Kim - The Effect of Bonds Market on Growth

a panel data set of 38 countries, and applying the generalized method of moments techniquesfor dynamic panels, we find that (i) stock market development is positively related toeconomic growth; (ii) the contributing role of bank credit to economic growth diminishes asdomestic bond markets develop; (iii) government bonds are positively related to economicgrowth, while the effects of corporate bonds change from negative to positive, as domesticfinancial structures expand in size and diversity.

2013 Elsevier Inc. All rights reserved.

Economic growth

that fn the

International Review of Economics and Finance 27 (2013) 529541

Contents lists available at SciVerse ScienceDirect

International Review of Economics and Finance

j ourna l homepage: www.e lsev ie r .com/ locate / i re fby not only promoting capital accumulation but also fostering efficient allocation of resources and technological innovations(Levine, 1997; Wachtel, 2001). It should be pointed out at the outset that economic growth is both country-specific andtime-specific. It is a complex process of innovation which is shaped by the dynamic interplay between the industrial structure andfinancial system peculiar to individual countries at a given point in time.1

While it appears that there exists a general consensus around the idea that an efficient mechanism of financial intermediationis critical in mobilizing savings and diversifying investment risks, growth literature shows sharply divided views on the specificdirection of causality between financial development and economic growth. (Beck & Levine, 2004; Hassan, Sanchez, & Yu, 2011)On the one hand, a priori intuition and empirical evidence have given rise to a positive relationship between financialdevelopment and economic growth. For instance, LaPorta, Lopez-de-Silanes, Shleifer, and Vishny (1998) find that countries withThe neoclassical economists' view,fulcrum. This view has been replaced ilegal systems which promote good corpor

We are grateful to the editor, Hamid Beladi and twofor their research and editorial assistances. This studEducation and Research at the University of Memphis Corresponding author.

E-mail address: [email protected] (C.S. Pyun)1 We are grateful to an anonymous referee for poin

1059-0560/$ see front matter 2013 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.iref.2013.01.008inancial markets play subordinate roles to the real sector of economy, has long lost itsliterature by the general proposition that financial markets promote economic growthPanel data analysisBond markets

1. IntroductionAccepted 22 January 2013Available online 31 January 2013

JEL classification:G2O1

Keywords:Financial marketLinking the missing market: The effect of bond markets oneconomic growth

Patara Thumrongvit a, Yoonbai Kim a, Chong Soo Pyun b,a University of Kentucky, United Statesb University of Memphis, United States

a r t i c l e i n f o a b s t r a c t

Article history:Received 5 August 2011Received in revised form 18 January 2013

Past studies have largely focused on the positive role of banks and stock markets on economicgrowth. This paper adds bond markets as a third key component of the financial system. Usingate governance and creditors' rights had faster economic growth, while Goldsmith

anonymous referees for their helpful comments. We are also indebted to Daewon Kim and Frank SanPietroy was partially supported by a summer research grant from the Wang Center for International Business.

.ting to us the innovative process of nancial system on economic growth.

ll rights reserved.

(1969), King and Levine (1993a, 1993b), Rousseau and Wachtel (2000) and Beck and Levine (2004) affirm the positive impactthat well-functioning banks and stock markets have had on economic growth.

Alternatively, Lucas (1998) and Favara (2003) are skeptical about the positive role of financial markets that has had on

530 P. Thumrongvit et al. / International Review of Economics and Finance 27 (2013) 529541economic growth. Rajan and Zingales (2003), Allen, Qian, and Qian (2005), Shyn and Oh (2008), Ergungor (2008) and Jalil,Feridun, and Ma (2010) each refute the validity of the premise that determinants of economic growth would feature such apositive role for financial markets. In particular, Ergungor (2008) shows that economic growth effects and financial systemdevelopment are bidirectional, and thus, a country's financial system's structure is irrelevant to economic growth. Rajan andZingales (2003) draw attention to a great reversal of economic growth experienced by Argentina during the twentieth centurydespite her earlier success in financial development, while Allen et al. (2005), and Shyn and Oh (2008) point to recent rapideconomic growth achieved by South Korea and China without the benefit of a well-developed domestic financial markets.

The overwhelmingmajority of literature on the relationship between financial markets and economic growth has focused on twokey institutions banks and stock markets with scant attention to bond markets. There are two major reasons for researchers'collective inattention to bond markets: First, bond financing is subsumed as part of debt financing, which has been historicallydominated by banks. This point is especially relevant in many economies where banks have preempted bondmarkets in general andcorporate bondmarkets in particularwith their high capitalization and liquidity (Ma, Remolona, &He, 2006). Second,while stocks aretraded at exchanges or their equivalent and their price discovery process can be analyzed by trading data, which is publicly available,bonds, are traded over-the-counters, where transaction data are not transparent nor made publicly available.

The motivations for our study are: First, little is known about the micro-structure of bond markets, such areas as priceformation and how the liquidity necessary for bond purchases is supplied. This is true even for the U. S. bond market, the world'slargest and most liquid market. (Biais, Decierck, Dow, Portes, & von Thaddden, 2006) Second, the development of domesticgovernment bond markets historically took place well before domestic stock markets came into being (Litan, Pomerleano, &Sundararajan, 2003). As a result, in many countries, especially in Europe, the total amount of bonds outstanding is greater thanthat of credit provided by banks, and the size of government bond markets trump that of the respective stock markets (Biais et al.,2006; Fink, Haiss, & Hristoforova, 2003).2 Third, bond markets are expected to experience rapid growth worldwide for tworeasons: i) Many countries now nurture their domestic bond markets as a defensive measure against the type of financial criseswhich many Asian and Latin American countries faced in the 1990's and ii) growing numbers of affluent investors in developingcountries are pouring their savings into bond markets in search of high yield and portfolio diversification (See Braun & Briones,2005; de la Torre, Gozzi, & Schumukler, 2007; Herring & Chtusripitak, 2000; Tudor, 2012). Finally, as Levine (1997) notes, anydebate which focuses solely on bank or equity market-based systems may be misdirected, and future research on financialstructure relative to economic growth should include more diverse countries other than the U. S., U. K. and Euro-zone countries.

The dearth of transaction data notwithstanding, several studies including Herring and Chtusripitak (2000), Braun and Briones(2005) and Fink, Haiss, Kirchner, and Moser (2006) examine the relationship between bond markets and economic growth. Onbalance, their finding is that the relationship is positive. It should be noted that none of these studies have used panel data methods.Thesemethods enable regression analysis of a pooled data set involving a spatial and temporal dimension and allow for control of theeffects of missing variables and the bias arising from simultaneity of key cross-sectional data being analyzed.3 Recently, Beck andLevine (2004) employed a panel data analysis on the effects of two sectors of financial markets banks and stock markets oneconomic growth.

