## _wp12168

## Source details

**Canonical URL:** [_wp12168](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12168.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12168.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12168.pdf.json)

---

### I. Introduction and research scope
- Commodity-exporting countries face large terms of trade fluctuations that render their real exchange rate volatile.
- The paper examines the relationship between volatility in the gold price and volatility in the South African real exchange rate (REER), focusing on volatilities (not joint movements).
- It investigates how capital account liberalization (break date 1995, financial rand abolished in March 1995) changed this volatility relationship for South Africa.
- Two main contributions:
  - Determine the direction of causality between gold price volatility and REER volatility.
  - Assess how capital account liberalization altered that volatility relationship.

### Data, measures, and empirical strategy
- Gold price volatility: monthly UNCTAD Commodity Statistics data covering 1979-2007; volatility measure is the twelve month rolling window standard deviation of the international gold price index.
- REER volatility: monthly REER constructed from monthly consumer price index data and monthly nominal exchange rate from IMF (2010); volatility measure is the twelve month rolling window standard deviation of REER.
- Break date for capital account liberalization: 1995 (alternative break years 1996 and 1997 used in robustness checks).
- Econometric approach:
  - Initial observation: volatility of gold price is much lower than REER volatility with a ratio of 1:16.
  - For full sample 1980–2007 (27 years): no unit root and no cointegration between REER and gold price volatility.
  - Splitting the sample at 1995: series are integrated of order one in both sub-samples and appear cointegrated in both pre- and post-liberalization sub-samples.
  - Johansen trace-test indicates one cointegration relationship in both sub-samples.
  - Model chosen: Vector Error Correction Model (VECM) with lag order selected by AIC and HQ and residuals checked for autocorrelation.

### Cointegration and diagnostic test results (selected)
- Johansen trace-test — Pre-liberalization sample (Table 1):
  - Number of obs=181; Lags=5.
  - Rank 0 trace statistic 17.9881; 5% critical value 12.530.
  - Rank 1 trace statistic 2.6993*; 5% critical value 3.8400.
- Johansen trace-test — Post-liberalization sample (Table 2):
  - Number of obs=181; Lags=3.
  - Rank 0 trace statistic 13.354; 5% critical value 12.530.
  - Rank 1 trace statistic 0.7073*; 5% critical value 3.8400.
- Lagrange-Multiplier (LM) tests for residual autocorrelation (Table 3) — selected results:
  - Sample (1979-1994): lag 1 chi2 7.5759; df 4; Prob>chi2 0.1084.
  - Sample (1995-2011): lag 1 chi2 0.5941; df 4; Prob>chi2 0.9637.
  - Note: at the 10% level, cannot reject the null of no autocorrelation for most tested orders; some individual lags show low p-values but overall model diagnostics indicated no evidence of misspecification.

### Main empirical results — short-run dynamics
- Pre-1995 (VECM short-run results, Table 4 — Panel 1 and Panel 2):
  - REER short-run dynamics primarily driven by own lags:
    - D_REER Volatility t-1 coefficient 0.8159; Std. Err. 0.0750; z-statistic 10.8800; P-value 0.0000.
    - D_REER Volatility t-4 coefficient -0.1755; Std. Err. 0.0744; z-statistic -2.3600; P-value 0.0180.
  - Gold-price lagged coefficients in REER equation are not statistically significant (e.g., D_Gold Price Volatility t-1 coefficient 0.0309; P-value 0.9710).
  - Gold volatility equation driven by REER lags (e.g., D_REER Volatility t-1 coefficient 0.8504; P-value 0.0000).
  - Conclusion: pre-1995 short-run movements in REER and gold price volatility are driven by their own dynamics; little evidence of short-run cross-effects.
- Post-1995 (VECM short-run results, Table 5 — Panel 1 and Panel 2):
  - REER still dominated by own lag:
    - D_REER Volatility t-1 coefficient 0.8162; Std. Err. 0.0727; z-statistic 11.2200; P-value 0.0000.
  - Some short-run cross-dynamics emerge but limited statistical significance:
    - D_Gold Price Volatility t-1 in REER equation coefficient 0.2225; P-value 0.8600 (not significant).
    - D_REER Volatility t-1 in gold equation coefficient 0.0070; P-value 0.1000 (marginal).
  - Correlation between series increases from nearly 0 to 0.2 after liberalization.
  - Overall short-run conclusion: no strong statistically significant short-run relationship; liberalization only slightly alters short-run dynamics.

