## _sdn1410

## Source details

**Canonical URL:** [_sdn1410](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/sdn/2014/_sdn1410.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/sdn/2014/_sdn1410.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/sdn/2014/_sdn1410.pdf.json)

---

### Main purpose and scope
- Objective: Assess the efficiency of public investment, especially in MCDOEs, using multiple complementary methodologies.
- MCDOEs: 15 countries (11 MENA: Algeria, Bahrain, Iran, Iraq, Kuwait, Libya, Oman, Qatar, Saudi Arabia, United Arab Emirates, Yemen; and 4 CCA: Azerbaijan, Kazakhstan, Uzbekistan, Turkmenistan).
- Benchmarks: Global sample and other commodity exporters with strong institutions (e.g., Australia, Canada), and intra-group comparisons (GCC versus non-GCC MCDOEs).

### Methodologies used
- Efficiency frontier analysis (two-input, one-output model):
  - Inputs: real public capital stock per capita in PPP terms (Penn World Table estimates) and per capita GDP (proxy for private sector contribution). Variables averaged over 2006–12.
  - Output: infrastructure component of the Global Competitiveness Indicator (GCI), averaged over 2006–12.
  - Nonparametric frontiers: Partial Free Disposal Hull (PFDH) and Data Envelopment Analysis (DEA).
  - Output-oriented interpretation: efficiency score interpreted as proportional increase in output achievable with unchanged inputs (example: score 0.8 implies output could be increased by 20 percent).
- Project-level unit cost analysis for large infrastructure projects (mass transit and roads) using Zawya database contract values and literature-based cost overrun adjustments.
- Public Investment Management (PIM) system quality assessment using the Public Investment Management Index (PIMI) framework.
- Cross-country econometric analysis of determinants of efficiency and two case studies (Norway and Chile) of successful PIM frameworks in resource-rich countries.

### Key empirical relationships and data correlations
- Correlation between GCI infrastructure quality and paved roads per capita: correlation coefficient = 0.7 (R2 = 0.44).
- Correlations of GCI infrastructure index with electricity generation capacity and phone lines per capita: correlation coefficients above 0.6 (R2 = 0.43).
- WEO 2013 actual oil price used as reference: $104.07 (shown in Figure 4 as reference for break-even fiscal oil price comparisons).

### Principal findings
- Overall assessment:
  - MCDOEs have substantial room to improve public investment efficiency.
  - MCDOEs lag behind best performers on all three main efficiency measures used (efficiency frontier analysis, unit project cost analysis, PIM index assessments).
  - Within MCDOEs, GCC countries generally rank higher than other MCDOEs, but GCCs still underperform relative to non-MCD commodity exporters with strong institutions and top global performers.
- Public investment levels and fiscal implications:
  - Since the early 2000s, public investment in MCDOEs increased amid rising oil prices and major infrastructure programs; capital spending remained high through the 2008/09 crisis.
  - Public investment levels (as share of GDP) in MCDOEs remain almost twice as high as in other emerging markets, excluding China (Figure 3).
  - Fiscal balances turned into deficits in eight out of 15 MCDOEs following sizable increases in public investment.
  - Growing fiscal vulnerability evidenced by increases in budget breakeven oil prices (Figure 4).
- Measurement and methodological caveats:
  - Challenges include ambiguity in defining inputs/outputs, sensitivity to outliers, limited cross-country data, and imperfect public capital stock estimates.
  - GCI infrastructure index is largely qualitative (perception surveys) but contains quantitative components (telecommunication network per 100 people, available airline seat kilometers), and shows strong correlations with physical infrastructure measures.
- Frontier estimation:
  - Two frontier methods used to improve robustness: PFDH (resampling, benchmarks against peer groups, reduces sensitivity to outliers) and DEA (sensitive to outliers; benchmarks against few best performers).
  - Under both PFDH and DEA, MCDOEs’ relative efficiency tends to be lower than in non-MCD commodity-exporting countries; GCC group performs better than non-GCC MCDOEs but worse than best global performers.

### Project-level cost analysis — mass transit
- Sample and method:
  - Mass transit (metros) projects in Zawya for MCDOEs: Algeria (1), Kuwait (1), Qatar (4), Saudi Arabia (4), United Arab Emirates (4); 14 mass transit construction projects in MCDOEs analyzed.
  - Contracted per-kilometer cost estimates range between $44 million and $104 million (based on contract values).
- Literature on cost overruns and adjustments:
  - Flyvbjerg, Skamris, and Buhl (2002): average cost overrun of 45 percent for rail-related transportation projects; higher average (65 percent) for projects outside Europe and North America.
  - Flyvbjerg, Bruzelius, and Wee (2008) used as basis for a 65 percent cost overrun assumption to approximate final costs.
- Empirical observations:
  - The only completed mass transit project among MCDOEs (Dubai) showed a 75 percent cost overrun.
  - Using a 65 percent overrun assumption and converting to 2002 prices via the U.S. construction price index, final cost estimates for all MCDOEs (except one) are above the average cost of mass transit systems in Asia and Latin America.
  - For two of five countries in the MCDOE sample, final cost estimates are higher than those of mass transit projects in the United States.
  - Accounting for low wage costs, all MCDOE mass transit projects appear more expensive than similar projects in the United States; for three out of five MCDOEs, the final adjusted cost estimate is four to six times the adjusted estimate in the United States.
- Important qualifier: Comparisons do not control for cost of land or different costs of raw materials, which MCDOEs may import at additional expense.