Our paper extends the Beck and Levine (2004) study by introducing the third major financial sector; the domestic bond market,which we divide into two subsectors; government and corporate bonds. In addition, as shown in Section 5 and Appendix II, we alsorefine Beck and Levine's system estimator by including the small-sample correction for two-step standard errors developed byWindmeijer (2005). We analyze, in this study, 38 countries: Argentina, Australia, Austria, Belgium, Canada, Chile, China, CzechRepublic, Denmark, Finland, France, Germany, Greece, HongKong, Hungary, Iceland, India, Ireland, Italy, Japan, South Korea,Malaysia,Mexico, the Netherlands, New Zealand, Norway, Philippines, Poland, Portugal, Singapore, South Africa, Spain, Sweden, Switzerland,Thailand, Turkey, the U. K. and the U. S.

Our findings are that; (i) stock market development is positively related to economic growth; (ii) the contributing role of bankcredit to economic growth diminishes as domestic bond markets develop; (iii) government bonds are positively related toeconomic growth, while the effects of corporate bonds change from negative to positive, as domestic financial structures expandin size and diversity.

The remainder of the paper is organized as follows. Section 2 starts with a preliminary description about the bond marketincluding its role and the historical development in various parts of the world. Section 3 provides a brief survey of literature.Section 4 describes definitions of key variables, descriptions of data, descriptive statistics, and preliminary findings from OLS.Section 5 discusses the model and presents the empirical results. Concluding remarks are found in Section 6.

2. The bond market

The bond market plays a critical role in the efficient functioning of capital markets. By channeling savings, it makes fundsavailable to long-term borrowers. In doing so, it sets the benchmark interest rates for debt instruments with varying maturities

2 A notable exception would be several Asian emerging economies, such as China and Korea, where the development of their stock markets preceded theemergence of their domestic government bond markets. See Park (2008, p. 15).3 A good survey article on the subject is found in Wachtel (2001).

531P. Thumrongvit et al. / International Review of Economics and Finance 27 (2013) 529541and risks, both of which serve as key criteria which inform household decision to save and lend, and business decisions to borrowand invest for long term projects. This role of rationing capital is of particular importance to the financegrowth nexus becauselong-term financing is essential to productivity gains in an economy which come from technological innovation and changingfactor inputs over time.

Bonds may be public or private indebtedness. Along with equity, they are used to finance long-term investments. In the U.S.,the bond market is about the same size as the stock market while, in Europe, bonds amount to approximately two thirds of totalsecurities outstanding. Further, the distribution of the total debt differs significantly between the two regions; in Europe, the bondmarket is dominated by government bonds and bonds issued by financial intermediaries. In the U.S., the proportion of bondsissued by the non-financial sector is much larger (Biais et al., 2006). Measured by relative percentage of country GDP, in 2010, theUS bond market is 177 percent of GDP, while Euro-zone country bond markets are approximately 70 percent. Japan's bondmarket is the largest at 252 percent of GDP.

The contrast is even more revealing for emerging economies in the Asia-Pacific. While the size of the bond market relative tothe country's GDP is very small e.g., 41 percent for India, 51 percent for China and 61 percent for Singapore the bond marketsin the region are expanding at an unprecedented rate having tripled their size since 2007. In the first six months of 2012 alone,companies in the Asia-Pacific region, excluding Japan, have issued a record $398 billion in bonds, up 29 percent from the sameperiod last year. An important point is that during the same period the volume of syndicated bank loans is down by 44 percent.This trend is expected to continue as the bond market boom is underwritten largely by the increasing number of affluent,indigenous investors in the region (Tudor, 2012).

In many countries in East Asia, the absolute amount of household savings is high, and their financial market assets haverecently grown significantly. However, the ability of these countries to intermediate saving efficiently has not kept pace witheconomic growth: Businesses still over-rely on banks as their principal source of financing. An important point, that should benoted here, is that many East Asian governments are implementing measures that expedite the expansion of corporate bondmarkets by instituting legal reforms that streamline issuance rules and offer tax incentives to the growing number of affluent localinvestors. (Asian Development Bank 2009) Appendix I provides data on relative size of bank assets, stock markets and domesticbond markets of selected countries.

3. A brief survey of literature

The vast majority of previous empirical studies which examined the role of financial markets on economic growth, havelargely ignored the effect of bond markets on the economic growth process. Bond markets were ignored under the notion thatwhat matters is the state of a country's overall financial development, and the differences in composition and institutions thatmake up a country's financial system are trivial so long as an economy has access to a well-functioning financial system (Levine,1997; Merton, 1992). This is because a well-functioning financial market enables economic agents to not only reduce transactioncosts through hedging, trading and pooling risks, but also increase the liquidity and size of the capital market, all of which areessential for economic growth (Fry, 1995; Goldsmith, 1969; King & Levine, 1993a, 1993b; McKinnon, 1973; Wachtel, 2001). Inparticular, Levine (1997) suggests two specific channels a capital accumulation channel and a technological innovation channel through which financial market improvements may promote economic growth. Under this line of thinking, it is sufficient to dividethe financial market into two broad sectors banks and stock markets as these sectors provide complementary growth-enhancingfinancial services.

In the seminal studies of Goldsmith (1969) and King and Levine (1993a, 1993b), the extent of financial development ismeasured by a single analytical variable that represents the size of the banking sector, such as bank credit or bank assets. Withregard to the roles of banks and stock markets, Levine (2002) finds that, after controlling for overall financial development,cross-country comparisons do not suggest that distinguishing between banks and stock markets is a first-order concern indeveloping an understanding of the economic growth process. Tadesse (2002), however, argues that bank-based systems are farbetter among countries with underdeveloped financial sectors while market-based systems outperform bank-based systemsamong countries with developed financial sectors. Other studies which analyzed the roles of banks and stock markets includeLevine and Zervos (1998), Rousseau and Wachtel (2000), and Beck and Levine (2004), all of which find that the relativedevelopment of banks and stock markets each help predict domestic economic growth with varying degrees of statisticalreliability. It is interesting to note that Ergungor (2008) views bank credit financing as a form of economic rent enforced in thosecountries which are unable to adapt their banking and securities laws to changing economic environments.

Regarding the role of stock markets, Ritter (2005) reports a negative relationship between inflation-adjusted stock returns andreal economic growth, measured in per capita terms, suggesting that the phenomenon of a fast growing economywith expandingstock markets is transitory, and once the valuations of a country's stock markets exceed a relative threshold, their role oneconomic growth is likely to show a negative relationship. In an early study involving bonds, Harvey (1989) finds that in the U.S.,the bond market is a better predictor of economic growth than the stock market. There is at best only weak theoretical support forexpecting a positive correlation between economic growthmeasured in per capita income and the contribution of stockmarket returns.Economic growth occurs for two major reasons productivity growth or increased factor inputs of capital and labor (Ritter, 2005).