### Main empirical results — long-run relationships and causality
- Pre-1995 (VECM long-run results, Table 4 — Panel 3 and Table 6 weak exogeneity):
  - Long-run relation coefficients:
    - Gold Price Volatility normalized coefficient 1.0000.
    - REER Volatility coefficient -0.0462; Std. Err. 0.0239; z-statistic 1.9600; P-value 0.0500.
  - Error-correction term significant only in the gold price volatility equation:
    - EC in gold equation coefficient -0.0242; Std. Err. 0.0063; z-statistic -3.8500; P-value 0.0000.
    - EC in REER equation coefficient 0.0352; P-value 0.6260 (not significant).
  - Weak exogeneity tests (Table 6):
    - D_rvolat (REER) chi2 0.2371; P>chi2 0.6263 (cannot reject weak exogeneity).
    - D_gvolat (gold) chi2 14.8384; P>chi2 0.0001 (reject weak exogeneity).
  - Interpretation: pre-1995 long-run causality runs from REER volatility to gold price volatility.
- Post-1995 (VECM long-run results, Table 5 — Panel 3 and Table 6 weak exogeneity):
  - Long-run relation coefficients:
    - REER Volatility normalized coefficient 1.0000.
    - Gold Price Volatility coefficient -12.299; Std. Err. 2.7483; z-statistic -4.4800; P-value <0.001.
  - Error-correction terms:
    - EC in REER equation coefficient -0.0313; Std. Err. 0.0109; z-statistic -2.8800; P-value 0.0040 (significant).
    - EC in gold equation coefficient 0.0014; Std. Err. 0.0006; z-statistic 2.1900; P-value 0.0280 (significant but small).
  - Weak exogeneity tests (Table 6):
    - D_rvolat (REER) chi2 8.2769; P>chi2 0.0040 (reject weak exogeneity).
    - D_gvolat (gold) chi2 4.8090; P>chi2 0.0283 (reject weak exogeneity).
  - Interpretation: post-1995 long-run causality runs from gold price volatility to REER volatility; REER volatility rejoins long-term equilibrium faster than gold volatility.
  - Economic interpretation: consistent with global investors speculating on South African developments amid gold-price volatility by investing in liquid South African assets; capital account liberalization increases transmission of commodity price volatility to REER.

### Robustness checks and extensions
- Alternative break dates: 1996 and 1997 used; results do not change substantially.
- Alternative financial variable: replace REER volatility with volatility in sovereign bond yields (EMBI data, available only from end 1990s).
  - Gold price and EMBI volatility are non-stationary and cointegrated.
  - Findings with EMBI: cointegration relationship goes both ways; some evidence gold price affects EMBI volatility somewhat more than the reverse.
- Suggested further extensions: consider a wider range of commodities and comparative analysis with smaller commodity producers and non-commodity-producing countries with liberalized capital accounts.

### Key empirical statistics and coefficients (selected)
- Initial volatility ratio: gold price volatility : REER volatility = 1:16.
- Johansen trace-test pre-liberalization Rank 0 trace statistic 17.9881; 5% critical value 12.530.
- Johansen trace-test post-liberalization Rank 0 trace statistic 13.354; 5% critical value 12.530.
- Pre-1995 VECM (REER equation): D_REER Volatility t-1 coefficient 0.8159; P-value 0.0000.
- Pre-1995 VECM (gold equation EC): EC coefficient -0.0242; P-value 0.0000.
- Post-1995 VECM (REER equation EC): EC coefficient -0.0313; P-value 0.0040.
- Post-1995 long-run coefficient on Gold Price Volatility (REER normalized): -12.299; Std. Err. 2.7483; P-value <0.001.
- Weak exogeneity tests (1979-1994): D_rvolat P>chi2 0.6263; D_gvolat P>chi2 0.0001.
- Weak exogeneity tests (1995-2011): D_rvolat P>chi2 0.0040; D_gvolat P>chi2 0.0283.

### Policy-relevant conclusions and implications
- Primary empirical conclusion: in South Africa, gold price volatility became a more determining factor for Rand-REER volatility after capital account liberalization than before.
- Policy implications:
  - Increased REER volatility can affect investment and trade and complicate fiscal policy through effects on the cost of borrowing.
  - Dependence on gold and its price developments becomes more important under open capital markets.
  - Commodity-exporting countries should weigh benefits and costs of capital account liberalization given potential heightened transmission between commodity price volatility and REER and borrowing costs.
  - The debate on capital controls during surges in capital inflows may benefit from considering commodity exporters and focusing policy attention on exchange rate volatility rather than level.
- Distributional/market-friction consideration: small firms may face high costs or inability to hedge exchange rate volatility using financial instruments, with implications for export-oriented sector development.

