The quality of the macroeconomic statistics is generally high, broadly conforming to international standards for compilation and dissemination, although coverage of illicit activities is poor. Executive Directors suggest reviewing the legal framework for statistical activity, including strengthening the institutional and financial independence of the National Administrative Department of Statistics. Further strengthening is needed in the areas of national accounts and government finance statistics. Developing a methodology to include unrecorded trade in the balance of payments on a timely basis is also required.

Abstract

The quality of the macroeconomic statistics is generally high, broadly conforming to international standards for compilation and dissemination, although coverage of illicit activities is poor. Executive Directors suggest reviewing the legal framework for statistical activity, including strengthening the institutional and financial independence of the National Administrative Department of Statistics. Further strengthening is needed in the areas of national accounts and government finance statistics. Developing a methodology to include unrecorded trade in the balance of payments on a timely basis is also required.

I. Overall Assessment

1. The quality of the macroeconomic statistics is generally high, broadly conforming to international standards for compilation and dissemination, although coverage of illicit activities is poor. Despite a few gaps in the extensive legal framework for statistical activity, in practice, the division of labor among compiling agencies is adequate and inter-agency collaboration is reasonably good. In a decentralized environment with multiple compilers for key macroeconomic indicators, however, there is a need to further strengthen collaboration to facilitate a better understanding of differences, including by the public. Untimely responses by some decentralized units slow dissemination of the national accounts. There is some scope to further improve the methodological basis and source data underpinning most datasets, as well as resources devoted to the national accounts and fiscal statistics. High-exposure illicit activities are covered only partially in the national accounts and related exports excluded from the balance of payments (BOP). However, statistical agencies are well placed to make further statistical progress on the basis of present plans and ongoing initiatives. In most cases, attention to user needs and good dissemination practices are supported by observance of the Special Data Dissemination Standard (SDDS).

2. Colombia subscribed to the SDDS on May 31, 1996 and started posting its metadata on the IMF’s Dissemination Standards Bulletin Board (DSBB) on April 25, 1997. Colombia is in observance of the SDDS, meeting specifications for coverage, periodicity, timeliness, and dissemination of advance release calendars. Colombia uses flexibility options on the timeliness of the production index and the analytical accounts of the banking system. For most datasets, summary methodological statements have yet to be posted on the DSBB. Appendix I provides an overview of dissemination practices compared to the SDDS.

3. In applying the IMF’s Data Quality Assessment Framework (DQAF), the mission arrived at the following conclusions:

4. Prerequisites of quality and assurances of integrity

  • The activities of the National Administrative Department of Statistics (NADS) are supported by a myriad of decrees and laws. While this ample legal framework nominally empowers the NADS to play a leading role in the statistical system, several of the key legal provisions are not implemented. In particular, the coordinating role of the NADS is effectively absent, even as the decentralization of statistical activity that began in 1993 increased the need for oversight. A deficiency is the lack of provisions mandating dissemination across the statistical system. The professional independence of the NADS could be seen as vulnerable, as its head is appointed by the President for a period coinciding with that of the government. A paucity of resources and comparatively low compensation for its staff slows NADS’s efforts to improve the national accounts and other statistics. Despite these shortcomings, there is a tradition of professionalism and high technical standards at the NADS that have facilitated its work and earned high credibility for its output, particularly the CPI.

  • The institutional environment under which the Office of Advisors to the Fiscal Policy Council (AFPC) at the Ministry of Finance and Public Credit (MFPC) compiles statistics is complex. While the AFPC compiles fiscal statistics for policymaking, other agencies compile fiscal statistics for auditing and control purposes. Financing data are prepared by the Bank of the Republic (BR). Despite the lack of a legal mandate, the AFPC releases data to the public directly and through the SDDS. As a result of insufficient coordination and legal constraints, there is considerable duplication of collection efforts and reconciliation across data prepared by various agencies is difficult. User needs are not monitored and there is no system to monitor the quality of statistical programs, except that specified under the program supported by the Stand-by Arrangement with the IMF. Pre-release access to data by officials involved in policymaking and IMF staff is not made known to the public. Staff and computing resources are insufficient to make needed improvements. AFPC officials follow ethical guidelines contained in several laws and regulations.