Bond markets worldwide are increasingly being recognized as an important component of financial development.Accordingly, they should be analyzed as an integral part of any well-functioning financial market. In many ways, the bankingsector and bond market complement each other. (See Eichengreen & Luengnaruemitchai, 2004; Fink et al., 2003; Herring &Chtusripitak, 2000; Wachtel, 2001). A well-functioning bond market sets the benchmark interest rates for all debts with varying

risk and maturities, which promotes efficient use of resources for economic growth. On the other hand, in their role as financialintermediaries, banks provide essential services. In addition to the obvious role of gathering deposits, banks help to reduceinformation asymmetries by issuing securities and holding sizeable volumes of bonds (Booth & Smith, 1986; Hawkins, 2002).

In the absence of bond markets, banks tend to have reduced opportunities for investing deposits. These surplus deposits maythen put the bank into a position where they may fund suboptimal or unsound loans. This scenario affects the risk profile in twoways; first, there is the risk of default on the portfolio of unsound loans; second, as the bank is involved in these riskier loanportfolios, their opportunity to manage overall risk through the use of derivatives, such as options, futures and swaps is reduced.

532 P. Thumrongvit et al. / International Review of Economics and Finance 27 (2013) 529541To illustrate, banks in general are constantly exposed to a problem of maturity mismatch because their deposits are largely shortterm while their loans are generally made with much longer maturities. In addition, banks often borrow in foreign currencies andmake loans in their home currency, creating the problem of currency mismatch in addition to the maturity mismatch problem.4

The development of corporate bond markets worldwide is examined by Braun and Briones (2005), Fink et al. (2006), and Biaiset al. (2006). Fink et al. (2003) examine the impact of bondmarket development on real output in 13 highly developed economiesover the period 19502000 and find evidence that bond market development influences real economic activity. However, Abbasand Christensen's (2007) study, which evaluated the role of domestic debt makers for 93 low-income countries and emergingmarkets over the 19752004 period, shows little statistical evidence linking the growth of corporate bond markets to economicgrowth.

3.1. Data

The annual data used in this paper consists of the panel of 38 developed and developing countries for the period from 1989 to2010. Domestic bondmarket data are obtained from the Bank for International Settlements (BIS). The BIS Quarterly Review reportsthe data of domestic debt securities, which are divided into domestic securities privately issued and those issued by thegovernments. They include long-term bonds and notes, treasury bills, commercial paper and other short-term notes issued in thedomestic markets. All other data are obtained from International Financial Statistics (IMF) and World Development Indicators(World Bank). The definitions of the variables and the data sources are summarized in Table 1.

3.2. Key variables dened

We measure the dependent variable, economic performance, by the growth rate of real per capita GDP. The independentvariables are divided into two groups one representing financial development and the other control variables. We employ threemeasures of financial development which represent respectively; banking, stock market and bond market. Following Beck andLevine (2004), banking sector development is measured by Bank credit, which equals claims on the private sector by bank depositdivided by GDP,5 while the liquidity of the stock market is measured by Turnover ratio, defined as the value of the trades ofdomestic shares on a domestic exchange divided by the total value of listed shares. A liquid market provides a ready exit-optionfor investors and thus reduces disincentives for long-run investment (Beck & Levine, 2004). Other variables, such as marketcapitalization the value of listed shares divided by GDP and total value traded the trading value of domestic shares ondomestic exchanges divided by GDP produce similar results. We apply a similar measure to the bond market. The proxymeasure of the development in the bond market is Domestic bond, which equals the outstanding amount of domestic bondsdivided by GDP. Other measures for the liquidity of the bond market, such as total value traded, may be more suitable, but are notused due to the unavailability of the data.

The control variables include other potential determinants of economic growth that are typically used in the growth literatureas summarized in Barro and Sala-i-Martin (1995). We employ initial real GDP per capita, average years of schooling, the share ofexports and imports in GDP, the inflation rate, and the ratio of government expenditures to GDP.6 Another control variable included inthe model is composite risk, which combines the political, financial, and economic risk rating for a country ranging from very highrisk (049.5) to very low risk (80100).7

3.3. Descriptive statistics and preliminary ndings

Table 2 presents the descriptive statistics regarding the average growth rates of the sample economies and the correlationcoefficients among the three key variables in the study. There are wide variations in the size of financial institutions across thesample countries which are not shown in the table. For instance, during the first 12-year period from 1989 to 2002, China attainedthe highest growth rate of 12 percent in 1992, while Hungary had the lowest growth rate of a negative 12 percent in 1991. Thesize of the domestic bond market for Hong Kong was less than 2 percent of GDP in 1990, while that of Singapore was 364 percent

4 The currency mismatch is considered one of the important causes of the Asian currency crisis in 1997. The World Bank had urged developing countries toaccelerate the deepening of their domestic bondmarkets well before the Asian currency crisis in 1997 as a means of ameliorating these mismatch problems on thepart of banks (Dalla et al., 1995).5 All nancial variables are measured as ratios to GDP, where we use trend GDP instead of actual GDP. Trend GDP is obtained by regressing actual GDP on

constant, trend, and squared trend.6 Data on average years of schooling are available every ve years, therefore, it appears only in the OLS results.7 The International Country Risk Guide is the real source of the data, but during the early process of collecting the data for this paper, the data became also

available from World Development Indicators.

Table 1Definitions and sources of variables.

Variable Definition and construction Source

Total

Gover

Trade

533P. Thumrongvit et al. / International Review of Economics and Finance 27 (2013) 529541in 1989. Poland had the smallest size of stock market and bank credit in 1991 and 1989, respectively. While the size of the bondmarket in Hong Kong was very small in 1990, its stock market and bank credit are the biggest in the group.8

The coefficients of simple correlation among the key variables are reported in Table 2. Both bank credit and bond marketcapitalization are negatively correlated with economic growth, while the domestic stock market is positively and weaklycorrelated with it. These findings, which will be analyzed in Section 5.2 below, contrast with our findings for a fourteen-yearperiod between 1989 and 2003. Namely, during the earlier and shorter period, both stock market and bank credit are positivelycorrelated with economic growth and only domestic bond market is negatively correlated with economic growth.

