## _wp1090 — 16. Threshold AR Results

## Source details

**Canonical URL:** [_wp1090 — 16. Threshold AR Results](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1090.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1090.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1090.pdf.json)

---

### I. Introduction and approach
- Objective: assess extent of regional financial integration in the Cooperation Council of the Arab States of the Gulf (GCC) by examining financial system structure, capital flow volumes, interest rate convergence, and cross-listed equity prices.
- Rationale for combined volume and price evidence:
  - High capital flows may reflect diverging fundamentals rather than integration.
  - Price convergence may reflect common/global shocks rather than integration.
- Data used:
  - Monthly 3-month interbank rates in local currency for six GCC countries, 1993–2009 (Saudi Arabia used as benchmark).
  - Daily cross-listed stock prices for a set of GCC equities (about 15 cross-listed assets identified; usable data for 6).
  - CPIS and mergers & acquisitions data for limited volume evidence.

### II. GCC financial system: structure and restrictions
- Institutional and macro background:
  - GCC member countries: Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, U.A.E.
  - Common external tariff (CET) rate of 5 percent implemented in 2003; common market established in early 2008 (not fully implemented).
  - Dollar peg for all except Kuwait (peg to an undisclosed basket).
- Financial sector structure and size:
  - Banking system dominated by commercial banks; Saudi Arabia and the U.A.E. account for nearly 75 percent of total banks’ assets and 70 percent of capital.
  - Banks well capitalized; NPL ratios low by international standards.
  - Saudi banking sector is most closed: 8.6 percent of liabilities originate abroad; Bahrain most open: 47.0 percent foreign liabilities (Table 2).
- Restrictions that remain:
  - Limits on purchase of local shares by nonresidents; additional controls on inward FDI; rules on reserve requirements, open forex positions, dividend remittances, lending to nonresidents, and foreign borrowing by banks.
  - Some restrictions relaxed within the GCC; many apply without discriminating source of inflows.

### III. Available volume indicators
- Mergers and acquisitions (Inter-GCC investments total): 26,382.4 (In millions of U.S. Dollars) (Table 4 totals).
- CPIS-derived portfolio shares (selected):
  - Bahrain: share of GCC equities in total portfolio investments rose from 5 percent to 30 percent between 2004 and 2007 (Table 5).
  - Kuwait: share of GCC equities rose from 37 to 58 percent between 2003 and 2007 (Table 5).
  - Bahrain holdings of GCC long-term debt rose from 8 to 15 percent of total holdings between 2004 and 2007 (Table 6).
- Caveats:
  - Limited regional capital flow data; CPIS and M&A data likely biased toward countries with closer regional ties (Bahrain and Kuwait).

### IV. Interest rate convergence: methods and key metrics
- Beta (β)-convergence:
  - Panel regressions of spread dynamics (spread S_it relative to Saudi 3-month interbank rate).
  - Half-life computed as ln(0.5)/ln(1+β).
- Sigma (σ)-convergence:
  - Trend in cross-sectional standard deviation σ_t of interbank rates over time.
- Empirical specifications and diagnostics (selected from Table 8):
  - Spread t-1 coefficients across panel specifications: -0.119***, -0.174***, -0.119***, -0.119***, -0.146*** with standard errors [0.0148 0.0182 0.0148 0.021 0.0199].
  - Spread t-1 * DUM2000: -0.0951*** [0.0284].
  - Observations across specifications: 955.
  - R-squared: 0.078 0.116 0.08 0.12.
  - Hansen test p-value 1.0; AR(1) p-value 0.04; AR(2) p-value 0.198; Breusch Pagan p-value 0.008; Hausmann LM p-value 0.0004.

### V. Interest rate convergence: empirical findings
- Beta-convergence results:
  - Panel estimates indicate convergence of spreads among GCC countries (Saudi 3-month interbank rate as benchmark).
  - Fixed effects model best characterizes the process (Breusch-Pagan and Hausman tests support individual effects and fixed effects).
  - Panel half-lives for deviations from the benchmark ranged between 3½ to 5½ months.
  - Individual regressions half-lives ranged from 2 months (Saudi Arabia–Qatar) to about 5 months (Saudi Arabia–Kuwait).
  - Convergence accelerated after 2000: half-lives since 2000 ranged between 2 to 4 months (Table 8, column 5 evidence; coefficient on spread with dummy for periods >2000 significantly negative).
  - Interpretation: profit-seeking flows would take 2 to 5 months to clear half of interest rate arbitrage opportunities despite pegged regimes.
- Comparisons with ECCU (control group):
  - ECCU panel half lives for treasury bills, lending and deposit rates ranged between 4–14 months (T-bill half life 5–10 months; deposit rates 10–14 months; lending rates 4–10 months).
  - GCC convergence significantly faster than ECCU.
- Sigma-convergence results:
  - Cross-sectional standard deviation of interest rates displayed a declining trend over 1993–2009 (negative and significant time trend in σ_t; see Figure 6).
  - Post-2000, convergence slowed; cross-sectional variation increased with the onset of the global financial crisis toward the end of the sample period.
- Policy implication highlighted:
  - Given half lives of 2–5 months despite open capital accounts, there is a case for developing money market instruments to permit more effective management of structural liquidity.

