## _wp08193

## Source details

**Canonical URL:** [_wp08193](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08193.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08193.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08193.pdf.json)

---

### Introduction and Key Conclusions
- Long-run drivers of inflation in Saudi Arabia and Kuwait:
  - Trading partners’ inflation is the main factor affecting domestic inflation.
  - Exchange rate pass-through has a significant but lower contribution.
- Short-run dynamics:
  - Positive demand shocks and excess money supply exert upward pressures on inflation in the short run but tend to dissipate quickly as real exchange rate and the money market reach a new equilibrium.
- Remedies discussed in the source:
  - Revaluation or adopting more flexible exchange rate regime to gain monetary policy independence.
  - Higher subsidies.
  - Addressing supply bottlenecks.
  - Containing government expenditures.

### Literature Review (selected points)
- Inflation treated as outcome of: supply-side (cost-push/mark-up), foreign factors, monetary factors, demand factors; inflation expectations also important.
- Empirical evidence cited (selected):
  - De Brouwer and Ericsson (1998): mark-up model for Australia.
  - Juselius (1992): strong dependence of Danish prices on West German price levels.
  - Lim and Papi (1997): money and exchange rates in Turkey during 1970–95.
  - Leo (2007): strong relation between money and inflation in Iran.
  - Al-Raisi and Pattanaik (2003): a depreciation of 10 percent in the NEER will increase the CPI by only 0.4 percent in Oman.

### Economic Model and Methodology
- Price level specification:
  - P = f(NEER, P*, Poil, M, D)
    - NEER = nominal effective exchange rate (foreign currency price per local currency; increase = appreciation)
    - P* = trading partners’ price level
    - Poil = price of oil
    - M = nominal money supply
    - D = aggregate domestic demand
- Baseline empirical specification: error correction model (ECM) using log variables with k lags; includes Δ terms for p, ne er, p*, oil, M, and demand.
- Excess measures:
  - Excess money supply: excm_t = m_t − m̂_t where m̂_t is estimated long-run money demand.
  - Excess domestic demand: excd_t = d_t − d̂_t where d̂_t is estimated potential demand (de-trending or HP-filter).
- Rationale for variable choices:
  - External factors emphasized due to hydrocarbon-dominated tradable sector and high import dependence for consumer goods.
  - Large expatriate workforce and flexible labor policies increase sensitivity to external prices.
  - Generous subsidies and pegged exchange rate regime limit active monetary policy and mitigate some domestic inflationary pressures.

### Data, Integration Order, and Sample Coverage
- Sample periods:
  - Saudi Arabia: 1966–2007 (annual); excess demand data: 1981–2007.
  - Kuwait: 1974–2007 (annual); excess demand data: 1979–2007.
  - Federal funds rate sample period: 1966–2007.
- Integration/order findings:
  - Most variables are I(1).
  - Output gap, interest rate, demand gap, and excess money supply are I(0).
  - Money supply is I(2) in the case of Kuwait.
- Cointegration:
  - Tests indicate existence of one cointegrating vector for the inflation equation (sample/specification details in source).

### Empirical Findings — Saudi Arabia
- Long-run elasticities (reported estimates):
  - Trading partners’ price level: a 1 percent increase in P* ⇒ 0.83 percent increase in Saudi price level.
  - NEER: a 1 percent increase in NEER (appreciation) ⇒ 0.19 percent decline in the price level.
  - Poil: insignificant in the long-run inflation equation in reported specification.
- Short-run ECM findings:
  - Domestic factors play a relatively limited role.
  - Excess money supply and lagged increases in money supply create short-run inflationary pressures.
  - Growth in real GDP has a negative impact on inflation (may reflect increased supply of nontradables).
  - Speed of adjustment: "It takes about 9 months to eliminate one-half of the deviation from long-run equilibrium."
- PPP tests:
  - Relative PPP (NEER = -1 and P* = 1) is strongly rejected.
- Model performance and diagnostics:
  - "The model tracks actual inflation well."
  - Residual diagnostics indicate: "the absence of serial correlation" and "the normality hypothesis is not rejected."
- Selected numerical outcomes (from reported models 1–8, exact values preserved):
  - Long Run neer(-1) coefficients: 0.19350, 0.19961, 0.13332, 0.13987, 0.31658, 0.24097, 0.24149, 0.15632
  - Long Run p* (-1) coefficients: -0.83295, -0.87421, -0.57863, -1.05107, -1.61082, -1.23633, -1.36629, -0.60067
  - Adjustment Coefficient (models 1–8): -0.590394, -0.361197, -0.534635, -0.300281, -0.112342, -0.190323, -0.162462, -1.441352
  - Short run Excess Demand (models 1–8): 0.025416, 0.090185, 0.042683, 0.090918, -0.312685, -0.238264, -0.098587, -0.037301
  - Short run Excess Money Supply (models 1–8): 0.010846, 0.013282, 0.015311, 0.01212, -0.009593, -0.002008, 0.00843, -0.233302
  - R-squared (models 1–8): 0.927232, 0.880717, 0.877816, 0.845329, 0.9344, 0.95600, 0.949124, 0.878525
  - Adj. R-squared (models 1–8): 0.797867, 0.701792, 0.72231, 0.613323, 0.891617, 0.927306, 0.912124, 0.779136
- Inflation episodes:
  - Saudi Arabia: Inflation reached 4.1 percent in 2007; reached 10.5 percent year on year in April 2008.