Table 2 also shows that the three key financial submarkets do not necessarily develop concurrently or in any discerniblelock-step fashion, judging from the very low correlation coefficients among them. Our tentative observations for probable causesfor the negative relationship between bond markets and economic growth as well as for the lack of robust statistical relationshipbetween bank credit and economic growth are that there may exist a non-linear crowding effect of government borrowingwhich may diminish the lending capacity of banks. This premise is consistent with Abbas and Christensen (2007) who report thatat moderate levels, government debt boosts economic growth, but beyond a certain level, public debt reduces funds available to

and services to predicted GDP.Inflation rate The annual percentage change in CPI WDIAverage years of schooling The average schooling years in the total population age 25 and over. The data are available online at the Center

for International Development at Harvard University(http://www.cid.harvard.edu/ciddata/ciddata.html).non-g

3.4. Cr

Tabperiod

where the eyears onumbstudy

Thvariabcountrgrowtstockregres

8 A su9 In a

transpapercentage of GDP.value traded The total value of shares traded during the period as a

percentage of GDP.WDI.

nment consumption The ratio of general government final consumptionexpenditure to predicted GDP.

Calculated with the data from WDI.

openness The ratio of the sum of imports and exports of goods Calculated with the data from WDIGDP per capita GDP divided by midyear population. Data are inconstant 1995 U.S. dollar.

The World Development Indicators (WDI).

GDP per capita growth Log difference of real GDP per capita. Calculated with the data from WDI.Predicted GDP Residual of a regression of GDP on constant, trend

and square of trend.Calculated with the data from WDI.

Domestic bond market The ratio of the amount outstanding of domestic debtsecurities to predicted GDP.

Calculated with the data from the Bank forInternational Settlements (BIS).

Bank credit The ratio of claims on the private sector by deposit moneybanks to predicted GDP.

Calculated with the data from the InternationalFinancial Statistic (IFS).

Turnover ratio The ratio of total value traded to market capitalization. Calculated with the data from WDI.Market capitalization The share price times the number of shares outstanding as a WDI.overnmental borrowers.9

oss-country ordinary least squares regressions

le 3 presents the results of the ordinary least squares regressions of economic growth for the panel of the 19892003and 38 countries. The cross-country regression is given by:

Gi Fi Xi i; 1

G represents real per capita GDP growth, F the set of indicators of financial development, X the set of control variables, andrror term. Following Beck and Levine (2004), each of the five reported regressions control for initial income and averagef schooling. The regressions, which include bank credit, stock markets and bond markets, control sequentially for a limiteder of variables such as government consumption, trade openness, inflation, and composite risk of individual countries in theindividually. This helps minimize the over-fitting problem.e cross-country regression results shown in Table 3 are generally consistent with those reported in the literature. All controlles, except average years of schooling, have the expected effects on economic growth. Thus, trade openness, the reduction ofy risk, and government consumption exhibit the expected positive effects. Inflation is negatively related to economich, which is consistent with previous studies. However, coefficient estimates are insignificant. The effects of bank credit andmarket turnover ratio are quite similar to those reported by Beck and Levine in terms of their signs and magnitudes of thesion coefficients. It should be noted that the effects of the bond market are positive as well, albeit statistically insignificant.

mmary statistics covering the period from 1989 to 2002 is available upon request.ddition, our analytical model may not capture the externality effect of government spending on the economy. Moreover, improved efciency andrency in corporate and government bond markets may be as important as the sheer size of the market or its liquidity.

Table 2Summary statistics: 3-year average for 19892010.

Economic growth (percent) Financial assets in percent of GDP

Bonds Bank credit Stocks

Descriptive statisticsObservations 228 266 262 266MeanStd. dMinMax

CorreEcono

534 P. Thumrongvit et al. / International Review of Economics and Finance 27 (2013) 529541Our analysis of the sub-component data, which are not reported on Table 5, shows that the growth of corporate bond markets haspositive impacts on economic growth for the entire period between 1989 and 2010 (see Table 5), while the estimates from theshorter period covering from 1989 to 2003 indicate negative correlations between the development of corporate bondmarket andeconomic growth.10

Lee, Pesaran, and Smith (1997) suggest that cross-sectional regressions of this kind (that is, one observation per country)allow us to observe neither the dynamic adjustments involved in an individual country's output nor the heterogeneity of growthrates across the sample countries. These problems are addressed through the use of panel data and the estimation techniquesdescribed in the next section.

4. Panel data analysis

By using instrumental variables for all regressors, panel data analysis allows for more precise estimates of the financegrowthrelationship while being able to avoid biases associated with cross-country regressions of time-series and cross-sectional data(Beck & Levine, 2004). Using the panel data model with a system estimator, which is outlined by Arellano and Bover (1995) anddeveloped by Blundell and Bond (1998), also allows us to deal with omitted variable bias and simultaneity bias, both of which areserious shortcomings in the standard cross-sectional growth regressions used in previous studies. It should be noted that thesystem estimator used by Beck and Levine (2004) suffers from a small-sample bias because coefficient estimates are obtainedfrom the one-step estimator while the p-values are derived from the two-step estimator. The problem is, due to large numbers ofinstruments created by the estimator, the standard errors obtained in the two-step method are severely biased downward.

The system estimator is designed for a panel of data with a large number of cross-section units and a short time series. Ourannual data do not fit well with the criteria.11 Thus, we construct a panel with a 3-year average from non-overlapping data, whichnot only enables us to examine the medium- to long-run effects of banks, stock markets and bond markets on economic growth,but also allows us to shorten the time-series dimension of the panel so that it fits the criteria imposed by the estimator.

Since the number of countries in our data is not very large and the appropriate number of instruments should be lower thanthe number of cross-section units, we employ a procedure developed by Calderon, Chong, and Loayza (2000) to reduce (orcollapse) the dimension of the instrument matrix. In the event that the reduced instrument matrix still generates moreinstruments than the number of countries, we can further reduce the number of instruments by limiting the lags of variables.12

Bonds 0.199 1Bank credit 0.066 0.271 1Stocks 0.079 0.168 0.418 14.1. M

Th

10 Rele11 TheCalderoperiods12 Aninstrumconsiste2.43 71.39 84.12 74.37ev. 2.44 61.30 47.55 76.12

4.53 0.0041 11.73 0.0011.71 396.85 244.00 757.79

lationsmic growth 1odel specications

e regression equation of interest can be written as

yi;tyi;t1 1 yi;t1 Xi;t i;ti;t i i;tE i E i;t

h i E ii;t

h i 0

2

vant data and analysis are available upon request.re is no guideline on the dimension of data suitable for the system estimator. For example, using data with 15 time periods along with the procedure inn et al. (2000), a model with one endogenous variable and four predetermined variables can generate up to 88 instruments. The same model with 6 timegenerates 34 instruments.important side effect of collapsing the instrument matrix by limiting the lags of variables is the loss of information. However, reducing the number ofents not only reduces the bias in standard errors produced by the two-step estimator, but it also makes the specication tests for the estimatorncy more credible.

