BIDIRECTIONAL APPROACH FOR UNIVERSITY MANAGEMENT: IMPROVED RELATIONSHIP WITH STUDENTS AND EDUCATIONAL COSTS MANAGEMENT

. The research, through a bidirectional approach, aims to develop the university management of the degree programs in order to attract and maintain the potential students, as well as to estimate and manage the costs of the educational services offered during their life cycle. The authors bring into the centre of the study the importance of the simulation process in predicting a possible future and reduce the probable risks and costs from time. The results confirm that the simulation methods used help the universities to determine the future risks and to predict when and what relationship marketing programs to use in order to attract, maintain and grow the number of valuable students and also to ensure cost-effective management of university programs. method (LCC).


Introduction
In the current period, many specific factors have occurred at the level of university education (demography, technical and technological discoveries, knowledge-based information society, artificial intelligence, data analytics, and block chain technology), which have completely transformed the structure and functionality of universities (Balzer, 2020;Chaminade & Lundvall, 2019;Kolomytseva & Pavlovska, 2020;Williams, 2019). The contemporary trend is for universities to be governed by the labour market, and their orientation is towards final consumers (economic entities, state institutions, who will benefit from skilled labour) (Altbach et al., 2009;Evans & Gill, 2017;Reichert, 2019). Knowing that relationships for human beings are essential, institutions, including universities, must understand that building a strategy on marketing relationships will lead to competitive advantage and satisfying students (Palmatier & Steinhoff, 2019;Toma, 2011). Universities, establishing their standards, creating new levels of excellent standards, are offering learning, education, and research (Kahn & Anderson, 2019). Now they are autonomous and ask from the pleasant student thinking, learning as understanding in a different way (Dall' Alba & Barnacle, 2007), due to their open-access mission (Schneider & Deane, 2015) and healthy attitudes, in order to face the economic, social, cultural, political, and technological challenges existing in the work market. Universities are increasingly taking inspiration from the business world in an attempt to achieve profitability in a competitive market. Universities are required to make realistic and valid estimates to ensure the essential sources of funding, and the costs involved must ensure both the supportability of the different categories of beneficiaries and the prosperity at the institution. That increasingly arises the need for evaluation, information, and communication on the cost of education in higher education (Ianos, 2010).
The study aims to develop the university management of undergraduate programs through an innovative two-way approach, on the one hand from the perspective of attracting and maintaining potential students and identifying optimal solutions for development, and on the other hand the estimation and management of the costs of educational services offered during their life cycle. Thus, the purpose of the research is to predict the future number of students enrolled, based on data from previous years and to estimate and manage the costs of educational services during their life cycle.
The objectives of the study are: Objective 1 -predict future problems in attracting and maintaining students and future solutions for developing the number of students; predict the right number of enrolled student in order to ensure effectiveness for the analyzed institution, observe the difference between the obtained calculation and the real data; improving the relationship with future students, getting in-value for students and the institution; Objective 2 -application of the life-cycle cost method (LCC) of an educational license program belonging to the economic profile, in order to identify whether the results of the LCC method add value to accounting information on the relevance of costs or the composition of the cost of production or provision of the service. Starting from the announced objectives, the research hypotheses are: Hypothesis 1: Statistical and econometric methods contribute to predicting future problems in attracting, enrolling and retaining students to improve the relationship students-university; Hypothesis 2: LCC and statistical methods applied for an educational license program belonging to the economic profile add value to accounting information on the relevance of costs or the composition of the cost of production or provision of the service. The novelty of the paper lies in the fact that it addresses a current problem with practical implications, which involves presenting a model in which statistical and econometric methods and the specific method of management accounting (LCC) are used as a bidirectional approach to improve university management. The structure of the research presents the main studies from the scientific literature, the applied materials and methods, followed by the results and conclusions of the research. Evans and Gill (2017) ask themselves if the university is a business, and they also answer, by providing the following information: that between the student and the institution is made a contract-students are obligated to attend the courses and the seminars, to pass the exams and to pay their fees, and the institution, through its professors, is obligated to offer excellent quality teaching, adequate facilities and well-designed courses, meaning education and academic excellence (Nejati, 2013;Sultan & Wong, 2019). The universities are using relationship marketing by increasing students' involvement in university activities, such as technologybased activities, programs based on the relationship students-librarians (Brock, 2019;Kaur, 2009;Pringle & Fritz, 2019). According to Baran et al. (2008), there are three levels of relational marketing. Level 1 is based on offering discounts in order to attract and maintain students, prices, money scholarships for good results, study scholarships for the same good results, reduction for student camps, travel, and transportation. Level 2 is based on user customization and social connections as special events for students: parties (just for students, for students and the professor, such as Christmas, seminars, conferences, workshops, focus groups, educational programs, PR events, sports activities, alumni relationships, counselling and assistance, counselling in career, social events (Christmas events, PR campaigns for children and families with lower income, international sessions for students, concerts, and programs for the beginning of the year, associations). Level 3 is based on connections sustained by structural solutions, as computers and IT programs in order to detect the students' payments, the preferences for some courses, the amount of money and the moment of paying the preference or for a program, or the frequency and the recency of presence, or the average calcification per semester or year, the evidence of results. These relationships allow the university to be known using recommendation and word-of-mouth (Oplatka & Hemsley-Brown, 2012), alumni relations (Ong, 2009), which represents acquisition and retention strategies for universities and a new approach for CRM, academic counselling and assistance, and new technology forming a new generation of students (Harland, 2012). Relationships are the way to gain new knowledge and experience and to confront different barriers (Sheldon & Turner-Vorbeck, 2019).