### Empirical Findings — Kuwait
- Main long-run drivers:
  - Trading partners’ inflation is the primary driver, followed by pass-through; magnitudes close to Saudi Arabia.
- Short-run impacts (reported exact statements and coefficients preserved):
  - Demand gap: "Demand that exceeds the long run (trend) level by 1 percent increases inflation by 0.1 percent in the short run."
  - Excess money supply: "Money supply in excess of equilibrium money demand by 1 percent increases inflation by 0.03 percent in the short run."
  - Speed of adjustment: "relatively high and close to that of Saudi Arabia."
- Selected numerical coefficients from Table 4 (models 1–5, exact values preserved):
  - neer(-1): 0.151403, 0.160207, 0.207545, 0.123418, 0.186421
  - p*(-1): -0.784288, -0.750785, -0.829833, -0.801091, -0.933541
  - p_oil(-1): -0.054615, -0.017295, -0.007039, -0.010434, NO
  - Adjustment Coefficient: -0.592508, -0.733497, -0.600839, -0.413733, -0.26666
  - Short run Excess Demand: 0.101456, 0.129361, 0.101061, 0.003895, 0.040121
  - Short run Excess Money Supply: 0.031926, 0.041703, 0.043225, -0.031954, 0.022236
  - R-squared (p equation): 0.962193; Adj. R-squared: 0.921477
- Inflation episodes:
  - Kuwait: Inflation averaged 5 percent in 2007; reached 10 percent year on year in February 2008.

### Robustness, Diagnostics, and Alternative Specifications
- Saudi robustness checks (Table 3 variations):
  - Variations included removing time trend, using real money, de-trending vs HP-filter for excess demand, changing sample periods, using WPI.
  - Key robust findings: "Results related to trading partners’ inflation and pass-through effect are robust."
  - Models 2 and 4 indicate: "1 percent excess demand increases inflation by 0.1 percent in the short run."
  - Magnitude of excess demand in 2007: "about 10 percent using hp-filter method and about 30 percent using de-trending method."
- Kuwait robustness checks (Table 4 variations):
  - Variations included real money, de-trending, different sample periods, WPI instead of CPI.
  - Robustness conclusions: "The results are quite robust to these different specifications." "Using de-trending produces slightly better results."
  - PPP assumption was strongly rejected based on log likelihood ratio test.
- Vector Error Correction Estimates (summary):
  - Saudi long-run coefficients (Table A1 sample 1980–2007): neer(-1) 0.193503 [ 2.61963]; p*(-1) -0.83295 [-3.46133]; p_oil(-1) -0.015192 [-0.69371]; Adjustment coefficient -0.590394.
  - Kuwait long-run coefficients (Table A2 sample 1979–2007): neer(-1) 0.151403 [ 2.41881]; p*(-1) -0.784288 [-50.8870]; p_oil(-1) -0.054615 [-3.16361]; Adjustment coefficient -0.592508.
- Residual and diagnostic tests (Table A4 excerpts):
  - PPP (LR test): Saudi Chi-square 6.93; P-value 0.031; Conclusion: Reject the null. Kuwait Chi-square 28.21; P-value 0.0000; Conclusion: Reject the null.
  - Residual Serial Correlation LM Tests: for Saudi and Kuwait at lags 1–3, conclusions generally "Fail to reject the null" of no serial correlation.
  - Normality Test joint (Jarque-Bera multivariate): Saudi Chi-square 62.65; P-value 0.223; Conclusion: Fail to reject the null. Kuwait Chi-square 55.1877; P-value 0.4675; Conclusion: Fail to reject the null.

### Discussion: Pass-through versus Trading Partners’ Inflation
- Empirical observation:
  - "External factors play a dominant role given the dependence of the Saudi and Kuwaiti economies on imports and foreign labor."
  - In the long run, trading partners’ inflation is the main driving force.
- Reasons pass-through is moderate and smaller than trading partners’ inflation (as stated in source):
  - Trading partners’ price changes often permanent; exchange rate changes often not.
  - Exporters/importers may adjust profit mark-ups ("price to market") to protect market share.
  - Global competition, low global inflation, cross-border production chains, and subsidies suppress pass-through.
  - Labor market and nominal anchor considerations reduce sensitivity of domestic prices to nominal shocks.
- Supporting empirical evidence cited:
  - Cross-country inflation correlation (Wang and Wen (2007)): average 0.57 for 18 OECD countries; 0.62 for a G7 sample (min 0.26, max 0.92).
  - Vigfusson et al. (2007) pass-through to export prices: NEER 1 percent ⇒ export price changes of 0.26 (EU), 0.24 (Japan), 0.47 (Asia NIEs), 0.16 (U.S.); PPI 1 percent ⇒ export price changes of 0.75 (EU), 0.35 (Japan), 1.01 (Asia NIEs), 0.72 (U.S.).

