AND THE UNDERGROUND ECONOMY IN : ARE THEY RELATED?

MUZAFAR SHAH HABIBULLAH, FACULTY OF ECONOMICS AND MANAGEMENT, UNIVERSITI PUTRA MALAYSIA YOKE-KEE ENG, FACULTY OF ACCOUNTANCY AND MANAGEMENT, UNIVERSITI TUNKU ABDUL RAHMAN

ABSTRACT

The underground economy (UE) also known as the shadow economy is a phenomenon known throughout the world. The growth of the UE is a matter of increasing concern, at least, throughout the developed nations. Researchers have recognized that the underground economy and crime are linked. Studies on the US, Canada and the transition economies indicate that crime and the shadow economy are facts of live. Given the important implication of UE, in this study, we use fuzzy set theory and fuzzy logic to construct an annual time-series for the (unobservable) Malaysia‟s UE over the period 1973-2003. Two input variables are used-the tax burden and government regulation. The resulting UE time-series is then tested for long-run relationship with the crime rate in Malaysia. Using the vector error-correction model, we are able to show that underground economy Granger causes the criminal activity in Malaysia. One policy implication from this study is that the must take proactive measures to curtail the development of the underground economy, for example, policies to attract unofficial activities back to the official sector.

INTRODUCTION

Striving for sustained long-run economic growth has always been on the top agenda for any country or government in the world. The importance of saving and investment process in economic development has led many democratic governments in both the developed and developing nations to adopt various liberalization programmes to encourage market forces to stimulate the economy. Because capital goods depreciate over time, a significant flow of saving must be generated and transferred into productive investment just to maintain a nation‟s capital stock and preserve existing living standards. For living standards to rise, a healthy flow of saving and investment must be sustained. As a general proposition, the more of current output saved and invested, the more rapid the rate of economic growth. However, sustaining long-run economic growth is not without cost. The United Nation Crime Commission meeting in Vienna in 2004 recognized that crime renders development unsustainable in many ways. They show that a high level of crime correlates with low level of human development and as a result, crime impedes economic growth in the long-run. Illegal activities such as extortion, , burglary, loan shark, drug trafficking, bombings or arson, kidnapping are some of the criminal activities that can hinder the development of a healthy economy. The purpose of the present paper is to determine empirically the relationship between criminal activities and shadow economy using annual data of an emerging economy, Malaysia for the period 1973 to 2003. As far as the authors are concern, this is a first attempt to relate empirically the relationship between crime and shadow economy in the literature. In fact, crime has received little attention and has been almost neglected by Malaysian economists. One difficulty why similar study has not been conducted is because of the unavailability of the long time series measure of shadow economy. However, in this study, following the work of Draeseke and Giles (2002), using the fuzzy set theory and fuzzy logic, we are able to construct

an annual series for the shadow economy in Malaysia. The advantage of applying fuzzy logic is twofold. First, it can avoid the complex calculations in conventional econometric models. Second, fuzzy rules with linguistic terms are easy for human to understand.

METHODOLOGY

In a simple definition, the shadow or underground economy is the part of the gross domestic product omitted, due to non-reporting and/or underreported in the official statistics. The latter includes the production of and trade in illegal products and services, unrecorded economic transactions that are due to the underdeveloped system of national statistics, as well as barter transactions and economic activities carried out by households. In general the shadow activities can be divided into four main categories: (i) underreported or not-reported; (ii) illegal; (iii) unrecorded; and (iv) household activities and barter transactions. Table 1 clearly shows the broad definition of the shadow economy. Schneider (2005) has estimated the size of the shadow economies for 110 countries for three periods of time (1990/1991, 1994/1995 and 1999/2000) using the DYMIMIC and the currency demand approach and these results are summarized in Table 2. At least two important insights can be drawn from the table. First, in all countries, the size of the underground economy has been increasing. And second, the average size of the shadow economy is larger for the developing compared to the developed nations. In Malaysia, 15 crime categories were collected as part of the Uniform Crime Reporting Program and considered representative of the most serious crime to form an index-Crime Index. The Crime indexes are split into two major subcategories- violent and property . Violent crime are those committed directly against a person that include murder, attempt to murder, gang-robbery with armed, gang-robbery without armed, robbery with armed, robbery without armed, rape, and aggravated assault. On the other hand, property crime are those in which there is no direct threat or harm to a person, and this include daytime house-breaking, night-time house-breaking, motor vehicle theft, non-vehicle and other theft. From the year 1973 to 2003 period, the number of crime reported in Malaysia has increased by 290 percent, from a total of 40,053 crimes in 1973 to 156,315 cases reported in 2003. However, when compared to other country, the Malaysian crime rate is low. According to the Seventh United Nations Survey of Crime Trends and Operations of Criminal Justice System, 1998-2000, in terms of total crime rate in Top100 countries all over the world, Malaysia is ranked 32, where in this category the United States leads and follow by Germany, United Kingdom, France and South Africa. Other Asian countries include Japan, India, South Korea and , respectively took the 8th, 10th, 11th, and 17th places. In another category, Malaysia ranked 21st for burglaries, ranked 13th for car thefts, ranked 24th for , ranked 26th for rapes, and ranked 22nd for robberies. Nonetheless, as cautioned by Andres (2002), such comparison may be misleading because recorded crime could be affected by multiple of factors such as penal laws, different recording practices etc. Table 3 illustrates the crime statistics by twelve categories of crime in Malaysia for the period 1973- 2003. In the table we sub-classify the period into 1973-82, 1983-92, and 1993-2003. In columns 2-4, we present the average number of cases, and in columns 5-7 is the average growth rates in crime cases, and the last three columns, represent the average share of criminal activities in total crime. Total crime include both violent and property crimes. While murder, attempted murder, armed robbery, robbery, rape and assault constitute violent crime, property crime consisted of daylight burglary, night burglary, lorry-van theft, car theft, motorcycle theft and larceny. As indicated in Table 3, the average number of all crime cases has been on an increasing trend. For the past three decades, the quantum of crime cases has shown an upward trend for all crime categories except for a brief dropped in number of cases for attempted murder for the period of 1983-92, and armed robbery

