ࡱ> BD?@A5@eBbjbj22 HXX G. . . @n 0   $6 ސސސP.J|6 ,zҕ(BR@XN+P+P+P+P+P+P+$/R1t+] _:>__t+ @+_ R N+_N+.H " 6ƕ 72ސ\ @| +0,Jo2ko266 6 o2 6D ܣzLƯt+t+6 6 Q_1Cd6 6 _EXAMINING THE EFFICIENCY OF CROATIAN HIGHER EDUCATION: AN APPLICATION OF STUDENT ATTAINMENT MODELLING Maja Mihaljevi, MA University of Split, Faculty of Economics Matice hrvatske 31, 21000 Split, Croatia Phone: ++385(0)21 430722 Fax:++385(0)21 430701 E-mail address: majam@efst.hr Key words: Education production function, Higher education, Effort, Peer effects 1. INTRODUCTION Similar to HE systems in the neighbouring countries, the Croatian HE framework has experienced some remarkable changes in the past decade. Understanding what fundamental factors affect the behaviour of the agents (i.e. students) in the centre of the education production process is of significant importance. Using an educational production function this paper introduces a model of attainment which incorporates the distinctive features of Croatian higher education (HE) into the standard attainment modelling practice. In a simple principal-agent framework students choose their effort levels to maximise their benefits i.e. future returns to education. The principal is the higher education institution (HEI), setting appropriate incentives to achieve its main goal of maximising educational outcomes. In the proposed model of education production students performance is captured by the students average grade for the first year of studies (GPA1). Under the assumption that this measure improves the signalling of students academic attainment to further educational programmes and prospective employers, we might conclude that it is, thus, creating incentives for a student to maximise his/her performance in the HEI. Explanatory variables used in the model relate to students personal and family characteristics, prior attainment, peer effects, socioeconomic background and there are several dummy variables controlling for the type of programme attended at the HEI. To the best of our knowledge, benefiting from a unique data set, we are including for the first time a measure of students effort in a HE attainment model. The first part of this paper, namely Sections 2 and 3, will address the theory behind student attainment modelling. Some of the limitations of the existing HE research are addressed in Section 4, while the distinctive characteristics of the Croatian HE system are presented in Section 5. These distinctive features are then incorporated in the empirical specification (Section 6) and the results are discussed (Section 7). At the end, some recommendations for future empirical work are presented and main conclusions are highlighted. 2. GENERIC THEORY In a higher education system, universities comprising of faculties, art academies, colleges and polytechnics can be considered as multi-product enterprises using what Rotschild and White (1995) label as a customer-input technology. The quality of the customer-input technology may have an impact on peer effects in the HEI which in turn has been found to have a significant effect on student achievement. This relationship between peer effects and student attainment will be examined in more detail in Section 6.4.1. The various resources in education can be combined in different ways to achieve certain educational ends. Examining the efficiency of these different combinations is of significant interest to policy makers and has important policy implications. Use of economic principles to assess the efficiency of HEIs relies on the analogy between educational enterprises and firms where educational enterprises produce educational outcomes similar to the way that firms produce outputs. Hence, the underlying economic principles from the neoclassical theory of the firm (Varian, 1999; Baumol et al., 1983) are applied to illustrate the functioning of educational enterprises. In higher education, a production function is used to express the relationships between institutions inputs and outputs where outputs might be more appropriately considered as outcomes of the educational process and will henceforth be referred to in that way. In a mathematical form it shows how an institution generates a vector of outcomes using a flow of inputs and some available technology. In HE, the outcomes are not easily quantifiable and may include some measure of student attainment, wages, well being and other benefits of the educational process where the central transformation process i.e. learning technologies or management practices, is excluded (Belfield, 2000). Also, educational outcomes are not sold at market prices, thus making it difficult to attach a market value to outcomes. Specifying outcomes in HE can, therefore, depend on the objective function of HE sector, which may be various, namely producing graduates, fostering academic excellence, disseminating knowledge through expanding enrolments, etc. (e.g. Sheehan, 1973). From the perspective of students objective function, in a simple HE framework, we assume that students choose their effort levels to maximise their benefits i.e. future returns to education where supplying more effort in the education process may imply a better earning power in the future. Following Costrell (1994, 1997), we can model a students utility function as depending on leisure and earnings where earnings are reduced if the student does not provide some effort in HE. The student supplies effort and chooses the achievement level that maximises his/her utility up to the tangency of that utility function and educational production function that is linked with earnings. Assuming that students grades or a similar type of credentials serve as a signal of student s academic attainment to the labour market or further educational programmes then, we might argue that students have an incentive to maximise their performance in the HEI, while trying to maximise the expected net benefits (from the labour market or elsewhere). Hence, there are additional outcomes of students level of attainment translated into direct utility, namely, higher earnings or general well being. Furthermore, along with the individual benefits from education there are other external benefits to the society that are usually not internalised into students' decisions (for a more detailed analysis of the conventionally discussed external benefits see Barr, 1998). The review of the development of the theory is presented in the next Section. It is necessary to emphasize some of the limitations in applying the theory of the firm to model education production. Firstly, to use the economic analogy in the context of higher education we would need to assume that HE is a business: it produces and sells educational services to consumers for a price and it buys inputs with which to make that product (Winston, 1999, p. 13). Hence, we could draw parallels between the HEIs and firms, students and customers, faculty and labour market, etc. However, contemporary HEIs are diverse, have multiple purposes and engage in a number of activities sometimes extending further than the customary teaching and research work (Cohn and Cooper, 2004). Furthermore, as Dixit (2002) notes, the education sector has some important distinguishing characteristics, namely multiple stakeholders, multiple objectives and multiple outputs. This makes the monitoring of educational production a complex issue. Hence, higher education is different from the familiar for-profit enterprise setting as Winston (1999) concludes. 3. LITERATURE REVIEW OF THE DEVELOPMENT OF THE THEORY After assuming that students and HEIs goal is the maximisation of educational outcomes we turn to the standard functional form of education production. A simplified education production function may take the following form: A = f (R, F, P, At-1, E)(1) Where A is student attainment; R represents institution's resource input; F includes family characteristics; P is peer input; At-1 is prior student attainment (capturing student ability) and E is students effort. The underlying assumption is that students are maximising their attainment (A) subject to constraints. In modelling attainment in HE, the dependant variable (A) is usually grades or performance in a written test, and the goal of the HEI is to allocate students and resources in such a way that it maximises student attainment conditional on the ability distribution of students. These actions by HEI are subject to budget constraint usually modelled as the amount of resource per student (typically modelled exogenously in a simple funding formula) multiplied by number of students. All of the explanatory variables are expected to have a positive effect on the dependent variable. P and A are determined exogenously, and E can be considered a function of prior attainment and cognitive ability and thus, may be determined endogenously. In our empirical investigation we introduced a separate effort variable to take account of the difference between effort and prior ability. It can be argued that the peer effects and the student effort variable are vital to the efficiency of education providers and also are quite complex in education production modelling i.e. the application of student attainment model to Croatia, hence they will be discussed in more detail in the next section. 3.1. Effort Theoretical examinations of the role of effort in educational process are quite rare and in contrast with the extensive literature on the role of effort in firms. For the latter, one of the most important contributions is the theory of efficiency wages developed by Shapiro and Stiglitz (1984). Holstrom and Tirole (1989) provide a survey of the work on the role of effort in firms. More recently, empirical estimations of effort have taken place for firms (summarised in De Fraja et al., 2005), and effort has been assessed by absenteeism (Ichino and Riphahan, 2004), the quit behaviour (Galizzi and Lang, 1998) and misconduct (Ichino and Maggi, 2000). However, given the distinctive features of HE sector and the differences between the firms and HEIs most of the findings on the role of effort in firms cannot be applied in student attainment modelling. In educational psychology, effort is one of the most studied variables affecting educational outcomes (Schenk, 2003). It has a significant impact on motivation (which may determine the trade-off between leisure and work) and thus, on student attainment. In our model we consider that the motivation is embedded in our effort variable, hence, we are not treating them as distinct. In education, the lack of data on effort usually impedes its inclusion as a variable in the studies on attainment thus leading to empirical specifications lacking one of the vital determinants in understanding the process. For studies on primary and secondary education, effort is conventionally examined by using proxies such as time spent on homework, time parents spend reading to their children, i.e. the data is mostly qualitative rather than quantitative and regression analysis is rarely used. There are few examples of modelling effort for primary and secondary education and estimating its impact on attainment i.e. Bonesrnning (2004), De Fraja et al. (2005). However, the data is rather limited and the research does not provide a clear-cut answer as to the relationship between the key players, namely teachers, students and parents and the interaction of their effort levels on pupils' attainment. This is especially a problem when modelling student attainment in HE, since there is a general lack of understanding of the role of student effort in educational attainment. To the best of our knowledge, there are no empirical studies in HE that are modelling student attainment using effort as an explanatory variable. 