Building Statistical Capacity in Fragile and Conflict-Affected States

Based on internal data, this paper finds that the capacity development program of the IMF’s Statistics Department has prioritized technical assistance and training to fragile and conflict-affected states. These interventions have yielded only slightly weaker results in fragile states than in other states. However, capacity development is constantly needed to make up for the dissipation of progress resulting from insufficient resources that fragile and conflict-affected states allocate to the statistical function, inadequate inter-agency coordination, and the pervasive impact of shocks exogenous to the statistical system. Greater coordination with other capacity development providers and within the IMF can help partially overcome low absorptive capacity in fragile states. Statistical capacity development is more effective when it is tailored to countries’ level of fragility.

Abstract

Based on internal data, this paper finds that the capacity development program of the IMF’s Statistics Department has prioritized technical assistance and training to fragile and conflict-affected states. These interventions have yielded only slightly weaker results in fragile states than in other states. However, capacity development is constantly needed to make up for the dissipation of progress resulting from insufficient resources that fragile and conflict-affected states allocate to the statistical function, inadequate inter-agency coordination, and the pervasive impact of shocks exogenous to the statistical system. Greater coordination with other capacity development providers and within the IMF can help partially overcome low absorptive capacity in fragile states. Statistical capacity development is more effective when it is tailored to countries’ level of fragility.

I. Introduction

Fragile and conflict-affected states (FCS) face formidable challenges for economic growth and development. According to the IMF’s definition, currently 42 countries are FCS, equivalent to around 20 percent of the Fund’s membership. The Managing Director’s Global Policy Agenda aims to enhance the IMF’s engagement to FCS. Fund capacity development (CD) to FCS has been a priority since 2014 and has accelerated under the Fund’s 2018 CD Strategy. IMF CD plays a key role in building strong institutions that can help execute macro-economic policies that promote growth and help prevent fragility. Statistical CD is important since policy makers need more and better data to craft macroeconomic policies, plus data quality challenges tend to be more acute in FCS. There is also a need for more and timely macroeconomic data for the public, capital markets, and the international community. The annual number of CD missions in statistics to FCS increased by 30 percent between fiscal year (FY) 2016 and FY2019, despite the challenges of working in FCS, including weak absorptive capacity, poor governance, limited resources, and conflict.

The 2018 IMF Independent Evaluation Office’s (IEO) evaluation of the Fund’s engagement with FCS recommended to take practical steps to increase CD impact, including on statistics. In response, the Fund’s Management Implementation Plan proposed a paper with analysis of the experience in providing statistical CD in FCS to identify lessons for making CD more impactful. This paper examines the characteristics of FCS affecting CD in statistics and highlights challenges common in FCS. It also scans recent CD developments, including how COVID-19 has impacted the delivery and effectiveness of engagement. The paper assesses CD impact in FCS and any differences with non-FCS. It attempts to identify factors impacting CD delivery and highlights policies that boost CD effectiveness.

Compared to non-FCS, the study finds that FCS have lower statistical capacity and face greater constraints especially with regards to resources and ability to establish internal coordination processes necessary for data compilation and dissemination. They also confront higher risks, such as uncertain political support to implement CD recommendations. Data quality issues undermine Fund surveillance of FCS more acutely than in non-FCS. Results-based management (RBM) ratings for CD are only slightly weaker for FCS than for non-FCS, indicating that efforts to tailor CD objectives and outcomes to the FCS context are generally successful. Since FY2016, the Statistics Department’s (STA) CD delivery to FCS had risen significantly, but the onset of the COVID-19 crisis in FY2020 weakened demand. Thanks to shorter, ad hoc interventions to address urgent issues identified by the statistical compilers, the share of FCS missions increased in STA’s total CD delivery in FY2021 despite IT and connectivity constraints.

When reviewing implementation of CD recommendations, the case studies and recent CD delivery to FCS confirm the importance of adopting a differentiated approach to FCS and the need for close coordination with other CD providers such as the World Bank. Case studies and analytical findings also highlight the importance of CD integration with Fund-supported programs and the need for longer-term CD engagement tailored to a country’s absorptive capacity and to changing circumstances in recipient countries. Close engagement with Fund country teams and inter-departmental coordination are key to ensure that objectives are prioritized in view of scarce absorptive capacity. Big Data is increasingly used as a non-traditional data source to help fill gaps in official source data, often caused by a persistent lack of coordination among institutions. The experience suggests that virtual delivery has the potential for improving CD to FCS. Not having to worry about travel costs opens opportunities to deliver shorter, well-targeted interventions that meet very specific demands and are better suited for countries with scarce absorptive capacity if internet connectivity and IT problems are resolved.

This paper has ten sections. After the introduction, the second section presents the statistical characteristics of FCS, and the third section is on data issues, including data quality. The fourth section highlights challenges in the absorption of CD and risks in FCS. The fifth section analyzes the overall delivery of CD to FCS, including by region and statistical topic. The sixth section examines CD delivery through the regional and thematic multi-partner vehicles and tailoring of CD. The seventh section assesses the impact of STA’s CD in FCS as compared to non-FCS through RBM data. The eighth section presents an empirical analysis of factors that impact statistical CD missions, such as the level of fragility and statistical capacity. The ninth section highlights strategic policy findings from the five country case studies to recommend policies that enhance CD effectiveness. The concluding section extracts the main lessons for making CD more impactful. The annexes present the details of the five country cases in Haiti, Djibouti, Kosovo, Madagascar, and Myanmar.

II. Statistical Characteristics of FCS

The IMF classifies as FCS countries affected by institutional fragility and/or violent conflict. Currently, weak institutional capacity is measured by a three-year average rating of 3.2 or less using the World Bank’s Country Policy and Institutional Assessment (CPIA) score, while the presence of a UN or regional peacebuilding operation is used as a proxy for conflict. Forty-two countries, equivalent to around 20 percent of the IMF’s membership, are currently classified as FCS, with nearly half concentrated in Africa and a quarter in the Middle East and Central Asia. In terms of income level, 35 (83 percent) are low income or lower middle-income countries (Table 1). Fifteen (36 percent) are assessed by the Fund as a ‘high risk location’ (HRL), a designation that constrains the delivery of CD (Figure 1). About a quarter of FCS are also considered small developing states or small states, which face unique vulnerabilities due to their small size, limited economic diversification and effects of climate change. 2

FCS are a heterogenous group with varying sources and degrees of fragility, but they share common characteristics. Many suffer from weak administrative and institutional capacity, leading to the disruption or failure by the state to deliver public services such as education, healthcare, energy, and security. Deep governance issues are more prevalent in FCS who tend to have less favorable macroeconomic outcomes, including lower economic growth rates, larger government debt, higher inflation rates and greater dependence on official development assistance (IEO, 2018). Political instability and conflict are other common characteristic of FCS along with vulnerability and exposure to natural disasters, all of which entail great economic and human costs. A country’s fragility status is also not permanent and may change over time.

Table 1.

FCS Overview

article image
Sources: World Bank; IMF Financial Data Query ToolNotes: 42 Total FCS; 36% HRL; 52% low income; 31% lower middle income; 17% upper middle income; 24% small states; 31% e-GDDS with NSDP; 29% with current program (ECF); 55% with RCF/RFI.

ECF: Extended Credit Facility; EFF: Extended Fund Facility; SA: Standby Arrangement

RCF: Rapid Credit Facility; RFI: Rapid Financing Instrument

World Bank methodology assessment of statistical capacity (scale 0 – 100)

Figure 1.
Figure 1.

