Fiscal Dynamics in Fragile Economies: What a 2025 Study Reveals

A digest of Ali & Nur (2025), Asian Journal of Economic Modelling

Somalia's public finances run on a narrow, import-dependent tax base and heavy inflows of external grants — a combination that makes the relationship between revenue and spending both consequential and hard to pin down. This study uses quarterly Somali Ministry of Finance data from 2013Q1 to 2024Q2 to model government spending, total revenue, grants, direct tax and indirect tax together in a Vector Autoregression (VAR), after finding no long-run cointegrating relationship among them. The result is a picture of short-run interdependence — revenue, grants and spending move together quarter to quarter — rather than a stable long-run equilibrium, with some of the paper's own headline claims worth checking against its tables before taking them at face value.

The study covers 2013Q1–2024Q2, giving 46 quarterly observations of Somalia's national fiscal data. It uses a VAR(3), a 5-variable system of spending, revenue, grants, direct tax and indirect tax. Total revenue and government spending are correlated at 0.94, and the government-spending VAR equation has an R² of 92% (Table 9).

Article at a Glance

The article is "Fiscal Dynamics in Fragile Economies: A VAR Analysis of Revenue Mobilization and Government Spending in Somalia," by Ali Yassin Sheikh Ali (Faculty of Economics, SIMAD University) and Abdulkadir Mohamed Nur (Central Bank of Somalia). It was published in the Asian Journal of Economic Modelling, Vol. 13, No. 2, pp. 278–295 (AESS Publications), in 2025 (received 28 April 2025; accepted 23 June 2025; published 3 July 2025). Its research area is public finance and fiscal policy in fragile and conflict-affected economies, and its study context is Somalia, using quarterly national fiscal data for 2013Q1–2024Q2 (46 observations) sourced from the Somali Ministry of Finance. The method is time series: ADF/PP/KPSS unit-root tests, Johansen and bounds cointegration tests, an unrestricted VAR, Granger causality, impulse response and variance decomposition. The DOI is https://doi.org/10.55493/5009.v13i2.5450.

The Research Problem

Somalia's fiscal system is still being rebuilt after decades of conflict: tax administration is weak, compliance is low, and the informal, import-heavy economy pushes revenue collection toward indirect taxes such as customs duties rather than direct income or corporate tax. Heavy reliance on external grants offers short-term relief but, the authors argue, can blunt the incentive for domestic revenue reform and expose the budget to donor volatility. Prior literature mostly studies taxation and expenditure as separate problems; the authors argue that few studies examine how direct tax, indirect tax and grants jointly shape government spending in a fragile-state setting like Somalia's.

Research Objective

The study sets out to identify the factors influencing government expenditure in Somalia, specifically the roles of indirect revenue, direct revenue and external grants, using a Vector Autoregression model to capture short-run relationships among government spending (GS), total revenue (TR), grants (GR), direct tax (DT) and indirect tax (IT). All variables are logged and expressed quarterly in millions of dollars.

How the Study Was Conducted

Quarterly data for 2013Q1–2024Q2 (46 observations) on government spending, total revenue, grants, direct tax and indirect tax were sourced from the Somali Ministry of Finance. Because unit-root tests showed a mixed order of integration and cointegration tests found no long-run relationship among the variables, the authors use an unrestricted VAR rather than a vector error-correction model, focusing the analysis on short-run dynamics and interdependencies.

The analysis followed six steps. In step 01, Stationarity, ADF, PP and KPSS tests showed that government spending, revenue, grants and direct tax are stationary at level, I(0), while indirect tax is I(1). In step 02, Cointegration, Johansen trace and max-eigenvalue tests, plus a bounds test (F = 1.30, below the lower critical bound), found no long-run cointegration. In step 03, Lag selection, the FPE, AIC and HQ criteria selected lag 3, while the Schwarz criterion instead suggested lag 1, a discrepancy the authors resolve in favour of lag 3. In step 04, Estimation, an unrestricted VAR(3) was estimated jointly on logged government spending, revenue, grants, direct and indirect tax. In step 05, Causality, pairwise Granger causality tests and VAR Granger Causality/Block Exogeneity Wald tests were run across all five variables. In step 06, Diagnostics, serial correlation LM, residual normality and heteroskedasticity tests were used, along with impulse response functions and variance decomposition.

Key Findings

The VAR equation results are as reported in Table 9 of the article. The government spending equation (LGS) has R² = 0.92 and F = 19.77. The total revenue equation (LTR) has R² = 0.85 and F = 10.55. The grants equation (LGR) has R² = 0.68 and F = 3.84. The direct tax equation (LDT) has R² = 0.91 and F = 18.87. The indirect tax equation (LIT) has R² = 0.89 and F = 14.58.

1. No long-run relationship exists between revenue and spending — only short-run dynamics. Both the Johansen trace/max-eigenvalue tests and the bounds test (F = 1.30, well below the lower 10% critical bound of 2.45) fail to find cointegration among government spending, revenue, grants, direct tax and indirect tax. This is why the authors use a VAR rather than a VECM, and it means every association the paper describes should be read as a short-run, quarter-to-quarter relationship rather than a stable long-run equilibrium.

2. Grants and spending, and revenue and grants, are locked in two-way feedback loops. Pairwise Granger causality tests show bidirectional causality between grants and government spending (grants→spending p = 0.033, spending→grants p = 0.001) and between total revenue and grants (p = 0.005 and p = 0.006 respectively). External aid and domestic fiscal activity in Somalia appear to be mutually reinforcing rather than aid simply responding to revenue shortfalls.