3.2. Peer Effects In terms of peer-group effects, they are perceived as a group of influences arising from 'social interactions' where the behaviour of one individual is affected by the behaviour or characteristics of other individuals in the same group. The characteristics of these interactions can be linked with the customer-input technology since, in student attainment models, peer quality can technically be considered as an input into the HEIs education production (Winston, 1999). Similar to the previously mentioned problems in using effort as an explanatory variable, studies on educational attainment that use peer effects as one of the determinants in the model are relatively limited. This is especially evident in models of attainment dealing with higher education where peer effects are very difficult to specify due to the fact that in most of the HE settings student chose their own peers. This introduces what Manski refers to as the 'reflection problem' (Manski, 1993) i.e. smarter students tend to choose to be around other smarter students thus it might be quite difficult to statistically distinguish between the effects of students own smartness and the smartness of its peers. For students in primary and secondary schools it is possible that they are assigned to classrooms in a way that is not related to achievement, hence the 'reflection problem' can be bypassed leading to more studies examining peer effects at this level. Peer effects at this pre-tertiary level have played an important analytical and empirical role since the publication of the Coleman Report in 1966 (Coleman et al. 1966). Studies such as Henderson et al. (1976), Hoxby (2000), Zimmer and Toma (2000), Checchi and Zollino (2001), Hanushek at al. (2003), Robertson and Symons (2003) and McEwan (2003) have found positive peer effects operating at the classroom level i.e. having better peers can improve students own attainment. Furthermore, some studies found that this effect was larger for low-ability students. Some of the few studies on peer effects in HE will be addressed in Section 4 along with a more detailed discussion of empirical work on modelling attainment. 4. CRITICAL EXAMINATION OF PREVIOUS EMPIRICAL WORK The bulk of theoretical and empirical work on education production functions comes from the United States and more recently, the United Kingdom hence, these HE environments are only partially comparable to that in Croatia. Elsewhere in Europe, education production studies in most cases are restricted to examining qualitative rather than quantitative determinants in education production and, if there is an application of regression analysis, it is usually estimated for elementary or secondary school level benefiting from the data from standardised tests like PISA and TIMSS. Furthermore, most of those studies tend to focus on resource effects with little focus on the determinants of educational attainment. Some of the exceptions modelling attainment in secondary education and discussing some of its determinants (i.e. mostly personal characteristics of students or socioeconomic background) are Feinstein and Symons (1999); Ammermueller et al. (2003); Brunello and Checchi (2003); Hakkinen et al. (2003); Hazans et al. (2003); Wolter (2003); Woessmann (2004) and Schneeweiss and Winter-Ebmer (2005). For studies dealing with higher education the quantity and the quality of the educational process is commonly captured by completion rates, test scores and grades obtained while standardized test-results from secondary school (e.g. A-levels in the UK and SAT scores in the US literature) are frequently used as proxies for value added in knowledge transmission at secondary level i.e. they are taken as a measure of student ability. Turning to other inputs in the education process, besides the ability (which is found to positively influence student attainment e.g. Rudd, 1984; Smith and Naylor, 2001), there are significant gender differences in educational performance. For example, in the studies by Rudd (1984), Smith and Naylor (2001), McNabb et al., (2002) females are found to outperform males, while Johnes and Taylor (1990) examine the ratio of male to female graduates and find no conclusive evidence on gender differences in attainment. Other personal characteristics that have been found to have a significant influence on student attainment are marital status, country of origin and age. For example, in Smith and Naylor (2004) married and UK-born students perform better than unmarried and overseas students. Blundell et al. (1997) use an ordered probit analysis of student attainment in HE (running separate regressions for men and women) and find that the type of secondary school attended has a statistically significant effect on HE attainment only for women. Furthermore, the effect of students age on attainment has been inconclusive. None of the empirical investigations addressed above attempted to evaluate student effort in HE nor did their authors acknowledge the lack of such an important variable. In regard to the peer characteristics in HE, they were mostly proxied by some variant of academic ability (e.g. SAT scores in the US HE and A-levels for UK) or some other more specific variables generated in the admissions process. The influenced behaviour i.e. student attainment, was usually captured by grades (GPA) or score in some written test. In HE, the studies trying to avoid the 'reflection problem' (issue discussed in Section 3.2) mostly focused on roommate level peer effects. For example, Sacerdote (2001) finds that students own GPA and his roommates are strongly correlated at there is some evidence of nonlinear peer effects. Zimmermann (2003) finds positive peer effects for some categories of students where having a roommate with low SAT score will diminish the academic performance of the student close to the middle of the SAT distribution. Winston and Zimmerman (2004) find that the strongest peer effects are associated with students in the middle of the SAT distribution and the negative effects of a low SAT roommate seem to influence students achievement more than the positive effect of the high SAT roommate. Peer group effects might influence student attainment not only through ability but also through socio-economic status and race (Hanushek et al. 2003), and gender composition (Robst et al., 1998, Smith and Naylor, 2001, Hoel et al. 2005). The evidence on the nature of any nonlinearities or gender peer effects is not completely clear. It seems that student in the middle of ability distribution are usually more sensitive to peer influences than those at the either end of the ability distribution. McNabb and Johnes (2004) seek to capture peer effects in UK by using secondary school outcomes i.e. A-levels. We argue that the effect of prior attainment (proxied by A-levels) may not be a good proxy for peer effects in HE since the HE environment is different in terms of student behaviour and there is a changing patter of social interactions as student matures (drawing on the work by Akerlof and Kranton, 2003). This is especially relevant if we are observing long duration of study like in Croatia. Furthermore, the authors scarcely address the endogeneity issues when estimating peer effects, thus perhaps introducing bias in their estimates. Although the empirical work to confirm the existence of peer effects in HE can be found, it is related to narrowly examined environments, predominantly the US and UK HE systems, and can rarely be applied in other different HE settings, i.e. Croatia. Furthermore, the analysis of peer effects focusing solely on the interactions of roommates might not be an appropriate way to address the problem. We argue that for a HE setting it would be more appropriate to look into classroom/course-group interactions than roommate interactions, since the former is more in line with the principle of studies in HE. This is also valid for Croatia where it is typical that the students are living at home while studying and most of the interactions are in course-groups to which the students are usually randomly assigned. Some of the specific features of Croatian HE system are presented in the next Section and these will be integrated into the empirical part. Overall, several problems can be identified as dominant in the studies modelling student attainment in higher education. Although the studies are generally technically precise in modelling attainment and addressing some complex issues (e.g. capturing peer effects) there seems to be no clear understanding and analysis of the underlying theory of education production e.g. the lack of a quantifiable measure of effort is rarely acknowledged. This seems to demonstrate that the transformation process ('of turning inputs to outputs') and its interaction with the educational environment are largely disregarded. Furthermore, the studies are usually vague in identifying who the decision making unit is in the production of education outcomes. That might be connected to another issue, namely the neglect of agency issues in modelling student attainment (also identified by Carnoy, 1995). 5. DISTINGUISHING FEATURES OF CROATIAN HE SECTOR Remarkably little is known about the determinants of student performance in Croatian HE and whether the current structure of educational production matches the needs of the economy. There are no national educational standards or external evaluations of exams hence, there is no feedback from HEIs on the educational practices/outcomes at tertiary level. There are no comparisons made through time, between regions and between different types of programmes. There is also a lack of trained professionals in the area of quality monitoring and assessment. Currently new issues are being addressed in the legal framework but have not however been effectively implemented, namely the degree of autonomy of the HEIs, introduction of quality assurance systems, performance-linked funding and market forces in HE operations. Although the Croatian government views education as a key element in the transition to a democratic society, it still needs to develop interpretable measures of education outcomes, such as student attainment, to monitor the efficiency of its education system. That type of information can be seen as a first contribution to assessing the effectiveness of the systems and the reforms taking place. According to the OECD report (2004) there are significant contributions for a country stemming from improvements in educational attainment. Positive effects are apparent in labour productivity, technological progress and in supporting countrys economic growth. It is estimated that in the OECD area one additional year of education increases economic output by between 3-6 percent (OECD, 2004). Similar to other HE systems in neighbouring countries, Croatian HEIs are faced with remarkable increases in the number of students thus leading to increasing concerns over stretched out resources and possibly deteriorating educational quality. Most of the other developing countries are also left with a similar daunting task - to cope with the