High Risk Locations

(Percent of time as an HRL between 2016–2020)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source Authors’estimates based on IMF data

The World Bank’s classification of FCS attempts to capture the differentiated nature of fragility and conflict. It is based on methodologies that distinguish countries according to the following categories: (i) countries with high levels of institutional and social fragility, based on indicators that measure the quality of policy and institutions, plus specific manifestations of fragility; (ii) countries affected by violent conflict, based on a threshold number of conflict-related deaths relative to the population. This latter category differentiates two more subcategories founded on the intensity of violence: (a) countries in high-intensity conflict and (b) countries in medium-intensity conflict.

III. Data Issues

Not surprisingly, data quality challenges are more acute in FCS. Using the World Bank’s statistical methodology score, which assesses a country’s capacity to adhere to international statistical standards, the average is lower in FCS than non-FCS, for all regions (Figures 2a, b).3 Between 2008 and 2020, the overall statistical score for FCS has improved by about nine percent, more than in non-FCS, and despite difficulties with data compilation due to COVID-19 lockdowns in 2020. Zooming into the different areas of economic and financial statistics, scores for FCS are the weakest in import/export price indices, national accounts, the consumer price index (CPI), and government finance statistics. These weaknesses are also reflected in the low score on participation in the IMF’s Data Standards Initiatives (Box 1, Tables 2a, b).

Figure 2a.
Figure 2a.

World Bank Methodology Assessment of Statistical Capacity for FCS and Non-FCS Countries

(Statistical Capacity Indicator Score)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source: World Bank Statistical Capacity Indicators.
Figure 2b.
Figure 2b.

World Bank Methodology Assessment of Statistical Capacity for FCS and Non-FCS Countries, by Region

(Statistical Capacity Indicator Score)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source: World Bank Statistical Capacity Indicators.

Another indicator for data problems is the IMF country teams’ assessment of data adequacy for surveillance included in staff reports. The highest data adequacy rating ‘A’ signals that data is adequate for surveillance, followed by rating ‘B’ if data has some shortcomings but is broadly adequate for surveillance, and then rating ‘C’ if data has serious shortcomings which significantly hamper surveillance. Of the 22 member countries whose data are rated ‘C,’ 16 are FCS, equivalent to 73 percent. Twenty-five FCS are rated ‘B,’ and one is rated ‘A’.

Table 2a.

Breakdown of Statistical Methodology Assessment

article image
Source: World Bank Statistical Capacity Indicators.
Table 2b.

Statistical Methodology Assessment Criteria

article image
Source: World Bank Statistical Capacity Indicators

IMF’s Data Standards Initiatives and FCS

The IMF’s Data Dissemination Initiatives help promote data transparency.4 They set standards for the dissemination of data needed to assess economic conditions and policies. The objective is for the authorities to publish such data in line with a pre-announced release calendar for the benefit of policymakers, market participants, the media, and the public. The standards consist of three ascending tiers: the enhanced General Data Dissemination System (e-GDDS) for countries with relatively weak statistical capacity; the Special Data dissemination Standard (SDDS) for members having or seeking access to capital markets; and the SDDS Plus primarily for systemically important financial systems. Data are published through an official website denominated the National Summary Data Page (NSDP). In Africa such websites use the African Development Bank’s (AfDB) Open Data Platform and provide a one-click process for posting data using the Statistical Data and Metadata eXchange (SDMX) standards. This mechanism has helped resource-strained national statistics agencies to adopt an open data approach and reduce reporting burdens.

Among the 42 FCS, only14 have fully implemented the e-GDDS by publishing data through the NSDP. Their performance in terms of reliable dissemination of data on the NSDP is in line with non-FCS. Twenty-four FCS participate in the e-GDDS without publishing any data through the NSDP (Table 3). Three FCS do not participate in Data Standards Initiatives. Within the FCS, only West Bank and Gaza has subscribed to the SDDS.

Table 3.

Fragile and Conflict-Affected States and IMF’s Data Standards Initiatives

article image

Implementing the recommendations of the e-GDDS improve data transparency and governance, as it requires good inter-agency coordination between the central bank, ministry of finance and national statistics office. In addition to limited staff and IT resources, FCS not implementing the recommendations of the e-GDDS may have confronted challenges to gain political support and to achieve improved inter-agency coordination. In the Middle East and Central Asia (MCD) region, for example, political instability and active conflict make inter-agency coordination among data producers more difficult, which has complicated the implementation of the e-GDDS recommendations. In fact, most of the e-GDDS countries that do not have a NSDP are HRL. Small island FCS in the APD (Asia Pacific Department) region launched NSDPs thanks to intense CD, funded by the Government of Japan. This CD assistance was integrated in the dialogue with country authorities as part of the IMF’s surveillance discussions with member countries.

IV. CD Challenges

A data mining analysis, using STA’s internal scorecards from FY2016–19, provides insights into challenges with absorbing CD.5 A sentiment analysis using a lexicon-based approach was performed on 348 scorecards to detect negative sentiments in the text. A dictionary of lexicons was created manually and applied using an algorithm to the text of the scorecards. The results were classified into three categories to compare absorptive capacity challenges in FCS and non-FCS (Table 4).6

Table 4.

Text Mining of Statistical Scorecards

article image
Source: IMF Staff
Figure 3a.
Figure 3a.

Countries Pacing Resource Challenges

(Percentage of total countries with CD delivery challenges in each category)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source: Authors estimates based on STA scorecards, FY2016-19
Figure 3b.
Figure 3b.

Countries Feeing Source Data Challenges

(Percentage of total countries with CD delivery challenges in each category)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source: Authors estimates based on STA scorecards, FY2016-19
Figure 3c.
Figure 3c.

Countries Facing institutional Cooperation Challenges

(Percentage of total countries with CD delivery challenges in each category)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source: Authors estimates based on STA scorecards, FY2016-19

Overall, terms associated with CD absorption issues appeared more frequently in scorecards for FCS than non-FCS and were even on the rise over time in two of the three examined categories between FY2016–19 (Figures 3a, b, c). In particular, the proportion of FCS facing resource inadequacy was higher and increased by 30 percent. The limitations included under-staffing, funding-related issues, and IT challenges. Source data problems also appeared more frequently in scorecards for FCS. The proportion of FCS facing weak interagency cooperation doubled over this period, but this challenge declined for non-FCS.

These findings were also reflected in analysis of risk ratings in STA’s RBM dataset.7 In four out of the five ratings categories, the risks to a CD project were higher in an FCS than in non-FCS (Figure 4). Resource adequacy risks, which cover the human, financial, infrastructure and technology resources needed to achieve an intended outcome, represented the highest threat to the implementation of CD recommendations. Meanwhile, management/technical staff commitment risks – assessing the authorities’ commitment to undertake CD recommendations – and political support risks – evaluating the likelihood that recommendations could be blocked by high-level decisionmakers or changes in political leadership – were significantly higher in FCS projects. External climate/conditions risk such as global recession, epidemics, internal/external security conditions that might slow or impede CD delivery were also significantly higher in FCS than non-FCS.

Figure 4.
Figure 4.