in the period 1992-2003. In all three periods, property crime represented more than 80 percent of all crime recorded (see column 8-10). The main contributor to property crime is larceny and followed by motorcycle theft and night burglary. Although the share of larceny and night burglary to total crime is on a decreasing trend, the share of motorcycle thefts is increasing. The share of motorcycle thefts has increased from 8 percent in 1973-82, to 15 percent in 1983-92 and 24 percent in 1993-2003 periods. As for other crime category, the share to total crime has been sustained. In Table 3, from columns 5-7, we observed that the average percentage growth rate of all crime categories for the period 1983-92 suggests that the growth in the number of cases is slowing down compared to the previous period. Except for murder and lorry-van theft, all category of crime has been slower despite their higher quantum in 1982-93 compared to 1973-82 periods. However, for the period 1992-2003, we experienced higher growth rates in all crime categories except for murder and armed robbery, which show an average growth of 3.2 percent and –1.9 percent respectively.

Theoretical Considerations Naylor (1996) argues that in the modern days, criminal activities are no more just redistribution of wealth, for example, burglary and ransom kidnapping, which are bilateral or trilateral – involving victim, perpetrator(s) and perhaps a middleman to dispose of the resulting merchandise or money, but increasingly, economically motivated crime means enterprise crime involving the production and distribution of goods and services facilitated by the underground economy. The dividing line between the explicitly criminal and the “informal” components of the modern underground economy has become equally blurred. Naylor stresses that there exists comfortable symbiosis between crime and the shadow economy and the links occur at many levels. According to Witte (1996), an important feature of the US urban areas is the existence of ghettos where an underground economy tends to flourish. In some cases underground economic activity is far more important than the legal economy in both the social and economic life in the ghetto. The underground economy in the ghetto provides a congenial environment for criminal activity because of its tenuous ties to legal institutions and its need of secrecy. Witte further notes that drugs and crime, particularly violent crimes are linked and these drugs are traded in an underground economy through organized gang members. According to Bate (2001) revenue losses resulting from organized crime forces government to cut programmes and services, increase taxation or even to run deficits. Not only do organized criminals siphon off money from society, they create billions in expenses for society to bear. These include the costs of prevention, detection and apprehension, justice and corrections systems and private security costs. There also are social costs that can be associated with fear of crime; refusal to go outside at night, moving to a safer neighbourhood, pain and suffering and loss of quality life. Noguera (1996) points out that crime and violent is not random phenomenon, but rather follow patterns. In his study, in many of Latino community, crime is organized through gangs and structured underground organizations. The transactions and “business” that takes place within the underground sector of the economy are by nature illegal. Noguera argues that the lack of economic opportunities in many impoverished urban communities creates conditions that are conducive to the occurrence of crime and the development of an underground economy. Merton (1967) contends that crime and violence are directly related to the absence of opportunities to achieve social mobility through legitimate channels. Furthermore, participants cannot appeal to official channels for arbitration or regulation. Consequently, there tends to be a high degree of violence associated with the underground economy, much of which is often gang related. Since the destruction of the Soviet Union, according to Glinkina (1999), the underground economic relations have developed above all because the state system was destroyed and the long established links between separate industries and whole territories were disrupted, the borders between new states are

permeable, presenting few obstacles to smuggling or other forms of criminal entrepreneurship. In fact, according to Bate (2001), smuggling is an important catalyst in stimulating the underground economy. It encourages existing criminal gangs and helps new ones to become established. Shelly (1998), on the other hand, argues that the largest element of the Soviet legacy is the concern of corruption and underground economy. During the Soviet period, members of the shadow economy who operated throughout the entire USSR routinely bribed government officials in order to sell produce, and law enforcement officials were paid to look the other way or to halt prosecution. In other word, the shadow economy led to criminal activity among the government officials in the Soviet. Johnson et al. (1998) and Friedman et al. (2000) suggest that the key determinants of underground activity are not tax rates but rather the extent of regulatory discretion. When regulations are lax and rule of law is weak, bureaucrats are allowed to make decisions on individual cases without supervision. This creates corruption, which causes firms to become unofficial. A smaller unofficial sector is found in a country with a lower regulatory burden on enterprises, less corruption, better rule of law, and a more efficient and competent tax administration.