expansion of HE sector while improving the quality of education, all within budgetary constraints (World Bank, 2000a). Despite the drastic increase of the population enrolling to some form of tertiary education in Croatia (from 1991/92 until 2004/05 there had been an 86 percent increase), the main problems are still in the long duration of studies and a small number of students earning a degree. According to the UNESCO estimates, the gross enrolment rate in tertiary education in Croatia in 2003 was 28.3 percent and only two thirds of all students who enrolled actually obtained their degrees (`oai, 2004). Moreover, the number of graduated (in 2003) relative to the number of enrolled (with a 5-year shift) is around 17 percent, indicating a slow progress through HE studies. Although university-type studies can be completed within 4-5 years, the average length of study is therefore markedly above this. Along with the efficiency and cost consideration emerging from such a situation, another concern is the relatively late age at which students enter the labour market. For some other European countries, Brunello and Winter-Ebmer (2002) report that the median graduation age in 1998 for 4-year HE degrees was 27.4 in Austria and 26.8 in Italy. This longer than necessary time to complete may affect available resources and defer the entry from HE to the labour market, reducing the labour supply and reducing tax revenues (Hakkinen and Uusitalo, 2003). This brings forth the need to examine not only the countrys educational structure but its specific features that influence student attainment. The former communist regime encouraged high levels of access to education, especially at secondary level. The decision to pursue post-secondary education was individual. According to the OECD (2001) estimates about one-third of general secondary school leavers did not continue their education. The choice of secondary school is closely linked with students future choice of study. However, a poor performing student is faced with problems in enrolling in tertiary level education due to a rigid structure of tertiary sector based on high student demand for enrolment and a competitive entrance exam procedure with a limited number of free (i.e. no tuition fee) places. The Faculties grade and administer entrance examination and along with these students need to fulfil some other general requirements. Entry is highly selective, there is a limited number of places, hence students need to obtain a good score in the entrance examination and have good overall high school grades (with a smaller weight given to their high school diploma grade). For students who did not achieve a high enough score at entrance examination to the HEI and cannot be admitted to government-sponsored places (i.e. zero tuition), they might still enrol at the HEI outside the official quota but they need to cover the costs of tuition fees. For those students there is a strong correlation between academic attainment and socioeconomic background (World Bank, 2000b). In terms of educational practices in Croatian HE before the implementation of the Bologna process guidelines, traditional (passive) lectures dominate and tutorials, team work and group activities are not common practice. There appears to be relatively little contact between students and professors. Students receive a modest amount of information about courses, course requirements, possible combinations of courses, and there is no career guidance provided before enrolments or during the course of study. In such a setting student effort and peer effects are variables that might have a crucial impact on student attainment since the student needs to rely more on his/her own effort and ability and the ability of his/her peers who may present a better source of information and guidance. At a later stage when more new data becomes available this type of investigation into the determinants of student attainment might also serve for an assessment of pre and post-Bologna effects on attainment in Croatian HE. When comparing Central and Eastern European higher education systems with those of the West (from which most of the empirical work on modelling student attainment is available), one is confronted with Western higher education systems that are quite heterogeneous and themselves experiencing rapid change as Scott (2002) notes. Hence, there is the need to develop an appropriate model of attainment specifically tailored to capture country-specific features in HE. 6. DEVELOPING A MODEL OF STUDENT ATTAINMENT This section presents an educational production function approach to test the significance of a complex set of factors on students attainment in Croatian HE. The novelty of this student attainment model is the inputs it uses, and their relevance in the Croatian HE context. Assuming a simple principal-agent framework, the student is a decision making unit choosing his effort level to maximise his attainment. The principal is the HEI setting appropriate incentives to achieve the main goal of maximising educational outcomes. This is a simplified version since there is usually a multiprincipal system in HE i.e. HEIs are agents of the government. 6.1. Data For student attainment modelling we used a large dataset on the entire student population on 4-year programmes for one large Croatian HEI in the period from 1994-2003. The dataset is cross sectional in character and contains a large number of variables for an even larger number of students. It was difficult to collect this data, but once obtained, it offers a multitude of analytic possibilities. For the first part of our analysis we focused on the determinants of educational attainment at the first year of study, thus obtaining insight into the relationship between the variables. This also means that a larger data set is presently available for estimation. Therefore, from the whole dataset (N=4140) a subpopulation was chosen for student attainment modelling (n=3856) i.e. only students who had completed their first year courses and were graded on them (GPA1) were selected. In cases when the last change in student status was entered prior to 2002 we assumed that the student dropped out since no activity was recorded for him/her in two years. Therefore, if a student dropped/intermitted and does not have all first year exams completed, he/she is excluded from investigation. From the whole sample of students (N=4140), 25.34 percent have completed their 4-year studies and 20.51 percent have dropped out. From here on, we will be just discussing the subpopulation under investigation (n=3856). For these ten cohorts we will be using a set of variables capturing students personal characteristics, previous schooling, socio-economic background, peer effects, student effort and several other course related characteristics. The variable descriptions are presented in Table 1 and descriptive statistics are presented in Table 2. Table 1: Variable Descriptions VariableDescriptionPersonal characteristicsAgeAge of the student at enrolmentGender1 if female; 0 otherwiseMarried1 if the student was married; 0 otherwiseBorn in Croatia1 if the student was born in Croatia; 0 otherwiseBorn in Bosnia and Herzegovina1 if the student was born in Bosnia and Herzegovina; 0 otherwiseUrban-rural1 if student is from an urban place of living; 0 otherwisePrevious schoolingSecondary school typeGymnasium1 if the student attended a gymnasium; 0 otherwiseVocational (omitted)1 if the student attended a vocational school; 0 otherwiseTechnical_other1 if the student attended a technical or other school; 0 otherwiseStudied economics 1 if the student attended a secondary school offering programmes in economics; 0 otherwiseStudied economics, commerce or tourism1 if the student attended a secondary school offering programmes in economics, commerce or tourism; 0 otherwiseSocio-economic backgroundParental educational attainmentUniversity-type degree or non-university college degree1 if the students parents obtained a university or non-university college degree; 0 otherwiseSecondary school education (omitted)1 if the students parents completed secondary education; 0 otherwiseBasic school education or no school completed1 if the students parents completed basic school education or have no basic school completed; 0 otherwiseEffortExam taking practiceAverage number of times student took exams during the first and/or second year of studyPeer effectsPeersCalculated as the mean GPA1 of students at the same course group as student iPeers squaredSquare of the average abilityPeers femaleThe proportion of females at the same course as student iCourse characteristicsFee status1 if the student is paying the fee; 0 otherwiseFull or part-time1 if the student is enrolled full-time; 0 otherwise Table 2: Descriptive Statistics VariableObservationsMeanStd. dev.GPA138563.0080.589Personal characteristicsAge385619.4392.362Gender38560.6180.486Urban38550.9580.200Born in Croatia38550.8710.335Born in B&H38550.1010.302Married38450.0260.160Secondary schoolGymnasium37770.5340.499Technical or other37770.0600.238Vocational37770.4060.491Studied economics 37760.2970.457Parental qualificationsFather_University-type degree or non-university college degree32560.4250.494Father_Secondary school education32560.4960.500Father_Basic school education or no school completed32560.0780.268Mother_University-type degree or non-university college degree32960.3070.462Mother_Secondary school education32960.5590.497Mother_Basic school education or no school completed32980.1320.339Course characteristicsFee status38560.7090.454Full or part-time38560.6110.488Peer effectsPeers138563.0070.233Peers1_sq38569.1001.430Peers1_female38560.6140.051EffortEffort121022.1790.856 6.2. Data Limitations From Table 2, it is evident that there will be some drawbacks for the empirical work. This primarily relates to a large incidence of missing observations for some of the explanatory variables, especially for our effort variable and parental education. Before deciding on the strategy in the empirical part of investigation we should understand the reasons data are missing e.g. is it because some questions in the initial questionnaire were not appropriate for the respondents, or were some subjects sensitive about some topics and did not answer, were there problems with data collection? For the Croatian HEI some of the reasons for missing observations might be that some student information was not entered in the database at all by the administrative staff (e.g. marital status, secondary school, number of times student took an exam, etc.) although the data existed. The reason for this might be the lack of time and other pending obligations of the staff, therefore, this would point out to a random pattern of missingness. Furthermore, it might be the case that students did not respond to some of the initial questions in the questionnaire (e.g. regarding their parents' education) and this data was never collected. The theory on imputation reveals three main concerns when dealing with missing data, namely the loss of efficiency, problems in data handling and analysis and, finally, bias due to differences between the observed and unobserved values (summarised in Barnard and Meng, 1999). In general, if we considered data imputation it requires caution since it changes the variance-covariance structure of the dataset and has an impact on estimated regression coefficients. These coefficients