## 1. Overall Trend of PPP Investments

## Source details

**Canonical URL:** [1. Overall Trend of PPP Investments](https://www.imf.org/-/media/files/publications/wp/2017/wp17243.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2017/wp17243.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2017/wp17243.pdf.json)

---

### Introduction and dataset
- PPPs can augment public investment and improve the efficiency of infrastructure spending; suited to large-scale power, transportation or telecom projects but expose investors to high regulatory uncertainty and lengthy payback periods.
- Governments and concessionaires often go into disputes during project lifecycles, leading to contract renegotiation or termination.
- Dataset: about 6,000 PPP contracts from 113 emerging and developing countries.
- Empirical strategy:
  - Theoretical model on PPP choice vs public procurement and optimal government commitment under accumulating fiscal risks from guarantees vs direct subsidies.
  - Estimation of (i) selection of PPP contract schemes and (ii) effect of governments’ fiscal commitments on the probability of contract disputes.
  - Tests using two fiscal institution measures:
    - a score on PIM quality related to PPPs (compiled from EIU (2012, 2014, 2015), IMF (2015), and World Bank (2016))
    - the budget transparency index (Wang et al, 2015)
  - Propensity score matching (PSM) with IPTW to address endogeneity in contract-type selection.

### Global trends and stylized facts
- PPP investments doubled from about $50 billion per year on average during 2000–05 to about $100 billion per year on average during 2010-15.
- Most projects concentrated in large MICs: BRICs, Argentina, Mexico, Colombia, Philippines, Indonesia, Turkey, and Nigeria.
- Developing Asia and LAC experienced a spike in PPPs since 2004, reaching about $50–60 billion per year.
- PPPs remain relatively small in other regions.
- Dispute risk:
  - Peaked in the late 1990s to the early 2000s in developing Asia and LAC.
  - Continued to rise in sub-Saharan Africa (SSA) as more contracts were signed.
  - Disputes more common in developing Asia, CEE-CIS, and LAC, particularly in the upper-MICs.

### Key empirical findings — contract features and dispute risk
- Guarantees and dispute risk:
  - Contrary to theoretical prediction that guarantees promote private participation, empirical results find disputes tend to increase for guaranteed contracts due to higher fiscal risks.
  - Negative consequences of guarantees linked to adverse selection of riskier projects and accumulation of contingent liabilities.
- Fiscal institutions:
  - Weak fiscal institutions (poor budget transparency and weak PIM related to PPPs) increase the likelihood of contractual failure by undermining project selection and dispute management.
- Geographic differences:
  - Matured markets (developing Asia and Latin America) commonly use guarantees; improving budget transparency is critical to manage PPP fiscal risk.
  - In SSA, concession contracts face higher dispute risk due to lower capacity in assessing project risk; strengthening PIM systems is essential.

### Theoretical framework — core mechanisms (preserved notation/context)
- Public investment: z yields project returns ܹ஺.
- SPV investment: k, return ߨ linked to service outcomes ܹ஻; SPV return depends on expected demand q, financing cost r, operational costs ܿଵ.
- PPPs can create short-term rents R when procured non-competitively; higher PPP regulation and operational quality ߯ଵ ∈ (0,1) limit rent-seeking.
- Voter policy bias ߪ௜ ~ ܷ_{−1/2,1/2}. PPPs selected when Pr[ܹ஺ − ܹ஻ + ߪ௜] ≥ 1/2.
- Comparative statics: ∂k*/∂߯ଶ < 0 implies optimal level of PPP investment declines as budget transparency improves; lack of budget transparency leads to over-commitment on PPPs.

### Conditions for disputes (firm-led and government-led, preserved expressions)
- Firm-led disputes:
  - Government-pay (ߤ = 1): firm refuses participation if participation constraint Π௧ ≤ 0; disputes when ߙ ≤ ܿ + r( ... ) for all t.
  - User-pay (ߤ = 0): financial loss probability ܨ(ϖ) = Pr[ݍ ≤ threshold] = Φ( (threshold − ݍ̄)/σ ); weak PIM leads to over-estimated break-even demand and higher ܨ(ϖ).
- Government-led disputes:
  - Contract terminated when ܹேு − ܹு ≥ 0 (condition in equation (4)) under government-pay; under user-pay analogous condition in equation (5) with fiscal risk term.
  - Accumulation of PPP contingent liabilities can induce government-initiated disputes.

### Testable hypotheses (numbered as in source)
- Hypothesis 1 (Adverse selection and PIM quality): Weaker PIM quality ߯ଵ could result in adverse selection of high risk projects.
- Hypothesis 2 (Budget transparency): Lack of budget transparency ߯ଶ in reporting contingent liabilities from guaranteed debt leads to over-commitment on PPPs.
- Hypothesis 3 (Effect of guarantee): Provision of guarantees encourages firm participation while increasing government-led disputes for higher fiscal risk and insufficient returns.
- Hypothesis 4 (Reputation costs): Likelihood of reneging contract is lower with stronger political constraint: ∂R/∂ψ(υ) ≥ 0.

### Data, sample, classification and estimation
- Data sources: World Bank PPI Database for contractual disputes; additional public sources (Factiva, MFI and sponsor websites) and project entities.
- Sample coverage: greenfield (BOT) and brownfield concessions for 113 emerging and developing countries; about 6,000 PPP contracts signed between 1984 and 2015.
- Dispute classification:
  - “Disputed” if distressed, renegotiated or cancelled due to conflicts between government and private parties.
  - “Government-led” disputes triggered by sovereign risk (economic crisis, political risk).
  - “Firm-led” disputes triggered by sponsor insolvency and technical problems.
  - Distressed contracts classified “non-disputed” if renegotiations due to sponsor unilateral actions or uninsurable external events (war, civil conflict).
- Estimation: reduced-form hazard model (Weibull) of probability of disputes conditional on contract duration ܣ; PSM-IPTW used to adjust for selection into contract types.

### Main quantitative results (preserving exact reported figures)
- Stylized sample moments and dataset scope:
  - Full sample includes 5,999 PPP contracts.
  - Total disputes occurred for about 6½ percent of the contract.
  - Guarantees: only 13 percent of contracts receive guarantees on average.
  - Direct subsidies: about 12 percent of contracts.
  - MFI involvement: 13 percent of contracts.
  - Local investor participation: about 65 percent of contracts in MICs and LICs.
  - Developing Asia local sponsorship: 78 percent.
  - Average GDP growth rate: 5.4 percent
  - Average inflation (CPI): 9.3 percent
  - GFN (in percent of GDP) average: 11 percent with range 0.3 to 102.2 percent
  - Public debt (in percent of GDP) average: 49.2 percent
- Summary statistics (selected exact values from Table A2):
  - Disputes: N 5,999 Mean 0.064 Std. dev 0.244 Median 0 Min 0 Max 1
  - Government-led disputes: N 4,598 Mean 0.044 Std. dev 0.205 Median 0 Min 0 Max 1
  - Contract duration (year): N 5,999 Mean 9.1 Std. dev 6.5 Median 8.0 Min 0.5 Max 34
  - Investment size (in mil USD): N 5,999 Mean 339.4 Std. dev 1,151.1 Median 77.5 Min 0.0 Max 35,586.5
  - Government guarantees: N 5,999 Mean 0.127 Std. dev 0.333 Median 0 Min 0 Max 1
  - Concession: N 5,999 Mean 0.271 Std. dev 0.444 Median 0 Min 0 Max 1
  - MFI supports: N 5,999 Mean 0.126 Std. dev 0.332 Median 0 Min 0 Max 1
  - Real GDP growth (%): N 5,998 Mean 5.4 Std. dev 2.3 Median 5.6 Min -9.9 Max 11.0
  - CPI inflation (%): N 5,997 Mean 9.3 Std. dev 10.2 Median 7.7 Min -1.0 Max 135.0
  - GFN/GDP (%): N 5,967 Mean 10.9 Std. dev 7.0 Median 11.8 Min 0.3 Max 102.2
  - Public debt/GDP (%): N 5,958 Mean 49.2 Std. dev 20.5 Median 42.7 Min 0.0 Max 163.4
  - PPP capital stock/GDP (%): N 5,681 Mean 3.6 Std. dev 5.4 Median 2.2 Min 0.0 Max 94.5