Eq. (2) can be rewritten as

yi;t yi;t1 Xi;t i;t 3

idiosy

Table 3Bond markets (aggregate), stock markets, banks and growth: Cross-country regression.

(1) (2) (3) (4) (5)

Constant 0.604 0.588 0.702 0.876 3.061(0.434) (0.442) (0.444) (0.733) (1.994)

Initial income 0.182*** 0.178*** 0.184** 0.182*** 0.216***(0.0638) (0.0614) (0.0675) (0.0638) (0.0666)

Average years of schooling 0.212 0.187 0.212 0.193 0.0579(0.186) (0.198) (0.185) (0.186) (0.202)

Government consumption 0.0370(0.0937)

Trade openness 0.0654(0.0544)

Inflation 0.0351(0.0652)

Composite risk 1.111*(0.610)

Bonds (overall) 0.0140 0.0117 0.0212 0.00881 0.00505(0.0348) (0.0343) (0.0356) (0.0432) (0.0345)

Turnover ratio 0.161** 0.154** 0.175** 0.159** 0.142*(0.0765) (0.0729) (0.0794) (0.0779) (0.0718)

Bank credit 0.0834 0.0821 0.0541 0.0505 0.00821(0.0678) (0.0662) (0.0824) (0.0930) (0.0898)

Countries 38 38 38 38 38R-squared 0.536 0.538 0.555 0.540 0.569

Notes: All variables are entered as logarithm. Exceptions are average years of schooling and inflation which are transformed as log(1+variable). p-values are inparentheses.

Table 4Bonds m

535P. Thumrongvit et al. / International Review of Economics and Finance 27 (2013) 529541(1) (2) (3) (4)Const

Initial

Gover

Trade

Inflat

Comp

Bond

Bank

Turno

HanseSerialNumbCountObser

NotesAll variaThe reg*, **, ***a The nub The nt, and bond market development. i is the fixed effects including the unobserved country-specific effect, i,t is thencratic shocks, and the subscripts i and t represent the country and time period, respectively.

arkets (aggregate), banks, stock markets and growth.where y is the log of real per capita GDP, X represents the set of explanatory variables including the indicators of bank, stockmarkeant 0.791* 0.501* 0.188 3.527*(0.47) (0.28) (0.27) (1.8)

income per capita 0.057*** 0.055*** 0.011*** 0.065***(0.052) (0.028) (0.042) (0.028)

nment consumption 0.235(0.208)

openness 0.0474(0.093)

ion rate 0.0142(0.027)

osite risk 0.956**(0.45)

market (overall) 0.0064 0.0054 0.0204 0.0065(0.016) (0.012) (0.017) (0.01)

credit 0.0064 0.0468 0.0012 0.0014(0.027) (0.046) (0.029) (0.035)

ver ratio 0.0663** 0.0666** 0.0067 0.0154(0.032) (0.027) (0.04) (0.024)

n test, p-value 0.138 0.047 0.23 0.351correlation test, p-value 0.141 0.313 0.184 0.196er of instruments 26 26 26 26ries 37 37 37 37vations 201 201 198 201

bles are entered as logarithm. Exception is inflation which is transformed as log(1+variable). Standard errors are in parentheses.ressions also include dummy variables for the different time periods that are not reported.indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively.ll hypothesis is that the instruments used are not correlated with the residuals.ull hypothesis is that the errors in the first-difference regression exhibit no second-order serial correlation.

Table 5Corporate bond markets, banks, stock markets and growth.

(1) (2) (3) (4)

All variables are entered as logarithm. Exception is inflation which is transformed as log(1+variable). Standard errors are in parentheses.The regressions also include dummy variables for the different time periods that are not reported.

536 P. Thumrongvit et al. / International Review of Economics and Finance 27 (2013) 529541As we pointed out earlier, the General Method of Moments (GMM) estimation used by Beck and Levine (2004) has a structuraldownward bias of standard errors. To mitigate this bias, Beck and Levine combine one-step and two-step system GMM results.We employ, in this paper, a small-sample correction for the two-step standard errors developed by Windmeijer (2005). In

*, **, *** indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively.a The null hypothesis is that the instruments used are not correlated with the residuals.b The null hypothesis is that the errors in the first-difference regression exhibit no second-order serial correlation.Constant 0.277 0.270 0.291 3.093(0.32) (0.28) (0.2) (2.2)

Initial income per capita 0.013*** 0.027*** 0.0025*** 0.068***(0.041) (0.033) (0.026) (0.03)



Inflation rate 0.0221(0.029)

Composite risk 0.871(0.549)

Bond market (overall) 0.0158 0.0128 0.0074 0.0024(0.014) (0.014) (0.01) (0.014)

Bank credit 0.032 0.0587 0.0337 0.0133(0.021) (0.037) (0.03) (0.036)

Turnover ratio 0.0572** 0.0628*** 0.0344 0.0124(0.024) (0.023) (0.03) (0.026)

Hansen testa p-value 0.201 0.089 0.116 0.202Serial correlation testb p-value 0.125 0.289 0.227 0.240Number of instruments 26 26 26 26Countries 38 38 38 38Observations 222 222 219 222

Notessimulation, he finds that the two-step GMM performs better than the one-step GMM in estimating coefficients with lower biasand smaller standard errors. Detailed discussions on the two-step GMM and the system estimator we employed are in theAppendix II.

4.2. Estimation results

Tables 47 report the results of the GMM estimation for overall bonds, corporate bonds, and government bonds, respectively.We use non-overlapping 3-year averaged data to focus on medium- to long-term growth effects.13

In Tables 46, among the control variables; government consumption, price instability (inflation) and overall risk of countryhave negative impacts on economic growth, while the impact of trade openness on economic growth is positive.14 The effect ofinflation varies when control specifications change as shown on Tables 57. In our model, volatility in the core macroeconomicindicator shows a negative effect on economic growth. This finding contrasts with other studies findings that macroeconomicvolatility is not significant. (Braun & Briones, 2005; Fink et al., 2006). Government consumption does not contribute to economicgrowth in any set of variables in our panel-data analysis. These results are consistent with Barro and Sala-i-Martin (1995) andother growth literature.

All control variables except government consumption have expected signs and estimates are statistically significant. All regressionspass the Hansen test and Arellano and Bover's (1995) test for serial correlations of error terms. This shows that our estimator isconsistent in the sense that the instrument variables are exogenous and that the error terms do not exhibit serial correlation.