will depend on the degree of missing observations, the difference between the characteristics of two populations (i.e. observable and unobservable) and the explanatory variables used in imputation. Little and Rubin (1987) report the cases where standard estimators applied to actual and imputed data had substantial differences and imputation methods might not lead to a reduction of bias in comparison to the incomplete dataset. For this analysis we shall assume that the data is missing completely at random (MCAR) i.e. that the distributions of missing and observed data are indistinguishable. The data is MCAR when the probability that an observation (Xi) is missing is unrelated to the value of Xi or to the value of any other variable in the dataset. As Rubin and Levin (1987) warn, this assumption does not mean that the pattern of missingness is itself random, but rather that missingness does not depend on the data values. In our empirical investigation the variable with most missing observations is effort i.e. the number of times student took an exam. The reporting of this number is done automatically through the HEI's computer system or individually by administrative staff and from this we cannot deduce that the missing information is related to the student or his/her characteristics. Therefore since the missingness mechanism cannot be completely linked with the student and his/her characteristics (i.e. that are actually variables in the student attainment model) we might assume that the effort data is MCAR. If the data are MCAR then the reduced data set resembles a randomly selected sub-sample of the original data and the conclusions made from it are valid. In this case, case deletion might be a suitable approach. Consequently, we are focusing only on the complete observations and after the case wise deletion of missing values the sample contains information on 1530 students. 6.3. Empirical Specification In the empirical work we will be using a set of variables capturing students personal characteristics, previous schooling, socio-economic background, peer effects, effort and several other course related characteristics. Similarly to equation (1) we model student attainment using the following theory-driven specification: A = f (X, F, S, P, E, C)(2) where student attainment is a function of students personal characteristics (X), family background (F), previous schooling (S), peer effects (P), effort (E), and course characteristics (C). For the first part of our analysis we focused on the determinants of educational attainment at the first year of study, thus obtaining insight into the relationship between the variables. This also means that a larger data set is presently available for estimation. Building on equation (2) our model of educational attainment for the first year of study has the following form:  EMBED Equation.3  (3) where student attainment is measured by the grade point average (GPA) at year one for student i in group j, taking the value from 2.00 (lowest) to 5.00 (highest). The 2003 entry cohort was chosen as the last one due to the fact that the students cannot enrol in their third year of study (i.e. 2005/2006 for the last 2003 entry cohort) without completing all the classes from the first year. Hence, for all the students in our sample, GPA at year one should be reported unless they are intermitting or dropped out before taking any exams. The groups at the first year of studies are determined alphabetically, and the lectures are separated in two groups in which students interact, therefore the construction of peer effects had to take this into account. Xi is a vector of student's personal characteristics and includes age at enrolment, gender, marital status, place and country of birth. Fi is a vector of family characteristics which includes parents labour market qualifications and this variable serves as a proxy for socio-economic background. Si is a vector of previous schooling characteristics i.e. it introduces information on the type of secondary school that student attended and if the student studied subjects that were related to his/her present subject area. There are also several dummy variables indicating course characteristics (Ci) i.e. is student enrolled full-time or part-time, paying tuition fees or exempt and programme name. The academic peer effects (P) are captured by the mean ability of students in the same group as student i, where this ability is proxied by the obtained GPA at first year. To allow for the non-linear nature of academic peer effects we include a squared measure of academic peer effects variable (P2). Peer effects might also be influenced by gender hence we include the proportion of female students in group j (Pf). Ei is student is effort proxied by the exam taking practice i.e. the average number of times that student i took exams during his course of study (taking the value from 1-8). The model is specified using Ordinary Least Squares where this estimation method is common in modelling student attainment (Todd and Wolpin, 2003). 6.4. Characteristics of the Variables The variable on parental qualifications is a proxy for familys socioeconomic status which might reflect the learning environment at home(Hanushek, 1986) and might also be a proxy for ability in nature and quality of student decision-making. Contrastingly, in some of the empirical work, direct influence of parental variables on student attainment was found to have a small or insignificant effect on student attainment (e.g. in Mayer, 1997). Nevertheless, the exact causal structure of the relationship cannot be specified i.e. the effect of the home learning environment on student attainment is not modelled due to data limitations which are typical for this line of inquiry (e.g. Manski, 1993). In relation to the type of secondary school attended and its effect on student attainment we expect that student from gymnasiums and 4-year technical school will probably outperform students who attended some other secondary school. The rationale for this comes from OECD (2001) data on the success of Croatian students in gymnasiums and 4-year technical schools which is found superior to the one in vocational schools (e.g. there are 28 percent of students receiving an excellent mark and 42 percent receiving very good marks). Four-year technical programmes are most popular with more than 41 percent of pupils enrolled in them. Since the structure of HE in Croatia, presented in the previous sections, seems to place high demand on students ability and motivation for learning, we argue that student effort and peer effects might have a crucial impact on student attainment. In such a setting, student is modelled as a decision making unit, managing and organising his/her own studies. Therefore, student is a key actor in shaping educational outcomes. This kind of approach, although simple, is not dominant in the empirical work on attainment. One reason is the generally scarce work addressing peer effects and effort in HE and issues with data availability to capture these effects. This argument is especially relevant for the effort variable since in the overwhelming majority of studies on educational production the authors do not even acknowledge the lack of a quantifiable measure of effort. Other reason might be the relative complexity of modelling peer effects in HE where it is methodologically challenging to distinguish between the impact of students peers and students own characteristics, hence there are significant issues with estimation bias and over or under estimation of peer effects (Hanushek et al. 2003). These issues are addressed next. 6.4.1. Effort and Peer Effects In examining student effort, Bishop (1996) claims that effort will be maximised in an educational system that is setting a clear link between what is learnt and what is graded. It can be argued that the overall grade (GPA) is an internally set standard that might affect students effort levels since the grade is then a signal to the labour market of students specific competencies. Internal examinations also serve to clarify the goals of an educational programme, encouraging internal efficiency necessary in Croatian HE. The evidence on internal examinations and their impact on achievement is vastly assessed for lower levels of education however it is rarely put in this framework (i.e. effort proxied by exams is affecting internal efficiency) for tertiary education. In Croatia, until 2005/06 school year, student could retake the same exam three times after which they need to have an oral examination by an appointed HEI committee. If the student fails even at that time, he/she needed to enrol into the course once more, thus prolonging his/her duration of studies. The data on the average number of times that student retook exams is available in our data set and we use it as a proxy for effort. This exam taking practice is viewed as the formative assessment of students commitment to progress in HE. We argue that students who are passing their exams with less retakes are the ones achieving higher GPA. In developing and post transitional countries, data availability and measures of peer effects and outcome seriously affect the methodology that can be used. There are difficulties in defining the peer-group, isolating causal peer-group effects from other influences, lack of appropriate data, and different identification methodologies adopted by researchers. As summarised by Manski (2000) and Moffit (2001), the empirical analysis of social interactions is plagued by conceptual and data problems. Overall, three problems are generally recognized in modelling the impact of peer effects on student attainment. Primarily, peer group characteristics might embody some of the other omitted or mismeasured factors that influence student attainment thus producing biased estimates of peer effects on attainment. Secondly, there is a reflection problem (Manski, 1993) i.e. student influences his peers and is at the same time influenced by them. Thirdly, students choose where to enrol and they chose their friends thus, it is possible that the group chosen and students own characteristics will be highly correlated. In this section we mostly focus on solving the first two problems which might influence our results the most, as they are directly related to the problem of biased estimates of peer effects i.e. making peer effects look important when they may not be. Furthermore, since in Croatia secondary school grades are school-specific (schools assess and administer exams) to avoid the problem of different grading practices across schools we will capture peer effects by using average grade scores at the HEI i.e. GPA at year one. Drawing on Hoxby (2000), we are using a standard baseline model of the impact of peer effects on educational outcome, where a simplified version has the following form:  EMBED Equation.3 (4) where GPAij is educational outcome for student i in group j, average GPAij is the mean ability of students in the same group as student i, where this ability is proxied by the GPA obtained at first year (capturing absolute peer effects), and Xij is a vector of other factors that affect student is outcome. However, since peer effects tend to influence student attainment through multiple channels we extend the baseline model and introduce two additional variables capturing peer effects. Drawing on the work by Johnes and McNabb (2004) and also other US and UK studies in which gender had been found to play a significant role in student attainment (e.g. in Rudd 1984, Smith and Naylor 2001) we included an additional measure of peer characteristics, namely gender composition of the peer group. Since peer effects will be modelled for students in HE and from diverse backgrounds and places of living we might disregard the impact of neighbourhood characteristics that are important in primary and secondary education and are known to bias upward the estimated peer effects in school (Levine and Painter, 2000; Krauth, 2004). Furthermore, many issues related to peer effects require a model that is either non-linear in peers mean achievement or in which other moments of the peer distribution matter. To address the issue of non-linearities in peer effects (recognized by Hoel et al., 2005; Winston and Zimmerman, 2004), we included a quadratic term of our peer effect proxy as noted in equation (3). 