### Determinants of contract selection (first-stage probit—selected exact coefficients)
- Ln(investment size): 0.030*** [0.002] for MFI supports (column 1).
- Ln(GFN/GDP): 0.055*** [0.013] for MFI supports; -0.084*** [0.018] for direct subsidy (column 4).
- Ln(public debt/GDP): 0.036** [0.017] reported in column 3 (specification context).
- Regulatory quality: 0.076*** [0.020] for MFI supports.
- Bureaucratic efficiency: -0.119*** [0.025] for MFI supports.
- Sponsor with same nationality: -0.043*** [0.009] for MFI supports; 0.049*** [0.010] for direct subsidy.
- Dummy: post-2000: -0.062*** [0.010] for MFI supports; 0.176*** [0.017] for guarantees.
- Observations across specifications: 5,877.
- Pseudo R-squared reported up to 0.348 (concession column).

### Hazard regression results (baseline and PSM-IPTW adjusted — selected exact estimates)
- Duration dependence: duration parameter ߙ > 1; Ln(ߙ) positive (examples: 0.076 [0.055], 0.080 [0.054]).
- Investment size: Ln(investment size) coefficient e.g., 0.147*** [0.037] (column 1).
- MFI supports: negative coefficients (mitigating disputes), effect larger and significant in some PSM-IPTW subsamples.
- Country-level variables:
  - Democracy: -0.049** [0.021] (column 1).
  - Real GDP growth: -0.256*** [0.052] (column 1).
  - CPI inflation: 0.039*** [0.002] (column 1).
- Contract-type effects (PSM-IPTW adjusted):
  - Guaranteed contracts: higher hazard; reported hazard ratio equivalent of 4.3 (column 5).
  - Guarantee coefficient (PSM-IPTW, column 5): 1.453** [0.595].
  - Concession coefficient (column 7): 1.037*** [0.246].
- Observations and log-likelihood examples:
  - Observations: 5,976 (column 1); 5,861 (columns with PSM-IPTW).
  - Log-likelihood examples: -1,499.9 (column 1), -2,968.2 (column 5).

### Robustness and heterogeneity (preserved findings)
- Guarantees and concessions hazard ratios remain >1 and significant in most specifications except SSA sample (guarantees not significant in SSA).
- Hazard ratio of guarantee significantly higher in matured PPP markets; slightly higher when LAC excluded.
- Concessions are riskier in less matured markets; in SSA concessions show higher risk reflecting larger demand risk.
- Guaranteed contracts show much higher risk when outcome is government-led disputes (risk triples), indicating government initiation drives many guaranteed-contract disputes.
- MFI support and democratic regime consistently reduce hazard ratios across specifications; effect stronger for government-led disputes and in matured markets or when LAC excluded.

### Role of PIM quality and budget transparency (selected exact estimates and tests)
- PPP PIM quality index: constructed from EIU’s Infrascope index and imputed IMF/World Bank scores; scaled 0 (low) to 100 (high).
- Correlation: positive correlation between PPP PIM quality index and budget transparency index (stock and flow). Budget transparency and PIM quality significantly lower in LICs (p-value=0.00).
- Hazard regressions:
  - PPP overall PIM quality index: -0.516*** [0.106].
  - PPP operational quality index: -0.350*** [0.126].
  - Budget transparency index (stock) in some subsample: -1.052*** [0.296].
- Heterogeneous guarantee effects:
  - Guarantee coefficient increases to 2.1 for lower PIM group (columns 2 and 4).
  - Guarantee coefficient rises to 1.9 for contracts in non-transparent budget reporting systems (column 6 low-transparency group).
  - Equality of coefficient F tests show significant differences (example p-value: 11.86 (0.001)*** in table footnotes).

### Policy recommendations and conclusions (preserved priorities)
- Medium-term priorities:
  - Improve PPP selection and management capacity (PIM quality) and enhance budget transparency to reduce adverse selection and hidden contingent liabilities.
  - Strengthen regulatory quality and bureaucratic efficiency to lower reliance on guarantees and reduce dispute risk.
  - Integrate PPPs into traditional PIM frameworks and incorporate PPP fiscal risk into the fiscal framework.
  - Develop central PPP units; build PPP-specific legal and regulatory frameworks and enhance capacity to manage fiscal risks.
- Short-term measures:
  - Draft contracts with support from multilateral financial institutions (MFIs) to leverage experience, increase reputational costs of government breaches, and mitigate dispute risk.
- Regional considerations:
  - In matured PPP markets (developing Asia and Latin America), high use of guarantees raises fiscal risk and dispute incidence; improving budget transparency is critical.
  - In sub-Saharan Africa, concessions predominate and are riskier due to weak PIM quality; as guarantees increase in SSA, dispute risk may rise over next 5-10 years.
- Final takeaway:
  - Improving structural weaknesses in budget transparency and PPP public investment management is a prerequisite for maximizing PPP efficiency and minimizing fiscal risks; MFI involvement provides an interim risk-mitigation tool.

*Source: wp17243 - 1. Overall Trend of PPP Investments (IMF working paper).*

### 1. Overall Trend of PPP Investments ....................................................................................

### 1. Overall Trend of PPP Investments

### Introduction
- PPPs can augment public investment and improve the efficiency of infrastructure spending.
- PPPs are suited to large-scale power, transportation or telecom projects that are capital intensive and technically complex, but their quasi-public nature and long economic lifecycle expose investors to high regulatory uncertainty and lengthy payback periods.
- Since the early 1990s, PPPs have increased in both middle (MICs) and low income countries (LICs) as an alternative source of financing to scale up public capital stock.
- Governments and concessionaires often go into disputes during project lifecycles, leading to contract renegotiation or termination.
- The paper provides empirical evidence that weaknesses in governments’ fiscal institutions – budget transparency and public investment management (PIM) of PPPs – can undermine project selection quality and increase contract disputes.

### Dataset and approach
- Dataset: about 6,000 PPP contracts from 113 emerging and developing countries.
- Empirical strategy:
  - A theoretical model on PPP choice over public procurement and optimal government commitment when fiscal risks accumulate from guarantees vs. direct subsidies.
  - Estimation of (i) selection of PPP contract schemes and (ii) effect of governments’ fiscal commitments on the probability of contract disputes.
  - Tests on whether improvements in fiscal institutions reduce disputes using two fiscal institution measures:
    - a score on PIM quality related to PPPs (compiled from EIU (2012, 2014, 2015), IMF (2015), and World Bank (2016))
    - the budget transparency index (Wang et al, 2015)
  - Propensity score matching (PSM) applied to address endogeneity due to selection of contract types.

