

• Concessional debt service typically shows
  relatively little cyclical variation over time.
• Export earnings are highly variable and, because
  commodity prices are at least to some extent mean
  reverting, this variation is in part cyclical.
• The objective is to more closely match the debt
  service obligations facing indebted governments
  with their ability to meet these obligations.
• Because schemes to smooth export earnings have
  not, in general, proved either successful or
  financially viable, this involves unsmoothing
  concessional debt service.

4
Overall Structure

• We look at schemes which modify governments debt
  service on concessional debt in relation to
  changes in their export earnings, or to the world
  prices which underlie these earnings a measure
  of their ability to pay.
• We attempt to do this in a way which is broadly
  neutral for the lender ie by acceleration and
  retardation of debt service payments.
• We also look for a simple scheme which is easily
  implemented without generating moral hazard
  problems.

5
Revenues or Prices?

• A link to export revenues would more closely
  align the scheme with debtor countries ability
  to pay. A price link will fail to insure
  borrowers against quantity shocks, and will give
  rise to basis risk if the countrys export
  price moves differently from the world price.
• Export revenues are only known with a lag. Basing
  the scheme on prices would increase timeliness.
• Basing the scheme on export revenues could
  generate poor incentives and moral hazard
  problems.
• A price link increases potential market
  offsetting possibilities.

6
Conditionality

• Conditionality becomes important in the context
  of moral hazard problems.
• With conditionality, the process for judging
  whether a particular government qualifies becomes
  time consuming - benefits are therefore less
  timely.
• This was a serious practical problem with the EU
  Stabex scheme where delays in making payments
  contributed to these payments becoming pro-
  cyclical.
• An advantage of price over revenue-based schemes
  is that they do not give rise to moral hazard.

7
Imported Commodities

• Developing countries which lack domestic oil
  resources may find that oil imports pre-empt a
  significant proportion of their export revenues.
  Other developing countries import much of their
  grain requirements. Both groups of countries are
  vulnerable to upward movements in these import
  prices.
• This suggests it may be sensible to consider
  extension of any scheme to cover important
  imported commodities.

8
Instrument Structure

• The scheme we investigate has the structure of an
  overlay the original (fixed) debt service
  payments are overlain by a set of floating for
  fixed swaps, where the floating rate is geared to
  the world price of the export commodity.
• The swaps can be structured to have an initial
  value of zero i.e. the expected discounted
  value of the fixed and floating payments are
  equal.
• The floating payment would be structured in terms
  of deviations from a (moving average) trend the
  scheme compensates for short term fluctuations in
  ability to pay, but not for any long term trend
  deterioration.
• An easy modification is to overlay using
  swaptions rather than swaps implying the
  floating payments only become effective for large
  departures (say ? 15) from trend.

9
Advantages and Disadvantages

• The scheme is simple.
• It is a straightforward extension of lending in
  currency baskets (already possible for IBRD
  loans).
• It will be timely prices are known instantly
  and averages immediately after the year end.
• Use of world prices circumvents moral hazard
  problems.
• But it is an empirical matter as to how
  effective it will be in matching debt service
  payments to export revenues
• and whether linking to concessional debt
  service is the appropriate delivery modality.

10
Part II

• The Hedge Ratio

11
In Relation to ?

• We might set hedge positions either in relation
  to
• Concessional debt service this corresponds to
  the overlay structure considered previously.
• Exports variation of concessional debt service
  becomes an instrument for offsetting variations
  in export earnings. We investigate this subject
  to the constraints that debt service cannot be
  negative and cannot exceed twice its scheduled
  level.
• Unless the ratio of debt service to exports is
  high, the former approach is likely to lead to a
  much lower degree of offsetting.

12
The Minimization Objective

• We might consider minimization variations in
  either the debt service to export ratio S/X or of
  the amount of foreign exchange available after
  servicing debt, X S.
• The second objective requires higher
  interventions debt service must vary in
  relation to the dollar value of export
  fluctuations, and not just proportionately to
  them.
• If expenditure O on imported commodities are
  included, the free foreign exchange criterion
  relates to X S O. We modify the debt service
  to export ratio to (S ?O)/X where ? is the
  average ratio of O to X.
• Although the debt literature primarily focuses on
  debt service-export ratios, we prefer the free
  foreign exchange concept.

13
The Overlay Hedge Parameters

• The simplest procedure is to make hedge weights
  proportional to export and import shares e.g.
  if coffee forms 50 of exports, there would be a
  50 coffee swap overlay.
• This supposes a unit hedge ratio, which is only
  optimal in the most simple cases. We can do
  better by taking into account the covariance
  structure of prices and revenues.
• We obtain variance minimizing hedge parameters by
  regressing deviations of XSO from its moving
  average trend on deviations of commodity export
  prices P and oil import prices O from their
  moving average trends.
• The resulting parameters would have minimized the
  Var(XSO) ex post.