Methods of Estimation The above discussion does not provide much information as to the direction and the long-run relationship between crime and the underground economy. In this study, we employ the concept of cointegration to infer long-run relationship between the two variables of interest – crime and the underground economy. According to Granger (1988), if two non-stationary variables, say X and Y are integrated of the same order and are found to be cointegrated, a usual conventional model which includes only dependent and independent variables in their stationary form is subject to mis-specification error. However, a conventional model is said to be well specified with the inclusion of the lagged residuals from the cointegrating regression between X and Y in their level form. Furthermore, with new development in the behaviour of the properties of time series, previous influential studies, particularly on causality in the economic literature has to be revalidated. In his words, Granger (1988: p. 204) concludes that, “Without zt being explicitly used, the model will be mis-specified and the possible value of lagged yt in forecasting xt will be missed. Thus, many of the papers discussing causality tests based on the traditional time-series modelling techniques could have missed some of the forecastability and hence reached incorrect conclusions about non-causality in mean. It does seem that many of the causality tests that have been conducted should be re-considered.” And more recently, Granger (1994) comments on the so-called the voluminous „cointegration- syndrome‟ studies exist in the literature. Granger points out that an empirical exercise should never just test for cointegration but should always followed with estimation of a full error-correction model of some form, as the extra gain in interpretation can be very worthwhile.

The Concept of Cointegration The concept of cointegration that was first introduced by Granger (1981) relates to the notion of a long run or equilibrium relationships among two or more variables. Granger (1981) pointed out that the movement of cointegrated series may be unequal in the short run but they are tied together in the long run, that is, they move parallel to each other over time. According to Granger (1986) and Engle and Granger (1987), a very important consequence of a cointegrated variables is that, one variable can be used to predict the other variable. In cointegration analysis, it is important that the series under study have the same order of integration.

Series Xt and Yt are integrated of the same order, denoted by XtI(d) and YtI(d), if the two time series require to be differenced d times to achieve stationarity. A series XtI(1), that is integrated of order one,

needs to be difference only once to achieve stationarity, that is, to become I(0). According to Granger (1986), „an I(0) series has a mean and there is a tendency for the series to return to the mean, so that it tends to fluctuate around the mean, crossing the value frequently and with rare extensive excursions.‟ For any I(1) series, it is always true that the linear combination of the two series will also result in an

I(1). However, it there exist a constant A, such that zt=Xt-AYt is stationary or I(0), then Xt and Yt will be said to be cointegrated, with A called the cointegrating parameter. If this were not the case, then the variables assumed to be generating the equilibrium could drift apart without bound, which is contrary to the equilibrium concept. If Xt and Yt are I(1) but cointegrated, then the relationship Xt=AYt is considered a long run or „equilibrium‟ relationship, and zt given above measures the extent to which the system Xt, Yt are out of equilibrium (Granger, 1986). Hence, the existence of a linear combination of two I(1) series that is I(0) suggests that the series generally move together over time, such that the relationship holds in the long run.

Unit Root Test for the Order of Integration Before we estimate the cointegrating regressions, we employed unit root tests to determine the order of integration of the individual series. This is because only variables that are of the same order of integration may constitute a potential cointegrating relationship. To test for the unit root of a time series, say Y, the augmented Dickey and Fuller (1981) unit root test is usually employed. The test is the t-statistic on parameter  from the following equation

L Y =  + Y +  Y +  (1) t 0 t-1 i1 i t-i t

where t is the disturbance term. The role of the lagged dependent variables in the augmented Dickey-

Fuller (ADF) regression equation (1) is to ensure that t is white noise. Thus, we need to choose the appropriate lag length L. Following Nelson and Plosser (1982), Perron (1988) uses a range of truncation lag parameters to evaluate the stationarity of the time series variables of interest. However, a less extensive approach in specifying the lag length is given by Schwert (1987, 1989). Following Schwert (1987, 1989) 1/4 the lag length was set equal to the integer portion of two values of  , that is, 4=int{4(T/100) } and 12= 1/4 int{12(T/100) }, and T is the number of observations. The null hypothesis, H0: Yt is I(1), that is, a unit root, is rejected (in favour of I(0)) if  is found to be negative and statistically significantly different from zero. The computed t-statistic on parameter , is compared to the critical value tabulated in MacKinnon (1991). When L = 0, we have the standard Dickey-Fuller test. The unit root tests for the first-difference of the variables is carried using the following regression equation

L 2Y =  + Y +  2Y +  (3) t 0 t-1 i1 i t-i t

where the null hypothesis is H0:Yt is I(2), that is, two unit roots, which is rejected (in favour of I(1)) if  is found to be negative and statistically significantly different from zero.