More specifically, Table 4 shows that turnover ratio, the proxy for stock markets, has significantly positive effects on economicgrowth while bank credit is insignificant, regardless of its sign, in all regressions. On the other hand, the effects of bond marketdevelopment on growth are also statistically insignificant. When the value of corporate bonds alone are used, the bond marketcontinues to have insignificant effects on economic growth. However, when the bond market is restricted to government bonds,bonds become significant in all regressions but one as reported in Table 6. This is consistent with the so called crowding-outeffect of government bonds that diminishes the share of corporate bonds in aggregate bond markets (Braun & Briones, 2005).

13 Since we have 21 years of data, 3-year averaging seems to be the best option to make the dimension of the panel suitable for the estimator; 5 or 6-yearaveraging would regenerate too few periods for each sample country.14 The composite risk ranges from 0 (high risk) to 100 (low risk); therefore the effect of this variable should be inversely interpreted.

Table 6Government bond markets, banks, stock markets and growth.

(1) (2) (3) (4)

Constant 0.547* 0.365 0.416* 4.200(0.301) (0.253) (0.222) (2.64)





Composite risk 0.871(0.549)

Bond market (overall) 0.0277*** 0.0193*** 0.0147 0.0105(0.01) (0.0073) (0.01) (0.012)

Bank credit 0.0231 0.0369 0.0265 0.0287(0.026) (0.033) (0.032) (0.03)

Turnover ratio 0.0617*** 0.0469** 0.0289 0.0236(0.02) (0.018) (0.026) (0.02)


NotesAll variables are entered as logarithm. Exception is inflation which is transformed as log (1+variable). Standard errors are in parentheses.The regressions also include dummy variables for the different time periods that are not reported.*, **, *** indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively.

a The null hypothesis is that the instruments used are not correlated with the residuals.b The null hypothesis is that the errors in the first-difference regression exhibit no second-order serial correlation.

Table 7Corporate and government bonds and growth.

(1) (2) (3) (4)

Constant 0.640* 0.488* 0.246 3.540*(0.37) (0.28) (0.18) (1.95)





Composite risk 0.988**(0.49)

Bond market (private) 0.0017 0.0016 0.0196 0.0023(0.016) (0.014) (0.013) (0.0102)

Bonds (government) 0.0184** 0.0156** 0.0071 0.0078(0.00744) (0.00756) (0.00887) (0.00885)

Bank credit 0.0061 0.0273 0.0019 0.0127(0.023) (0.038) (0.028) (0.026)

Turnover ratio 0.0453** 0.0420 0.0106 0.0176(0.022) (0.025) (0.026) (0.024)


NotesAll variables are entered as logarithm. Exception is inflation which is transformed as log (1+variable). Standard errors are in parentheses.The regressions also include dummy variables for the different time periods that are not reported.*, **, *** indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively.

a The null hypothesis is that the instruments used are not correlated with the residuals.b The null hypothesis is that the errors in the first-difference regression exhibit no second-order serial correlation.

537P. Thumrongvit et al. / International Review of Economics and Finance 27 (2013) 529541

The relative roles of government bonds and corporate bonds are reported in Table 7, where both types of bonds are included in

from negative for a study period covering from 1989 to 2003 to positive for this study period covering from 1989 to 2010. This

538 P. Thumrongvit et al. / International Review of Economics and Finance 27 (2013) 529541finding confirms earlier studies: To wit, the expanded bond markets together with advance in economy's financial structurecontribute to economic growth through technological innovations and enhanced labor productivity in the private sector (Fink etal., 2006; Ritter, 2005).

In summary, the literature is still divided on the directional causality of the three major financial markets banking, bond andequity stock that have on economic growth. The division is understandable in view of the fact, as Hassan et al. (2011) aptlyobserves, empirical literature on the subject tries to combine the endogenous growth theory and micro structure of financialsystems, both of which are still undergoing evolutionary changes. Our study contributes to the empirical literature with thefindings (i) that there is a significantly positive relationship between economic growth and the expansion of stock market, and(ii) that the relationship between bank credit and economic growth as well as that between bond markets and economic growthare time-varying and bidirectional: Namely the contributing role of bank credit to economic growth is significantly positiveduring early stages of economic growth with relatively underdeveloped financial structure, but this positive relationship changesto a negative relationship as domestic corporate bond markets develop along with the development of financial structure of thethirty eight sample countries covered in our study.

15 Data and analysis for the years 19892003 are available from the corresponding author.the same regression. The results confirm those reported in Tables 5 and 6 and indicate that it is government bonds, not corporatebonds, which are significantly related to the growth performance of an economy. This finding is not surprising and it is consistentwith the literature on the primary impacts of government's capital expenditures: That is, economic growth largely comes fromenhanced labor productivity and technological innovation in the private sector, but government's capital expenditure such asgovernment spending on communication and transportation networks and national security provides externalities of publicgoods (Fink et al., 2006; Ritter, 2005).

In summary, Tables 57 spring upon findings that have not been reported in the literature to the best of our knowledge. To wit,when bond markets are explicitly incorporated in the system of panel data regressions, both the stock market turnover ratio andoverall bond markets are positive and robust in determining economic growth, while the effect of bank credit on economicgrowth changes from positive to negative in all twelve regressions except one. Recall that most past studies concernedthemselves with the size of financial markets and its relative efficiency in providing funds for economic growth from twoprincipal sources banks and stock markets disregarding the role of bond market financing, that provided necessary long-termfinancing for businesses and governments.

5. Summary and conclusions

In this paper, we have examined the effects of the development of a domestic bond market on economic growth, a subject thathas largely been ignored in the empirical literature. While we have dealt neither with the different types of bonds (e.g., fixed,floating rates or zero-coupon bonds) nor bond derivatives (e.g., bond futures and options) in this study, it stands to reason thatbond markets complement banks and stock markets in deepening the financial market of an economy. For instance, in theabsence of bond markets, banks tend to become overcapitalized and that may lead them to make suboptimal or unsound loans.With well-developed bond markets, banks invest in bonds, and in so doing, reduce information asymmetries which promote theefficient uses of resources. This is the reason the World Bank urged developing countries to accelerate the deepening of theirdomestic bond markets long before the financial crises in Asia, Russia and Latin America in the late 1990's (Dalla et al., 1995).

Consistent with previous studies, our findings shows that both stock markets and banks help promote economic growth, eventhough the statistical relationship between banking growth and economic growth changes from positive to negative as domesticbond markets, especially those of corporate bonds, expand and overall financial structure matures. We find that the simplecorrelation coefficients among the three sectors bank, bond and stock market are mildly negative for bond and bank credit andmildly positive for stock market. We can deduce from this that differences in the composition and institutions of financial marketsdo matter, as the three major sectors banks, stock markets and bond markets do not grow simultaneously, but theycomplement each other at different stages of their respective growth. This finding buttresses the general proposition thatnecessary conditions for bond market development are similar to those that foster the development of a country's banking sectorand that when we introduce bonds into the analysis, debate may well shift from the relative merit of a bank-based andequity-based financial system to a bipolar system comprising of debt, which include bank loans and bonds, and equity financing(Burger & Warnock, 2006).

