Significant Difference in the Yields of Sukuk Bonds versus Conventional Bonds

Abstract

Bond yields of Treasury and corporate bonds are observed in a listed exchange. This article reports the findings on the market yield behaviour of two types of debt securities in the same exchange, the sharia-compliant sukuk bonds and the normal conventional bonds. There are 17 exchanges where sukuk bonds are traded, and the outstanding value is estimated at US$ 1,200 billion. The average yields of sukuk Treasury bonds are significantly higher (premium) than that of conventional Treasury bonds. On the other hand, investors in the sukuk corporate bonds receive slightly lower returns (discount) of about 25 basis points in the case of long-term sukuk bonds. To the best of our knowledge, this is the first study to verify these differences using appropriate advanced econometric methods. These results have far-reaching implications for the market practices as well as for teaching of bond pricing behaviour since this new form of debt markets is growing at about 17 per cent a year.

JEL Classification: F23, F31, G12

Keywords

Bond yields zero-risk risky security bank deposit yields

1. Introduction

This article documents a puzzle of observed premiums investors earn on sharia-compliant sukuk Treasury securities compared to yields on identical conventional securities traded in the same bond market; also, we observe a discount on AAA-rated sharia-compliant sukuk debt securities relative to identical conventional debt securities. Godlewski, Turk-Ariss and Weill (2013) provide a description of the sukuk debt markets. Haron (1996) and Lim (1998) documented differences in sharia-compliant banks, though they did investigate bond market behaviour. In Malaysia (and 16 other financial centres), there exists identical sukuk debt securities (Treasury bills, Treasury bonds, Agency and AAA corporate bonds) traded in the same exchange along with matching conventional securities. The interesting thing about the two classes of securities is that (a) conventional debt securities and (b) sukuk debt securities are traded side by side in the same markets under same market clearing processes.

No study exists as to whether the yield difference reported (Safari, Ariff, & Mohamad, 2013) is merely due to methodological issues, or risk differences or simply measurement errors. The motivation for this research is to document the yield premium in two different identical Treasury securities (zero-risk) traded in the same markets. We also intend to investigate the significant discount observed on the AAA-rated (risky) corporate debt securities in the same market.

The long-run relationships between (a) the zero-risk sukuk Treasuries versus the zero-risk conventional bonds and (b) the AAA-rated risky conventional corporate bond versus the sharia-compliant AAA-rated sukuk bond have not yet been fully explored. Verification that there is no difference in yields would call into question the very idea of profit-loss-and-risk-sharing principle underlying the sharia regulation binding these contracts. With increased uncertainty in investor’s returns under the profit-and-risk-sharing contracting, there ought to be a higher yield on sharia-compliant sukuk securities for investors. Furthermore, given their pricing is essentially based on the cash flows of two types of similar securities differentiated by profit-risk-sharing, there ought to be a long-run relationship that requires careful testing as well. We apply some advanced econometric methods to investigate these long-run relationships between the two types of bond securities traded in the same market.

These two research questions are practice-relevant requiring careful investigation with one or more relevant econometric methods to ensure that the results in this article are robust so that one could have reliable findings. A sustained difference in the yields of identical conventional instruments and the new sharia-compliant sukuk securities would challenge some of the market practices and rating procedures in this new market as long as the observed results are not simply due to testing errors. Moreover, systematic premiums or discounts in such instruments are a legitimate business research topic because it has implications for financial practices for the tested marketplace and also other similarly organised sharia-compliant securities markets. London has just listed its first sukuk bond in July 2014 with a term of 3 years with a yield to match the guilt instrument: this issue was oversubscribed 11 times.

The rest of the article is organised as follows. The next two sections cover relevant literature and test methodology. The results are presented in Section 4 on zero-risk Treasury securities and Section 5 presents the results on AAA corporate bonds. The article concludes with a summary in Section 6.

2. Bond Yields Studies

2.1 Sharia-based Bond Securities

There are not many studies on this new type of bond securities. Sukuk bond markets, first of which was issued in public market in 1998, have attracted the attention of mainstream journals as well as serious press; Wall Street Journal published an article by Lane on 16 November 2006 and Bloomberg data series covered this bond market as well (Lane, 2006). By 2014, there were 17 locations around the world where these new type of bonds were issued and held to maturity or traded daily along with conventional bonds in the same markets. Among them, the exchange (Bursa) in Malaysia is the largest with the attractive feature that the bonds are designed to be traded continually, whereas in most other markets sukuk issues are mostly held to maturity. About three-quarters of the issued sukuk are traded in Malaysia. The latter practice makes it difficult to study the yield behaviour in most other markets, except in Malaysia. We have also compiled and maintained a data set based on traded prices on both types of bonds for the period 2005–2014, and the same data have been used in this study.

Sukuk securities are issued by governments (T-bills and bonds; agency bonds) and by corporations. According to Bloomberg services, the size of the sukuk issues amount to US$ 1,200 billion as also reported in Safari, Ariff and Mohamad (2014), although the size of the private issues by banks and traded in the OTC markets is uncertain, suggesting that the OTC and the exchange-listed markets could be much larger. The same source suggests that the liquidity of these instruments is very low. Hence, an observed premium could well be a liquidity premium and not a genuine higher yield to investors: this hypothesis is not investigated in this article.