RIM Risk Ratings by Category, FY2017–21

(Percent)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Results Based Management System

While all of these factors hamper STA’s CD delivery to FCS, there are differences in the resource adequacy risks among institutions involved in the compilation and dissemination of economic and financial statistics. Traditionally the national statistics offices, which are generally in charge of national accounts and price statistics, are most affected by the resource adequacy risks, reflecting low priority for data in a tight fiscal policy framework. Central banks, which are in most countries responsible for financial, monetary, and external sector statistics, have usually a more generous funding envelope and autonomy in allocating resources. Ministries of finance tend to be in-between. This pattern is also reflected in the statistical areas where the World Bank statistical methodology score is weakest, notably national accounts and prices, while external sector statistics have the highest scores.

Regarding the World Bank, the recent decision by donors to allocate a significant amount of resources of the IDA 19 replenishment to data issues might offer opportunities to mitigate some of the resource issues in FCS. This funding to enhance staffing and IT infrastructure could complement ongoing work by the World Bank in the area of surveys, which is coordinated with the IMF’s CD to ensure synergies particularly in supporting the production of real sector statistics (RSS).

V. CD Delivery to FCS

Between FY2016 and FY2019, and before the COVID-19 pandemic, STA CD missions to FCS rose by 30 percent, far outpacing growth in demand from non-FCS where CD missions rose by about 2 percent, even though the growth rate for FCS has been somewhat more volatile (Figures 5a, b). 8 The onset of the COVID-19 crisis in March 2020 led to a reduction in CD missions to FCS and non-FCS in FY2020 as missions were cancelled, reflecting the authorities’ shift in focus to crisis management and the sudden stop of travel. The switch to remote delivery allowed the gradual recovery of demand for CD missions in FY2020. The number and the share of CD missions to FCS even rose in FY2021, but missions were much shorter, as reflected in the declining FCS share of the dollar cost of STA CD (Figure 5c). This change was caused by less demand for time-intensive CD. For instance, multi-year CD programs for the re-basing of national accounts used to last 1–2 weeks prior to the pandemic, covering many issues. Instead, with COVID-19 demand shifted to ad hoc requests to address urgent crisis-related issues. For example, due to lockdowns, on-site data surveys for the CPI became unfeasible, so CD assisted in identifying alternative data sources, including web scraping and Big Data (Box 2).

Figure 5a.
Figure 5a.

STA CD Missions

(Left Number of CO Missions Right Percent of CD Missions to FCS)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source: IMF Travel Information Management System (TIMS), IMF Participant Applicant Tracking System (PATS).
Figure 5b.
Figure 5b.

STA CD Missions Annual Growth Rates

(Annual growth rate percent)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source: IMF Travel Information Management System (TIMS), IMF Participant Applicant Tracking System (PATS).
Figure 5c.
Figure 5c.

STA CD Spending

(Left; Millions of USD; Right; Percent of CD Spending)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source: Estimates based on Analytics Costing & Estimation System (ACES) data.

Looking at trends at the regional level, Africa (AFR) is the largest beneficiary of STA CD to FCS, followed by APD and MCD (Figure 6), where the rising trend was reversed by the crisis. In APD, the number of FCS missions has been relatively stable, probably reflecting less severe connectivity and IT problems. In MCD, the number of missions has been more volatile with the presence of many HRLs. Notably, despite shortcomings in internet connectivity and IT equipment, the shift to remote CD facilitated access to countries classified as HRL, which are mainly in the MCD region. In the past, travel restrictions to HRLs required cumbersome and expensive arrangements to gather officials in third country locations. With remote delivery, the number of CD missions increased quite substantially for the MCD region in FY2021, partly owing to new engagements in Yemen and Sudan (Figure 7). In terms of the distribution of CD delivery by country, Myanmar, West African countries, and Madagascar have been amongst the most intense FCS beneficiaries of CD (Figure 8).

Figure 6.
Figure 6.

Average STA CD Delivery to FCS. by Region

(Percent of STA s total CO missions)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Travel Information Management System (TIMS)

Most statistics CD to FCS is focused on national accounts statistics, followed by government finance statistics (including debt statistics), and external sector statistics (Figure 9). This prioritization in CD demand, led by the country authorities, is also broadly consistent with the main data issues identified in the World Bank’s methodological data scores. Despite the World Bank scores also highlighting lagging participation in the IMF Data Dissemination Initiatives, CD in this area has been limited reflecting the nature of this type of CD intervention, which is usually limited to 1–2 missions and can only advance when the authorities commit resources to meet specific data requirements.

(Number of CO Missions)
Figure 7.

STA CO Missions to Fragile Countries

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Travel Information Management System (TIMS)
Figure 8.
Figure 8.

Global Intensity of STA CD Mission Delivery to FCS

(Based on average number of missions between FY2017–21)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Travel Information Management System (TIMS)
Figure 9.
Figure 9.

STA CD Delivery to FCS by Statistical Topic and Region

(Percent of total missions, FY2016–21)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Travel Information Management System (TIMS)

VI. CD Delivery Through RCDCs and Thematic Trust Funds

Over 90 percent of STA’s CD delivery to FCS is funded by donors compared to 85 percent for non-FCS (Figures 10a, b), reflecting their strong interest in supporting low and lower-middle income countries and a particular emphasis on FCS. Looking at the different means of IMF CD delivery, the Regional Capacity Development Centers (RCDCs) deliver about two thirds of total FCS CD. This share has remained stable throughout the pandemic (Figure 11). RCDCs have a multi-year engagement with countries to improve the methodological foundation for data compilation. Their presence on the ground and proximity to the country is particularly suitable for FCS as they require long-term support to develop statistical capacity. Their continuous engagement helps understand local/regional contexts, including the political economy of data, which is a major risk to implementing CD recommendations in FCS.

Figure 10a.
Figure 10a.

Share of STA CD Missions to FCS, by Funding Source

(Percent of total STA CD missions, FY2019–21)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Travel Information Management System (TIMS)
Figure 10b.
Figure 10b.

Share of STA CO Missions to non-FCS, by Funding Source

(Percent of total CD missions. FY2019–21)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Travel Information Management System (TIMS)

A total of 40 of the 42 FCS are covered by an RCDC (Figures 12a, b) except for Kosovo and Somalia. The larger presence of RCDCs in AFR and APD has resulted in a greater capacity to respond to the large CD needs of FCS in those regions. In contrast, MCD has had only one RCDC predominantly serving FCS, the Middle East Regional Technical Assistance Center (METAC). For HRLs, travel restrictions limit the benefit of CD delivery through METAC (Figure 13). The new Caucasus, Central Asia, and Mongolia Regional Capacity Development Center (CCAMTAC) launched virtually in early 2021, now also benefits Tajikistan in MCD.

Figure 11.
Figure 11.

STA CD to FCS by HQ and RCDC

(Left Number or missions; Right; Percent delivered by RCDC)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Travel Information Management System (TIMS)
Figure 12.
Figure 12.

FCS Membership in RCDC

(Number of countries, FY2021)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Travel Information Management System (TIMS)
Figure 12b.
Figure 12b.

STA CD to FCS, by RCDC

(Percent of total CD missions, FY2016-21)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Travel Information Management System (TIMS)
Figure 13.
Figure 13.

STA HQ and RCDC CD Delivery to FCS

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Travel Information Management System (TIMS).

STA CD delivered through IMF headquarters (HQ) is closely coordinated with and complementary to RCDC CD, usually focusing on more advanced areas of statistics. For example, while RCDCs assist with methodological issues related to external sector statistics, CD by HQ experts supports more advanced topics such as widening statistical coverage by including informal economy estimates, or the effect of digitalization on the services account.