### Policy Implications and Conclusions
- Principal findings recap:
  - Both foreign and domestic factors influence inflation; external factors dominate.
  - Recent inflationary pressures driven by: rising trading partners’ inflation, supply constraints in food and other commodities, slide of the U.S. dollar, rapidly rising domestic demand since 2003, and to a lesser extent an expansionary monetary stance imported via the dollar peg.
- Policy recommendations:
  - "Containing inflationary pressures under the peg regime would have to be based on containing domestic demand and addressing nontradable supply bottlenecks, especially in the real estate sector."
  - "Containing inflation through expanding the subsidy system has to be balanced with the fiscal and efficiency cost of subsidies."
  - Domestic measures expected to play only a temporary role; the slide in the U.S. dollar is "unlikely to become a permanent fixture."
- Regional implication:
  - "The similarity between the inflationary processes in Saudi Arabia and Kuwait suggests that the cost associated with the planned monetary union would be limited."

### Appendices and Technical Notes
- Calculating demand gap series (Appendix 2):
  - De-trending: log(X) - log(X̂) = Gap_det where X̂ = β1 + β2*time.
  - HP-Filter: Gap_HP = log(X) - log(X̂_s) where X̂_s is the smoothed series using Hodrick-Prescott exponential smoothing.
- Money demand and excess money supply (Appendix 3):
  - Cointegration tests for money demand (Table A5) show evidence of cointegration for Saudi and Kuwait (trace and max stats preserved in source).
  - Excess money supply constructed as mmexcm = m − m̂ (nominal) or real money minus estimated real-money demand (real).
  - Caveat: estimated excess money supply is noisy due to oil GDP; non-oil GDP would be preferable but unavailable.
- Trading partner weights (Table A3 selected percentages):
  - Saudi Arabia: United States 20.2, Japan 12.8, Germany 8.4, United Kingdom 7.6, China 6.2, Italy 5.7, Korea 4.7, France 4.5
  - Kuwait: United States 14.4, Japan 14.4, Germany 10.2, United Kingdom 7.2, China 5.6, Italy 7.3, Korea 3.6, France 4.9

_Italic: Source: _wp08193 - Appendix Tables (IMF staff; content unit provided)._

### References..............................................................................................................

### _wp08193 - References..............................................................................................................

### References
- References................................................................................................................................29

### Appendices
- Appendix 1. Detailed ECM Results.........................................................................................32
- Appendix 2. Calculating Demand Gap Series .........................................................................35
- Appendix 3. Money Demand Equation and Excess Money Supply........................................36

### Tables
- 1. Unit Root Tests ....................................................................................................................12
- 2. Cointegration Test for Inflation Equation............................................................................15
- 3. Saudi Arabia: Comparing Alternative Models ....................................................................19
- 4. Kuwait: Comparing Alternative Models..............................................................................24

### Figures
- 1. Imports to Aggregate Domestic Demand Ratio, 1980–2007.................................................6
- 2. Non-oil Exports to Aggregate Domestic Demand Ratio, 1980–2007 ...................................6
- 3. GCC: Inflation, 1964–2007 .................................................................................................11
- 4. Saudi Arabia: Inflation and its Theoretical Determinants, 1967–2007 ...............................14
- 5. Saudi Arabia: Actual Inflation and Predicted Long- and Short-Run Inflation ....................17
- 6. Kuwait: Inflation and its Theoretical Determinants, 1974–2007.........................................21
- 7. Kuwait: Actual Inflation and the Predicted Long-run and Short-run Inflation ...................23

### Boxes
- 1. Empirical Evidence on Exchange Rate Pass-through ..........................................................26

*Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08193.pdf*

### Appendix Tables

### Appendix Tables

### Introduction and Key Conclusions
- Long-run drivers of inflation in Saudi Arabia and Kuwait:
  - Inflation in trading partners is the main factor affecting domestic inflation.
  - Exchange rate pass-through has a significant but lower contribution.
- Short-run dynamics:
  - Positive demand shocks and excess money supply exert upward pressures on inflation in the short run, but tend to dissipate quickly as real exchange rate and the money market reach a new equilibrium.
- Remedies discussed in the source text that have been proposed by others (listed in the source):
  - Revaluation or adopting more flexible exchange rate regime to gain monetary policy independence.
  - Higher subsidies.
  - Addressing supply bottlenecks.
  - Containing government expenditures.