*Source: _wp12168 (IMF PDF).*

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

### _wp12168 - References .............................................................................................................

### I. Introduction and research scope
- Commodity-exporting countries face large terms of trade fluctuations that render their real exchange rate volatile.
- Increased volatility in the real exchange rate adversely affects private agents’ consumption and investment decisions.
- South Africa, the second largest producer of gold, has experienced more frequent episodes of nominal and real exchange rate volatility since the liberalization of the capital account (the financial rand was abolished in March 1995).
- The paper examines the relationship between volatility in the gold price and volatility in the South African real exchange rate, taking into account both volatilities (not joint movements) — noted as a novelty of this study.
- The paper also explores how capital account liberalization has changed this relationship for South Africa.

### Data context and notable statistics
- Foreign direct investment amounted to 0.3 percent of GDP in 2010 while portfolio flows amounted to 3 percent.
- Since liberalization, South Africa has received an increasing portion of capital flows in the form of portfolio investments.
- The heightened volatility in the South African Rand during 1982-1985 is related to political developments and international sanctions tied to apartheid.

### Contributions and hypotheses
- Two main contributions:
  - Determine the direction of causality between volatility in gold prices and volatility in the real exchange rate.
  - Assess the extent to which capital account liberalization altered the volatility relationship.
- Rationale:
  - Understanding the relationship is important for resource-rich countries facing high commodity price volatility.
  - Opening the capital account may magnify the impact of commodity revenue volatility on the exchange rate because of changes in capital flow composition.

### Key findings (as reported)
- Prior to capital account liberalization, long-run causality runs from the South African Rand to gold price volatility.
- For the post-liberalization period, causality runs from gold price volatility to the South African Rand.
- Policy-relevant inference:
  - Following capital account liberalization, the composition of capital flows in South Africa tilted toward portfolio investment.
  - The findings suggest gold price volatility plays a key role in explaining both excessive exchange rate volatility and the disproportionate share of speculative (short-run) inflows South Africa has faced since capital account liberalization.

### Policy implications and considerations
- Resource-rich countries should consider how capital account liberalization interacts with commodity price volatility when evaluating exchange rate and capital flow policies.
- Small firms may face high costs or inability to hedge against exchange rate volatility using financial instruments, with important implications for export-oriented sector development (see Raddatz (2011) as discussed in the source).

*Source: _wp12168 - References (IMF PDF).*

### Section  4  explains  our  estimation  strategy;  Section  5  discusses  the  main  empirical  results;

### _wp12168 - Section  4  explains  our  estimation  strategy;  Section  5  discusses  the  main  empirical  results;

### Literature and contribution
- Connects to the "commodity currency" literature showing a relationship between exchange rate levels and commodity prices (cites: Chen and Rogoff (2003); Cashin, Cespedes and Sahay (2004); Frankel (2007); Ngandy (2005)).
- Refers to literature on capital account liberalization and its growth/crisis effects (cites: Cuddington and Liang (1998); Dell Arrica et al. (2000); Henry (2007)).
- Links to macroeconomic volatility literature showing volatility hurts growth (cites: Aghion and Banerjee (2005); Ramey and Ramey (1995); Aghion et al. (2009); Bui and Pippenger (1990); Smith (1999); Gilbert (1989); Raddatz (2011)).
- Paper’s contribution: focuses on the second moment (volatility) of the relationship between commodity price and real exchange rate using South Africa as a case study and exploits the March 1995 capital account liberalization as a quasi-natural experiment.

### Empirical strategy — data and measures
- Gold price volatility: monthly UNCTAD Commodity Statistics data covering 1979-2007; volatility measure is the twelve month rolling window of the standard deviation of the international gold price index.
- Real exchange rate volatility: monthly REER constructed from monthly consumer price index data and monthly nominal exchange rate from IMF (2010); volatility measure is the twelve month rolling window of the standard deviation of REER.
- Break date for capital account liberalization: 1995 (following Farrell and Todani (2004)); alternative break years considered in robustness checks (1996 and 1997).