Our findings on the effects of government and corporate bonds are quite revealing and informative in understanding the rolesof the two markets on economic growth. First, we find that the relationship between the overall bond market and economicgrowth is significantly positive for government bonds while it is insignificantly negative for corporate bonds. It also seems thatthe effect of government bonds is dominant so that the effects of overall bond market are positive although not statisticallysignificant. Note that the significantly positive effect of government bonds arises even when government consumption, which hasnegative effects on economic growth, is included in the specification. This finding is consistent with previous studies ongovernment bonds cited in Section 3.

Second, our study shows that the relationship between corporate bond market development and economic growth changes15

Appendix I. Stock market, bank asset and domestic bond market, 2010 (% of GDP)

Th(b) unsamemome

Thelimin

finite-sample bias. Bond, Hoeffler, and Temple (2001) suggest a way to detect this bias by comparing the difference GMM resultsto alternative estimates of the autoregressive parameter, in Eq. (A1). A consistent estimate of should be between thoseobtained from the OLS and within group estimators.16 Another bias, suggested by Beck and Levine (2004), is due to measurementerrors in which the bias may be exacerbated by differencing.

To correct these biases, we use an estimator developed by Blundell and Bond (1998). This estimator combines the regressionin differences with the regression in levels. The idea for the regression in levels is that it transforms the instruments by

Country Stock market capitalization Total bank assets1 Domestic debt securities

China 81.02 184.49 51.15

539P. Thumrongvit et al. / International Review of Economics and Finance 27 (2013) 52954116 We nd that the estimate of from the OLS is0.00738 and the within group estimator is0.14. Our estimates from the difference GMM are well belowthat from the within group estimator, hence, we decided to drop all the results from the difference GMM.E Xi;ts i;ti;t1 h i

0 for s2; t 3;; T; A3

where the moment condition for predetermined variables is

E Xi;ts i;ti;t1 h i

0 for s1; t 2;; T: A4

The above moment conditions mean that lagged levels of the explanatory variables can be used as instruments for theequation in first differences. Lags 2 and more are appropriate instruments for endogenous variables, while lags 1 and more areappropriate for predetermined variables. Using lagged levels as instruments for first-difference variables eliminates simultaneitybias and gives rise to the difference estimator.

Blundell and Bond (1998) show that the difference estimator performs poorly in the case of persistent explanatory variables,such as output, or when sample size is small. Under these circumstances, the estimator may be subject to a large downwardE yi;ts i;ti;t1 h i

0 for s2; t 3;; T; A2Although there is no country-specific effect in Eq. (A1), the lagged dependent variable is still endogenous, since the yi,t 1 termin (yi,t 1 yi,t 2) is correlated with the i,t 1 variable in (i,t i,t 1). Furthermore, any predetermined variables in X that are notstrictly exogenous may be correlated with i,t 1 as well. Assuming that the error term, i,t, is not serially correlated and theexplanatory variables, X, are weakly exogenous, Arellano and Bond (1991) propose the following moment conditions forendogenous variables,ere are three problems in estimating the growth equation: (a) explanatory variables that are typically endogenous,observed country-specific effects, and (c) the simultaneity issue of whether or not financial development takes place at thetime that economic growth occurs. These problems are corrected by applying the first-differenced generalized method ofnts (GMM) estimators to dynamic panel data models.e first differenced GMM using the methods of Holtz-Eakin, Newey, and Rosen (1998) and Arellano and Bond (1991)ates the country-specific effect to yield:

yi;tyi;t1 yi;t1yi;t2

Xi;tXi;t1

i;ti;t1

A1Notes: 1, excluding central bank assets. 2, data not available.Sources: World Bank (World Development Indicator), IMF (Principal Global Indicator) Bank for International Settlements

(Total Domestic Debt).

Appendix II. GMM estimation of the system estimator

India 93.46 72.10 41.02Korea 107.37 122.13 109.51Singapore 166.18 217.01 61.12U.S. 117.53 158.88 177.07Japan 74.57 314.71 251.59Euro zone nations 51.62 303.87 2

differencing in such a way that they are exogenous to the fixed effects.17 Additional moment conditions for the regression inlevels are

h i

Levine,Levine,Levine,Litan, RLucas, RMa, G., R

540 P. Thumrongvit et al. / International Review of Economics and Finance 27 (2013) 52954117 This is valid under the assumption that, although the levels of the explanatory variables may be correlated with the country-specic effect, there is nocorrelation between the differences of these variables and the country-specic effect. Unlike the regression in differences, where many lagged levels are used asinstruments, the regression in level uses only the most recent difference as an instrument.R. (1997). Financial development and economic growth: Views and agenda. Journal of Economic Literature, 35(2), 688726.R. (2002). Bank-based or market-based financial systems: Which is better? NBER Working Papers 9138.R., & Zervos, S. (1998). Stock markets, banks, and economic growth. American Economic Review, 88, 537558.. E., Pomerleano, M., & Sundararajan, V. (2003). The future of domestic capital markets in developing countries. Washington, D.C.: Brookings Institution.. (1998). On the mechanics of economic development. American Economic Review, 22, 342.emolona, E., & He, J. (2006). Developing corporate bondmarket in Asia: a synopsis of theKunmingdiscussions. Bank for International SettlementsWorking Paper 26.E yi;tsyi;ts1 i;t 0 for s 1; A5

E Xi;tsXi;ts1

i;th i

0 for s 1: A6

In this paper, we use the GMM estimator based on Eqs. (A2)(A6) in order to generate the system estimator.There are crucial assumptions underlying GMM estimator calculations that needed to be checked. In order to ascertain the

assumptions that the GMM estimator is consistent, and instruments are exogenous, and that the error terms do not exhibit serialcorrelation, we apply two specification tests. First, the Hansen test of over-identifying restrictions, where the null hypothesis isthat the instrumental variables are uncorrelated with the residuals. According to Roodman (2006), the Hansen test has aweakness in that the more instruments we use, the weaker the test becomes. Our second specification test is the Arellano andBover's (1995) test, which examines whether the error terms are serially correlated. Since the test is applied to the differencederror terms, the first-order serial correlation is expected. That is, it=it i,t 1 is correlated to i,t 1=i,t 1 i,t 2 due to thecommon i,t 1 term. Therefore, to check for the first-order serial correlation in levels, we should look for second-order correlationin difference (Roodman, 2006). A consistent GMM estimator is the one that cannot reject the null hypotheses under the Hansentest and the Arellano and Bond test.