  • The Law of the Bank of the Republic along with relevant decrees and internal resolutions provide a good basis for data collection, although there is no legal mandate to disseminate statistics. Nevertheless, the BR has a longstanding tradition of compiling and disseminating statistics for the benefit of policymakers and to inform the public. The Governor of the BR is elected by a board of directors, most of whom are appointed for staggered terms, thereby providing assurances of technical independence. The BR maintains close relations with relevant agencies, most notably the superintendents of banks, securities, and companies. This facilitates timely transmittal of source data and the levying of penalties for nonreporting/misreporting, if needed. Resources devoted to statistical production are adequate, with highly qualified professional staff. Emphasis on quality is high, as reflected by promotion of training and research. While relevance of the BOP is assisted by extensive contacts with key users, a policy of consultations with users of the monetary and financial statistics (MFS) is lacking. The BR recruits qualified staff and promotes professionalism and observance of ethical standards.

5. The methodological foundations of the macroeconomic statistics are generally sound. The national accounts have been compiled in line with best practice for over three decades, and work is underway to update the base year. However, coverage of illicit output is limited to value added in agricultural production. Commendably, the government finance statistics (GFS) cover the entire nonfinancial public sector (NFPS), although using a representative sample of local governments and enterprises for “above-the-line” estimates. Classification is in line with international standards, except that functional classification of NFPS expenditure is not prepared and BOP misclassification of certain foreign direct investment (FDI) transactions. The MFS generally follow the recommendations of the MFSM, and improvements in classification, sectorization, and basis for recording are underway. There is a need to publish monetary aggregates consistent with international standards, excluding central government deposits. While BOP statistics generally follow international standards, unrecorded trade and related counter flows are excluded.

6. Accuracy and reliability of the national accounts and the GFS are hindered by data gaps. A major shortcoming in the case of the national accounts is the lack of an up-to-date business register to support sample surveys with adequate coverage, especially of enterprises in the service sector. Source data for compiling statistics on the operations of local governments (departments and municipalities) are poor, but a project to improve these data and reduce the reporting burden is in place at the MFPC and the General Accounting Office of the Nation (GAO). The project envisages providing technical assistance to local governments, which will require more resources. Statistical techniques used to validate BOP data from surveys, ITRS, and administrative records promote accuracy and minimize errors.

7. Serviceability of published macroeconomic statistics is very good in most cases. A recent slippage will be addressed shortly, as the final estimate of annual GDP for 2003 is to be published by end-20051, following a long delay in the publication of the preliminary estimate. While data dissemination generally observes SDDS prescriptions, users have expressed a desire for higher frequency and better timeliness for certain variables (see below). BOP and monetary statistics are consistent within the dataset, over time, and with data from other sources. A weakness of the GFS is the lack of systematic reconciliation of certain data (including on public investment) with other datasets. The AFPC has not implemented a regular and transparent revision schedule. Revision studies are seldom published.

8. Significant efforts to improve the accessibility of macroeconomic data have yielded good results. For most datasets, detailed data and metadata are readily available in various formats, and statistics are disseminated according to a pre-announced calendar. Unpublished nonconfidential data are generally made available to users upon request. Publications and web pages provide contact information and a list of publications and other services. The AFPC disseminates GFS data in accordance with the advance release calendar included in the DSBB of the IMF. However, methodological notes on GFS are not disseminated. Users noted that detailed data are scarce and that published GFS are excessively concise.

9. Section II of this report provides a summary assessment of data quality by agency and dataset. This is followed by staff recommendations in Section III. The authorities’ response to this report and a volume of detailed assessments are presented in separate documents.

II. Assessment by Agency and Dataset

10. Assessments of the quality of four macroeconomic datasets—national accounts, government finance, monetary, and balance of payments statistics—were conducted using the Data Quality Assessment Framework (DQAF July 2003). In this section, the results are presented at the level of the DQAF elements and using a four-point rating scale (Table 1). Assessments of the prerequisites of data quality and the assurances of integrity (Dimensions “0” and “1” of the DQAF) are presented in Tables 2a-c. For each dataset, the assessment of methodological soundness, accuracy and reliability, serviceability, and accessibility (Dimensions “2” to “5” of the DQAF) are shown in Tables 3a-d.

Table 1.

Colombia: Data Quality Assessment Framework July 2003—Summary Results

Key to symbols: O = Practice Observed; LO = Practice Largely Observed; LNO =Practice Largely Not Observed; NO = Practice Not Observed; NA = Not Applicable

article image
Practice observed: current practices generally in observance meet or achieve the objectives of DQAF internationally accepted statistical practices without any significant deficiencies.Practice largely observed: some departures, but these are not seen as sufficient to raise doubts about the authorities’ ability to observe the DQAF practices. Practice largely not observed: significant departures and the authorities will need to take significant action to achieve observance. Practice not observed: most DQAF practices are not met. Not applicable: used only exceptionally when statistical practices do not apply to a country’s circumstances.
Table 2a.