7. RESULTS Before examining the preliminary findings of the empirical analysis, several sources of bias can be noted. Firstly, it is common across the relevant literature in education production modelling to consider only the students who enrolled, thus leaving out all students who might have attempted to enrol or did not obtain necessary qualifications. Thus, the prevailing empirical work on student attainment is conditional on students who enrolled. This is also the case here and the analysis of the application and admission process which might have an effect on student attainment is left out. Secondly, the issue of missing data has been recognized at the start of empirical analysis. Under the assumption that the data is missing completely at random, i.e. there is no pattern in the missingness of data, we deleted incomplete observations from the dataset, and proceeded with the estimation. At the end, there were still more than 1500 observations left enabling us to examine the relationship between the variables. Furthermore, the peer effects variable might introduce some endogeneity bias in the estimation of our model. However, drawing on the work by Johnes and McNabb (2004) the peer effects are captured by several variables thus potentially reducing the problem. After trying several different specifications for the model some of the dummy variables were left out since they were repeatedly insignificant and with a low value of coefficients (i.e. female peer effects, country of birth). To avoid multicollinearity in the model, in the case of variables on the type of secondary school attended, we left out the category of vocational schools as a reference group. For parental qualifications the reference group is students parents who have a secondary-level education. The diagnostic tests were performed and while normality and functional form were not an issue for the attainment model, heteroscedasticity presented and obvious problem. To address it, we used the robust correction where the new reported standard errors are more robust to failure to meet the assumption of homogeneity of variance of the residuals. Table 3 reports the results of the Ordinary Least Squares regression of student attainment at the first year of study for 1530 students. The R squared for the regression in Table 3 is 0.29 and the adjusted R squared is 0.28. F statistic (15, 1514) is 41.58 and the prob(F)= 0.000. Table 3: Regression Results, N=1530 VariableCoefficientsRobust std. err.t-statisticConstant-5.623**2.402-2.340Course characteristicsFee status-0.388***0.032-12.170Full or part time-0.225***0.032-7.040Peer effectsPeers5.635***1.6123.490Peers squared-0.813***0.270-3.020EffortExam taking practice-0.208***0.015-13.580Previous schoolingGymnasium0.068*0.0381.800Technical and other-0.0510.061-0.850Studied a related subject0.090**0.0402.240Socioeconomic statusFather University or college degree-0.0020.027-0.060Element. school education or no education0.0340.0550.620Mother University or college degree0.0050.0290.190Element. school education or no education-0.074*0.042-1.750Personal characteristicsAge-0.0110.009-1.260Married0.0930.1120.830Urban-rural0.0810.0721.130Notes: Significant at ***1%, **5% and *10% . We focus first on the results for variables on personal characteristics. The coefficient on age at enrolment, although statistically insignificant, has a negative sign and a higher t-statistic than other variables in this group (P-value is 0.16) indicating perhaps that older students perform less well in the HEI. Other dummy variables like 'urban-rural' or 'country' did not have a significant effect for students in the dataset but they both have a positive sign and the coefficient on urban-rural has a low P-value (0.19) suggesting that students from cities perform better than students from rural areas. In terms of previous schooling, the coefficients on 'gymnasium' is statistically significant at a 10 percent level, thus attending a gymnasium has a positive effect on student attainment with regard to the reference category - vocational secondary schools. More specifically, the expected GPA1 is 0.068 units higher for students who attended a gymnasium in comparison to students from a vocational secondary school. The coefficient on 'technical and other' category which mostly includes students who attended a technical school or transferred from some other HEI is negative, hence, it might be argued that attending a technical secondary school or some other tertiary level educational programme and transferring has a negative effect on student attainment in the HEI in relation to previously attending some form of vocational secondary school. Studying a subject in secondary school that is related to student's present field of study has a statistically significant and positive effect on the GPA1 in comparison to students who have not studied a similar subject. The variable on parental qualifications which is a proxy for familys socioeconomic status is statistically significant only for the category of mothers with elementary school education or no education. More precisely, the expected GPA1 is 0.074 units lower for students whose mother falls in the low or no education category in comparison to students whose mother has completed the secondary school (reference group). All the variables for course characteristics, peer effects and effort have a statistically significant impact on student attainment either at one or five percent level of significance. In the above model, paying a fee has a negative effect on student attainment and the expected GPA1 is 0.388 units lower for those students in comparison to students admitted to tuition free places. Since this variable might be considered as a proxy for student ability, its sign is justifiable i.e. high ability students are the ones who are exempt from paying the tuition fee since they are ranked high in the admissions procedure to the HEI and will probably have a higher GPA in the course of study than the lower ability, fee paying students. In such an environment, the fee paying student may not be motivated to exert effort and raise his GPA, hence a negative coefficient. In addition, being a full time student has an unexpected negative effect on educational attainment with a coefficient of -0.225 which might suggest that part-time students are actually the ones who are perhaps more motivated to increase their GPA and perform well at the HEI. One of the explanations might be that since the majority of part time students are working along with studying and, as argued previously, if the GPA is a signal to the labour market of student's specific competencies, than these students are inclined to raise their GPA (and possibly shorten their duration of studies) while anticipating an increase in wage. As expected, peer effects are found to positively influence student attainment and the GPA1 increases by 5.635 units for a unit increase in the group mean ability (i.e. peer effects) when all other explanatory variables are held constant. This result suggests that student's peers have a very strong effect on student's attainment and this finding is common in other empirical investigations, however, primarily for UK and US HE systems. The non-linear specification of the variable is also significant and negative demonstrating that peer effects are significant for students in the middle of the attainment distribution who are, as Winston and Zimmerman (2004) note, usually more susceptible to peer influence than those at either end of the attainment distribution. The variable on effort exerted at the first year of study is statistically significant at one percent level (has the highest t-value in the model) and has an expected negative sign since it is measured as the number of times student took exams. When this variable (i.e. the number of times the student took exams) increases by one unit, the GPA1 is expected to decrease by 0.208 units. This confirms our initial assumption that students who had less retakes are probably the ones achieving higher grades. 8. CONCLUSION This paper presented an education production function approach to test for the first time the significance of a complex set of factors on students attainment in Croatian HE. The novelty of the developed attainment model was the inputs it used, primarily the design of the peer effects and effort variables, and their modification to fit the Croatian context. However, several limitations need to be addressed. Firstly, the analysis was performed only for one large Croatian HEI for which the data was available and hence its results cannot be extended to the entire Croatian HE system. Secondly, there was a large number of missing observations which might have introduced bias in the results. To avoid this problem for future research, multiple imputation might be considered as a commonly applicable technique for large datasets. Finally, as mentioned in Section 7 the analysis is performed only for students who enrolled in the HEI thus leaving out information on all other students who might have attempted to enrol or did not obtain the necessary qualifications. Nevertheless, information stemming from student attainment modelling as presented in this paper may enable a better assessment of the changes in Croatian HE environment and may lead to a better understanding of the possible effects of policy changes in HE. To extend on this research, at a later stage we will introduce other variables, primarily additional measures of student ability, that have just recently become available and are expected to be helpful in understanding the determinants of student attainment. Furthermore, our goal is also to investigate the determinants of educational attainment for students who completed their studies. Those results will be compared with the results obtained in this paper thus offering more insight into the progress of studies of Croatian students. When more data becomes available, this type of research might enable a valuable comparison of Croatian HE system before and after the implementation of the Bologna process guidelines. BIBLIOGRAPHY: Akerlof, G. and Kranton, R. (2002): Identity and Schooling: Some Lessons for the Economics of Education, Journal of Economic Literature, Vol. 40, Issue 4, pp. 1167-1201. Ammermueller, A.; Heijke, H. and Woessmann, L. (2003): Schooling Quality in Eastern Europe: Educational Production during Transition, IZA Discussion Paper No. 746. Barnard, J. and Mang, X. (1999): Applications