Pakistani government cooking the books
Undermining data integrity
By Dr Kaiser Bengali
Data has a critical place in policymaking. It is on the basis of analysis of data that policymakers are enabled to identify emerging problems, the source and nature of such problems, and adopt appropriate remedial measures.The importance of data is such that the collapse of the Soviet Union is traced in part to the generalised distortion of economic data. To cite one instance, the Soviet Union had instituted a system where industry managers were provided production targets and received bonuses if the target was exceeded. This created a perverse incentive to report higher output than actual.The over-reporting of output became so pervasive across the vast country that policymakers in Moscow no longer had the correct picture of the quantity or composition of output in the economy. Bereft of accurate data, their decisions became increasingly unrelated to realities on the ground and, in effect, irrelevant to determining the course of the economy – and the country. More recently, the distortion of intelligence data by the Bush Administration in the US has led to a situation where the world’s sole superpower is facing a certain defeat in Iraq.Developing countries are unable to devote sufficient resources to match the extensive databases of the developed countries. Nevertheless, many developing countries have over time significantly improved their data collection and reporting systems with a view to enhancing their capacity for analysis and decision-making. Pakistan stands out as an exception, where the data regime has suffered a regression, particularly during the last half-decade. A measure of this regression can be made from the fact that the position of the Director-General of the Federal Bureau of Statistics – the principal data collection and production agency – has remained vacant since June 2003.The absence of the Director-General has exposed middle ranking FBS officials to direct pressure from senior finance ministry officials to “massage” the data to suit the regime’s purposes. A review of data problems over the period 2002 to 2006 brings forth two disturbing aspects of the state of affairs: one is deliberate distortion of data and the other is misrepresentation of the data. This is a sorry legacy that General Musharraf’s economic managers are likely to leave behind.Customs duty receipts: Instances abound. The ability to achieve and even exceed tax revenue targets has been heralded as one of the hallmarks of success of the Musharraf regime’s economic managers. However, there are aspects of tax data that raise serious questions. One of the federal budget documents, the Explanatory Memorandum on Federal Receipts, reports data on Customs Duty receipts for 12 specific commodity groups. A perusal of the said documents for the years 2000-01 to 2005-06 shows an inexplicable phenomenon.The Explanatory Memorandum 2001-02 shows that for seven out of 12 items revised receipts for 2000-01 are uniformly 7.3 per cent lower than budgeted receipts for the year. And the Explanatory Memorandum 2006-07 shows that for all the reported items revised receipts are uniformly higher than budgeted receipts by 10.25 per cent (Table 1 on P. IV). Further, over the fiscal years 2002-2006, growth in customs duty receipts for at least 10 out of 12 items is a uniform 3.1 in 2002-03, 9.7 in 2003-04, 27 per cent in 2004-05, and 17.3 percent in 2005-06 (Table 2 on P. IV). The uniformity across commodities in the percentage difference between revised and budgeted figures and with respect to percentage change in receipts is clearly not plausible and it appears that the data has been ‘created’ by applying a uniform growth rate for each year.Fiscal deficit: Managing the fiscal deficit has also involved an element of creative accounting. The Budget for the fiscal year 2003-04 showed fiscal deficit for 2002-03 at 4.2 per cent. This, it appeared, was an underestimate on account of what may be called the ‘KESC factor’. The Budget reported a receipt of Rs57.3 billion under the head "Recovery of Investment from KESC" and an entry of Rs57.7 billion on the expenditure side under the head "Unallocables". It is understandable for budgeted expenditures for, say, emergency relief to be placed under the ‘Unallocables’ head; but an expenditure that has already been incurred is clearly identifiable and for it to be reported as ‘Unallocable’ was found to be intriguing.Enquiries revealed that neither had the government received this amount from KESC, nor had the government incurred this expenditure. The two corresponding receipts and expenditure items were merely accounting entries that cancelled each other out, with no net accounting impact on the budget. The