The Cointegration Tests After determining that the series are of the same order of integration, we test whether the linear combination of the series that are non-stationary in levels are cointegrated. In most studies, the Engle- Granger two-step estimation procedure was frequently used to test for cointegration. However, the Engle and Granger (1987) two-step estimation procedure for testing cointegration has been criticized for being

static and suffers from several econometric problems. First, Banerjee et al. (1986) have noted that even though the two-step procedure produces super consistent parameter estimates, for small sample the bias on the estimated parameter can be quite severe. Second, when cointegration between variables is not unique, the Engle-Granger two-step procedure performs less satisfactory. The estimates are not invariant to the chosen normalization, that is, which variable to be used as regressor and which to be used as regressand. Finally, regressing integrated series by using OLS method tend to invalidate statistical inferences (see Perman, 1991). As an alternative to the Engle-Granger two-step procedure, Johansen (1988) and Johansen and Juselius (1990) suggest the maximum likelihood estimation procedure to test for cointegrating relationship. This approach does not suffer from any of the problem mentioned above. Detailed exposition on the Johansen- Juselius technique has been provided in Dickey et al. (1991), Cuthbertson et al. (1992) and Charemza and Deadman (1992). However, a brief discussion on the Johansen-Juselius technique is provided below. We begin with by defining a k-lag vector autoregressive (VAR) representation

Xt =  + 1Xt-1 + 2Xt-2 + ... + kXt-k + t (t=1, 2,...,T) (4)

where Xt is a px1 vector of non-stationary I(1) variables,  is a px1 vector of constant terms, 1, 2...k are pxq coefficient matrices and t is a px1 vector of white Gaussian noises with mean zero and finite variance. Equation (4) can be reparameterised as

Xt =  + 1Xt-1 + 2Xt-2 + ... + k-1Xt-k+1 + kXt-k + t (5)

where i = - + 1 + 2 + ... + i (i=1, 2,...k-1) and  is defined as

 = - + 1 + 2 +...+ k. (6)

Johansen (1988) shows the coefficient matrix k contains the essential information about the cointegrating or equilibrium relationship between the variables in the data set. Specifically, the rank of the matrix k indicates the number of cointegrating relationships existing between the variables in Xt. In this study, for a two case variables, Xt = (crime rate and unemployment) and so p=2. Therefore, then the hypothesis of cointegration between crime rate and unemployment is equivalent to the hypothesis that the rank of k = 1. In other words, the rank r must be at most equal to p-1, so that r p-1, and there are p-r common stochastic trends. If the r=0, then there are no cointegrating vectors and there are p stochastic trends. The Johansen-Juselius procedure begins with the following least square estimating regressions

p1 X =  +  X +  (7) t 1 i1 i t-i 1t

Xt-p = 2 + iXt-i + 2t (8)

T Define the product moment matrices of the residuals as S = T -1   (for i,j=1,2), Johansen ij t1 it jt (1988) shows that the likelihood ratio test statistic for the hypothesis of at most r equilibrium relationships is given by

p -2lnQ = -T ln(1- ) (9) r ir1 i

where 1 > 2 >...p are the eigenvalues that solve the following equation

S22-S21S11’S12=0. (10)

The eigenvalue are also called the squared canonical correlations of 2t with respect to 1t. The limiting distribution of the -2lnQr statistic is given in terms of a p-r dimensional Brownian motion process, and the quantiles of the distribution are tabulated in Johansen and Juselius (1990) for p-r=1,...,5 and in Osterwald- Lenum (1992) for p-r=1,...10.

Equation (9) is usually referred to as the trace test statistic which is rewritten as follows

p L = -T ln(1- ) (11) trace ir1 i

where r+1,...p are the p-r smallest squared canonical correlation or eigenvalue. The null hypothesis is at most r cointegrating vectors. The other test for cointegration is the maximal eigenvalue test based on the following statistic

Lmax = -T.ln(1-r+1) (12)

th where r+1 is the (r+t) largest squared canonical correlation or eigenvalue. The null hypothesis is r cointegrating vectors, against the alternative of r+1 cointegrating vectors. Comparing the two tests, Johansen and Juselius (1990) indicate that the trace test may lack power relative to the maximal eigenvalue test which will produce clearer results. One cautionary note is that the Johansen-Juselius multivariate cointegration procedure is weak in small sample.

Sources of Data Data on crime and their subcategories for the period 1973 to 2003 are collected from the Royal Police of Malaysia (PDRM). The total crime activities are classified into 12 categories: murder, attempted murder, armed robbery, robbery, rape and assault (these comprise the violent crime); daylight burglary, night burglary, lorry-van theft, car theft, motorcycle theft and larceny (comprises the property crime). The data on underground economy was constructed for annual time series for the period 1970 to 2003 using the fuzzy logic approach. In contrast to the conventional parametric method, fuzzy logic avoids the need for rigid mathematical modeling and the distribution assumption. Fuzzy logic translates natural language descriptions of decision policies into an algorithm using a mathematical model. Such a model consists of fuzzification, inference, and defuzzification (Altrock, 1996). In the present study, we follow closely the work of Draeseke and Giles (2002) when estimate a measurement of underground economy for New Zealand. To simplify our analysis, our application uses two causal variables-variables which are

widely believed to be primary determinants of the underground economy. These two variables are the ratio of tax revenue to GDP (TR), and the ratio of government consumption expenditure to GDP to proxy for government regulation (REG). To illustrate, first, we set up the membership functions for the two factors. In fuzzy logic, both factors can be described using linguistic terms, such as Very Low (VL), Low (L), Normal (N), High (H), and Very High (VH). Meanwhile the linguistic terms for the „willingness to go underground‟ WTGU are Very Small (VS), Small (S), Average (A), Big (B), and Very Big (VB). All these terms are summarized in Table 4. Each linguistic term is associated with membership functions. The peaks of the linguistic terms are calculated by adding or subtracting one or two standard deviation (SD) and after applying the fuzzy rule laid out in Draeseke and Giles (2002), the linguistic term are now becoming human understandable. For example, the following rule:

IF REG = VH AND TR = VH THEN UE = VB

Can be read as

If government regulation is Very High and Tax Burden is Very High then the Underground Economy is Very Big.