To enhance efficiency and agility of HQ-based CD, STA has transitioned to a structure for donor funding for increased flexibility in allocating resources to countries and topics in statistics. This flexibility is particularly relevant for FCS as they face high risks to CD delivery which requires frequent adjustments to CD plans. In FY2015, when the transition to the new funding structure started, STA had 17 bilateral donor projects and trust funds with often very strict requirements by donors on project objectives and countries to be covered. By FY2022 this number declined to about five, including the two main multi-donor trust funds— the Data for Decisions (D4D) Fund and the Financial Sector Stability Fund (FSSF). These two funds have increasingly financed CD to FCS, as FCS-focused bilateral donor projects have been de-emphasized. When the pandemic hit, donors supporting these funds helpfully agreed to move some funding away from longer-term CD projects towards shorter ad hoc interventions.

The FSSF aims at fostering financial sector stability in low and lower-middle-income countries (LLMICs) and all FCS.9 It is designed to finance CD delivered by the Monetary and Capital Markets Department (MCM) as well as STA to ensure close integration. For example, when MCM plans a Financial Sector Stability Review, STA CD focuses on data gaps hampering analyses of financial vulnerabilities and risks to macroeconomic stability. This allows recipient countries to use scarce absorptive capacity to address the most urgent data gaps. Since the FSSF launch in 2017, 14 FCS have benefitted from FSSF CD, while the decision to widen eligibility to all FCS in 2020 helped further increase the number of FSSF funded technical assistance (TA) missions to FCS (Figure 14).

Figure 14.
Figure 14.

FSSF TA Mission Delivery

(Number of FSSF funded TA mission, FY2018–21)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Travel Information Management System (TIMS).

The D4D Fund prioritizes CD in FCS, mostly in the AFR and APD regions.10 It aims to put more and better data in the hands of decision-makers to support evidence-based macroeconomic policies. Forty-four percent of the D4D’s eligible countries are FCS. Since its launch in 2018, activities have covered 35 of the 42 FCS (Figure 15). In FY2021, officials from 30 FCS participated in the series of COVID-19 webinars on public sector debt and external sector debt statistics funded by the D4D Fund. The D4D Fund’s Financial Access Survey (FAS) database houses data on financial inclusion for 40 of the 42 FCS.

Figure 15.
Figure 15.

Fund CD Beneficiaries

(Percent of total beneficiaries1, FY2019–21)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source: IMF Travel Information Management System (TIMS), IMF Participant Applicant Tracking System (PATS).1 Beneficiaries include TA mission to FCS and tracking participants from FCS

In recent years, short-term experts have delivered almost half of the CD to FCS, followed by the long-term experts stationed in the RCDCs (Figure 16a). Short-term experts are hired by the RCDC and IMF HQ to deliver CD. They are backstopped and supervised by long-term experts or HQ-based staff, which have delivered proportionally more CD to non-FCS (Figure 16b). The use of short-term experts provides flexibility in mobilizing experts at short notice depending on demand. (For example, experts with specialty knowledge, language skills, and experience in specific country groups). However, it entails somewhat less continuity in CD delivery, which might be particularly difficult in the weak and often complex institutional environment of FCS. Still, this is somewhat mitigated by the fact that quite a few short-term experts who undertake CD missions to FCS are repeatedly used and thus are familiar with FCS issues.11 In addition, the STXs receive intense backstopping by staff in HQ and RCDCs, which ensures quality control and continuity in the engagements with FCS.

Figure 16a.
Figure 16a.

STA CD to FCS, by Modality

(Percent of mission delivery days. FY2019–21)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Travel Information Management System (TIMS).
Figure 16b.
Figure 16b.

STA CD to Non-FCS, by Modality

(Percent of mission delivery days. FY2019–21)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Travel Information Management System (TIMS).

STA Training

Training is an essential component of the statistical CD program for FCS, especially during the early phase of recovering from fragility when re-building staff capacity is a priority and a precondition to assimilate future CD. To leverage the curriculum of the traditional face-to-face training courses, STA is developing a series of online courses on the foundations of economic and financial statistics, funded by the D4D Fund. Courses on government finance statistics, public sector debt statistics, balance of payments and international investment position, and national accounts statistics have already been launched and translations into French and Spanish are being phased in. Preparation of courses in price statistics, monetary and financial statistics, and financial soundness indicators are under way.

These online courses have reached a share of FCS participants of over 30 percent for the AFR and MCD regions and about 15 percent in APD (Figure 17a). This is broadly in line with the share in traditional face-to-face courses despite challenges with internet access in many FCS, which illustrated continued strong demand for training (Figure 17b). Virtual training tailored to address challenges to data compilation during the pandemic attracted a share of participants from FCS exceeding traditional face-to-face and online courses in all regions (Figure 17c). This illustrates the potential to ease absorptive capacity constraints by shorter training events, similar to the pattern observed with TA missions to FCS (see Section V).

Figure 17a.
Figure 17a.

Active Participants in STA Online Courses

(Left; Participants; Right Percent from FCS, FY2019-20)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

IMF Participant Applicant Tracking System (PATS)
Figure 17b.
Figure 17b.

Participants of In-Person Trainings on Statistical Topics, by Region

(Left; Participants; Right Percent from FCS, FY2019-20)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

IMF Participant Applicant Tracking System (PATS)
Figure 17c.
Figure 17c.

Participants In STA Virtual Courses

(Left; Participants; Right Percent from FCS, FY2019-20)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

IMF Participant Applicant Tracking System (PATS)

Tailoring CD

As the FCS country group includes different levels of fragility, CD needs to be tailored accordingly. Conflict and political and social disruption can hinder CD provision, which requires long-term engagement and flexibility to adjust work plans when political risks materialize, or conflicts or natural disasters require to return to more basic needs. STA CD delivery has been tailored as follows (Figure 18):

  • During the early phase of fragility when countries recover from conflict, training is a dominant component of CD to raise staff capacity and, consequently, absorptive capacity for future CD. For monetary and financial sector statistics and government finance statistics, which rely on administrative data sources (financial sector balance sheet data collected by the central banks and official government financial data), CD focuses on establishing basic reporting systems. This needs to be closely coordinated with regulatory reforms that establish a legal basis for proper bank reporting and public financial management. CD on national accounts, prices and external sector statistics is more complex as it relies on surveys and other sources, including in the private sector, which is often dominated by informal transactions. Sometimes conflict and fragility put at risk existing data compilation. For instance, remote CD to Yemen helped measure price changes for items whose price collection was spotty during the conflict and deteriorated further with the pandemic. Consequently, the Central Statistics Organization was able to estimate missing prices so that a complete CPI could be released. Such ad hoc interventions sometimes lead to further CD demand, for example, the authorities in Yemen indicated interest in implementing recommendations for further enhancing CPI coverage.

  • As recovery advances, the focus of CD shifts towards widening coverage. For example, government finance statistics would aim to include the most relevant extrabudgetary funds while financial sector statistics would explore options to cover the most macro-critical non-financial institutions. External sector statistics would complement customs data through surveys to estimate the informal sector flows and seek to reduce errors and omissions by adding private sector data sources to estimate the financial accounts. In parallel, CD would assist with enhancing data frequency and timeliness.

  • Small island states often feature a rather persistent lack of capacity to assimilate and retain CD in statistics. A key reason is severe underfunding of the statistical function. Against this backdrop, work plans for the new Pacific Financial Technical Assistance Center (PFTAC) funding cycle—which serves seven mostly small FCS with very low statistical capacity in APD—identifies countries for capacity supplementation. Until basic absorptive capacity has been established, the STA expert is engaged in actual data compilation to ensure production of the most essential data for surveillance. Once staffing levels are sufficient to absorb CD, which also requires strengthening coordination among compiling institutions, the traditional CD approach can be pursued that aims at sustainable outcomes. STA also uses synergies among countries in the region when supplementing capacity, including common tools, methods, and where possible data sources, including Big Data.