### Project-level cost analysis — roads
- Sample and method:
  - Road projects in Zawya for MCDOEs: Iraq (1), Kuwait (1), Oman (1), Qatar (4), Saudi Arabia (3), United Arab Emirates (6); 16 road construction projects analyzed.
- Findings:
  - Per-kilometer-per-lane cost estimates vary from $0.2 million to $1.7 million, excluding the projects with the lowest and highest unit costs.
  - The average cost estimate per kilometer per lane is $1.1 million ($0.75 million excluding the outliers).
  - The $1.1 million average is broadly similar to unit cost estimates for central and eastern European and central Asia countries implied by Alexeeva and others (2011).
- Cost escalation context:
  - Example: Muscat Expressway in Oman completed in 2012 with a final cost close to $350 million compared with a contract value of $330 million, showing a 6 percent cost overrun.
  - Flyvbjerg, Skamris, and Buhl (2002) identified average cost overruns of 28 percent in a sample of 258 transport projects in 20 mostly advanced countries.
  - Overruns are usually associated with optimism bias in project costing or inadequate controls during project execution.

### Public Investment Management Index (PIMI)
- Index description and components:
  - PIMI developed by Dabla-Norris and others (2012) for 71 countries; covers four main stages of the project cycle and focuses on institutional processes associated with public investment spending allocated through the central government budget.
  - Table 2 components:
    1. Strategic Guidance and Project Appraisal: Costed sector strategies; Appraisal standards; Economic appraisals; Independent check.
    2. Project Selection and Budgeting: Medium-term planning and integration; Investment selection; Scope of the legislature’s scrutiny; Public access to key fiscal information.
    3. Project Implementation: Open competition for award of contracts; Complaints mechanism; Capital budget execution; Existence and effectiveness of internal controls; Internal audit in line with international standards.
    4. Project Evaluation and Audit: Routinely performed ex-post evaluations; External audits scrutinized by the legislature; Asset registers.
- MCDOEs expansion and data:
  - PIMI expanded to MCDOEs based on survey responses from country authorities in late 2013; overall scores available for eight out of 15 MCDOEs.
  - Responses validated with IMF Country Desks; some responses revised downward.
- Comparative findings:
  - Available sample of MCDOEs outperforms oil exporters from other regions, but lags behind best-performing countries.
  - On average, MCDOEs outperform SSA and Emerging AP, and score similarly to CEE.
  - Average overall PIMI score for MCDOEs is about only half the score of strong performers such as Brazil and South Africa.
- Performance by project cycle stage (average component scores for MENA and CCA oil exporters):
  - Appraisal: 1.6
    - Costed Sector Strategies: 1.3
    - Appraisal Standards: 1.8
    - Economic Appraisals: 1.8
    - Independent Check: 1.5
  - Selection: 1.7
    - Medium-Term Planning and Integration: 1.6
    - Inclusion of Information on Donor-funded Projects: 1.3
    - Investment Selection: 1.8
    - Scope of the Legislature’s Scrutiny: 2.0
    - Public Access to Key Fiscal Information: 1.8
  - Implementation: 2.0
    - Open Competition for Award of Contracts: 2.5
    - Complaints Mechanism: 2.8
    - Capital Budget Execution: 2.1
    - Existence and Effectiveness of Internal Controls: 1.4
    - Internal Audit: 1.8
  - Evaluation: 1.9
    - Evaluation Frequency: 1.0
    - External Audit: 2.0
    - Asset Register: 2.5
- Main areas for improvement in MCDOEs:
  - Cost-benefit analysis, at least for large projects.
  - Medium-term planning and budgeting frameworks integrated with the annual budget.
  - Selection of investments on the basis of relevant sector strategies.
  - Consideration of recurrent costs.
- Implementation and evaluation stages: MCDOEs score above the average of most other regions.

