Context:
Since the 2011 Arab Spring, the IMF has become a near-constant actor in economic policymaking across several Middle East and North Africa countries. Tunisia, Egypt, Morocco, and Jordan each experienced a distinct post-2011 trajectory, full democratic transition, authoritarian restoration, monarchic reform, and externally induced structural strain, respectively, while repeatedly engaging with the Fund over the following decade. More than a decade later, persistent fiscal fragility across these cases raises the question of whether IMF engagement has produced durable consolidation or merely repeated temporary relief.
Research question:
To what extent, and under what conditions, do IMF programs improve public finances in MENA countries experiencing political transition?
Specific objectives:
* Evaluate a policy: measure the average effect of IMF program participation on fiscal health via a composite Fiscal Health Index built through PCA, across a 14-country panel (2011–2024).
* Analyze policy outcomes: compare qualitative trajectories across Tunisia, Egypt, Morocco, and Jordan to explain the heterogeneity behind the positive average effect.
* Provide recommendations: advocate for more politically informed conditionality design.
* Raise awareness: show that "do IMF programs work?" is underspecified , the better question is for whom, under what conditions, and through which channels.
Synthèse
Synthesis
This paper examines whether, and under what conditions, IMF programs improve public finances in Middle East and North Africa (MENA) countries undergoing political transition. The study focuses on four central cases — Tunisia, Egypt, Morocco, and Jordan — over the 2011–2024 period, in the wake of the Arab Spring, which transformed the IMF's relationship with the region from an occasional, technocratic engagement into a near-constant presence in economic policymaking.
Framework and research question
The four countries studied offer a rare comparative setting: each experienced a different post-2011 political trajectory (full democratic transition for Tunisia, authoritarian restoration for Egypt, gradual monarchic reform for Morocco, externally induced structural strain for Jordan), while being subject to broadly similar conditionality. This configuration allows the paper to test a long-standing question in international economics: do IMF programs produce durable fiscal consolidation, or merely repeated temporary relief? The study builds on prior work by Harrigan and El-Said (2010), which covered 1983–2004, by extending the analysis through the post-2011 decade and adding a quantitative panel dimension absent from the existing literature.
Methodology
Two approaches are combined. First, a detailed qualitative analysis of program content and implementation in the four countries, generating stylized facts on program type, political constraints, and external shocks. Second, a quantitative panel analysis covering 14 countries (the four MENA cases plus ten comparators from Eastern Europe, Asia, Lebanon, Algeria, Iraq, and Sub-Saharan Africa), totaling 193 country-year observations.
The central analytical tool is a composite Fiscal Health Index, built through Principal Component Analysis (PCA) from four standardized fiscal variables (general government gross debt, primary balance, total revenue, net lending/borrowing). The first principal component, explaining 41.1% of total variance, captures overall fiscal sustainability and serves as the basis for the index, rescaled to 0–100.
This index is then tested in a panel regression model:
Fiscal Health = α + β₁·(IMF Program) + β₂·(GDP Growth) + β₃·(Inflation) + β₄·(Real Interest Rate) + country effects + error
Four specifications are estimated (Pooled OLS, individual Fixed Effects, Two-way Fixed Effects, Random Effects), supported by a battery of specification tests. The Hausman test fails to reject the null (χ² = 3.05, p = 0.549), supporting the Random Effects model as the preferred specification, with country-clustered robust standard errors correcting for the serial correlation detected by the Wooldridge test.
Hypotheses
Four hypotheses guide the analysis:
(H1) IMF program participation is associated, on average, with improved fiscal health.
(H2) this effect is heterogeneous across program type and political context.
(H3) GDP growth is the most robust macroeconomic determinant of fiscal health.
(H4) IMF-related fiscal gains are temporary in the absence of sustained structural reform.
Main results
All four specifications show a positive, statistically significant association between IMF program participation and fiscal health.
The Random Effects model (preferred specification) estimates an effect of +4.29 points on the index (p < 0.05), notably smaller than the naive Pooled OLS estimate of +7.43 points. This gap is informative: nearly half of the raw association reflects selection bias (countries entering programs are systematically weaker fiscally), rather than a genuine causal effect.
GDP growth remains the most robust predictor across all specifications (roughly +1 index point per additional percentage point of growth), confirming H3. Inflation and real interest rates are not statistically significant in any specification, likely because their strong cross-country heterogeneity (near-zero inflation in Morocco versus Lebanese hyperinflation) is absorbed by country fixed effects.
Heterogeneity across the four MENA cases
The positive average effect masks sharply contrasting national trajectories:
* Egypt: the clearest and most positive program effect. The 2016 EFF ($12 billion, the largest in the region), combining currency flotation, VAT introduction, and subsidy removal, delivered a cumulative 5.5% of GDP improvement in the primary balance. The executive had the (authoritarian) political capacity needed to implement painful measures.
* Morocco: the most atypical and successful case. Four consecutive precautionary lines (2012–2018), never drawn upon, served as a credibility signal rather than a financing mechanism. The fiscal deficit fell from 7.3% to 3.6% of GDP, with gains persisting even after program exit.
* Tunisia: a trajectory of continued deterioration, despite five program episodes since 2011. Opposition from the UGTT labor union and political fragmentation blocked implementation of structural reforms (public wage bill, energy subsidies), culminating in President Kais Saied's rejection of a new agreement in 2022 and public debt reaching 83% of GDP.
* Jordan: macroeconomic stability was achieved (strong reserves), but underlying debt dynamics remained unresolved (public debt near 90% of GDP), due to major structural constraints (energy import dependence, the cost of hosting Syrian refugees).
Five cross-cutting patterns emerge from the qualitative analysis: (1) program type matters as much as content, undrawn precautionary arrangements (Morocco) sometimes outperform fully drawn programs; (2) political regime, more than reform depth, determines compliance; (3) the public wage bill is the most consistently unfulfilled conditionality across all four countries, regardless of regime type; (4) external shocks (COVID-19, the Ukraine war) systematically disrupt multi-year adjustment paths; (5) non-conditional alternative financing (Gulf states, China) lowers the cost of non-compliance, especially in Egypt and post-2022 Tunisia.
Conclusion and recommendations
The paper concludes that "do IMF programs work?" is the wrong question. The answer depends fundamentally on domestic political capacity to implement conditionality, a capacity that programs themselves can neither create nor guarantee. Key recommendations include:
Design more politically informed conditionality that accounts for each country's institutional context and implementation capacity, rather than applying uniform templates.
Recognize that undrawn precautionary arrangements (the Moroccan model) can generate durable credibility gains without the social costs of fully disbursed programs.
Take seriously the near-universal failure of public-wage-bill conditionality, which touches a legitimacy issue that MENA governments can least afford to address precisely when programs are signed.