Three special characteristics of sharia-compliant sukuk bonds are worth noting to understand how the pricing may differ from conventional bonds. First is the asset-backing required under which the assets of a sukuk-issuing firm must be transferred to a special purpose registered company, the shares of which are proportionately owned by the lenders till the debt is paid off. That is, part of the assets of a firm is separated for a period of time from the control of the firm. Second, all investors in the sharia-compliant instruments have to share losses and then only share profits in a pre-agreed formula. Third, issuing costs of such instruments are known to be higher than the issuing costs of conventional standardised bonds. This is due to the infant status of this market as well as to the complex rules that underpin contracting terms such that there are several embedded contracts in each sukuk issue. For these three known reasons, one would hypothesise that the sharia-compliant bond yield should perhaps be higher than the yield to investors in the cases of no-risk–no-profit-sharing common bonds.

A systematic premium could well be due to these embedded costs and protection against losses to firms in a sharia-compliant market instrument. On the other hand, a contrary reasoning can also be advanced. The risk- and profit-sharing reduces the long-run costs to the firm perhaps because bond investors are holding part of the assets—hence they are sticky investors—during the tenor of the bond. That aspect in a sense safeguards the investors from quickly selling off the portion of the assets they hold in preference for waiting for the majority of assets performing well in the next period. Investors would be well off by sticking to the firm at times of trouble because they are assured of safe asset-backing of their loaned amount.

Another argument advanced is ‘… sukuk are structured to ensure an equivalent returns to conventional bond, with the difference that the return on the sukuk is generated from an underlying asset, not from the obligation to pay interest’ (Miller, Challoner, & Atta, 2007, p. 24). Thus, the magnitude of the investors’ return is non-determined, and depends on the returns from the assets backing the sukuk issue. Under this view, a premium or discount is possible depending on the returns from the assets that back an issue; hence, investors could expect a premium or a discount. Wilson (2008, p. 172) argues that the yield ought to be the same because financiers in this market make special efforts to render sukuk to be identical to conventional securities so unfamiliar investors can assess the risk of these new investments. This line of reasoning would predict identical yields for sharia-compliant and conventional bonds.

This brief review of key contributions suggests that there are three possibilities. No premium or discount; both premium and discount are possible on the yields of assets backing the contract. There could be premium in some cases because of risk–profit-sharing while there could be discount because of stickiness of investors holding a portion of the issuing firm’s assets.

2.2 Empirical Evidence

The literature on these research issues is growing slowly. One of the latest publications (Godlewski et al., 2013) provides some description of the market. Additionally, it reports that the stock price of sukuk-bond issuing firm suffers a significant negative return around the time of the news release (see p. 753). Similar results have been reported in Alam et al. (2013): this article is based on studying six sukuk versus conventional bond markets. Obviously, this finding would have us believe that existing investors consider sukuk issues as bad news, and therefore, would require higher yields to hold such securities. This is something akin to credit downgrade news when stock prices react negatively (Steiner & Heinke, 2001). Promoters of sharia-compliant debt securities claim that such securities are intrinsically desirable in that these are priced based on entirely different principles derived from sharia rules on risk–profit-sharing principles and asset-backing so that the pricing ought to be different to that of conventional bonds. Thus, the pricing mechanism is arguably different, so the yields ought to be different from the yields on conventional bonds; see Khan and Mirakhor (1994) and Ariff, Iqbal and Mohamad (2012).

Another unpublished thesis indicates a significant yield difference in the cases of zero-risk sector (Treasury bills and bonds) and the risky AAA securities (Safari, 2012). The differences range between 6 basis points to about 56 basis points, the latter in the case of long-dated Treasury bonds. A study by Mansor et al. (2012) suggests that the yield on sharia-compliant mutual funds is similar to the yields on conventional mutual funds. Overall, there seems to be no conclusive evidence of Islamic securities having higher or lower yields compared to yields on the much larger common bond instruments. The arguments in the theoretical literature are open ended as well. However, the results of this article support yield premiums for the zero-risk Treasury bills and bonds, and a yield discount for the corporate debt securities compared with the identical conventional bonds.

3. Methodology

3.1 Data Series

The data used in this study are short- and longer-term-traded yield series from securities belonging to the conventional and sukuk bills, sukuk Treasury bonds and the conventional relative to the AAA-rated sukuk bonds with conventional bills and bonds. The data series are monthly observations from 2005 to 2014 as follows: (a) Treasury bills, (b) Treasury bonds, (c) Government Agency bonds and (d) AAA-rated bonds. The last one provides yields of risky corporate bonds, whereas the others are risk-free bonds. The observations are pooled as follows: bills are grouped as one category up to 1-year maturity (termed short term), Treasury bonds with up to 5 years are grouped as medium term and those above 6–20 years are grouped as long term; there are no 6-year maturity bonds. The AAA-rated bonds are divided into three groups as well. The total number of observations is 5,048 monthly yields computed from trade prices.