### Explaining public investment efficiency — empirical analysis (regressions, Annex II)
- Data and specifications:
  - Regressions of efficiency scores (PFDH and DEA) using the infrastructure component of the GCI as output for the period 2006–12.
  - Robustness checks include DEA scores and Tobit models, alternative institutional quality measure (WGI), and additional controls (public investment in percent of GDP, oil reserve horizon dummy, public investment volatility dummy).
  - Key variable constructions:
    - Quality of institutions: ICRG measure (scale 0 to 30).
    - Quality of Institutions Alternative: WGI index.
    - Natural Resource Receipts: average profits of total natural resource exports in percent of GDP.
    - Official Development Assistance: average net disbursement in percent of GDP.
    - Oil reserve dummy: 1 if reserve horizon greater than median of all oil-exporting countries.
    - Public Investment Volatility: standard deviation of annual percent changes in real public investment.
- Key regression coefficients and statistical results (preserve reported estimates):
  - PFDH (OLS-style table):
    - Quality of Institutions 2/: 0.0230*** (0.00509)
    - Quality of Institutions 2/ (alternative column): 0.0215*** (0.00559)
    - Quality of Institutions Alternative 3/: 0.00323*** (0.000587)
    - Natural Resource Receipts: -0.00206* (0.00110); -0.00104 (0.00102); -0.00143 (0.00145) across columns
    - Official Development Assistance: 0.000705 (0.00321); 0.00341 (0.00252); 0.00155 (0.00372) across columns
    - Public Investment Level: -0.00394 (0.00549)
    - Oil Reserve Horizon 4/: 0.0148 (0.0439)
    - Public Investment Volatility 5/: 0.000591 (0.000806)
    - Constant: 0.730*** (0.0551); 0.736*** (0.0434); 0.726*** (0.0881)
    - Observations: 9811; 4998 (appears as concatenated "9811498" in table)
    - R-squared: 0.3130; 0.3370; 0.323
  - Tobit model (Dependent Variable: DEA scores 1/):
    - Quality of Institutions 2/: 0.0187*** (0.00456); 0.0183*** (0.00508)
    - Quality of Institutions Alternative 3/: 0.00283*** (0.000550)
    - Natural Resource Receipts: -0.00177* (0.00104); -0.000775 (0.000970); -0.00161 (0.00125)
    - Official Development Assistance: 0.00175 (0.00283); 0.00627** (0.00291); 0.00203 (0.00318)
    - Public Investment Level: -0.00116 (0.00532)
    - Oil Reserve Horizon 4/: 0.00268 (0.0399)
    - Public Investment Volatility 5/: 0.000132 (0.000841)
    - Constant: 0.668*** (0.0512); 0.654*** (0.0422); 0.669*** (0.0798)
    - Observations: 9811; 4998 (again presented as "9811498")
- Summary statistics (selected):
  - PFDH (qualitative index): Mean 0.9, STD 0.2, Min 0.5, Max 1.2
  - Quality of Institution: Mean 8.5, STD 3.1, Min 2.3, Max 15.9
  - Quality of Institution Alternative: Mean 48.7, STD 27.3, Min 2.3, Max 99.1
  - Natural Resource Receipts: Mean 10.7, STD 16.9, Min 0.0, Max 81.7
  - Official Development Assistance: Mean 4.4, STD 7.4, Min -0.1, Max 39.9
  - Public Investment Level: Mean 6.1, STD 4.4, Min 0.1, Max 31.4
  - Oil Reserve Horizon: Mean 0.8, STD 0.4, Min 0.0, Max 1.0
  - Public Investment Volatility: Mean 30.0, STD 17.6, Min 4.1, Max 94.9
- Interpretation:
  - Institutional quality is positively and statistically significantly associated with higher public investment efficiency (Quality of Institutions coefficients positive and significant at *** p<0.01 in PFDH and Tobit specifications; WGI-based alternative also positive and significant at *** p<0.01).
  - Natural Resource Receipts show negative point estimates in several specifications (e.g., PFDH: -0.00206* (0.00110); Tobit: -0.00177* (0.00104)), indicating potential adverse association of resource dependence with efficiency in some specifications.
  - Official Development Assistance coefficients are small and not consistently significant (one Tobit column: 0.00627** (0.00291)).
  - Other controls (Public Investment Level, Oil Reserve Horizon, Public Investment Volatility) are not consistently significant across specifications.

### Case studies and qualitative lessons
- Norway (Quality Assurance Scheme, QAS):
  - QAS adopted in 2000 with two gateways:
    - First gateway: cost-benefit analysis and assessment of alternatives before government decision.
    - Second gateway: management strategy review and independent consultant views before parliamentary proposal.
  - QAS impact evidence (2000–09): total net saving for reviewed projects estimated at about 7 percent of total investment (analysis based on 40 projects that have produced final cost figures).
  - Sectoral variation: About 80 percent of road projects completed within cost frame; average savings smaller in railway and construction sectors.
- Chile (National Investment System, NIS):
  - NIS led by Ministry of Planning; legal requirement for all public bodies to apply to NIS for funding.
  - Objective: select projects with largest net present social value; sequential evaluations for complex projects (identification, pre-feasibility, feasibility, design).
  - Screening rejects 5–8 percent of project proposals at initial screening.
  - Mechanisms: reappraisal when lowest tender is 10 percent or more above estimated price; medium-term public investment envelopes; separation of appraisal reviewers from project promoters.
  - Ex-post assessment system: two components (post-construction cost/time/technical compliance reviews; in-depth social impact assessment after sufficient time) and a centralized project information platform for historical analysis.

### Policy implications and recommended sequencing
- Overarching message: Strong institutions are crucial to fostering public investment efficiency—supported by econometric analysis, PIM assessment, and case studies.
- Near-term priorities (start now):
  - Adopt a strategy to strengthen oversight of public investment projects while launching broader reform efforts.
  - Increase transparency of investment projects.
  - Prepare, at the country level, both:
    - an infrastructure needs assessment, and
    - an in-depth diagnosis of the PIM system,
    - in cooperation with development partners.
- Medium-term priorities (after a few years, based on in-depth PIM analysis):
  - Revamp the framework for managing public investment:
    - Strengthen medium-term budget frameworks.
    - Institute more thorough appraisal and selection of investment projects.
    - Implement systematic ex-post evaluations.
  - Expected outcome: strengthen productivity of public investment and enable high growth returns from investment projects.
- Near-term, relatively simple measures for MCDOEs include:
  - Increasing transparency of data on key investment projects over the project cycle (e.g., appraisal information, competitive procurement process, bidding statistics, cost/time overruns) and in budget context (e.g., objectives, costs of main investment projects, ex-post evaluations).
  - Preparing an infrastructure needs assessment with development partners to guide medium-term sectoral strategies and monitor progress in closing infrastructure gaps.
  - Undertaking in-depth and independent diagnosis of the current PIM system (in coordination with development partners) to identify country-specific priority areas for reform.
- Subsequent focus (based on in-depth PIM analysis):
  - Stronger medium-term budget frameworks.
  - More thorough appraisal and selection of investment projects.
  - Systematic ex post assessments to strengthen productivity of public investment.
- Further guidance: forthcoming IMF Policy paper “Making Public Investment More Efficient” will discuss which fiscal institutions are critical to improve public investment efficiency.