### Stylized facts — Global trends in PPPs
- PPP investments doubled from about $50 billion per year on average during 2000–05 to about $100 billion per year on average during 2010-15.
- Most projects are in large MICs, such as BRICs, Argentina, Mexico, Colombia, Philippines, Indonesia, Turkey, and Nigeria.
- Countries in developing Asia and LAC experienced a spike in PPPs since 2004, reaching about $50–60 billion per year.
- PPPs remain relatively small in other regions.

### Evolution of dispute risk
- Dispute risk:
  - Peaked in the late 1990s to the early 2000s in developing Asia and LAC.
  - Continued to rise in sub-Saharan Africa (SSA) where more contracts started to be signed.
- Disputes are more common in developing Asia, the Central and Eastern Europe & the Commonwealth Independent States (CEE-CIS), and LAC, particularly in the upper-MICs.

### Key empirical findings
- Contrary to the theoretical prediction that guarantees promote private participation, empirical results find that disputes tend to increase for guaranteed contracts due to higher fiscal risks.
- The negative consequence of guarantees is associated with:
  - adverse selection of riskier projects
  - accumulation of contingent liabilities
- Weak fiscal institutions (poor budget transparency and weak PIM related to PPPs) increase the likelihood of contractual failure by undermining project selection and dispute management.
- Geographic differences in implications:
  - Matured markets (such as developing Asia and Latin America) commonly use guarantees; improving budget transparency is critical to manage PPP fiscal risk.
  - In sub-Saharan Africa, concession contracts face higher dispute risk than other regions due to lower capacity in assessing project risk; strengthening PIM systems is essential to improve selection and operation of PPPs.

### Policy recommendations
- For optimal PPP design and fiscal risk management:
  - Develop central PPP units.
  - Incorporate the fiscal risk of PPPs into the fiscal framework.
  - Integrate PPPs into traditional PIM frameworks.
  - Improve budget transparency to adequately manage fiscal risk where guarantees are prevalent.
  - Strengthen PIM systems, especially in SSA, to improve project selection and operation.
  - Build PPP-specific legal and regulatory frameworks and enhance capacity to manage fiscal risks, as low capacity results in contractual failure, discouraging private investment and reducing social welfare.

*Source: wp17243 - 1. Overall Trend of PPP Investments.*

### Appendix Figure A1 shows that PPPs (a measure of a government’s past experiences in PPP

### wp17243 - Appendix Figure A1 shows that PPPs (a measure of a government’s past experiences in PPP

### Clustering of PPPs and determinants of disputes
- PPPs and observed disputes tend to cluster in particular regions.
- Explanations for clustering:
  - Disputes are more common in countries vulnerable to financial crises.
  - Dispute risk rises when an economy is hit by external shocks (such as commodity price fluctuations, natural disasters, and adverse shocks in growth and inflation) (Nose 2014).
  - Political regime change can be a determinant of disputes.
  - Institutional factors: countries with authoritarian political regimes and poor economic and political institutions are more likely to initiate disputes (Acemoglu and Robinson, 2005; Eden, Kraay and Qian, 2012).
  - Evidence from World Bank’s political risk insurance agency (MIGA) pre-claim data on breach of contract is consistent with these patterns.
- In many cases, economic crisis and political regime change exposed PPP projects to financial stress or inconsistent policy (or both), leading to renegotiations of contract.

### Contract design and fiscal institutions
- Common government supports in PPPs:
  - Guarantees, direct subsidies, or Viability Gap Funds (VGFs) to support private participation in infrastructure (example cited: Engel, Fischer, and Galetovic (2014) for VGFs in India).
- Trends and regional patterns:
  - More private investors sought supports from multilateral financial institutions (MFIs) in the early 1990s; recently guarantees and direct subsidy have become important risk mitigation instruments for PPPs (Figure 3).
  - PPP contracts that received government guarantees and direct subsidies were concentrated in developing Asia and LAC (Figure 4); concessions were also common in both regions.
  - In SSA countries, guaranteed contracts have gradually increased but are still rare; concessions and MFI-supported projects are more common.
- Fiscal risk implications:
  - Recent increase in government involvement in PPPs could elevate fiscal risk as more guarantees and subsidies are provided, especially if granted to pursue projects which are not always financially viable.
- Motivations and weaknesses:
  - PPPs were often motivated by fiscal constraints on borrowing and taxation that limited governments’ financial ability to undertake public investments.
  - PPPs are often used to circumvent budget constraints and to postpone recording the fiscal costs of investments rather than for efficiency reasons.
  - In many countries, PPPs have not always performed better than public procurement; they have functioned as ad-hoc, off-budget arrangements without robust processes for risk assessments.
- Success factors:
  - Effective procurement decisions, economic and financial feasibility, efficient management and oversight under strong fiscal and legal institutions (Sabol and Puentes, 2014).
  - Transparent disclosure of fiscal risk in the medium-term budgetary frameworks to monitor and manage contingent liabilities.
- Definitions and notes:
  - Government guarantee forms include: minimum payment guarantee (demand risk), debt guarantee (insolvency risk), revenue guarantee (minimum revenues), exchange rate guarantee (currency risk) (World Bank, 2014b). (Footnote 4)
  - Direct subsidies include availability payments (regular subsidy over the life-cycle of the project) and capital subsidy for initial construction cost (World Bank 2014b). (Footnote 5)

### Theoretical framework: overview
- Objective: model how fiscal institutions—PIM quality and budget transparency—affect the survival of PPPs through government procurement and investment decisions and firm participation.
- Core comparisons: public procurement vs PPPs (SPV investing k, returns linked to service outcomes).
- Key model features and notation preserved exactly as presented:
  - Public investment: z yields project returns ܹ஺.
  - SPV investment: k, return ߨ linked to service outcomes ܹ஻.
  - SPV return depends on expected demand q, project financing cost r, and operational costs ܿଵ for SPV.
  - PPPs can create short-term rents R when procured non-competitively.
  - Higher PPP regulation and operational quality ߯ଵ ∈ (0,1) limit politician’s rent-seeking.
  - Voter policy bias ߪ௜ ~ ܷ_{−1/2,1/2}. PPPs selected when attract majority voting: Pr[ܹ஺ − ܹ஻ + ߪ௜] ≥ 1/2 leading to condition in equation (1).
- Implications:
  - PPPs chosen for higher efficiency gain than public procurement (݂(ݖ,݇)≫); or regardless of feasibility due to adverse selection from overestimation of PPP returns (optimism bias (Flyvbjerg, 2009)) and rent-seeking by politicians, both due to weak PIM quality.