implication of these ‘mutually cancelling out’ entries was, however, meaningful. Essentially, the entry on the receipts side absolved KESC of its liabilities to the government to the extent of Rs57.3 billion. By showing a corresponding entry on the expenditure side, the government had effectively written off KESC's losses. The subterfuge served the purpose of under-reporting the fiscal deficit. The correct procedure would have been to just have one entry on the expenditure side to reflect the writing off of the KESC debt. In the event, however, the fiscal deficit would have risen to 5.6 per cent – which was the actual fiscal deficit for 2002-03 – instead of 4.5 per cent as claimed (see Table 3).GDP growth rates: Creative accounting also appears to have been applied to arriving at the desired economic growth rates. The principal government publication that provides a detailed and comprehensive empirical documentation of the state of the various sectors and aspects of the economy is the Pakistan Economic Survey. Usually produced on the eve of the presentation of the annual budget, the Economic Survey provides provisional figures for the current fiscal year and revised figures for the preceding fiscal year. The provisional figures are estimated on the basis of 9-10 months data that is available at the time the Survey is finalised for publication.The revised figures are based on the un-audited data for the 12 months of the preceding fiscal year. Dividing the provisional year figure (the numerator) by the revised year figure (the denominator) provides the growth rate. Higher the denominator, lower the growth rate and vice versa. As such, the upward or downward correction of the denominator – the revised figure — can lower or raise the current year growth rate.This device has been increasingly used in recent years to manipulate growth rates to desired levels. For example, the Economic Survey 2002-03 reported GDP growth at 5.1 per cent. The last year when GDP growth was above five per cent was in 1995-96, when it stood at 6.6 per cent. Coming in the wake of rather poor performance of the economy after 1996, the whiff of recovery was welcome. However, questions arose as to the actual magnitudes of growth; questions with respect to the consistency and reliability of data and estimation methodology upon the basis of which the growth rates were claimed.The preceding year’s Economic Survey (2001-02) provided provisional figures for the year 2001-02, which were updated and presented as revised figures in the Economic Survey 2002-03. This is standard practice. Given that provisional figures are annualised estimates – based on 9-10 months data – it is understandable that the revised figures will vary somewhat from provisional ones. Relatively greater variations can be accepted in, say, crop agriculture, on account of crops that are harvested in the last quarter of the fiscal year, and whose output projections can go awry because of natural factors.For other sectors, however, estimates for 2-3 months are unlikely to change too widely from the monthly average of 9-10 months data. This is the case in, say, public administration and defence, where the value addition accrues largely from salaries – and monthly salaries for the last 2-3 months of the year are unlikely to vary significantly from that in the preceding 9-10 months.A perusal of national income data indicates substantial variation between the provisional and revised growth rates. For the year 2001-02, provisional fishing sector growth of four per cent turned was revised into negative 12 per cent, 3.8 per cent provisional growth in finance and insurance expanded to 16.9 per cent revised growth, and 18.2 per cent provisional growth in public administration and defence collapsed to 6.5 per cent growth in revised terms (see Table 4). Such large corrections indicate that the process of compilation and tabulation of data in the preceding year was not carried out with the required level of accuracy. In the event, the precision of current year’s estimates are also thrown in doubt.The public administration and defence, finance and insurance, and fishing sectors then accounted for seven, two and 0.9 per cent of the economy. Clearly, the impact of any manipulation in the public administration and defence sector is greater. Provisional estimates of value addition in the public administration and defence sector in 2001-02 was recorded at Rs50,217 million, which was revised downwards by nearly Rs5 billion or nine per cent to Rs45,256 million. Such a significant decline in value addition in the sector during the last quarter of the fiscal year is questionable.The impact of this revision was two-fold. One, reporting provisional public administration and defence sector