Figure 18.
Figure 18.

STA’s Tailoring of CD for Fragile States at Different Levels of Fragility

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Statistical Innovation and Big Data in FCS1

For many FCS reliance on internationally recommended data sources and surveys is challenging, partly owing to weak data collection systems aggravated by a lack of coordination. These problems have been compounded by the COVID-19 crisis. Big Data can help FCS overcome source data challenges by providing alternative data sources. For example, using mobile phone data on travel services is more cost effective than cross-border surveys. Some FCS in Africa are interested in how to use data on mobile money and banking in the compilation of monetary statistics. Below are a few examples of how STA is using Big Data innovation in FCS.

In the Republic of Congo, administrative data has been utilized to develop high frequency indicators (HFIs). STA provided CD in 2019 and 2020 to use sales data from the monthly value added tax database for the compilation of GDP estimates and for the development of HFIs. Tax data offer timely and inexpensive source data for large parts of the formal economy but are often underused because of the absence of sharing agreements with the tax office and difficulties inherent to large datasets. STA assisted the national statistical office in engaging with the tax office and trained the compilers on data management, including for the treatment of data gaps and outliers which are common in large tax records. This was closely coordinated with the Central Africa Technical Assistance Center (AFC) which continued to follow up on the recommendations and assisted with implementation. CD to FCS in Africa also addresses the use of less traditional data sources such as regulatory data, data on electricity generation, and on production by large corporations to track the monthly evolution of economic activity.

Interest in African FCS to use nowcasting techniques to generate more timely data for economic policy making has created new demand for CD in the use of Big Data. For example, STA has been supporting Sierra Leone with the preparation of a database which consolidates administrative data from various sources complemented by Big Data from private sources to develop a set of HFIs. In addition to supporting nowcasting models, this database will serve as source data in addressing persistent data gaps that hampered compilation of traditional statistics.

In West Bank and Gaza property prices obtained from websites, a form of Big Data, were used to develop a residential property price index (RPPI). A remote TA mission funded by the D4D Fund in 2020 assisted the Palestine Monetary Authority with the development of the RPPI. In the absence of suitable source data, the authorities assessed the potential use of both asking prices (collected from two websites where properties are listed for sale) and appraisal prices generated from the process for mortgaging property. They found asking prices to be more suitable. STA helped with the development of the index, which was published in February 2021.2

STA has experimented with a method to use Big Data on vessel traffic (developed by Arslanalp, Marini, and Tumbarello (2019)) to produce trade estimates in real time for Malta.3 The methodology was extended by Arslanalp, Koepke and Verschuur (2021) and applied to eleven Pacific Island countries (about half of which are FCS) in view of lags in official statistics. This methodology can provide early warning signs of turning points in economic activity, helping fill gaps in official data. STA also produced a user guide to enable Fund economists to implement this methodology in other countries, including FCS, to help produce high frequency trade indicators of economic activity.

1 Unlike statistical data that are compiled for specific purposes, Big Data is a byproduct produced in business and administrative systems, social networks, and the internet of things.2 See https://www.pma.ps/en/Media/Press-Releases/announcing-the-findings-of-the-palestine-monetary-authoritys-residential-property-price-index-4th-quarter-2020.3 Automatic Identification System (AIS) vessel traffic data was used from marine traffic. AIS is an ‘air traffic control system’ for ships and allows for real-time tracking of commercial vessels

VII. Monitoring Targeted Results

Data from the RBM framework was used to monitor targeted CD results, as a way to measure CD impact.12 A strategic management tool for designing CD interventions and measuring impact, RBM is based on the concept of the logical framework or log frame. RBM assumes that a causal framework links CD activities to desired outcomes, and ultimately to better policies in the beneficiary country. There is a distinction between outputs (what the IMF is responsible for—TA reports, training, missions, etc.), outcomes, and benefits to the recipient country.

Objectives define high-level goals of CD, and are not directly observable (Figure 19). Thus, in order to measure a country’s progress towards reaching objectives, outcomes are used as measurable steps forward in CD achieved when recommendations developed jointly with country authorities are implemented. Indicators specify how the achievement of outcomes will be validated. Milestones are smaller, time-bound steps used to working towards an outcome. Objectives, outcomes and milestones are defined in the IMF RBM catalog. RBM milestone and outcome ratings are examined by the project manager, with both rated on a scale of 1 to 4 as follows: 1 – not achieved, 2 – partially achieved, 3 – largely achieved, and 4 – fully achieved.

Figure 19.
Figure 19.

RBM Logframe

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source IMF Staff

The RBM analysis suggested that FCS had the lowest milestone ratings in prices, then government finance statistics and the Data Standards Initiatives, then national accounts, followed by balance of payments, and financial institutions ratings were the highest rated (Figures 20a, b).13 This fits with the observation that national statistics offices and ministries of finance tend to be more resource constrained and have lower capacity than central banks producing balance of payments and financial sector data.

Figure 20a.
Figure 20a.

FCS Average Milestone Rating by Region and Topic, FY2016–211

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source: IMF Results Based Management Report.1 RBM Milestone data for EUR and WHD are not shown since there is only one FCS in each these regions and individual is confidential.
Figure 20b.
Figure 20b.

Non-FCS Average Milestone Rating by Region and Topic, FY2016–211

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source: IMF Results Based Management Report.1 RBM Milestone data for EUR and WHD are not shown since there is only one FCS in each these regions and individual RBM data is confidential.

FCS had a marginally lower overall average milestone rating at 3.1, compared to non-FCS at 3.2. This was due to FCS having lower milestone ratings in balance of payments, national accounts and prices, and Data Standards Initiatives. This may indicate that national statistical agencies in FCS are more under-resourced than those in non-FCS, similar to the findings in Section IV on the text mining analysis of the statistical scorecards. FCS’ lower rating in Data Standards Initiatives relative to non-FCS is due to only 14 FCS having fully implemented e-GDDS (see Box 1). However, FCS on average had similar milestone ratings for financial institutions and government finance statistics. By region, FCS in APD had notably lower milestone ratings than non-FCS. This is probably related to the presence of seven FCS in the Pacific Islands that are also mostly small states and characterized by low absorptive capacity.14

FCS performance also tends to lag slightly behind non-FCS when examining broader outcome results. The average outcome rating for FCS was 2.9, while for non-FCS it was 3.0. When examining the distribution of the outcome ratings with 1 and 2 ratings classified as unachieved, and 3 and 4 ratings classified as achieved, FCS had more unachieved outcomes than non-FCS, and less achieved outcomes than non-FCS (Figure 21). The marginal differences in ratings between FCS and non-FCS might also reflect STA’s effort to take into consideration constraints from fragility when designing milestones and determining the timeframe for achieving outcomes. Similarly, analysis of the impact of STA trainings in FY2019–21 reveals that FCS participants on average showed somewhat lower learning gains than non-FCS participants: when comparing average post-training quiz results of participants with pre-training quiz results, FCS participants’ learning gains were 18 percentage points while non-FCS’ learning gains were 21 percentage points.

Figure 21.
Figure 21.