### Literature Review (selected points)
- The literature considers inflation the outcome of four main factors: supply-side (cost-push/mark-up), foreign factors, monetary factors, and demand factors; inflation expectations are also important.
- Empirical evidence cited in the source:
  - De Brouwer and Ericsson (1998): mark-up model for Australia.
  - Juselius (1992): strong dependence of Danish prices on West German price levels.
  - Lim and Papi (1997): key role for money and exchange rates in Turkey during 1970–95.
  - Leo (2007): strong relation between money and inflation in Iran.
  - Country-specific studies cited for GCC or comparable countries (Darrat 1985; Al-Mutairi 1995; Keran and Al Malik 1979; Al-Raisi and Pattanaik 2003).
- Al-Raisi and Pattanaik (2003) finding quoted: a depreciation of 10 percent in the nominal effective exchange rate (NEER) will increase the CPI by only 0.4 percent in Oman.

### Economic Model and Methodology
- The domestic CPI price level (P) is modeled as a function:
  - P = f(NEER, P*, Poil, M, D) where:
    - NEER = nominal effective exchange rate (foreign currency price per local currency; increase = appreciation),
    - P* = trading partners’ price level,
    - Poil = price of oil,
    - M = nominal money supply,
    - D = aggregate domestic demand.
- Baseline empirical specification: an error correction model (ECM) of the form (log variables, k lags), capturing long-run relations and short-run dynamics including Δ terms for p, ne er, p*, oil, M, and demand.
- Excess money supply measurement:
  - excm_t = m_t − m̂_t where m̂_t is estimated long-run money demand (money demand as function of real GDP, GDP deflator, and interest rate).
- Excess domestic demand measurement:
  - excd_t = d_t − d̂_t where d_t is real domestic demand (government and private consumption and investment) and d̂_t is potential demand estimated via de-trending or HP-filter.
- Rationale for variables:
  - External factors emphasized due to the tradable sector composition (hydrocarbons) and high import dependence for consumer goods.
  - Flexible labor policies and large expatriate workforce increase sensitivity of domestic prices to external factors (remittance purchasing power).
  - Generous subsidies and pegged exchange rate regime limit the active use of monetary policy and mitigate some domestic inflationary pressures.

### Data, Integration Order, and Sample Coverage
- Sample periods used in tests/estimations:
  - Saudi Arabia sample: 1966–2007 (annual); excess demand data: 1981–2007.
  - Kuwait sample: 1974–2007 (annual); excess demand data: 1979–2007.
  - Federal funds rate sample period: 1966–2007.
- Unit root / order of integration findings (summary of source table):
  - Most variables are I(1).
  - Output gap, interest rate, demand gap, and excess money supply are I(0).
  - Money supply is I(2) in the case of Kuwait.
- Cointegration:
  - Cointegration tests indicate the existence of one cointegrating vector for the inflation equation (sample and specification details in source).

### Empirical Findings — Saudi Arabia (selected results preserved from source)
- Long-run inflation equation (estimates summarized in the source):
  - The main driving force: trading partners’ inflation.
    - A one percent increase in trading partners’ price level results in a 0.83 percent increase in the price level in Saudi Arabia.
  - Exchange rate pass-through:
    - A one percent increase in the NEER (appreciation) results in a 0.19 percent decline in the price level in Saudi Arabia.
  - Oil prices were insignificant in the long-run inflation equation for Saudi Arabia in the reported specification.
- Short-run ECM results (summary statements and exact timing preserved):
  - Domestic factors play a relatively limited role in driving inflation in Saudi Arabia.
  - Excess money supply and lagged increases in money supply create inflationary pressures in the short run.
  - The negative impact of growth in real GDP on inflation may reflect increased supply of nontradables.
  - Speed of adjustment:
    - It takes about 9 months to eliminate one-half of the deviation from long-run equilibrium.
- Tests for PPP:
  - Relative PPP (coefficients of NEER = -1 and P* = 1) is not supported by the estimates; a test for the PPP hypothesis was strongly rejected in the reported analysis.

### Modeling and Estimation Choices (methodological notes from source)
- Money supply was evaluated as part of the long-run cointegrating vector but produced non-sensible coefficient signs; therefore the preferred approach used excess money supply as a stationary short-run regressor.
- Demand shock proxies:
  - Real GDP or non-oil GDP were considered but may be inaccurate due to the large share of oil activity; hence the authors used a demand gap based on domestic demand series (government and private consumption and investment) starting in 1981.
- ECM specifications reported used two lags and included intercept with no trend; a war dummy was used for Kuwait data when appropriate.

### Inflation Developments and Context (selected numerical points)
- Saudi Arabia:
  - Inflation reached 4.1 percent in 2007.
  - Inflation reached 10.5 percent year on year in April 2008.
- Kuwait:
  - Inflation averaged 5 percent in 2007.
  - Inflation reached 10 percent year on year in February 2008.
- Historical pattern:
  - The GCC experienced two episodes of relatively high inflation coinciding with oil booms: the 1970s and the period starting in 2003.