### VI. Cross-listed stocks: data, measures, and methodology
- Sample and preprocessing:
  - Around 15 cross-listed assets identified in the GCC; usable data for 6 cross-listings.
  - Days without positive trading volume in both markets removed; final series shorter because only simultaneous trading days retained.
  - Trading-periods (time between simultaneous trading days) can exceed a week due to infrequent simultaneous trading.
- Price measures and models:
  - Cross-listed stock premiums P1/P2 - 1 used as measure of arbitrage opportunity and implied transaction costs.
  - Models: AR with GARCH for persistence; Threshold AR model to estimate neutral band (threshold c) and non-linear adjustment due to transaction costs.
  - Threshold AR results summarized in Table 16.

### VII. Cross-listed stocks: empirical findings and statistics
- Mean arbitrage/premium:
  - GCC mean arbitrage opportunity: around 2 to 4 percent.
  - Paper-wide summary: average arbitrage opportunity in GCC around 3 percent.
  - Comparative observations:
    - GCC mean arbitrage smaller than Hong-Kong/Singapore and Caribbean markets, larger than Petrobras (Argentina–Brazil) and larger than western developed markets where arbitrage opportunities hover around 1 percent.
- Persistence and half-lives:
  - Half-lives of price differentials measured in number of trading days: around two days (Table 15, Column B).
  - Half-lives measured in week days: around 6–7 days (Table 15, Column C).
  - Interpretation: when simultaneous trading occurs, arbitrage opportunities are seized quickly (≈2 trading days). However, infrequent simultaneous trading makes arbitrage persistent in calendar days (≈6–7 week days).
- Selected summary statistics (from Table 14):
  - Qatar - Abu-Dhabi mean 4.6 standard dev. 3.1 10% percentile 1.0 90% percentile 8.8.
  - Qatar Telecom (QTel) mean 4.6 standard dev. 3.1 10% percentile 1.0 90% percentile 8.8.
  - Bahrain - Kuwait (avg) mean 2.8 standard dev. 2.6 10% percentile 0.4 90% percentile 5.9.
  - Gulf Finance House (GFH) mean 2.6 standard dev. 2.6 10% percentile 0.3 90% percentile 5.7.
  - United Gulf Bank (UGB) mean 3.4 standard dev. 2.9 10% percentile 0.5 90% percentile 7.7.
  - Kuwait-UK GFH mean 7.4 standard dev. 7.1 10% percentile 0.9 90% percentile 18.1.
  - Comparative: Hong-Kong - Singapore (avg) mean 6.2 standard dev. 5.4 10% percentile 1.0 90% percentile 13.1; France - Germany (avg) mean 1.2 standard dev. 1.2 10% percentile 0.1 90% percentile 2.6.
- ARCH/GARCH and AR parameter highlights (from Table 15):
  - Qatar - Abu-Dhabi, QTel: β -0.37***; ARCH lag 1.57; half-life 7.3; 1/half-life 0.150; GARCH lag 0.17.
  - Bahrain - Kuwait (avg), GFH: β -0.29***; ARCH lag 2.26; half-life 6.4; 1/half-life 0.09; GARCH lag 0.88***.
  - Ithmaar: β -0.28***; ARCH lag 2.12; half-life 2.9; 1/half-life 0.14**; GARCH lag 0.83***.
  - Ahli United Bank: β -0.19***; ARCH lag 3.37; half-life 7.1; 1/half-life 0.13**; GARCH lag 0.82***.
  - United Gulf Bank (UGB): β -0.27***; ARCH lag 2.21; half-life 2.2; 1/half-life 0.300; GARCH lag 0.43.
  - Germany-UK GFH: β -0.94***; ARCH lag 0.20; half-life 0.3; 1/half-life 0.20**; GARCH lag 0.81***.
- Threshold AR findings (selected, from Table 16):
  - Qatar - Abu-Dhabi, QTel: c 2-0.80-0.47; additional coefficients 0.18 0.26.
  - Bahrain - Kuwait (avg), GFH: c 0.6-0.36-0.33; GFH coefficients 0.30 0.05 -0.29 *** 0.09 *** 0.89 ***.
  - Ithmaar: c 0.7 0.41 -0.35 *** 0.15 ** 0.83 ***.
  - Ahli United Bank: c 1.4 0.36 * -0.26 *** 0.13 ** 0.83 ***.
  - United Gulf Bank (UGB): c 0.3 -0.93 -0.29 *** 0.31 0.42.
  - Petrobras (Brazil - Argentina): c 1.8 -0.40 -0.21 0.12 *** 0.86 ***.
- Interpretation of threshold results:
  - Neutral band c estimates imply GCC markets fare well on average (neutral band small).
  - For three equities, threshold is very low compared to average premium—arbitrage opportunities quickly become large enough to attract capital.
  - Conclusion: presence of arbitrage opportunities driven more by illiquidity and infrequent simultaneous trading than by prohibitively large transaction costs.