Frequency Distribution of STA Target Outcome Achievements

(Percent of completed protects from May 2015)

Citation: IMF Working Papers 2022, 045; 10.5089/9798400202742.001.A001

Source: IMF Results Based Management Report.Note: Projects designated as successful were rated by staff as having fully or largely met targeted outcomes.

VIII. Determinants of STA CD Mission Delivery

To gain a deeper understanding of the relationship between STA CD mission delivery and statistical capacity, cross-country regressions were conducted using STA mission data as the dependent variable from 142 countries from 2016 to 2020 (Table 5).15 The cross-country regressions address the question whether there are more CD missions to countries with lower statistical capacity and if there are significant differences between FCS and non-FCS.16 The regressions employ other control variables, such as RBM milestone ratings, population, presence of an IMF program, and security risk—factors that are hypothesized to also affect the number of CD missions. This study employs country averages—instead of using a panel data to investigate within-country effects—to examine CD delivery variation across countries. The regression results report Hubert-White robust standard errors, to correct for the heteroskedasticity revealed using the Breusch Pagan tests.

Table 5.

Determinants of Statistical CD Missions Dependent Variable: Total CD Missions (FY2016–2020)1

article image
Sources: IMF Travel Information Management System (TIMS); World Bank Statistical Capacity Indicators; World Bank World Development Indicators; IMF Monitoring of Fund Arrangements (MONA).Note: Robust standard errors in parenthesis. *** p<0.01, ** p<0.05, * p<0.1. Study period is IMF Fiscal year 2016–20.

Total Missions are measured as the total number of missions in a country during the study period which runs from May 1st until April 30th of the following year. Data is from the STA CD Mission database.

The methodology category of the World Bank Statistical Capacity Indicator is used to represent average statistical capacity of a country in terms of methodology and data dissemination. Average values from 2016–20 are used for each country.

The milestone rating of capacity development missions is deemed as the capacity of the authorities to implement the milestones set for a capacity development mission. Average values from the study period are used for each country. Data is from the STA CD Mission database.

Fragility represents the proportion of time a country is considered an FCS between the study period.

High-risk location represents the proportion of time a country is considered a high security during the study period.

Log population is the 2016–2020 average log population from the World Development Indicators, to measure the systemic nature of the country.

IMF Program is a dummy variable representing the proportion of time a country is under an IMF program between 2016–2020 using the IMF Monitoring of Fund Arrangements (MONA) database.

The summary statistics (Table 6) suggest that countries in the sample received an average of 17 CD missions between FY2016–2020, an average methodology capacity score of 54.2 from a scale of 0–100, and an average milestone rating of 3.7 out of 4. There are large cross-country variations in the sample as suggested by the minimum and maximum values of the variables. Further, the summary statistics reveal differences in countries’ fragility, security risk, and participation in an IMF program.

The data has some limitations, including that milestone ratings are not recorded for all missions. The study attempted to overcome this issue by taking the average reported milestones in a country during the study period. While analyzing aggregate data is useful in providing an overview of CD delivery, it masks potential differences in CD delivery across statistical sectors, i.e., national accounts, prices, etc. This empirical analysis focuses on aggregate measures of CD delivery because there were not enough data on a sectoral/topical basis to provide meaningful results. Furthermore, while regression analysis using outcome ratings could provide further insights, data for STA outcome ratings were insufficient to provide an econometric estimation.

The empirical results in Table 4 highlight four main findings:

  • 1) Countries with lower statistical capacity and milestone ratings receive more STA CD missions. The analysis shows that the number of CD missions is negatively correlated with statistical capacity and milestone ratings.

  • 2) Fragility is negatively correlated with the number of CD missions, but the relationship is not significant. (columns 2–4).

  • 3) High risk locations are associated with a lower number of CD missions (column 4).

  • 4) Countries with an IMF program and large populations are associated with higher CD missions and the correlation is statistically significant (columns 2–4).

The first finding suggests that STA CD missions are being deployed to countries with higher statistical needs as reflected by their lower statistical capacity and lower milestone ratings. The second and third findings suggest that CD missions are not being significantly affected by country fragility, though negative secuirty factors related to conflict can weigh on the number of missions. The last finding indicates that STA CD is well integrated with IMF program activities.

Table 6.

Summary Statistics

article image
Sources: IMF Travel Information Management System (TIMS); World Bank Statistical Capacity Indicators; World Bank World Development Indicators; IMF Monitoring of Fund Arrangements (MONA) Note: Study period is IMF Fiscal year 2016–20.

Total Missions are measured as the total number of missions in a country during the study period which runs from May 1st until April 30th of the following year. Data is from the STA CD Mission database.

The methodology category of the World Bank Statistical Capacity Indicator is used to represent average statistical capacity of a country in terms of methodology and data dissemination. Average values from 2016–20 are used for each country.

The milestone rating of capacity development missions is deemed as the capacity of the authorities to implement the milestones set for a capacity development mission. Average values from the study period are used for each country. Data is from the STA CD Mission database.

Fragility represents the proportion of time a country is considered an FCS between the study period.

High-risk location represents the proportion of time a country is considered a high security during the study period.

Log population is the 2016–2020 average log population from the World Development Indicators, to measure the systemic nature of the country.

IMF Program is a dummy variable representing the proportion of time a country is under an IMF program between 2016–2020 using the IMF Monitoring of Fund Arrangements (MONA) database.

IX. FCS Case Studies: Strategic Policy Findings

The case studies (presented in the Annexes) yield lessons that can be applied to improve the impact of CD in statistics. Five FCS were selected for the case studies, one per region (Table 7). Djibouti is a small state, facing unique vulnerabilities with acute capacity constraints. Haiti is a medium-intensity conflict fragile state, and an HRL, while Kosovo is the only FCS in Europe (EUR). Madagascar is one of the FCS in AFR that has received substantial CD that has been effective in supporting surveillance and lending. Myanmar was selected because it was the highest intensity user of Fund CD until early 2021 and is classified as a medium-intensity conflict FCS and an HRL.17 Other selection criteria include:

  • Djibouti, Madagascar, and Haiti have country engagement strategies (CES), an MIP initiative for FCS whereby IMF country teams write a paper that integrates the roles of policy advice, financial support, and capacity building.

  • Haiti, Kosovo, and Madagascar have had Fund-supported programs with conditionality over the last six years, whereas Djibouti and Myanmar have benefited from Fund financing instruments for the provision of emergency assistance such as Rapid Credit Facilities (RCF) or a Rapid Financing Instrument (RFI), which have limited conditionality.

  • One of the five countries, Kosovo, is an upper middle-income country, while Djibouti, Myanmar and Haiti are lower middle-income, and Madagascar is low income, which is representative of the fragile states’ group.

Table 7.

Case Studies Overview and Selection Criteria

article image
Sources: World Bank Statistical Capacity Indicators; IMF Financial Query ToolNotes: The median methodology score for fragile states is 45.; The history of lending arrangements is as of August 31, 2020.

RCF: Rapid Credit Facility; ECF: Extended Credit Facility; SA: Standby Arrangement; RFI: Rapid Financing Instrument

The Country Engagement Strategy (CES) is an MIP initiative for FCS whereby country teams write a memo that integrates the role of policy advice, financial support and capacity building. It benefits from inputs from diverse stakeholders, external development partners, and the country authorities.