_Italic: Source: _wp08193 - Appendix Tables (IMF staff; content unit provided). _

### Appendix 1). Figure 5 shows the long run and short run predicted inflation compared to the

### _wp08193 - Appendix 1). Figure 5 shows the long run and short run predicted inflation compared to the

### Model performance and diagnostics (Saudi Arabia)
- The model "tracks actual inflation well."
- Diagnostic tests for the residuals suggest:
  - "the absence of serial correlation"
  - "the normality hypothesis is not rejected (see table A.4)."
- Figure 5 (Saudi Arabia) displays:
  - Actual Inflation, Long Run Predicted Inflation, Short run predicted Inflation (1982–2007).
- Adjustment (speed of adjustment) note:
  - "The speed of adjustment is the number of periods (years) required to reduce one-half of a deviation from the long-run equilibrium. It is calculated as log (0.5)/log(1+1/δ), see Rogoff (1996)."

### Robustness and alternative specifications (Saudi Arabia)
- Alternative specifications summarized in Table 3 include:
  - Removing the time trend from the model (model 2).
  - Using real money instead of nominal money to estimate excess money supply (model 3).
  - Using de-trending instead of hp-filter to estimate excess demand (model 4).
  - Changing the period to 1966–2007 and 1968–2007 and using real GDP (models 5 and 6) or real non-oil GDP (model 7).
  - Using the WPI instead of the CPI (model 8).
- Key robust findings:
  - "Results related to trading partners’ inflation and pass-through effect are robust to alternative specifications."
  - Models 2 and 4 indicate that excess demand is significant and that "1 percent excess demand increases inflation by 0.1 percent in the short run."
- Magnitude of excess demand in 2007:
  - "excess demand in 2007 was about 10 percent using hp-filter method and about 30 percent using de-trending method."
- Interpretation:
  - Despite a small coefficient on excess demand, the "large level of excess demand, induced by strong public and private spending, over the last few years has contributed to the recent inflationary pressures."
  - The relatively larger impact for trading partners’ inflation during 1966–2007 could reflect the high inflation during the 1970s and data issues (exaggerated trading partners’ inflation and use of GDP-based output gaps).

### Selected estimated coefficients and model metrics (Saudi Arabia, excerpt from Table 3)
- Long Run: neer(-1) coefficients (models 1–8): 0.19350, 0.19961, 0.13332, 0.13987, 0.31658, 0.24097, 0.24149, 0.15632
- Long Run: p* (-1) coefficients (models 1–8): -0.83295, -0.87421, -0.57863, -1.05107, -1.61082, -1.23633, -1.36629, -0.60067
- Long Run: p_oil(-1) coefficients (models 1–8): -0.01519, 0.04259, -0.03776, 0.11116, 0.30563, 0.10008, 0.24927, 0.00890
- Adjustment Coefficient (models 1–8): -0.590394, -0.361197, -0.534635, -0.300281, -0.112342, -0.190323, -0.162462, -1.441352
- Short run: Excess Demand coefficients (models 1–8): 0.025416, 0.090185, 0.042683, 0.090918, -0.312685, -0.238264, -0.098587, -0.037301
- Short run: Excess Money Supply coefficients (models 1–8): 0.010846, 0.013282, 0.015311, 0.01212, -0.009593, -0.002008, 0.00843, -0.233302
- Sample periods noted in Table 3: "1981–2007", "1966–2007", "1968–2007", "1985–2007" as applicable.
- Model fit metrics (selected):
  - R-squared (models 1–8): 0.927232, 0.880717, 0.877816, 0.845329, 0.9344, 0.95600, 0.949124, 0.878525
  - Adj. R-squared (models 1–8): 0.797867, 0.701792, 0.72231, 0.613323, 0.891617, 0.927306, 0.912124, 0.779136

### Kuwait: estimation summary and key equations
- Data and modeling choices:
  - Use excess demand instead of a GDP-based output gap to proxy demand shocks; demand data available from 1979.
  - "Given that money is I(2), we use only excess money with no lagged changes for money supply."
  - Figure 6 shows Kuwait inflation and theoretical determinants (1974–2007).
- Main estimated long-run inflation equation (as presented in the source, exact transcription preserved):
  - "[3.16]      [50.89]           [-2.42]              
055.078.015.053.1
*oil
ppneerp⋅+⋅+⋅−="
- Short-term inflation equation (exact transcription preserved):
  - "[3.68]      [-2.79]                      [0.25]                   [2.85]             [-0.75]                [0.36]          
D0.055       0.034     - 005.0067.0009.0004.0         
[0.42]            [4.34]                  [-1.56]                      [-0.23]            [-1.53]              [1.03]           
12.011.1086.0012.0204.0203.0          