### Econometric approach and time-series properties
- Initial observation: volatility of gold price is much lower than REER volatility with a ratio of 1:16; drop in volatility more visible after liberalization.
- Unit root / cointegration testing:
  - For full sample 1980 until 2007 (27 years), no unit root and no cointegration found between REER and gold price volatility.
  - When splitting sample at 1995, series are integrated of order one over the two sub-samples and appear cointegrated in both pre- and post-liberalization sub-samples.
  - Johansen trace-test indicates one cointegration relationship between gold and real exchange rate volatility in both sub-samples.
- Model choice: Vector Error Correction Model (VECM) preferred over VAR because of cointegration; VECM of order p specified with standard notation (Δ operator, Π matrix, Φ matrices), with order chosen by AIC and HQ and ensuring residuals are not autocorrelated.

### Main results — short-run and long-run relationships
- Model diagnostics:
  - Lagrange-multiplier (LM) tests for autocorrelation (before and after liberalization): at the 10% level, cannot reject the null of no autocorrelation for any tested orders; no evidence of model misspecification.
- Short-run results:
  - Pre-1995: movements in REER and gold price are driven by their own dynamics and do not affect each other; lagged gold-price volatility coefficients are not significant in REER equation and vice versa.
  - Post-1995: the restriction of no short-run cross-effects is no longer observed; lagged coefficients of REER and gold price volatility are statistically significant in both panels but cross-terms largely not significant except the second lag of REERΔ in the gold volatilityΔ equation.
  - Correlation between series increases from nearly 0 to 0.2 after liberalization.
  - Overall: in the short run there is no strong statistically significant relationship between gold price and REER volatility, and liberalization only slightly alters this.
- Long-run results:
  - Pre-1995: error-correction term significant only in the gold price volatility equation — evidence that REER volatility causes gold price volatility in the long run, but not vice versa; weak exogeneity tests: cannot reject exogeneity for REER but can reject for gold price volatility (top part of Table 6).
  - Post-1995: cointegration vector indicates a long-run relationship going from gold price volatility to REER volatility — evidence that gold price volatility causes REER volatility in the long run. Error-correction coefficient significant in panel 1 (REER equation) but not in panel 2. Weak exogeneity tests require rejection of exogeneity for both variables but more so for REER (bottom part of Table 6). REER volatility rejoins long-term equilibrium faster than gold volatility.
  - Interpretation: post-liberalization results consistent with global investors speculating on South African developments amid gold-price volatility by investing in liquid South African assets; capital account liberalization increases transmission of commodity price volatility to REER.

### Robustness checks
- Alternative break dates: 1996 and 1997 used; results do not change substantially.
- Alternative financial variable: substitute REER with volatility in sovereign bond yields using Emerging Market Bond Index (EMBI) data (available only from end 1990s). Gold price and EMBI volatility are non-stationary and cointegrated.
  - Findings with EMBI: cointegration relationship goes both ways; some evidence gold price affects EMBI volatility somewhat more than the reverse (detailed results available upon request).
- Authors note potential extensions: consider a wider range of commodities and compare with smaller commodity producers and non-commodity-producing countries with liberalized capital accounts (left for future research).

### Policy-relevant conclusions and implications
- Primary empirical conclusion: in South Africa gold price volatility has become a more determining factor for Rand-REER volatility after capital account liberalization than before.
- Policy implications:
  - Increased REER volatility may have consequences on investment and trade and complicate fiscal policy via effects on the cost of borrowing.
  - Dependence on gold and its price developments becomes more important under open capital markets, posing challenges for commodity-exporting countries.
  - Commodity-exporting countries should weigh benefits and costs of capital account liberalization in light of potential heightened transmission between commodity price volatility and key variables such as REER and borrowing costs.
  - The debate on capital controls during surges in capital inflows may benefit from considering circumstances of commodity exporters and focusing on exchange rate volatility rather than level.
- Contextual observation: commodity prices entered an age of growing volatility; the perceived lack of safe assets at the onset of the financial crisis 2007-08 engendered a flight to commodities and a sharp increase in prices, amplifying the relevance of the paper’s findings.

*Source: _wp12168 - Section  4  explains  our  estimation  strategy;  Section  5  discusses  the  main  empirical  results;*

### REFERENCES

### _wp12168 - REFERENCES

### Key empirical context and data sources
- Gold price data: UNCTAD Commodity Statistics (used for price levels and volatility measures).
- Real effective exchange rate (REER): combines monthly consumer price index data and the monthly nominal exchange rate from IMF (2010).
- REER and gold price volatility measures: twelve month rolling window of the standard deviation of respective series.

