International Financial Institutions and the Promotion of Autocratic Resilience
International Financial Institutions and the Promotion of Autocratic Resilience (DOI: https://doi.org/10.1017/S0020818325101276) was written by Christina Cottiero and Christina J. Schneider in 2026. It was published in International Organization.
The authors examine the role of international financial institutions in supporting autocratic regimes.
IFIs are classified by membership composition.
'Universal' lenders (e.g., the World Bank) are excluded
- Authors note that the regional and cross-regional IFIs which remain are generally borrower-led, so are expected to be sensitive to the needs of borrowers.
V-Dem is used to construct democracy scores for member countries. Polyarchy scores below 0.5 are labeled autocratic.
- 'Consolidated' autocratic/democratic means consistently below/above the threshold over time
- 'Hybrid' means mixture
- Examples of consolidated autocratic IFIs: Arab Petroleum Investments Corporation; Development Bank of the Central African States (Banque de Développement des États de l'Afrique Centrale, BDEAC)
Expectation is that autocratic and hybrid IFIs differ from democratic IFIs in some way that causes them to contribute more to autocratic regimes. The authors justify this, and theorize a causal link, on the basis of negative externalities from localized regime instability. Conflict can bleed over borders, and regional security crises undermine confidence in the current order. Furthermore, autocratic leaders face greater risks in instability. Therefore autocratic leaders are comparatively more willing to lend money so as to stave off instability.
A motivating example is the CAR. In March 2013, Michel Djotodia overthrew Francois Bozizé. The World Bank immediately suspended disbursements and, almost a year later, released a loan with conditions about how the funds are used. In comparison, the members of the BDEAC met in June and by October had reactivated thr CAR account (frozen due to nonpayment) and approved substantial loans without conditions.
The authors start with the COW IGO dataset for the set of IFIs, although some regional development banks were added. The primary analysis uses the 18 IFIs identified as consolidated autocratic or hybrid. These IFIs have 143 borrower relationships in the range of 1967 to 2021.
Authors compile data on loans from OECD and AidData, with several additions from primary sources. Values are adjusted to 2011 levels and logged. In circumstances where a loan has multiple recipient countries, and some amount of the funds are not earmarked for specific recipients, then that value is excluded. The Trade and Development Bank (TDB) and West African Development Bank (Banque Ouest Africaine de Développement, BOAD) are the most problematic IFIs in this manner, but authors note that the analysis is robust to excluding them entirely.
The Uppsala Conflict Data Program (UCDP) collects data on the number of armed conflicts involving the state and domestic nonstate actors, ranging from 0 to 12. The authors derive a 'Anti-Government Conflict' indicator from this.
As a proxy for international (democratizing) regime pressure, the authors sum the values of loans reported by OECD in the 'government and civil society' sector. This sum is also then logged. Data availability of this variable becomes a limiting factor, forcing analysis to focus on 1990 onwards. The authors call this 'Democracy Promotion Aid' and they do explore alternative proxies in an appendix.
The authors model the logged loan value per year of these 143 borrower relationships. Autocratic and democratic recipients are modeled separately. They find that, for autocratic recipients, Anti-Government Conflict and Democracy Promotion Aid have significant and positive effects on loan amount. For democratic reciepients, only Democracy Promotion Aid does.
- Fixed effects for IFI, and also for region-year pairs. The authors use reghdfe.
- Controls include...
- GDP and GDP per capita are collected from the World Bank. National GDP is then logged.
- Similarity of UNGA voting between recipient and U.S. ('Support for the Liberal International Order')
- Logged distance between IFI headquarter country and recipient country
- Also a model estimated with the entire sample, and including interactions of V-Dem Polyarchy scores with each explanatory variable. Results are again consistent with the conclusion that only autocratic recipients feature an effect from Anti-Government Conflict.
The authors also ran a variety of models with additional predictors, e.g. the V-Dem Polyarchy score, or alternate specifications.
Lastly, the authors repeat the model with the set of democratic IFIs. Like with democratic recipients, there is a positive and significant effect from Democracy Promotion Aid only.