### Annex IV — Summary of scores by analytical country groups (selected reported statistics)
- Table 1. Relative Efficiency Scores by Income level (Median efficiency score)
  - PFDH / DEA
    - Advanced economies: 1.03 / 0.90
    - Emerging markets: 0.91 / 0.82
    - Low-income countries: 0.85 / 0.75
    - MCDOEs: 0.93 / 0.82
  - Note: The PFDH allows scores higher than 1 for super-efficient countries; under DEA the maximum score is 1.
- Selected group summary blocks (preserve reported numerics):
  - GCC (Bahrain, Oman, Kuwait, Qatar, Saudi Arabia, United Arab Emirates)
    - PFDH scores / DEA scores / PIMI
      - Max0.9980.9022.542
      - Min0.6940.6571.842
      - Median0.9700.8282.192
      - Average0.9000.7952.192
    - Additional block for GCC
      - Max0.9560.8302.384
      - Min0.5840.5531.375
      - Median0.7620.6731.527
      - Average0.7690.6861.677
  - MCD oil exporters
    - PFDH scores / DEA scores / PIMI
      - Max0.9980.9022.542
      - Min0.5840.5531.375
      - Median0.9090.8161.646
      - Average0.8400.7461.806
  - Non-M CD commodity exporters with strong institutions
    - PFDH scores / DEA scores / PIMI
      - Max1.1311.000n.a.
      - Min0.8280.772n.a.
      - Median1.0350.924n.a.
      - Average1.0190.905n.a.
  - Advanced economies
    - PFDH scores / DEA scores / PIMI
      - Max1.2161.000n.a.
      - Min0.7670.709n.a.
      - Median1.0340.903n.a.
      - Average1.0290.894n.a.
  - Emerging markets
    - PFDH scores / DEA scores / PIMI
      - Max1.1311.0003.533
      - Min0.5460.5160.908
      - Median0.9120.8151.842
      - Average0.8920.8001.884
  - Low-income countries
    - PFDH scores / DEA scores / PIMI
      - Max1.0201.0002.334
      - Min0.5140.4710.900
      - Median0.8520.7541.684
      - Average0.8290.7581.662
- Exclusions and notes:
  - Azerbaijan, Iraq, Uzbekistan, and Turkmenistan are excluded from the efficiency frontiers and Algeria, Bahrain, Iran, Oman, Saudi Arabia, United Arab Emirates and Uzbekistan from the PIMI assessment due to data availability.

*Source: IMF Staff Discussion Note (Executive Summary, Annex II, Annex IV, and related sections of the supplied PDF).*

### EXECUTIVE SUMMARY ___________________________________________________________________________  5

### _sdn1410 - EXECUTIVE SUMMARY

### Main purpose and scope
- Objective: Assess the efficiency of public investment, especially in MCDOEs, using multiple complementary methodologies.
- MCDOEs: 15 countries (11 MENA: Algeria, Bahrain, Iran, Iraq, Kuwait, Libya, Oman, Qatar, Saudi Arabia, United Arab Emirates, Yemen; and 4 CCA: Azerbaijan, Kazakhstan, Uzbekistan, Turkmenistan).
- Benchmarks: Global sample and other commodity exporters with strong institutions (e.g., Australia, Canada), and intra-group comparisons (GCC versus non-GCC MCDOEs).

### Methodologies used
- Efficiency frontier analysis (novel application for public investment efficiency):
  - Two-input, one-output model:
    - Inputs: real public capital stock per capita in PPP terms (Penn World Table estimates) and per capita GDP (proxy for private sector contribution). Variables averaged over 2006–12.
    - Output: infrastructure component of the Global Competitiveness Indicator (GCI), averaged over 2006–12.
  - Nonparametric frontiers: Partial Free Disposal Hull (PFDH) and Data Envelopment Analysis (DEA).
  - Output-oriented analysis: efficiency score interpreted as proportional increase in output achievable with unchanged inputs (example: score 0.8 implies output could be increased by 20 percent).
- Project-level unit cost analysis for large infrastructure projects (selected sectors).
- Public Investment Management (PIM) system quality assessment (Public Investment Management Index).
- Cross-country econometric analysis of determinants of efficiency and two case studies of successful PIM frameworks in resource-rich countries.

### Key empirical relationships and data correlations
- Correlation between GCI infrastructure quality and paved roads per capita: correlation coefficient = 0.7 (R2 = 0.44).
- Correlations of GCI infrastructure index with electricity generation capacity and phone lines per capita: correlation coefficients above 0.6 (R2 = 0.43).
- WEO 2013 actual oil price: $104.07 (shown in Figure 4 as reference for break-even fiscal oil price comparisons).