The case studies highlight the importance of coordination with other CD providers, especially the World Bank. In Djibouti, coordination with the World Bank supported a major overhaul in national accounts and international trade statistics. In Madagascar, the launch of the NSDP in 2019 was undertaken with support from the AfDB, while coordination with the World Bank supported the release of revised national accounts in 2019. In Haiti, coordination with the World Bank supported the compilation and release of the new GDP series in 2020. In Kosovo, close coordination with development partners helped STA focus on improving the government finance statistics, while Eurostat provided TA on real sector statistics, and the Swedish development agency focused on CPI/PPI. In Myanmar, the ADB has focused on real sector statistics which allowed STA to focus on other sectors, particularly government finance and external sector statistics. Kosovo and Myanmar also highlight the importance of encouraging coordination amongst official data compilers.

In most of the case studies, the effectiveness of statistical CD has been improved by integrating it with Fund-supported programs, financing instruments for the provision of emergency assistance, and surveillance work. In Myanmar, the blend RCF/RFI has helped with the publication of government finance statistics data and improved coverage of the foreign direct investment (FDI) data. In Madagascar, the 2016–20 Extended Credit Facility (ECF) conditionality supported the new statistics law adopted in 2018, the new law on institutional arrangements for compilation of government finance statistics in 2019, and the publication of rebased national accounts in 2019. In Haiti, re-engagement through Article IV surveillance and the RCF helped support the resumption of the publication of real, monetary, and fiscal data in early 2020, the release of new rebased GDP in 2020, and the production of new monetary data. In Djibouti, the integration of STA CD with the RCF has helped improve the consistency of external sector statistics.

Is the CES useful for statistical capacity building? It was in the case of Haiti, as the CES helped to prioritize CD provided in 2020 on data compilation and reporting in national accounts and monetary statistics and facilitated the integration of CD with surveillance.18 Similarly, for Djibouti the CES has prioritized data reporting, with priority in CD on external accounts and in the medium term, in other statistical areas, including the Data Standards Initiatives. On the other hand, for Madagascar, the CES did not define a role for future statistical CD when it recommended a successor ECF arrangement, though it did acknowledge that past statistical CD had been useful in national accounts and prices.

Complementing CD delivered by short-term experts and long-term experts based in the region, usually in RCDCs, with HQ-delivered CD has worked well in all country cases. For example, through AFS the long-term expert has provided CD on real sector statistics in Madagascar, while short-term experts have provided TA on prices, and HQ-based CD has assisted with government finance, external sector, and monetary statistics. In Djibouti, CD on real sector has been provided by the long-term expert in METAC while short-term experts and HQ staff have provided CD on fiscal, monetary, and external sector statistics. Kosovo is not a member of an RCDC but did benefit from regional long-term experts in national accounts (2012–14) and in government finance statistics (2016–19), complemented by CD support through short-term experts and headquarter staff.

Remote CD has been effective during the COVID-19 crisis, though it has limitations. Due to the virtual format, interest in CD has diminished in some cases, such as in Djibouti. There are also connectivity and IT challenges, time-zone difference constraints, and shorter attention spans by the CD recipients. On the positive side, the remote format bypasses security concerns that come with on-site missions, such as in Haiti. Here national accounts and monetary missions in 2020 were effective in helping with the compilation of new GDP and new monetary statistics. Virtual missions also facilitate area department participation in STA missions, thus enhancing CD integration with Fund surveillance and program work. Virtual meetings also facilitate continuous follow up by the CD provider. In Madagascar, the authorities consider virtual CD to be effective, despite its challenges. Virtual missions are more effective when provided by an expert who already knows local officials. This has been the case in Kosovo where government finance statistics CD has been provided by an experienced short-term expert who had been a long-term expert in the region. In Myanmar, the effectiveness of virtual missions by the long-term experts in the Capacity Development Office in Thailand (CDOT) was facilitated by prior relationship with the authorities.

Anchored by a regional expert, long-term tailored CD engagements tend to be effective. Initially, in FCS with lower capacity, CD tends to be more hands-on, with training used to develop staff capacity, as was the case in Kosovo. Long-term experts provided more intense training and TA on statistical compilation at the start, and as statistics improved, CD became more focused on government finance statistics, and from 2019, only short-term experts and HQ staff provide CD. Another example is Djibouti, a small state where STA has resorted to continuous hands-on engagements, including remote support in between TA activities, and complementing TA with hands-on training. Following CD delivered between 2015 and 2020, the central bank started reporting regularly improved monetary and financial data. In Haiti TA complemented by training for the national statistics office allowed the release of rebased GDP in 2020. This contrasts with CD in 2006 that attempted to help develop high frequency indicators, which at the time was beyond absorptive capacity.

Affected by low absorptive capacity, FCS require more time to build capacity. STA has been engaged over many years in the FCS in the case studies, but progress has been subject to setbacks, especially in conflict FCS, owing to political and social disruption. For instance, recent events in Myanmar have jeopardized statistical capacity building progress,19 while in Haiti, political instability, lack of policy ownership together with environmental fragility pose challenges for retaining the knowledge transferred through CD.

Interdepartmental Fund coordination on CD can facilitate greater impact of CD in statistics. In Myanmar, STA long-term experts in CDOT have coordinated effectively with the Fiscal Affairs Department (FAD) long-term expert and the MCM resident advisor to improve balance of payments statistics, public sector external debt, and fiscal statistics. In Djibouti, STA participated in the Financial Sector Stability Review led by MCM, with improvements to financial soundness indicators’ data quality. STA also delivered a TA mission in 2018 which improved the compilation of monetary and financial statistics.

For French-speaking FCS, there is a need for CD to tackle language barriers, especially for training. Djibouti officials would benefit from attending French courses at the African Training Institute (ATI). Workshop attendance by Haitian officials has been constrained because most regional events in the Caribbean are conducted in English, though some officials have attended French training at the ATI. These countries should have more flexible access to trainings and seminars in other RCDCs in French, especially as this is facilitated by the greater usage of remote or virtual modalities.

X. Lessons and Conclusion

FCS face severe challenges in improving their statistical systems. On average, World Bank statistical methodology scores are lower for FCS, with data shortcomings hindering the adequacy of their official data for Fund surveillance. STA CD has worked to respond to the basic statistical needs of FCS, with most CD missions delivered in national accounts and government finance statistics, broadly consistent with the weakest main topical areas identified by the World Bank statistical methodology scores. While CD in statistics to FCS had been trending upwards until the onset of the COVID-19 crisis, progress had been constrained by inadequate resources that FCS allocate to the statistical function, insufficient inter-agency coordination, and a dearth of source data.

Risk ratings from the RBM framework also point to FCS’s insufficient resources as a principal risk to achieving CD outcomes, followed by a weak commitment to implement CD recommendations, both at the technical and senior management level. While the achievement of targeted results of STA CD in FCS have been broadly in line with non-FCS, reflecting STA efforts to tailor RBM milestones and outcomes to FCS’s circumstances, the analysis yielded moderately lower RBM ratings for FCS in balance of payments, national accounts, prices, and IMF Data Standards Initiatives. The fact that only 14 FCS have implemented the recommendations of the e-GDDS illustrates the difficulties faced by compiling agencies to mount a well-coordinated effort to achieve reliably regular data dissemination.

STA’s greater reliance on multi-donor trust funds has provided more flexibility in allocating resources, which is often needed in high-risk FCS environments. While overall funding risks for multi-donor trusts funds might come to be affected by fiscal challenges in advanced countries, the main risk to STA CD stems from funding problems in RCDCs. With two thirds of STA CD to FCS delivered through RCDCs, it will be important to work with donors to maintain and, if possible, increase resources for the RCDCs for FCS.