### VIII. Causes of premiums and persistence (identified in text)
- Lagging information on events originating in other markets.
- Market illiquidity (large trades affecting price in one market).
- Inefficiencies in market functioning (number of opening days, sluggish order transfer).
- Actual transaction costs (order costs, exchange restrictions—though limited recently).
- Government intervention (reduces float and market liquidity).

### IX. Synthesis, conclusions, and policy-relevant takeaways
- Volume evidence (limited) suggests some regional integration, especially for Bahrain and Kuwait directing a large share of investments to the GCC; recent expansion of equity markets may bias volume indicators upward.
- Interest rate (price) evidence:
  - Beta-convergence indicates financial integration with half-lives of 2 to 5 months to clear interest rate differentials in the GCC.
  - Interest rate convergence faster in the GCC than in the ECCU.
  - Sigma-convergence improved mainly between 1993 and 2000, with limited progress afterwards and some widening during the global crisis.
- Equity price evidence:
  - Average arbitrage opportunity in GCC equity cross-listings around 3 percent.
  - Arbitrage removed within a couple of trading days when simultaneous trading occurs, but calendar persistence is longer (≈6–7 week days) due to infrequent simultaneous trading.
  - Pricing of cross-listings within the GCC more efficient than between GCC and Europe, suggesting regional integration at least at par with global integration.
- Main frictions limiting faster price/yield convergence:
  - Market illiquidity and infrequent simultaneous trading days.
  - Differences in information processing and market functioning, particularly between GCC and advanced markets.
- Policy-relevant recommendation highlighted:
  - Further development of money market instruments and measures to improve market liquidity and trading frequency could reduce arbitrage persistence and advance regional financial integration.

*Source: IMF Working Paper — "16. Threshold AR Results" (chapter/section content extracted from _wp1090).*

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

### _wp1090 - References

### Figures
- 1. Credit Default Swaps on Five-Year Senior ...........................................................................6
- 2. Nominal Exchange Rates and Interbank Interest Rates .........................................................7
- 3. Banking System Assets and Capital ......................................................................................8
- 4. GCC Stock Market Indices ..................................................................................................11
- 5. GCC Interbank Rates ...........................................................................................................12
- 6. Sigma Convergence .............................................................................................................16
- 7. Selected Cross-Listed Stocks in the GCC ............................................................................18
- 8. Gulf Finance House Cross-Listings .....................................................................................19

### Tables
- 1. Structure of the Banking Sector .............................................................................................8
- 2. Loans, Deposits and Foreign Assets and Liabilities ..............................................................9
- 3. Restrictions on Resident and Nonresident Deposit Accounts ..............................................10
- 4. Mergers and Acquisitions ....................................................................................................10
- 5. Portfolio Investment Asset in GCC Partner - Equity ...........................................................11
- 6. Portfolio Investment Asset in GCC Partner - Long-Term Debt Securities .........................11
- 7. Prudential Measures, Latest Available Data ........................................................................24
- 8. Tests of Beta Convergence ..................................................................................................25
- 9. ECCU: T-Bill Rates Tests of Beta Convergence .................................................................26
- 10. ECCU: Deposit Rates Tests of Beta Convergence ............................................................27
- 11. ECCU: Lending Rates Tests of Beta Convergence ...........................................................28
- 12. Selected Sources and Destination of Cross-Listings .........................................................29
- 13. Cross-Listed Stocks in the GCC ........................................................................................30
- 14. Arbitrage Opportunities Summary Statistics .....................................................................31
- 15. ARCH Results ....................................................................................................................32

*Source: _wp1090 - References*

### 16. Threshold AR Results ...............................................................................................

