Chapter

Chapter 11. Financial Integration Ahead of East African Monetary Union

Author(s):
Paulo Drummond, S. Wajid, and Oral Williams
Published Date:
January 2015
Share
  • ShareShare
Show Summary Details
Author(s)
Masafumi Yabara

Well-functioning financial markets can accelerate economic growth and alleviate poverty. A large body of research has found a positive relationship between financial market development and economic growth, including in sub-Saharan Africa (Levine and Zervos, 1998; Adjasi and Biekpe, 2006a; Collins, 2004).1 Developed financial markets promote growth by mobilizing domestic savings and investments, efficiently allocating mobilized resources to local companies, and allowing diversification of risks. In addition, deep and liquid local financial markets can lessen an economy’s vulnerability to external shocks by reducing currency and duration mismatches in raising funds. Cross-country evidence shows that financial development can reduce income inequality by increasing the income of the poor (Making Finance Work for Africa, 2007).

Despite these merits, financial markets remain underdeveloped in most low-income countries because of structural constraints. Limited income and a small private sector make investors and issuers scarce. Running capital markets entails huge start-up and operating costs for both regulators and market participants. This can be prohibitive for countries with limited capacity and small markets: authorities are required to establish and manage regulatory frameworks and trading platforms, and issuers need to go through painstaking due diligence for initial public offerings and maintain detailed financial reporting afterwards. An empirical study suggests that a certain minimum-efficient size of bond markets exists, because large issuance and trading volumes are more economical (Eichengreen and Luengnaruemitchai, 2004).

Regional integration can help countries overcome these constraints. Integrated financial markets, if managed properly, will allow pooling of savings across the region, cost and information sharing among members, wider diversification of risks, enhanced competition and innovation across financial institutions, richer choice of financial products provided to regional and foreign investors, and more integration into the global economy facilitated by more attractive markets (Irving, 2005; Making Finance Work for Africa, 2007).

East African Community (EAC) countries have been pursuing financial market development through regional integration. These countries face growing financing needs to build up infrastructure for sustained growth, making well-functioning local financial markets important. Financing through markets could complement commercial bank financing, which dominates EAC financial sectors, where competition is low (Sanya and Gaertner, 2012). Recognizing the benefits of financial markets and the limitations of individual country approaches, the EAC member states are committed to establishing a common market, which would include free movement of capital under the treaty establishing the community. Furthermore, integration of financial markets, in particular government debt markets, is essential for a monetary union to transmit common monetary policy effectively across the region and realize the full benefits.

This chapter empirically investigates whether actions taken under the EAC framework have succeeded in advancing financial integration. Few previous studies exist on this subject. IMF (2009) assesses comovements of government bond yields among Kenya, Uganda, and Tanzania. Wang (2010) measures deviations from covered interest rate parity in foreign exchange markets for the same countries. Both of these studies conclude that EAC financial integration is limited.

This paper contributes to the literature by comprehensively assessing the degree of integration of the financial markets (treasury, interbank, stock markets) of the five EAC countries where possible, and examining whether integration has progressed.

Although there is no universal definition of financial integration, in general, financial markets are said to be integrated when the law of one price holds. In perfectly integrated financial markets with no barriers to cross-border transactions, returns of comparable assets should be equalized across markets, as long as there is no difference in country and exchange rate risks as well as in transaction costs. In this sense, it should be noted that deep financial integration can be achieved without any institutionalized unification of markets: the markets of the United States and the United Kingdom are said to be highly integrated, although these markets are separate under different legal frameworks.

The literature relies on two broad categories of measures to assess financial integration: price-based and quantity-based measures. The former directly estimate whether and at what speed rates of return of comparable assets converge across borders. The latter investigate correlation between domestic savings and investment, building on the idea of Feldstein and Horioka (1980) that in a world of high capital mobility, there should be no relation between domestic savings and investments, because domestic investments are financed by a pool of global savings under a unified interest rate. This paper employs the price-based measures because they have a clear-cut interpretation; price data are simple and relatively reliable compared with savings and investment data in low-income countries, and the measures have high frequency, allowing assessment of progress in integration over a relatively shorter time series.

The next section briefly describes EAC financial markets, focusing on debt and stock markets, the following section reviews the EAC authorities’ efforts to integrate their financial markets under the EAC framework, and the final section empirically assesses the depth and progress of financial market integration in the EAC using the methodology of beta and sigma convergence and cointegration analysis.

East African Community Financial Markets

Current Market Structure

Debt Markets

Each of the five EAC countries operates a government debt market, although at different stages of development. Central banks in each hold auctions under different frameworks (Table 11.1) to sell Treasury bills and bonds on behalf of the governments as instruments of monetary and fiscal policy implementation. These auctions are open to nonresidents, except in Tanzania, where nonresidents are prohibited from holding government securities. Issued securities are traded over the counter, on local stock exchanges, or both, although the secondary markets are largely inactive, as argued below.

Table 11.1Auction of Government Securities
BurundiKenyaRwandaTanzaniaUganda
T-billT-bondT-billT-bondT-billT-bondT-billT-bondT-billT-bond
Maturity13-52 weeks2-5 years13-52 weeks1-30 years4-52 weeks2-5 years5-52 weeks2-10 years13-52 weeks2-10 years
Auction frequencyWeeklyWeeklyWeeklyMonthlyWeeklyMonthlyBiweeklyMonthlyBiweeklyMonthly
Minimum bid amountFbu 100 million (US$81,200)KSh 0.1 million (US$1,200)KSh 0.05 million (US$600)Rwf 0.1 million (US$200)TSh 0.5 million (US$300)Tsh 1 million (US$700)Ush 0.1 million (US$400)
NonresidentEligibleEligibleEligibleIneligibleEligible
Sources: African Development Bank, African Fixed Income and Derivatives Guidebook; and central bank websites.Note: Government securities are Treasury bills (T-bill) and bonds (T-bond). Fbu = Burundi francs; Ksh = Kenya shillings; Rwf = Rwanda francs; Tsh = Tanzania shillings; Ush = Uganda shillings.
Sources: African Development Bank, African Fixed Income and Derivatives Guidebook; and central bank websites.Note: Government securities are Treasury bills (T-bill) and bonds (T-bond). Fbu = Burundi francs; Ksh = Kenya shillings; Rwf = Rwanda francs; Tsh = Tanzania shillings; Ush = Uganda shillings.

