Identify the price shocks
Use separately identified export- and import-price innovations, distinguished from global-demand and global-supply disturbances in the published shock dataset.
Sovereign debt can rise even when the terms of trade do not.
The terms of trade compare the prices a country receives for its exports with the prices it pays for imports. They measure relative prices but conceal movements common to both.
When export and import prices fall by the same proportion, their ratio is unchanged on impact. Yet falling export and import prices can still affect government finances.
I ask whether combining these two prices into a single indicator loses information that matters for sovereign debt. The paper tests two possibilities: whether the source of a price movement matters, and whether joint price declines create debt pressure that the ratio misses.
The common factor k cancels. Both prices can fall sharply while the terms of trade record no change on impact.
See the debt consequencesThe baseline combines country-specific structural price shocks with panel local projections for twenty African economies. The shock sample covers 1990–2019.
Use separately identified export- and import-price innovations, distinguished from global-demand and global-supply disturbances in the published shock dataset.
Include both shocks in the same regression and estimate how gross government debt changes at each horizon. Country effects and lagged controls account for persistent differences and earlier economic conditions.
Scale each individual shock to a specified change in its corresponding price. For a common decline, find the combination of shocks that jointly lowers both prices. Compare these results with a terms-of-trade-only projection.
The baseline uses the published posterior-median structural shocks of Di Pace, Juvenal, and Petrella. Their identification uses sign, relative-elasticity, and narrative restrictions. The debt projections include country fixed effects and two lags of the outcome, price shocks, domestic growth, inflation, the real exchange rate, and world growth. Contemporaneous growth, inflation, and exchange rates are allowed to transmit the shock.
The outcome is the debt ratio at horizon h minus its value in the year before the shock. Forward outcome windows can extend beyond 2019 even though the price-shock sample ends there. Interpretation depends on the structural dataset’s identification assumptions.
Algeria, Burkina Faso, Cameroon, Chad, Côte d’Ivoire, Democratic Republic of the Congo, Egypt, Equatorial Guinea, Gabon, Ghana, Kenya, Madagascar, Malawi, Mauritius, Morocco, Niger, Nigeria, Senegal, South Africa, Sudan.
Coverage follows bilateral price-data availability, rather than a country’s debt-distress status. Source: Appendix Table A1.
Choose a price shock to see what the terms-of-trade ratio records—and what the paper estimates happens to debt.
Export and import prices each fall by 10%, leaving their ratio unchanged on impact.
Terms of trade = export prices ÷ import pricesPoint estimates after four years
95% interval: 2.7 to 9.2 pp
pp = percentage points of GDP · Line shows the 95% confidence interval.
The separate-price model captures debt pressure that the unchanged ratio misses.
20 African economies · 1990–2019 · Point estimates: Figure 3. Confidence intervals: Table 3 and Section 4.2.3. Shock sizes follow the paper’s log-change convention. Read the paper for estimation and uncertainty ↗
The export and import scenarios each scale one identified structural shock to a 10% move in its own price. The other price is allowed to respond. The common-decline scenario instead uses a combination of shocks that jointly lowers both prices by 10% on impact. Its debt response is therefore computed separately, rather than by adding the two rescaled single-shock estimates.
Both prices fall 10%. Terms of trade are unchanged on impact. Estimated debt rises 6.0 percentage points of GDP; the terms-of-trade-only model predicts no change.
When both prices fall equally on impact, the terms-of-trade-only model predicts no debt response at any horizon. This benchmark does not imply that the subsequent terms-of-trade path remains flat. The separate-price estimates show a debt increase that persists through year four.
95% interval: 2.73 to 9.20 pp. The interval excludes zero.
Table 3 · Joint 10% export- and import-price decline · pp of GDP. The chart uses the reported estimates and intervals at years 0–4; connecting lines guide the eye. The four-year estimate of 5.96 is rounded to 6 in the headline result.
| Years after shock | Export-price decline | Import-price increase | ToT-only: single shock | Joint decline | Joint decline: 95% interval |
|---|---|---|---|---|---|
| 0 | 2.66 | −0.38 | 1.56 | 1.67 | −1.65 to 5.00 |
| 1 | 3.98 | −2.62 | 2.40 | 4.39 | 0.88 to 7.90 |
| 2 | 4.25 | −4.05 | 2.26 | 5.83 | 2.36 to 9.30 |
| 3 | 4.45 | −3.14 | 3.05 | 5.10 | 1.91 to 8.28 |
| 4 | 5.71 | −3.40 | 3.93 | 5.96 | 2.73 to 9.20 |
Single-shock columns rescale an own-price structural shock; cross-price responses are allowed. The joint decline is constructed separately. The ToT-only column applies to the single-shock comparisons; its joint-decline prediction is zero.
The historical data document nine country-year episodes in which both price indices had 100 × log changes below −10. Their median four-year debt increase was 13.1 percentage points of GDP. Individual outcomes varied.