### 16. Threshold AR Results

### I. Introduction and approach
- Objective: assess extent of regional financial integration in the Cooperation Council of the Arab States of the Gulf (GCC) by examining financial system structure, capital flow volumes, interest rate convergence, and cross-listed equity prices.
- Rationale: combine volume and price data because:
  - High capital flows may reflect diverging fundamentals rather than integration.
  - Price convergence may reflect common/global shocks rather than integration.
- Data used:
  - Monthly 3-month interbank rates in local currency for six GCC countries, 1993–2009 (Saudi Arabia used as benchmark).
  - Daily cross-listed stock prices for a set of GCC equities (about 15 cross-listed assets identified; usable data for 6).
  - CPIS and mergers & acquisitions data for limited volume evidence.

### II. GCC financial system: structure and restrictions
- Institutional and macro background:
  - GCC member countries: Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, U.A.E.
  - Common external tariff (CET) rate of 5 percent implemented in 2003; common market established in early 2008 (not fully implemented).
  - Dollar peg for all except Kuwait (peg to an undisclosed basket).
- Financial sector structure and size:
  - Banking system dominated by commercial banks; Saudi Arabia and the U.A.E. account for nearly 75 percent of total banks’ assets and 70 percent of capital.
  - Banks well capitalized; NPL ratios low by international standards.
  - Saudi banking sector is most closed: 8.6 percent of liabilities originate abroad; Bahrain most open: 47.0 percent foreign liabilities (Table 2).
- Restrictions that remain:
  - Limits on purchase of local shares by nonresidents; additional controls on inward FDI; rules on reserve requirements, open forex positions, dividend remittances, lending to nonresidents, and foreign borrowing by banks.
  - Some restrictions relaxed within the GCC; many apply without discriminating source of inflows.

### III. Available volume indicators
- Mergers and acquisitions (Inter-GCC investments total): 26,382.4 (In millions of U.S. Dollars) (Table 4 totals).
- CPIS-derived portfolio shares (selected):
  - Bahrain: share of GCC equities in total portfolio investments rose from 5 percent to 30 percent between 2004 and 2007 (Table 5).
  - Kuwait: share of GCC equities rose from 37 to 58 percent between 2003 and 2007 (Table 5).
  - Bahrain holdings of GCC long-term debt rose from 8 to 15 percent of total holdings between 2004 and 2007 (Table 6).
- Caveats: limited regional capital flow data; CPIS and M&A data likely biased toward countries with closer regional ties (Bahrain and Kuwait).

### IV. Price data, models, and methodology
- Interest rate convergence measures:
  - Beta (β)-convergence: panel regressions of spread dynamics (spread S_it relative to Saudi 3-month interbank rate) to detect mean reversion; half-life computed as ln(0.5)/ln(1+β).
  - Sigma (σ)-convergence: trend in cross-sectional standard deviation σ_t of interbank rates over time.
- Equity price measures:
  - Cross-listed stock premiums P1/P2 - 1 used as measure of arbitrage opportunity and implied transaction costs.
  - Models: AR with GARCH for persistence; Threshold AR model to estimate neutral band (threshold c) and non-linear adjustment due to transaction costs.

### V. Interest rate convergence: empirical findings
- Beta-convergence results:
  - Panel estimates indicate convergence of spreads among GCC countries (Saudi 3-month interbank rate as benchmark).
  - Fixed effects model best characterizes the process (Breusch-Pagan and Hausman tests support individual effects and fixed effects).
  - Panel half-lives for deviations from the benchmark ranged between 3½ to 5½ months.
  - Individual regressions half-lives ranged from 2 months (Saudi Arabia–Qatar) to about 5 months (Saudi Arabia–Kuwait).
  - Convergence accelerated after 2000: half lives since 2000 ranged between 2 to 4 months (Table 8, column 5 evidence; coefficient on spread with dummy for periods >2000 significantly negative).
  - Interpretation: profit-seeking flows would take 2 to 5 months to clear half of interest rate arbitrage opportunities despite pegged regimes.
- Comparisons with ECCU (control group):
  - ECCU panel half lives for treasury bills, lending and deposit rates ranged between 4–14 months (T-bill half life 5–10 months; deposit rates 10–14 months; lending rates 4–10 months).
  - GCC convergence significantly faster than ECCU.
- Sigma-convergence results:
  - Cross-sectional standard deviation of interest rates displayed a declining trend over 1993–2009 (negative and significant time trend in σ_t; see Figure 6).
  - Post-2000, convergence slowed; cross-sectional variation increased with the onset of the global financial crisis toward the end of the sample period.
- Policy implication noted in text:
  - Given half lives of 2–5 months despite open capital accounts, there is a case for developing money market instruments to permit more effective management of structural liquidity.

