By W.A Wijewardena –
Alleged massaging of growth numbers: Should a credibility restoration exercise be launched for Sri Lanka’s statistics agency?
Allegation of massaging GDP numbers
A recent newspaper report has alleged that the top-most official in the country’s official Statistics Bureau, the Department of Census and Statistics or DCS, has instructed the officer handling the GDP numbers to increase the growth rate in Quarter I of 2013 from 5.5% to 6% when there were no ground conditions warranting the issue of such instructions (available here ).
The top-most official concerned has denied any wrong-doing but declined to give details of his side of the story expecting the problem to die away ‘just like the bubbles forming instantly in an opened bottle of soda will die away on their own accord’. He has good reason to rely on this bubble-dying theory since the media do not follow up these stories and the civil society is somewhat apathetic to them.
However, in this digital world, there is already a digital footprint of this allegation in the clouds and therefore it cannot die off naturally as things would have happened a few decades ago. To add to the digital footprint, previously, at the Committee Stage of the Budget 2014 debate, the opposition Parliamentarian Anura Kumara Dissanayake too had made the same allegation in Parliament quoting a statement reported to have been made by the officer concerned at an initial disciplinary inquiry.
It is a serious allegation since the country’s credibility is at stake
Nothing has been proved or disproved as yet. But, this allegation is serious enough to deserve the attention of top politicians on both sides of the House, top policy makers, academia, international watchdogs of the country’s economic data such as IMF and World Bank and civil society activists.
This is because, if the allegation is true, it is the country’s credibility which is at stake and it is in the interest of all that such credibility is not compromised for whatever the reasons. Countries that made such compromises in the recent history like the former Soviet Union, Argentina and Greece have lived to experience the most horrible outcomes in their respective economies.
The use of an arbitrarily appreciated exchange rate to report GDP in dollars
This writer in a previous My View in this series faulted DCS for using an appreciated exchange rate to show a higher per capita GDP in 2011 and the Central Bank for not qualifying it correctly when there was an arbitrary appreciation of the rate in the midst of a widened trade gap and a loss of the country’s foreign exchange reserves to maintain that appreciated rate (available here ). That was a technical error made by DCS in computing GDP numbers. But the present allegation is not on a technical error but on a fraudulent act on the part of the senior officers of DCS.
These acts, categorised as bureaucratic frauds, demonstrate the failure of the economic policy governance in the country. Since the most important economic numbers concerning growth, poverty, unemployment and inflation are computed by DCS, any scandal bringing its credibility to question should not be treated lightly by DCS as well as the civil society.
Need for revealing which SNA is being followed
The website of DCS has reported that the national accounts of the country that produce GDP numbers are being compiled in accordance with the guidelines issued by the United Nations known as the UN System of National Accounts or SNA. However, DCS has not revealed which revision of SNA is being currently followed by it for this purpose since SNA has been revised periodically in1968, 1993 and 2008 after it was first published in 1953. Hence, what is applicable for current compilation is its latest version issued in 2008.
This qualification is important since there is a wide variation in the method of compilation from one revision to the other. If an old outdated manual is used by DCS, its numbers will suffer from a number of deficiencies with respect to comprehensiveness, coverage, treatment and final results. Further, there is no mention of the internal governance procedure being adopted by DCS in compiling GDP data or whether DCS follows any other international best practices such as the General Data Dissemination System or GDDS of IMF. Even the Annual Report on National Accounts of Sri Lanka 2012 issued by DCS in September 2013 does not clarify these issues.
Sri Lanka has not updated IMF Meta-data template since 2007
Sri Lanka has been a participant in GDDS since 2002 when GDP data were compiled by the Central Bank for its own publication and interpretation purposes. In terms of GDDS, the participating country has to report to IMF a detailed account of data integrity and access by the public. After the compilation of GDP data was handed to DCS in 2007 – since there was no necessity for two government agencies to do the same job – DCS too has made a detailed report on its data integrity system to IMF.
However, the last update of this report has been more than six years ago on 1 June 2007 according to the IMF website implying that DCS has not taken the commitments under GDDS which bind it to good policy governance in data compilation and release seriously (available here ). This writer in the My View under reference drew the attention of DCS to this deficiency but it appears that even after one and a half years since then, no attempt has been made to upgrade Sri Lanka’s position.
The claims made by DCS to IMF
According to this non-updated report, DCS has claimed that it has ‘independence’ to determine the ‘release of data’ and ‘methodology’, no governmental official has access to GDP data before they are released to the public, senior government officials do not make commentaries on the data so released and data are qualified as ‘provisional or revised’, as the case maybe, in all publications. However, DCS has not agreed to give advance notice of the changes in methodology except making some technical clarifications.
This is an admission of serious lapse in its governance procedure since it gives DCS the right to effect changes to methodology arbitrarily to suit the interested parties and thereby keeping the public to whom the data are addressed in the dark. Though it has promised to IMF that there is an advance release calendar of GDP data to the public, posted to its website annually, the current web page does not contain such an advance release calendar.
By way of future plans, it has promised to improve the methodology and train the staff and sought for technical assistance from IMF for training and procuring the needed equipment. Since there has not been an update of the original report, there is no evidence that these plans have been successfully implemented. Hence, there appears to be a vast gap between what has been promised to IMF and through IMF to the global and local community and what has been practised by DCS.
SNA is not a proof of data integrity
DCS has claimed that its GDP data are being compiled according to UN SNA without qualifying which SNA it has adopted, that is, whether it is 1993 SNA or 2008 SNA. In any case, SNA is not a data integrity measure but a uniform practice to be followed by all nations. Thus, the adoption of SNA will help a country to compile data according to this uniform method thereby making its GDP numbers comparable with other nations which also follow the same methodology.