growth for 2001-02 boosted the GDP growth rate for the year. And two, the downward revision of the 2001-02 sectoral growth rate from provisional 18.2 per cent by as much as two-thirds to revised 6.5 per cent allowed the 2002-03 rate of growth in public administration and defence to be shown at positive 5.2 per cent. If, however, the 2002-03 growth rate is calculated on the basis of the uncorrected provisional 2001-02 growth rate of 18.2 per cent, the 2002-03 growth rate for the sector turns out to be negative 5.2 per cent!Likewise, the overall GDP growth rate is also affected. Based on the revised 2001-02 figures, the 2002-03 GDP growth rate is reported at 5.1 per cent. Based on the provisional 2001-02 figures, the GDP growth rate appears to be half percentage point lower at 4.6 per cent. It appears that economic managers were aiming to produce a five per cent plus GDP growth rate.Poverty estimation: Poverty is now a political issue and official poverty estimates have continued to vary, thereby casting doubts on their veracity. The Economic Survey 2002-03 reported the 2000-01 poverty head count ratio at 32.1 per cent, which was revised downwards in the Economic Survey 2003-04 to 27.3 per cent. However, a 2005 publication of a government agency – the Center for Research on Poverty Reduction and Income Distribution – revised the 2000-01 poverty rate upwards to 34.45 percent.Further, the Economic Survey 2003-04 reported the poverty rate to have reduced by 4.2 percentage points to 23.1 per cent in 2004, while the Economic Survey 2005-06 reported poverty at 23.9 per cent. The 10.6 percentage point reduction of poverty from 34.45 percent in 2000-01 to 23.9 per cent in 2004-05 translates to a 31 per cent decrease or a six per cent decline in poverty per year.The heroic nature of the claim merited a closer look at poverty data. Given the availability of the required level of disaggregated data, it was possible to analyse as to how the 4.2 percentage point decrease in poverty, claimed in 2003-04, was estimated. And it appeared that creativity had also extended to sampling the data for analysis.A perusal of the method by which the poverty decline estimate had been arrived at showed that it was conceptually and statistically inconsistent. Given the government's unwillingness to accept the results of the Pakistan Integrated Household Survey (PIHS) 2001, which showed a rise in poverty between FY1999 and FY2001, the government conducted an alternative Household Consumption Expenditure Survey (HCES) in 2004. This survey, carried out over the 18-day period April 19 to May 6: 2004, sampled 5,046 rural and urban households in the four provinces of the country.The HCES 2004 data was tabled against the segment of the PIHS 2001 data collected over the same 18-day period, i.e., April 19 to May 6. The total PIHS sample comprises 14,536 households, of which 1,096 (614 rural and 482 urban) households were surveyed during this 18-day period. Statistically, the standard error of a 1,096 households sample is significantly larger than that of a 14,536 household sample, compromising the accuracy of the results.Moreover, while the FBS sampling methodology is meticulous and ensures statistically representative coverage of the 14,536 households surveyed, it is certainly unlikely that there was an attempt to ensure the overall representativeness of the 1,096 households surveyed during the particular 18 days. It is possible that, if 100 households are to be surveyed in a territorial unit over a period of 30 days, the enumerators survey one part of the area that is either wealthier or poorer during, say, the first two weeks.In the event, the partial sample will be biased. Clearly, therefore, the HCES 2004 sample and the PIHS 2001 sub-samples are not comparable; rendering the result conceptually and statistically flawed. The exercise amounts to subjective selection of data in order to produce pre-determined results. Under the circumstances, the current even more historic claim of reduction in poverty by nearly one third is also rendered questionable.Maintaining the integrity of data is of paramount importance for planning and policy decisions and for ensuring the overall health of the economy. It is the purpose of this article to highlight the fact that this integrity is under threat and that it is imperative that the entire process of data collection, tabulation and publication is rendered free of political manipulation. This requires that, as s first step, the Federal Bureau of Statistics and the provincial statistical bureaus are made into autonomous organizations – a la State Bank of Pakistan – to ensure that these publicly funded, public organisations are devoted to the service of the state rather than servitude of a government.
