The Impact of Islamization on Income Inequality and Economic Growth Nexus in Malaysia

The Impact of Islamization on Income Inequality and Economic Growth Nexus in Malaysia The Impact of Islamization on Income Inequality and Economic Growth Nexus in Malaysia Wan Ahmad Wan Omar 1, Hafizah Abdul Rahim 2 and Fauzi Hussin 3 1 School of Business Innovation and Technopreneurship, Universiti Malaysia Perlis, E-mail: wanahmad@unimap.edu.my 2 Department of Shari ah and Economics, Universiti Malaya, E-mail: arhafizah@gmail.com 3 School of Education and Modern Languages, Universiti Utara Malaysia, E-mail: fauzi@uum.edu.my Abstract: This paper is to study the effect of Islamization index, as a proxy of Islam on the relationship between income inequality and economic growth in Malaysia based on autoregressive distributed lag (ARDL) modelling approach. Gini coefficient index is employed as an indicator for income inequality and real per capita GDP as an indicator for economic growth. The hypothesis is whether Islamization has any impact for the income inequality and economic growth in Malaysia. The result from the ARDL analysis reveals that there is a significant level relationship for Islam in the income inequality and economic growth nexus. Islam has indeed provided a significant impact on the improvement of income inequality for the ummah through zakat institution and baitul maal, and economic growth through Islamic capital as transmission channels. Keywords: Islamization, Income Inequality, Economic Growth, Autoregressive Distributed Lag (ARDL), Malaysia. JEL Classification: A130, C320, O470, O530, Z120. INTRODUCTION The studies on the relationship between income inequality and economic growth have been very active since the publication of seminal work of Simon Kuznets (1955), who suggests that in the existence of income inequality, economic growth movement follows an inverted U-shape curve. In other words, in the early stage of economic development, income equality in a country is inevitably higher to provide a motivating factor for the rich to invest in order to spur economic growth.there is inconclusive debate on the effect of income inequality on economic growth, as some studies suggest a positive effect (see Barro (2000), Li and Zou (2002), and others), while other researchers reveal negative effect in their study results (see Alesina and Rodrik(1991), Stiglitz (2012), and others). From religious perspective, there is a role of religion to gentle down the effect of income inequality on the wellbeing of the population through church 61 International Journal of Applied Business and Economic Research

Wan Ahmad Wan Omar, Hafizah Abdul Rahim and Fauzi Hussin and charity organizations. Religion provides social capital in the economy, which is one of the factors for economic development (Fukuyama 2001). In Islam, there is a specific institution to deal with the income inequality and the poor. Islam recognizes the existence of income equality among the population and provides solution in zakat, sadaqahand waqf mechanisms to resolve it, through bait-ul maal institution. However, there is a scanty research publication on the effect of Islam on the income inequality and economic growth. This paper is to provide an empirical evidence and discussion on the impact of Islam on income inequality and economic growth, in which Malaysia is taken as a case study. Islamization refers to a process of inculcating Islamic values and policies in the economic development process, which could be measured through indexation over time. From Islamic perspective, the process of indexation in the economic development is to indicate the achievement of the Muslims (Ahmad Sarji 2005) and the degree of adherence to the doctrines and teachings of Islam (Rehman and Askari 2010).The process of economic development in Islamic perspective is based on maqasid shari ah (M. Chapra 1993) with inclusion of moral and spiritual values in addition to existing factors of economic development in the realization of a developed state. Time series index with deterministic and stochastic trend components have been known as a plausible approach in empirical research to study a relation between economic variables (see Nelson and Plosser (1982), Franzini and Harvey (1983), and Gujarati (2003)). This empirical study employs Islamization index as discussed by Wan Omar, et. al. (2014) with updated data for 2012 and 2013, as the dataset has passed trend analysis and diagnostic tests, which shows deterministic and stochastic trend components to be a plausible Islamization dataset. The empirical studies on panel data on income equality by Barro (2000, 2008), Deininger and Squire (1996), and others have confirmed the existence of Kuznets curve pertaining to the effect of real per capita gross domestic products (GDP) as indicator of economic growth on the income inequality (indicated by Gini coefficient index). Starting from low value, Gini coefficient index tends to increase with the increases of real per capita GDP. However, the effect is eventually flattened out when the level of real per capita GDP has reached a sufficiently high level, and any further increases in economic growth will tend to reduce income inequality. There is no consensus on the effect of income inequality on the economic growth. The study is perhaps motivated by contemporary phenomenon such as globalization, the enormous disparities between countries, or the inconsistencies found with the expected inequality among countries and their development, in an effort to better understand this relationship. The relationship between income inequality and economic growth in Malaysia has been discussed by many researchers (see Anand (1983), Da Vanso and Kusnic (1983), Faaland, et. al. (2003), and others), and the most recent is by Abdul Khalid (2014) who has reported that income inequality and poverty in Malaysia have improved from 1970 up to 1990, but has stubbornly remained the same over the last 20 years period (1990 2010). The income inequality between the rural and urban population has made very little headway despite huge allocation of development fund provided by the Federal Government to the rural areas. Income inequality between the Muslims (or Malay bumiputra in terms of ethnicity) and non-muslims is still high and persistent; and the current income inequality in Malaysia is considered one of the highest in Asia in spite of recording a robust economic growth over the last decade. Abdul Khalid has outlined three major causes of inequalities in Malaysia: historical transmission by the British colonial; labour market impediment which is very much skewed to the private sector; and policy implementation failure on the part of government (Abdul Khalid 2014, 137-159). International Journal of Applied Business and Economic Research 62