14
Welfare Analysis

• We need a criterion against which we can judge
  the success of any scheme.
• It is useful to use a money metric i.e. what
  is the dollar value of any scheme to the clients.
  However, these valuations will always depend on
  an assumed value for risk aversion.
• We use a variation on the money metric procedure.
  We ask, What would be the equivalent percentage
  reduction in debt service which would leave
  borrowers indifferent between current
  arrangements and the proposed scheme.

15
Part III

• Country Simulations

16
Our Sample

• We simulate debt service payments for ten
  moderately or severely indebted
  commodity-dependent African countries over
  1984-2001.
• Selection based on (a) high proportion of
  concessional in total debt, (b) at least one
  export commodity making up 10 of total exports
  or one import commodity accounting for 10 of
  total imports and (c) adequate trade statistics.
• The countries are Benin, Burkina Faso, Burundi,
  Cameroon, Ghana, Kenya, Madagascar, Malawi,
  Rwanda and Tanzania.

17
Schemes

• We simulate two schemes.
• A commodity swap scheme with price deviations
  measured relative to a four year moving average
  of past prices lagged one year.
• A swaption scheme and relating to the same moving
  average with a ?15 band.
• We simulate both schemes on two bases
• debt service basis
• export revenue basis

18
There is significant variance reduction but the
value of these is quite small equivalent to
debt reductions of less than 3. These median
statistics disguise substantial variation over
countries.
19
These results show that, on average, there is a
deterioration in performance by moving away from
the swap-based scheme (A) to the swaption (B).
20
Benefits by Country
Burkina Faso, Burundi, Malawi and Rwanda are seen
as obtaining greatest benefit from the proposed
scheme.1 Estimated benefits are very low in
Benin, Ghana and Kenya.
21
Impact on IDA Flows
If the scheme is based on scheduled payments,
there is little net impact if it is based on
export revenues, the net impact is large.
22
Overall Effectiveness

• The best performing scheme is the swaps scheme
  based on trend export earnings and using
  regression-based weights.
• This gave benefits which we evaluate
  conservatively as equivalent to concessional debt
  service reductions of between zero (Benin) and
  5¾ (Rwanda). Median benefit is around 2¼ of
  concessional debt service.
• This evaluation is based on a conservative unit
  estimate for risk aversion trebling to give a
  median 7 benefit is an effective upper bound.
• Simulation of IDA flows indicates only modest
  offsetting across commodities.
• Tying the scheme to a debt service instrument
  implies low disruption of payments to IDA while
  hedging against trend exports results in IDA
  assuming substantial a commodity price risk which
  may disrupt net repayment flows.

23
Why are the Outcomes only Modest?

• Concessional debt repayments are an inadequate
  lever to stabilize export revenue fluctuations.
• Basis risk the prices countries obtain for their
  exports are only moderately correlated with world
  prices.
• Quantity risk export revenues (and oil import
  costs) vary as much through country-specific
  variations in quantities exported (or imported)
  as through prices.
• All three explanations contribute to the modest
  performance, but we regard quantity variation as
  the most important.

24
Basis Risk
The long term hedging basis appears good for
coffee and cocoa but poor for cotton and tobacco.
The correlations for tea are mixed.
25
Quantity Risk
Quantities are more variable than prices for most
country-commodity pairs. This limits the
insurance offered by any price-based insurance
scheme.
26
How Important are Prices?
The table gives the proportion of the variance of
deviations of log commodity revenues from
centered five year moving averages which cannot
be explained by deviations in international
prices from centered five year moving averages.
We interpret this measure as measuring the
importance of quantity and basis variations.
27
Part IV

• Country Studies

28
Burundi

• This is the country for which the scheme works
  best scheduled debt service scheme worth 3.6
  and export revenue based scheme worth 4.7 debt
  reduction.
• Major export is arabica coffee (79 of total
  revenues).
• Oil import costs averaged 31 of total export
  revenues.
• Burundis hedging basis relative to the New York
  coffee is good. However, movements in that price
  only explain around 40 of movements in coffee
  revenue.
• Oil import expenditures were weakly negatively
  correlated with the world oil price (r -0.17)
  hedges ineffective.
• The scheme fails to cope with two major coffee
  export revenue movements (1993, 1996) unrelated
  to changes in world prices.