This study is done on Malaysian securities and the data series are collected from central bank websites and from the Bond Pricing Agency, which is a private vendor on debt securities traded statistics working with the researchers on this research. We also source DataStream. The yields are computed on a daily basis, and reported in the database: we picked the monthly observations at the end of each month over the test period because of our concern for liquidity. The yields are from traded data at the exchange.

3.2 Tests

The validity of the cointegrating series is determined by investigating the order of integration of the variables, which by definition, should be of the same. That is, the variables in the cointegrating equation should be of the same order; levels or first or greater than first difference must be the same on both sides of the equation. One may assume an equilibrium long-run relationship to exist between variables (sukuk and conventional yields) if the variables of the same order are integrated. There are a number of possibilities on co-integration tests. Engle and Granger (1987) developed an OLS framework to estimate the static version of co-integration model (SOLS). Another approach is to use the Johansen–Juselius procedure, which provides necessary information on the co-integrating property of variables while estimating the long-run relationship.

We use the auto-regressive distributed lag (ARDL) method with a bound test with test statistics on the long-run and the short-run relationships (Pesaran, Shin, & Smith, 2001). Pesaran and Shin (1999) show that ‘The ARDL approach has the additional advantage of yielding consistent estimates of the long-run coefficients that are asymptotically normal irrespective of whether the underlying regressors are I(1) or I(0)’. Under this approach, a number of variables with different orders of integration can also be applied. After we verified the existence of co-integrating relationship, we use the ‘error-correction’ model, which takes care of short-run dynamics of the variables and their adjustment speed towards the long-run relationship. Hence, a speed of adjustment coefficient can thus be estimated using this dynamic model.

Simply testing for the existence of significant difference in yields in sukuk debt securities can be done using the difference in the monthly yields between the two types of securities, and then testing whether the difference is significantly higher (premium) or lower (discount). Once the difference is computed, a test of difference being positive can be done using the means and medians of the difference. Mean and median difference formulae are

t = \frac{M e a n}{\sqrt{\frac{σ^{2}}{n - 1}}}

(1)

t = \frac{M e d i a n}{\sqrt{\frac{σ^{2}}{n - 1}}},

(2)

where the mean and median refer to the computed difference between the Islamic and conventional yields. The standard error shown as the denominator is computed from the differenced series. The t-values are expected to be significant when the differences in the several paired series are tested.

The ARDL approach to co-integration requires the tests to be done using the following equation:

Δ C O N V_{t} = α_{i} + λ_{1} I S L A M I C_{t - 1} + Σ_{i = 1}^{m} β_{1 i} Δ I S L A M I C_{t - i} + e_{t},

(3)

where a_i is a drift component and e_t are random errors. The regression is for the current observations of the conventional (CONV) yields with lagged variables of the Islamic securities yields. One can specify simultaneous modelling of long-run and short-run dynamics in a conditional ARDL–ECM framework. In order to verify the existence of a long-run relationship, we use the critical values proposed by Pesaran et al. (2001) by comparing the calculated F-statistics from the pre-determined lower and upper bound measures. Finding the two series to be co-integrated in the long run would indicate that there is error-correction (ECM) and convergence of the series in the long run, despite the differences being statistically significant. The ECM estimated would indicate the long-run dependence of the two series.

3.3 Variables and Hypotheses

There are five hypotheses developed for testing. These are identified and described in Table 1.

As reviewed in the literature section, there are three possible outcomes: no difference, discount and premium. There is no reason for a significant difference to exist for the yields on zero-risk conventional and sukuk contracts, as per one view. After all, existing tests on mutual funds, stock market returns and takaful contracts show there are no significant differences. This reasoning appears to suggest that if the same functions are fulfilled by two alternative instruments (i.e., intermediation function), then the results must be the same: Wilson’s arguments also suggest this. Hence, there are strong conceptual grounds on which we suspect that the series need to be similar. On the other hand, the three characteristics that differentiate the sukuk would require a higher yield for this class: so a test is for premium. A discount could well exist if sukuk is seen as rendering debt to be safer because the investors are sticky.

The maintained hypotheses as stated in Table 1 are to be tested for different maturities ranging from short term (1 year and less), and medium term (up to 5 years) to long term (over 6 years) in the cases of Treasury securities as examples of zero-risk market yields and AAA corporate bonds. The tests are done on both means and medians in order to avoid the problems associated with right-skewed distributions of yields. This allows us to test the three hypotheses with regard to the puzzle of premium in the Islamic securities markets.

Table 1.

Test Hypotheses Developed for Testing Differences in Deposits and Zero-risk Yields

Hypotheses
H1	Islamic Treasury bill yields (short term) are not different from conventional Treasury bill yields (short term).
H2	There is no difference between Islamic zero-risk sukuk Treasury bond yields and zero-risk conventional Treasury bonds: medium and long terms.
H3	There is no difference between the yields of the Islamic corporate sukuk bonds and the yields of conventional corporate bonds, AAA-rated bonds: short, medium and long terms.
H4	The yields of zero-risk Treasuries of sharia-compliant and conventional securities are not co-integrated in the long run.
H5	The yields of risky AAA corporate sukuk and the AAA corporate conventional bond are not co-integrated.