The Impact of Islamization on Income Inequality and Economic Growth Nexus in Malaysia The objective of the paper is to provide an empirical evidence on the existence of short-run and long-run of Islam as proxied by Islamization index on income inequality and economic growth relationship. The paper is structured into five sections. Section one provides the introductory background of the study. Related literature review on the subject is discussed in section two. Section three elaborates model specification, research hypotheses, data sources and description. Findings on the research will be revealed in section four, and discussion on the findings and its policy implication in section five. Section six concludes the study. LITERATURE REVIEW The empirical literature pertaining to the effect of income inequality on economic performance since publication of Simon Kuznets (1955) seminal work on the subject has been very active as a result from increase in availability of data on income distribution. Kuznets hypothesized that the effect of income inequality on economic development followed an inverted U-shaped relation which could be interpreted as the evolution of the distribution of income over the economic transition from a rural to an industrial economy whereby income inequality should increase during the early stages of development and decrease later as the nation develops. Early empirical studies tended to support the conjecture that overall income inequality and economic growth are inversely related, see the works of Benabou (1996) for review of these studies. However, current theoretical model of income inequality and economic growth suggests that the impact of income inequality can both facilitate and retard economic growth depending on the economic status of the country. In the analysis of Robert Barro (2000, 2008), income inequality appears to support economic growth only within rich countries, but to slow it down in poorer countries. Therefore, incomeequalizing policies might be justified on growth promotion grounds in poor countries, but for richer countries, active income-redistribution seems to involve a trade-off between benefits of better equality and a reduction in overall economic growth (R. Barro 2000). The debate continues in the empirical literature as to whether the ultimate effect of overall income inequality on economic growth is positive, negative or not significant. Nonetheless, it seems that the conclusion of the studies is very much depending on the application of econometric method and data consideration (Sarah Voitchovsky, 2005). Sarah breaks the inequalities in the income distribution into the top and bottom ends, and employs standard growth model with a set of explanatory variables to control for inequality at the top and bottom ends of the income distribution simultaneously. Her empirical results support the main hypothesis that income inequality at the top and bottom ends of the income distribution have different effects on economic growth top end inequality supports growth while the bottom end inequality retards growth. From religious perspective, religion has a role to lessen the effect of inequality on poverty and economic growth. Josten (2004) develops a model to capture the essence of both human capital and socio-political instability channels based on the notion that social capital is the variable that affects economic growth and income inequality. The basic idea of Josten s thesis is that as income inequality grows, the community s social capital decreases in consequent, thus affects economic growth negatively. Theoretical framework on how Islamization could have an impact on the income distribution has been discussed by Sadeq (1989), Shafi (1979), Mannan (1970), and other Islamic economists. Islam determines the appropriate factor pricing that meets maqasid shari ah, which subsequently determines the functional distribution of income and wealth in an economy. Economic growth can hardly assuage the profanity of 63 International Journal of Applied Business and Economic Research

Wan Ahmad Wan Omar, Hafizah Abdul Rahim and Fauzi Hussin income inequality and poverty if the growth leads to concentration of income or wealth among a few of the population. In the neoclassical economic framework, factor pricing has a profound impact on the income distribution (Pen,1971). Although the Qur an and the Sunnah do not explicitly mention factors of production, but there is an implicit indication about them: shari ah provides factor which is equitable, hence every factor of production gets a just reward of their services and limits the possibilities of gross income inequalities (Sadeq 1989, p.64). In Islam, there is a mechanism on how the income generated per period of time is distributed among the factors of production involved in the production process through the mechanism of factor pricing, and among other by mean of transfer payments by the government and specific institution. The specific institution that deals with the problem of poverty and income inequality in Islam is baitul-maal which collects their revenues from zakat, sadaqah, awqaf and other contributions from the Muslims. Jehle (1994), Habib (2004) and others have provided empirical evidence that zakat and other contributions in Islam improves income inequality among the Muslim ummah. Wan Omar, et al. (2015) suggests that Islamization has a significant influence on the economic development in Malaysia, based on cointegration analyzes using ARDL approach. His study provides empirical evidence based theoretical framework for relationship between Islam and economic development, specifically reference to Malaysia as discussed by Aslam Haneef (2001), Pramanik (2002), Wilson (1998), and other scholars in Islamic economics. Basically, an index is a statistical method used to measure magnitude of an indicator s change over time or place. The purpose of index is to make performance comparison of indicators between two or more periods easier. Islamization index is proposed to measure developmental progress of Islamization process from spiritual or faith, ummah resource or human capital for Muslims and economic aspect of Islam in terms of capital accumulation. There is a distinct different between religiosity index and Islamization index in the current literature. The general religiosity index for each country in the world has been published by Gallup International Association, Duke University, Pew Research Center and other organizations, based on sampling values surveys conducted during particular periods in each country. The index has been employed as a proxy for religion in the studies of relationship between religion and economic growth variables (see the works of (Grier 1997), (Blum and Dudley 2001), (Barro and McCleary 2003), and others). Albeit a positive relationship was reported on the effect of religion on economic growth but it lacked significant robustness due to data heterogeneity between the proxy of religion, factors and indicators of economic growth (Barro and McCleary 2003). The problem of robustness may be effectively remedied through the introduction of time series data which passes the trend-stationary process (Granger and Newbold 1974). Islamization index was introduced in a time series dataset as an attempt to measure the development of Islam in Muslim-majority country, and to address the problem of robustness in the empirical analysis of religion of Islam on economic growth variables (Wan Omar, Fauzi Hussin and Asan Ali 2014). The dimension and components of Islamizationindex is illustrated in Table 1 which comprises of three sub-indices faith (iman), education, and capital. Faith (iman) is measured on five basic tenets of Islam shahadah, compulsory prayers (shalat), zakat, fasting in the month of Ramadhan, and performing hajj at least once in a life time. Shahadah and fasting could be measured only by means of values survey, while shalat, zakat and performing hajj are indicated by ratios of Muslim population per mosque in the country, total zakat payers to adult Muslim population, and total hajj depositors in the hajj management fund institution (tabung haji) per Muslim population respectively. The higher the faith sub-index indicates the higher religiosity of Islam among the Muslims in the country. International Journal of Applied Business and Economic Research 64