          [1.78]               [3.04]                     [-3.16]       [-50.89]          [2.42]                   [-5.31]           
032.00.10)055.078.015.053.1(592.0
2121
*
2
*
12121
111
*
⋅+Δ⋅+Δ⋅+Δ⋅−Δ⋅+
Δ⋅+Δ⋅+Δ⋅−Δ⋅−Δ⋅−Δ⋅+
⋅+⋅+⋅−⋅−⋅+−⋅−=Δ
−−−−
−−−−−−
−−−
tt
oil
t
oil
t
tttttt
ttt
oil
t
gdpgdppp
ppneerneerpp
ExcmExcdppneerpp"
- Interpretation of Kuwait results:
  - "The results indicate that the main driving force for inflation is the trading partners’ inflation followed by the pass-through effect."
  - "The size of the impact of trading partners’ inflation and the pass-through effect is very close to that of Saudi Arabia."
  - "The results also indicate that the speed of adjustment is relatively high and close to that of Saudi Arabia."
  - Both excess demand and excess money supply are significant:
    - "Demand that exceeds the long run (trend) level by 1 percent increases inflation by 0.1 percent in the short run."
    - "Money supply in excess of equilibrium money demand by 1 percent increases inflation by 0.03 percent in the short run."
  - "Given the recent increase in domestic demand, the model suggests that part of the inflation in the last few years was driven by strong domestic demand."
- Figure 7 (Kuwait) shows actual inflation and predicted long-run and short-run inflation (1981–2007).

### Robustness and alternative specifications (Kuwait)
- Table 4 compares alternative models (models 1–5) with variations:
  - Using real money instead of nominal money to estimate excess money supply (model 2).
  - Using de-trending instead of hp-filter (model 3).
  - Changing the period to 1974–2007 (model 4).
  - Using WPI instead of CPI (model 5).
- Representative coefficient values from Table 4 (models 1–5):
  - neer(-1): 0.151403, 0.160207, 0.207545, 0.123418, 0.186421
  - p*(-1): -0.784288, -0.750785, -0.829833, -0.801091, -0.933541
  - p_oil(-1): -0.054615, -0.017295, -0.007039, -0.010434, NO
  - Adjustment Coefficient: -0.592508, -0.733497, -0.600839, -0.413733, -0.26666
  - Short run WAR Dummy: 0.055218, 0.057169, 0.058059, 0.099228, 0.099508
  - Short run Excess Demand: 0.101456, 0.129361, 0.101061, 0.003895, 0.040121
  - Short run Excess Money Supply: 0.031926, 0.041703, 0.043225, -0.031954, 0.022236
  - Sample period variants: "1979–2007", "1981–2007", "1974–2007"
- Robustness conclusions:
  - "The results are quite robust to these different specifications."
  - "Using de-trending produces slightly better results."
  - Using WPI for 1978 to link CPI series does not alter results materially but produces less favorable model selection criteria; "Oil prices variable is insignificant in the WPI model and model selection criteria suggest removing it."
  - "The PPP assumption was strongly rejected based on log likelihood ratio test."

### Discussion: Pass-through versus trading partners’ inflation
- Empirical observation:
  - "External factors play a dominant role given the dependence of the Saudi and Kuwaiti economies on imports and foreign labor to meet domestic demand of tradables and nontradables."
  - "In the long run, the main driving force for inflation in Saudi Arabia and Kuwait is trading partners’ inflation."
- Why pass-through is moderate and smaller than trading partners’ inflation:
  - Changes in trading partners’ price levels are usually permanent, while changes in exchange rates are often not.
  - Exporters may absorb exchange rate impacts by adjusting profit mark-ups ("price to market") to protect market share.
  - Importers and retailers with some monopolistic power may adjust mark-ups to preserve market share.
  - Global competition and low global inflation enable substitution and limit pass-through.
  - Cross-border production chains limit pass-through to final products.
  - Subsidies for basic commodities in Saudi Arabia and Kuwait limit pass-through for affected items (with fiscal and efficiency costs).
  - Labor market structure (large government employment, sticky private wages) implies NEER changes have income effects that can offset tradables price movements via nontradables price adjustments.
  - Establishment of a strong nominal anchor reduces domestic price sensitivity to nominal shocks, contributing to "very low and declining pass-through of exchange rate to inflation."
- Supporting empirical evidence (literature cited in text):
  - Cross-country inflation correlation (Wang and Wen (2007)): average 0.57 for 18 OECD countries; 0.62 for a G7 sample (minimum 0.26, maximum 0.92).
  - Vigfusson et al. (2007) pass-through to export prices: 1 percent NEER change leads to 0.26 (EU), 0.24 (Japan), 0.47 (Asia NIEs), 0.16 (U.S.) changes in export prices; 1 percent change in PPI leads to 0.75 (EU), 0.35 (Japan), 1.01 (Asia NIEs), 0.72 (U.S.) changes in export prices.
  - Box I: examples and case studies indicating weak and declining pass-through (Mishkin 2008; Lafleche, 1996/1997; Cunningham and Haldane, 2000; Gagnon, 2004; Burnstein, Eichenbaum, and Rebelo, 2007; Yam 2008).

### Policy implications and conclusions
- Principal findings:
  - Both foreign and domestic factors influence inflation in Saudi Arabia and Kuwait, with external factors dominant.
  - Recent inflationary pressures have been largely driven by rising trading partners’ inflation, supply constraints in food and other commodities, the slide of the U.S. dollar, rapidly rising domestic demand since 2003, and to a lesser extent an expansionary monetary stance imported via the dollar peg.
- Policy recommendations and implications:
  - "Containing inflationary pressures under the peg regime would have to be based on containing domestic demand and addressing nontradable supply bottlenecks, especially in the real estate sector."
  - "Containing inflation through expanding the subsidy system has to be balanced with the fiscal and efficiency cost of subsidies."
  - Domestic measures are expected to play only a temporary role; the slide in the U.S. dollar is "unlikely to become a permanent fixture."
  - Historical NEER behavior: "The experience of the last four decades indicates that changes in the NEER have centered around zero in the long run (see Figure 4)."
- Regional implication:
  - "The similarity between the inflationary processes in Saudi Arabia and Kuwait suggests that the cost associated with the planned monetary union would be limited."