of Multiple Imputation in Medical Studies: from AIDS to NHANES, Statistical Methods in Medical Research, Vol. 8, Issue 1, pp. 17-36. Barr, N. (1998): Benefits of Education: What We Know and What We Don't Know, HM Treasury, UK (available at: http://www.hm-treasury.gov.uk/media/24A/71/252.pdf, accessed 15/11/2005). Baumol, W.; Panzar, J. and Willig, R. (1983): Contestable Markets: An Uprising in the Theory of Industry Structure: Reply, American Economic Review, American Economic Association, Vol. 73, Issue 3, pp. 491-496. Belfield, C. (2000): Economic Principles for Education: Theory and Evidence, Edward Elgar, Cheltenham, UK. Bishop, J. (1996): The Impact of Curriculum-based External Examinations on School Priorities and Student Learning, International Journal of Educational Research, Vol. 23, pp. 653-752. Blundell, R.; Dearden, L.; Goodman, A. and Reed, H. (1997): Higher Education, Employment and Earnings in Britain, Institute for Fiscal Studies, London. Bonesrnning, H. (2004): The Determinants of Parental Effort in Education Production: Do Parents Respond to Changes in Class Size? Economics of Education Review, Vol. 23, No.1, pp. 1-9. Brunello, G. and Checchi, D. (2003): School Quality and Family Background in Italy, IZA Discussion Paper No. 705, Institute for the Study of Labour, Bonn. Brunello, G. and Winter-Ebmer, R. (2002): Why Do Students Expect to Stay Longer in College? Evidence from Europe, IZA Discussion Papers No. 658. Carnoy, M. (Ed.) (1995): International Encyclopedia of Economics of Education, Pergamon Press, UK. Checchi, D. and Zollino, F. (2001): Sistema Scolastico e Selezione Sociale in Italia, Rivistadi Politica Economica, Vol. 91, Issue 7-8, pp. 43-84. Coleman, J.; Campbell, E., Hobson, C.; Mcpartland, J.;Mood, A.; Weinfeld, F. and Yorke, R. (1966): Equality of Educational Opportunity, United States Government Printing Office, Washington, DC. Costrell, R. (1994): A Simple Model of Educational Standards, American Economic Review, Vol. 84, Issue 4, pp. 956-71. Costrell, R. (1997): Can Centralized Educational Standards Raise Welfare?, Journal of Public Economics, Vol. 65, pp. 271-293. De Fraja, G.; Oliveira, T. and Zanchi, L. (2005): Must Try Harder: Evaluating the Role of Effort in Educational Attainment, Centre for Economic Policy Research (CEPR), Discussion Paper No. 5048, London. Dixit, A. (2002): Incentives and Organizations in the Public Sector: An Interpretative Review, Journal of Human Resources, 37(4): 696-727. Feinstein, L. and Symons, J. (1999): Attainment in Secondary School, Oxford Economic Papers, Vol. 51, Issue 2, pp. 300-321. Galizzi, M. and Lang, K. (1998): Relative Wages, Wage Growth, and Quit Behavior, Journal of Labor Economics, Vol. 16, pp. 367391. Hkkinen, I. and Uusitalo, R. (2003): The Effect of a Student Aid Reform on Graduation: A Duration Analysis, Department of Economics, Uppsala University, Working paper No. 2003:8. Hkkinen, I.; Kirjavainen, T. and Uusitalo, R. (2003): School Resources and Student Achievement Revisited: New Evidence from Panel Dana, Economics of Education Review, Vol. 22, Issue 3, pp. 329-335. Hanushek, E. (1986): The Economics of Schooling: Production and Efficiency in Public Schools, Journal of Economic Literature, Vol. 24, Issue 3, pp. 1141-77. Hanushek, E.; Kain, J.; Markman, J. and Rivkin, S. (2003): Does Peer Ability Affect Student Achievement?, Journal of Applied Econometrics, Vol. 18, Issue 5, pp. 527-544. Hazans, M.; Rastrigina, O. and Trapeznikova, I. (2003): Family Background and Schooling Outcomes before and During the Transition: Evidence from the Baltic Countries, Baltic International Centre for Economic Policy Studies - (BICEPS) Publications, Latvia (available from: http://www.biceps.org/publications.html). Heider, F. (1958): The Psychology of Interpersonal Relations, Wiley, New York. Henderson, V.; Mieszkowski, P. and Sauvegeau, Y. (1976): Peer Group Effects and Educational Production Functions, Economic Council of Canada, Ottawa. Holstrom, B. and Tirole, J. (1989): The Theory of the Firm, in Schmalensee, R. and Willig, R. (Eds.), Handbook of Industrial Organisation, Vol. 1, pp. 61-133, North-Holland, Amsterdam. Hoxby, C. (2000): Peer Effects in the Classroom: Learning from Gender and Race Variation, NBER Working Paper, No. 7867. Ichino, A. and Maggi, G. (2000): Work Environment and Individual Background: Explaining Regional Shirking Differentials in a Large Italian Firm, Quarterly Journal of Economics, Vol. 115, pp. 10571090. Ichino, A. and R. Riphahan (2004): The Effect of Employment Protection on Worker Effort: A Comparison of Absenteeism during and after Probation, Journal of the European Economic Association, Vol. 3, pp.120-143. Johnes, G. and McNabb, R. (2004): Never Give up on the Good Times: Student Attrition in the UK, Oxford Bulletin of Economics & Statistics, Vol. 66 Issue 1. Johnes, J. and Taylor, B. (1990): Non-Completion Rates in UK Universities: A Reply, Higher Education, Vol. 19, pp. 386-90. Krauth, B. (2004): Economics Working Paper Archive, Econ WPA, Econometrics Series, No. 0408002. Levine, D. and Painter, G. (2000): Are Measured School Effects Just Sorting? Identifying Causality in the National Education Longitudinal Survey, Institute of Industrial Relations Working Paper Series No. 078 00, University of California. Little, R. and Rubin, D. (1987): Statistical Analysis with Missing Data, John Wiley and Sons. Manski, C. (1993): Identification of Endogenous Social Effects: The Reflection Problem, Review of Economic Studies, Vol. 60. pp. 531-542. Manski, C. (2000): Economic Analysis of Social Interactions, Journal of Economic Perspectives, Vol. 14, Issue 3, pp. 115-136. Mayer, S. (1997): What Money Can't Buy: Family Income and Children's Life Chances, Harvard University Press, Cambridge, MA. McEwan, P. (2003): Peer Effects on Student Achievement: Evidence from Chile, Economics of Education Review, Vol. 22, Issue 2, pp.131-41. McNabb, R.; Pal, S. and Sloane, P. (2002): Gender Differences in Educational Attainment: The Case of University Students in England and Wales, Economica, Vol. 69, pp. 481-503. Moffit, R. (2001): Policy Interventions, Low-level Equilibria and Social Interactions, In Social Dynamics, edited by S. Durlauf and P. Young, MIT Press, Cambridge, MA. OECD (2001): Thematic Review of National Policies for Education: Croatia, Paris, France. OECD (2004): Education at a Glance: OECD Indicators, Paris, France. Robertson, D. and Symons, J. (2003): Do Peer Groups Matter? Peer Group versus Schooling Effects on Academic Attainment, Economica, Vol. 70, Issue 277, pp. 31-53. Robst, J.; Keil, J. and Russo, D. (1998): The Effect of Gender Composition of Faculty on Student Retention, Economics of Education Review, Vol. 17, pp. 429439. Rotschild, M. and White, L. (1995): The Analytics of Pricing in Higher Education and Other Services in Which Customers are Inputs, Journal of Political Economy, Vol. 103, Issue 3, pp. 573-586. Rudd, E. (1984): A Comparison Bbetween the Results Achieved by Women and Men Studying for First Degrees in British Universities, Studies in Higher Education, Vol. 9, pp. 4757. Sacerdote, B. (2001): Peer Effects with Random Assignment: Results from Darmouth Roommates, Quarterly Journal of Economics, Vol. 116, Issue 2, pp. 681-704. Schneeweis, N. and Winter-Ebmer, R. (2005): Peer Effects in Austrian Schools, Economics Series 170, Institute for Advanced Studies. Schenk, D. (2003): Try Harder? Motivational Effects of Effort Attributional Feedback, ERIC Digest ED 479353, ERIC Counseling and Student Services Clearinghouse. Scott, P. (2002) Reflections on the Reform of Higher Education in Central and Eastern Europe, Higher Education in Europe, Vol. 27, Nos. 1-2, pp. 137-152. Shapiro, C. and Stiglitz, J. (1984): Equilibrium Unemployment as a Worker Discipline Device, American Economic Review, Vol. 74, pp. 433-444. Sheehan, J. (1973): The Economics of Education, George Allen and Unwin Ltd., London. Smith J. and Naylor, R. (2001): Dropping Out of University: A Statistical Analysis of the Probability of Withdrawal of UK University Students, Journal of the Royal Statistical Society (Series A), Vol. 164, pp. 389-495. Smith, J. and Naylor, R. (2004): Determinants of Educational Success in Higher Education (chapter 11) in Johnes G. and Johnes J., Editors (2004): International Handbook on the Economics of Education, Edward Elgar, Cheltenham, UK. `oai, V. (2004): Does It Pay to Invest in Education in Croatia? Return to Human Capital Investment as a Factor in Human Resource Competitiveness, chapter in The Competitiveness of Croatian Labour Force, Institute for Public Finance, Zagreb. Todd, P. and Wolpin, K. (2003): On the Specification and Estimation of the Production function for Cognitive Achievement, Economic Journal, Vol. 113, Issue 485, pp. f3-f33. Varian, H. (1999): Intermediate Microeconomics: A Modern Approach, 5th Edition, W.W. Norton, New York. Weiner, B. (1979): A Theory of Motivation for some Classroom Experiences, Journal of Educational Psychology, Vol. 71, pp. 3-25. Winston, G. (1999): Subsidies, Hierarchy and Peers: The Awkward Economics of Higher Education: Journal of Economic Perspectives, Vol. 13 Issue 1. Winston, G. and Zimmerman, D. (2004): Peer Effects in Higher Education, prepared for C. Hoxby ed.: College Decisions: How Students Actually Make Them and How They Could, University of Chicago Press, forthcoming. Woessmann, L. (2004): How Equal Are Educational Opportunities? Family Background and Student Achievement in Europe and the United States, CESifo Working Paper Series CESifo Working Paper No. 1162, CESifo GmbH. Wolter, S. (2003): Sibling Rivalry: A Six Country Comparison, IZA Discussion Paper, No. 734, Institute for the Study of Labour, Bonn. World Bank (2000a): Higher Education in Developing Countries: Peril and Promise, Report of the Task Force on Higher Education and Society, World Bank, Washington, D.C. (available from: http://www.tfhe.net/report/downloads/report/whole.pdf, accessed 10/11/2005). World Bank Staff (CB) (2000b): Hidden Challenges to Education Systems in Transition Economies, World Bank, Washington, USA, (available from: http://site.ebrary.com/lib /staffordshire/Doc?id=5007387&ppg=70, accessed 1/11/2005) Zimmer, R. and Toma, E. (2000): Peer Effects in Private and Public Schools across Countries, Journal of Policy Analysis and Management, Vol. 19, Issue 1, pp. 75-92. Zimmerman, D. (2003): Peer Effects in Higher Education: Evidence from a Natural Experiment, Williams Project on the Economics of Higher Education, Review of Economics and Statistics, Vol. 85, Issue 1, pp. 9-23.  Theoretical considerations of effort in psychology can be found in Heider (1958) and its application to an achievement context can be found in Weiner (1979).  PISA Programme for International Student Assessment.  TIMSS Third International Mathematics and Science Study.  Gross enrolment rate in tertiary education is the sum of all tertiary level students enrolled at the start of the school year expressed as the percentage of the mid-year population of the five year age group after the official end of secondary school leaving age i.e. 18-22 year group for Croatia.  The sorting was based on the list of towns and cities published online by Wikipedia at http://en.wikipedia.org/wiki/List_of_cities, accessed 23/11/05.  Filled in by the student at the time of his/her enrolment at the HEI.  