### VI. Cross-listed stocks: empirical findings
- Sample and preprocessing:
  - Around 15 cross-listed assets identified in the GCC; usable data for 6 cross-listings.
  - Days without positive trading volume in both markets removed; final series shorter because only simultaneous trading days retained.
  - Trading-periods (time between simultaneous trading days) can exceed a week due to infrequent simultaneous trading.
- Mean arbitrage/premium:
  - GCC mean arbitrage opportunity: around 2 to 4 percent (text summary).
  - Authors’ statement: "at around 2 to 4 percent, the mean arbitrage opportunity is smaller in the GCC than in the markets of Hong-Kong and Singapore and in the Caribbean. The GCC average arbitrage opportunity is however larger than the price differential obtained for the company Petrobras listed in Argentina and Brazil. The GCC markets are also less efficient than western developed markets where arbitrage opportunities hover around 1 percent."
  - Paper-wide summary: average arbitrage opportunity in GCC around 3 percent (conclusion).
- Persistence and half-lives:
  - Half-lives of price differentials measured in number of trading days: around two days (Column B in Table 15).
  - Half-lives measured in week days: around 6–7 days (Column C in Table 15).
  - Interpretation: when simultaneous trading occurs, arbitrage opportunities are seized quickly (≈2 trading days). However, infrequent simultaneous trading makes arbitrage persistent in calendar days (≈6–7 week days).
- Regional vs global comparisons:
  - Within-GCC arbitrage opportunities smaller and vanish more quickly than price differentials between a GCC market and world markets.
  - Example: Gulf Finance House (GFH) cross-listings:
    - Bahrain–Kuwait, Bahrain–London, Bahrain–Frankfurt, London–Frankfurt premiums plotted (Figure 8).
    - GFH premium between London and Frankfurt was 2 percent on average before the fall of Lehman, and reached 5.5 percent afterwards (text).
  - Arbitrage opportunities between the Gulf and advanced markets noted as “very large (exceeding 6 percent)” and as persistent as GCC price differentials.
- Neutral band (Threshold AR) results:
  - Neutral band c estimates (Table 16, Column A) imply GCC markets fare well on average (neutral band small).
  - For three equities, threshold is very low compared to average premium—arbitrage opportunities quickly become large enough to attract capital.
  - Conclusion from threshold estimates: presence of arbitrage opportunities driven more by illiquidity and infrequent simultaneous trading than by prohibitively large transaction costs.
- Potential causes of observed premiums and persistence (text list):
  - lagging information on events originating in other markets;
  - market illiquidity (large trades affecting price in one market);
  - inefficiencies in market functioning (number of opening days, sluggish order transfer);
  - actual transaction costs (order costs, exchange restrictions—though limited recently);
  - government intervention (reduces float and market liquidity).

### VII. Synthesis and conclusions
- Volume evidence (limited) suggests some regional integration, especially for Bahrain and Kuwait directing a large share of investments to the GCC; recent expansion of equity markets may bias volume indicators upward.
- Interest rate (price) evidence:
  - Beta-convergence indicates financial integration with half-lives of 2 to 5 months to clear interest rate differentials in the GCC.
  - Interest rate convergence faster in the GCC than in the ECCU.
  - Sigma-convergence improved mainly between 1993 and 2000, with limited progress afterwards and some widening during the global crisis.
- Equity price evidence:
  - Average arbitrage opportunity in GCC equity cross-listings around 3 percent.
  - Arbitrage removed within a couple of trading days when simultaneous trading occurs, but calendar persistence is longer (≈6–7 week days) due to infrequent simultaneous trading.
  - Pricing of cross-listings within the GCC more efficient than between GCC and Europe, suggesting regional integration at least at par with global integration.
- Main frictions limiting faster price/yield convergence:
  - Market illiquidity and infrequent simultaneous trading days.
  - Differences in information processing and market functioning, particularly between GCC and advanced markets.
- Policy-relevant takeaway: despite relatively open capital accounts and a common peg, further development of money market instruments and measures to improve market liquidity and trading frequency could reduce arbitrage persistence and advance regional financial integration.