Semantically, from that rule, when the government regulation is Very High, indicates that individuals tend to go UE to avoid the regulations. Similarly, when the Tax Burden is Very High, individuals would like to go UE to avoid the taxes. Hence, when both conditions are matched, the UE is considered Very Big.

The linguistic values are then translated into a numerical value, and this step is called defuzzication.

The index of WTGU is calculated as  (WTGU yi ) /  WTGU where yi is a weight. If WTGU is VERY

SMALL then yi is zero, if WTGU is SMALL then is 0.25, if WTGU is AVERAGE then is 0.5, if WTGU is BIG then is 0.75 and, finally, if WTGU is VERY BIG then is 1. For further details on computing index of the underground economy, refer to Draeseke and Giles (2002).

THE EMPIRICAL RESULTS

Before testing for cointegration by using the Johansen-Juselius procedure, we test for the order of integration of all categories of crime variables and the underground economy. Table 5 show the results of the unit root test for the test of the order of integration of the economic time series under investigation. Clearly the ADF test statistics indicate that the underground economy and all categories of crime in Malaysia are difference stationary, in other words, they are I(1) in levels. Having noted that all series are of the same order of integration, we run the cointegration test following the procedure provide by Johansen and Juselius (1990). These results are tabulated in Table 6. The null hypothesis of no cointegration cannot be rejected in almost all cases of the crime categories. However, in some cases, the null hypothesis of no more than zero and no more than one cointegrating relation is soundly rejected. The problem with these results is that they contradict the unit root tests. If there are two cointegrating vectors in a two-equation VAR, it means that the data are stationary I(0) series. Either the unit root tests are wrong and the series are really stationary I(0) processes, or the cointegration tests are wrong and the series are really nonstationary I(1) processes. Since there is no practical method for choosing

between these two sets of results, and knowing that the Johansen-Juselius cointegration procedure are distorted in small sample, we proceed our analysis employing the vector-error correction model to infer cointegration among the series. According to the „Granger Representation Theorem‟ not only does cointegration imply the existence of an error-correction model but also the converse applies, that is, the existence of an error-correction model implies cointegration of the variables. Since our task is to determine the long-run relationship and the causal direction between the two variables in question, we estimate a vector error-correction model (VECM). For the following bivariate vector error-correction models (VECM)

k k y =  +  y +  x +  ecm +  (13) t 0 i1 i t-i i1 i t-i 1 t-1 1t

k x =  +  x +  y +  ecm +  (14) t 0 i1 i t-i i t-i 2 t-1 2t

where ecmt-1 is the lagged residual from the cointegration between yt (say, crime rate) and xt (say, underground economy) in level. Granger (1988) points out that based on equation (13), the null hypothesis that xt does not Granger cause yt is rejected not only if the coefficients on the xt-i are jointly significantly different from zero, but also if the coefficient on ecmt-1 is significant. The VECM also provides for the finding that xt Granger cause yt, if ecmt-1 is significant even though the coefficients on xt-i are not jointly significantly different from zero. Furthermore, the importance of ‘s and ‘s represent the short-run causal impact, while  gives the long-run impact. In determining whether yt Granger cause xt, the same principle applies with respect to equation (14). Above all, the significance of the error-correction term indicates cointegration, and the negative value for ’s suggest that the model is stable and any deviation from equilibrium will be corrected in the long-run. The results of estimating equations (13) and (14) are presented in Table 7. In our study, we attempt to determine whether crime and the underground economy in Malaysia are related and when these two variables are related or exhibit long-run relationship, we would expect the estimated parameters of the error- correction terms in equations (13) and (14) are significant and show negative sign. Results in Table 7 indicate that only in the cases of total crime, murder, attempted murder, daylight burglary, night burglary and larceny when crime rate act as dependent variable; and armed robbery when underground economy act as dependent variable; that we found cointegration exists and the long-run relationship is stable. Furthermore, in all these cases, except one, the results suggest that the underground economy Granger cause the criminal activities in Malaysia. Only in the case of armed robbery that the result indicates that criminal activity Granger cause the underground economy in Malaysia. In the third column, our results suggest that in many of the cases the crime rate Granger causes the underground, despite showing long-run relationship, however, the model is unstable. In summary, we conclude that the underground economy leads the criminal activities in Malaysia.