The deletion of observations with missing values is a default in most statistical packages. In this case, we used Stata software which deleted all cases with missing values on any of the variables in our analysis. 26L N   & ' ( 4 5 z { | ~ ɼּzkaYTY h;N]h$0h;N]hs0JmH sH hrNhCJaJmH sH hrNhB*mH phsH  hrNh6B*mH phsH hrNhn6mH sH hrNhmH sH hrNhnmH sH hrNh0JmH sH hrNhn0JmH sH hrNh5mH sH hrN5CJaJmH sH hrNhn5CJaJmH sH N ' ( ) * { | } ~    $7$8$H$a$gdn$a$gdn$a$gdgdn2 & Fgds$a$$a$$a$gdnLBaBdB Q f 4 R U V BGHghiyzUqҿwwwogohYxmH sH h~ZmH sH hYxh?}mH sH hYxhnmH sH hrNhmH sH hrNhB*mH phsH hmH sH h?}mH sH h>mmH sH hrNhnmH sH h;NmH sH hrNh;NmH sH  h;N]h$0h;N]h;N]mH sH h>mh;N]mH sH (   !tu6:!!!¸xppeeeZeh)Fbh^ mH sH hrNh^ mH sH h^ mH sH hbmH sH hCEmH sH h%h%mH sH h%hnmH sH hrNhg?0JmH sH hrNhn0JmH sH ha0JmH sH hrNh>mmH sH h>mmH sH h)FbmH sH hz3mH sH h;NmH sH h}mH sH hrNhnmH sH ! !!& &?&@&"'#'<'@' $$Ifa$gd^ $$Ifa$gdn$a$gd^ $a$gdngdn2 & Fgda!!!!$$ & &>&?&$')'I'J'K'b'c''''''''((e(f(((******++!-"-嚎嚀xph[mH sH hmH sH hCEhn6H*mH sH hrNhnH*mH sH hCEhn6mH sH hrNhn6]mH sH hrNhn5mH sH hrNhg?5mH sH ha5mH sH hB mH sH h^ mH sH hrNhnmH sH hrNhnB*mH phsH '@'A'B'C'a*b*"-#-/-0-0066668 $7$8$H$a$gdn & Fgd^ $a$gdnSkd$$Ifl0" 8 t644 la"-#-$-%-&-.-/-/0000112233555555666666647F7Y7Z7v::!<)<;<<<>>umh=tmH sH hshnmH sH h?~7mH sH hrmH sH !jhrNhn0JUmH sH hrNhnB*mH phsH hmH sH h^ mH sH hrNhnmH sH h^ hn5mH sH h^ 5mH sH h^ h^ 5mH sH hs5mH sH hrNhmH sH *88;;\?]???CCEEJJ L L#O$OSSZW[WZ[[$a$gd=t $7$8$H$a$gdUP$a$gdn2 & Fgda $7$8$H$a$gdn>>>> ?Z?[?\?]?`???? @'@a@c@AAAAAACCEEGGHIdd5hrNhn0J&B*CJOJQJ^JaJmH phsH  *hrNhnmH sH hrNhnB*mH phsH !jhrNhn0JUmH sH hrmH sH hrNhn5\mH sH hrNhg?5\mH sH ha5\mH sH hrNhg?mH sH hsmH sH h=tmH sH hrNhnmH sH IIIJ J4J5JJJZJ_JkJmJJJL4LFLGL\LeLgLSSTTUTZW[WXY1YQYmYnYZZ|ph`hnmH sH hsmH sH  *hrNhnmH sH hrNhUPB*mH phsH hrNhUPmH sH hUPmH sH h[mH sH h=thnmH sH hWmH sH hrmH sH hrNhnmH sH 5hrNhn0J&B*CJOJQJ^JaJmH phsH /h]kR0J&B*CJOJQJ^JaJmH phsH "ZZZ[[Sac c#ceeajbjkkopppppqDrErstttkvlvovvvɽɣɛɓɈɀɀɀɀwk_Sh)Fbhn5mH sH h)Fbhg?5mH sH h)Fbha5mH sH h=t5mH sH h7LmH sH hrNh7LmH sH h[lmH sH h?~7mH sH !jhrNhn0JUmH sH h?~7]mH sH hrNhn]mH sH hrNhnmH sH h=thn5mH sH h=thg?5mH sH h=tha5mH sH h=tmH sH  [__RaSajjmmppttkvlvvvww%y/y0y~ & Fgdg?$a$gdW $7$8$H$a$gd)Fb $7$8$H$a$gd7L$a$gdn $7$8$H$a$gdnvvxxx$y%y&y*y/y0yyyz{||~~~~~ !8R¶ꢮꙐ{k[hrNhn6CJaJmH sH hrNhn5CJaJmH sH h[6mH sH hrNhn6mH sH h=t6mH sH h)Fb6mH sH h)FbmH sH h\kh: mH sH h^ hn5mH sH h^ hg?5mH sH hs5mH sH hWmH sH hrNhWmH sH hrNhnmH sH hrNhg?mH sH ~~!"+78Vkd$$IflF "  t0    44 la $Ifgdn $$Ifa$gdn$a$gdn8QRSWwfkdQ$$Ifl" t044 la $$Ifa$gdnwxysggg $$Ifa$gdnkd$$IflF " t0    44 làsggg $$Ifa$gdnkd{$$IflF " t0    44 là΀π߀sggg $$Ifa$gdnkd&$$IflF " t0    44 la2ssg^g $Ifgdn $$Ifa$gdnkd$$IflF " t0    44 lastusggg $$Ifa$gdnkd|$$IflF " t0    44 laRЁ){ghك.Nʄ˄Ԅ4K0> #ʇ|sjh6mH sH hUP6mH sH hp76mH sH h.~h.~6CJaJmH sH h.~CJaJmH sH  *hrNhnCJaJmH sH hHkCJaJmH sH hrNhHkCJaJmH sH hrNhn6CJaJmH sH )jhrNhn0JCJUaJmH sH hrNhnCJaJmH sH 'сsg $$Ifa$gdnkd'$$IflF " t0    44 laсҁӁ $$Ifa$gdnfkd$$Ifl" t044 la(zmz $If^gdn $$Ifa$gdnykdQ$$Ifl0" t044 la()*?zsgZg $If^gdHk $$Ifa$gdHkkd$$IflF " t0    44 laz{|ςsgZg $If^gdn $$Ifa$gdnkd$$IflF " t0    44 laςЂт?sg^g $Ifgdn $$Ifa$gdnkd<$$IflF " t0    44 la?@Ah؃sg^g $Ifgdn $$Ifa$gdnkd$$IflF " t0    44 la؃كsg $$Ifa$gdnkd $$IflF " t0    44 la $$Ifa$gdnfkd= $$Ifl" t044 laOzmz h$If^hgdn $$Ifa$gdnykd $$Ifl0" t044 laՄsgZg h$If^hgdn $$Ifa$gdnkdQ $$IflF " t0    44 laLsgZg h$If^hgdn $$Ifa$gdnkd $$IflF " t0    44 lasg $$Ifa$gdnkd $$IflF " t0    44 la…ׅ/ $$Ifa$gdnfkdR $$Ifl" t044 la/0=sg $$Ifa$gdnkd $$IflF " t0    44 la=>?E $$Ifa$gdnfkdp$$Ifl" t044 lasggg $$Ifa$gdnkd$$IflF " t0    44 la†ÆІ sggg $$Ifa$gdnkd$$IflF " t0    44 la "sg $$Ifa$gdnkd3$$IflF " t0    44 la"#$/_ $$Ifa$gdnfkd$$Ifl" t044 la_`assggg $$Ifa$gdnkdW$$IflF " t0    44 laʇӇsgggU@@$$3&`#$/Ifa$b$gd$3&`#$/Ifb$gd $  a$gdnkd$$IflF " t0    44 la-kd$$Ifl\>u"%C %%^%\ 6`30>44 lae4p($$3&`#$/Ifa$b$gdʇ %)-036?CHINOW[`abgy}ÈԈ߈ !%(+.1GJMPTVXphrNh6CJaJmH sH hCJaJmH sH hrNhCJaJmH sH hrNh5CJaJmH sH M$$3&`#$/Ifa$b$gd$3&`#$/Ifb$gd B0$3&`#$/Ifb$gdkd$$Ifl\>u"%C %%^%\ 6`30>44 lae4p( !%*17t___$$3&`#$/Ifa$b$gdh$3&`#$/If^hb$gdtkd$$Iflu"% 6`30>44 lae4p 78?B,h$3&`#$/If^hb$gdkd$$Ifl\>u"%C %%^%\ 6`30>44 lae4p(?DJPQ-kd$$Ifl\>u"%C %%^%\ 6`30>44 lae4p($$3&`#$/Ifa$b$gdQW\bh$$3&`#$/Ifa$b$gdh$3&`#$/If^hb$gdhiyB,h$3&`#$/If^hb$gdkd$$Ifl\>u"%C %%^%\ 6`30>44 lae4p(y~-kd$$Ifl\>u"%C %%^%\ 6`30>44 lae4p($$3&`#$/Ifa$b$gd$$3&`#$/Ifa$b$gdh$3&`#$/If^hb$gdB,h$3&`#$/If^hb$gdkd$$Ifl\>u"%C %%^%\ 6`30>44 lae4p(ˆÈ-kd$$Ifl\>u"%C %%^%\ 6`30>44 lae4p($$3&`#$/Ifa$b$gdÈԈՈ߈xbMMM$$3&`#$/Ifa$b$gdh$3&`#$/If^hb$gdtkd$$Iflu"% 6`30>44 lae4p $3&`#$/Ifb$gdB,h$3&`#$/If^hb$gdkd$$Ifl\>u"%C %%^%\ 6`30>44 lae4p( -kd$$Ifl\>u"%C %%^%\ 6`30>44 lae4p($$3&`#$/Ifa$b$gd!&,2$$3&`#$/Ifa$b$gdh$3&`#$/If^hb$gd23FB,$3&`#$/If^b$gdkd$$Ifl\>u"%C %%^%\ 6`30>44 lae4p(FKQWX-kd$$Ifl\>u"$C %%^%\ 6`30>44 lae4p($$3&`#$/Ifa$b$gdXpqxbMMM$$3&`#$/Ifa$b$gdh$3&`#$/If^hb$gdtkd $$Iflu"$ 6`30>44 lae4p $3&`#$/Ifb$gd+/359;Za|ߊ /379=?SWZ]acer{|}ˋҋ܋݋hrNh6CJaJmH sH hCJaJmH sH hrNhCJaJmH sH S‰B,h$3&`#$/If^hb$gdkd!$$Ifl\>u"$C %%^%\ 6`30>44 lae4p(-kd"$$Ifl\>u"$C %%^%\ 6`30>44 lae4p($$3&`#$/Ifa$b$gd+06<$$3&`#$/Ifa$b$gdh$3&`#$/If^hb$gd<=|B,h$3&`#$/If^hb$gdkd#$$Ifl\>u"$C %%^%\ 6`30>44 lae4p(|-kd$$$Ifl\>u"$C %%^%\ 6`30>44 lae4p($$3&`#$/Ifa$b$gd$$3&`#$/Ifa$b$gdh$3&`#$/If^hb$gdŠB,h$3&`#$/If^hb$gdkd%$$Ifl\>u"$C %%^%\ 6`30>44 lae4p( -kd&$$Ifl\>u"$C %%^%\ 6`30>44 lae4p($$3&`#$/Ifa$b$gd !"#$$3&`#$/Ifa$b$gd$3&`#$/Ifb$gd#$/B)$h$3&`#$/If^ha$b$gdkd($$Ifl\>u"%C %%^%\ 6`30>44 lae4p(/4:@A-kd%)$$Ifl\>u"$C %%^%\ 6`30>44 lae4p($$3&`#$/Ifa$b$gdASX^d$$3&`#$/Ifa$b$gd$h$3&`#$/If^ha$b$gdderB-$$3&`#$/Ifa$b$gdkd;*$$Ifl\>u"$C %%^%\ 6`30>44 lae4p(rstuv-kdQ+$$Ifl\>u"$C %%^%\ 6`30>44 lae4p($$3&`#$/Ifa$b$gdv}$$3&`#$/Ifa$b$gdh$3&`#$/If^hb$gdB,h$3&`#$/If^hb$gdkdg,$$Ifl\>u"%C %%^%\ 6`30>44 lae4p(-kdw-$$Ifl\>u"%C %%^%\ 6`30>44 lae4p($$3&`#$/Ifa$b$gdċʋ$$3&`#$/Ifa$b$gdh$3&`#$/If^hb$gdʋˋҋB0$3&`#$/Ifb$gdkd.$$Ifl\>u"%C %%^%\ 6`30>44 lae4p(ҋӋԋՋ֋-kd/$$Ifl\>u"%C %%^%\ 6`30>44 lae4p($$3&`#$/Ifa$b$gd֋ދ$$3&`#$/Ifa$b$gdh$3&`#$/If^hb$gd B6... & Fgdg? $  a$gdnkd0$$Ifl\>u"%C %%^%\ 6`30>44 lae4p( NOZ^YoYݏ֐א!ʙЙٙwoodoohshbmH sH hbmH sH  *hrNhnmH sH hrNhnB*mH phsH !jhrNhn0JUmH sH hrNhnmH sH h: mH sH hshnmH sH hrNhg?mH sH h^ hn5mH sH h^ hg?5mH sH hs5mH sH h5mH sH h6mH sH " ֐אgh2PQ01 $$Ifa$gdb $$Ifa$gdn 7$8$H$gdn$a$gd=t$a$gdn & Fgdg?2348P01L{ ǻۚۊۂwkcT>+j'lG hrNhnCJUVaJmH sH jhrNhnUmH sH h/]mH sH hrNhn6mH sH hrNhbmH sH hnmH sH hDP)mH sH hUPmH sH (hrNhnCJOJQJ^JaJmH sH h=thn5mH sH h=thg?5mH sH h=tha5mH sH h=tmH sH hrNhnmH sH hbmH sH !jhrNhb0JUmH sH  $$Ifa$gdb $$Ifa$gdn$a$gdngdnSkd1$$Ifl0" 8 t644 la UYrs}~ '4567_`a./_dghiuvҥӥȺԲȺԲȺȺȤȺԜ|h^ hg?5mH sH ha5mH sH hrNho@mH sH hg?mH sH hrNhn6H*mH sH hbmH sH hrNhn6H*mH sH hrNhn6mH sH hrNhnmH sH jhrNhnUmH sH !jO2hrNhnEHUmH sH .01զ֦  & Fgdg? $7$8$H$a$gdn$a$gdnWkdw5$$IfTl0" 8 t644 laTզ֦+jnoptz߭ƸƸ׸׸藌yqyyiZyhrNhnCJaJmH sH h #%mH sH h DmH sH hrNhg?mH sH hamH sH hrNh=tmH sH h=tmH sH hbmH sH !jhrNhn0JUmH sH hG?OJQJ^JmH sH  hrNhG?OJQJ^JmH sH  hrNhnOJQJ^JmH sH hrNhnmH sH h^ hn5mH sH !ǽ˽ $$Ifa$gd^ $$Ifa$gdn $7$8$H$a$gdn$a$gdngdg?ýĽŽƽԽ׽ٽڽ VWþ%&^cϾ岤岤岤vncha5\mH sH hHmH sH hrNhnCJaJmH sH h)mH sH hrNhHmH sH hrNhnH*mH sH hrNhn6H*mH sH hrNhn6mH sH !j6hrNhnEHUmH sH +j>8G hrNhnCJUVaJmH sH hrNhnmH sH jhrNhnUmH sH !˽̽ν12345672 & Fgda$a$gdn $7$8$H$a$gdngdnWkd8$$IfTl0" 8 t644 laTSc7L%& 01ٻ٫ΆΆ~shsssshahKsmH sH hahamH sH ho@mH sH hahHmH sH hajha0JUmH sH hBcmH sH hKsmH sH hamH sH hKshKsmH sH hHmH sH hahnmH sH hrNhnmH sH hrNhg?5\mH sH hrNhn5\mH sH (128A[\&ls2G8Q,-89=>EF6Mwxz˼{sh'\mH sH h_mH sH h_ho@mH sH hLomH sH h_hnmH sH hrNhnmH sH hrNhn6CJaJmH sH hrNhnCJaJmH sH hrNhn5CJaJmH sH ha6mH sH hrNhn6mH sH hG?mH sH hnmH sH )78\er $$Ifa$gdo@ $Ifgdn$a$gdnF=111 $$Ifa$gdo@ $Ifgdnkd;9$$IfTl\|?{ %%%%0644 lap(TF=111 $$Ifa$gdo@ $Ifgdnkd@:$$IfTl\|?{ %%%%0644 lap(TF9--- $$Ifa$gdo@ $If^gdnkdE;$$IfTl\|?{ %%%%0644 lap(T F9--- $$Ifa$gdo@ $If^gdnkd<<$$IfTl\|?{ %%%%0644 lap(T&'()F=111 $$Ifa$gdo@ $Ifgdnkd3=$$IfTl\|?{ %%%%0644 lap(T)*09?EF9--- $$Ifa$gdo@ $If^gdnkd*>$$IfTl\|?{ %%%%0644 lap(TEFT^dkF9--- $$Ifa$gdo@ $If^gdnkd!?$$IfTl\|?{ %%%%0644 lap(TklstuvF=111 $$Ifa$gdo@ $Ifgdnkd@$$IfTl\|?{ %%%%0644 lap(TvwF9--- $$Ifa$gdo@ $If^gdnkdA$$IfTl\|?{ $%%%0644 lap(TF=111 $$Ifa$gdo@ $Ifgdnkd B$$IfTl\|?{ %%%%0644 lap(TF9--- $$Ifa$gdo@ $If^gdnkdC$$IfTl\|?{ %%%%0644 lap(TF9--- $$Ifa$gdo@ $If^gdnkdC$$IfTl\|?{ %%%%0644 lap(T%+1F9--- $$Ifa$gdo@ $If^gdnkdD$$IfTl\|?{ %%%%0644 lap(T12GHIJF=111 $$Ifa$gdo@ $IfgdnkdE$$IfTl\|?{ %%%%0644 lap(TJKRoF9, h$If^hgdn $If^gdnkdF$$IfTl\|?{ %%%%0644 lap(Tov|:- h$If^hgdnkdG$$IfTl\|?{ $%%%0644 lap(T $$Ifa$gdo@:kdH$$IfTl\|?{ $%%%0644 lap(T $$Ifa$gdo@ $$Ifa$gdo@ h$If^hgdnh$If^h`gdn"*07F9--- $$Ifa$gdo@ h$If^hgdnkdI$$IfTl\|?{ $%%%0644 lap(T78QRSTF=111 $$Ifa$gdo@ $IfgdnkdJ$$IfTl\|?{ $%%%0644 lap(TTUY`fmF9--- $$Ifa$gdo@ $If^gdnkdK$$IfTl\|?{ %%%%0644 lap(Tmnv|F9--- $$Ifa$gdo@ $If^gdnkdL$$IfTl\|?{ %%%%0644 lap(TF9--- $$Ifa$gdo@ $If^gdnkdM$$IfTl\|?{ %%%%0644 lap(T F>>6>$a$gdV$a$gdnkdN$$IfTl\|?{ %%%%0644 lap(T7-LUW]bgwzHSFPQde./0@XZafwڿڿh_hVmH sH hLoOJQJ^JmH sH hVOJQJ^JmH sH  hrNh: OJQJ^JmH sH hVmH sH hDP)hDP)mH sH h DmH sH h_mH sH hDP)mH sH h'\mH sH h_hnmH sH h_hSmH sH . %(4M[]Jhju4<`|}~4]l˸˰˰˝˰˰h'\mH sH h+@mmH sH hmH sH h@zYhFmH sH hFmH sH h DmH sH h@zYhnmH sH h@zYh@zYmH sH  *hrNhnmH sH hVOJQJ^JmH sH hVmH sH hLomH sH 7 QH $7$8$H$a$gd D$ dd[$\$a$gd D$dd[$\$a$gd D$a$gdg?$a$gdS$a$gdn2 & Fgda $7$8$H$a$gdn5Us12FL&'"$%Sz  !78EZjļ̼ԼԱ걼ԎhrNhn5\mH sH ha5\mH sH h DmH sH h#\mH sH hFmH sH hFh@DmH sH h+@mmH sH h7mH sH hLomH sH hFh&smH sH hFh@zYmH sH hFhnmH sH hFhFmH sH ,8K #$<FX!'0,7Eb~"+@E͹ŀ|x|hkh> hShxUhcmH sH  hShch$mH sH h>mH sH hrNhSmH sH hSmH sH hxUmH sH h@h$hShxUhM h$0hM hnmH sH hrNhn5\mH sH hrNhg?5\mH sH /EOST^y"F "v}~l־־־ֳqbh DhnCJaJmH sH "h Dhn6CJaJhmH sH h DhnCJaJhmH sH hrNhg?0JmH sH hrNhWmH sH hnmH sH hrNh>mH sH h2vmH sH h>mH sH hemH sH h/]mH sH h@mH sH h>h$mH sH h>hsmH sH h>%3GHOP35nou}k޽p^pKpKp%h DhB B*CJaJmH phsH "h DhB 6CJ]aJmH sH h DhB CJaJmH sH h DhbCJaJmH sH h DhdCJ\aJmH sH h DhdCJaJmH sH h Dhn6CJaJmH sH h DhnCJaJmH sH "h Dhn6CJ\aJmH sH h DhnCJ\aJmH sH "h Dhn5CJ\aJmH sH k HVj $G{ %>gi͹েেwwhYhh Dh #%CJaJmH sH h DhB CJaJmH sH h Dhn;CJaJmH sH h DhnCJ]aJmH sH h DhnCJ\aJmH sH "h Dhn6CJ]aJmH sH 'h Dh?6CJaJmH nHsH tH$h Dh?CJaJmH nHsH tHh DhnCJaJmH sH h Dhn6CJaJmH sH ".%]21I-$dd7$8$H$[$\$a$gd D!$dh[$\$a$gd D$7$8$H$^`a$gd D $7$8$H$a$gd D$dd[$\$a$gd D$ dd[$\$a$gd D 2wn )H+,@ix%8@°° }h}h}(h Dhn6B*CJaJmH phsH %h DhnB*CJaJmH phsH h Dhn6CJaJmH sH h DhnCJ]aJmH sH "h Dhn6CJ]aJmH sH h DhnCJaJmH sH h Dh #%CJaJmH sH h DhB CJaJmH sH h DhB 6CJaJmH sH '-|D}XG/)b d$L7$8$H$^`La$gd D$dd7$8$H$[$\$a$gd D $7$8$H$a$gd D$dd[$\$a$gd D$ dd[$\$a$gd D@Aj|}hôxffVffDf"h Dh #%6CJ]aJmH sH h DhnCJ]aJmH sH "h Dhn6CJ]aJmH sH *h Dhb6CJ]aJmH nHsH tH$h DhbCJaJmH nHsH tH$h Dh?CJaJmH nHsH tHh DhnCJaJmH sH ,h Dhn0J&CJOJQJ^JaJmH sH #h Dhn0J1CJ\aJmH sH &h Dhn0J15CJ\aJmH sH vAJ STUer!