### Cointegration test results
- Table 1: Johansen Cointegration Test Results on Pre-Capital Account Liberalization Sample
  - Trend: none
  - Number of obs=181
  - Lags=5
  - Rank 0: parameters 16.0000; loglikelihood -314.4047; eigenvalue .; trace statistic 17.9881; 5% critical value 12.530
  - Rank 1: parameters 19.0000; loglikelihood -306.7603; eigenvalue 0.0810; trace statistic 2.6993*; 5% critical value 3.8400
  - Rank 2: parameters 20.0000; loglikelihood -305.4106; eigenvalue 0.0148
  - Note: The null hypothesis of the trace test is the existence of a number of cointegration relations equal to the rank, versus the maximum rank (2 here). The critical value at a 5% risk level is included.

- Table 2: Johansen Cointegration Test Results on Post-Capital Account Liberalization Sample
  - Trend: none
  - Number of obs=181
  - Lags=3
  - Rank 0: parameters 8.0000; loglikelihood -202.3277; eigenvalue .; trace statistic 13.354; 5% critical value 12.530
  - Rank 1: parameters 11.000; loglikelihood -196.0045; eigenvalue 0.0675; trace statistic 0.7073*; 5% critical value 3.8400
  - Rank 2: parameters 12.0000; loglikelihood -195.6508; eigenvalue 0.0039
  - Note: The null hypothesis of the trace test is the existence of a number of cointegration relations equal to the rank, versus the maximum rank (2 here). The critical value at a 5% risk level is included.

### Residual autocorrelation (Lagrange Multiplier) tests
- Table 3: Lagrange Multiplier Test for Residual Autocorelation
  - Sample (1979-1994)
    - lag 1.0000: chi2 7.5759; df 4; Prob>chi2 0.1084
    - lag 6.0000: chi2 8.0892; df 4; Prob>chi2 0.0884
    - lag 12.0000: chi2 8.6740; df 4; Prob>chi2 0.0698
    - lag 18.0000: chi2 3.0007; df 4; Prob>chi2 0.5577
    - lag 24.0000: chi2 12.1413; df 4; Prob>chi2 0.0163
  - Sample (1995-2011)
    - lag 1.0000: chi2 0.5941; df 4; Prob>chi2 0.9637
    - lag 6.0000: chi2 12.8373; df 4; Prob>chi2 0.0121
    - lag 12.0000: chi2 9.4205; df 4; Prob>chi2 0.0514
    - lag 18.0000: chi2 8.6451; df 4; Prob>chi2 0.0706
    - lag 24.0000: chi2 7.2657; df 4; Prob>chi2 0.1225
  - Note: H0 - no autocorrelation at the specified lag order

### VECM results — Pre-Capital Account Liberalization (1979 – 1994)
- Table 4: VECM Results for Pre-Capital Account Liberalization Sample

- Panel 1: Dependent variable: D_REER Volatility
  - EC: Coefficient 0.0352; Std. Err. 0.0724; z-statistic 0.4900; P-value 0.6260
  - D_REER Volatility t-1: Coefficient 0.8159; Std. Err. 0.0750; z-statistic 10.8800; P-value 0.0000
  - D_REER Volatility t-2: Coefficient -0.1457; Std. Err. 0.0970; z-statistic -1.5000; P-value 0.1330
  - D_REER Volatility t-3: Coefficient 0.0076; Std. Err. 0.0964; z-statistic 0.0800; P-value 0.9370
  - D_REER Volatility t-4: Coefficient -0.1755; Std. Err. 0.0744; z-statistic -2.3600; P-value 0.0180
  - D_Gold Price Volatility t-1: Coefficient 0.0309; Std. Err. 0.8485; z-statistic 0.0400; P-value 0.9710
  - D_Gold Price Volatility t-2: Coefficient -0.4208; Std. Err. 1.1083; z-statistic -0.3800; P-value 0.7040
  - D_Gold Price Volatility t-3: Coefficient -0.2007; Std. Err. 1.0649; z-statistic -0.1900; P-value 0.8510
  - D_Gold Price Volatility t-4: Coefficient -0.0400; Std. Err. 0.6057; z-statistic -0.0700; P-value 0.9470