### Procurement and investment decision mechanics
- Payment schemes under BOT contract:
  - Government-pay: government provides recurring subsidy (availability payments) ߙ to firm; ߤ = 1.
  - User-pay: firm collects user-fees; ߤ = 0.
  - Government offer: (ߙ,݌) = (ܣ(݌)ߤ, p is user fee and ߙ is availability payments).
  - Asset returns to government at contract termination time ݐ̅.
- Government maximizes present value of output subject to budget constraint (equation (2)); budget constraint includes:
  - Lump-sum tax revenues T.
  - Infrastructure investment ݅௧ݖ = (௧ାଵ)(1 − ߜ1)ݖ௧.
  - Operational cost ܿଶ to maintain infrastructure capital.
  - User-fees stochastic: ݍ = ݍߩ = ݍ̄ + ϵ, with ϵ ~ N(0,σ^2).
  - Government may offer direct subsidy ߙ (if ߤ = 1), and revenue guarantee ݍ(݌,ݖ) to ensure minimum user-fee revenues ݍ݌ݖ to the firm.
  - Fiscal sustainability requires fiscal deficit below fiscal limit L.
- Budget transparency:
  - Budget transparency higher as ߯ଶ → 1; close to zero if a contingent liability is hidden by undertaking the PPP off-budget.
- Production function and firm choice:
  - Cobb-Douglas production ݂(݇௧ݖ௧) = ݇^{α}ݖ^{β}; private return ߮ increases as service quality improves.
  - Firm chooses k* = argmax_k [k^{α} q^{β} (ߙߤ + k r) − ܿଵ k −... ] (preserved formulation context).
- Public optimal investment k* and comparative statics (equation (3)):
  - Lagrange multiplier ܧ௧ ߣ௧ாଵ is lower as lump-sum tax increases and higher as fiscal costs (direct subsidy and fiscal risk of guarantees) increase (see appendix E).
  - From Eq. (3), ∂k*/∂߯ଶ < 0, implying the optimal level of PPP investment declines as budget transparency improves.
  - Lack of budget transparency leads to over-commitment on PPPs; combined with adverse selection, this leads to over-commitment on infeasible PPPs likely to trigger guarantees.

### Conditions for outbreak of disputes (firm-led and government-led)
- Firm-led disputes:
  - Under government-pay scheme (ߤ = 1):
    - Firm refuses participation if participation constraint Π௧ ≤ 0 where Π௧(ߙ) = ܿ + r( ... ) for k*_{௧}. Firm initiates disputes when subsidy insufficient to cover financing and operational costs: ߙ ≤ ܿ + r( ... ) for all t.
  - Under user-pay scheme (ߤ = 0):
    - Financial loss occurs if user-fee revenues below minimum threshold with probability ܨ(ϖ) = Pr[ݍ ≤ threshold] = Φ( (threshold − ݍ̄)/σ ).
    - Firm’s participation constraint violated if Π௧^{(user)} ≤ 0 expressed in terms of ܨ(ϖ), revenues, and costs for k*_t for all t.
    - When PIM quality is weak, break-even public service demand ݍ is over-estimated ex-ante, increasing likelihood of financial loss ܨ(ϖ).
- Government-led disputes:
  - Under government-pay scheme:
    - Government compares welfare if project completes (ܹு) vs terminated (ܹேு).
    - Successful PPP yields welfare ܹு = Σ_{t=0}^{t̅} β_t [ outputs + SPV’s profit ] + β_{t̅} ܫ.
    - Government may breach at period τ, pay reputation cost ܪ, and continue via public procurement; welfare after breach given by expression in text.
    - Contract terminated if ܹேு − ܹு ≥ 0 leading to condition in equation (4).
    - Equation (4): first square bracket captures efficiency gains of PPP vs public procurement; second captures value for money. Government-led dispute occurs when sum of these two terms is smaller than RHS (residual asset value net of reputation cost).
  - Under user-pay scheme:
    - Welfare ܹு^{(user)} defined with fiscal risk term (third term).
    - Condition for contract breach expressed in equation (5): accumulation of PPP contingent liabilities could induce government to initiate disputes.

### Testable predictions and hypotheses
- Hypothesis 1 (Adverse selection and PIM quality): Weaker PIM quality ߯ଵ could result in adverse selection of high risk projects.
- Hypothesis 2 (Budget transparency): Lack of budget transparency ߯ଶ in reporting contingent liabilities from guaranteed debt leads to over-commitment on PPPs.
- Hypothesis 3 (Effect of guarantee): Provision of guarantees encourages firm’s participation in PPPs, while increasing government-led disputes for higher fiscal risk and insufficient returns to recover project costs.
- Hypothesis 4 (Reputation costs): Likelihood of reneging contract is lower with stronger political constraint: ∂R/∂ψ(υ) ≥ 0. (Reference to “sovereign theft” by Tomz and Wright (2010))

### Data and methodology (data description and hazard estimation)
- Data sources:
  - Contractual dispute data from World Bank’s Private Participation in Infrastructure (PPI) Database.
  - Additional data from public sources (Factiva, websites of multilateral agencies and sponsors) and information from project entities.
- Sample coverage:
  - Covers both greenfield contracts (BOT) and brownfield concessions for 113 emerging and developing countries (see Table A1 and Table A2 referenced).
  - Data include about 6,000 PPP contracts signed between 1984 and 2015.
- Dispute classification:
  - Contracts classified as “disputed” if distressed, renegotiated or cancelled due to conflicts between government and private parties.
  - Disputes classified as “government-led” if triggered by sovereign risk including economic crisis (financial crisis, sharp currency devaluations) and political risk (nationalization, expropriation, policy changes in tariff setting).
  - Disputes classified as “firm-led” if triggered by sponsor’s insolvency and technical problems.
  - Contracts experiencing distress are classified as “non-disputed” if renegotiations caused by sponsor’s unilateral actions (e.g., change in business strategy) or uninsurable external events beyond parties’ control (such as war and civil conflict).
- Estimation approach:
  - Reduced-form approach (hazard model) used to test the hypotheses derived in the theoretical model.

*Source: wp17243 - Appendix Figure A1 shows that PPPs (a measure of a government’s past experiences in PPP; content excerpt provided).*

### section III. It estimates the probability of disputes conditional on the duration of contract ܣ

### section III. It estimates the probability of disputes conditional on the duration of contract ܣ

### Model specification and estimation
- Duration variable: ܣ
௜, defined as the number of years a project survives before it ends due to contract breach or termination. Observed duration ܣ
௜
ൌ min(ܣ
௜
∗, c) with c = 2015 for censored contracts.
- Distributional assumption: ܣ
௜ follows a Weibull distribution. Hazard function estimated parametrically as:
  - ܣ(ߣ
௜
ݔ; ௜௖௥) = exp(ݔ
௜௖௥ᇱߚ) ܣߙ
௜
ఈିଵ where ߙ measures duration dependence.
- Standard errors: clustered over industry and contract year levels.
- Covariates (ݔ
௜௖௥) include:
  - ܺ
ଵ,௜: contract-level characteristics (financial arrangements with MFIs, government guarantees, direct subsidies, concession agreement, ln(investment size), sponsor nationality, post-2000 dummy).
  - ܺ
ଶ,௖: country-level variables (average real GDP growth, CPI inflation during contractual period, democratic regime measured by Polity IV-derived score from 0-20, duration of national leaders, regulatory quality, bureaucratic efficiency, public debt/GDP, gross financing needs (GFN) in percent of GDP, commodity exporter status).
  - ܦ
௝௥: region and sector dummies to account for clustering.