Market size differs considerably among the countries (Table 11.2). Kenya leads the region, with government securities outstanding at 27.3 percent of GDP and with maturities of up to 30 years. Tanzania and Uganda follow with amounts outstanding at 10.3 percent and 8.1 percent of GDP, respectively. These two countries succeeded in extending the maturities of Treasury bonds to 10 years in the early 2000s. Markets in Burundi and Rwanda were recently instituted. The central bank of Burundi started auctioning government securities at the end of 2006, with maturities now up to 10 years. Rwanda launched its over-the-counter securities market in 2008 and started listing government securities there. A first five-year Treasury bond was marketed in Rwanda in 2010, and the over-the-counter market was converted to a full-fledged stock exchange in January 2011. The size of the market is relatively large in Burundi, at 8.5 percent of GDP, but Rwanda’s is small, at only 2.2 percent of GDP.2

Table 11.2Treasury Bills and Bonds Outstanding at End-2010
Burundi1KenyaRwandaTanzaniaUganda2
(Millions of US$)
Treasury bills and bonds1268,6121202,0801,230
Treasury bills2,04694445544
Treasury bonds6,566251,635686
(Percent of gross domestic product)
Treasury bills and bonds8.527.32.210.38.1
Treasury bills6.51.72.23.6
Treasury bonds20.80.58.14.5
Sources: East African Community central banks; IMF, World Economic Outlook database; and author’s calculations.Note: “…” indicates that data is not available.

As of September 2010. Includes other debts held by nonfinancial institutions.

As of June 2010.

Sources: East African Community central banks; IMF, World Economic Outlook database; and author’s calculations.Note: “…” indicates that data is not available.

As of September 2010. Includes other debts held by nonfinancial institutions.

As of June 2010.

Corporate bonds are issued in the EAC countries, except for Burundi, and traded at local stock exchanges. But the markets are in the early stages of development and are inactive, with local companies mainly relying on commercial bank financing. The amounts outstanding are negligible when measured as a percent of GDP (Table 11.3); issuers are limited to financial institutions, especially foreign affiliated institutions. Transactions in the secondary markets rarely take place, including in the much larger Nairobi Stock Exchange (NSE), in Kenya.

Table 11.3East African Community Corporate Bond Markets at End-2010
KenyaRwandaTanzaniaUganda
Number of issuers10155
Amount outstanding (million US$)743.81.751.644.3
Amount outstanding (percent of GDP)2.40.030.260.29
Sources: Country stock exchanges; IMF, World Economic Outlook database; and author’s calculations.Note: There are no operational corporate bond markets in Burundi.
Sources: Country stock exchanges; IMF, World Economic Outlook database; and author’s calculations.Note: There are no operational corporate bond markets in Burundi.

Stock Markets

Disparities across the region are larger in stock markets than in debt markets. The NSE, established in 1954, has the longest history and is by far the largest in the region. It has 55 listed companies, with market capitalization of 46 percent of GDP as of end-2010 (Table 11.4). The Dar es Salaam Stock Exchange (DSE) in Tanzania and the Uganda Securities Exchange (USE), established in the late 1990s, have market values of about 15 percent of GDP. The Rwanda Stock Exchange (RSE) had only two companies listed as of end-2010,3 with transactions seldom taking place. Burundi has no stock exchange and capital is raised mainly from commercial banks.

Table 11.4East African Community Stock Markets at End-2010
KenyaRwandaTanzaniaUganda
Number of companies listed5521513
Market capitalization (million US$)14,4983,2531,810
Market capitalization (percent of GDP)46.016.111.9
Turnover (million US$)1,2830.0123.918.4
Turnover ratio18.80.71.0
Sources: Country stock exchanges; IMF, World Economic Outlook database; and author’s calculations.Note: There exists no operational stock market in Burundi. “…” indicates data is not available.

The ratio of turnover to market capitalization.

Sources: Country stock exchanges; IMF, World Economic Outlook database; and author’s calculations.Note: There exists no operational stock market in Burundi. “…” indicates data is not available.

The ratio of turnover to market capitalization.

Challenges for Financial Markets in the East African Community

Despite differing levels of development, the EAC countries face the same challenges as other low-income countries in developing domestic financial markets: low capitalization and liquidity. Owing to the costs of issuing and listing securities, issuers in the markets are overly confined to government entities, former state-owned enterprises, and foreign-affiliated banks. Low income and savings prevent individuals from participating in the markets, leaving investor bases heavily dominated by commercial banks and pension funds (Table 11.5). Whereas foreign investors’ transactions account for fair amounts of total turnover on the stock markets,4 available statistics, though coverage is restricted, indicate that nonresident holdings of securities are low in the EAC compared with the aggregate of sub-Saharan Africa (see Figure 11.4).5 As a result, market size has remained small, and governments are largely dependent on external sources of financing, including concessional borrowing (see Figure 11.5).

Table 11.5Bond Holding by Category of Investors(in percent, as of June 30, 2009)
Burundi1KenyaRwanda1Tanzania1Uganda
Commercial banks65Commercial banks53Commercial banks83Central bank41Commercial banks77
Central bank24Pension funds27Pension funds2Commercial banks30Pension funds15
Insurance companies6Insurance companies11Insurance companies2Pension funds23Central bank5
Pension funds4Individuals2Others13Nonbank2Insurance companies2
Others0.2Others6Insurance companies2Individuals1
Individuals1
Others1
Source: African Development Bank, African Fixed Income and Derivatives Guidebook.

Figures include only government security holdings, not corporate bonds.

Source: African Development Bank, African Fixed Income and Derivatives Guidebook.

Figures include only government security holdings, not corporate bonds.

Figure 11.1Comparison of Financial Markets

Sources: Country authorities; World Federation of Exchanges; IMF, World Economic Outlook database, Regional Economic Outlook: Sub-Saharan Africa, and Coordinated Portfolio Investment Survey; World Bank, World Development Indicators database; and author’s calculations.

Note: Turnover ratio is calculated as the ratio of turnover to market capitalization.

Figure 11.2Interst Rates and Stock Indices in the East African Community

Sources: IMF, African Department database; Bloomberg L.P.; Thomson Reuters; and author’s calculations.

Note: The Treasury bill rate for Rwanda is the weighted average rate of Treasury bills with maturities of 28, 91, 182, and 364 days. All stock indices are adjusted for exchange rates. Stock returns are computed as the log difference of the indices.

Figure 11.3Sigma Convergence in the Treasury Bill Markets

Sources: IMF, African Department database; and author’s estimates.

Note: Dotted lines denote linear time trends of sigma convergence, in periods before and after July 2007.

Figure 11.4Sigma Convergence in the Interbank Markets

Sources: IMF, African Department database; and author’s estimates.

Note: Dotted lines denote linear time trends of sigma convergence, in periods before and after July 2007.

Figure 11.5Sigma Convergence in the Stock Markets

Sources: Bloomberg L.P.; Thomson Reuters; and author’s estimates.

Low liquidity is also due to shallow investor bases. Local commercial banks and pension funds, the dominant investors in the region, generally tend to hold securities until maturity. Market infrastructure is another impediment. While trading infrastructure consisting of real-time gross settlement systems, clearing houses, and central securities depositories are all operational, except in Burundi; these systems are yet to be connected outside the borders, rendering foreign investors’ investments costly and time consuming.

