the low income probability for the good type (1 i qH); where (m i 1) is equal to
the rate of return on the investment, V represents the investment’s ...xed costs in
the ...rst period and qH is the probability for the good type to have a high income
in the second period.
Intuitively, as the probability for the good type to have a high income in
the second period increases, it will decrease also its need for the debt relief, since
in the event of a high income the good type will always repay its debt. The
whole expression is negative, this suggesting the existence ofan inverse correlation
between a country’s investments and the level of the debt relief.
®(Q(2) + bS) represents what creditors could seize in case of default, where
® is the fraction of available resources which can be used to repay the debt,
Q(2) is the country’s low income value in the second period and bS represents
the bene...cial e∏ect of the programmes adoption on period-two outcomes (where
1 > b , 0): S is the costs of the IMF adjustment programme and it indicates a
direct reduction of welfare rather than a .nancial cost. It should be viewed as a
loss of social welfare due (for example, to adverse social ejects such as reduction of
social services and adverse shifts in income distributi on). In Section 4.1.4 we will
discuss better how these qualitative variables will become the control variables of
our empirical model.
4. The Empirical model
In the empirical model we want to test the existence of an e∏ect of a Fund pro-
gramme adoption on the subsequent concession ofa debt rescheduling. As we saw
in Section 3 two di∏erent empirical literature have developed, which have consid-
ered, independently, IMF arrangements and debt rescheduling. Here, instead, we
want to estimate a bivariate probit model for the joint determination of a Fund
programme adoption and of the debt rescheduling conditional on the programme
16