### Bias adjustment: Propensity Score Matching with IPTW
- Motivation: selection of contract type depends on government and private firm characteristics, biasing contract-level estimates.
- Method: PSM with inverse probability of treatment weighting (IPTW) applied to hazard regression to adjust for selection bias in a non-linear hazard model where valid instruments are absent.
- Rationale: PSM-IPTW offers parsimonious specification with least mean squared error for hazard estimation (Austin 2012).
- Limitation: selection on unobservables may persist (e.g., investor strategic skills), but inclusion of comprehensive country- and investor-level covariates aims to minimize bias.

### Determinants of contract selection (first-stage probit; main qualitative findings)
- Gross financing needs (GFN, in percent of GDP):
  - Positively associated with offering guarantees.
  - Negatively associated with direct subsidies.
- Public debt/GDP:
  - Positive coefficients on public debt and negative interaction with budget transparency index (stock) indicate indebted governments use guarantees more, but this tendency decreases with higher budget transparency.
- Institutional quality:
  - Regulatory quality positively associated with obtaining MFI supports.
  - Higher bureaucratic efficiency associated with higher likelihood of concessions and lower probability of offering guarantees.
- Sponsor nationality:
  - Local sponsor increases likelihood of direct subsidies and concessions.
- Commodity exporters:
  - PPPs in commodity-exporting countries tend to receive more guarantees and direct subsidies.
- Selected quantitative indicators from Table 1 (average marginal effects; standard errors in brackets):
  - Ln(investment size): 0.030*** [0.002] for MFI supports (column 1).
  - Ln(GFN/GDP): 0.055*** [0.013] for MFI supports; -0.084*** [0.018] for direct subsidy (column 4).
  - Ln(public debt/GDP): coefficients vary by specification (e.g., 0.036** [0.017] in column 3).
  - Regulatory quality: 0.076*** [0.020] for MFI supports.
  - Bureaucratic efficiency: -0.119*** [0.025] for MFI supports.
  - Sponsor with same nationality: -0.043*** [0.009] for MFI supports; 0.049*** [0.010] for direct subsidy.
  - Dummy: post-2000: -0.062*** [0.010] for MFI supports; 0.176*** [0.017] for guarantees.
  - Observations across specifications: 5,877.
  - Pseudo R-squared ranges reported up to 0.348 (concession column).

### Hazard regression results (baseline and PSM-IPTW adjusted)
- Duration dependence:
  - Duration dependence parameter (ߙ) is always greater than one (Ln(ߙ) reported positive across columns), confirming positive duration dependence: probability of disputes increases as contracts mature (obsolescing bargain).
  - Example Ln(ߙ) values from Table 2: 0.076 [0.055] (column 1), 0.080 [0.054] (column 2).
- Investment size:
  - Positive and significant: Ln(investment size) coefficients e.g., 0.147*** [0.037] (column 1).
- MFI supports:
  - Negative coefficients (mitigating disputes) but generally not always statistically significant in baseline; effect larger and significant in some PSM-IPTW subsamples.
- Country-level macro/political variables:
  - Democracy: negative coefficient e.g., -0.049** [0.021] (column 1), indicating democracies see fewer disputes.
  - Duration of national leader: 0.909*** [0.298] reported (model interaction context).
  - Real GDP growth: negative and significant (e.g., -0.256*** [0.052] column 1).
  - CPI inflation: positive and significant (e.g., 0.039*** [0.002] column 1).
- PPP capital stock/GDP: coefficient negative but not statistically significant in column 4.
- Effects of contract types (PSM-IPTW adjusted hazard estimates):
  - Guaranteed contracts: higher hazard; reported hazard ratio equivalent of 4.3 (column 5).
  - Direct subsidy: positive but not statistically significant (column 6: coefficient 0.912 [0.593]).
  - Concessions: significantly higher hazard (column 7 coefficient 1.037*** [0.246]).
- Selected quantitative values from Table 2:
  - Guarantee coefficient (PSM-IPTW, column 5): 1.453** [0.595].
  - Concession coefficient (column 7): 1.037*** [0.246].
  - Observations: varying by column (e.g., 5,976 in column 1; 5,861 in columns with PSM-IPTW).
  - Log-likelihood examples: -1,499.9 (column 1), -2,968.2 (column 5).

### Robustness checks and heterogeneity analyses
- Alternative samples and tests:
  - Split by market maturity (top 5 percent vs less matured PPP countries), exclude 1,954 LAC contracts, restrict to SSA, restrict to user-pay contracts, and use government-initiated disputes as outcome.
- Key robustness findings:
  - Hazard ratios of guarantees and concessions remain >1 and significant in most specifications except SSA sample (where guarantees not significant).
  - Hazard ratio of guarantee is significantly higher in matured PPP markets and slightly higher when LAC countries excluded.
  - Concessions are riskier in less matured markets.
  - In SSA, guarantees less common and guaranteed contracts face lower dispute risk relative to baseline; concessions show higher risk in SSA reflecting larger demand risk.
  - Guaranteed contracts show much higher risk when outcome is government-led disputes (risk triples), suggesting government initiation drives many guaranteed-contract disputes.
  - MFI support and democratic regime consistently reduce hazard ratios across specifications; effect stronger for government-led disputes and in matured markets or when LAC excluded.

### Role of public financial management (PIM quality and budget transparency)
- PPP PIM quality index:
  - Constructed from EIU’s Infrascope index, expanded by imputations from IMF (2015) PIMA index and World Bank (2016) PPP preparation/procurement/contract management scores; scaled 0 (low) to 100 (high).
- Correlation:
  - Positive correlation between PPP PIM quality index and budget transparency index (stock and flow). Budget transparency and PIM quality significantly lower in LICs (p-value=0.00).
- Hazard regressions with PIM quality and budget transparency (selected results from Table 3):
  - PPP overall PIM quality index has a negative and significant effect on disputes (e.g., -0.516*** [0.106]).
  - PPP operational quality index negative and significant (e.g., -0.350*** [0.126]).
  - Budget transparency index (stock) coefficient reported as -0.161 [0.130] in column 5 (not significant), and -1.052*** [0.296] in a subsample.
- Adverse selection and contingent liability effects:
  - Guarantee coefficient increases substantially for low-PIM group: rises to 2.1 for lower PIM group (columns 2 and 4 show this increase), indicating negative adverse selection.
  - Guarantees rise to 1.9 for contracts signed in countries with non-transparent budget reporting systems (column 6 low-transparency group), indicating pure contingent liability effect.
  - Equality of coefficient F test confirms significant differences in guarantee coefficients between high vs low PIM/budget transparency groups (example p-values: 11.86 (0.001)*** reported in table footnotes for tests).

### Policy implications and conclusions
- Empirical summary:
  - Larger government financing needs, lower budget transparency, and weaker bureaucratic efficiency are associated with higher probability of governments offering guarantees.
  - Guaranteed contracts are associated with higher dispute risk after PSM-IPTW adjustment (hazard ratio equivalent reported as 4.3); concessions also carry higher dispute risk.
  - Higher PPP management quality (PIM) and better budget transparency significantly reduce dispute risk.
  - MFI involvement and stronger political institutions reduce dispute risk and the incidence of government-led disputes.
- Policy recommendations and scenarios:
  - Medium-term priorities:
    - Improve PPP selection and management capacity (PIM quality) and enhance budget transparency to reduce adverse selection and hidden contingent liabilities.
    - Strengthen regulatory quality and bureaucratic efficiency to lower reliance on guarantees and reduce dispute risk.
  - Short-term measures:
    - Draft contracts with support from multilateral financial institutions (MFIs) to leverage experience, increase reputational costs of government breaches, and mitigate dispute risk.
  - Regional considerations:
    - In matured PPP markets (developing Asia and Latin America), high use of guarantees increases fiscal risk and dispute incidence; vigilance and institutional strengthening needed.
    - In sub-Saharan Africa, concessions predominate and are riskier due to weak PIM quality; as guarantees increase in SSA, dispute risk may rise over next 5-10 years.
- Final takeaway:
  - Improving structural weaknesses in budget transparency and PPP public investment management is a prerequisite for maximizing PPP efficiency and minimizing fiscal risks; MFI involvement provides an interim risk-mitigation tool.

