A> A
17 One could think of period-two values as the present discounted value (i.e a stock) of all future disbursement and repayment streams (i.e flows).
18 N ote that repayment capacity, i.e. the maximum resource transfers to creditors, is assumed to be given. The bargaining problem between debtors and creditors about the amount of resource transfers will be considered below. Here it is sufficient to assume that resource transfers in each period are equal to the maximum amount of resources available to a country for the purpose of debt servicing, minus a fraction reserved to maintain a minimum subsistence level.
xi+X2/(l+r)<D, creditors know they will not be p aid back th e full am ount of new
lending, and therefore refuse to lend at any level of interest. C reditors reckon th a t the
problem is one of insolvency, since outstanding debt cannot possibly be repaid, and
are left w ith no choice other than to forgive p art of ou tstanding debt. D ebt relief then
takes the form of either reduced principal (lowering Do), or redu ced rate of interest
(on Do and /o r Do-xi), or a com bination of the two. In sum, w ith no uncertainty about
future resource transfers, a liquidity problem cannot exist: either th e country is able
to pay, or, if it is not, the problem is exclusively one of insolvency an d can only be
dealt w ith by forgiving the fraction of debt exceeding its ability to pay.
C onsider now h ow uncertainty about the country's value of future stream of earnings,
and thus its future repaym ent capacity, changes the analysis. U ncertainty is
introduced by assum ing repaym ent capacity to be stochastically determ ined by a
num ber of factors, som e of w hich are exogenous to the borrow er (i.e. any exogenous
shock affecting the country's repaym ent capacity, such as a d ro p in the w orld prices
of a country's key export com modities), and others th at are endogenously
determ ined by the country itself (e.g. its investm ent decisions an d adjustm ent efforts).
A dding uncertainty, period-tw o earnings becom e a stochastic variable, w hich for the
sake of sim plicity is assum ed to take only tw o possible realisations of a random
process: one associated w ith a 'g o o d state' of nature, denoted x g and occurring w ith
probability p , and the other w ith a 'b ad state', x b . The tim e flow of e x p e c t e d
repaym ents and new disbursem ents changes accordingly:
Do; r Period 1 P eriod 2
R epaym ent capacity (earnings) XI p X G + ( l - p ) X B
N ew lending Do-xi 0
As a result, the expected value of repaym ent consistent w ith the solvency condition is
now:
(l+r)(D-xi)<pxG+(l-p)xB => (D-xi )<(pxc+( 1 -p)xn)/( 1+r) (13)
In contrast to the condition of certainty considered above, here the question as to
w hether or n o t the debtor is solvent is not well defined. Certainly, w ith (13) holding
true, and pro vided th at lenders are risk-neutral, the country w ill be able to attract
vo luntary lending of the am ount (D-xi). H ow ever, even then it is n ot to be taken for
granted that the country will actually earn enough to repay its debt, d epending
obviously on the realisation of either state of nature. Therefore, it is u p to creditors'
subjective assessm ent of a country's solvency - i.e. condition (13) - to determ ine
w hether the country w ill experience a liquidity crisis. W hat if the inequality
condition (13) is not fulfilled? A t first sight, it w ould app ear th at creditors w ould not
extend further loans (D-xi) to the country, since the expected value of new funds
w ould fall short of the am ount lent (i.e. their face value). If so, a liquidity crisis w ould
occur in period one, and creditors w o u ld be able to collect only a fraction Z, in
p resent value term s, of o utstanding debt. A ssum ing th at Z<(pxG+(l-p)xB)/(l+r) 19,
creditors can, how ever, im prove on their outcom e by rolling over the debt, and
holding o u t until period two. This is so because partial default is possible, b u t not
certain. If the good state is realised, the creditors m ay be paid back in full after all. If
the bad state occurs, they will still have im proved u p o n the outcom e associated w ith
a period-one liquidity crisis, as long as they are able to extract from the country an
interest rate (i) hig h enough to enable creditors to receive all potential resource
transfers in either state. This interest rate is m axim ised by the lenders setting it so as
to exhaust the resource transfer in the good state: (D-xi)(l+i)=XG.20Accordingly, the
scenarios w ould involve:
Do; r Period 1 Period 2
Liquidity crisis (no re XT. Z < ( p X G + ( l - p ) X B )
lending)
N ew lending Do-xi x g full paym ent (Do-xi)(l+i)
x b partial default, b u t maxt ( p X G + ( l - p ) x B )
19 Assum ing costs of default arising, for instance, from an imperfect enforcement mechanism, such as the inability to seize all available assets of the debtor, or sim ply from a variety of transaction costs associated with default.
20 That is, with i>r if x d ( l + i ) > D - x i, and r>i if x d ( l + i ) < D-xi. Creditors, however, will always perceive new lending always as concessional, since it exceeds expected value of repayment. N ote also that if there is som e probability of the good state occuring, new lending will always be at i>r.
K rugm an (1988a) em phasises that only existing lenders have the incentive to extend
new borrow ing. By doing so, they m axim ise the potential resource transfer on total
ou tstanding debt, so to say 'd efen d in g ' its net present value. By contrast, new lenders
w ould n ot offer any loans, since the expected value of new len ding falls short of its
face value:
(pxG+(l-p)xB)/(l+r) < (D-xi)(l+i) = x g (14)
V iew ed in isolation, new loans to the country are low er than expected repaym ent,
w hich obviously deters new lenders from entering a contract w ith th a t country,21
W ith this sim ple b u t insightful analysis, K rugm an (1988a) w as am ong the first to
highlight tw o crucial characteristics involving sovereign deb t strategy. First, it
dem onstrates th at the analytical dichotom y of insolvency versus illiquidity is
essentially flawed: if it could be know n w ith certainty that a country is solvent,
lenders w ould extend n ew loans in all cases, so as to ensure full repaym ent of
outstanding loans. In contrast, a country know n to be insolvent w ould consequently
also be illiquid, w hile the opposite w ould not be true. A sim ilar logic applies w hen
future p aym ent capacity is uncertain: as long as creditors deem the expected value of
resource transfers higher than o utstand in g debt, new lending w ill still take place.
W ith uncertainty, how ever, it is creditors' subjective expectations th at can bring
about a liquidity crisis of a solvent borrow er. Illiquidity occurs ou t of expected
insolvency, w h eth er or not expectations prove to be w rong ex post. Second,
K rugm an's analysis explains the typical p attern of 'defensive lending', characterising
the lenders' attitude tow ard highly indebted borrow ers, particularly LICs. In
particular, defensive lending is show n to occur even in a situation of expected
insolvency, since existing creditors have the incentive to defend their existing loans
by m axim ising potential returns (or m inim ising expected losses) on their overall
stake in the debtor country.
21 There are many alternative w ays to show the rationale for new lending by existing creditors. Suppose,