CD missions have been well targeted. Countries with lower statistical capacity and milestone ratings have received more CD missions. Fragility is not a significant driver of the number of CD missions, though being a HRL can weigh somewhat on the amount of STA missions. Countries with an IMF lending program are associated with higher CD missions, confirming that STA CD is well integrated with program activities.

Despite the difficulties that FCS face with the absorption of CD, the overall statistical capacity of FCS has increased since 2008, as reflected in the increase in the average World Bank’s statistical methodology score. This has coincided with better targeting of STA’s CD delivery as it has become more flexible, regionally focused, demand driven, and results-based with the evolution of the RBM framework. In addition, RBM data suggests that majority of STA projects in FCS have achieved their TA outcomes.

Given the weak absorption capacity of FCS, STA has tailored its CD delivery and engaged more through the virtual modality. For example, in the Pacific Island fragile states served by PFTAC, with low absorptive capacity, STA is initially focusing on capacity supplementation, to ensure basic data compilation. STA utilizes its RCDC long-term experts that have a strong knowledge of FCS in their region and focus on providing longer-term statistical CD in methodology issues. The share of FCS in STA CD increased during the pandemic thanks to a shift to shorter virtual ad hoc interventions. Virtual CD missions have been effective in expanding engagement with FCS, especially when there are security concerns.

Some of the main policy recommendations arising from the delivery of STA CD to FCS relate to the need to tailor the type of CD and its modalities, and to improve coordination. These can form the basis for medium-term STA CD strategy for a post-pandemic world.

  • CD should be tailored to where an FCS lies along the fragility spectrum. The more a FCS is subject to shocks and the lower its capacity and resources, the more statistical CD to FCS will need to be tailored, long-term, and focused on hands-on capacity building. CD in fragile states in conflict and with low absorptive capacity will tend to need CD focused on basic data compilation or establishing basic reporting systems.

  • Greater use of more long-term experts stationed in the RCDCs appears warranted. To strengthen long-term engagement and continuity, experienced short-term experts should be selected and training on FCS issues should be provided to new ones.

  • Training to FCS should be an integral part of the CD, especially during the early phase of fragility, as it is critical to build staff capacity. This includes not only face-to-face training, but also online and virtual trainings. The D4D-funded online training has become a powerful tool to leverage the traditional face-to-face training, with 30 percent of participants from FCS in AFR and MCD. The share of FCS participants has been even higher in short virtual training events on specific topics for data compilation during the pandemic.

  • Ongoing efforts to advance blended learning should build on the training experience by designing a curriculum with short courses aimed at highly relevant subjects, such as fiscal consolidation. By integrating video material from the online courses with interactive sessions focused on hands-on data workshops, the efficiency of training should further improve to meet the needs of FCS.

  • Remote delivery of CD should continue even when travel resume. As indicated in the case studies, several FCS appreciated the new remote delivery of CD that addressed specific challenges in a flexible manner. Especially for HRLs, this shift opened a more efficient way to deliver CD, which should complement the traditional approach. Also, remote delivery has allowed country authorities and CD experts to stay engaged regarding the implementation of CD recommendations.

  • Advancing FCS participation in the IMF’s Data Standards should be a key objective. To focus on the FCS that have not yet implemented the e-GDDS—not yet publishing essential data for surveillance—STA launched in 2020 a Japan-funded project to improve data dissemination globally, including in 22 eligible FCS. Weak cooperation among data-producing agencies in FCS hampers regular dissemination even after compilation has been achieved. The implementation of e-GDDS with the establishment of an NSDP enhances cooperation between the central bank, ministry of finance and national statistics office, thereby freeing up scarce capacity in FCS and supporting efficiency gains.

  • Coordination with the World Bank and other CD providers, and within the Fund would help address low absorptive capacity. The IDA 19’s allocation of significant funds to improving statistics can help address the resource scarcity at the heart of subpar data in FCS. Holding regular meetings between the heads of international organizations to exchange information on CD provision could enhance CD cooperation. Pressing ahead with the integration of statistical CD with Fund-supported programs and surveillance can improve statistics in FCS. Better coordination among compiling agencies can remove obstacles for sharing source data and support regular dissemination for the IMF’s Data Standards Initiatives. CES have the potential to help further integrate CD with surveillance.

STA’s experience with improving CD delivery to FCS while securing flexible donor support has positioned STA well for the future. It will be important, however, to leverage the experience from remote delivery during the pandemic to sustain the benefits from shorter interventions focused on urgent demands, integrating traditional face-to-face with remote and online delivery and more continuous engagements.

Annexes on FCS Case Studies

I. Case Study on Djibouti1

article image
article image
article image

II. Case Study on Haiti2

article image
article image
article image
article image

III. Case Study on Kosovo3

article image
article image
article image
article image

IV. Case Study on Madagascar4

article image
article image
article image
article image

V. Case Study on Myanmar5

article image
article image
article image
article image
article image

References

  • Aav, E., Alves, M. A., Boukezia, R., De Santis, Y., Fisher, L., Grote, M., Helis, J., Suljagic, I. J., Koukpaizan, V. P., Luca, O., Mesa-Puyo, D., Murara, L., Okello, A. K., Pedastsaar, E., Petit, P., Schlotterbeck, S., Seiwald, J., Shah, A., Sparkman, R. and Taiclet, B. (2017). “Building Fiscal Capacity in Fragile States – Case Studies”, Fiscal Affairs Department Paper No. SM/17/90, International Monetary Fund, Washington, DC.

    • Search Google Scholar
    • Export Citation
  • African Development Bank Group – Statistics Department (ECST). (2017). “Economic Benefits of Open Data in Africa”, Abidjan, March.

    • Search Google Scholar
    • Export Citation
  • Arslanalp, S., Koepke, R. and Verschuur, J. (2021). “Tracking Trade from Space: An Application to Pacific Island Countries”, Working Paper No. 21/225, International Monetary Fund, Washington, DC. DOI: https://0-doi-org.library.svsu.edu/10.5089/9781513593531.001

    • Search Google Scholar
    • Export Citation
  • Arslanalp, S., Marini, M. and Tumbarello, P. (2019). “Big Data on Vessel Traffic: Nowcasting Trade Flows in Real Time”, Working Paper No. 19/275, International Monetary Fund, Washington, DC.

    • Search Google Scholar
    • Export Citation
  • Bassanetti, A. (2021). IMF Working Paper: “When does Capacity Development Achieve Good Outcomes? Evidence from the IMF Results-Based Management Data”, International Monetary Fund, Washington, DC.

    • Search Google Scholar
    • Export Citation
  • Chami, R., Espanizo, R. and Montiel, P. (2021). “Macroeconomic Policy in Fragile States”, Oxford: Oxford University Press.

  • Gupta, S., Keen, M., Baer, K., Barrand, P., Boukezia, R., Dybczak, K., Fisher, L., Fouad, M., Hansen, T., Hurcan, Y., Mansour, M., Murara, L., Okello, A., Olden, B., Pecho, M., Pessoa, M., Renteria, C. and others. (2017). “IMF Policy Paper: Building Fiscal Capacity in Fragile States”, International Monetary Fund. Washington, DC, June.

    • Search Google Scholar
    • Export Citation
  • Hammer, C., Kostroch, D., Quiros, G. and STA Internal Group. (2017). “IMF Staff Discussion Note – Big Data: Potential, Challenges, and Statistical Implications”, International Monetary Fund, Washington, DC, September.