Senegal, 2009: both prices fell sharply, while their relative movement was small. The observed debt ratio rose by 17.8 percentage points of GDP.
Historical episode data · All price changes are 100 × Δlog(price), not ordinary percentage changes; debt changes are pp of GDP. These are historical observations with no controls or causal identification. They are distinct from the estimated 10% shock effects above.
| Country / year | Export-price log change | Import-price log change | ToT log change | Observed debt change |
|---|---|---|---|---|
| Cameroon · 2015 | −30.3 | −15.0 | −15.3 | 20.9 pp |
| Senegal · 2009 | −21.1 | −19.3 | −1.8 | 17.8 pp |
| South Africa · 2009 | −16.1 | −12.0 | −4.1 | 16.3 pp |
| Morocco · 2009 | −12.9 | −15.3 | 2.4 | 15.1 pp |
| Egypt · 2009 | −28.5 | −14.1 | −14.4 | 13.1 pp |
| South Africa · 2015 | −12.1 | −10.8 | −1.3 | 12.8 pp |
| Cameroon · 2009 | −26.6 | −11.0 | −15.6 | 6.3 pp |
| Egypt · 2015 | −12.1 | −10.1 | −2.0 | −0.8 pp |
| Burkina Faso · 1998 | −18.6 | −11.4 | −7.2 | −5.1 pp |
The table reports nine selected historical episodes. The Democratic Republic of the Congo in 2009 is excluded because subsequent debt relief sharply reduced its debt stock.
The paper separates the cumulative contribution of primary balances from the rest of the debt response. The export and import shocks show different accounting patterns.
Section 4.4 · pp of GDP per positive one-standard-deviation structural shock. These units differ from the 10% scenarios above. The decomposition uses point estimates; components can differ from the total by rounding.
| Positive structural shock | Debt response | Primary-balance contribution | Accounting remainder |
|---|---|---|---|
| Export | −5.49 | −2.36 | −3.14 |
| Import | −1.24 | 0.97 | −2.21 |
The remainder includes debt-ratio movements outside the cumulative primary-balance contribution. The remainder combines several possible channels; it does not identify any one mechanism.
The positive import shock’s primary-balance contribution raises debt, while its total debt response is negative. The remainder therefore dominates the budget contribution. The current draft finds no support for a foreign-currency valuation explanation. Restricted-sample evidence supports a nominal-GDP contribution, but the available data do not separately identify nominal-GDP pass-through and debt transactions outside the primary balance.
Source: Section 4.4 and Appendix A.3. The four-year import-price interval in the 10% scenario includes zero; the point-estimate decomposition should be read alongside that uncertainty.
The findings point to a change in how external-price exposure is represented: retain the export and import paths, and evaluate scenarios in which the two move together.
The same terms-of-trade change can arise from different export- and import-price shocks. Export and import exposures need their own estimated debt responses.
Test simultaneous export- and import-price declines explicitly. An unchanged ratio does not establish that debt pressure is absent.
Primary balances alone do not account for the estimated debt movement. Examine nominal GDP and debt transactions alongside revenue and spending, and backtest the timing against country-specific evidence.
The estimated 6-percentage-point increase applies to this sample and scenario. It is not a calibrated forecast for every country. The design studies surprise shocks; anticipated price paths and the mechanisms behind the import-side response require further investigation.
The paper considers alternative structural-shock reconstructions, three shift-share designs using trade shares and world prices, and a WTI-based instrument for six oil exporters. The shift-share exercises also extend beyond Africa. These checks ask whether the findings depend on one identification strategy. They use different shock scales, so their numerical estimates are not directly comparable.
Sources: Section 4.3 and Appendix Tables A2–A5. Country and lag choices, global controls, persistence, and inference procedures are examined in Appendix A.4 and Table A9.
The manuscript contains the full specifications, estimates, and robustness checks behind this companion.
Sovereign debt can rise even when the terms of trade do not. Using separately identified export- and import-price shocks for twenty African economies over 1990–2019, I estimate that a 10 percent export-price decline raises gross government debt by 5.7 percentage points of GDP after four years, while an equal import-price increase lowers it by 3.4 points. A terms-of-trade regression reverses the import-price sign and removes common price movements. A simultaneous 10 percent decline in both prices leaves the ratio unchanged on impact but raises debt by 6 points. Export-price shocks transmit through delayed commodity-linked revenue; the import-price debt response falls predominantly outside the primary balance. Stress tests that summarize commodity exposure with the terms of trade therefore miss the debt movements that matter most when export and import prices fall together.
Dissertation: Essays on Fiscal Policy in Developing Economies
Explore the dissertation →
Revenue contributes to the export-side response.
A positive export-price shock lowers debt. Cumulative primary-balance changes account for about 43% of the four-year debt response. The revenue response peaks around one year after the shock; the budget does not account for the whole debt movement.