Source: Authors’ own.

Advanced econometric tests are conducted to provide more accurate results on the same phenomenon by way of co-integration tests. These tests identify if there exists a short- as well as a long-run dependence of the two series, although the significant differences may exist between the sukuk and conventional security yields. The maintained hypotheses are that there should be close relationships between the two series despite there being significant yield premium. The task of exploring why the differences exist will be the topic for a future research. We have mean and median difference tests, co-integration tests and error correction models. The final results are expected to show that there could be similar behaviour of the two series. This would suggest that the two contract modes are meant to fulfil the same functions provided by different contracting modes.

4. Findings

4.1 Descriptive Statistics on Means and Medians

The findings on the three sets of interrelated tests are reported and discussed in this section. Table 2 reports the descriptive statistics of Islamic and Conventional Government, Agency bonds, and Corporation (AAA) rates as in Panels A, B and C, respectively. Accordingly, the series is divided into three different terms to maturity, including short, medium and long term. Advanced test results follow thereafter. The results in Panel A refer to the descriptive statistics on risk-free (zero risk) Treasury securities of both sharia-compliant sukuk and conventional ones grouped as short-, medium- and long-term yields. This time-based classification allows us to have large number of observations to ensure accurate results for three classes. The statistics in Panel B refer to the agency securities grouped into short and medium terms; there are no long-term agency bonds. Panel C presents the data on AAA corporate bonds.

One regular pattern seen in the summary statistics in Table 2 is that the yields of zero-risk Treasuries are increasing with increases in tenure of the bonds. The short-term mean is about 3 per cent, whereas the long-term mean is about 4 per cent; thus, it shows that the market is able to price the duration risk or term risk correctly for both sharia-compliant and conventional bonds. The second observation, a little puzzling, is that the agency bond yield is lower: the medium-term agency bond yields are about 20 basis points lower than the yields in the case of the medium-term Treasuries. This is a peculiarity of this market. One reason could be that the sovereign ratings of Malaysia is higher while the ratings of the agencies (mostly very large corporate owned by the government) are less risky as these are very low beta firms traded in the main exchange for which the governments supports the agency bond issue, although these are very large corporations such as the state telecommunication, power and FELDA corporation. In the cases of sukuk versus conventional agency bonds, there is a difference of 0.02 per cent or 2 basis points.

Table 2.
Descriptive Statistics of the Series Used in the Tests

Panel A: Government Bonds Zero-risk Securities

Mean Median Max Min Std. Dev. Skewness Kurtosis JB Obs.

Short-term Conv. 2.95 2.93 3.98 1.82 0.49 –0.68 3.09 24.67 318

Short-term Islamic 2.97 2.93 4.08 1.82 0.50 –0.54 3.01 15.65 318

Medium-term Conv. 3.35 3.34 4.58 2.20 0.40 –0.02 3.76 7.61 318

Medium-term Islamic 3.42 3.40 4.65 2.30 0.39 0.03 3.68 6.26 318

Long-term Conv. 4.06 4.01 5.18 2.91 0.43 0.35 2.71 10.33 424

Long-term Islamic 4.11 4.07 5.18 3.00 0.41 0.29 2.55 9.67 424

Panel B: Agency Bonds Government-backed Securities

Mean
Median
Max
Min
Std. Dev.
Skewness
Kurtosis
JB
Obs.

Short-term Conv. 2.92 2.92 3.83 1.82 0.49 –0.73 3.06 18.87 212

Short-term Islamic 2.93 2.92 3.89 1.82 0.51 –0.59 2.98 12.29 212

Medium-term Conv. 3.04 3.05 3.98 1.92 0.44 –0.58 3.41 11.96 192

Medium-term Islamic 3.07 3.07 4.08 1.97 0.45 –0.38 3.31 5.44 192

Panel C: AAA Risky Bonds Issued by the Same Firms with the Same Ratings and Term

Mean
Median
Max
Min
Std. Dev.
Skewness
Kurtosis
JB
Obs.