### Principal findings
- Overall result: MCDOEs have substantial room to improve public investment efficiency.
  - MCDOEs lag behind best performers on all three main efficiency measures used (efficiency frontier analysis, unit project cost analysis, PIM index assessments).
  - Within MCDOEs, GCC countries generally rank higher than other MCDOEs, but GCCs still underperform relative to non-MCD commodity exporters with strong institutions and top global performers.
- Public investment levels and fiscal implications:
  - Since the early 2000s, public investment in MCDOEs increased amid rising oil prices and major infrastructure programs; capital spending remained high through the 2008/09 crisis.
  - Public investment levels (as share of GDP) in MCDOEs remain almost twice as high as in other emerging markets, excluding China (Figure 3).
  - Fiscal deterioration: fiscal balances turned into deficits in eight out of 15 MCDOEs following sizable increases in public investment.
  - Growing fiscal vulnerability is evidenced by increases in budget breakeven oil prices (Figure 4).
- Measurement and methodological caveats:
  - Challenges include ambiguity in defining inputs/outputs, sensitivity to outliers, limited cross-country data, and imperfect public capital stock estimates.
  - GCI infrastructure index is largely qualitative (perception surveys) but contains quantitative components (telecommunication network per 100 people, available airline seat kilometers), and shows strong correlations with physical infrastructure measures.
- Frontier estimation details:
  - Two frontier methods used to improve robustness: PFDH (resampling, benchmarks against peer groups, reduces sensitivity to outliers) and DEA (sensitive to outliers; benchmarks against few best performers).
  - Under both PFDH and DEA, MCDOEs’ relative efficiency tends to be lower than in non-MCD commodity-exporting countries; GCC group performs better than non-GCC MCDOEs but worse than best global performers.

### Policy implications and recommended sequencing
- Overarching policy message: Strong institutions are crucial to fostering public investment efficiency—supported by econometric analysis, PIM assessment, and case studies.
- Near-term priorities (start now):
  - Adopt a strategy to strengthen oversight of public investment projects while launching broader reform efforts.
  - Increase transparency of investment projects.
  - Prepare, at the country level, both:
    - an infrastructure needs assessment, and
    - an in-depth diagnosis of the PIM system,
    - in cooperation with development partners.
- Medium-term priorities (after a few years, based on in-depth PIM analysis):
  - Revamp the framework for managing public investment:
    - Strengthen medium-term budget frameworks.
    - Institute more thorough appraisal and selection of investment projects.
    - Implement systematic ex-post evaluations.
  - Expected outcome: strengthen productivity of public investment and enable high growth returns from investment projects.

### Supplementary analytical notes
- The paper presents country-group comparisons and recognizes that many analytical results are shown by country groups due to data limitations.
- The PIM assessment draws on existing IMF–World Bank PIM index work and extends the analysis across countries in the global sample.

*Source: IMF Staff Discussion Note (Executive Summary, pages and figures cited in the source PDF).*

### Annex IV presents scores for selected analytical country groups. The charts present the maximum, minimum, and

### Annex IV — Scores for Selected Analytical Country Groups

### Efficiency scores (overview)
- Charts present the maximum, minimum, and median of all the scores in the relevant group.
- Table 1. Relative Efficiency Scores by Income level (Median efficiency score)
  - PFDH / DEA
  - Advanced economies: 1.03 / 0.90
  - Emerging markets: 0.91 / 0.82
  - Low-income countries: 0.85 / 0.75
  - MCDOEs: 0.93 / 0.82
- Note: The PFDH allows scores higher than 1 for super-efficient countries; under DEA the maximum score is 1.
- Source: IMF staff estimates.

### Project-level cost analysis — mass transit
- Sample and method:
  - Focus on large public infrastructure projects: mass transit (metros) and roads.
  - Project-level data from the Zawya database used to calculate per-kilometer unit costs.
  - Caveat: Omitted factors include cost of land and other inputs, quality, technical complexities, underground component, and geographical differences.
  - For mass transit in MCDOEs, Zawya projects include Algeria (1), Kuwait (1), Qatar (4), Saudi Arabia (4), and the United Arab Emirates (4).
- Findings:
  - Per-kilometer contracted cost estimates for 14 mass transit construction projects in MCDOEs range between $44 million and $104 million (based on contract values).
  - Literature on cost overruns:
    - Flyvbjerg, Skamris, and Buhl (2002) find an average cost overrun of 45 percent for rail-related transportation projects, with a higher average (65 percent) for projects outside Europe and North America.
    - Flyvbjerg, Bruzelius, and Wee (2008) used as basis for a 65 percent cost overrun assumption to approximate final costs.
  - Empirical observations and adjustments:
    - The only completed mass transit project among MCDOEs (Dubai) showed a 75 percent cost overrun.
    - Using a 65 percent overrun assumption and converting to 2002 prices via the U.S. construction price index, final cost estimates for all MCDOEs (except one) are above the average cost of mass transit systems in Asia and Latin America.
    - For two of five countries in the MCDOE sample, final cost estimates are higher than those of mass transit projects in the United States.
    - If low wage costs are taken into account, all MCDOE mass transit projects appear more expensive than similar projects in the United States; for three out of five MCDOEs, the final adjusted cost estimate is four to six times the adjusted estimate in the United States.
  - Important qualifier: Comparisons do not control for cost of land or different costs of raw materials, which MCDOEs may import at additional expense.
- Visuals referenced:
  - Figure 10. Mass Transit Final Cost Estimates — Final cost estimate (millions of 2002 U.S. dollars per kilometer): Minimum / Maximum / Median comparisons across MCDOEs, United States, Asia & Latin America, Europe.
  - Figure 11. Average Cost Overruns — Index (relative to construction sector wage United States =100) and average cost overruns by project type (Rail, Fixed Link, Road, All projects) with standard deviation.