CONCLUSION

The underground economy (UE) also known as the shadow economy is a phenomenon known throughout the world. The growth of the UE is a matter of increasing concern, at least, throughout the developed nations. One major concern of a growing UE is that the economic policy is drawn on erroneous “official” indicators like unemployment, official labor force, income, consumption or at least indicators that are “wrong” in magnitude. In such a situation a prospering UE may lead to severe difficulties for politicians

because it “causes” or “provides” unreliable official indicators and the direction of the intended policy measures may therefore be questionable. Researchers have recognized that the underground economy and crime are linked. Studies on the US, Canada and the transition economies indicate that crime and the shadow economy are facts of live. Given the important implication of UE, in this study, we use fuzzy set theory and fuzzy logic to construct an annual time-series for the (unobservable) Malaysia‟s UE over the period 1973-2003. Two input variables are used-the tax burden and government regulation. The resulting UE time-series is then tested for long-run relationship with the crime rate in Malaysia. Using the vector error-correction model, we are able to show that underground economy Granger causes the criminal activity in Malaysia. One policy implication from this study is that the government of Malaysia must take proactive measures to curtail the development of the underground economy, for example, policies to attract unofficial activities back to the official sector. This effort may assist macro stabilization. Eilat and Zinnes (2000) have laid down the following category of recommendations, and among others are: (1) Countries with a sizable unofficial economy should further strengthened market liberalization. Liberalization of official markets reduces costs of undertaking official business, providing an incentive to become efficient; (2) A primary cause of shadow economic activity is an attempt to avoid predatory and obstructive regulations and their associated licensing and bribe fees. Putting in place several simple regulatory reforms can lessen the regulatory compliance burden; (3) One factor that would encourage a firm to remain official is the availability of banking services. Therefore, banking sector privatization and restructuring that would improve such services through increased competition, financial deepening, and increased intermediation will tend to reduce shadow activity; and (4) We have shown that government failures are what stimulate the growth of a shadow economy. One very important aspect is to provide and enforce adequate protection of property rights because a primary benefit of being official is access to the legal system.

REFERENCES

Altrock, C.V. (1996). Fuzzy Logic and Neuro Fuzzy Applications in Business and Finance. Englewood Cliffs, NJ: Prentice Hall Press. Andres, A.R. (2002). Crime in Spain: An Empirical Investigation. Unpublished manuscript. Available at http://www.sam.sdu.dk/~jtl/adveco/antonio.pdf Bajada, C. and Schneider, F. (2005). The Shadow Economies of the Asia-Pacific. Pacific Economic Review 10(3): 379-401. Banerjee, A., Dolado, J.J., Hendry, D.F. and Smith, G.H. (1986). Exploring Equilibrium Relationship in Econometric through Static Models: Some Monte Carlo Evidence. Oxford Bulletin of Economics and Statistics 48: 253-277. Bate, R. (2001). Smuggling: Unintended Consequence of Government Intervention. The Liberty Briefing 1: 3-5. Brantingham, P. and Easton, S.T. (1998). The Costs of Crime: Who Pays and How Much? 1998 Update. 1998 Fraser Institute Critical Issues Bulletin. The Fraser Institute. Charemza, W.W. and Deadman, D.F. (1992). New Direction in Econometric Practice. England: Edward Elgar. Cuthbertson, K., Hall, S.G. and Taylor, M.P. (1992). Applied Econometric Techniques. New York: Harvester Wheatsheaf. Davidson, J.E.H., Hendry, D.F., Srba, F. and Yeo, S. (1978). Econometric Modelling of the Aggregate Time Series Relationships between Consumers‟ Expenditure and Income in the United Kingdom. Economic Journal 88: 661-692. Dickey, D.A. and Fuller, W.A. (1981). Likelihood Ratio Statistics for Autoregressive Time Series with a Unit Root. Econometrica 49: 1057-1077. Dickey, D.A., Jansen, D.W. and Thronton, D.L. (1991). A Primer on Cointegration with an Application to Money and Income. Federal Reserve Bank of St. Louis Review 73: 58-78. Draeseke, R. and Giles, D.E.A. (2002). A Fuzzy Logic Approach to Modelling the New Zealand Underground Economy. Mathematics and Computers in Simulation 59: 115-123. Eilat, Y. and Zinnes, C. (2000). The Shadow Economy in Transition Countries: Consequences for Economic Growth and Donor Assistance. CAER II Discussion Paper No. 83, Harvard Institute for International Development. Engle, R.F. and Granger, C.W.J. (1987). Co-integration and Error Correction: Representation, Estimation and Testing. Econometrica 55: 251-276. Foglesong, T.S. and Solomon Jr., P.H. (2001). Crime, Criminal Justice and Criminology in Post-Soviet Ukraine. National Institute of Justice, US Department of Justice, USA. Freeman, R.B. (1996). Why Do So Many Young American Men Commit Crimes and What Might We Do About It. Journal of Economic Perspectives 10(1): 25-42. Friedman, E., Johnson, S., Kaufman, D. and Zoido-Lobaton, P. (2000). Dodging the Grabbing Hand: The Determinants of Unofficial Activity in 69 Countries. Journal of Public Economics 76: 459-493. Glinkina, S. (1999). Russia‟s Underground Economy During the Transition. In E.L. Feige and K. Otto (eds.). Underground Economies in Transition: Unrecorded Activity, Tax Evasion, Corruption and Organized Crime. Aldershot: Ashgate. Granger, C.W.J. (1994). Some Comments on Empirical Investigations Involving Cointegration. Econometric Reviews 13(3): 345-350. Granger, C.W.J. (1988). Some Recent Development in a Concept of Causality. Journal of Econometrics 36: 199-211. Granger, C.W.J. (1986). Developments in the Study of Cointegrated Economic Variables. Oxford Bulletin of Economics and Statistics 48: 213-228. Granger, C.W.J. (1981). Some Properties of Time Series Data and Their Use in Econometric Model Specification. Journal of Econometrics 16: 121-130. Hendry, D.F. (1983). Econometric Modelling: The Consumption Function in Retrospect. Scottish Journal of Political Economy 30: 193-220. Hendry, D.F. (1979). Predictive Failure and Econometric Modelling in Macroeconomics: The Transactions Demand for Money. In P. Omerod (ed.). Economic Modelling. London: Heinemann. Johansen, S. (1988). Statistical Analysis of Cointegration Vectors. Journal of Economic Dynamics and Control 12: 231-254. Johansen, S. and Juselius, K. (1990). Maximum Likelihood Estimation and Inference on Cointegration with Application to the Demand for Money. Oxford Bulletin of Economics and Statistics 52: 169-210.