The Impact of Islamization on Income Inequality and Economic Growth Nexus in Malaysia Table 1 Dimension and Components of Islamization Index Dimension Religion Human Capital Islamic Capital Ratios Population per mosque; Muslim adult literacy rate; Islamic capital to gross domestic Zakat payers per Muslim; School enrolment rate among product (GDP); Hajj depositors per Muslim. Muslim adults. Islamic banking and takaful to total financial services. Index Islamic faith Islamic education Islamic capital Source: Wan Omar, Fauzi Hussin and Asan Ali 2014, p.1301 The second sub-index, Islamic education is to capture the human capital development based on Islamic teachings that will impact on the economic development and growth. Literacy rate, school enrolment and average years of schooling have been extensively used in economic growth empirical research as proxies for education or human capital development (Mankiw, Romer and Weil 1992). While the last subindex, Islamic capital indicator, comprises of ratios on Islamic banking and finance in terms of net assets value to total banking and financial market, and total shari ah compliant capital and fund market to total investment capital in the country. The indicator is a proxy of Islamic capital to indicate the participation of Muslims in the economy through capital accumulation. The result of trend analysis on Islamization index dataset in Malaysia shows that the index has passed stationarity and diagnostic tests as plausible variable to be employed as one of the economic variables to study economic relationship between economics variables as argued by (Nelson and Plosser 1982), (Franzini and Harvey 1983), and others. The implications from the stationary test are that the shocks in the Islamization index have transitory effects; and satisfy a strong law of large numbers (SLLN) and suitably standardized sums of elements of the time series obey a central limit theorem (CLT); which are important to prevent the potential risk of misspecification of the data-generating mechanism and exhibiting spurious correlation (Durlauf and Phillips 1987). Therefore, Islamization index could be employed as a proxy for the religion of Islam to capture the important variables of economic growth factors such as human resource development and capital accumulation, as well as to affect the income inequality. Specifically, an increase in Islamization as a causal factor is posited to lower down the income inequality and subsequently may lead to better standard of living. Model Specification METHODOLOGY AND DATA Consider Cobb-Douglas (C-D) model in log-linear transformation form for the impact of Islamization on income inequality and economic growth nexus as follows; ln G t ln y ln Islam (1) t 0 1 2 t 3 t t where G t is Gini coefficient, t is time trend factor y t is real per capita GDP (as indicator of economic growth), Islam t is Islamization index, is time related coefficient of the exogenous variables, and t is a white noise or error terms. Following Pesaran and Shin (1997), the general form of augmented autoregressive distributed lag, ARDL (p, q 1, q 2,..., q i ) model is 65 International Journal of Applied Business and Economic Research

Wan Ahmad Wan Omar, Hafizah Abdul Rahim and Fauzi Hussin where k (, p) g (, q ) x w (2) t i i iz t t i 1 (, p) = 1 1 2 2... p p (3) i (, q i ) = i0 + i1 +... + iqi q i, i = 1, 2,..., k (4) is a lag operator for dependent variable (endogenous) such that g t = g t 1 and w is a vector of deterministic variables such as intercept term and time trends, or exogenous variables with fixed lag. t is error terms or white noise. The exogenous x it constitutes real per capita GDP and Islamization index in log-linear form. Based on model specification in equation (2), we synthesize the basic estimable model in equation (1) into a conditional vector error correction model, ARDL(p, q 1, q 2, q 3 ) for the Gini coefficient equation for hypotheses testing, and the result is ln G t = c 0 + c 1 t + 1 ln G t 1 + 2 ln y t 1 + 3 ln Islam t 1 l l l lng ln y ln Islam (5) 4i t i 5i t j 6i t k 1t i 1 j 1 k 1 where c 0 is an intercept, c 1t is time trend, is coefficient of exogenous variables, and t is a white noise. Research Hypotheses Vector error correction model (VECM) in equation (5) is the specifications for level relations between each variable in the equation, there are four possible hypotheses to determine the long-run impact of Islamization on the relationship between Islamization index on income inequality and economic growth nexus: 1. Unidirectional level relation of real per capita GDP and Islamization index to Gini coefficient, if the estimated coefficients on the lagged y t and Islam t are statistically different from zero as a group ( 5j 0, 6k 0) or ( 5j = 0, 6k = 0), in equation (5). 2. Conversely, unidirectional level relation from the lagged G t and Islam t to y t is not statistically different from zero ( 5j = 0, 6k = 0), and the set of lagged Gini coefficient in equation (6) is statistically different from zero ( 5j 0, 6k 0). ln y t = c 0 + c 1 t + 1 ln y t 1 + 2 ln G t 1 + 3 ln Islam t 1 l l l ln y ln G ln Islam (6) 4i t i 5i t j 6i t k 1t i 1 j 1 k 1 3. Income inequality improvement promoting relation is suggested when the long-run coefficients in ARDL(p, q 1, q 2, q 3 ) model in equation (5) and (6) are statistically significantly different from zero in all regressions. International Journal of Applied Business and Economic Research 66