*Source: IMF Working Paper — "16. Threshold AR Results" (chapter/section content extracted from the provided PDF).*

### REFERENCES

### _wp1090 - REFERENCES

### Key References Cited
- Adam, K., Jappelli T., Menichini A. M., Padula M., and M. Pagano (2002), Analyse, Compare,and Apply Alternative Indicators and Monitoring Methodologies to Measure the Evolution of Capital Market Integration in the European Union, Report to the European Commission.
- Baele, L., A. Ferrando, P. Hördahl, E. Krylova and C. Monnet (2004), Measuring Financial Integration in the Euro Area, Occasional No. Paper 14.
- Chai, J., 2006, “The Eastern Caribbean Currency Union Banking System in a Time of Fiscal Challenge,” in R. Sahay, D.O. Robinson and P. Cashin (editors), The Caribbean: From Vulnerability to Sustained Growth, Washington DC: International Monetary Fund, pp. 143-180.
- ECB (2005), Indicators of financial integration in the Euro Area, European Central Bank.
- Edison H. and C. Reinhart (2001), Stopping hot money, Journal of Development Economics, Vol 66, pp. 533–553.
- Espinoza R. and G. Kwon (2009), Regional Financial Integration in the Caribbean: Evidence from Financial and Macroeconomic Data, IMF Working Paper, WP 09/139.
- International Monetary Fund (2007), Exchange Rate Arrangements and Restrictions, washington DC.
- Kwon, G., S. Panth, J. Chai, A. Gerson, G.M. Casselle and R. Espinoza, 2008, “Financial Integration in the Caribbean,” in A. Bauer, P. Cashin and S. Panth (editors), The Caribbean: Enhancing Economic Integration, IMF Western Hemisphere Department, Washington DC: International Monetary Fund, pp. 3-43.
- Levy Yeyati E., S.L. Schmukler and N. Van Horen (2006), International Financial Integration through the Law of One Price, World Bank Working Paper WPS2897.
- Obstfeld M. and A. Taylor (1997), Nonlinear Aspects of Goods-Market Arbitrage and Adjustment: Hechkscher’s Commodity Points Revisited, Journal of the Japanese and International Economies, Vol 11. pp 441-479.
- Quah, Danny (1993), “Galton’s Fallacy and Tests of the Convergence Hypothesis”, Sandinavian Journal of Economics, 95 (4), 427-43.
- Tong, H. (1978), On a Threshold Model, in C.H. Chen (ed.) Pattern Recognition and Signal Processing, Amsterdam: Sijthoff and Noordhoff.
- Tsomocos D. (2003), Equilibrium analysis, banking and financial instability, Journal of Mathematical Economics, Vol. 39(5-6), pp. 619-655.

### Table 7 — Prudential Measures, Latest Available Data
- Countries: Bahrain 1/; Kuwait 2/; Oman; Qatar 2/; Saudi Arabia 3/; U.A.E. 4/
- Nonperforming Loans (percent): 3.0 3.1 3.2 1.0 2.1 2.5
- Capital Adequacy: 24.0 16.0 15.9 17.0 20.6 13.3
- Provisioning Rate: 84.7 107.6 91.0 142.9 101.5
- Return on Assets: 3.2 2.1 2.4 2.8 2.3
- Return on Equity: 12.7 27.8 14.7 21.7 30.5 21.1
- Sources: Global Financial Stability Report (GFSR); National Authorities.
- Notes: 1/ End-2008 for Non-Performing Loans and Capital Adequacy. End-2007 for Return on Equity. 2/ End-September 2008. 3/ End-2006 for Return on Equity, otherwise end-2007. 4/ End-June 2008.

### Table 8 — GCC: Tests of Beta Convergence (selected coefficients and diagnostics)
- Dependent variable/panel specifications: (1) OLS, (2) Fixed Effects, (3) Random Effects, (4) GMM, (5) GMM Fixed Effects.
- Spread t-1: -0.119*** -0.174*** -0.119*** -0.119*** -0.146*** [standard errors: 0.0148 0.0182 0.0148 0.021 0.0199]
- Spread t-1 * DUM2000: -0.0951*** [0.0284]
- ∆Spread t-1: 0.0207 0.050 0.0207 0.021 0.0710** [0.0321 0.0323 0.0321 0.067 0.0327]
- ∆Spread t-2: 0.0801** 0.105*** 0.0801** 0.08** 0.118*** [0.0318 0.0319 0.0318 0.022 0.0319]
- ∆Spread t-4: -0.0936*** -0.0702** -0.0936*** -0.102** -0.0513 [0.0321 0.0321 0.0321 0.05 0.0324]
- Saudi-Bahrain: 0.146*** 0.149*** [0.0369 0.0368]
- Saudi-Oman: 0.186*** 0.202*** [0.0387 0.0388]
- Saudi-Qatar: 0.153*** 0.153*** [0.0374 0.0372]
- Saudi-U.A.E.: 0.149*** 0.146*** [0.0372 0.0370]
- Constant: 0.033*** -0.08*** 0.0325*** 0.0325 -0.0765*** [0.0110 0.0253 0.0110 0.0295 0.0252]
- Observations: 955 across specifications.
- R-squared: 0.078 0.116 0.08 0.12
- Hansen test: p-value 1.0
- AR(1): p-value 0.04
- AR(2): p-value 0.198
- Breusch Pagan p-value 0.008
- Hausmann LM p-value 0.0004
- Note: Standard errors in brackets. Significance: *** 1 percent, ** 5 percent, * 10 percent.