Reading note: The abstract and results section describe the fitted government-spending model as having "a high R-squared value of 85%," but Table 9 reports an R² of 0.92 for the LGS equation — 85% is actually the R² of the total-revenue (LTR) equation. The F-statistic quoted alongside it (19.77) does match LGS, so this looks like a mismatched figure rather than a different model; worth checking Table 9 directly rather than citing the abstract's number.

1. Indirect tax does not directly Granger-cause government spending, despite the paper's framing. The abstract highlights indirect taxes as showing "a stable relationship with government spending," but the pairwise Granger test from indirect tax to spending is not significant (p = 0.198). Indirect tax does significantly Granger-cause grants (p = 0.003) and is highly correlated with direct tax (r = 0.92) and total revenue (r = 0.83) — the "stability" claim holds for those relationships, but not as direct causality into the spending equation itself.

2. Diagnostics are mostly clean, but not spotless. Serial correlation is not a major issue at lags 1–3, though lag 4 shows borderline evidence of it. The heteroskedasticity test passes only narrowly (χ² = 499.26, p = 0.06, just above the 5% cutoff), and residual normality is rejected for one VAR component (Jarque-Bera p = 0.01) as well as for the joint multivariate normality test (p = 0.03) — a caveat worth keeping in mind when reading the impulse-response and variance-decomposition results with confidence.

What the Study Contributes

The contribution is largely empirical and data-specific: a VAR model using Somalia's disaggregated fiscal data — separating direct tax, indirect tax and grants rather than treating revenue as one aggregate — to trace short-run interdependencies with government spending. The authors argue this fills a gap in fragile-state fiscal literature, which they say has tended to study taxation and expenditure as separate problems rather than as an interconnected system, particularly for Somalia specifically.

Important Limitations

Identified by the authors: The availability and reliability of Somalia's quarterly fiscal data is flagged as a significant shortcoming. The analysis is linear and does not account for sector-specific dynamics, which the authors say may limit its applicability beyond Somalia's particular fiscal environment. Future research is recommended to use more flexible SVAR models incorporating sector-level governance variables.

Skilful cautions arising from the study design: 46 quarterly observations feeding a 5-variable VAR(3) model — 15 lagged regressors plus a constant per equation — is a thin data-to-parameter ratio, a common constraint in small-sample macro VARs but one that limits confidence in individual coefficient estimates. Lag-length criteria disagree: FPE, AIC and HQ favour lag 3, while the Schwarz criterion favours lag 1, and the paper adopts lag 3 without showing the main results are robust to the shorter lag its own SC statistic recommends. The abstract's claim of an 85% R² for the government-spending model does not match Table 9's reported 0.92 for that equation (see Reading Note), so readers citing model fit should use the table figure. The claim that indirect taxes have "a stable relationship with government spending" is supported by correlation and variance-decomposition evidence but not by the direct pairwise Granger causality test from indirect tax to spending, which is not significant, so the specific causal pathway is narrower than the framing suggests. Finally, residual normality is rejected for one VAR component and jointly across the system, and the heteroskedasticity test passes only narrowly (p = 0.06), both of which suggest the confidence intervals around the impulse responses should be read with some caution.

Why This Research Matters

For Somalia and comparable aid-dependent, fragile states, the finding that grants and government spending are locked in a two-way feedback loop — not simply aid flowing in response to revenue gaps — has direct budget-planning implications: reducing aid dependency requires managing that mutual linkage, not just negotiating aid inflows. The paper's policy recommendations, on modernising tax administration, expanding indirect-tax coverage and diversifying revenue, follow from this short-run, interdependent picture of Somalia's fiscal system rather than from any long-run equilibrium relationship, since none was found.

Skilful Research Insight

Editorial commentary — not a finding of the study. Three things stand out. First, always check a paper's headline summary statistics against its own tables: the 92%/85% mismatch on the government-spending equation's R² is an easy figure to carry forward uncritically if you only read the abstract. Second, "no cointegration" is itself an important finding here — it means every association described in the abstract and discussion is a short-run relationship, not a stable long-run one, and should be read with that qualifier attached. Third, the pairwise Granger causality results are more specific than the abstract's language suggests: indirect tax significantly Granger-causes grants and revenue, but not government spending directly — which policy lever a reader should focus on (tax administration versus aid negotiation) depends on getting that pathway right.

Read the Original Research

The original article is "Fiscal Dynamics in Fragile Economies: A VAR Analysis of Revenue Mobilization and Government Spending in Somalia," published in the Asian Journal of Economic Modelling, 13(2), 278–295 (2025), by Ali Yassin Sheikh Ali and Abdulkadir Mohamed Nur. Its DOI is https://doi.org/10.55493/5009.v13i2.5450, and the publisher/journal page is AESS Publications, www.aessweb.com. Skilful summarises; all findings belong to the original authors.

SEO Information

The SEO title is "Fiscal Dynamics in Somalia: 2025 VAR Study Summary." The meta description is "A Skilful digest of a 2025 Asian Journal of Economic Modelling VAR study on revenue mobilization and government spending in fragile-economy Somalia." The URL slug is /fiscal-dynamics-fragile-economies-var-somalia. The primary keyword is fiscal dynamics fragile economies VAR Somalia. The secondary keywords are revenue mobilization government spending; Granger causality Somalia; indirect taxes fragile states; Johansen cointegration bounds test; aid dependency fiscal policy.