By Dr Kaiser Bengali
Data has a critical place in policymaking. It is on the basis of analysis of data that policymakers are enabled to identify emerging problems, the source and nature of such problems, and adopt appropriate remedial measures.The importance of data is such that the collapse of the Soviet Union is traced in part to the generalised distortion of economic data. To cite one instance, the Soviet Union had instituted a system where industry managers were provided production targets and received bonuses if the target was exceeded. This created a perverse incentive to report higher output than actual.The over-reporting of output became so pervasive across the vast country that policymakers in Moscow no longer had the correct picture of the quantity or composition of output in the economy. Bereft of accurate data, their decisions became increasingly unrelated to realities on the ground and, in effect, irrelevant to determining the course of the economy – and the country. More recently, the distortion of intelligence data by the Bush Administration in the US has led to a situation where the world’s sole superpower is facing a certain defeat in Iraq.Developing countries are unable to devote sufficient resources to match the extensive databases of the developed countries. Nevertheless, many developing countries have over time significantly improved their data collection and reporting systems with a view to enhancing their capacity for analysis and decision-making. Pakistan stands out as an exception, where the data regime has suffered a regression, particularly during the last half-decade. A measure of this regression can be made from the fact that the position of the Director-General of the Federal Bureau of Statistics – the principal data collection and production agency – has remained vacant since June 2003.The absence of the Director-General has exposed middle ranking FBS officials to direct pressure from senior finance ministry officials to “massage” the data to suit the regime’s purposes. A review of data problems over the period 2002 to 2006 brings forth two disturbing aspects of the state of affairs: one is deliberate distortion of data and the other is misrepresentation of the data. This is a sorry legacy that General Musharraf’s economic managers are likely to leave behind.Customs duty receipts: Instances abound. The ability to achieve and even exceed tax revenue targets has been heralded as one of the hallmarks of success of the Musharraf regime’s economic managers. However, there are aspects of tax data that raise serious questions. One of the federal budget documents, the Explanatory Memorandum on Federal Receipts, reports data on Customs Duty receipts for 12 specific commodity groups. A perusal of the said documents for the years 2000-01 to 2005-06 shows an inexplicable phenomenon.The Explanatory Memorandum 2001-02 shows that for seven out of 12 items revised receipts for 2000-01 are uniformly 7.3 per cent lower than budgeted receipts for the year. And the Explanatory Memorandum 2006-07 shows that for all the reported items revised receipts are uniformly higher than budgeted receipts by 10.25 per cent (Table 1 on P. IV). Further, over the fiscal years 2002-2006, growth in customs duty receipts for at least 10 out of 12 items is a uniform 3.1 in 2002-03, 9.7 in 2003-04, 27 per cent in 2004-05, and 17.3 percent in 2005-06 (Table 2 on P. IV). The uniformity across commodities in the percentage difference between revised and budgeted figures and with respect to percentage change in receipts is clearly not plausible and it appears that the data has been ‘created’ by applying a uniform growth rate for each year.Fiscal deficit: Managing the fiscal deficit has also involved an element of creative accounting. The Budget for the fiscal year 2003-04 showed fiscal deficit for 2002-03 at 4.2 per cent. This, it appeared, was an underestimate on account of what may be called the ‘KESC factor’. The Budget reported a receipt of Rs57.3 billion under the head "Recovery of Investment from KESC" and an entry of Rs57.7 billion on the expenditure side under the head "Unallocables". It is understandable for budgeted expenditures for, say, emergency relief to be placed under the ‘Unallocables’ head; but an expenditure that has already been incurred is clearly identifiable and for it to be reported as ‘Unallocable’ was found to be intriguing.Enquiries revealed that neither had the government received this amount from KESC, nor had the government incurred this expenditure. The two corresponding receipts and expenditure items were merely accounting entries that cancelled each other out, with no net accounting impact on the budget. The implication of these ‘mutually cancelling out’ entries