### Tables 9–11 — ECCU: Tests of Beta Convergence (T-Bill, Deposit, Lending Rates; selected results)
- Table 9 (T-bill Spread):
  - T-bill Spread t-1: -0.0530*** -0.116*** -0.0530*** -0.0737*** [0.0199 0.0286 0.0199 0.00742]
  - ∆T-bill Spread t-1: -0.280*** -0.238*** -0.280*** -0.272*** [0.0488 0.0504 0.0488 0.0749]
  - ∆T-bill Spread t-2: -0.283*** -0.248*** -0.283*** -0.279*** [0.0503 0.0513 0.0503 0.0597]
  - Observations: 455. R-squared: 0.149 0.167 0.15. Hansen test p-value 1.0. AR(1) p-value 0.07. AR(2) p-value 0.74. Breusch Pagan p-value 0.95. Hausmann LM p-value 0.09.
- Table 10 (Deposit Spread):
  - Deposit Spread t-1: -0.0526*** -0.0589*** -0.0526*** -0.0722*** [0.0146 0.0155 0.0146 0.0276]
  - ∆Deposit Spread t-3: -0.112** -0.109** -0.112** -0.168*** [0.0467 0.0469 0.0467 0.0152]
  - Observations: 455. R-squared: 0.05 0.05 0.05. Hansen test p-value 1.0. AR(1) p-value 0.23. AR(2) p-value 0.73. Breusch Pagan p-value 0.29. Hausmann LM p-value 0.90.
- Table 11 (Lending Spread):
  - Lending Spread t-1: -0.0691*** -0.149*** -0.0691*** -0.111*** [0.0181 0.0263 0.0181 0.0131]
  - ∆Lending Spread t-1: 0.0149 0.0586 0.0149 0.0401*** [0.0475 0.0480 0.0475 0.0111]
  - St. Kitts-Dominica: 0.217*** [0.0600]; St. Kitts-Grenada: 0.145*** [0.0548]; St. Kitts-St. Vincent: 0.158*** [0.0548]
  - Observations: 455. R-squared: 0.04 0.07 0.05. AR(1) p-value 0.04. AR(2) p-value 0.48. Breusch Pagan p-value 0.33. Hausmann LM p-value 0.003.

### Table 12 — Selected Sources and Destination of Cross-Listings (counts)
- GCC: Kuwait to Bahrain 10; Qatar to U.A.E. 2.
- ASEAN+3: China mainland (all to HKG) 185; Hong Kong 8; Japan 1; Malaysia 1; Singapore 1; Korea 2.
- CARICOM: Trinidad and Tobago (authorities’ data) 9.
- EU and Israel: Spain (to France) 3; France (to Spain) 2; Germany (to France) 4; France (to Germany) 400; Spain (to Germany) 119; Euro Area (to UK) 180; UK (to Euro Area) 1943; Israel (to Euro Area) 105.
- Brazil: Brazil (to Argentina) 1.
- NAFTA: Canada (to the U.S.) 1479; Mexico (to the U.S.) 54; U.S.A. (to Canada) 41.
- Pacific Island Forum: Australia (to New Zealand) 26; New Zealand (to Australia) 21.
- Others: India (to Thailand) 1; South Africa (to Ghana) 2; South Africa (to Namibia) 30; South Africa (to Zimbabwe) 2.
- Sources: DataStream; and country authorities for GCC and Caribbean markets.