The Impact of Islamization on Income Inequality and Economic Growth Nexus in Malaysia 4. No level relation is suggested when the coefficients of Gini coefficient, real per capita GDP and Islamization index are not statistically significant in the ARDL(p, q 1, q 2, q 3 ) model. The error correction model (ECM) for equation (5) to test the impact of Islamization on Gini coefficient and per capita GDP in short-run hypothesis is p q q lng lng ln y ln Islam ecm ; t 1, 2,..., T (7) t 1i t i 2i t j 3k t k t 1 1t i 1 j 1 k 1 Where is intercept, ecm t 1 is error correction variable and is its short-run coefficient. We use bounds testing procedure as discussed by (Pesaran, Shin and Smith 2001) to test the hypotheses on level relationship for long-run and short-run. The main advantage of this testing method and estimation strategy lies in the fact that it can be applied irrespective of whether the regressors are integrated I(0) or I(1), and it does not require pre-testing thus avoid pre-testing problem as in the standard cointegration analysis. The first step in bounds testing procedure, is to estimate the ARDL models using ordinary least square (OLS) method based on selected deterministic properties (with intercept and time trend). Joint significant of the long-run coefficients is determined by F-statistic and its p-value for all variables in the error correction (ECM) form of the underlying ARDL model. If the computed F-statistic is significant and higher than upper bounds critical value (which is tabulated in Appendix B, Pesaran and Pesaran (2009)) then we shall reject null hypothesis of no long-run relation between Islamization, income inequality and economic growth. The last three steps in bounds testing procedure is to test for the impact of Islamization in the income inequality improvement promoting and short-run impact hypotheses based on ARDL model in equations (5) to (7). Data Sources and Description We utilize the Islamization index and its sub-indices-islamic faith (iman), Islamic education (ilmu) and Islamic capital in (Wan Omar, Fauzi Hussin and Asan Ali 2014) dataset from 1969 to 2013 (base year, 1969 = 100). Details on datasets for Gini coefficient and real per capita GDP are presented in Table 2. The data for Gini coefficient is sourced from Malaysia Economic Planning Unit website and Statistic Department. However, the data is not continuous on year to year basis, hence some extrapolation and interpolation on the available data are conducted for the missing years. Table 2 Sources of Data Collection Description Variables Sources of Data Indicator of Income Inequality Gini coefficient Malaysia Economic Planning Unit (EPU) and Statistics Department. Indicator of Economic Growth Real per capita GDP in US$ Summer, Heston and Aten (Penn World (in 2005 price) Table 7.1) Indicator of Islam Islamization Index, Faith, Education Wan Omar, et. al. (2014) and Capital Sub-indices 67 International Journal of Applied Business and Economic Research

Data Preliminaries Wan Ahmad Wan Omar, Hafizah Abdul Rahim and Fauzi Hussin RESEARCH FINDINGS The plot of variables of indicator for income inequality (Gini coefficient) and indicator for economic growth (real per capita GDP in US$ at 2005 price) in log-linear form is depicted in Figure 1. Distribution of Gini coefficient time series dataset over the last 44 years was skewed to the lower values indicating that there is some improvement on the income inequality in Malaysia. The highest recorded value was in 1976, an adverse effect from rapid growth as a result from new economic policy implementation and the lowest in 2013. The movement of Gini coefficient over the last two decades was erratic despite robust in the economic growth.the economic growth in Malaysia throughout the sample period (1969-2013) has experienced a steady growth, in spite of several economic shocks occurred in the 1976, 1987, 1998 and recently in 2008. It is clear from Figure 1 and 2, that Gini coefficient is generally decreasing with the rising trend of real per capita GDP and Islamization index. The trend of all variables in the model as depicted in Figure 1 and 2 show the properties of time trend and intercept in its time series data. Hence, these suggest at least initially that time trend and intercept properties need to be included in the ARDL analysis for unit root and bounds tests. Prior to implementation of Islamic financial system and other Islamic based policies by the Government in the end of 1980s, Islamization index time series was basically flat and very thin. The index experienced a rapid growth trend in the 1990s onwards, mainly contributing by the exponential growth of the Islamic capital in Malaysia. This indicates that Islamization process in Malaysia in the sample period was rather a slow process for the first 25 year period, and largely driven by the rapid development of Islamic capital market in the last 20 year period (Figure 2). The Islamic faith (iman) sub-index was very thin in the first two decades until 1997 as indicated by the low number of depositors in pilgrimage management fund (Tabung Haji) and low collection of zakat payment with the bait-ul maal despite Muslim population is more than 60% of the total population in Malaysia. Thereafter, the sub-index is erratic in nature and strongly correlated with the economic condition of Muslims who are Tabung Haji s depositors and zakat payers. In the early decade of sample period (1969-1979), the sub-index of Islamic education was slightly on downwards trend Figure 1: The Plots of Gini Coefficient and Real per capita GDP in Malaysia for the period 1969-2013 International Journal of Applied Business and Economic Research 68