- Panel 2: Dependent variable: D_Gold Price Volatility
  - EC: Coefficient -0.0242; Std. Err. 0.0063; z-statistic -3.8500; P-value 0.0000
  - D_REER Volatility t-1: Coefficient 0.8504; Std. Err. 0.0737; z-statistic 11.5400; P-value 0.0000
  - D_REER Volatility t-2: Coefficient -0.3169; Std. Err. 0.0963; z-statistic -3.2900; P-value 0.0010
  - D_REER Volatility t-3: Coefficient 0.1380; Std. Err. 0.0925; z-statistic 1.4900; P-value 0.1360
  - D_REER Volatility t-4: Coefficient -0.0936; Std. Err. 0.0526; z-statistic -1.7800; P-value 0.0750
  - D_Gold Price Volatility t-1: Coefficient -0.0058; Std. Err. 0.0065; z-statistic -0.9000; P-value 0.3710
  - D_Gold Price Volatility t-2: Coefficient 0.0018; Std. Err. 0.0084; z-statistic 0.2200; P-value 0.8280
  - D_Gold Price Volatility t-3: Coefficient 0.0031; Std. Err. 0.0084; z-statistic 0.3700; P-value 0.7140
  - D_Gold Price Volatility t-4: Coefficient -0.0003; Std. Err. 0.0065; z-statistic -0.0500; P-value 0.9630

- Panel 3: Coefficients of the long-run relation
  - Gold Price Volatility: Coefficient 1.0000
  - REER Volatility: Coefficient -0.0462; Std. Err. 0.0239; z-statistic 1.9600; P-value 0.0500

- Note: The table includes values and statistical significance of the short run coefficients of the VECM model as well as the error correction term (EC).

### VECM results — Post-Capital Account Liberalization (1995 – 2011)
- Table 5: Post-Capital Account Liberalization VECM Results

- Panel 1: Dependent variable: D_REER Volatility
  - EC: Coefficient -0.0313; Std. Err. 0.0109; z-statistic -2.8800; P-value 0.0040
  - D_REER Volatility t-1: Coefficient 0.8162; Std. Err. 0.0727; z-statistic 11.2200; P-value 0.0000
  - D_REER Volatility t-2: Coefficient -0.1386; Std. Err. 0.0742; z-statistic -1.8700; P-value 0.0620
  - D_Gold Price Volatility t-1: Coefficient 0.2225; Std. Err. 1.2613; z-statistic 0.1800; P-value 0.8600
  - D_Gold Price Volatility t-2: Coefficient -0.0490; Std. Err. 1.2941; z-statistic -0.0400; P-value 0.9700

- Panel 2: Dependent variable: D_Gold Price Volatility
  - EC: Coefficient 0.0014; Std. Err. 0.0006; z-statistic 2.1900; P-value 0.0280
  - D_REER Volatility t-1: Coefficient 0.0070; Std. Err. 0.0042; z-statistic 1.6500; P-value 0.1000
  - D_REER Volatility t-2: Coefficient -0.0116; Std. Err. 0.0043; z-statistic -2.6800; P-value 0.0070
  - D_Gold Price Volatility t-1: Coefficient 0.7370; Std. Err. 0.0736; z-statistic 10.0100; P-value 0.0000
  - D_Gold Price Volatility t-2: Coefficient -0.0153; Std. Err. 0.0755; z-statistic -0.2000; P-value 0.8400

- Panel 3: Coefficients of the long-run relation
  - REER Volatility: Coefficient 1.0000
  - Gold Price Volatility: Coefficient -12.299; Std. Err. 2.7483; z-statistic -4.4800; P-value <0.001

- Note: The table includes values and statistical significance of the short run coefficients of the VECM model as well as the error correction term (EC).

### Tests of weak exogeneity
- Table 6: Testing Weak Exogeneity
  - Sample (1979-1994)
    - Equation D_rvolat: Parms 1; chi2 0.2371; P>chi2 0.6263
    - Equation D_gvolat: Parms 1; chi2 14.8384; P>chi2 0.0001
    - Note: H0 - weak exogeneity of the variable
  - Sample (1995-2011)
    - Equation D_rvolat: Parms 1; chi2 8.2769; P>chi2 0.0040
    - Equation D_gvolat: Parms 1; chi2 4.8090; P>chi2 0.0283
    - Note: H0 - weak exogeneity of the variable

*Source: _wp12168 - REFERENCES*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12168.pdf_