What is at issue in terms of the recent allegation is not international comparability of data but the integrity of the data compiled by the country. The leg for this integrity comes from GDDS which requires its participants to follow a standard data disclosure and protection procedure. This procedure requires the participant country to erect effective firewalls between the top policy makers and politicians on the one hand and the officials who compile GDP data on the other.
Good data integrity practices
According to the data integrity template of GDDS, the following are the firewalls and good practices which have to be put in place by a participating country.
No advance release calendar reported by DCS
First, there should be an advance release calendar of the data well publicised through the website of the agency compiling GDP data. This advance release calendar, while imposing discipline on those who compile the data to plan and finish the job on time, does prevent the use of the data release dates to accommodate political or other objectives of those in power.
For instance, if the advance release calendar says that the First Quarter data should be released on 30 April, it cannot be advanced in order to meet the aspirations of anyone holding high office in the government if the data are favourable or delay the issue if the data are unfavourable. In other words, the data compiling agency should serve the public who are considered supreme and sovereign and not the political masters.
High officials should not have access to data before releasing to the public
Second, the high officials of the government, including the political masters, should not have prior access to the GDP data before they are released to the public. This requirement is there to prevent the top officials from interfering with the data compiling procedure or the final numbers estimated.
Even if the data are unfavourable to them, they have no way of influencing the compilers since they come to know of the data only after they have been made public. This gives the total independence to data compilers to estimate the data according to the approved economic methodology which in Sri Lanka’s case has been outlined by UN SNA.
High officials shouldn’t interfere in estimation processes
Third, the high officials of the government should not have opportunity to make ad hoc or arbitrary commentaries on the estimates made by the compilers. These commentaries take the form of announcing that the actual growth rates are higher or lower than what has been estimated by the compilers. Top public policy officials and politicians in power always desire to show better economic performance which they consider as a success story of their policies.
When the actual estimates do not support their desires, they may seek either to refute the data or influence the data compilation in subsequent years. If the top officials do not agree with the methodology adopted, they have the freedom to make suggestions which have to be carefully evaluated by independent experts on the subject. If they agree with the top officials, the change in the methodology should be well announced to keep the public informed of the impending changes and that the data series are not comparable with previous data series without making suitable revisions to them. It is this requirement which DCS has not agreed to adopt.
A multi-level review process is needed to prevent personal interests from influencing estimates
Fourth, though IMF template has not explicitly mentioned about it, to maintain data integrity, there should not be opportunities for a single person or a group of persons – whether they are inside the compiling agency or outside the agency – to decide on the estimated numbers without following the due procedures.
To realise this goal, there should be multi-level review and verification procedures with clearly laid-down review terms for those who participate in the review sessions. Such a system is necessary to prevent the compilers from presenting arbitrary numbers without following the accepted methodology or the top officials in the agency from coming up with arbitrary numbers. All decisions made at the review and verification sessions should be recorded and kept as evidence of following the due procedure in compiling the numbers for later reference if necessary.
Central Bank had such a multi-level approval process when it compiled GDP before 2007
This writer recalls that when the GDP data were compiled by the Central Bank prior to 2007 this procedure was adopted. The data were compiled by a team of technically qualified officers in the National Accounts Division of the Department of Statistics. After the data were compiled by this team, they were verified by a team of higher ups in the Department for consistency and compliance. Once the data have been finalised by the Department, they are presented to a high level committee chaired by the Deputy Governor in charge of monetary policy and research and consisting of the Director of Economic Research and the relevant Assistant Governors as the other members.
The job of this high level committee is not to increase or decrease the estimated growth rate but to verify whether the due methodology has been followed in the compilation and whether there is internal consistency in the estimates made. For instance, if there is a high growth in the garment s and textile industry, it should be supported by the export data during the period. Or coconut production cannot go up if the coconut prices too have increased during the period indicating a short supply.
After the estimates have been checked for compliance and consistency, they are released to the public in accordance with the advanced release calendar. The Chinese Walls created are such that these numbers are not shared with anyone within or outside the Central Bank. Even the Monetary Board of the Central Bank will come to know of these numbers only after they have been released to the public.
DCS should do its job independently
Thus, the good practice requires DCS to follow an independent procedure to compile the GDP numbers. DCS is not supposed to follow the statements made by the Ministry of Finance or the Central Bank on the expected economic growth in the country. This is because the compilation of GDP numbers is a micro-level exercise involving the collection of information from a variety of sources.
According to the reported statement of the official under reference, DCS has been collecting such information from more than 200 outside agencies. Both the Central Bank and the Ministry of Finance do not have access to such a vast number of agencies that provide information to DCS. Hence, both these reputed institutions make announcements on the projected economic growth of the country based on macro-level projections of the potential economic growth which the country may achieve in a given year. These projections may or may not be realised depending on the changes in the ground conditions. Hence, after DCS has come up with the actual estimates, it could prepare a reconciliation statement highlighting the reasons for deviations from the original macro-level projections made by the Ministry or the Central Bank. In the interest of transparency and disclosure, this statement should be made public.
The reported instructions by a Deputy Director-General of DCS to the officer concerned that it was the wish of the Director-General that he should pay heed to the announcements made by the Ministry of Finance and the Central Bank does not go along with this requirement. DCS is not there to prove or disprove these two institutions; it is there to report the actual growth, estimated by following a set methodology, to the public.
A credibility restoration exercise needed
In the aftermath of these reported allegations in Parliament and in the media, there appears to be a need for a ‘credibility restoration exercise’ for the country’s official statistics agency.
*W.A Wijewardena – Formerly Deputy Governor of the Central Bank of Sri Lanka and presently Visiting Lecturer at PIM, University of Sri Jayewardenepura, Asian Institute of Technology, Bangkok and Naresuan University, Thailand. He can be reached at firstname.lastname@example.org