Skip to main content

Advertisement

SpringerLink
Log in

On the economic costs of political instabilities: a tale of sub-Saharan Africa

  • Published:
Empirical Economics Aims and scope Submit manuscript

Abstract

The relationship between political instability (PI) and economic development has been extensively debated in the social sciences, yielding contrasting empirical results. In this article, we synergise two novel techniques—synthetic control (SC) and matrix completion (MC)—to study the economic effects of PI events on the sub-Saharan African scene. Our identification strategies capture potential spatial and temporal heterogeneities of treatment effects. We show that PI inflicts statistically and economically significant collateral damage, albeit heterogeneous across economies and over time. Notably, the group average treatment effect on the treated (ATT) reveals output shrinkage of approximately 17–20 percentage points. Exploiting the temporal dimension of the two techniques, we find that although output loss persists into the long run, the bulk of the damage is wrecked in the short run. This finding particularly casts doubt on the full economic recovery hypothesis in the near aftermath of a PI episode. Furthermore, we unearth spatial heterogeneities of the effects where we show that the negative effects are disproportionately more pronounced for economies which experience protracted PI episodes. Lastly, our paper demonstrates that while both methods provide reliable counterfactuals, the MC estimator yields significantly better pre-treatment fit in our data than the SC framework.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13
Fig. 14

Similar content being viewed by others

Availability of data and material

All data and materials leveraged in the analysis are publicly available, and all the sources to draw the data are disclosed in the manuscript.

Notes

  1. We use the term SSA region broadly to encompass all economies wholly or partially located to the south of the Sahara desert–including Djibouti and Somalia, whose categorisation varies across different sources.

  2. See also, inter aliaBenhabib and Rustichini (1996); Devereux and Wen (1998); Svensson (1998); Murdoch and Sandler (2002).

  3. See, for instance, the control units used in Costalli et al. (2017) and Matta et al. (2022).

  4. We relax this assumption in our empirical application.

  5. In our analysis, as we document below, \(j = 1\) could be any of the nineteen SSA countries which experienced PI episodes during the 1980–2013 window, while the donor pool comprises of all other African countries that did not experience any PI episode during the entire study period (1965–2013).

  6. Please see Liu et al. (2022) for a detailed formal discussion of the model specifications.

  7. See Marshall and Cole (2009) and Themnér and Wallensteen (2012) for the updated versions of the respective databases.

  8. Our choice of a decade-long post-treatment window is also inspired by Abadie et al. (2015).

  9. We assume that the effects of the prior PI episode may not erode away within that 5-year horizon (see, e.g. Kang and Meernik 2005; Collier 1999; De Groot et al. 2022).

  10. Similar inclusion criteria were used by Adhikari et al. (2018) in their study of reform waves using the SC framework.

  11. Although Eritrea also meet our case selection criteria, the country is dropped due to data paucity.

  12. This precludes Angola, Benin, Chad, Mauritius, Morocco, Mozambique, Nigeria, Uganda and Zimbabwe.

  13. Alternatively, one is suggested to employ the Demeaned synthetic control (due to Ferman and Pinto 2021) in order to circumvent this challenge. Note that this paper does not pursue this route since our goal is to compare the prototype SC against its prototype MC counterpart.

  14. See Abadie (2021) for a related discussion on data requirements in the practical application of SC frameworks.

  15. Note that our results remain robust to the exclusion of exogenous matching covariates (see Sect. 5.2.3).

  16. For the SC framework, we exploit the distribution of post-/pre-treatment RMSPE ratio to estimate the p-value following Abadie et al. (2010) and Galiani and Quistorff (2017), as we discuss in Sect. 5. The p-value for the MC estimator is borrowed from Liu et al. (2022).

  17. Please refer to Ben-Michael et al. (2022) for a formal contextual discussion of the SSC variant.

  18. We also attempted this exercise using the interactive fixed effects counterfactual (IFEct) approach, whose results remain robust to those of the MC estimator. Nevertheless, the IFEct estimator is outperformed by the latter. Results for the IFEct estimations may be obtained from the authors upon a written request.

  19. All MC estimates are based on unit-level bootstraps with 200 replications.

  20. In fact, Hodler (2019) provides evidence of full economic recovery 17 years following the 1994 Rwandan genocide.

  21. A subset of treated countries that share physical borders with members of the donor pool includes: Burundi, Burkina Faso, Central African Republic, Republic of Congo, Gambia, Mali and Rwanda.

  22. We also report the results of this exercise for the SC framework in Fig. 16. Indeed, the SC results largely corroborate those obtained by the MC estimator.

  23. These results can be accessed from the authors upon request.

  24. We owe an anonymous referee for this observation.

References

Download references

Acknowledgements

The authors are grateful to the Editor and the two anonymous referees for the constructive comments that substantially improved the paper. The authors are also indebted to Hung-Ju Chen, Diana Carillo, Dinarti Tarigan, Shu-Shiuan Lu, Ziyi Liu and participants at the 2022 Taiwan Economics Association Conference for helpful suggestions and discussions. The usual disclaimer applies.

Author information

Authors and Affiliations

  1. Department of Economics, National Dong Hwa University, Hualien, Taiwan

    Joe Maganga Zonda & Ming-Jen Chang

  2. Department of Economics, National Cheng Kung University, Tainan, Taiwan

    Chang-Ching Lin

Authors
  1. Joe Maganga Zonda
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Chang-Ching Lin
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Ming-Jen Chang
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Ming-Jen Chang.

Ethics declarations

Competing Interests

The authors declare that they have no conflict of interests.

Human and animal rights

This article does not contain any studies with human participants or animals performed by any of the authors.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendices

Appendix A: Donor pool weights for SC framework

This Appendix presents the weights of the control units used in the construction of the SC counterfactuals (Table 3).

Table 3 Donor weights in the construction of the SC counterfactuals
Full size table

Appendix B: Other robustness checks

See Figs. 15, 16, 17 and 18.

Fig. 15
figure 15

Effects of PI episodes on GDP (log): treated (black line) vs control units (grey lines). Notes: The horizontal axis illustrates the number of years relative to the onset of the PI episode

Full size image
Fig. 16
figure 16

SC Framework: leave-one-out distribution of the counterfactuals

Full size image
Fig. 17
figure 17

Dynamic treatment effect of PI: full donor pool vs leave-neighbours-out. Notes: The horizontal axis illustrates the number of years relative to the onset of the PI episode. For the MC estimator, the grey-shaded area represents the 90% confidence intervals obtained from the baseline estimates with full donor pool

Full size image
Fig. 18
figure 18

SC Framework: dynamic treatment effects of PI episodes on GDP growth rate. Notes: The dashed line plots the GDP growth rate computed by recalibrating the SC framework on the log difference of GDP. The solid (dark) line depicts GDP growth rate computed by taking the year-on-year change of the baseline treatment effect. For the latter scenario, in each country, the observations start nine years before the onset of the PI episode since the growth rate in this case is computed post-SC estimation which was originally confined to 10 years pre-PI episode

Full size image

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zonda, J.M., Lin, CC. & Chang, MJ. On the economic costs of political instabilities: a tale of sub-Saharan Africa. Empir Econ 66, 137–173 (2024). https://doi.org/10.1007/s00181-023-02452-4

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00181-023-02452-4

Keywords

JEL Classification

Search

Navigation