*Source: wp17243 - section III. It estimates the probability of disputes conditional on the duration of contract ܣ (IMF working paper text provided).*

### References

### References

### Key literature cited
- Acemoglu, D. and Robinson, J., 2005 Economic Origins of Dictatorship and Democracy, Cambridge University Press
- Arezki, R. and Gylfason, T., 2011, Resource Rents, Democracy and Corruption: Evidence from Sub-Saharan Africa, CESifo Working Paper #3575
- Austin, 2013, The Performance of Different Propensity Score Methods for Estimating Marginal Hazard Ratios. Statistics in Medicine, 32(16), pp. 2837–49.
- Besley, T. and S. Coate, 1997, An Economic Model of Representative Democracy. Quarterly Journal of Economics 112(1): 85–114.
- Besley, T. and Ghatak, M., 2010, Property Rights and Economic Development, in Chapter 68, Handbook of Development Economics, vol. 5
- Besley, T. and Kudamatsu, M., 2008, Making Autocracy Work. Institutions and Economic Performance, edited by Elhanan Helpman (Havard University Press, 2008).
- Besley, T. and Persson, T., 2009, The Origins of State Capacity: Property Rights, Taxation, and Politics, American Economic Review, vol. 99, no. 4, pp. 1218–44
- Besley, T. and Persson, T., 2010, State Capacity, Conflict, and Development, Econometrica, 78(1), pp. 1–34
- Cangiano, M., Alier, M., Anderson, B., Hemming, R., and Petrie, M., 2006, Public-private Partnerships, Government Guarantees, and Fiscal Risk. (Washington: International Monetary Fund).
- Dabla-Norris, E., Brumby, J., Kyobe, A., Mills, Z., and Papageorgiou, C., 2012, Investing in Public Investment: An Index of Public Investment Efficiency. Journal of Economic Growth. 17, pp. 235–66.
- Economist Intelligent Unit, 2012, 2014, 2015, Evaluating the Environment for Public-private Partnerships: The 2012, 2014, and 2015 Infrascope.
- Eden, M., Kraay, A., and Qian, R., 2012, Sovereign Defaults and Expropriations: Empirical Regularities, World Bank Policy Research Working Paper, #6218 (Washington: World Bank).
- Engel, E., Fischer, R. and Galetovic, A., 2014, The Economics of Public-Private Partnerships: A Basic Guide, Cambridge University Press, (New York: Cambridge University)
- Frankel, J., 2012, The Natural Resource curse: A Survey of Diagnoses and Some Prescriptions, in Chapter 2 of “Commodity Price Volatility and Inclusive Growth in Low-Income Countries” (Washington: International Monetary Fund).
- Flyvbjerg, B., 2009, Survival of the Unfittest: Why the Worst Infrastructure Gets Built and What We can do about It. Oxford Review of Economic Policy. 25(3), pp. 344–67.
- Gayat, E., Resche-Rigon, M., Mary JY., Porcher, R., 2012, Propensity Score Applied to Survival Data Analysis through Proportional Hazards Models: A Monte Carlo Study. Pharmaceutical Statistics. 11(3), pp. 222–29.
- Guasch, L., Laffont, J., and Straub, S., 2007, Concessions of Infrastructure in Latin America: Government-led Renegotiation. Journal of Applied Econometrics. 22, pp. 1267–94.
- Hart, O., 2003, Incomplete Contracts and Public Ownership: Remarks and an Application to Public-Private Partnerships, Economic Journal, 113, pp. 69–76.
- Humphreys, M. and Bates, R., 2005, Political Institutions and Economic Policies: Lessons from Africa. British Journal of Political Science, 35(03), pp. 403–28.
- IMF, 2014a, World Economic Outlook: Legacies, Clouds, Uncertainties, Chapter 3 “Is It Time for an Infrastructure Push? The Macroeconomic Effects of Public Investment”, (Washington: International Monetary Fund).
- IMF, 2014b, Fiscal Monitor: Public Expenditure Reform: Making Difficult Choices, Chapter 2 “Public Expenditure Reform” (Washington: International Monetary Fund).
- IMF, 2015, Toward More Efficient Public Investment (Washington: International Monetary Fund).
- Iossa, E. and Martimont, D., 2015, The Simple Micro-Economics of Public-Private Partnership. Journal of Public Economic Theory, 17(1), pp. 4–48.
- Jensen, N., 2008, Political Risk, Democratic Institutions, and Foreign Direct Investment. The Journal of Politics, 70(4), 1040–52.
- Jensen, N. and Johnston, N., 2011, Political Risk, Reputation and the Resource Curse, Comparative Political Studies, 44(6), pp. 662–88.
- Jensen, N., Johnston, N., Lee, C., and Sahin, A., 2017, Crisis and Contract Breach: The Domestic and International Determinants of Expropriation, mimeo
- Karlan, D. and Zinman, J., 2009, Observing Unobservables: Identifying Information Asymmetries with a Consumer Credit Field Experiment. Econometrica, 77(6), pp 1993–2008
- Li, Q., 2009, Democracy, Autocracy, and Expropriation of Foreign Direct Investment. Comparative Political Studies, 42(8), pp. 1098–208.
- Li, Q. and Resnick, A., 2003, Reversal of Fortunes: Democratic Institutions and Foreign Direct Investment Inflows to Developing Countries. International Organization, 57(1), pp 175–211.
- Maskin, E. and Tirole, J., 2008, Public-private Partnerships and Government Spending Limits, International Journal of Industrial Organization, 26(2), pp. 412–20.
- Milesi-Ferretti, G. M., 2003, Goods, Bad, or Ugly? On the Effects of Fiscal Rules with Creative Accounting, Journal of Public Economics, 88, pp. 377–94.
- Nose, M., 2014, Triggers of Contract Breach: Contract Design, Shocks, or Institutions? World Bank Policy Research Working Paper #6738 (Washington: World Bank)
- Nose, N., Queiroz, C., and Nose, M., 2017, Fiscal Commitments to Encourage PPP Projects in Transport Infrastructure, in “Transport Infrastructure and Systems” (London: Taylor & Francis Group)
- Osborne, M. J., and A. Slivinski., 1996, A Model of Political Competition with Citizen-Candidates. Quarterly Journal of Economics 111(1): 65–96.
- Rosenbaum, P. and Rubin, D., 1983, The Central Role of the Propensity Score in Observational Studies for Causal Effects, Biometrika, 70, pp. 41–50.
- Sabol, P. and Puentes, R., 2014, Private Capital, Public Good: Drivers of Successful Infrastructure Public-Private Partnerships, Brookings Institution (Washington)
- Tomz, M. and Wright, M., 2010, Sovereign Theft: Theory and Evidence about Sovereign Default and Expropriation, in the Natural Resources Trap: Private Investment without Public Commitment, eds. Cambridge, MA: MIT Press, pp. 69–110
- Valero, V., 2015, Government Opportunism in Public-Private Partnerships. Journal of Public Economic Theory, 17(1), pp. 111–35.
- Wang, R., Irwin, T., and Murara, L., 2015, Trends in Fiscal Transparency: Evidence from a New Database of the Coverage of Fiscal Reporting. IMF Working Paper WP/15/188 (Washington: International Monetary Fund).
- Woodhouse, E., 2006, The Obsolescing Bargain Redux? Foreign Investment in the Electric Power Sector in Developing Countries, International Law and Politics, vol. 38, 121.
- World Bank, 2014a, The Power of Public Investment Management: Transforming Resources into Assets for Growth (Washington: World Bank).
- World Bank, 2014b, Public-Private Partnerships Reference Guide version 2, World Bank Institute and PPIAF (Washington: World Bank).
- World Bank, 2016, Benchmarking Public-Private Partnerships Procurement 2017 (Washington: World Bank).

