## Chapter 2. A Repeat of the Past? Fiscal Implications of Commodity Price Surges in the Middle East and North Africa — Online Annex

## Source details

**Canonical URL:** [Chapter 2. A Repeat of the Past? Fiscal Implications of Commodity Price Surges in the Middle East and North Africa — Online Annex](https://www.imf.org/-/media/files/publications/reo/mcd-cca/2022/october/english/ch2onlineanexxes.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/reo/mcd-cca/2022/october/english/ch2onlineanexxes.pdf.md)
- [Structured JSON version](/-/media/files/publications/reo/mcd-cca/2022/october/english/ch2onlineanexxes.pdf.json)

---

### Methodology: local projections and empirical specification
- Estimation approach: impulse response functions directly from local projections (Jordà 2005), with advantages noted by Stock and Watson (2007) and Auerbach and Gorodnichenko (2013) as a flexible alternative to VAR/ARDL.
- Estimated reduced-form equation (summarized from the source):
  - Dependent variable: yi,t+k — change of subsidies and transfers as a percent of GDP between time t – 1 and t + k.
  - Key regressor: interaction between country-specific socioeconomic variable xi,t and a large positive oil price shock si,t (coefficient of interest = β).
  - si,t defined as a dummy: si,t =1 if Δ%pt > median(Δ%pt+ ); si,t =0 otherwise.
  - Large positive oil price changes defined as annual percentage change above 26 percent.
  - xi,t proxies: measures of targeting of social protection programs and the Gini coefficient.
  - Low targeting dummy: equals 1 when beneficiary incidence for the lowest quartile of the income distribution is below 27.5 percent (the median beneficiary incidence across MENA countries), 0 otherwise.
  - Controls (zi,t): growth rate and lags of the country-specific commodity price shocks, forward leads of the interactions between time 0 and the end of the forecast horizon to correct local-projection bias (Teulings and Zubanov 2014).
  - Number of lags included (l) = 1; results robust to different lag length.
  - Fixed effects included; dynamic impacts interpreted relative to a country-specific trend.

### Data, sample, and variables
- Estimation sample: unbalanced panel of MENA oil-importing countries, annual frequency, from 2000 to 2020.
- Regressions including beneficiary incidence of social assistance use a restricted country sample due to limited data (see Table A1 for country-year coverage).
- Oil price measure: percentage change of the Brent spot market price in US dollars (month average).
- Subsidies and transfers and real GDP growth: from the Regional Economic Outlook: Middle East and North Africa data set.
- Beneficiary incidence of social protection programs: World Bank’s ASPIRE database.
- Years identified as large positive oil price change episodes for oil prices: 2004, 2005, 2008, 2010, 2011, 2018, and 2021 for oil prices.
- Table A1 (country and year coverage for beneficiary incidence indicator) lists countries and associated years of coverage (as in the source).

### Key empirical findings
- Empirical estimates of the dynamic response of price subsidies and transfers following a large positive shock to oil prices show:
  - A statistically significant stronger increase in subsidies in MENA countries with lower targeting of social protection programs.
  - Similar evidence for countries where inequality, as measured by the Gini coefficient, is above the regional median.
- Figures (cumulative impulse response functions) show:
  - Panel 1: interaction of positive (above-median) oil price shock with low targeting of social protection programs — dependent variable = subsidies and transfers, percent of GDP; 95% confidence bands reported.
  - Panel 2: interaction of positive (above-median) oil price shock with higher Gini coefficient (above MENA median) — dependent variable = subsidies and transfers, percent of GDP; 95% confidence bands reported.
- Interpretation: country-specific socioeconomic characteristics (low targeting and higher inequality) amplify the fiscal response (increase in subsidies and transfers) to large positive oil price shocks.

### Model interpretation and robustness
- Dynamic impacts should be interpreted relative to country-specific trends because fixed effects are included.
- Forward leads of interactions up to the forecast horizon (k) are included to correct bias in local-projection impulse responses.
- Results reported are robust to different lag lengths (baseline uses l = 1).

*Source: Chapter 2 Online Annex, ch2onlineanexxes (IMF PDF).*

---


_Source: https://www.imf.org/-/media/files/publications/reo/mcd-cca/2022/october/english/ch2onlineanexxes.pdf_