### Project-level cost analysis — roads
- Sample and method:
  - Road projects in Zawya include Iraq (1), Kuwait (1), Oman (1), Qatar (4), Saudi Arabia (3), and the United Arab Emirates (6).
- Findings:
  - Per-kilometer-per-lane cost estimates for 16 road construction projects in MCDOEs vary from $0.2 million to $1.7 million, excluding the projects with the lowest and highest unit costs.
  - The average cost estimate per kilometer per lane is $1.1 million ($0.75 million excluding the outliers).
  - The $1.1 million average is broadly similar to unit cost estimates for central and eastern European and central Asia countries implied by Alexeeva and others (2011).
- Cost escalation context:
  - The literature and region experience suggest cost escalation for road and rail is less concerning than for mass transit.
  - Example: Muscat Expressway in Oman completed in 2012 with a final cost close to $350 million compared with a contract value of $330 million, showing a 6 percent cost overrun.
  - Flyvbjerg, Skamris, and Buhl (2002) identified average cost overruns of 28 percent in a sample of 258 transport projects in 20 mostly advanced countries.
  - Overruns are usually associated with optimism bias in project costing or inadequate controls during project execution.

### Public Investment Management Index (PIMI)
- Index description:
  - The PIMI was developed by Dabla-Norris and others (2012) for 71 countries.
  - Covers four main stages of the project cycle and focuses on institutional processes associated with public investment spending allocated through the central government budget.
  - Table 2. PIMI components:
    1. Strategic Guidance and Project Appraisal: Costed sector strategies; Appraisal standards; Economic appraisals; Independent check.
    2. Project Selection and Budgeting: Medium-term planning and integration; Investment selection; Scope of the legislature’s scrutiny; Public access to key fiscal information.
    3. Project Implementation: Open competition for award of contracts; Complaints mechanism; Capital budget execution; Existence and effectiveness of internal controls; Internal audit in line with international standards.
    4. Project Evaluation and Audit: Routinely performed ex-post evaluations; External audits scrutinized by the legislature; Asset registers.
- MCDOEs expansion and data:
  - This Staff Discussion Note expands the PIMI to MCDOEs based on survey responses from country authorities in late 2013.
  - Overall, scores are available for eight out of 15 MCDOEs.
  - Responses were validated with IMF Country Desks; some responses were revised downward.
- Comparative findings:
  - The available sample of MCDOEs outperforms oil exporters from other regions, but lags behind best-performing countries.
  - On average, MCDOEs outperform countries in sub-Saharan Africa (SSA) and emerging Asia and the Pacific (AP), and score similarly to countries in Central and Eastern Europe (CEE).
  - Caution: Limited coverage of oil exporters in other regions (six in SSA, one in Emerging AP, and two in Latin America and the Caribbean (LAC)).
  - Despite relatively good performance, the average overall PIMI score for MCDOEs is about only half the score of strong performers such as Brazil and South Africa.
- Performance by project cycle stage:
  - MCDOEs’ performance is particularly weak at the appraisal and selection stages.
  - Median scores suggest MCDOEs lag behind CEE and LAC in appraisal and selection stages, but perform better than SSA.
  - Main areas for improvement in MCDOEs:
    - Cost-benefit analysis, at least for large projects.
    - Medium-term planning and budgeting frameworks integrated with the annual budget.
    - Selection of investments on the basis of relevant sector strategies.
    - Consideration of recurrent costs.
  - In implementation and evaluation stages, MCDOEs score above the average of most other regions.
- Annexes:
  - Annex III elaborates components of the PIMI and presents MCDOE country performance across all stages of the PIMI.

### Explaining public investment efficiency — empirical analysis
- Cross-country regression findings:
  - Quality of institutions:
    - Based on specification (1) in Table 3: a one standard deviation increase (about three percentage points in the quality of governance measures from the International Country Risk Guide, namely bureaucracy quality, corruption, and law and order) could improve public investment efficiency by 0.20 (i.e., about 22 percent of the average score in the sample).
    - Interpretation: Consistent with literature that weak institutional quality can lead governments to use capital spending for rent-seeking, producing inefficient spending.
  - Natural resource receipts:
    - Specification (1) suggests a negative association: an increase by one standard deviation in natural resource receipts (about 17 percent) could reduce the efficiency of public investment by 0.02, roughly 3 percent of the average score in the sample.
    - However, the impact estimates are no longer significant in other specifications.
    - Consistent with Gelb and Grassman (2010) on high volatility of natural resource revenues contributing to poor quality of public spending.
  - Official development assistance (ODA):
    - ODA is generally positively associated with investment efficiency, although the estimated coefficient is statistically insignificant.
- Robustness and alternative specifications:
  - Using alternative indicators of institutional quality (World Governance Indicator (WGI) from the World Bank in specification (2)) and including additional control variables (public investment in percent of GDP, oil reserve horizon dummy, public investment volatility dummy) do not generally affect the result that institutional quality is significantly associated with public investment efficiency.
  - Oil reserve horizon: dummy = 1 if country’s reserve horizon is greater than the median of all oil-exporting countries (based on BP data).
  - Public investment volatility: dummy = 1 if the standard deviation of annual percent changes in real public investment is greater than the 90th percentile in the sample.
  - Additional explored variables (statistically insignificant) include degree of urbanization, population density, and business environment.
  - The Budget Institutions Index (Dabla-Norris and others (2011)) used as an alternative measure of institutional quality was statistically significant but reduced the sample by about two-thirds of observations.
- Interpretation:
  - The adverse impact of higher public investment levels on efficiency could reflect diminishing returns on investment and/or wasteful public spending, though these estimates are not statistically significant.