These constraints create rather illiquid capital markets in the EAC. Figure 11.1 shows turnover ratios of nearly zero in the Tanzania and Uganda stock markets. There is no secondary market in Burundi: most investments are held to maturity.

Regional Initiatives to Integrate Financial Markets in the East African Community

As noted, the EAC countries have pursued financial market development through regional integration. In the treaty establishing the EAC, the member states committed to establishing a common market with free movement of capital. Specifically, the treaty calls for (1) removal of controls on capital transactions among the member countries (Article 86) and (2) harmonization of capital market infrastructure including regulations, taxation, accounting, trading systems, and cross-listings of securities (Article 85).6 The common market was officially launched in June 2010, awaiting full implementation by 2015. The Common Market Protocol requires legislation by each member to fully implement the common market by 2015. The annexes to the protocol provide timetables of actions to be undertaken by each state, including capital account liberalization.

Liberalizing capital transactions and harmonizing market infrastructure are essential and natural steps toward financial market integration. Regulations for cross-border capital transactions prevent domestic investors from freely participating in foreign markets and foreign investors from investing in domestic markets, making barriers to cross-border financial flows. Market infrastructure that is not harmonized hampers cross-border transactions and constitutes another barrier to financial market integration. Because financial transactions are affected by many factors such as regulatory frameworks, trading systems, and taxation, harmonization of these market settings is essential to realizing the law of one price. The ultimate form of financial market integration is the unification of the entire market infrastructure, in which all participants can engage in financial activities across borders in exactly the same way as they do in their home countries.

Capital Account Liberalization

Although liberalizing capital transactions across a region is the first step for integrated financial markets, experience indicates that capital account liberalization could cause crisis by making an economy vulnerable to external financial flows.7 To minimize the adverse effects, countries are advised to sequence liberalization and advance it along with comprehensive promarket reforms to maintain the stability of an economy (Ishii and Habermeier, 2002).

Kenya, Rwanda, and Uganda have already liberalized capital transactions within the region. Uganda was the first to fully open capital accounts in 1997, as part of a broader package of market-oriented reforms. Rwanda achieved full capital account liberalization in 2010. Even though restrictions on nonresidents’ investments in domestic markets remain in Kenya, East African investors are treated as local investors, meeting the commitment under the treaty.

Plans for gradual removal of capital controls are under way in Tanzania, while Burundi is lagging behind. Tanzania partially liberalized capital transactions in the 1990s. The Bank of Tanzania, recognizing the risk of opening the capital account, is formulating a plan for the gradual lifting of capital controls, in accordance with the Common Market Protocol. In Burundi, where capital markets are the least developed, the authorities still have significant control over capital transactions, and regulatory frameworks are yet to be established in some areas.

Harmonization of Market Infrastructure

In the EAC countries, the intent to harmonize market infrastructure was evident even before the establishment of the community. The capital market authorities of Kenya, Tanzania, and Uganda established the East African Member States Securities Regulatory Authorities in 1997 to enhance cooperation and advance market integration. Rwanda and Burundi joined in 2008 and 2011. The Capital Markets Development Committee, consisting of chief executives of the regulatory authorities and security exchanges, was established in 2001. It is a standing committee of the EAC, making policy recommendations on regulation and integration of the capital markets.8

Cooperation and harmonization are fairly advanced in the EAC compared with other regional integration arrangements in Africa (United Nations Economic Commission for Africa, 2008; African Development Bank, 2010). The East African Member States Securities Regulatory Authorities agreed on an approval procedure for cross-border listings in the EAC in 2000 and compiled common debt-ratio criteria for those wishing to issue debt securities. The organizations are also taking the lead in taxation of financial transactions, financial reporting standards, trading systems, and financial education. The USE has harmonized its listing rules with those of the NSE, although only six companies were cross-listed on the EAC stock exchanges at the end of 2010. Kenya, Uganda, and Tanzania are working toward demutualizing their respective stock exchanges, and merging them into a single regional stock exchange in the future.9 Regional initiatives are ongoing to integrate payment and settlement systems across the region.

Measuring Financial Market Integration in the East African Community

The previous section noted that EAC authorities made salient efforts to integrate their local financial markets. To gauge the success of their efforts, this section measures the progress of integration in the EAC financial markets, employing the methods of beta and sigma convergence and cointegration analysis.

Methodology

Beta and Sigma Convergence

Two concepts have been widely used in the literature to assess integration of financial markets. The first one, beta (β) convergence, is measured by the following regression with panel data:

where Si,t denotes a spread of yields on a relevant portfolio investment between country i and a benchmark market at time t, and l represents lag. If financial markets are perfectly integrated, this spread should be zero as long as securities traded have the same risks and maturity structures, following the law of one price (mean reversion). Therefore, a negative β coefficient indicates mean reversion taking place across the markets, and an absolute value of the coefficient represents the speed of convergence at which the spread is dissolved and investment returns on securities in country i converge with those in the benchmark market. The γl measures lagging effects of ΔSi in previous periods.

In this analysis, the benchmark market is assumed to be Kenya, as in IMF (2009), given its dominant size and development in the region. Thus the analysis focuses on the spreads of returns between Kenya and the other countries. Three-month lags are uniformly taken, with lags beyond the duration not being statistically significant in any of the estimates.

The second concept, sigma (σ) convergence, employs the cross-sectional standard deviation of yields across countries at each time, calculated as follows:

where n represents a number of the countries, Ri,t represents a return on a portfolio investment in country i at time t, and Rt identifies an average return in the region at time t. Regressing the computed sigma on a time trend tells us whether and at what pace the dispersion is decreasing and thus whether financial integration is deepening over time. Perfect convergence is realized when the sigma stays at zero. Beta and sigma convergence capture different aspects of financial integration: while beta convergence measures to what extent integration has been achieved in a fixed time framework, sigma convergence illustrates whether markets are moving toward deeper integration. Beta convergence is a necessary but not a sufficient condition for sigma convergence: beta convergence could be associated with sigma divergence (Sala-i-Martin, 1996).

An extensive number of empirical studies use these concepts, especially in the context of financial market integration in the European Union (EU). Adam and others (2002) apply these indicators to 10-year bond yields and interbank rates, as well as the mortgage rates and corporate loans rates of the EU countries, concluding that EU financial integration has increased, particularly since 1999.

Babetskii, Komárek, and Komárková (2008) use these indicators to assess stock market integration of the new EU member states, such as Czech Republic and Hungary, and find positive evidence. For other regions, Espinoza, Prasad, and Williams (2010) measure interest rate convergence in the Gulf Cooperation Council interbank markets and see evidence of integration, although little progress has been made since 2000.