*Sources: Authorities data; and Fund staff estimates.*

### Appendix 1). In addition, domestic demand in all GCC countries depends highly on

### _wp08193 - Appendix 1). In addition, domestic demand in all GCC countries depends highly on

### Key findings on GCC domestic demand and oil revenue
- Domestic demand in all GCC countries depends highly on government spending from oil revenues.
- Changes in oil prices and revenues and other global or regional shocks are likely to have on average similar impact on domestic demand given the similarity between these countries, including in the government spending response to higher revenues.
- The impact on investors and consumers’ confidence is likely to be similar across these countries.

### Vector Error Correction Estimates — Summary (Saudi Arabia, Kuwait)
- Table A1. Saudi Arabia: Vector Error Correction Estimates (Sample 1980–2007; t-statistics in [  ]). Long run equation coefficients:
  - neer(-1): 0.193503 [ 2.61963]
  - p*(-1): -0.83295 [-3.46133]
  - p_oil(-1): -0.015192 [-0.69371]
  - t: 0.02437
  - C: -2.452129
  - Adjustment coefficient (for equations): -0.590394, -0.29976, -0.146682, -0.352519 (t-stats: [-5.68420][-0.59112][-2.81524][-0.11693])
- Short run dynamic highlights (Saudi Arabia):
  - D(p(-1)): -0.426935 [ -2.38712]
  - D(p*(-1)): 2.19899 [ 3.94399]
  - D(LRGDP(-1)): 2/-0.241897 [-2.88765] (note table formatting preserved)
  - R-squared (p equation): 0.927232; Adj. R-squared: 0.797867
  - Mean dependent (p): 0.004174; S.D. dependent: 0.020031
- Table A2. Kuwait: Vector Error Correction Estimates (Sample 1979-2007; t-statistics in [ ]). Long run equation coefficients:
  - neer(-1): 0.151403 [ 2.41881]
  - p*(-1): -0.784288 [-50.8870]
  - p_oil(-1): -0.054615 [-3.16361]
  - C: -1.534589
  - Adjustment coefficient (for equations): -0.592508, -0.003937, 0.241539, -7.56748 (t-stats: [-5.31008][-0.00615][ 2.43666][-2.35899])
- Short run dynamic highlights (Kuwait):
  - D(p*(-1)): 1.114126 [ 4.33977]
  - WAR coefficient: 0.055218 [ 3.67973]
  - R-squared (p equation): 0.962193; Adj. R-squared: 0.921477
  - Mean dependent (p): 0.031397; S.D. dependent: 0.032175

### Selected trading partner weights (Table A3)
- Trading partner weights by country (percent):
  - Saudi Arabia: United States 20.2, Japan 12.8, Germany 8.4, United Kingdom 7.6, China 6.2, Italy 5.7, Korea 4.7, France 4.5
  - Kuwait: United States 14.4, Japan 14.4, Germany 10.2, United Kingdom 7.2, China 5.6, Italy 7.3, Korea 3.6, France 4.9
  - Qatar: United States 15.3, Japan 11.8, Germany 8.9, United Kingdom 9.0, China 3.9, Italy 7.7, Korea 5.5, France 7.2
  - U.A.E.: United States 11.8, Japan 8.4, Germany 7.8, United Kingdom 9.2, China 5.5, Italy 5.3, Korea 4.7, France 6.5
  - Bahrain: United States 3.3, Japan 8.8, Germany 8.5, United Kingdom 7.2, China 4.9, Italy 5.5, Korea 3.5, France 7.6
  - Oman: United States 13.7, Japan 14.6, Germany 7.1, United Kingdom 8.3, China 3.2, Italy 4.3, Korea 3.6, France 4.9
- Source: INS.

