Therefore, we expect to ...nd a signi^cative and positive sign for the ancient
of this dummy). Instead, for the way in which we have constructed variable
C, past Fund arrangements (that is arrangements which have been made many
years before the debt rescheduling) should not in≠uence debt rescheduling in the
present. More speci.cally, we assumed that when reschedulings are conditional
on IMF programmes, only recent ones are assumed to in≠uence them.
The growth of government consumption is one of the variables which deter-
mines the so called “supply side” of a Fund arrangement, that is the probability
that the Fund would approve the request of a loan, rather than a∏ecting the prob-
ability that a country would ask for IMF intervention. This disti nction between a
“demand” and a “supply side” on IMF arrangements was made.rst in the paper
by Knight and Santaella (1997) in Section 2.1. Unfortunately, we could not use
here the other three variables in their paper as three more “instruments”; that is
“nominal depreciation exceeding 5%”, “two year change in government revenues”
and “two year change in real domestic credit”. The .rst two due to the unavail-
ability of these data and the last one because, at the start of the analysis we could
not .nd a signi...cant e∏ect.
Restraint on central government expenditure is a key element for the Fund
to approve an arrangement (and thus we would expect to .nd a signi.cative and
negative sign for the coe^ient of this variable) while, as far as we know, there is
not such an explicit requirement to obtain a debt rescheduling (as con.rmed in
the literature survey on the determinants of a debt rescheduling in Section 2.2).
4.2. De...nition of the variables
This model is estimated as a cross section using annual observations, the overall
period goes from 1983 to 1995. There are values corresponding to all the four pos-
sible combination of C and I . The data are taken from the International Financial
Statistics (IFS), the World Bank Tables, the World Development Indicators and
20