Cointegration Analysis

Another approach widely used in assessing stock market integration is investigating the long-run equilibrium of returns among stock markets using cointegration modeling. Specifically, the following error correction model with l lags is considered:

where Pt is a (n × 1) vector of stock indices at time t, v is a (n × 1) vector of parameters, Π is a (n × r) parameter matrix with rank r < n, Γ1, …, Γl are (n × n) matrices of parameters, and εt is a (n × 1) vector of random errors. The essence of the approach is to identify r, a number of cointegrating vectors. If n variables with unit roots have r cointegrating relationships, they have nr common stochastic trends. Thus if r equals n – 1, stock markets are perfectly integrated under one common long-run trend. Alternatively if r equals zero, all data series are independent (Kasa, 1992). Johansen and Juselius (1990) derive two likelihood-ratio tests to infer on r, known as the trace statistics and the maximum-eigenvalue statistics.

A number of empirical studies implement this approach.10Serletis and King (1997) and Bley (2009) investigate stock market integration across the EU. Manning (2002), Click and Plummer (2005), and Yu, Fung, and Tam (2010) apply the technique to assess integration of Asian stock markets. However, there has been little empirical work on assessing long-run linkages among African stock markets. To my knowledge, only Wang, Yang, and Bessler (2003) and Adjasi and Biekpe (2006b) conduct cointegration analysis focusing on African stock markets. The former find no cointegration relationship among five African stock markets (South Africa, Egypt, Morocco, Nigeria, Zimbabwe) and the U.S. market between July 1999 and May 2002. The latter select seven African economies (Egypt, Ghana, Mauritius, Nigeria, Kenya, South Africa, Tunisia) and identify two cointegrations for November 1997 to August 2005.

Data

This chapter assesses integration of the three types of financial markets in the EAC—Treasury bill, interbank, and stock markets—using the methodology described previously. For the Treasury bill markets, monthly data on 91-day Treasury bill rates for Burundi, Kenya, Tanzania, and Uganda are retrieved from the IMF African Department database (Figure 11.2). For Rwanda, weighted average interest rates of 28-, 91-, 182-, and 364-day Treasury bills are alternatively employed because the data on the 91-day Treasury bill are available only after 2009. The database consists of data published or reported by the authorities. The data have, in total, 615 observations from September 2001 to November 2011, when data for all five countries were available. Monthly data on interbank market rates are also obtained from the same source. The data contain a total of 524 observations for January 2001 to December 2010, but Burundi is not included, reflecting its largely inactive interbank market (Figure 11.2).

For the stock markets, monthly averages of representative stock indices—the NSE 20 Share Index, the DSE All Share Index, and the USE All Share Index—are computed from daily data retrieved from Bloomberg L.P.11 Rwanda and Burundi were not included in the analysis because the RSE has no stock index, with only two companies listed at end-2010, and Burundi does not have a stock market. The data set covers November 2003 to January 2012. The DSE All Share Index, however, is available only after December 2006, resulting in 260 observations in total. For this analysis, the stock indices are converted to a common currency, using spot rates between the local currencies and the U.S. dollar.12 Stock returns are computed as the log difference of the indices (Figure 11.2).

Results

Treasury Bill and Interbank Markets: Beta Convergence

Beta convergence suggests a tendency for returns to converge across the EAC Treasury bill markets. The first three columns in Table 11.6 show the estimated beta coefficients using the panel data with ordinary least squares, fixed effects, and random effects models. The coefficients are negative and statistically significant, and robust to estimation method. The estimated beta indicates it takes more than a year before the magnitude of a deviation is reduced by half (a “half-life” of deviations).13 Four columns on the right side of Table 11.6 present the results of ordinary least squares regressions using individual yield spreads of Burundi, Rwanda, Tanzania, and Uganda from the benchmark Kenyan market. These individual regressions show substantive variation in the magnitude of beta coefficients, ranging from −0.021 for Burundi (a 32.4 month half-life) to −0.096 for Tanzania (a 6.9 month half-life), seemingly reflecting respective markets’ degrees of integration with the Kenyan market.

Table 11.6Beta Convergence in the Treasury Bill Markets
Pooled RegressionIndividual Regression
OLSFixed effectsRandom effectsBurundi-KenyaRwanda-KenyaTanzania-KenyaUganda-Kenya
β−0.052***−0.054**−0.052***−0.021−0.054*−0.096***−0.061
(0.014)(0.015)(0.017)(0.018)(0.024)(0.026)(0.040)
Number of observations476476476119119119119
R-squared0.1510.1520.1510.1230.1660.2220.240
Half-life of deviations (months)12.912.412.932.412.46.911.1
Sources: IMF, African Department database; and author’s calculations.Note: Robust standard errors are reported in brackets. *** denotes significance at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. Coefficients on constant, lagged variables, and country dummies (for the fixed effects model) are not reported. Implied half-life of deviations is calculated as ln(0.5)/ln(1 + b). OLS = ordinary least squares.
Sources: IMF, African Department database; and author’s calculations.Note: Robust standard errors are reported in brackets. *** denotes significance at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. Coefficients on constant, lagged variables, and country dummies (for the fixed effects model) are not reported. Implied half-life of deviations is calculated as ln(0.5)/ln(1 + b). OLS = ordinary least squares.

The speed of convergence in the Treasury bill markets has significantly increased in recent years. To evaluate the developments of beta over time, separate beta coefficients were estimated before July 200714 and after by modifying equation (11.1) as follows:

where βbfr and βbft denote beta coefficients before and after July 2007, respectively. Two dummy variables, Dbfr and Dbft, take values one before and after July 2007, respectively (and zero otherwise). Table 11.7 shows that speed of convergence has significantly increased since July 2007, reducing a half-life of deviations from 15 months to 6 months.

Table 11.7Beta Convergence in the Treasury Bill Markets—Before Full Membership and After
Pooled RegressionIndividual Regression
OLSFixed effectsRandom effectsBurundi–KenyaRwanda–KenyaTanzania–KenyaUganda–Kenya
β
Before July 2007−0.046** (0.015)−0.047** (0.016)−0.046** (0.015)−0.021 (0.017)−0.058* (0.025)−0.072* (0.028)−0.057 (0.043)
After July 2007−0.111*** (0.029)−0.114*** (0.029)−0.111*** (0.030)−0.158* (0.065)−0.036 (0.049)−0.164** (0.058)−0.106 (0.065)
Number of observations476476476119119119119
R-squared0.1590.1610.1590.1700.1670.2380.244
Test: β is the same before and after July 20074.57 (0.033)4.65 (0.032)3.32 (0.069)5.24 (0.024)0.23 (0.635)2.14 (0.146)0.43 (0.514)
Half-life of deviations (months)
Before July 200714.814.314.833.011.69.211.9
After July 20075.95.75.94.019.13.96.2
Sources: IMF, African Department database; and author’s estimates.Note: Figures in brackets are standard errors for coefficients and p values for Wald tests. *** denotes significance at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. Coefficients on constant, lagged variables, and country dummies (for the fixed effects model) are not reported. Implied half-life of deviations is calculated as ln(0.5)/ln(l + β). OLS = ordinary least squares.
Sources: IMF, African Department database; and author’s estimates.Note: Figures in brackets are standard errors for coefficients and p values for Wald tests. *** denotes significance at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. Coefficients on constant, lagged variables, and country dummies (for the fixed effects model) are not reported. Implied half-life of deviations is calculated as ln(0.5)/ln(l + β). OLS = ordinary least squares.