Colombia: Assessment of Data Quality—Bank of the Republic (BR)

article image
Table 2b.

Colombia: Assessment of Data Quality—Ministry of Finance and Public Credit (MFPC)

article image
Table 2c.

Colombia: Assessment of Data Quality—National Administrative Department of Statistics (NADS)

article image
Table 3a.

Colombia: Assessment of Data Quality—National Accounts

article image
Table 3b.

Colombia: Assessment of Data Quality—Government Finance Statistics

article image
Table 3c.

Colombia: Assessment of Data Quality—Monetary Statistics

article image
Table 3d.

Colombia: Assessment of Data Quality—Balance of Payments

article image

11. To inform its assessment, the mission conducted a user survey and held a meeting with users in order to elicit their views on the macroeconomic statistics. The user survey revealed that the price and the monetary statistics are rated highly. Users viewed the methodological basis and the reliability of the macroeconomic data as appropriate, most notably in the case of the monetary and BOP statistics. Users pointed to insufficient detail and dissagregation in the presentation of the fiscal statistics. A large majority of users indicated dissatisfaction with the timeliness of the national accounts, the fiscal statistics, and the BOP; they were unaware of publication of release calendars for these data. The majority of users were satisfied with accessibility to data, most notably monetary, price, and balance of payments statistics. However, users saw scope for improved access to fiscal metadata.

12. In a follow-up meeting with the mission, users noted that efficient operation of the domestic financial markets requires timely dissemination of key data, with advanced release calendars strictly observed. They noted that long delays in the publication of GDP data created uncertainty in financial markets. Despite observing SDDS prescriptions, the timeliness of quarterly GDP and BOP data was seen as insufficient to meet user needs. Some expressed a desire for advance calendars to include the specific time of the day of release. On the trade-off between timeliness and robustness, users indicated a strong preference for earlier release of preliminary data. Also, more methodological information on the national accounts was needed, particularly on indicators of industrial activity and financial services. This could help better understand differences among a number of indicators of industrial activity. There was demand for a monthly index of economic activity. Users of fiscal statistics were unaware of advance release calendars, and noted that time series for key fiscal aggregates were unavailable, detail was insufficient, and there was a dearth of methodological information. This was problematic as various GFS compilers release different data for the same variable.

III. Staffs Recommendations

13. Staffs recommendations are aimed at redressing shortcomings identified in the assessments, which are based on the review of statistical practices, discussions with the data producing agencies, and input from users. Recommendations propose ways to further increase observance of international standards. Their implementation would help enhance the analytical usefulness of Colombia’s statistics. Apart from the high priority recommendations presented below, the companion Detailed Assessments volume presents additional recommendation for most datasets in the final section of the respective chapter.

Cross-cutting Recommendations

  • Review the legal framework for statistical activity, including to strengthen the institutional and financial independence of the NADS. This will require (i) appointing its head for a period overlapping that of the government; (ii) making its budget subject of direct approval by Congress in the context of a multi-year plan for statistical improvement; (iii) enabling it to apply sanctions on non-reporting government entities; and, (iv) increasing its budget.

  • Strengthen inter-agency collaboration, including through activation of the National Commission of Statistical Information envisaged by decree 1820 of 1990. Raising the effectiveness of this commission would require limiting its membership to the entities that effectively carry the heavier responsibilities in the production of macroeconomic statistics: the NADS, the BR, the MFPC, and the CO.

  • Within the national information policy and the government’s connectivity agenda, launch a robust effort to strengthen statistical culture in central government ministries, local governments and enterprises with support from the highest political level. This could be done through a decree requiring statistical compilers within the public sector to adhere to compilation and reporting standards to be set by the National Commission of Statistical Information.

  • Adhere to a policy whereby only the responsible institution (that is a member of the National Commission of Statistical Information) may release official data.

  • Expand revision studies to include more analysis on the direction and magnitude of revisions as well as long-term revision patterns. Make these studies available to the public and inform the public of the revision cycle.

  • Specify the time of the day of publication in the advance release calendars.

  • Summary methodologies should be posted on the DSBB and periodically updated.