*ܽͭ͝{l]M]h DhB 6CJaJmH sH h DhB CJaJmH sH h Dh?CJaJmH sH h Dh?CJ]aJmH sH "h Dh?6CJ]aJmH sH h Dh?CJ\aJmH sH h Dhn6CJaJmH sH h DhnCJ]aJmH sH h DhnCJaJmH sH "h Dhn6CJ]aJmH sH "h Dh #%6CJ]aJmH sH deK^  ~   F   w O$dd@&[$\$a$gd D$a$gd D$^`a$gd D$dd7$8$H$[$\$a$gd D$dd[$\$a$gd D$ dd[$\$a$gd D$7$8$H$^`a$gd D*01o. I ^   H \       }   . F Z t  ξξίξΏξ|ξξeMe/h Dh #%0J&6CJOJQJ^JaJmH sH ,h Dh #%0J&CJOJQJ^JaJmH sH %h DhnB*CJaJmH phsH h Dh)CJaJmH sH  h Dhb0J1CJaJmH sH h DhbCJaJmH sH h Dhn6CJaJmH sH h DhnCJaJmH sH "h Dhn6CJ]aJmH sH h DhnCJ]aJmH sH  + S w  z'EMOǸy^A^8h Dhn0J&6B*CJOJQJ^JaJmH phsH 5h Dhn0J&B*CJOJQJ^JaJmH phsH h DhB 6CJaJmH sH h DhB CJ\aJmH sH h DhB CJaJmH sH h Dhn6CJaJmH sH h DhnCJaJmH sH ,h Dhn0J&CJOJQJ^JaJmH sH %h Dh #%6CJ\]aJmH sH h Dh #%CJaJmH sH OcuJ0G(cAAgdb$a$gdngdn$a$gdngds$ dd[$\$a$gd D$dd[$\$a$gd D=bcBbtuɬɝ~cLc=~+"h Dhn6CJ]aJmH sH h DhB CJaJmH sH ,h DhB 0J1B*CJ\aJmH phsH 5h DhB 0J&B*CJOJQJ^JaJmH phsH h Dhn6CJaJmH sH h DhnCJaJmH sH h DhdCJaJmH sH 8h Dhd0J&6B*CJOJQJ^JaJmH phsH 5h Dhd0J&B*CJOJQJ^JaJmH phsH 5h Dh #%0J&B*CJOJQJ^JaJmH phsH Zn/0,.9X`qࠑukg_T_T_T_hhcCJaJhcCJaJhcjhc0JUhrNhmH sH  h Dh?0J1CJaJmH sH h Dh?CJaJmH sH h DhCJaJmH sH "h Dhn6CJ]aJmH sH h Dh)CJaJmH sH h Dhn6CJaJmH sH h DhnCJaJmH sH h DhnCJ]aJmH sH  q}(klrs#'(FGH#&'()@@@@ظذؖ؏ؖ؈ujhhcCJaJ!jhhc0JCJUaJU h:Zhc h ++hchUPh,5C 55^5\9/  / e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh5"#v":V l 6`30>,59/ e4p $$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh5"#v":V l 6`30>,59/ e4p $$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,,5C 55^5\9/ e4p($$If!vh5"#v":V l 6`30>,59/ e4p $$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/ e4p($$If!vh55T5:57#v#vT#v:#v7:V l 6`30>,5C 55^5\9/  e4p($$If!vh5?5(#v?#v(:V0l t06,5 58(Dd b  c $A? ?3"`?2r)1@L}O@N2`!F)1@L}O@+بBhxڥAOA7v R*"6x"$PzҒliHh$( pL̍޼#˼ d7ӷohm hC3wGΒ[kTqQa!!%fbl ^  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~      !"#$&'()*+,-./0123456789:;<=>JCFGHIKLMNOPQRSTUVWXYZ[\]^_abcderghijklmnopq`stuvwxyz{|}~Root Entry F72EData %_WordDocumentHObjectPool7272_1198324519 F7272Ole CompObjfObjInfo ! FMicrosoft Equation 3.0 DS Equation Equation.39qwl= GPA_1 ij = 0 +X i +F i +S i + 1 P i + 2 P Equation Native _1194889534 F7272Ole  CompObj fi2 + 3 Pf j + 4 E i +C i + i FMicrosoft Equation 3.0 DS Equation Equation.39q:—_  GPA ijObjInfo Equation Native 1Tablef2SummaryInformation( =+'GPA ______  ij +X ij Oh+'0 (4 P \ h t&Katedra ekonomske teorije i politike,ofateShtrikahtrhtr Normal.dotnmajam.d2jaMicrosoft Word 10.0LgS0' F`LJ||*'gKWZ+ɸ]8uR}F6w7uA|Dʳ~ĈE1u+Ww tFyz\~<'! fQ8fKӳ/@6E*fTfa[nm=i]Աv;3edsCybs@]_BY+#>{KqJ'u:BL~7GNsv"'F ?\C\5qiu?uHu:?TYB{+]oq6 ,q V i( ׏ $$If!vh5|5#v|#v:V0l t06,5 58TDd < b  c $A? ?3"`?2GS<۽7W6`!GS<۽7"`@Hb xxڝKP]xm("E,I)4Ђf AG]"Ckj^s\ @Xzb)d,T X&3TP+j dz%3a i8=ˑ<\+-"W?+b5}Wq+G="k4q_:]{9n 1Ǿ(1_]f81Op4B~vJN)@IzČ\]U<4j)f)9kg\g]X>7s|D$$If!vh5E5"#vE#v":V0l t06,5 58T$$If!vh55055<#v#v0#v#v<:V l06,55559/ p(T$$If!vh55055<#v#v0#v#v<:V l06,55559/ p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T$$If!vh55055<#v#v0#v#v<:V l06,55559p(T4Dd^  C ,AZnak mali"`RDPIbw1'4ǫ^OFVDPIbw1'4ǫJFIF,,C    $.' ",#(7),01444'9=82<.342C  2!!22222222222222222222222222222222222222222222222222" }!1AQa"q2#BR$3br %&'()*456789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz w!1AQaq"2B #3Rbr $4%&'()*56789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz ?( +4l5!ʟʫwm& Rs?Y~G,voW<{ewm& ?e;ߓƧ9(O_;m7Q /(5?YErAoɿY~Gx:+~Mwm& =OpsǹQ\,vo_;m7Qj+=βe;ߓAoɿcS_uW' /(~M{s?Y~G,voW<{e:~g2.mޙGn~PN̠)QEo(Z|+7-kR}ߨQEEPEPEPEPEPEPEPEPѿB?'Z_!-zE|'=z_ (4 (>R% Zr|FBzJ_Ï_$Fmu|/q6ںė=|((̠((((濌P?_ʼ7;> >[(7G$K^^oKD#_%H^K^EVEP_k_^P?W/_IKqGS~EVS|/q6ں>8ѿm]}|oK՞>QEfPQX,y%gos7$2dq"Ÿ0iUs{&O2tTW; ?` GftTW; ?` GftTW; ?` Gfxoo(w/^}]mJUVQ\a "mH\E}jQO# (2>%"?פWk+?ƗQEQEo(Z|+7-kR}ߨQE hͿ6N4oW__7[gQY|kJnlJW]ů))\U}GE_QZQ@Q@Q@Q@Fk+ ȄO؟K| ((K7-kQJ_ )>|o(ЃoN4oW_\'7so Izه‚(>_%7W?%+J{Q]Yhm+velD+3._CFqTDy5#.wS 3.G!(wTG CaQ[xS(3.Gpc 3.G!(wTG =;$TV!(wTG ? G ?Q!9%¢ G ?Qg?]?i.Ǹ|k+C^~ 6˚2;Օ/QXQ@)|FBz>#Bֿ!\}%/=OQZ}MIƍ\Cj_$Fmuu/V{0PQEAEPEPEPEPEPEPEPEPEP_k_^P?W/_IKqGS~EVS|/q6ں߇!l̑hEeۨ&J;A3ЗYATkYwg4Wi|UѯEdWi _=\G$ѯEdWi _=\G$ѯEW[;!l$NN=XQET2][d$oFp 7,QUg?PG){X2E29cwE"ʜ}Zi(`QE|J_ % ZrDx>7QEh@QEQEQEQE'?_ y\&?_ y_;,O^(M'ogos+ѫj|/ q]Z~h(QEo(Z|+xX6XrU :W5@栟%zMHwS6{ZE?Q@r9eֿ͢="(pc6{ZE?Q@=Ö]+Kk G#/ŽhYv3h/G_3k_g 99e{/k=ş(n`D$R|籮~ezQEbYC 5oxyufh`TY?aYٯ)4쒽c774}}ffOpo?Vdxn9"ҶȌ0 54!+aEP( ( ( ( ( ( ( ( ( ( ( (?00000000020000p000020000p00000000p200000 0 0 000000 0 0 00 0 0 0(0(0(0(0(0000020000000008080080000@00@00p00000000000X0`0p0`0`0h00h0h0h0000000000000000004@@@ NormalCJ_HaJmH sH tH X@X Heading 1$$@&a$5:CJ OJQJaJmH sH V@V Heading 2$$@&a$5CJOJQJaJmH sH P@P Heading 3$@&5CJOJQJaJmH sH DA@D Default Paragraph FontVi@V  Table Normal :V 44 la (k@(No List 4@4 Header  !4 @4 Footer  !PB@P Body Text$5$7$8$9DH$a$ aJmH$sH$R>@"R Title$5$7$8$9DH$a$56CJaJmH$sH$LP@2L Body Text 2$a$OJQJaJmH sH 6U@A6 Hyperlink >*B*phHRH  Balloon TextCJOJQJ^JaJ8Oa8 n First Heading5\>@r>  Footnote TextCJaJ@&@@ Footnote ReferenceH*HC@H Body Text Indentx^@O@ nSecond heading  & F >O> n Third heading `6Oq6 nStyle1$a$mH sH \O\ n References$dh^`a$CJaJmH sH TOT n Text boxovi  6CJOJ QJ ]mHsH|O| nStyle Heading 1 + Justified$hha$6:CJKHOJQJ\]:O: nbook daJmH sH tH N^@N n Normal (Web)!dd[$\$ mHsHtH$O!$ ngoohl3$O1$ ngoohl4$OA$ ngoohl5.)@Q. n Page NumberXOaX nmedium-normal1#56CJOJQJ\]^JaJo(B@B nTOC 1'x56\]mH sH tHLL nTOC 2(x^5CJ\aJmH sH tHB@B nTOC 3 )^CJaJmH sH tHBB nTOC 4 *^CJaJmH sH tHBB nTOC 5 +^CJaJmH sH tHBB nTOC 6 ,^CJaJmH sH tHBB nTOC 7 -^CJaJmH sH tHBB nTOC 8 .^CJaJmH sH tHBB nTOC 9 /^CJaJmH sH tHj@j n Table Grid7:V000.X@. nEmphasis6]:O": n FIRST FIRST 2 & FVo2V bDefault 37$8$H$!B*CJ_HaJmHphsHtH(88[ve B#^FIe eHyHfgh}'()*{|}~      ?@"#<@ABCa!b!"$#$/$0$''----//22\6]666::<<AA C C#F$FJJZN[NQRRVVRXSX__bbeeiikklkkkll%n/n0nssuu!u"u+u7u8uQuRuSuWuwuxuyuuuuuuuuuuvvv2vsvtvuvvvvvvvvvvv(w)w*w?wzw{w|wwwwww?x@xAxhxxxxxxyyyOyyyyyzzzLzzzzzzz/{0{={>{?{E{{{{{{{{{ | |"|#|$|/|_|`|a|s|||||||||||||}}} }!}%}*}1}7}8}?}D}J}P}Q}W}\}b}h}i}y}~}}}}}}}}}}}}}}}}}}}}}~ ~~~~!~&~,~2~3~F~K~Q~W~X~p~q~~~~~~~~~~~+06<=| !"#$/4:@ASX^derstuv}ĀʀˀҀӀԀՀրހ օׅgh2PQ0101՛֛ Dz˲̲β12345678\er &'()*09?EFT^dklstuvw%+12GHIJKRov|"*078QRSTUY`fmnv| QH.%]21I-|D}XG/)b deK^~Fw]OcuJ  0       G(cKLNQTWYZ\]^_`abcf0000000000000000000000000000000000000000000000000000000000000 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00p00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 000000000000p00000 0 0 000000 0 0 0000000000000000@0@0@0@0@0@0@0@0@0@0@0@0 @0 @0 @0@002000000000p0p0p0p00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00p00p00200@00p0p@0@0@0@0p@0p@0@0@0@0@0@0@0@0@0@0@0@0@0@0p@!0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0@0p@0p@0@0@0p@0@0@0@0@0@0@0@0@0@0@0@0@0p@0@0@0@0@0@0@0p@0@0@0@0@0@0@0@0@0@0@0@000@00@0p@0@0@0@0p@0@0@0p@00%l @0@0@0@0@0@0@0@0@0@0@00u000      ?@"#@AB0$''//2]66:<<AARRVRXSX__kkll%n/n0ns7uQuSuWuwuxuuuuvv2vsvvvvvvvv(w)w*w?wzw{w|www?xAxhxxxyyyOyyyyyzzzLzzzzzz/{0{={?{E{{{{{{{{ | |"|$|/|_|`|a|s||||||}} }!}%}*}1}7}8}P}Q}h}i}}}}}}}}}}}~~2~3~W~X~p~q~~~~~~~~~+06<=| #$/@ASX^derstuv}ĀʀˀҀՀրހׅP10֛ Dz˲̲β18 )0ET^dkv%+1Jo"*078TUY`fmnv| QH.%]21ID}XG/)b d^Fw]OcuJ  0     G(cKfO9008t O900@ 0@ 0@ 0@0@0@0@0@0Oy0 0Oy0W0@ 0Oy00Oy00Oy00Oy00Oy00Oy00Oy0(0Oy0(0Oy0g0Oy0g0Oy0(0 Oy0(0Oy010@ 0Oy0F0Oy0R0Oy0?0]Oy0R0Oy0@0@ 0Oy00@0Oy0,0Oy00Oy00Oy0,0@ 0Oy0-08`Oy0-0Oy0r0Oy0+0Oy00Oy00Oy00Oy00 Oy00 Oy0 0Oy0 0Oy00Oy00 Oy00 Oy00Oy00 Oy00 Oy0 0 Oy00 Oy00 Oy00Oy0 0Oy0 0Oy0 0Oy0 0Oy0 0Oy0 0Oy0 0Oy00Oy00Oy00Oy0 0Oy00Oy00Oy00Oy0 0Oy0 0\Oy0"0Oy0"0Oy00Oy0 0 Oy0O08Oy0.0 Oy0Q0Oy0Q0Oy0(0 Oy0.0 Oy0.0 Oy0.0 Oy00Oy020 Oy020 Oy020 Oy00Oy060 Oy0 0 Oy0f0Oy0 0 Oy0 0 Oy00Oy00Oy0m0Oy00Oy00Oy00Oy00Oy00Oy0n0Oy0o0Oy0o0Oy0n0Oy00Oy00Oy0N0Oy0N0Oy00Oy00Oy00Oy00Oy00Oy00Oy0z0Oy0 0IOy00Oy00 >Oy0 0Oy00Oy0 0Oy0 0Oy0 0Oy0 0Oy0 0Oy00Oy0 0 >Oy0"0Oy0 0 >Oy0"0Oy00 >Oy0"0Oy00 >Oy0"0Oy00 >Oy0"0Oy00 ?Oy0"0Oy00 ?Oy0"0Oy00 ?Oy0"0Oy00 ?Oy0"0Oy00Oy0"0Gy00Oy0*0 Oy00Oy008Oy0.0 Oy0.0 Oy00Oy00Oy00Oy020 Oy020 Oy00Oy00Oy00Oy00Oy00Oy008Oy0.0 Oy0.0 Oy00Oy00Oy00Oy020 Oy020 Oy0"0IOy0N0Oy0>0 ?Oy0 0Oy0N0Oy0 0Oy00Oy00Oy0 0Oy0 0Oy0 0Oy0A0 Oy0A0 Oy0A0 Oy0A0 Oy0A0 Oy00!Oy00!Oy00#Oy00t"Oy0P0 ?Oy0J0Oy00t"Oy0J0 Oy0J0 Oy00 Oy00Oy00Oy0M0IOy00Oy00Oy00Oy00Oy00 @0 Oy00IOy00Oy00IOy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy00Oy0w0Oy00Oy00Oy00Oy00Oy0c0XOy00 @0XOy00Oy00Oy00Oy00Oy00]Oy00Oy0 1Oy00Oy00Oy00Oy00Oy00Oy0c0Oy00Oy00Oy00Oy00Oy01Oy01Oy00{Oy0 0{Oy0 0 Oy0 0 Oy0 0Oy00Oy00{Oy01IOy00 Oy00 Oy00 Oy00$Oy00!Oy00{Oy00{Oy00 Oy00 Oy00 Oy0R1IOy0T1IOy0$0 Oy0$0 Oy0$0 Oy00Oy0+0{Oy011IOy0.0 Oy0.0 Oy0-0 Oy00\%Oy0*0 Oy0&0{Oy0'0 Oy020 Oy0)0 Oy0)0 Oy0'0  @0l Oy0*0Oy0*0 Oy0/0{Oy000 Oy020 Oy020 Oy020 Oy000  @0 Oy01IOy01 Oy01IOy090 Oy090 Oy080 Oy01 Oy01IOy0>0 Oy0>0 Oy0>0 Oy0#1 Oy0(1IOy0H0 Oy0H0 Oy0G0 Oy00 Oy0c0Oy0c0Oy0c0XOy01Oy018(Oy01Oy01Oy01Oy01Oy00Oy0F1+tOy00Oy0&0Oy0&0Oy0&0Oy0M1Oy0-0Oy0-0Oy0e0@0@0POy0$0 Oy00 #@0P@0P@0Oy0l0@0POy00Oy0 0@0Oy00Oy0,0Oy00Oy010Oy010Oy0H0Oy00@0Oy0 0h Oy0 1Oy00Oy00Oy01Oy0t1/tOy00Oy0u1Oy0;0Oy0,0Oy00Oy00Oy00Oy0&0Oy0-0Oy00&My0t10tMy0}1My0u1@0Oy01 @0@0P@0X@0X@0XOy00Oy0 0My0*0My0%0@0 @0 @0P Oy00Oy00$ Oy00Oy00 @0 @0Oy00@0@0 0   !"->IZvRʇ 1Ek@* q@eB  " @'8[~8ẁsс(zς?؃/= "_ 7?QhyÈ2FX<| #/Adrvʋҋ֋ ˽7)Ekv1Jo7Tm -dOAeB   #dB òŲe::L   %@H  0(  0(  B S  ?e _Toc124584186 _Toc124584187 _Toc124584188 _Toc124584189 _Toc124584190 _Toc121167343 _Toc124584191 _Toc121167344 _Toc124584192 _Toc121167345 _Toc124584193 _Toc121167346 _Toc124584194 _Toc121167347 _Toc124584195 _Toc121167348 _Toc124584196 _Toc118781525 _Toc121167295 _Toc121167350 _Toc124584198 ]6Qkk%n%n33wwwwf  >6Rk.n.nOOԛԛ\\\\fv w  x yL z< {  | }, ~l   l ,   l ,   l ,   l ,   l , TTTTDDDDDD¬DìĬŬƬDǬȬɬtʬ4ˬ̬7dͬT7dά7dϬ6dЬ6dѬ3dҬ\3dӬ3dԬ2dլ2d֬Nd׬dNdج$Nd٬MdڬMd۬LdܬLdݬTLdެ߬t4RddRd$RdDtgd,y    d $   _d4 x4dlot4$Sd  dGd | T d d d T        T   \    \    L     L ATAAhhy}}##6677g7g7}788.B.BEBEBPHPHJJMMZZ[[y^y^^^eeuuuuvvOvOvq}q}++ TT^@@OO}}ww; rr|| ==--;%oo..::LLUU\66?GGQQY-       f      !"#$%&'()*+,-./0123456789:;<=>?@BACDFEGHIJKLMNOPRQSTUVXWYZ[\]^`_abcdefghijklmnopqrsutvxwy{z|~}w{{##66 7 7n7n77880B0BGBGBRHRHJJMMZZ[[^^^^eeuuvv1v1vevevx}x}22\`` FFTT }}="zz?? 7EE#''tt55??SSZbb>FFMMWaa)::  # #    f  !"#$%&'()*+,-./0123456789:;<=>?@BACDFEGHIJKLMNOPRQSTUVXWYZ[\]^`_abcdefghijklmnopqrsutvxwy{z|~}9*urn:schemas-microsoft-com:office:smarttagsState_*urn:schemas-microsoft-com:office:smarttagscountry-regionhttp://www.5iantlavalamp.com/V*urn:schemas-microsoft-com:office:smarttagsplacehttp://www.5iantlavalamp.com/=*urn:schemas-microsoft-com:office:smarttags PlaceName=*urn:schemas-microsoft-com:office:smarttags PlaceType8e*urn:schemas-microsoft-com:office:smarttagsdateh*urn:schemas-microsoft-com:office:smarttagsCity0http://www.5iamas-microsoft-com:office:smarttags 11115200523DayMonthYeareeehlmw& Q U bhbgF%N%W%_%d%j% &&M&S&X&`&}&&&&&&'!'z++++//}1111_4d4x4|4444444448:D:T:\:a:h:q:y:::::::::::>>??AACCFFFFFG"H(HHHIIII>JIJQQ)S-STTXX\ \^^^!^G_O_T_\_mmnnpp|ww{{| |q~~~~~=Nކ$)rs46_a./GNgiuxҚӚ`hŲ̃#)OZԲٲVWovóдִ(./3FL $ aeHT .4Ua}LVX_emv{]e 19BJIQW[ahqw -3#)Q\ekDJSX (}/5)/AJbh!*KPVZ^gFK]b\bJS"R f    J P   JRW^elnpqstz{ LLNNOOQRTUWX`af!!,,55@@]A_A{H}HLLSSAcCcgkikDŽ̈́[`%'df+- %~y}Z_$&%2 `cLOc f    'eLLNNOOQRTUWX`af33333333333333333333333333333333333333333h|  ZP@C`""!$#$0$$&'+,F-.23h5p78]9?A CC#F$FIJJJZN[NPPQR6U:VGY#Z^e__ab c$effghjkklm%n0nnTo{pqPrWstuu"u8uSuvv)w*w?w{w|wwwwxxyyyzzz0{?{E{{{{ |$||||||||}}}}}q~~~+=|ʀˀހ `t2114֛Ȝ ƥDzβ7:&8e  LLNNOOQRTUWX`af LLNNOOQRTUWX`afmajaml~(lo/`?Rڰnx Rڰng36U ,2@FA?̞#X@bWfhh&v%xZ)/Aؐ %d6U Wj%`l5rO A2pFxli (J2'b- Jf7O0ֳ4QQ0TovtVdؾv^LQKgTmA) B M ^ dp7WxUKs/])a #%DP)z3?~7g??G?uB DCE7L+N;N$fP jR]kRSWvX@zY['\\}n]_)FbcBce_gHkk[l+@mG"s2vYx.~rN f/ H%@uO#\o@t#}rIn>?}C@Db$S=t7Lo?Vs~ZUP|F: ^&s`W)5Sz   #<@A!$Au!u"u+u7u8uQuRuSuWuwuxuyuuuuuuuuuuvvv2vsvtvuvvvvvvvvvvv(w)w*w?wzw{w|wwwwww?x@xAxhxxxxxxyyyOyyyyyzzzLzzzzzzz/{0{={>{?{E{{{{{{{{{ | |"|#|$|/|_|`|a|s|||||||||||}}} }!}%}*}1}7}8}?}D}J}P}Q}W}\}b}h}i}y}~}}}}}}}}}}}}}}}}}}}}}~ ~~~~!~&~,~2~3~F~K~Q~W~X~p~q~~~~~~~~~~~+06<=| !"#$/4:@ASX^derstuv}ĀʀˀҀӀԀՀրހDz˲̲\er &'()*09?EFT^dklstuvw%+12GHIJKov|"*078QRSTUY`fmnv|e fEc01y!(!0 * *!0!0@X[]e00@0d0@00 @0@Unknown gz Times New RomanTimes New Roman5Symbol3& z ArialCAdvTT5843c571cTimesNewRomanTimes New RomanO&  Zurich Cn BTImpact9GaramondABook Antiqua5& z!TahomaE& Century Gothic?5 z Courier New;Wingdings7Georgia"1h{f{fj{f((!4d} }  2qHP ??%Katedra ekonomske teorije i politike,Shtrikamajam                          FMicrosoft Word Document MSWordDocWord.Document.89q