The A2F household survey was calibrated to publicly available results from FBS. Calibration was mainly done based on:
Population data from the National Census of 1998 and Projections available in Economic Survey2014-15;
Income data available from PSLM 2012-13;
Other Government of Pakistan conducted surveys.
Since it was decided to give an over-weight to smaller provinces, so as to ensure that province-level results are statistically significant, the self-weighted sample approach could not be used and a detailed weighting exercise was conducted to calibrate the survey's results .
Regarding the comparability of the 2015 A2F survey results with those of the 2008 A2F survey, the following should be noted:
The information available on the 2008 A2F Survey calibration1 reveals that, in spite of this calibration exercise, weighting resulted in a slightly higher average income than the national averages (as recorded then by PSLM 2005 even after inflation adjustment). This means that those financial inclusion indicators which are strongly dependents on income must be compared with a caveat: where income has a positive relationship with a particular financial inclusion indicator, the 2008 A2F survey results may have overstated its penetration. We believe that high prevalence of borrowing in the 2008 survey may be partly explained by this phenomenon.
The 2008 A2F survey took the urban / rural proportion to be 37% urban and 63% rural. The 1998census, which is currently the only benchmark available, measured the urban / rural distribution as 35% vs. 65 %. Thus, in comparison to 1998, the 2008 survey gave more weightage to urban areas. The 2015 A2F survey has stuck to the 1998 census figures for urban / rural distribution, because we believe that, for any valid comparison with other benchmark studies available in the country, the distribution of the population in urban and rural zones must be maintained as found in1998. Our reason for this choice is that possible evolutions in the distribution of the population in urban and rural zones cannot be accurately predicted until a census happens in the country. Indeed, more recent estimates of a possible relative increase in the urban population have been widely contested. Consequently, financial inclusion indicators that are heavily dependent on the type of location (urban vs. rural) should be considered with caution when the 2015 data is compared to the 2008 data.
The 2008 survey was conducted between early November 2007 and March/April 2008, though over 90% of the survey was completed in early 2008. Thus, most of the field work was completed before the assassination of Mohtarma Benazir Bhutto and the subsequent deterioration in security and law and order situation in the country, particularly along the western border areas of Pakistan.