Short-term Conv. 3.46 3.44 4.69 2.28 0.40 –0.13 3.91 11.82 318

Short-term Islamic 3.42 3.40 4.65 2.24 0.39 –0.16 3.94 13.21 318

Medium-term Conv. 4.07 3.97 5.44 3.32 0.41 1.04 3.89 67.46 318

Medium-term Islamic 4.01 3.93 5.40 3.30 0.38 0.91 3.72 50.58 318

Long-term Conv. 5.08 4.99 6.73 3.92 0.59 0.37 2.52 13.64 424

Long-term Islamic 5.08 4.97 6.69 3.88 0.65 0.42 2.34 19.95 424

Panel A: Government Bonds Zero-risk Securities
Short-term Conv.	2.95	2.93	3.98	1.82	0.49	–0.68	3.09	24.67	318
Short-term Islamic	2.97	2.93	4.08	1.82	0.50	–0.54	3.01	15.65	318
Medium-term Conv.	3.35	3.34	4.58	2.20	0.40	–0.02	3.76	7.61	318
Medium-term Islamic	3.42	3.40	4.65	2.30	0.39	0.03	3.68	6.26	318
Long-term Conv.	4.06	4.01	5.18	2.91	0.43	0.35	2.71	10.33	424
Long-term Islamic	4.11	4.07	5.18	3.00	0.41	0.29	2.55	9.67	424
Panel B: Agency Bonds Government-backed Securities
	Mean	Median	Max	Min	Std. Dev.	Skewness	Kurtosis	JB	Obs.
Short-term Conv.	2.92	2.92	3.83	1.82	0.49	–0.73	3.06	18.87	212
Short-term Islamic	2.93	2.92	3.89	1.82	0.51	–0.59	2.98	12.29	212
Medium-term Conv.	3.04	3.05	3.98	1.92	0.44	–0.58	3.41	11.96	192
Medium-term Islamic	3.07	3.07	4.08	1.97	0.45	–0.38	3.31	5.44	192
Panel C: AAA Risky Bonds Issued by the Same Firms with the Same Ratings and Term
	Mean	Median	Max	Min	Std. Dev.	Skewness	Kurtosis	JB	Obs.
Short-term Conv.	3.46	3.44	4.69	2.28	0.40	–0.13	3.91	11.82	318
Short-term Islamic	3.42	3.40	4.65	2.24	0.39	–0.16	3.94	13.21	318
Medium-term Conv.	4.07	3.97	5.44	3.32	0.41	1.04	3.89	67.46	318
Medium-term Islamic	4.01	3.93	5.40	3.30	0.38	0.91	3.72	50.58	318
Long-term Conv.	5.08	4.99	6.73	3.92	0.59	0.37	2.52	13.64	424
Long-term Islamic	5.08	4.97	6.69	3.88	0.65	0.42	2.34	19.95	424

Source: Authors’ own.

Notes: Islamic denotes sharia-compliant issues in the same market. Some series are slightly skewed, and the normality tests fail to support normal distribution. Our tests are also done on medians for robustness checks.

There are slightly higher yields for almost all the zero-risk sukuk securities. The AAA-rated corporate bond yields (Islamic and conventional) are much higher than the yields on respective zero-risk securities. The difference between the sukuk and conventional AAA-rated corporate bonds is –0.04 per cent for the short term; –0.06 for the medium term with a –2 basis point for the long term. Hence, the AAA-rated sukuk bonds have lower yield (discount).

In the cases of AAA securities, we find opposite results. Hence, the descriptive statistics in this table provide initial indications on the existence of statistically significant differences between the two types of securities.

Figure 1 is a plot of the series over the test period of 10 years. We plot some selected series to avoid the article being too long. The plots show that the zero-risk sukuk and the risky sukuk yields appear to dominate the conventional yields across all maturities, short-, medium- and long-term bonds. There is a systematic higher yield for sharia-compliant sukuk securities, which will be tested to see if these are statistically significant.

Figure 1.

Source: Authors’ own.

Figure 2 provides the plots for the yields of AAA bonds for identical term and risk and issuers: the term of the sharia-compliant issues do not have as many issues as the conventional issues. Hence, there are breaks in the series. It is mildly evident that there is a slightly higher yield, which may or may not be much different across the time. However, in the case of medium-term sukuk bonds, the difference appears to be much larger than that in the case for the other maturities. Only the tests to be reported in the ensuing part could show if these differences are premiums or discounts.

Figure 2.

Source: Authors’ own.

4.2 Tests on Sukuk Premium/Discount

Table 3 is a summary of the findings on the important issue of whether the observed premiums on the sharia-compliant sukuk securities are systematically yielding a higher/lower return to the investors on Islamic securities. This table reports the test statistics for the mean and median of conventional and Islamic government securities. The number of observations in each test is about 400 monthly yields. These tests are two-sided tests done on means and medians of the differences in the paired samples of conventional and sukuk securities. The maintained hypothesis is that the means and the medians are significantly different from zero: all are positive differences. The number of observations is close to 400 monthly yields in the tested data set observed at the end of trading months of securities that had at least one trade in the month. Most of them had several trades, and some had continuous traded volumes.

The t-tests on the medians and on the means all show there are significant differences in the yields of the paired securities. The first three tests (short, medium and long terms) show that there is a significant difference between the yields of the pairs of zero-risk Treasuries.

Table 3.

Mean and Median Difference Test Statistics on Islamic and Conventional Yields

Type	Mean	Median
Government, Agency and Corporate Securities	T–Statistics	T–Statistics
Short-term government bonds	–4.423***	0
Medium-term government bonds	–13.45***	–12.28***
Long-term government bonds	–14.48***	–11.86***
Short-term agency bonds	–2.982***	0
Medium-term agency bonds	–5.65***	0
Short-term AAA bonds	12.86***	12.40***
Medium-term AAA bonds	11.75***	7.798***
Long-term AAA bonds	–1.088	4.922***

Source: Authors’ own.

Notes: The t-values are significant at or higher than 0.001 probability (***). Where the entry is “0” in the case of the median tests, the two series had the same medians, so no test is needed.