### Table 13 — Cross-Listed Stocks in the GCC (selected observations)
- Qatar - Abu-Dhabi: Qatar Telecom (QTel) Start date 5-Jun-03 End date 22-Oct-08 Days Trading 291 No. of Days Since First Simultaneous Trading 1966 Average No. of Week Simultaneous Trading 4.8
- Bahrain - Kuwait (avg): Gulf Finance House (GFH) Start date 14-Jun-04 End date 13-Jul-09 Days Trading 881 No. of Days Since First Simultaneous Trading 1855 Average No. of Week Simultaneous Trading 3.3
- GFH listings in Germany-UK, Bahrain-UK, Kuwait-UK: Days Trading 390, 308, 356; No. of Days Since First Simultaneous Trading 718, 735, 735; Average No. of Week Simultaneous Trading 1.3, 1.7, 1.5
- Brazil - Argentina: Petrobras Start date 28-Apr-06 End date 11-Jul-07 Days Trading 297 No. of Days Since First Simultaneous Trading 439 Average No. of Week Simultaneous Trading 1.1
- Sources: Bloomberg; countries’ authorities; and authors’ calculations.

### Table 14 — Arbitrage Opportunities Summary Statistics (selected series)
- Qatar - Abu-Dhabi mean 4.6 standard dev. 3.1 10% percentile 1.0 90% percentile 8.8
- Qatar Telecom (QTel) mean 4.6 standard dev. 3.1 10% percentile 1.0 90% percentile 8.8
- Bahrain - Kuwait (avg) mean 2.8 standard dev. 2.6 10% percentile 0.4 90% percentile 5.9
- Gulf Finance House (GFH) mean 2.6 standard dev. 2.6 10% percentile 0.3 90% percentile 5.7
- United Gulf Bank (UGB) mean 3.4 standard dev. 2.9 10% percentile 0.5 90% percentile 7.7
- Kuwait-UK GFH mean 7.4 standard dev. 7.1 10% percentile 0.9 90% percentile 18.1
- Trinidad and To. - Jamaica mean 6.6 standard dev. 5.8 10% percentile 0.9 90% percentile 13.0
- Hong-Kong - Singapore (avg) mean 6.2 standard dev. 5.4 10% percentile 1.0 90% percentile 13.1
- France - Germany (avg) mean 1.2 standard dev. 1.2 10% percentile 0.1 90% percentile 2.6
- Source: Authors' calculations.

### Table 15 — ARCH Results (selected parameters and diagnostics)
- Qatar - Abu-Dhabi, Qatar Telecom (QTel): β -0.37***; ARCH lag 1.57. half-life 7.3; 1/half-life 0.150; GARCH lag 0.17.
- Bahrain - Kuwait (avg), GFH: β -0.29***; ARCH lag 2.26; half-life 6.4; 1/half-life 0.09; GARCH lag 0.88***.
- Ithmaar: β -0.28***; ARCH lag 2.12; half-life 2.9; 1/half-life 0.14**; GARCH lag 0.83***.
- Ahli United Bank: β -0.19***; ARCH lag 3.37; half-life 7.1; 1/half-life 0.13**; GARCH lag 0.82***.
- United Gulf Bank (UGB): β -0.27***; ARCH lag 2.21; half-life 2.2; 1/half-life 0.300; GARCH lag 0.43.
- Germany-UK GFH: β -0.94***; ARCH lag 0.20; half-life 0.3; 1/half-life 0.20**; GARCH lag 0.81***.
- Notes: 1/ ln(1/2)/ln(1+β). 2/ ln(1/2)/ln(1+β)*(n.of days between simultaneous trading). 3/ The AR model for this listing does not include a constant (the model with constant generated a β lower than -1).

### Table 16 — Threshold AR Results (selected coefficients)
- Qatar - Abu-Dhabi, Qatar Telecom (QTel): c 2-0.80-0.47; additional coefficients 0.18 0.26.
- Bahrain - Kuwait (avg), GFH: c 0.6-0.36-0.33; GFH coefficients 0.30 0.05 -0.29 *** 0.09 *** 0.89 ***.
- Ithmaar: c 0.7 0.41 -0.35 *** 0.15 ** 0.83 ***.
- Ahli United Bank: c 1.4 0.36 * -0.26 *** 0.13 ** 0.83 ***.
- United Gulf Bank (UGB): c 0.3 -0.93 -0.29 *** 0.31 0.42.
- Petrobras (Brazil - Argentina): c 1.8 -0.40 -0.21 0.12 *** 0.86 ***.
- Notes: 1/ The AR model for this listing does not include a constant (the model with constant generated a β lower than -1). Columns (b)(c) and (d)(e) denote AR/ARCH lag and GARCH lag parameters as presented.

*Source: _wp1090 - REFERENCES (IMF).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1090.pdf_