The Impact of Islamization on Income Inequality and Economic Growth Nexus in Malaysia as a result of the transformation of some Islamic religious schools into national-based schools by the Government (see Rosnani Hashim (2004) for discussion on its rationale and process of conversion). After that period, the enrolment rate in the Islamic religious school has increased significantly. However, the sub-index of Islamic education was erratically moving beginning 1980s onwards. In the early period, specifically from 1969 to early 1980s, there was very little movement of Islamic capital sub-index. During these periods, the contributors for Islamic capital to the economy were restricted to deposits in Tabung Haji and collection of zakat payment by bait-ul maal in each state within Malaysia. In 1983 the first Islamic bank was established in Malaysia that marked the beginning of Islamic banking development and others such as Islamic finance, takaful and shari ah compliant fund as reflected by the gradual increases of the sub-index from 1985 onwards. The Islamic capital grew exponentially over the last two decades. Unit Root Tests Figure 2: Trends of Islamization sub-indices The purpose of unit root tests is to check the stationarity status of all variables in the ARDL model and the order of its integration after differencing if the variable is non stationary in the first level test. This is to ensure that all variables pass unit root test and their integration are not I(2) or I(d) stationarity so as to avoid spurious results, and ARDL could not be applied to the model. The lag length is automatically calculated by the system, based on Akaike information criteria (AIC) but pegged to the maximum of 9 lags, the number of lag assumption that is considered sufficient to remove any serial correlation in the residual of the regression model (Gujarati 2003). 69 International Journal of Applied Business and Economic Research

Wan Ahmad Wan Omar, Hafizah Abdul Rahim and Fauzi Hussin The results of unit root tests (t-statistic) for six variables in the ARDL model is presented in Table 3, which produce mixed results, with strong evidence in favor of the unit root hypothesis with 5% or less significance, except for Islamic education variable which is significant without differencing when time trend and intercept properties are adopted into the model. The unit root tests on all the variables, generally produce a mixed result of cointegration I(0) and I(1), hence it has complied with the unit test requirement to proceed for bound testing procedure in ARDL approach(pesaran and Pesaran 2009). Table 3 Unit root test results (t-statistic) using ADF method and AIC selection criteria With trend and intercept With intercept only Variable Level 1 st Diff I(d) Level 1 st Diff 2 nd Diff I(d) Gini Coef. 3.2600* 4.9339*** I(0) 0.6125 4.9472*** I(1) Per capita GDP 2.8488 6.1215*** I(1) 2.5307 5.5958*** I(1) Islamization 0.5135 8.4533*** I(1) 2.9478 3.0573** I(1) Faith (iman) 1.9307 5.0393*** I(1) 1.4918 5.0207*** I(1) Educ. (ilmu) 4.7831*** 6.0475*** I(0) 0.9447 6.0897*** I(1) Capital (siasah) 2.0770 3.3239** I(1) 1.0748 1.3740 4.1263** I(2) Notes on significant test: *** denotes 1% significant, ** denotes 5% significant, and * denotes 10% significant. Bounds Testing and Hypotheses Result The ARDL procedure consists of two stages. The first stage is to investigate the existence of long-run relation between the variables using ordinary least square (OLS) method. The existence of long-run relation is tested by comparing the computed F-statistic with the tabulated critical value in respective properties case for significance test (p-value) of lagged levels of the variables in the error correction form of the underlying ARDL model. The result of F-statistic from OLS method is presented in Table 4 based on AIC selection criteria and maximum lag order of 3 for all possible endogenous in the ARDL models. The outcomes, generally we can reject null hypothesis of no level relation between variables in the ARDL models, except for F Islam (Islam Gini, Y) and F Siasah (C Gini, Y, F, E) ARDL models, in which its long-run coefficient are not significant. The result also suggests that inclusion of intercept and time trend properties in the model will result better significant. The result of F-statistic for testing the joint null hypothesis with intercept and time trend properties in F Gini (Gini Y, Islam) is 4.5288 which is higher than tabulated upper bound (4.458) at 90% confidence level, indicating the existence of long-run level relation between Islamization, income inequality and economic growth. To consider the significance of the lagged level variables in the error correction models explaining ln y t and ln Islam t, we backtrack to edit the regression equation, change the exogenous and consecutively to become endogenous, and follow the same step in calculating F-statistic for joint significance test. The result of F-statistic for economic growth regression in F y (Y Gini, Islam) and F Islam (Islam Gini, Y) is 3.2348 and 0.2488 respectively. Both F-statistics fall well below the lower bound of the critical band (which is 3.484 4.458), and hence the null hypothesis that the level variables do not enter significantly into the equations for ln y t and ln Islam t cannot be rejected. This conclusion holds irrespective of whether the underlying variables are I(0) and I(1). The results suggest that there exists a long-run level relationship International Journal of Applied Business and Economic Research 70