Looking at individual countries, the absolute values of beta coefficients have increased for all the countries but Rwanda. In particular, the beta convergence of Burundi has turned out to be statistically significant after July 2007, reducing a half-life of deviations from 33 months to 4 months. These results should be interpreted with caution as they are sensitive to the sample period, due to the small number of observations.

Beta convergence in the EAC interbank markets, shown in Table 11.8, also suggests mean reversion taking place: the coefficient is negative and statistically significant regardless of the models used. It also provides an intuitively sensible result that the speed of convergence is faster in the interbank markets (about a 6-month half-life) than in the Treasury bill markets (about a 13-month half-life). The estimated speed of convergence is comparable to that in the Gulf Cooperation Council markets (a half-life of deviations ranging from 3.6 to 5.5 months), estimated by Espinoza, Prasad, and Williams (2010).

Table 11.8Beta Convergence in the Interbank Markets
Pooled RegressionIndividual Regression
OLSFixed effectsRandom effectsRwanda - KenyaTanzania - KenyaUganda - Kenya
β−0.103**−0.116**−0.103***−0.033−0.139*−0.139*
(0.033)(0.039)(0.027)(0.067)(0.060)(0.061)
Number of observations381381381127127127
R-squared0.1410.1440.1410.0990.1960.193
Half-life of deviations (months)6.35.66.320.44.64.6
Sources: IMF, African Department database; and author’s estimates.Note: Robust standard errors are reported in brackets. *** denotes significance at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. Coefficients on constant, lagged variables, and country dummies (for the fixed effects model) are not reported. Implied half-life of deviations is calculated as ln(0.5)/ln(1 + b). OLS = ordinary least squares.
Sources: IMF, African Department database; and author’s estimates.Note: Robust standard errors are reported in brackets. *** denotes significance at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. Coefficients on constant, lagged variables, and country dummies (for the fixed effects model) are not reported. Implied half-life of deviations is calculated as ln(0.5)/ln(1 + b). OLS = ordinary least squares.

Speed of convergence has increased in the interbank markets, as in the Treasury bill markets (Table 11.9). The implied half-life of deviations has declined from seven months before July 2007 to around four months since then. The individual regressions also show increasing beta convergence for the countries, except Rwanda. However, as mentioned for the Treasury bill markets, this result is to be interpreted with caution, as the number of observations is relatively limited and most of the estimated beta coefficients lack statistical significance.

Table 11.9Beta Convergence in the Interbank Markets—Before Full Membership and After
Pooled RegressionIndividual Regression
OLSFixed effectsRandom effectsRwanda -KenyaTanzania -KenyaUganda -Kenya
β
Before July 2007−0.088*−0.100**−0.088***−0.048−0.126*−0.097
(0.034)(0.039)(0.014)(0.056)(0.062)(0.064)
After July 2007−0.165*−0.187*−0.1650.004−0.204−0.402***
(0.083)(0.090)(0.111)(0.126)(0.174)(0.129)
Number of observations381381381127127127
R-squared0.1440.1490.1440.1020.1990.244
Test: β is the same before and after July 20070.740.940.530.240.185.08
(0.391)(0.334)(0.468)(0.622)(0.673)(0.026)
Half-life of deviations (months)
Before July 20077.56.57.514.25.16.8
After July 20073.93.33.93.01.3
Sources: IMF, African Department database; and author’s estimates.Note: Figures in brackets are standard errors for coefficients and p values for Wald tests. *** denotes significant at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. Coefficients on constant, lagged variables, and country dummies (for the fixed effects model) are not reported. Implied half-life of deviations is calculated as ln(0.5)/ln(1 + β).
Sources: IMF, African Department database; and author’s estimates.Note: Figures in brackets are standard errors for coefficients and p values for Wald tests. *** denotes significant at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. Coefficients on constant, lagged variables, and country dummies (for the fixed effects model) are not reported. Implied half-life of deviations is calculated as ln(0.5)/ln(1 + β).

Treasury Bill and Interbank Markets: Sigma Convergence

Sigma convergence reveals that integration of the EAC Treasury bill markets has deepened somewhat since 2001, but the progress has stagnated in recent years. Figures 11.3 and 11.4 show the calculated sigma convergence in the EAC Treasury bill and interbank markets, respectively, as well as the linear time trends before and after July 2007. The figures present sigma convergence both with respect to dispersion across the region and dispersion between Kenya and each of the other markets. The figures suggest a downward trend in the dispersion in the EAC markets since 2001; however, the trend has reversed since July 2007 in most markets.

Table 11.10 shows the results of regressions of σt on a linear time trend, summarizing the time trends arising in Figures 11.3 and 11.4. The coefficient of σt in the Treasury bill markets changed from −0.07 before July 2007 to 1.2 after, and the same reverse of the time trend can be found in the interbank markets. These changes are statistically significant. The widening of the dispersion has further accelerated since July 2011, especially in the interbank markets. This is due to high inflation in the region caused by global commodity price hikes and drought in the region: countries hit by these exogenous shocks responded by tightening monetary policy and increasing interest rates considerably (Figure 11.1), widening the dispersion of interest rates across the region. The rising trend of dispersion, however, cannot be solely attributed to these shocks, as the trend emerged before 2011, in particular in Rwanda and Uganda.15