The t-values exceeding 4.423 indicate that the conventional yields are smaller than that of the sukuk security yields for all three maturities, short, medium and long terms. This level of significance is acceptable at or above 0.01, which suggests a strong premium for the sukuk security in the zero-risk Treasury market prices. Next, the agency bonds also have significant lower yields in the case of sukuk issues: note that there are no long-term bonds in this group. Hence, in the cases of zero-risk Treasuries, the ‘government’ issued sukuk bonds have systematic higher yields across all maturities.

Third, the yields in the short- and medium-term AAA corporate markets are lower for the sukuk bonds. The yield is higher for the long-term sukuk bonds (7 year and higher). It is only in the cases of bonds with less than 6-year maturities where we find no premium. This is an anomaly compared with all other trends showing an Islamic security as having a premium, which is significantly higher than zero.

4.3 The Long-run Dependence of Both Yield Series

The results on the ARDL tests are presented in this part on short- and long-run relationships. These tests aim to identify, despite the significant premium existing for the sukuk securities, whether there is a long-run relationship between the two pricing series, one for the sukuk debt instruments based on no pre-negotiated yields and the conventional with pre-negotiated returns.

These test results help to answer a central pricing issue in the bond market. Measured as monthly yields of the sharia-compliant and the conventional debt instruments, there is sufficient evidence of a higher yield which may be referred to as the sukuk Treasury debt security premium. The two types of debt instruments are fundamentally the same, except that one is priced using pre-agreed interest payments, and the other is based on post-profit distribution of a return, presumably dependent on risk and maturity differences. Hence, the two series, at least in the long run, have to be co-integrated. The ARDL test results reveal whether the relationship exists between the two series. In Panel A, the tests show that the two series are co-integrated. That is, there is a long-run relationship.

Table 4 provides the ARDL tests of the medium- to longer-term government and also the AAA-corporate bonds. These tests show that there is a long-term relationship between (a) conventional bond and (b) Islamic bond with medium maturity. The F-statistics are all significant as shown in the first four rows of the table. The tests also identify a lag of 2 for the long-run equilibrium. Table 5 shows the results on some of the tests on the assumptions of the model. Table 5 verifies that the assumptions are holding for the ARDL approach to be applied to the series.

Table 4.

ARDL Results for Islamic versus Conventional Bonds (Medium and Long Term) Bound Test to the ARDL Approach

Conv. Bond = f (Islamic Bond)	F-statistic (MTGOV) k=1, n=318	19.50***
		F-statistic (LTGOV) k=1, n=424	30.70***
		F-statistic (MTAAA) k=1, n=318	24.79***
		F-statistic (LTAAA) k =1, n=424	14.56***
Pesaran et al. (2001) Critical ValueLag = 2	Lower Bound I (0)	Upper Bound I (1)
99% level	4.81	6.02
95% level	3.15	4.11
90% level	2.44	3.28
Lagrange multiplier statistic (MTGOV)Likelihood ratio statistic	35.521*37.688*
Lagrange multiplier statistic (LTGOV)Likelihood ratio statistic	54.494*58.358*
Lagrange multiplier statistic (MTAAA)Likelihood ratio statistic	43.803*47.164*
Lagrange multiplier statistic (LTAAA)Likelihood ratio statistic	27.744*28.701*

Source: Authors’ own.

Table 5.

Diagnostic Tests for the Government and AAA Bond Yields

Panel A: Medium-term Government Bond Yields
Test	LM Version	P-value	F-version	P-value
Serial correlation	CHSQ(1) = 2.677	0.102	(1, 308) = 2.640	0.105
Functionality	CHSQ(1) = 0.389	0.533	(1, 308) = 0.381	0.537
Normality	CHSQ(2) = 123.89	0.000***
Heteroskedasticity	CHSQ(1) = 0.090	0.764	(1, 313) = 0.089	0.765
Panel B: Long-term Government Bond Yields
Test	LM Version	P-value	F-version	P-value
Serial correlation	CHSQ(1) = 5.224	0.022**	(1, 416) = 5.227	0.023**
Functionality	CHSQ(1) = 0.096	0.756	(1, 416) = 0.095	0.758
Normality	CHSQ(2) = 52.38	0.000***
Heteroskedasticity	CHSQ(1) = 13.18	0.000***	(1, 419) = 13.54	0.000***
Panel C: Medium-term AAA Bond Yields
Test	LM Version	P-value	F-version	P-value
Serial correlation	CHSQ(1) = 0.069	0.792	(1, 308) = 0.068	0.794
Functionality	CHSQ(1) = 0.149	0.699	(1, 308) = 0.146	0.702
Normality	CHSQ(2) = 2326.4	0.000***
Heteroskedasticity	CHSQ(1) = 20.04	0.000***	(1, 313) = 21.27	0.000***
Panel D: Long-term AAA Bond Yields
Test	LM Version	P-value	F-version	P-value
Serial correlation	CHSQ(1) = 0.730	0.393	(1, 415) = 0.721	0.396
Functionality	CHSQ(1) = 3.270	0.071*	(1, 415) = 3.248	0.072*
Normality	CHSQ(2) = 4109.3	0.000***
Heteroskedasticity	CHSQ(1) = 24.49	0.000***	(1, 419) = 25.88	0.000***

Source: Authors’ own.