The Impact of Islamization on Income Inequality and Economic Growth Nexus in Malaysia Table 4 The Results of Bounds Testing for the ARDL Model Bounds Test (F-statistic) No. ARDL Model Int. + trend Int. only AIC Lag Outcome 1. F Gini (Gini Y, Islam) 4.5288*** 0.7501 3 Reject H 0 Int + trend 2. F y (Y Gini, Islam) 3.2348** 2.3137* 3 Reject H 0 3. F Islam (Islam Gini Y) 0.2488 1.9837 3 Accept H 0 4. F Gini (Gini Y, F, E, C) 4.1230*** 1.1975 3 Reject H 0 Int + trend 5. FY(Y Gini, F, E, C) 5.8771*** 2.1934* 3 Reject H 0 6. F Iman (F Gini, Y, E, C) 7.2726*** 2.6392** 3 Reject H 0 7. F Ilmu (E Gini, Y, F, C) 2.3969* 3.5591** 3 Reject H 0 8. F Siasah (C Gini, Y, F, E) 0.8341 1.8287 3 Accept H 0 Notes: Y = real per capita GDP. Gini = Gini coefficient. F = Islamic faith (iman). E = Islamic education (ilmu). C = Islamic capital (siasah). Int. = intercept. Critical value at 90% confidence level for Case III (intercept and trend) for degree of freedom (d.f) or k = 3 is given in (Pesaran and Pesaran 2009). Lower bound = 3.484, upper bound = 4.458. For case II (intercept only), for d.f. or k = 3, at 90% confidence level; lower bound = 2.711, upper bound = 3.800. For d.f. or k = 5, case III (intercept and trend); lower bound = 2.782, upper bound = 3.827. For case II (intercept only), lower bound = 2.262, upper bound = 3.367, at 90% confidence level.*** denotes 1% significance. ** denotes 5% significance. * denotes 10% significant. between income inequality, economic growth and Islamization, and variables real per capita GDP and Islamization index can be treated as the long-run forcing variables for explanation of Gini coefficient. There is also level relation between Islamization sub-indices faith, education and capital, on income inequality and economic growth as shown in ARDL model number 4 to 6 in Table 4 as its computed F-statistics exceed critical upper bound value of 3.827 at 90% confidence level. The main purpose of the first stage is to ensure that the long-run relation between variables in the ARDL model to be estimated is not in fact spurious. After we have satisfied with that condition, only then we start for second stage of bounds testing procedure. Income inequality improvement promoting hypothesis for equation (5) is conducted in second stage of the bounds test procedure, by estimating the long-run coefficients between variables in the underlying ARDL model with chosen lag order of 2 and deterministic property (intercept and time trend).the result for long-run coefficient estimatesfor ARDL(1, 2, 2) specification based on AIC selection in Table 5, reveals that Islamization and real per capita GDP have a very significant impact on the income inequality (indicated by Gini coefficient). The computed F-statistic (18.8479) and W-statistic (56.5438) of the model are higher than upper bound critical values (6.538 and 19.6105) at 5% significance. All variables in the ARDL model are very highly significant with small standard error. The underlying ARDL equation also passes all the diagnostic tests that are automatically computed by the system. The estimated coefficients of the long-run relationship on ARDL model show that Islam proxied by Islamization index and real per capita GDP have a very significant impact on income inequality (proxied by Gini coefficient). We obtain the following level relation for long-run Gini coefficient equation from the ARDL(1,2,2) estimation; 71 International Journal of Applied Business and Economic Research

Wan Ahmad Wan Omar, Hafizah Abdul Rahim and Fauzi Hussin Table 5 Estimates of Long-run Coefficients Using ARDL Approach ARDL(1,2,2) selected based on Akaike Information Criterion Dependent variable is ln Gini t (Gini coefficient) 37 observations used for estimation from 1977 to 2013 Regressor Coefficient Standard Error T-Ratio[Prob] ln y t (per capita GDP).29358.045612 6.4365[.000] ln Islam t (Islamization).22413.027780 8.0683[.000] c 0 (intercept).63751.45356 1.4056[.171] c 1 t (time trend).020211.0024706 8.1807[.000] Testing for existence of a level relationship among the variables in the ARDL model F-statistic 95% Lower Bound 95% Upper Bound 90% Lower Bound 90% Upper Bound 18.8479 5.4470 6.5368 4.5544 5.4470 W-statistic 95% Lower Bound 95% Upper Bound 90% Lower Bound 90% Upper Bound 56.5438 16.3410 19.6105 13.6633 16.3410 If the statistic lies between the bounds, the test is inconclusive. If it is above the upper bound, the null hypothesis of no level effect is rejected. If it is below the lower bound, the null hypothesis of no level effect can t be rejected. The critical value bounds are computed by stochastic simulations using 20000 replications. ln Gini t = 0.637 0.020t + 0.294 ln y t + 0.224 ln Islam t + t (8) (.453) (.002) (.045) (.028) where t is the error-correction terms. The standard errors of the long-run estimates are given in the bracket. The results also reveal highly significant estimated coefficients for variable in real per capita GDP (1% significant level) and correct sign of Islamization index coefficient (1% significant level), which contribute to significant impact to the Gini coefficient. An increase of 1% in Islamization index will lead to approximately 0.22% improvement in the Gini coefficient, ceteris paribus. The results of log-linear error correction model of Gini coefficient associated with the above-mentioned long-run relationship are shown in the Table 6. With the exception of ln y t and ln Islam t, all other lagged change coefficients in the ECM are statistically significant, justifying the choice of p = 2 in lag length for ARDL model estimation. Generally, the model goodness of fit is relatively good with R 2 at 0.70914; no serial correlation as DW-statistic (2.5724) is more than 2; and the underlying ARDL equation passed all the diagnostic tests which is automatically computed by the system. Error correction model also portrays the short-run dynamic coefficients associated with the long-run relationship. The error correction coefficients (in Table 6), estimated at 0.72528(0.1032) is statistically highly significant (p-value = 0.000), has the correct sign and suggests a high speed of convergence to equilibrium once the long-run income inequality in equation (8) is shocked. The signs of short-run coefficients dynamic impact are found to be maintained into the long-run relationship, as indicated by all the variables in the model. This result reveals that approximately 72.5% of disequilibria from the previous year s shock converge back to the long-run equilibrium in the current year. International Journal of Applied Business and Economic Research 72