Table 11.10Sigma Convergence in the Treasury Bill and Interbank Markets
EACBurundi-KenyaRwanda-KenyaTanzania-KenyaUganda-Kenya
Treasury Bill Market
Time trend of σ
Before July 2007−0.072***−0.123***−0.013−0.004−0.062***
(0.007)(0.018)(0.011)(0.010)(0.014)
After July 20071.207***0.054***0.068***−0.0050.074***
(0.327)(0.011)(0.010)(0.015)(0.009)
Constant
Before July 20077.363***10.493***3.682***3.616***5.863***
(0.336)(1.035)(0.668)(0.474)(0.935)
After July 20070.040***0.081−0.0562.224***−0.390
(0.011)(0.229)(0.206)(0.484)(0.240)
Number of observations123123123123123
R-squared0.9220.7660.6560.7860.562
Test: trend of σ is the same before and after July 200774.2368.4229.520.0065.74
(0.000)(0.000)(0.000)(0.983)(0.000)
Interbank Market
Time trend of σ
Before July 2007−0.052***−0.001−0.059***−0.049***
(0.005)(0.008)(0.008)(0.009)
After July 20070.049**0.090***0.0010.052**
(0.018)(0.027)(0.021)(0.016)
Constant
Before July 20075.435***2.484***5.575***5.040***
(0.279)(0.450)(0.346)(0.577)
After July 20071.693***0.1461.979***0.770*
(0.446)(0.505)(0.557)(0.389)
Number of observations131131131131
R-squared0.8660.5890.7720.594
Test: trend of σ is the same before and after July 200729.3410.607.0131.17
(0.000)(0.001)(0.009)(0.000)
Sources: IMF, African Department database; and author’s estimates.Note: The regressions are estimated with OLS. Figures in brackets are robust standard errors for coefficients and p values for Wald-tests. *** denotes significance at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. EAC = East African Community; OLS = ordinary least squares.
Sources: IMF, African Department database; and author’s estimates.Note: The regressions are estimated with OLS. Figures in brackets are robust standard errors for coefficients and p values for Wald-tests. *** denotes significance at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. EAC = East African Community; OLS = ordinary least squares.

Stock Markets: Beta and Sigma Convergence

Beta convergence implies mean reversion taking place among the stock markets of Kenya, Uganda, and Tanzania (Table 11.11). The beta coefficients, estimated at −0.55 in the panel regressions, are statistically significant regardless of the model employed. The results suggest surprisingly fast convergence in the stock markets, with less than a month of half-life. The beta coefficients remain significant if regressions are separately estimated using the individual spreads of returns in Tanzania and Uganda. These individual regressions suggest that integration between the Ugandan and the Kenyan stock markets (a 0.4-month half-life) is stronger than that between the Tanzanian and the Kenyan markets (about a month half-life), as suggested in Figure 11.2.

Table 11.11Beta Convergence in the Stock Markets
Pooled RegressionIndividual Regression
OLSFixed EffectsRandom EffectsTanzania-KenyaUganda-Kenya
β−0.551**−0.556**−0.551***−0.521*−0.792
(0.182)(0.179)(0.058)(0.203)(0.182)
Number of observations1141141145794
R-squared0.3850.3850.3850.3950.426
Half-life of deviations (months)0.870.850.870.940.44
Sources: Bloomberg L.P.; Thomson Reuters; and author’s estimates.Note: Robust standard errors are reported in brackets. *** denotes significance at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. Coefficients on constant, lagged variables, and country dummies (for the fixed effects model) are not reported. Implied half-life of deviations is calculated as ln(0.5)/ln(1 + β). OLS = ordinary least squares.
Sources: Bloomberg L.P.; Thomson Reuters; and author’s estimates.Note: Robust standard errors are reported in brackets. *** denotes significance at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. Coefficients on constant, lagged variables, and country dummies (for the fixed effects model) are not reported. Implied half-life of deviations is calculated as ln(0.5)/ln(1 + β). OLS = ordinary least squares.

Sigma convergence suggests that divergence among the EAC stock markets has not diminished in the past few years. Estimated sigma convergence shown in Figure 11.5 seems to imply that the dispersion of stock returns among the NSE, the DSE, and the USE has declined slightly. However, regressing the estimated σt on a linear time trend reveals that the coefficient on σt is not statistically significant, although it has a negative sign (Table 11.12). This finding does not change if σt is computed as dispersion between the Kenyan market and each of the other two markets. Sigma convergence also shows that dispersion between the Tanzanian and Kenyan markets is larger than that between the Ugandan and Kenyan markets, consistent with the results of beta convergence. These results partly reflect that the global financial crisis had less effect on the Tanzanian stock market than on its neighbors (Figure 11.5), owing to its relatively limited integration with the global economy. These results of beta and sigma convergence in the stock markets are unchanged if the data are not adjusted for exchange rates.

Table 11.12Sigma Convergence in the Stock Markets
EACTanzania-KenyaUganda-Kenya
Time trend of σ−0.011−0.026−0.007
(0.016)(0.023)(0.005)
Constant4.436***5.343***2.432***
(0.673)(0.932)(0.326)
Number of observations616198
R-squared0.0060.0190.013
Sources: Bloomberg L.P.; Thomson Reuters; and author’s estimates.Note: The regressions are estimated with ordinary least squares. Figures in brackets are robust standard errors. *** denotes significance at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. EAC = East African Community.
Sources: Bloomberg L.P.; Thomson Reuters; and author’s estimates.Note: The regressions are estimated with ordinary least squares. Figures in brackets are robust standard errors. *** denotes significance at the 0.1 percent, ** at the 1 percent, and * at the 5 percent levels. EAC = East African Community.

Stock Markets: Cointegration Analysis

Two steps are necessary before conducting cointegration analysis: (1) verifying whether the data series are nonstationary, containing unit roots, and (2) selecting the number of lags used in the model. For the first step, augmented Dickey-Fuller and Phillips-Perron tests are conducted on the stock price data. The results uniformly confirm that all the data series are modeled as integrated of order one and appropriate for cointegration analysis. For the second step, four lags are chosen following the Akaike’s information criterion and a likelihood-ratio test. Adjusting or not adjusting for exchange rates does not change these results.

The result indicates that there is no long-run relationship among the EAC stock markets. Given the nonstationarity and the number of lags identified, Johansen’s trace tests are implemented to determine cointegration rank. The tests indicate that there is no cointegration vector in the EAC stock markets (Table 11.13). The maximum-eigenvalue statistics suggest one cointegrating relationship, but when the data are not adjusted for the exchange rates, both the trace statistics and maximum-eigenvalue statistics indicate no cointegration. I further conducted the same analysis following different information criteria for lags and using semimonthly data, but found no cointegration vector among the EAC stock prices.

Table 11.13Cointegration Tests for the East African Community Stock Markets
Trace StatisticsMaximum-eigenvalue Statistics
Number of Co-integrating VectorsAdjusted for Exchange RatesNot Adjusted for Exchange Rates5 percent Critical ValueAdjusted for Exchange RatesNot Adjusted for Exchange Rates5 percent Critical Value
r = 028.990*28.261*29.68022.61615.957*20.970
r = 16.37412.30415.4104.808*9.19914.070
r = 21.5663.1053.7601.5663.1053.760
Source: Author’s calculations.Note: * denotes the number of cointegrating vectors suggested by the statistics.
Source: Author’s calculations.Note: * denotes the number of cointegrating vectors suggested by the statistics.