*Source: IMF staff estimates; text from the Staff Discussion Note (Annex IV).*

### Annex II presents a detailed description and summary of statistics of all variables, as well as additional robustness

### _sdn1410 - Annex II presents a detailed description and summary of statistics of all variables, as well as additional robustness

### Regression analysis: specification and robustness checks
- Annex II reports regressions of efficiency scores (PFDH and DEA) using the infrastructure component of the Global Competitiveness Index (GCI) as output and a set of explanatory variables for the period 2006–12.
- Robustness checks include:
  - DEA scores and estimates from an alternative Tobit model (Tobit used because DEA efficiency scores are between 0 and unity).
  - An alternative institutional quality measure (WGI index) reported as "Quality of Institutions Alternative".
  - A Tobit specification reported separately in the Annex.
- Notes on variable construction:
  - Dependent variable: Efficiency Scores (PFDH) 1/ (PFDH scores estimated using the infrastructure component of the GCI (2006-12)).
  - Quality of institutions: ICRG measure on a scale from 0 to 30 with higher values indicating better quality governance (constructed using bureaucracy quality, corruption, and law and order).
  - Quality of Institutions Alternative: WGI index.
  - Natural Resource Receipts: average profits of total natural resource exports in percent of GDP (World Bank WDI).
  - Official Development Assistance: average net disbursement in percent of GDP (OECD).
  - Oil reserve dummy (Oil Reserve Horizon 4/): equals 1 if a country's reserve horizon is greater than the median of all oil-exporting countries, based on BP data.
  - Public Investment Volatility 5/: Standard deviation of annual percent changes in real public investment.

### Key regression coefficients and statistical results (PFDH and Tobit specifications)
- PFDH (OLS-style table presented):
  - Quality of Institutions 2/: 0.0230*** (0.00509)
  - Quality of Institutions 2/ (alternative column): 0.0215*** (0.00559)
  - Quality of Institutions Alternative 3/: 0.00323*** (0.000587)
  - Natural Resource Receipts: -0.00206* (0.00110); -0.00104 (0.00102); -0.00143 (0.00145) across columns
  - Official Development Assistance: 0.000705 (0.00321); 0.00341 (0.00252); 0.00155 (0.00372) across columns
  - Public Investment Level: -0.00394 (0.00549)
  - Oil Reserve Horizon 4/: 0.0148 (0.0439)
  - Public Investment Volatility 5/: 0.000591 (0.000806)
  - Constant: 0.730*** (0.0551); 0.736*** (0.0434); 0.726*** (0.0881)
  - Observations: 9811; 4998 (appears as concatenated "9811498" in table)
  - R-squared: 0.3130; 0.3370; 0.323
  - Note: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1.

- Tobit model (Dependent Variable: DEA scores 1/):
  - Quality of Institutions 2/: 0.0187*** (0.00456); 0.0183*** (0.00508)
  - Quality of Institutions Alternative 3/: 0.00283*** (0.000550)
  - Natural Resource Receipts: -0.00177* (0.00104); -0.000775 (0.000970); -0.00161 (0.00125)
  - Official Development Assistance: 0.00175 (0.00283); 0.00627** (0.00291); 0.00203 (0.00318)
  - Public Investment Level: -0.00116 (0.00532)
  - Oil Reserve Horizon 4/: 0.00268 (0.0399)
  - Public Investment Volatility 5/: 0.000132 (0.000841)
  - Constant: 0.668*** (0.0512); 0.654*** (0.0422); 0.669*** (0.0798)
  - Observations: 9811; 4998 (again presented as "9811498")
  - Note: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1.

### Summary statistics reported (sample means, std, min, max)
- PFDH (qualitative index): Mean 0.9, STD 0.2, Min 0.5, Max 1.2
- Quality of Institution: Mean 8.5, STD 3.1, Min 2.3, Max 15.9
- Quality of Institution Alternative: Mean 48.7, STD 27.3, Min 2.3, Max 99.1
- Natural Resource Receipts: Mean 10.7, STD 16.9, Min 0.0, Max 81.7
- Official Development Assistance: Mean 4.4, STD 7.4, Min -0.1, Max 39.9
- Public Investment Level: Mean 6.1, STD 4.4, Min 0.1, Max 31.4
- Oil Reserve Horizon: Mean 0.8, STD 0.4, Min 0.0, Max 1.0
- Public Investment Volatility: Mean 30.0, STD 17.6, Min 4.1, Max 94.9

### Interpretation of empirical findings
- Institutional quality is positively and statistically significantly associated with higher public investment efficiency:
  - PFDH specification: Quality of Institutions coefficients are positive and significant at *** p<0.01.
  - Tobit specification: Quality of Institutions coefficients are positive and significant at *** p<0.01.
  - The WGI-based alternative institutional measure (Quality of Institutions Alternative) is also positive and significant at *** p<0.01 in both specifications.
- Natural Resource Receipts show negative point estimates in several specifications, with one PFDH column reporting -0.00206* (0.00110) and Tobit one column reporting -0.00177* (0.00104), indicating potential adverse association of resource dependence with efficiency in some specifications.
- Official Development Assistance coefficients are small and not consistently significant; one Tobit column reports 0.00627** (0.00291).
- Other controls (Public Investment Level, Oil Reserve Horizon, Public Investment Volatility) are not consistently significant across specifications.

### Case studies and qualitative lessons (Norway and Chile)
- Norway: Quality Assurance Scheme (QAS) adopted in 2000 with two gateways:
  - First gateway: cost-benefit analysis and assessment of alternatives before government decision.
  - Second gateway: management strategy review and independent consultant views before parliamentary proposal.
  - QAS impact evidence (2000–09): total net saving for reviewed projects estimated at about 7 percent of total investment (analysis based on 40 projects that have produced final cost figures).
  - Sectoral variation: About 80 percent of road projects completed within cost frame; average savings smaller in railway and construction sectors.
- Chile: National Investment System (NIS) led by Ministry of Planning:
  - Legal requirement for all public bodies to apply to NIS for funding.
  - Objective: select projects with largest net present social value; sequential evaluations for complex projects (identification, pre-feasibility, feasibility, design).
  - Screening rejects 5–8 percent of project proposals at initial screening.
  - Mechanisms: reappraisal when lowest tender is 10 percent or more above estimated price; medium-term public investment envelopes; separation of appraisal reviewers from project promoters.
  - Ex-post assessment system: two components (post-construction cost/time/technical compliance reviews; in-depth social impact assessment after sufficient time) and a centralized project information platform for historical analysis.