The statistics in Table 6 provides the test statistics on the research question. The t-values of the coefficients show (see the highly significant t-values) that the two series are co-integrated in the long run). The medium-term AAA coefficient is close to 1.00 (0.999), while the long-term coefficient is slightly lower at 0.85, nonetheless close to 1.

Table 6.

Estimated Long-run Coefficients Using the ARDL Approach

(1)	Conv.–1.00	Islamic0.992(32.86)***	Intercept–0.042(0.411)	(2)	Conv.–1.00	Islamic1.026(54.50)***	Intercept–0.163(–2.10)**
(AAA) Medium term				(AAA) Long term
(3)	Conv.–1.00	Islamic0.999(32.58)***	Intercept0.059(0.478)	(4)	Conv.–1.00	Islamic0.847(16.34)***	Intercept0.769(2.89)***

Source: Authors’ own.

Table 7 provides the error correction test results. The first set of results—see (1)—are for zero-risk Treasury securities as to whether there exists a long-run relationship in those pairs. The results show that the ECM term is negative and significant at 1 per cent. This thereby satisfies the condition for the presence of long-run relationship and the mean-reverting behaviour of the Islamic government and AAA bonds to the equilibrium level with those of conventional framework.

Table 7.

Error Correction Representation for the Selected ARDL Models

Gov. (Medium term)ARDL (2,2)					Gov. (Long term)ARDL (1,1)
(1)	Conv.	Islamic0.903(29.58)***	Intercept–0.015(0.40)	ECM (–1)–0.367(–6.7)***	(2)	Conv.	Islamic1.010(63.7)***	Intercept–0.065(–2.0)***	ECM (–1)–0.40(–10.2)***
	R-squared = 0.784					R-squared = 0.913
AAA (Medium term) ARDL (2,2)					AAA (Long term) ARDL (2,1)
(3)	Conv.	Islamic0.894(38.41)***	Intercept0.143(0.483)	ECM (–1)–0.243(–7.5)***	(4)	Conv.	Islamic0.754(42.5)***	Intercept0.077(2.6)***	ECM (–1)–0.10(–4.6)***
	R-squared = 0.863					R-squared = 0.831

Source: Authors’ own.

6. Conclusion

We started with the aim of providing a comprehensive examination of sukuk security premium that was first reported in an unpublished recent study. This article reports on our research with newer tests, and also econometric refinements to verify the alleged sukuk yield premium for zero-risk securities across three maturity classes, namely the short (bills), medium (1–5 year) and long terms (over 6–20 years), using monthly data over 2005–2014. Further, we investigate the yield differences in the sukuk debt securities for the risky AAA-rated corporate debt issues.

The results would have us argue that there is a systematic higher yield for sukuk across (i) zero-risk Treasury bills market, (ii) zero-risk medium-term Treasury bonds and (iii) zero-risk long-term Treasury bonds, (iv) the agency sukuk bonds have lower yields. The yield differences range from as small as 8 basis points for the 1-month short-term Treasury bills to about 12 basis points for the Treasury bonds. The AAA corporate bonds have a difference lower than the conventional AAA bonds, which is a discount. Interestingly, the longer the maturity, the higher are the yield differences of both types of securities; this shows that the investors in this market place price correctly for the listed instruments for the maturity differences.

For instance, the 7–20 year long-term bonds yield is about 4.10 per cent, while the less than 1-year bill yield is 2.90 per cent. Interestingly, the market is also able to price the risk of corporate bonds. The long-term corporate bond yield is 5.10 per cent and the long-term Treasury yield is 4.10. That suggests that there is 100 basis points higher yield for the risky AAA-rated corporate issues. Evidently, the Malaysian listed debt markets are pricing the securities in line with what the liquidity premium and on-risk premium theories suggest: see Fisher (1930) on interest rates and Sharpe (1964) on CAPM. Further, the advanced econometric tests used in this study suggest that there is a long-run dependence of the Islamic yield series on the conventional yield series as per the Pesaran et al. (2001) tests.

Surprisingly, the sukuk debt yields have statistically significant difference by about 3–25 basis points across the listed debt markets. Why this difference, that is, a sukuk yield premium/discount exists is a puzzle that requires further research to find out what factor(s) are driving the higher/lower yields. This research issue is a continuing challenge given that the results reported in this article confirm a systematic yield difference for Islamic securities. The discount observed in corporate bond market is inconsistent with the finding that the market treats sukuk issues as bad news. Similarly, our evidence would argue against the stand of Wilson (2008) that suggests the yield ought to be similar.

Footnotes

Acknowledgements

We acknowledge with sincere thanks the help of the Bond Pricing Agency Malaysia and also the American University of Sharjah for giving us access to the data series for this research. In addition, we sourced data also from the DataStream at the University Putra Malaysia. This article received useful comments from a discussant at the 16-th MFA Conference in June, 2014; the article was presented at the EBES Conference in Spain in November, 2014. The authors are solely responsible for any errors in the article.