The Impact of Islamization on Income Inequality and Economic Growth Nexus in Malaysia Table 6 Error Correction Representation for the Selected ARDL Model ARDL(1, 2, 2) selected based on Akaike Information Criterion Dependent variable is ln Gini t 37 observations used for estimation from 1977 to 2013 Regressor Coefficient Standard Error T-Ratio[Prob] ln y t.035156.040459.86893[.392] ln y t 1.20046.047572 4.2137[.000] ln Islam t.0090874.034561.26293[.794] ln Islam t 1.18311.037040 4.9436[.000] c 1 t.014659.0022589 6.4895[.000] ecm t 1.72528.10320 7.0277[.000] ecm t = ln Gini t 0.29358 ln y t 0.2243 ln Islam t 0.63751c 0 + 0.020211c 1 t R-Squared.70914 R-Bar-Squared.62603 S.E. of Regression.0087318 F-Stat. F(6,30) 11.3776[.000] Mean of Dependent Variable.0025249 S.D. of Dependent Variable.014279 Residual Sum of Squares.0021348 Equation Log-likelihood 128.0645 Akaike Info. Criterion 119.0645 Schwarz Bayesian Criterion 111.8154 DW-statistic 2.5724 Predictive Power and Stability Test Error correction model can also be used in forecasting the rate of change of income inequality conditional on current and past changes in real per capita GDP and Islamization index. The result of dynamic forecasts for the level of natural log of Gini coefficient (indicator for income inequality) is presented in Table 7. The root mean sum squares of forecast errors of about 0.26% per year compares favorably with the value of the same criterion computed over the estimation period. The model also forecasts favorably the extent of the rise in income inequality for the last three years forecasting period (see Figure 3). The plots of cumulative sum of recursive residuals (CUSUM) and cumulative sum of squares (CUSUMSQ) of recursive residuals of the error correction model in stability tests also suggest that the regression coefficients of the underlying ARDL (1, 2, 2) model are generally stable over the sample period and do not reveal any evidence of statistically significant break (see Figures 4 and 5). Hence, this stability test confirms that the error correction model of underlying ARDL (1, 2, 2) has predictive power in forecasting the extent of changes in income inequality in the last three years period with very minimum errors as expected. DISCUSSION AND POLICY IMPLICATION The impact of Islamization on income inequality and economic growth relationship in Malaysia is empirically investigated using autoregressive distributed lag (ARDL) modelling approach based on collected time series dataset from 1969 to 2013. Exponential growth of Islamic capital over the last two decades was the major 73 International Journal of Applied Business and Economic Research

Wan Ahmad Wan Omar, Hafizah Abdul Rahim and Fauzi Hussin Based on ARDL Regression Based on 34 observations from 1977 to 2010. ARDL(1, 2, 2) selected using Akaike Information Criterion. Table 7 Dynamic Forecasts for Level of Log Gini Coefficient Dependent variable in the ARDL model is ln Gini t included with a lag of 1. List of other regressors in the ARDL model: ln y t, ln y t 1, ln Islam t, ln Islam t 1, ln Islam t 2, c 0, c 1 t Observation Actual Prediction Error 2011 3.7687 3.7691.3951E-3 2012 3.7804 3.7793.0010688 2013 3.7827 3.7871.0043928 Summary Statistics for Residuals and Forecast Errors Estimation Period Forecast Period Estimation Period 1977 to 2010 2011 to 2013 Mean.0000.0012397 Mean Absolute.0059306.0019522 Mean Sum Squares.6221E-4.6865E-5 Root Mean Sum Squares.0078876.0026201 Figure 3: Dynamic forecasts for change in income inequality International Journal of Applied Business and Economic Research 74

The Impact of Islamization on Income Inequality and Economic Growth Nexus in Malaysia Figure 4: Plot of cumulative sum of recursive residuals of income inequality, per capita GDP and Islamization index Figure 5: Plot of cumulative sum of square of recursive residuals for ARDL(1, 2, 2) model 75 International Journal of Applied Business and Economic Research