was, however, meaningful. Essentially, the entry on the receipts side absolved KESC of its liabilities to the government to the extent of Rs57.3 billion. By showing a corresponding entry on the expenditure side, the government had effectively written off KESC's losses. The subterfuge served the purpose of under-reporting the fiscal deficit. The correct procedure would have been to just have one entry on the expenditure side to reflect the writing off of the KESC debt. In the event, however, the fiscal deficit would have risen to 5.6 per cent – which was the actual fiscal deficit for 2002-03 – instead of 4.5 per cent as claimed (see Table 3).GDP growth rates: Creative accounting also appears to have been applied to arriving at the desired economic growth rates. The principal government publication that provides a detailed and comprehensive empirical documentation of the state of the various sectors and aspects of the economy is the Pakistan Economic Survey. Usually produced on the eve of the presentation of the annual budget, the Economic Survey provides provisional figures for the current fiscal year and revised figures for the preceding fiscal year. The provisional figures are estimated on the basis of 9-10 months data that is available at the time the Survey is finalised for publication.The revised figures are based on the un-audited data for the 12 months of the preceding fiscal year. Dividing the provisional year figure (the numerator) by the revised year figure (the denominator) provides the growth rate. Higher the denominator, lower the growth rate and vice versa. As such, the upward or downward correction of the denominator – the revised figure — can lower or raise the current year growth rate.This device has been increasingly used in recent years to manipulate growth rates to desired levels. For example, the Economic Survey 2002-03 reported GDP growth at 5.1 per cent. The last year when GDP growth was above five per cent was in 1995-96, when it stood at 6.6 per cent. Coming in the wake of rather poor performance of the economy after 1996, the whiff of recovery was welcome. However, questions arose as to the actual magnitudes of growth; questions with respect to the consistency and reliability of data and estimation methodology upon the basis of which the growth rates were claimed.The preceding year’s Economic Survey (2001-02) provided provisional figures for the year 2001-02, which were updated and presented as revised figures in the Economic Survey 2002-03. This is standard practice. Given that provisional figures are annualised estimates – based on 9-10 months data – it is understandable that the revised figures will vary somewhat from provisional ones. Relatively greater variations can be accepted in, say, crop agriculture, on account of crops that are harvested in the last quarter of the fiscal year, and whose output projections can go awry because of natural factors.For other sectors, however, estimates for 2-3 months are unlikely to change too widely from the monthly average of 9-10 months data. This is the case in, say, public administration and defence, where the value addition accrues largely from salaries – and monthly salaries for the last 2-3 months of the year are unlikely to vary significantly from that in the preceding 9-10 months.A perusal of national income data indicates substantial variation between the provisional and revised growth rates. For the year 2001-02, provisional fishing sector growth of four per cent turned was revised into negative 12 per cent, 3.8 per cent provisional growth in finance and insurance expanded to 16.9 per cent revised growth, and 18.2 per cent provisional growth in public administration and defence collapsed to 6.5 per cent growth in revised terms (see Table 4). Such large corrections indicate that the process of compilation and tabulation of data in the preceding year was not carried out with the required level of accuracy. In the event, the precision of current year’s estimates are also thrown in doubt.The public administration and defence, finance and insurance, and fishing sectors then accounted for seven, two and 0.9 per cent of the economy. Clearly, the impact of any manipulation in the public administration and defence sector is greater. Provisional estimates of value addition in the public administration and defence sector in 2001-02 was recorded at Rs50,217 million, which was revised downwards by nearly Rs5 billion or nine per cent to Rs45,256 million. Such a significant decline in value addition in the sector during the last quarter of the fiscal year is questionable.The impact of this revision was two-fold. One, reporting provisional public administration and defence sector growth for 2001-02 boosted the GDP growth rate for the year. And two, the downward