Johnson, S., Kaufman, D. and Zoido-Lobaton, P. (1998). Regulatory Discretion and the Unofficial Economy. American Economic Review 88(2): 387-392. MacKinnon, J.G. (1991). Critical Values for Cointegration Tests. In R.F. Engle and C.W.J. Granger (eds.). Long-Run Economic Relationships: Readings in Cointegration. Oxford: Oxford University Press. Merton, R. (1967). On Theoretical Sociology. New York: The Free Press. Naylor, R.T. (1996). From Underworld to Underground: Enterprise Crime, “Informal Sector” Business and the Public Policy Response. Crime, Law & Social Change 24: 79-150. Noguera, P.A. (1996). Violence Prevention and the Latino Population: An Analysis of Patterns and Trends and Prescriptions for Effective Intervention. Motion Magazine. Available at http://www.inmotionmagazine.com/pedro21.html Osterwald-Lenum, M. (1992). A Note with Quantiles of the Asymptotic Distribution of the Maximum Likelihood Cointegration Rank Test Statistics. Oxford Bulletin of Economics and Statistics 54: 461-472. Perman, R. (1991). Cointegration: An Introduction to the Literature. Journal of Economic Studies 18: 3-30. Phillips, A.W. (1954). Stabilisation Policy in a Closed Economy. Economic Journal 64: 290-323. Sargan, J.D. (1964). Wages and Prices in the United Kingdom: A Study of Econometric Methodology. In P.E. Hart, G. Mills and J.K. Whittakers (eds.). Econometric Analysis for National Planning. London: Butterworths. Schneider, F. (2005). Shadow Economies Around the World: What Do We Really Know? European Journal of Political Economy 21: 598-642. Schneider, F. and Enste, D. (2002). Hiding in the Shadows: The Growth of the Underground Economy. Economic Issues No. 30. International Monetary Fund. Shelly, L.I. (1998). Organized Crime and Corruption in Ukraine: Impediments to the Development of a Free Market Economy. The Journal of Post-Soviet Democratization 6(4): 648-663. Schwert, G.W. (1987). Effects of Model Specification Tests for Unit Root in Macroeconomic Data. Journal of Monetary Economics 20: 73-103. Schwert, G.W. (1979). Tests of Causality: The Message in the Innovations. In K. Brunner and A.H. Meltzer (eds.). Three Aspect of Policy and Policymaking: Knowledge, Data and Institutions. Amsterdam: North-Holland. United Nations Crime Commission. (2004). Background Press Release. Available at http://www.unis.unvienna.org/unis/pressrels/2004/uniscp485.html. Witte, A.D. (1996). Urban Crime: Issues and Policies. Housing Policy Debate 7(4): 731-748.

Table 1. A taxonomy of types of underground economic activities

Type of activity Monetary transactions Non-monetary transactions Illegal activities Trade with stolen goods; drug dealing Barter of drugs, stolen goods, smuggling and manufacturing; prostitution; etc. Produce or growing drugs for own gambling; smuggling; fraud; etc. use. Theft for own use.

Tax evasion Tax avoidance Tax evasion Tax avoidance Legal activities Unreported income Employee Barter of legal All do-it-yourself from self- discounts, services and goods work and employment; Wages, fringe benefits neighbour help salaries and assets from unreported work related to legal services and goods Source: Schneider (2005) and the reference therein.

Table 2. Average size of the shadow economy for developing, transition and OECD-countries

Countries/Year Average size of the shadow economy-value added in % of official GDP using DYMIMIC and currency demand method (Number of countries Mostly developing countries: 1990/1991 1994/1995 1999/2000

Africa (24) 33.9 37.4 41.2 Central and South America (17) 34.2 37.7 41.5 Asia (25) 20.9 23.4 26.3 Transition countries (23) 31.5 34.6 37.9 Highly developed OECD countries 13.2 15.7 16.8 (21) Source: Schneider (2005: Table 5.1, p. 618).