### Empirical and methodological sources emphasized
- Propensity score methods and survival/hazard models: Rosenbaum and Rubin (1983); Austin (2013); Gayat et al. (2012); Karlan and Zinman (2009).
- PPP-specific microeconomics, contracts, guarantees, and fiscal risk: Cangiano et al. (2006); Engel, Fischer, and Galetovic (2014); Iossa and Martimont (2015); Maskin and Tirole (2008); Valero (2015).
- Political economy and resource curse literature relevant for expropriation and sovereign risk: Acemoglu and Robinson (2005); Frankel (2012); Jensen (2008); Li (2009); Tomz and Wright (2010).

---

### Technical Annex — A. Country Groupings (sample selection)
- From all countries available in PPI database, the study chooses 113 emerging and developing countries (categorized by income groups).
- Table A1 lists countries with number of PPP contracts next to country name. Select entries and counts (exact formatting preserved):
  - Afghanistan7 Angola8 Albania15
  - Bangladesh59 Armenia7 Algeria22
  - Benin8 Belize3 Argentina178
  - Burkina Faso3 Bhutan2 Azerbaijan7
  - Burundi4 Bolivia17 Belarus4
  - Cambodia31 Cameroon8 Bosnia and Herzegovina4
  - Central African Republic3 Cape Verde2 Botswana3
  - Chad3Congo, Rep.7Brazil775
  - Comoros3 Cote d'Ivoire16 Bulgaria48
  - Congo, Dem. Rep.10 Djibouti3 Chile194
  - Ethiopia3 Egypt, Arab Rep.21 China1,148
  - Gambia, The2 El Salvador12 Colombia136
  - Guinea7 Georgia15 Costa Rica40
  - Guinea-Bissau2 Ghana16 Dominican Republic29
  - Kenya20 Guatemala34 Ecuador27
  - Kyrgyz Republic6 Honduras30 Gabon12
  - Liberia8 India870 Iran, Islamic Rep.8
  - Madagascar9 Indonesia110 Jamaica14
  - Malawi4Iraq13Jordan27
  - Mali2 Lao PDR29 Kazakhstan8
  - Mozambique12 Lesotho1 Lebanon3
  - Myanmar7 Mauritania2 Lithuania8
  - Nepal36Moldova5Macedonia, FYR4
  - Niger4 Mongolia4 Malaysia107
  - Rwanda9 Morocco20 Maldives3
  - Sierra Leone7 Nicaragua17 Mauritius12
  - Tajikistan8 Nigeria55 Mexico260
  - Tanzania25 Pakistan87 Namibia1
  - Togo7 Papua New Guinea3 Panama26
  - Uganda27 Paraguay5 Peru109
  - Zimbabwe5 Philippines119 Romania42
  - Senegal16 Russian Federation119
  - Sri Lanka87 Seychelles3
  - Sudan6 South Africa79
  - Swaziland1 Thailand153
  - Syrian Arab Republic3 Tunisia9
  - Tonga3 Turkey176
  - Turkmenistan1 Uruguay36
  - Ukraine30 Venezuela, RB12
  - Uzbekistan8
  - Vietnam86
  - Yemen, Rep.8
  - Zambia7
- Total number of PPPs3411,7973,861 (exact string preserved)
- Income group counts shown: Low income (31)Lower middle income (43)Upper middle income (39)

### Technical Annex — B. Map
- Figure A1: Spatial Distribution of PPPs and the Disputes (in 1985–2015)
- Source: World Bank PPI database

### Technical Annex — C. Summary Statistics (dataset scope and highlights)
- Full sample includes 5,999 PPP contracts.
- Most contracts signed in developing Asia and Latin American countries and in energy or transport sectors.
- Total disputes occurred for about 6½ percent of the contract.
- Guarantees: only 13 percent of contracts receive guarantees on average.
  - Guarantees provided mainly with payment guarantee (80 percent), revenue guarantee (11 percent), or tariff rate guarantee (4 percent).
- Direct subsidies provided to about 12 percent of contracts.
- MFI involvement: 13 percent of contracts involved MFIs (more common before 2000s).
- Local investor participation: about 65 percent of contracts in MICs and LICs fully or partially involved local investors.
  - Developing Asia local sponsorship: 78 percent (about 80-90 percent in India, Malaysia, China, Thailand, and Sri Lanka).
- Cyclical macro averages in the sample:
  - Average GDP growth rate: 5.4 percent
  - Average inflation (CPI): 9.3 percent
  - GFN (in percent of GDP) average: 11 percent with range 0.3 to 102.2 percent
  - Public debt (in percent of GDP) average: 49.2 percent
- Political institutions:
  - Many countries are democratic (democracy score above 10)
  - Average length of national leaders in office: about 5 years with range 1 to 34 years
- Commodity exporter classification: about 23 percent