National Accounts

  • Publish preliminary estimates of the national accounts for 2003 without further delay.

  • In collaboration with the NTCD, prepare a directory of establishments to provide an adequate basis for sample surveys.

  • Elaborate a data model to guide the building of a data warehouse aimed at improving the consistency of source data at the micro level that supports compilation of the national accounts as well as the balance of payments and fiscal statistics.

  • Press ahead with work to change the base year to 2000, preferably by end-2005. Looking ahead, consider a one-off shortening of the cycle for changing the base year in order to take advantage of the wealth of data stemming from the results of the ongoing comprehensive censuses that will become available in the first half of 2006.

  • Improve the treatment of the economic activity generated by the large illicit sector, moving beyond agricultural output.

  • Measure GDP volume changes using annual chain indices; the measuring of government transactions should be made on an accrual basis.

  • Reduce the publication lag of the preliminary estimate of quarterly GDP from 90 to 45 days, which would need to be reflected in the advance release calendar.

  • Maintain an external relations policy aimed at promptly dispelling doubts, misinterpretation, and questions about the compilation of the national accounts (including the GDP of the manufacturing sector).

  • Observe the (advanced) release calendar for annual GDP.

Government Finance statistics

  • Strengthen the authority of the MFPC/AFPC to collect information from any public sector institution. In addition, all proposals for legal action with a possible impact on the reporting, storing, or dissemination of fiscal data must be reviewed by the AFPC, with its opinion to be taken into account at the highest ministerial level.

  • Make the MFPC/AFPC responsible for undertaking systematic reconciliation of fiscal aggregates (including public investment) produced by the different institutions involved (GAO, CO, BR, NADS, NPD, NTCD, GDPC, as well as other internal units of the MFPC). This will require a greater resource endowment.

  • The AFPC should compile data on the financing of each of the NFPS entities, which will require requesting from reporting agencies data on below-the-line operations. The BR should disaggregate net credit to the public sector by individual entity. These actions by the AFPC and the BR will permit better validation of data on the fiscal outturn.

  • The MFPC should develop the functional classification of the nonfinancial public sector expenditure.

  • Monitor user needs, with efforts directed at ensuring that published data meet demands for more detail, longer time series, and methodologies information.

  • Increase efforts to make users aware of data dissemination practices. In particular, explain that differences between the datasets used for policymaking (AFPC) and auditing and control (GAO, CO). Also, inform the public that selected officials in charge of formulating macroeconomic policy, policymakers, and the IMF receive fiscal data in advance of public release. Intensify efforts at publicizing the advance release calendar, including by hyper linking the MFPC’s website to the SDDS.

  • Provide more computing resources to improve the compilation of data in AFPC, which should support efforts undertaken by the MFPC’s Fiscal Support Directorate, FOSIT, and GAO.

Monetary statistics

  • Publish monetary aggregates excluding the deposits of the central government, as recommended by the MFSM.

  • Press ahead with work underway to improve sectorization, including positions with nonresidents, now available in the call report form 338. Looking to the medium-term, consider revising the chart of accounts to ensure that reporting from banks meets statistical needs in a comprehensive manner, thereby reducing the need for supplementary information.

  • Incorporate accrued interest into all underlying financial instruments and sectors.

  • Introduce formal procedures for consultations with users, such as user surveys and periodic meetings.

Balance of Payments

  • In consultation with the NADS, develop, without delay, a methodology to include unrecorded trade in the balance of payments on a timely basis. A statistical technique would need to be designed to estimate illegal exports and counterpart transactions.

  • Reclassify FDI intercompany debt transactions from other investment to FDI. Based on the 10 percent ownership criteria of FDI, reclassify some FDI equity transactions to portfolio equity. Adjust the related investment income series accordingly.

  • Continue efforts to identify government bonds issued abroad and purchased by residents as well as those purchased by residents on the secondary market.

APPENDIX I

Table 1.

Colombia: Dissemination Practices Compared to the SDDS Coverage, Periodicity, and Timeliness of Data

article image
Note: Periodicity and timeliness: (D) daily; (W) weekly or with a lag of no more than one week from the reference data or the closing of the reference week; (M) monthly or with a lag of no more than one month; (Q) quarterly or with a lag of no more than one quarter; (A) annually; and (...) not applicable. Italics indicate encouraged categories.
1

As noted in the companion document “Response by the Authorities” (p. 2), the estimates were published in February 2006.