Table 3. Descriptive statistics on criminal activities in Malaysia, 1973-2003

Crime Average number of cases Average growth rates in Average share of criminal category crime cases in percentage activities to total crime 1973-82 1983-92 1993- 1974-82 1983- 1993- 1973- 1983- 1993- 2003 92 2003 82 92 2003

Crime: 62638 77262 127550 6.4 1.2 8.2 100 100 100 Violent: 6023 10102 17065 10.1 4.1 8.1 9.49 13.10 13.45 Murder 240 348 514 4.0 7.2 3.2 0.39 0.46 0.42 Attempted 64 45 55 4.5 4.2 12.2 0.10 0.06 0.05 murder Armed 503 817 687 12.6 3.8 -1.9 0.81 1.05 0.61 robbery Robbery 3220 5758 10179 14.6 4.0 10.5 5.01 7.42 7.81 Rape 324 607 1258 8.2 5.5 6.9 0.52 0.80 1.03 Assault 1673 2526 4372 6.4 4.3 5.8 2.66 3.31 3.53 Property: 56616 67160 110485 6.1 0.8 8.2 90.51 86.90 86.55 Daylight 3634 4445 7062 8.6 3.2 4.9 5.69 5.79 5.76 Burglary Night 12395 16711 20331 10.8 0.5 3.7 19.57 21.58 16.83 Burglary Lorry-van 167 576 2781 16.4 16.6 18.2 0.26 0.77 2.04 theft Car theft 1168 2918 5243 15.5 6.1 11.4 1.83 3.77 3.95 Motorcycle 5342 11635 32696 15.2 4.4 15.4 8.37 14.99 24.49 theft Larceny 33911 30876 42372 2.9 -0.7 6.0 54.78 40.00 33.49

Notes: Authors‟ calculation.

Table 4. Linguistic terms

TR: Tax burden Very Low (VL) Low (L) Normal (N) High (H) Very High (VH) -2SD -1SD Mean +1SD +2SD

REG: Govt regulation Very Low (VL) Low (L) Normal (N) High (H) Very High (VH) -2SD -1SD Mean +1SD +2SD

UE: Underground economy Very Small (VS) Small (S) Average (A) Big (B) Very Big (VB)

Table 5. Results of ADF unit root test Crime rate category Level First difference (Intercept and Trend) (Intercept)

Crime: -2.35 -3.24 [0.39] [0.02] Violent: -2.75 -3.71 [0.22] [0 ] Murder -3.56 -4.69 [0.05] [0 ] Attempted murder -2.20 -4.49 [0.46] [0 ] Armed robbery -2.32 -4.48 [0.40] [0 ] Robbery -2.17 -3.47 [0.48] [0.01] Rape -3.31 -4.97 [0.08] [0 ] Assault -2.87 -3.21 [0.18] [0.02] Property: -2.29 -3.19 [0.42] [0.03] Daylight Burglary -3.31 -3.20 [0.08] [0.03] Night Burglary -3.06 -3.71

[0.13] [0 ] Lorry-van theft -2.41 -4.25 [0.36] [0 ] Car theft -2.01 -3.39 [0.56] [0.01] Motorcycle theft -2.19 -3.00 [0.47] [0.04] Larceny -2.38 -3.34 [0.37] [0.02]

Underground economy -2.68 -6.09 [0.25] [0 ]

Notes: All unit root estimations were done using Eviews. Eviews select lag 1 as default and were used throughout the analysis. The square brackets, [.],.contain the p-values.

Table 6. Results of Johansen-Juselius multivariate cointegration test Crime rate category Trace statistics Max-eigenvalue statistics

Ho: r=0/HA: r≥1 Ho: r≤1/HA: r=2 Ho: r=0/HA: r=1 Ho: r≤1/HA: r=2

Crime: 11.38 2.60 8.79 2.60 Violent: 5.95 0.38 5.57 0.38 Murder 11.46 4.22* 7.23 4.22* Attempted murder 15.31 5.14* 10.17 5.14* Armed robbery 11.14 2.44 8.69 2.44 Robbery 5.83 0.17 5.65 0.17 Rape 3.97 0.02 3.95 0.02 Assault 10.27 3.44 6.82 3.44 Property: 11.72 2.20 9.52 2.20 Daylight Burglary 15.26 3.24 12.02 3.24 Night Burglary 10.79 3.94* 6.84 3.94* Lorry-van theft 5.86 0.63 5.23 0.63 Car theft 6.70 0.56 6.13 0.56 Motorcycle theft 6.69 0.22 6.46 0.22 Larceny 16.05* 3.97* 12.08 3.97* Notes: Critical values are: Trace statistics: 5% 1%

Ho: r=0/HA: r≥1 15.41 20.04

Ho: r≤1/HA: r=2 3.76 6.65

Max-eigenvalue statistics: 5% 1%

Ho: r=0/HA: r=1 14.07 18.63

Ho: r≤1/HA: r=2 3.76 6.65 Asterisk (*) denotes statistically significant from zero at the 5% level.

Table 7. Results of implied cointegration tests and Granger long-run causality

Crime category 1 2

Crime: -1.95* 1.76* Violent: -0.87 2.06** Murder -1.70* 1.96** Attempted murder -2.70** 0.75 Armed robbery -0.37 -2.49** Robbery -0.76 2.13** Rape -1.07 1.57 Assault -1.46 2.02** Property: -1.54 2.42** Daylight Burglary -3.18** 0.68 Night Burglary -2.46** 0.45 Lorry-van theft 0.65 2.02** Car theft -0.57 2.26** Motorcycle theft -0.94 2.17** Larceny -1.80* 2.82**

Notes: Asterisks (*, **) denote statistically significant from zero at the 10% and 5% level respectively