### Table A2. Summary Statistics, Full Sample (selected exact values)
- Sample size and key variable moments (N, Mean, Std. dev, Median, Min, Max):
  - Disputes: N 5,999 Mean 0.064 Std. dev 0.244 Median 0 Min 0 Max 1
  - Government-led disputes: N 4,598 Mean 0.044 Std. dev 0.205 Median 0 Min 0 Max 1
  - Contract duration (year): N 5,999 Mean 9.1 Std. dev 6.5 Median 8.0 Min 0.5 Max 34
  - Investment size (in mil USD): N 5,999 Mean 339.4 Std. dev 1,151.1 Median 77.5 Min 0.0 Max 35,586.5
  - Sponsor with same nationality: N 5,999 Mean 0.646 Std. dev 0.478 Median 1 Min 0 Max 1
  - Government guarantees: N 5,999 Mean 0.127 Std. dev 0.333 Median 0 Min 0 Max 1
  - Direct subsidy: N 5,999 Mean 0.121 Std. dev 0.326 Median 0 Min 0 Max 1
  - Concession: N 5,999 Mean 0.271 Std. dev 0.444 Median 0 Min 0 Max 1
  - MFI supports: N 5,999 Mean 0.126 Std. dev 0.332 Median 0 Min 0 Max 1
  - Log GDP per capita in 2000: N 5,940 Mean 8.487 Std. dev 0.635 Median 8.331 Min 5.790 Max 9.762
  - Real GDP growth (%): N 5,998 Mean 5.4 Std. dev 2.3 Median 5.6 Min -9.9 Max 11.0
  - CPI inflation (%): N 5,997 Mean 9.3 Std. dev 10.2 Median 7.7 Min -1.0 Max 135.0
  - Democracy: N 5,979 Mean 13.8 Std. dev 6.3 Median 17.3 Min 1 Max 20
  - Duration of national leader (year): N 5,992 Mean 5.3 Std. dev 3.5 Median 4.5 Min 1 Max 34
  - GFN/GDP (%): N 5,967 Mean 10.9 Std. dev 7.0 Median 11.8 Min 0.3 Max 102.2
  - Public debt/GDP (%): N 5,958 Mean 49.2 Std. dev 20.5 Median 42.7 Min 0.0 Max 163.4
  - PPP capital stock/GDP (%): N 5,681 Mean 3.6 Std. dev 5.4 Median 2.2 Min 0.0 Max 94.5
  - Commodity exporter: N 5,999 Mean 0.232 Std. dev 0.422 Median 0 Min 0 Max 1
  - Regulatory quality: N 5,999 Mean -0.109 Std. dev 0.495 Median -0.219 Min -2.410 Max 1.483
  - Bureaucratic efficiency: N 5,999 Mean -0.042 Std. dev 0.445 Median -0.061 Min -1.970 Max 1.216
  - PPP PFM quality 1/: N 5,853 Mean 0.001 Std. dev 1.000 Median -0.173 Min -3.016 Max 1.527
  - PPP operational quality 1/: N 5,853 Mean 0.001 Std. dev 0.999 Median 0.555 Min -2.889 Max 0.972
  - Budget transparency index (flow) 1/: N 5,999 Mean 0.009 Std. dev 0.995 Median -0.040 Min -1.830 Max 2.666
  - Budget transparency index (stock) 1/: N 5,999 Mean 0.008 Std. dev 0.997 Median -0.173 Min -1.538 Max 2.928
- Sector dummies (N 5,999):
  - Oil, gas, and mining Mean 0.051 Std. dev 0.221 Median 0 Min 0 Max 1
  - Energy Mean 0.454 Std. dev 0.498 Median 0 Min 0 Max 1
  - Transport Mean 0.259 Std. dev 0.438 Median 0 Min 0 Max 1
  - Water Mean 0.129 Std. dev 0.335 Median 0 Min 0 Max 1
  - Information and Communication Mean 0.107 Std. dev 0.309 Median 0 Min 0 Max 1
- Region dummies (N 5,999):
  - Developing Asia Mean 0.491 Std. dev 0.500 Median 0 Min 0 Max 1
  - Eastern Europe and CIS Mean 0.087 Std. dev 0.281 Median 0 Min 0 Max 1
  - Latin America and the Caribbean Mean 0.326 Std. dev 0.469 Median 0 Min 0 Max 1
  - Middle East and North Africa Mean 0.024 Std. dev 0.152 Median 0 Min 0 Max 1
  - Sub-saharan Africa Mean 0.073 Std. dev 0.260 Median 0 Min 0 Max 1

### Technical Annex — D. Balancing Test of PSM (key results)
- Table A3 reports balancing tests before and after Propensity Score Matching (PSM) for two treatments: guarantees and direct subsidies.
- After PSM, almost all contract characteristics are well balanced for both treatments.
- Indicators of matching quality:
  - Pseudo-R square declined close to zero after matching (from 0.200 to 0.013 for guarantees; from 0.177 to 0.007 for direct subsidy).
  - Mean standardized bias significantly declined (from 19.1 to 2.9 for guarantees; from 21.8 to 2.2 for direct subsidy).
- Selected balancing comparisons (Guarantee treatment, before vs after, exact figures preserved):
  - Ln(investment size) Before: Treated 4.25 Control 4.25 %bias -0.10 tp>|t| -0.02 0.99; After: Treated 4.24 Control 4.34 %bias -5.30 tp>|t| -1.06 0.29
  - Ln(GDP per capita in 2000) Before: Treated 8.42 Control 8.50 %bias -11.60 tp>|t| -2.87 *** 0.00; After: Treated 8.43 Control 8.43 %bias -1.10 tp>|t| -0.22 0.83
  - Ln(GFN/GDP) Before: Treated 2.44 Control 2.37 %bias 15.20 tp>|t| 3.91 *** 0.00; After: Treated 2.44 Control 2.44 %bias -0.10 tp>|t| -0.02 0.99
  - Ln(public debt/GDP) Before: Treated 3.88 Control 3.81 %bias 15.50 tp>|t| 3.69 *** 0.00; After: Treated 3.88 Control 3.87 %bias 4.40 tp>|t| 0.97 0.33
  - Regulatory quality Before: Treated -0.14 Control -0.10 %bias -6.90 tp>|t| -1.74 * 0.08; After: Treated -0.14 Control -0.14 %bias -0.50 tp>|t| -0.10 0.92
  - Sector dummies showed large imbalances before matching that were reduced after matching (e.g., Energy, Transport).
- Note: The PSM procedure includes square terms of ln(investment size), ln(GDP per capita in 2000), GFN, and ln(public debt). 1/ Mean distance in marginal distribution of the covariates as suggested by Rosenbaum and Rubin (1985).

### Technical Annex — E. Derivation of the Shadow Price (mathematical first-order condition)
- Dynamic optimization of Eq. (2) yields a first-order condition with respect to decision variable (symbol shown in original).
- Using the condition with the budget constraint and Eq. (3), the shadow price is expressed (full symbolic expression preserved in source).
- Partial derivatives of the shadow price yield three comparative statics with sign results:
  - First comparative static: derivative expression sign = 0 (as in source)
  - Second comparative static: derivative expression sign = 0
  - Third comparative static: derivative expression sign = 0
- Concludes with mathematical end-of-proof marker ∎

### Technical Annex — F. Role of PPP Investment Management in case of Concessions (empirical finding)
- Main empirical insight:
  - PPP investment management (PIM) quality and budget transparency are important in avoiding adverse selection of guaranteed projects.
  - Concessions tend to increase disputes, especially in SSA countries.
  - Interaction evidence: dispute risk decreases as a country improves its PPP investment management practices.
- Table A.4: Effect of Concessions on Disputes: by the Quality of Fiscal Institutions (selected coefficients, exact values preserved)
  - Concessions: 1.805*** [0.362] in column (1); 1.522*** [0.344] in column (2)
  - x PPP overall PIM quality index: -0.797** [0.336]
  - x PPP operational quality index: -1.180*** [0.324]
  - Duration dependence Ln(α): 0.096 [0.138] in (1); 0.114 [0.155] in (2)
  - Observations: 5,741; Sample: All; PSM-IPTW applied: Y/Y; Sector and region dummies: Y/Y; All disputes
- Note: Regression controls include ln(investment size), MFI supports, democracy, real GDP growth, and CPI inflation as in the baseline specification. * Significant at 10%; ** 5%; *** 1%. Robust standard errors are clustered at industry and contract year levels (reported in brackets).

*Italic: Source — wp17243 - References (IMF working paper PDF).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2017/wp17243.pdf_