### Public Investment Management Index (PIMI) findings for MENA and CCA oil exporters (average component scores)
- Overall component scores (averages):
  - Appraisal: 1.6
    - Costed Sector Strategies: 1.3
    - Appraisal Standards: 1.8
    - Economic Appraisals: 1.8
    - Independent Check: 1.5
  - Selection: 1.7
    - Medium-Term Planning and Integration: 1.6
    - Inclusion of Information on Donor-funded Projects: 1.3
    - Investment Selection: 1.8
    - Scope of the Legislature’s Scrutiny: 2.0
    - Public Access to Key Fiscal Information: 1.8
  - Implementation: 2.0
    - Open Competition for Award of Contracts: 2.5
    - Complaints Mechanism: 2.8
    - Capital Budget Execution: 2.1
    - Existence and Effectiveness of Internal Controls: 1.4
    - Internal Audit: 1.8
  - Evaluation: 1.9
    - Evaluation Frequency: 1.0
    - External Audit: 2.0
    - Asset Register: 2.5
- Source of PIMI methodology: Dabla-Norris and others (2012).

### Main findings and policy implications (summary)
- There is substantial room to improve public investment efficiency in MENA and CCA oil exporters; improving efficiency of sizable investment programs could boost growth and advance development agendas.
- MCDOEs’ performance appears weak relative to best non-MCDOE performers across efficiency frontier measures, project-level data, and PIMI scores; GCC countries generally rank higher than other MCDOEs.
- Stronger institutions could foster more efficient public investment, supported by regression results, case studies, and PIMI analysis.
- Near-term, relatively simple measures for MCDOEs include:
  - Increasing transparency of data on key investment projects over the project cycle (e.g., appraisal information, competitive procurement process, bidding statistics, cost/time overruns) and in budget context (e.g., objectives, costs of main investment projects, ex-post evaluations).
  - Preparing an infrastructure needs assessment with development partners to guide medium-term sectoral strategies and monitor progress in closing infrastructure gaps.
  - Undertaking in-depth and independent diagnosis of the current PIM system (in coordination with development partners) to identify country-specific priority areas for reform.
- Subsequently, on the basis of in-depth PIM analysis, focus reforms on:
  - Stronger medium-term budget frameworks.
  - More thorough appraisal and selection of investment projects.
  - Systematic ex post assessments to strengthen productivity of public investment.
- Further guidance will be provided by the forthcoming IMF Policy paper “Making Public Investment More Efficient,” which will discuss which fiscal institutions are critical to improve public investment efficiency.

*Source: Annex II and related sections of the IMF Staff Discussion Note contained in the supplied content.*

### Annex IV. Summary of Scores by Analytical Country Groups

### Annex IV. Summary of Scores by Analytical Country Groups

### Exclusions and notes
- Azerbaijan, Iraq, Uzbekistan, and Turkmenistan are excluded from the efficiency frontiers and Algeria, Bahrain, Iran, Oman, Saudi Arabia, United Arab Emirates and Uzbekistan from the PIMI assessment due to data availability.

### Summary statistics by country group (as presented)
- GCC (Bahrain, Oman, Kuwait, Qatar, Saudi Arabia, United Arab Emirates)
  - PFDH scores / DEA scores / PIMI
    - Max0.9980.9022.542
    - Min0.6940.6571.842
    - Median0.9700.8282.192
    - Average0.9000.7952.192
  - (Additional block for GCC)
    - Max0.9560.8302.384
    - Min0.5840.5531.375
    - Median0.7620.6731.527
    - Average0.7690.6861.677

- MCD oil exporters
  - PFDH scores / DEA scores / PIMI
    - Max0.9980.9022.542
    - Min0.5840.5531.375
    - Median0.9090.8161.646
    - Average0.8400.7461.806

- Non-M CD commodity exporters with strong institutions
  - PFDH scores / DEA scores / PIMI
    - Max1.1311.000n.a.
    - Min0.8280.772n.a.
    - Median1.0350.924n.a.
    - Average1.0190.905n.a.

- Advanced economies
  - PFDH scores / DEA scores / PIMI
    - Max1.2161.000n.a.
    - Min0.7670.709n.a.
    - Median1.0340.903n.a.
    - Average1.0290.894n.a.

- Emerging markets
  - PFDH scores / DEA scores / PIMI
    - Max1.1311.0003.533
    - Min0.5460.5160.908
    - Median0.9120.8151.842
    - Average0.8920.8001.884

- Low-income countries
  - PFDH scores / DEA scores / PIMI
    - Max1.0201.0002.334
    - Min0.5140.4710.900
    - Median0.8520.7541.684
    - Average0.8290.7581.662

### Methodologies referenced
- Efficiency frontier analysis
- Global
- Non-GCC (Algeria, Azerbaijan, Iran, Iraq, Kazakhstan, Libya, Turkmenistan, Uzbekistan, Yemen)

*Source: IMF staff estimates.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/sdn/2014/_sdn1410.pdf_