### PPP assumption and residual tests (Table A4)
- 1- PPP Assumption (LR test):
  - Saudi Arabia: Chi-square Statistic 6.93; P-value 0.031; Conclusion: Reject the null
  - Kuwait: Chi-square Statistic 28.21; P-value 0.0000; Conclusion: Reject the null
- 2- Residual Serial Correlation LM Tests (Null: no serial correlation at lag order h):
  - Saudi Arabia:
    - Lag 1: LM-Statistic 20.3925; P-value 0.2031; Conclusion: Fail to reject the null
    - Lag 2: LM-Statistic 16.21565; P-value 0.438; Conclusion: Fail to reject the null
    - Lag 3: LM-Statistic 16.05634; P-value 0.449; Conclusion: Fail to reject the null
  - Kuwait:
    - Lag 1: LM-Statistic 26.03687; P-value 0.0535; Conclusion: Fail to reject the null
    - Lag 2: LM-Statistic 19.45255; P-value 0.2459; Conclusion: Fail to reject the null
    - Lag 3: LM-Statistic 10.70245; P-value 0.8275; Conclusion: Fail to reject the null
- 3- Normality Test joint (Jarque-Bera) 1/ (Null: residuals are multivariate normal):
  - Saudi Arabia: Chi-square Statistic 62.65; P-value 0.223; Conclusion: Fail to reject the null
  - Kuwait: Chi-square Statistic 55.1877; P-value 0.4675; Conclusion: Fail to reject the null
- Note: 1/ The residual normality test is the multivariate extension of the Jarque-Bera normality test. Method suggested by Urzua (1996).

### Calculating demand gap series (Appendix 2)
- De-trending (Equation A2.1):
  - A series X is de-trended by regression of X on a constant and time. Predicted X (X̂) is subtracted from actual X to get gap:
    - log(X) - log(X̂) = Gap_det where X̂ = β1 + β2*time
- HP-Filter (Equation A2.2):
  - Gap_HP = log(X) - log(X̂_s) where X̂_s is the smoothed series using Hodrick-Prescott exponential smoothing.

### Money demand equation and excess money supply (Appendix 3)
- Cointegration tests (Table A5) using nominal money supply (m), real GDP (gdp), GDP deflator (p), and interest rate (r):
  - Saudi Arabia (Sample 1966–2007; 2 lags): Trace Stat. for r=0 is 71.83* (95% C.V. (63.88)); Max Stat. for r=0 is 31.00 (32.12)
  - Kuwait (Sample 1974–2007; 1 lag): Trace Stat. for r=0 is 62.73* (47.86); Max Stat. for r=0 is 31.74* (27.58)
  - Interpretation:
    - Kuwait: null of no cointegration strongly rejected; trace indicates there might be two cointegrating relationships while max indicates only one.
    - Saudi: trace rejects null of no cointegration; maximum eigenvalue test indicates no cointegration relationship but statistics are close and can be rejected at 1 percent significance level. Testing for more than one cointegration is highly rejected in both tests.
- Standardized eigenvectors (money demand in nominal terms) — implied cointegrating vectors (as given):
  - Saudi (nominal): [ -0.22 ] [ 13.23 ] [ 4.55 ]  yields: rpgdpmKuwait *354.0*553.1*768.096.6: [-4.70][1.92][4.55] *221.23*10.1*814.3*12.64.19: rpgdptmSaudi −++−=
    - (Table text preserved verbatim as presented)
  - Alternative real-money formulation (Equation A3.2) — long run real money demand:
    - Kuwait: coefficients presented as [-2.31] [2.00] ... giving rgdprmKuwait −+−= *66.4*406.31.3: (verbatim table entries)
    - Saudi (real): coefficients presented leading to rgdptimermSaudi −+⋅−−= *30.17*20.3094.070.11:
- Excess money supply construction:
  - Using nominal money demand: mmexcm = m − m̂ where m is nominal money supply and m̂ is estimated money demand (from equation (A3.1)).
  - Using real money: excess money supply = real money supply − estimated real money demand (from equation (A3.2)).
  - Caveats: Estimated excess money supply (nominal or real) suffers from noise introduced by the oil GDP. Money demand function likely more stable with non-oil GDP, but data unavailability prevents use. Thus money demand functions provide only a proxy for excess money supply.

### Cointegration test details and coefficients (Table A5 excerpt)
- Saudi Arabia cointegration test statistics:
  - Trace Stat. r=0: 71.83* (95% C.V. (63.88))
  - Max Stat. r=0: 31.00 (32.12)
  - Trace Stat. r≤1: 40.84 (42.92)
  - Max Stat. r≤1: 18.41 (25.82)
- Kuwait cointegration test statistics:
  - Trace Stat. r=0: 62.73* (47.86)
  - Max Stat. r=0: 31.74* (27.58)
  - Trace Stat. r≤1: 31.00* (29.80)
  - Max Stat. r≤1: 20.50 (21.13)
- Standardized eigenvectors and adjustment coefficients (selected entries preserved verbatim from table):
  - Saudi Arabia standardized eigenvectors (var order m, gdp, p, r):
    - Row examples: 1.000 -3.814 -1.100 23.221 ; 1.215 -3.050 1.000 -22.944
    - Std. adjustment coefficients (equ.): m -0.048 0.059 -0.053 -0.450 ; gdp 0.024 -0.006 0.032 -0.321
  - Kuwait standardized eigenvectors (var order m, gdp, p, r):
    - Row examples: 1.000 -0.768 -1.553 0.354 ; 0.130 -0.379 -0.394 1.000
    - Std. adjustment coefficients (equ.): m -0.155 0.025 0.000 0.001 ; gdp 0.198 0.022 0.005 0.029

*Sources: Authorities data and Fund staff estimates.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08193.pdf_