revision of the 2001-02 sectoral growth rate from provisional 18.2 per cent by as much as two-thirds to revised 6.5 per cent allowed the 2002-03 rate of growth in public administration and defence to be shown at positive 5.2 per cent. If, however, the 2002-03 growth rate is calculated on the basis of the uncorrected provisional 2001-02 growth rate of 18.2 per cent, the 2002-03 growth rate for the sector turns out to be negative 5.2 per cent!Likewise, the overall GDP growth rate is also affected. Based on the revised 2001-02 figures, the 2002-03 GDP growth rate is reported at 5.1 per cent. Based on the provisional 2001-02 figures, the GDP growth rate appears to be half percentage point lower at 4.6 per cent. It appears that economic managers were aiming to produce a five per cent plus GDP growth rate.Poverty estimation: Poverty is now a political issue and official poverty estimates have continued to vary, thereby casting doubts on their veracity. The Economic Survey 2002-03 reported the 2000-01 poverty head count ratio at 32.1 per cent, which was revised downwards in the Economic Survey 2003-04 to 27.3 per cent. However, a 2005 publication of a government agency – the Center for Research on Poverty Reduction and Income Distribution – revised the 2000-01 poverty rate upwards to 34.45 percent.Further, the Economic Survey 2003-04 reported the poverty rate to have reduced by 4.2 percentage points to 23.1 per cent in 2004, while the Economic Survey 2005-06 reported poverty at 23.9 per cent. The 10.6 percentage point reduction of poverty from 34.45 percent in 2000-01 to 23.9 per cent in 2004-05 translates to a 31 per cent decrease or a six per cent decline in poverty per year.The heroic nature of the claim merited a closer look at poverty data. Given the availability of the required level of disaggregated data, it was possible to analyse as to how the 4.2 percentage point decrease in poverty, claimed in 2003-04, was estimated. And it appeared that creativity had also extended to sampling the data for analysis.A perusal of the method by which the poverty decline estimate had been arrived at showed that it was conceptually and statistically inconsistent. Given the government's unwillingness to accept the results of the Pakistan Integrated Household Survey (PIHS) 2001, which showed a rise in poverty between FY1999 and FY2001, the government conducted an alternative Household Consumption Expenditure Survey (HCES) in 2004. This survey, carried out over the 18-day period April 19 to May 6: 2004, sampled 5,046 rural and urban households in the four provinces of the country.The HCES 2004 data was tabled against the segment of the PIHS 2001 data collected over the same 18-day period, i.e., April 19 to May 6. The total PIHS sample comprises 14,536 households, of which 1,096 (614 rural and 482 urban) households were surveyed during this 18-day period. Statistically, the standard error of a 1,096 households sample is significantly larger than that of a 14,536 household sample, compromising the accuracy of the results.Moreover, while the FBS sampling methodology is meticulous and ensures statistically representative coverage of the 14,536 households surveyed, it is certainly unlikely that there was an attempt to ensure the overall representativeness of the 1,096 households surveyed during the particular 18 days. It is possible that, if 100 households are to be surveyed in a territorial unit over a period of 30 days, the enumerators survey one part of the area that is either wealthier or poorer during, say, the first two weeks.In the event, the partial sample will be biased. Clearly, therefore, the HCES 2004 sample and the PIHS 2001 sub-samples are not comparable; rendering the result conceptually and statistically flawed. The exercise amounts to subjective selection of data in order to produce pre-determined results. Under the circumstances, the current even more historic claim of reduction in poverty by nearly one third is also rendered questionable.Maintaining the integrity of data is of paramount importance for planning and policy decisions and for ensuring the overall health of the economy. It is the purpose of this article to highlight the fact that this integrity is under threat and that it is imperative that the entire process of data collection, tabulation and publication is rendered free of political manipulation. This requires that, as s first step, the Federal Bureau of Statistics and the provincial statistical bureaus are made into autonomous organizations – a la State Bank of Pakistan – to ensure that these publicly funded, public organisations are devoted to the service of the state rather than servitude of a government.
Labels: economy, manipulation, Pakistan
posted by India News Watch at 9:22 PM
<< Home