Review of Islamic Economics, Vol. 8, No. ɪ, 1004
studies were confined to the US financial sector. There, the large number of
banks has traditionally facilitated econometric modelling (Avkiran 1999). Of
late, the literature on efficiency issues is also growing apace in developing
economies, especially those in transition to free markets.
10. Usually the conditions of linear homogeneity and symmetry are imposed on the
data prior to estimation exercise implying constant returns to scale. See, for
example Saaid et al., (2.003), n∙ ɪ 1∙
it. Interestingly, Majid et al. (1003: 405) did promise to provide the ranking of 34
banks, included in their sample on this basis, but their paper eventually provided
no such ordering.
12. See Saaid et al. (2003): They use the technique to extract share equations for
capital and deposits (pp. 131, 134), but it is unclear why they dropped the one
for labour prior to estimation (n.rτ, p. τ39).
13. If uj were equal to zero in fact or by assumption, SFA would be reduced to
central tendency analysis.
14. For details see Bock (1998, pp. ɪ - 5). Information on relevant software is
available in Herrero and Pascoe (2002).
15. Sathye (2003), Jemric and Vujcic (2002), and Leong et al. (2002) provide good
illustrations of the point.
16. For detailed comments on this paper see Hasan (2003).
17. It is a difficult question because the mainstream banks do not maintain separate
full cost statements for the Islamic windows, e.g. their share in the overheads is
not available to researchers. Maybe, it is because this significant component is
missing or underestimated that the authors reached the pleasing conclusions that
they did.
18. What the authors present in Table 2 of their work looks like regression
coefficient. How and why they used MLM for their determination is not
explicit.
19. For clarity, the symmetry and linear homogeneity conditions would have been
better stared separately in n. 8, p. τ39. Some of the parameters in Table 2, p.
134 do not seem to satisfy the specified conditions.
20. It may be indicated that the SFA does not provide the allocative efficiency
estimates AEi; the same are estimated through division of the CEi (= OEi) by the
corresponding TEi. As such, the authors could well have kept the inefficiency
scores shown in their Table 4, p. 135 as (ι - CE); alternatively, they could have
clarified that discrepancy arises due to the relationship between efficiency CE
and inefficiency CIE. (See n. 5 above.)
21. The authors could easily have tested the validity of their claim, regarding the
adverse efficiency effect of forced diversion of finance to agriculture, using a
fixed effects model.
22. The equation could be valid for an output oriented formulation where C* > C.
But in the cost-oriented approach that Saaid et al. have taken, we find OE=CVC
(see Kebede, p. 13). For, as C > C*, OE remains < ι. Compare the authors’
formulation with those in our Figures ɪ and 3; see also equation 4.3.
23. The authors impose on their model the restriction ∑λ∣=ι.
24. Some studies target profit as the efficiency measure. For example, Hassan and
Bashir (2003) use the criterion for an aggregative analysis. It must be indicated
that profitability “can be characterized as a performance indicator of single unit