Appendix

Table A1.

Correlation of the Conventional and Islamic Bond Yields, 2005–2014

	STCON	STIS	MTCON	MTIS	LTCON	LTIS	STCON	STIS	MTCON	MTIS	LTCON	LTIS	STCON	STIS	MTCON	MTIS
Panel A: Govt. Bonds
STCON	1.00
STI	1.00	1.00
MTCO	0.87	0.88	1.00
MTIS	0.85	0.87	0.98	1.00
LTCON	0.11	0.14	0.50	0.52	1.00
LTIS	0.10	0.13	0.48	0.52	0.98	1.00
Panel B: AAA Bonds
STCON	0.90	0.91	0.78	0.80	0.15	0.15	1.00
STIS	0.91	0.91	0.79	0.81	0.16	0.15	1.00	1.00
MTCON	0.54	0.57	0.74	0.78	0.67	0.67	0.70	0.71	1.00
MTIS	0.51	0.54	0.68	0.74	0.61	0.62	0.71	0.71	0.96	1.00
LTCON	–0.02	0.01	0.32	0.36	0.80	0.80	0.15	0.15	0.78	0.76	1.00
LTIS	–0.03	0.00	0.31	0.35	0.79	0.79	0.13	0.14	0.75	0.75	0.99	1.00
Panel C: Agency Bonds
STCON	1.00	1.00	0.87	0.85	0.11	0.10	0.90	0.91	0.54	0.51	–0.02	–0.03	1.00
STIS	1.00	1.00	0.88	0.87	0.15	0.13	0.91	0.92	0.58	0.54	0.02	0.00	1.00	1.00
MTCON	0.55	0.55	0.64	0.63	0.22	0.20	0.52	0.54	0.43	0.38	0.09	0.10	0.55	0.55	1.00
MTIS	0.57	0.58	0.67	0.67	0.23	0.22	0.55	0.56	0.45	0.41	0.11	0.12	0.57	0.58	0.99	1.00

Source: Authors’ own.

Notes: STCON: short-term conventional bond, STIS: short-term Islamic bond, MTCON: medium-term conventional bond, LTCON: long-term conventional bond, LTIS: long-term Islamic bond.

References

Alam

Hassan

M. K.

& Haque

M. A.

(2013). Are Islamic bonds different from conventional bonds? International evidence from capital market tests. Borsa Istanbul Review, 13(3), 22–29.

Ariff

Iqbal

& Mohamad

(2012). Sukuk Islamic debt securities: Theory, practice and issues. USA and UK: Edward Elgar Publishing.

Engle

R. F.

& Granger

C. W. J.

(1987). Co-integration and error correction: Representation, estimation, and testing. Econometrica, 55(2), 251–276.

Godlewski

C. J.

Turk-Ariss

& Weill

(2013). Sukuk vs. conventional bonds: A stock market perspective. Journal of Comparative Economics, 41(3), 745–761.

Haron

(1996). The effects of management policy on the performance of Islamic banks. Asia Pacific Journal of Management, 13(2), 63–76.

Khan

M. S.

& Mirakhor

(1994). Monetary management in an Islamic economy. Islamic Economics, 6(1), 3–22.

Lane

B. K.

(2006). HD Islamic-bond market becomes global by attracting non-Muslim borrowers.

Lim

K. G.

(1998). Book review: Islamic banking system: Concepts and applications. Asia Pacific Journal of Management, 15(1), 104–105.

Miller

N. D.

Challoner

& Atta

(2007). UK welcomes the Sukuk-How the UK finance bill should stimulate Islamic finance in London, much to the delight of the city’s banks. International Financial Law Review, 26(1), 12.

10.

Mansor

Bhatti

& Ariff

(2015). New evidence on the impact of fees on mutual fund performance of two types of funds. Journal of International Financial Markets, Institutions and Money, 35(1), 102–115.

11.

Pesaran

& Shin

(1999). An autoregressive distributed lag modelling approach to cointegration analysis. In Strom

(Ed.), Econometrics and economic theory in the 20th century: The Ragnar Frisch centennial symposium. Cambridge: Cambridge University Press.

12.

Pesaran

M. H.

Shin

& Smith

R. J.

(2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics, 16(3), 289–326.

13.

Safari

(2012). Classification, yield to maturity determinants, and valuation of Islamic debt securities (sukuk). An unpublished PhD thesis at the Faculty of Economics and Management, University Putra Malaysia.

14.

Safari

Ariff

& Mohamad

(2013). Do debt markets price sukuk and conventional bonds differently? Journal of King Abdulaziz University Islamic Economies, 26(2), 113–149.

15.

Safari

Ariff

& Mohamad

(2014). Sukuk securities: New way of debt contracting. Princeton, NJ: John Wiley & Sons.

16.

Steiner

& Heinke

V. G.

(2001). Event study concerning international bond price effects of credit rating actions. International Journal of Finance & Economics, 6(2), 139–157.

17.

Wilson

(2008). Innovation in the structuring of Islamic sukuk securities. Humanomics, 24(3), 170–181.