29
Coffee export revenues fell sharply in 1993 and
1996
Coffee prices rose slightly in 1993 and fell only
modestly in 1996.
30
Ghana

• The scheme delivers relatively little for Ghana
  scheduled debt service scheme worth 0.1 and
  export revenue based scheme worth 0.8 debt
  reduction.
• Major exports are gold (40) and cocoa (25). Oil
  imports are 24 if total exports but data are
  incomplete.
• The hedging basis is good for both gold and oil
  but cocoa revenues are hedged and therefore
  correlate poorly with the world cocoa price. Oil
  import costs are only weakly correlated with the
  oil price.
• Overall, the scheme appears unnecessary.

31
Because of hedging, cocoa revenues are smoother
than the cocoa price.
Gold export revenues expanded rapidly during 90s
despite a flat-to-falling price.
32
Benin

• Benin appears to give the poorest outcomes in our
  sample, with an overall negative valuation
  scheduled debt service scheme worth -0.2 and
  export revenue based scheme worth 2.6 debt
  reduction.
• Major export is cotton 35 of its total
  revenues. The cost of oil imports averaged 16 of
  total export revenues.
• The cotton export price is poorly correlated with
  the world price and movements in the world price
  explain 20 of Benins cotton revenues. The oil
  correlation is comparable.
• Movements in X-S-O are dominated by the low
  figure in 1991 and the high spike in 1996
  unrelated to either cotton or oil. It may be that
  national accounts data are too poor to allow
  adequate analysis.

33
X-S-O appears to have been low in 1991 and high
in 1995
These movements appear unrelated to cotton
exports or oil imports.
34
Lessons from the Country Studies

• Movements in world commodity prices are not, in
  general, the major factor responsible for
  movements in reported export revenues and import
  expenditures, even in countries which are highly
  dependent on these commodities.
• This is partly because of basis risk and partly
  because of unrelated quantity variations.
• If the objective is to protect countries from
  sharp falls in export revenues (or rises in
  import costs) one should operate on these
  revenues and costs price is an inadequate
  proxy.
• Any scheme must relate to specific country
  circumstances quantity and revenue variations
  which appear random to the observer may have been
  predictable or even planned by governments.

35
Part V

• Conclusions and the Way Forward

36
Our Work Suggests

• Indexing debt service to commodity export and/or
  import prices generates only modest and uneven
  benefits to indebted countries there may be
  benefits for particular countries, but it does
  not appear to be a general panacea.
• Moderation of concessional debt service
  repayments is a poor lever for protecting
  commodity-exporting countries from revenue falls.
  It may be better to look for alternative
  modalities to provide this protection.
• World commodity prices are insufficiently
  correlated with commodity export revenues for a
  non-discretionary price-based hedging schemes to
  give good protection. It may be preferable to
  look directly at export revenues.
• Country circumstances are diverse any scheme
  should form part of overall development
  assistance.

37
Hausmann-Rigobon

• Hausmann and Rigobon (HR) have proposed that IDA
  countries should be able to borrow in local
  currency on an inflation-indexed basis.
• This effectively protects countries against real
  exchange rate depreciation.
• Work by Paul Cashin and co-authors in the IMF
  shows that, for many commodity dependent
  developing countries, real exchange rates are
  well-correlated with commodity prices (also
  revenues?).
• The HR scheme may therefore offer a superior
  mechanism for protecting primary exporters
  against revenue shortfalls but this needs to be
  carefully evaluated.
• On the negative side, the HR scheme is limited by
  its dependence on concessional debt service, and
  gives rise to potential moral hazard and adverse
  selection problems.

38
What are the Objectives?

• If the objective is to ease debt payments, it may
  be preferable to consider either
• a) further debt relief,
• b) a move towards increased grant aid, or
• c) a modified version of the Hausmann-Rigobon
  scheme.
• If the objective is to provide commodity-exporting
  countries with a measure of insurance against
  the impact of adverse revenue fluctuations, this
  may be better achieved through a compensatory
  finance mechanism.
• However, compensatory finance is associated with
  well known problems (timeliness, moral hazard
  etc) which will need to be resolved.

39
A New Compensatory Finance Scheme?

• A quick-disbursing low conditionality facility
• geared to shortfalls in export revenues (plus
  sharp jumps in oil or grains import costs?)
• for pre-qualified IDA countries (e.g through
  PRSP or relevant Fund programmes), eliminating
  the need for complex conditionality negotiations
  
• relative to projections of likely future output
  (and import) trends agreed by the Bank and
  government.
• Although such shortfalls may largely be due to
  price changes (and hence be hedgable by the
  Bank), unanticipated quantity variations (e.g.
  from drought) could also qualify.
• This would not preclude IDA moving to local
  currency lending that should be seen as an
  independent proposal.

40
Thank You for Your Attention