References

Abbas, S. M. A., & Christensen, J. E. (2007). The role of domestic debt markets in economic growth: An empirical investigation for low-income countries andemerging markets. IMF Working Papers (June, 07/127).

Allen, F., Qian, J., & Qian, M. (2005). Law, finance and economic growth in China. Journal of Financial Economics, 77, 57116.Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and an application to employment fluctuations. The Review of

Economic Studies, 58, 277297.Arellano, M., & Bover, O. (1995). Another look at the instrumental-variable estimation of error-components models. Journal of Econometrics, 68, 2952.Barro, R. J., & Sala-i-Martin, X. (1995). Economic Growth. New York: McGraw Hill.Beck, T., & Levine, R. (2004). Stock markets, banks, and growth: Panel evidence. Journal of Banking and Finance, 28, 423442.Biais, B., Decierck, F., Dow, J., Portes, R., & von Thaddden, E. -L. (2006). European corporate bond markets: Transparency, liquidity and efficiency. Centre for Economic

Policy Research, London School of Business.Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 8, 115143.Bond, S., Hoeffler, A., & Temple, J. (2001). GMM estimation of empirical growth models. Economics Group, Nuffield College, University of Oxford.Booth, J., & Smith, R. (1986). Capital raising, underwriting and certification hypothesis. Journal of Financial Economics, 15, 261281.Braun, M., & Briones, I. (2005). The development of bond markets around the world. Working Paper. Universidad Adolfo Ibanez.Burger, J. D., & Warnock, F. E. (2006). Local currency bond markets. IMF Staff Papers, 53, 133146.Calderon, C., Chong, A., & Loayza, N. (2000). Determinants of current account deficits in developing countries. World Bank Research Policy Working Paper 2398.Dalla, I., Khatdhate, D., Rao, D. C., Kondury, K., Jun, L., & Chuppe, T. (1995). The emerging Asian bond market. World Bank Working Paper No. 14880.de la Torre, A., Gozzi, J. C., & Schumukler, S. L. (2007). Financial development: Maturing and emerging policy issues.World Bank Research Observer, 22(1), 67102.Eichengreen, B., & Luengnaruemitchai, P. (2004). Why doesn't Asia have bigger bond market? NBER Working Paper 10576.Ergungor, O. E. (2008). Financial structure and economic growth: Structure matters. International Review of Economics and Finance, 17, 292305.Favara, G. (2003). An empirical reassessment of the relationship between finance and growth. IMF Working Papers 03/123.Fink, G., Haiss, P., & Hristoforova, S. (2003). Bond markets and economic growth. IEF Working Paper 49.Fink, G., Haiss, P., Kirchner, H., & Moser, U. (2006). Financing through bond issues and the nexus with economic growth. Paper presented at the meeting of European

Financial Management Association, Madrid, Spain.Fry, M. J. (1995). Money, interest and banking in economic development. Baltimore: Johns Hopkins University Press.Goldsmith, R. W. (1969). Financial structure and development. New Haven, CT: Yale University Press.Harvey, C. R. (1989). Time-varying conditional covariances in tests of asset pricing models. Journal of Financial Economics, 24, 289317.Hassan, M. K., Sanchez, B., & Yu, J. S. (2011). Financial development and economic growth: New evidence from panel data. The Quarterly Review of Economics and

Finance, 51, 88104.Hawkins, J. (2002). Bond markets and banks in emerging economies: The development of bond markets in emerging economies. Bank for International Settlements

Working Paper 11.Herring, R. J., & Chtusripitak, N. (2000). The case of the missing markets: The bond market and why it matters for financial development. The Wharton School

Working Paper 01-08. University of Pennsylvania.Holtz-Eakin, D., Newey, W., & Rosen, H. S. (1998). Estimating vector autoregressions with panel data. Econometrica, 56, 13711395.Jalil, A., Feridun, M., & Ma, Y. (2010). Financegrowth nexus in China revisited: New evidence from principal components and ARDL bounds tests. International

Review of Economics and Finance, 19, 189195.King, R. G., & Levine, R. (1993a). Finance and growth: Schumpeter might be right. Quarterly Journal of Economics, 108, 717738.King, R. G., & Levine, R. (1993b). Finance, entrepreneurship, and growth: Theory and evidence. Journal of Monetary Economics, 32, 513542.LaPorta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. (1998). Law and finance. Journal of Political Economy, 106, 11131155.Lee, K., Pesaran, M. H., & Smith, R. (1997). Growth and convergence in a multi-country empirical stochastic Solowmodel. Journal of Applied Econometrics, 12, 357392.

McKinnon, R. I. (1973). Money and capital in economic development. Washington, D. C.: Brooking Institution.Merton, R. C. (1992). Financial innovation and economic performance. Journal of Applied Corporate Finance, 4, 1222.Park, J. -H. (2008). Bond market development: The experience of the Republic of Korea. Seoul: Korea Institute of Finance.Rajan, R., & Zingales, L. (2003). The great reversals: The politics of financial development in the twentieth century. Journal of Financial Economics, 69, 550.Ritter, J. R. (2005). Economic growth and equity returns. Pacific-Basin Finance Journal, 13(5), 489503.Roodman, D. (2006). How to do xtabond2: An introduction to difference and system GMM in stata. Center for Global Development Working Paper 103.Rousseau, P. L., & Wachtel, P. (2000). Equity markets and growth: Cross-country evidence on timing and outcomes, 19801995. Journal of Business and Finance, 24,

19331957.Shyn, Y., & Oh, W. (2008). The effects of financial development and growth opportunities on industrial growth in Korea. Korea Institute of Finance Working Paper.Tadesse, S. (2002). Financial architecture and economic performance: International evidence. Journal of Financial Intermediation, 11(4), 429454.Tudor, A. (2012). Asian companies now prefer bonds. Wall Street Journal, 260, C1 (June 8).Wachtel, P. (2001). Growth and finance: What do we know and how do we know it? International Finance, 4(3), 335362.Windmeijer, F. (2005). A finite sample correction for the variance of linear efficient two-step GMM estimators. Journal of Econometrics, 126, 2551.

541P. Thumrongvit et al. / International Review of Economics and Finance 27 (2013) 529541

Linking the missing market: The effect of bond markets on economic growth1. Introduction2. The bond market3. A brief survey of literature3.1. Data3.2. Key variables defined3.3. Descriptive statistics and preliminary findings3.4. Cross-country ordinary least squares regressions

4. Panel data analysis4.1. Model specifications4.2. Estimation results

5. Summary and conclusionsStock market, bank asset and domestic bond market, 2010 (% of GDP)GMM estimation of the system estimatorReferences

Documents

Kim - The Effect of Bonds Market on Growth