Wan Ahmad Wan Omar, Hafizah Abdul Rahim and Fauzi Hussin contributor for the rapid growth of Islamization process in Malaysia. Total Islamic capital in Malaysia was estimated at RM55.6 billion in the year 2000 and increased by more than fourteen-fold to RM800.1 billion in 2013, growing at the average rate of 11.2% per annum. In terms of ratio to GDP, the size of Islamic capital has also increased from 35.2% in 2000 to 81.0% in 2013. Hence, the study of the impact on the Islamization on the economic performance, specifically income inequality in Malaysia is timely. The result of cointegration analysis using ARDL modelling approach suggests that there is a level relation between Islam, proxied by Islamization index and economic growth on the performance of income inequality in Malaysia. All variables in the underlying ARDL model have passed the prerequisite unit root test prior to conducting bounds test for hypotheses testing. The results of joint null hypothesis based from F-statistic and W-statistic of bounds test reject the null hypothesis, when an appropriate lag order is selected with imposition of intercept and time deterministic trend into the underlying ARDL model. In backtracking regression where each exogenous becomes a new endogenous to recheck the joint null hypothesis, the results of both F-statistic and W-statistic for log real per capita GDP and log Islamization index fall well below lower bound of critical value band, suggesting do not reject the null hypothesis that the level variables of economic growth and Islamization do not enter significantly into the underlying ARDL equation. Therefore, these results suggest there exists a long-run relationship between income inequality, economic growth and Islamization, and the variables log real per capita GDP and log Islamization index can be treated as the long-run forcing variables for the explanation of log Gini coefficient (the indicator for income inequality). In the next hypothesis on income equality promoting, the result of long-run coefficients of the log real per capita GDP and log Islamization index suggests that Islamization support long-run distributive income with improved Gini coefficient as a result of better distribution of income. The F-statistic and W-statistic of the underlying ARDL model are well above the upper bound critical values, and long-run coefficients of log per capita GDP and log Islamization index are all very highly significant with expected positive values and relatively low standard error. The result from error correction model (ECM) for shortrun of the ARDL model provides further evidence that there is a dynamic relationship between income inequality, economic growth and Islamization. The ECM coefficient, estimated at 0.7252(0.103) is highly statistically significant, has the correct sign and suggests a relatively high speed of convergence in the economy to equilibrium once shocked. This result means that approximately 72.5% of disequilibria from previous year s shock converge back to the long-run equilibrium in the current year. Again, the ECM model also passes stability test, and the model has very good predictive power to forecast future income inequality. The impact of Islamization on the income inequality and economic growth nexus comes from Islamic capital as a transmission channel. The result of joint null hypothesis between log Islamic capital sub index as endogenous and other variables log Gini coefficient, log real per capita GDP, log Islamic faith sub index and log Islamic education sub index, has reveals that both computed F-statistic (7.7592) and W-statistic (38.7962) are higher than its upper bound critical values for F-statistic (5.2696) and W-statistic (26.3481) respectively at 95% confidence level. All variables in the ARDL estimates are significant and the model passes all diagnostic tests which is automatically computed by the system. The sub index for Islamic capital comprises of total assets in Islamic banking and takaful, net asset value in investment funds of Tabung Haji, Baitul-maal, Islamic-compliant investment in Permodalan Nasional Berhad (PNB), and Islamic fund International Journal of Applied Business and Economic Research 76

The Impact of Islamization on Income Inequality and Economic Growth Nexus in Malaysia management registered under Securities Commission (see Wan Omar, et al., 2014). These organizations are the major contributors for the payment of zakat which is disbursed by the baitul maal to provide various programmes to the poor and other asnaf in enhancing their income. The above mentioned results implicate that Islamization has the potential to be a policy target for prompting long-run income inequality improvement without jeopardizing the economic growth in Malaysia. Islamization process has been in Malaysia since 1960s with the inception of Tabung Haji and in the last two decades, we saw an exponential growth of Islamic capital market, in which Malaysia is one of the biggest Islamic capital market in the world. Hardcore poor in Malaysia have been completely eliminated since 1990s. CONCLUSION This paper provides an empirical evidence on relevance of application of Islam as a policy variable to provide a positive impact on the income distribution and economic growth. The impact of Islam on the relation of income inequality and economic growth in Malaysia, and its policy implicationare evidently drawn from the present study for which the results are merely considered as suggestive empirical predictabilities. The ARDL results also indicate the existence of behavioral relationship of Islam on income inequality and economic growth, that suggests how much the level of the indicator of income inequality will change when there is a change in Islamic indicator. Hence, the results from this study are very beneficial in impelling the belief about the relevance and relative significance of Islamic sector development policy for equitable distribution of income as per maqasid shari ah for the wellbeing of our Muslim ummah. What we can learn from this study is that the short-run and long-run estimates from the results may provide a simple barometer to reflect the degree of the actual occurrence in the development of Islamic sector in the economy. Through this study, for instance, there is an indication that Islamic sector development is moving in favorable direction in supporting distributive income without jeopardizing the economic growth. REFERENCES Abdul Khalid, Mohammed. (2014), The Colour of Inequality: Ethnicity, class, income and wealth in Malaysia. Kuala Lumpur: MPH Publishing. Ahmad Sarji, A. H. (2005), Indicators of Development of the Muslim Ummah in Malaysia. In The Encyclopedia of Malaysia: Religiouns and Beliefs, by Hassan M Kamal and Basri Ghazali. Singapore: Archipelago Press. Alesina, Alberto, and Dani Rodrik. (1991), Distributive Politics and Economic Growth. NBER Working Paper Series #3668, March. Anand, Sudhir. (1983), Inequality and Poverty in Malaysia: Measurement and Decomposition. Washington. DC: Oxford University Press. Aslam Haneef. (2001), Islam and Economic Development in Malaysia: A reappraisal. Journal of Islamic Studies 12(3): 269-290. doi:10.1093/jis/12.3.269. Barro, R. J. (2000), Inequality and Growth in a Panel of Countries. Journal of Economic Growth 5: 5-32. Barro, R. J, and R McCleary. (2003), Religion and Economic Growth. National Bureau of Economic Research (NBER), Working Paper Series No. 9682. Barro, Robert J. (2008), Inequality and Growth Revisited. Working Paper Series on Regional Economic Integration No.11, Jan: 1-14. 77 International Journal of Applied Business and Economic Research