Summary and Interpretation of the Empirical Results

The results suggest that beta convergence is taking place, or even strengthening, in the EAC financial markets. This means that there is a foundation on which risk-adjusted yield spreads across local markets are arbitraged through cross-border financial transactions. This finding contradicts IMF (2009), which finds that such convergence is not taking place in the EAC Treasury bond markets. It is presumably because IMF (2009) has a much smaller number of observations (46), due to the infrequency of Treasury bond issues and transactions. Comparing estimated beta convergence across the EAC markets, the speed of convergence of returns is much faster in the stock markets (a half-life of less than a month) than in the Treasury bill markets (about a 13-month half-life) and the interbank markets (about a 6-month half-life). This is most likely because investors, especially foreign ones, actively trade stocks in the secondary markets.

Measured sigma indicates that cross-sectional dispersion of returns among the EAC financial markets has been unchanged or even widening in some markets, over the past few years. How is this result compatible with statistically significant beta convergence? Remember beta convergence is a necessary, but not a sufficient condition for sigma convergence. Estimated beta indicates that yield spreads across local markets tend to be arbitraged by cross-border financial transactions. Nevertheless, sigma convergence shows such dispersion of yields still remains. This suggests that the remaining dispersion is caused more by differences in underlying country economic conditions (such as inflation and interest rate) and risks (exchange rate and political risks) than financial barriers that prevent market participants from exploiting arbitrage opportunities. Different fundamentals and risks naturally result in different returns from investments, preventing realization of the law of one price. The decrease in dispersion before 2007, especially in the Treasury bill and interbank markets, is most likely the result of the more stable fiscal and monetary policies adopted by most African governments over the past decade, rather than of deepened financial integration in the EAC. This trend has since stagnated as national economies experienced asymmetrical impacts from shocks such as the global financial crisis, political turmoil in Kenya, and drought in the region. Cointegration analysis, which suggests that there has been no long-run relationship among the EAC stock markets since 2007, complements this observation.

Conclusions

EAC financial markets are underdeveloped, although with significant divergence of developmental stages among the countries. Whereas government debt markets are functional in all the EAC countries, market size ranges from 2.2 percent of GDP in Rwanda to 27.3 percent in Kenya. Kenya has the most advanced stock market, Rwanda only just launched its stock market, and Burundi has no stock exchange. Despite these differences, the countries face the same challenges—low capitalization and liquidity of the financial markets. Issuers are largely confined to public institutions and foreign-affiliated banks. Investors are dominated by commercial banks and local pension funds, leaving participation of individual and foreign investors limited. These constraints, as well as insufficient market infrastructure, result in small and illiquid financial markets.

To overcome these challenges, the EAC countries have been pursuing regional integration of domestic financial markets by removing capital regulations and harmonizing market infrastructure. Kenya, Rwanda, and Uganda have fully liberalized capital transactions across the region, while Tanzania and Burundi are obliged to do so by 2015. Momentum toward harmonizing market infrastructure was evident even before the launching of the EAC, and cooperation is fairly advanced. The outputs have included a common procedure for cross-border listings, while further progress in wide areas including taxes, financial reporting, trading systems, and financial education is expected.

The empirical results on financial integration in the EAC are mixed. Estimated beta convergence indicates that convergence of investment returns is taking place in all the three financial markets assessed in this paper—Treasury bill, interbank, and stock markets. It is even shown that the speed of convergence has increased in recent years in the Treasury markets (and to less extent in the interbank markets). This result suggests that the authorities’ efforts at financial integration have reduced barriers to financial transactions across the borders, giving financial institutions motivation to take advantage of arbitrage opportunities across the markets. Nevertheless, estimated sigma convergence reveals that the mechanism of beta convergence has not succeeded in diminishing the dispersion of returns across local markets. On the contrary, the result shows increasing divergence in the Treasury bill and interbank markets for most countries. The cointegration analysis also indicates no long-run cointegration relationship among the EAC stock markets.

These results suggest that slow progress in the convergence of the EAC economies, rather than the existence of financial barriers, has impeded realization of the law of one price, or deepening of financial integration. Although the EAC states agreed upon a set of convergence criteria of economies in the context of establishing a monetary union, convergence of macroeconomic performance, such as inflation and fiscal balance, has been uneven. While it is imperative for the EAC countries to accelerate the momentum of removing barriers to financial transactions across the region, achieving a higher level of economic convergence and stability is essential in realizing integrated regional financial markets and reaping the full benefits of the monetary union to be established. Establishing effective mechanisms for conducting regional surveillance and enforcing sound national economic policies would help the countries achieve these objectives.

References

    AdamK.T.JappelliA. M.MenichiniM.Padula and M.Pagano2002Analyse, Compare, and Apply Alternative Indicators and Monitoring Methodologies to Measure the Evolution of Capital Market Integration in the European Union.” Centre for Studies in Economics and Finance Department of Economics and Statistics University of Salerno Italy.

    • Search Google Scholar
    • Export Citation

    AdeleganO. J.2008Can Regional Cross-Listings Accelerate Stock Market Development? Empirical Evidence from Sub-Saharan AfricaIMF Working Paper No. 08/281 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    AdeleganO. J.2009The Impact of the Regional Cross-Listing of Stocks of Firm Value in Sub-Saharan AfricaIMF Working Paper No. 09/99 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    AdjasiC. K. D. and N. B.Biekpe2006aStock Market Development and Economic Growth: The Case of Selected African CountriesAfrican Development Review Vol. 18 No. 1 pp. 144161.

    • Search Google Scholar
    • Export Citation

    AdjasiC. K. D. and N. B.Biekpe2006bCointegration and Dynamic Causal Links amongst African Stock MarketsInvestment Management and Financial Innovations Vol. 3 No. 4 pp. 102119.

    • Search Google Scholar
    • Export Citation

    African Development Bank2010Financial Sector Integration in Three Regions of Africa (Tunis, Tunisia).

    African Securities Exchanges Association2011African Securities Exchanges Association Yearbook 2011.” http://www.africansea.org.

    • Search Google Scholar
    • Export Citation

    BabetskiiI.L.Komárek and Z.Komárková2008Financial Integration of Stock Markets among New EU Member States and the Euro AreaCzech Journal of Economics and Finance Vol. 57 No. 7–8 pp. 341362.

    • Search Google Scholar
    • Export Citation

    BleyJ.2009European Stock Market Integration: Fact or Fiction?Journal of International Financial Markets Institutions and Money Vol. 19 No. 5 pp. 759776.

    • Search Google Scholar
    • Export Citation

    ClickR. W. and M. G.Plummer2005Stock Market Integration in ASEAN after the Asian Financial CrisisJournal of Asian Economics Vol. 16 No. 1 pp. 528.

