Page 248 - Proceeding 2015

FAMP ANDRONICEANU Armenia and OHANYAN Gurgen
CCASP IMF POLICIES AND THEIR EFFECTS ON EDUCATION IN ROMANIA

mentioned above, there is a notable amount of studies claiming that IMF participation led to reducing

budget allocations on educations, freezing salaries of public employees and cutting jobs in public
institutions. Hence, we present comprehensive discussion of employed methodology in the next chapter.

4. DATA AND METHODOLOGY

In the literature there are several statistical methods employed to estimate effects of IMF programs. In
this section we briefly introduce those main methodological approaches by justifying our choice of a

method. Particularly, Ul Haque and Khan (1998) segregate the following four methods: before-after
method, with-without, generalized evaluation and comparison of simulations.

Yet, Vreeland (2006) goes further by splitting generalized evaluation into three distinct approaches, there

are as follows: controlling for selection on observed variables, controlling for selection on unobserved
variables and instrumental variables. Each of them represents sophisticated methods to tackle selection
problem in evaluation of IMF programs.

The before-after (BA) and with-without (WW) approaches are more intuitive and have drawback as they
do not count selection problem, but their advantage is the ease of calculation. Yet, generalized evaluation PROCEEDINGS OF THE 11 TH ADMINISTRATION AND PUBLIC MANAGEMENT INTERNATIONAL CONFERENCE ”Strategic Management for Local Communities” 30 th – 31 st October 2015 Bucharest
and controlling for selection demand large-n observations, which are available in cross country and panel

data analysis. Moreover, these methods, as well, have their limitations. Particularly, controlling selection
methods are based on assumption that errors are distributed normally or bell-shaped. Instrumental
variables could solve the problem with selection, but factors that condition the selection into IMF program

can influence its effects, as well (Vreeland, 2006). Thus, assessment of IMF conditionality effects on
program countries is not straightforward.

Given that our research attempts to identify IMF conditionality effects on Romanian education, we have

faced with real constraints on data availability. Data limitation hinders employment of above mentioned
sophisticated methods for evaluation remaining the most appropriate ones BA and WW. Meanwhile, BA
approach could be applied in this case, as Romania since 2009 has signed three continuous SBAs with

IMF with two years duration, respectively in 2011 and 2013. Hence, there is no available data on education
indicators to evaluate after program performance.

We have consulted the literature on overcoming selection problem in WW approach. As not considering

the selection could cause serious misinterpretation of results. Particularly, it is not fair to compare a
country that has problems and turned to the IMF with one that is more preferable economic situation.

246