    • Search Google Scholar
    • Export Citation

    CollinsS. M.2004International Financial Integration and Growth in Developing Countries: Issues and Implications for AfricaJournal of African Economies Vol. 13AERC Supplement 2ii55ii94.

    • Search Google Scholar
    • Export Citation

    EichengreenB. and P.Luengnaruemitchai2004Why Doesn’t Asia Have Bigger Bond Markets?NBER Working Paper No. 10576 (CambridgeMassachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    EspinozaR.A.Prasad and O.Williams2010Regional Financial Integration in the GCCIMF Working Paper No. 10/90 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    FeldsteinM. and C.Horioka1980Domestic Saving and International Capital FlowsEconomic Journal Vol. 90 No. 358 pp. 314329.

    • Search Google Scholar
    • Export Citation

    International Monetary Fund2008Regional Economic Outlook: Sub-Saharan Africa (Washington).

    International Monetary Fund2009Uganda: Selected IssuesIMF Country Report No. 09/37 (Washington).

    IrvingJ.2005Regional Integration of Stock Exchanges in Eastern and Southern Africa: Progress and ProspectsIMF Working Paper No. 05/122 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    IshiiS. and K.Habermeier2002Capital Account Liberalization and Financial Sector StabilityIMF Occasional Paper No. 211 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    JohansenS. and K.Juselius1990Maximum Likelihood Estimation and Inference from Cointegration—with Applications to the Demand for MoneyOxford Bulletin of Economics and Statistics Vol. 52 No. 2 pp. 169210.

    • Search Google Scholar
    • Export Citation

    KasaK.1992Common Stochastic Trends in International Stock MarketsJournal of Monetary Economics Vol. 29 No. 2 pp. 95124.

    • Search Google Scholar
    • Export Citation

    LevineR. and S.Zervos1998Stock Markets, Banks, and Economic GrowthAmerican Economic Review Vol. 88 No. 3 pp. 537558.

    • Search Google Scholar
    • Export Citation

    Making Finance Work for Africa2007Financial Sector Integration in Two Regions of Sub-Saharan Africa.” http://www.mfw4a.org/resources/documents/documents-details/financial-sector-integration-in-two-regions-of-sub-saharan-africa-how-creating-scale-in-financial-markets-can-support-growth-and-development.html.

    • Search Google Scholar
    • Export Citation

    ManningN.2002Common Trends and Convergence? South East Asian Equity Markets, 1988–1999Journal of International Money and Finance Vol. 21 No. 2 pp. 183202.

    • Search Google Scholar
    • Export Citation

    MurindeV.2012Financial Development and Economic Growth: Global and African EvidenceJournal of African Economies Vol. 21African Economic Research Consortium Supplement 2 pp. i10i56.

    • Search Google Scholar
    • Export Citation

    Sala-i-MartinX.1996Regional Cohesion: Evidence and Theories of Regional Growth and ConvergenceEuropean Economic Review Vol. 40 No. 6 pp. 13251352.

    • Search Google Scholar
    • Export Citation

    SanyaS. and M.Gaertner2012Assessing Banking Competition within the Eastern African CommunityIMF Working Paper No. 12/32 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    SerletisA. and M.King1997Common Stochastic Trends and Convergence of European Union Stock MarketsThe Manchester School Vol. 65 No. 1 pp. 4457.

    • Search Google Scholar
    • Export Citation

    SharmaG. D. and B. S.Bodla2010Are the Global Stock Markets Inter-linked? Evidence from the LiteratureGlobal Journal of Management and Business Research Vol. 10 No. 1 pp. 2939.

    • Search Google Scholar
    • Export Citation

    United Nations Economic Commission for Africa2008Assessing Regional Integration in Africa III (Addis Ababa: Economic Commission for Africa).

    • Search Google Scholar
    • Export Citation

    WangY. D.2010Measuring Financial Barriers among East African Community CountriesIMF Working Paper No. 10 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    WangZ.J.Yang and D. A.Bessler2003Financial Crisis and African Stock Market IntegrationApplied Economics Letters Vol. 10 No. 9 pp. 527533.

    • Search Google Scholar
    • Export Citation

    YuI.L.Fung and C.Tam2010Assessing Financial Market Integration in Asia—Equity MarketsJournal of Banking and Finance Vol. 34 No. 12 pp. 28742885.

    • Search Google Scholar
    • Export Citation
1Murinde (2012) provides a comprehensive survey of the literature in this field, from both theoretical and empirical perspectives, highlighting evidence specific to African economies.
2The relatively large scales of securities outstanding in Tanzania and Burundi are partly due to central bank holdings of government securities.
3In January 2011, Bralirwa, a brewing company, became the third company listed on the RSE.
4In 2010, foreign investors’ transactions accounted for 28.2 percent on the NSE and 21.8 percent on the DSE of the total value traded (African Securities Exchanges Association, 2011).
5Data are compiled from the Coordinated Portfolio Investment Survey conducted by the IMF and are available at http://www.imf.org/external/np/sta/pi/datarsl.htm. About 75 countries voluntarily participate in the survey, reporting cross-border holdings of portfolio investment securities, classified by issuers.
6Adelegan (2008) and (2009) find positive effects of cross-listings of securities in deepening stock markets and increasing values of companies cross-listed in sub-Saharan Africa.
7For the history of capital account liberalization in the EAC, see IMF (2008) and (2009).
8The committee was reorganized into the Capital Markets Insurance and Pensions Committee, with an expanded mandate covering insurance and pension development.
9The Rwanda Stock Exchange has been demutualized since its inception.
10For a literature review, see Sharma and Bodla (2010).
11Monthly frequency is adopted, taking the low liquidity of the markets into consideration, as shown in Figure 11.1. In the USE, for example, the number of deals and trading days were only 11 and 8, respectively, in November 2003, the beginning of the study period. Using weekly data in such a low-liquidity market is unlikely to appropriately reflect market fundamentals, biasing the analysis. Thus, monthly averages with less noise and more information to estimate a long-run relationship are employed in this analysis. Cointegration analysis using semimonthly data is conducted as a robustness check (discussed subsequently).
12Converting to a common currency may not be necessary in relatively advanced financial markets, because investors may hedge foreign exchange risks using forward contracts. But that is not necessarily the case for the EAC, because the forward markets are at nascent stages of development (Wang, 2010). I conducted the analyses using data series both adjusted and not adjusted for exchange rates, resulting in the same conclusions (discussed subsequently).
13A half-life of deviations is calculated as ln(0.5)/ln(1 + β).
14The timing when Burundi and Rwanda joined the EAC is chosen for analysis.
15Dropping observations in 2011 from the analysis does not change the declining trends of sigma convergence in the Treasury bill and interbank markets, although some of the time-trend coefficients lose statistical significance.

    Other Resources Citing This Publication