Literature review
Students are regarded as customers, to whom the universities are offering them education services, and because the clients have rights for what they are paying for, they may have complaints. However, also, the university has rights, because the contract is for the two parts (Evans & Gill, 2017). Relational marketing is perceived as a bond of equality between partners (Baran et al., 2008), and the partnership between the university and its students is characterized by trust, commitment, communication, collaboration, and knowledge sharing, resulted through the mutual satisfaction of objectives. The relationship with students is different according to the stage of student life-cycle: recruitment stage, enrolment, retention, and post-graduation (Ackerman & Schibrowsky, 2007;Lechtchinskaia et al., 2012;Perna & Baraldi, 2014). Relations with students, perceived as the relations with customers, may conduct to collecting information for a database, which may become strategic knowledge and may be used in order to increase the ability of the institution to sell its educational services more efficient, to create at the highest level of satisfaction and loyalty with obtaining a high profit (Linger et al., 2004). Thus, the students-based orientation, become a 1-to-1 relation, based on creating and offering added-value; this type of relationship developed due to new technologies (Svend & Oliver, 2019) and is belonging to those organizations which are using data from the past about its customers and are used in order to predict what client wants. Thus, the organization will treat its customers differently and will grow the mutual value (Chen & Popovich, 2003;Foglieni et al., 2017;Fukuda, 2017;Lamb et al., 2012;Peppers & Rogers, 2004;Sauro, 2015;Valdani & Arbore, 2013;Winer, 2001).

Materials and methods
Objective 1 -The used tools are descriptive statistics, Monte Carlo simulation, Markov chains, Gauss distribution, and informatics programs (Excel and MATLAB). The institution which has been analysed is a university from Romania, denoted from strategic reasons, as UR. On the 17th days for the students' enrolment (the summer period), were extracted data for four years between 2016-2019. The total number of enrolled students is shown in the table below (Table 1). Monte Carlo method is used in different economic activities, so using it in the education field is a novelty. Monte Carlo method is demonstrating its efficiency in analysing phenomenon characterized by a large number of variables and parameters (Duica & Florea, 2018;Zio, 2013). Markov chains are a mathematical method used in planning and forecasting (Beardwell & Claydon, 2007;Ching et al., 2013;Duica et al., 2019;Florea & Mihai, 2017;Nastase (Bidireanu) et al., 2019;Tracey, 2004), in this case, it is analysed the forecasting of the enrolled student evolution.
Objective 2 -The first variant of analysis of the results obtained by applying the model specific to the LCC method is the sensitivity analysis. The analysis can be repeated for each parameter of the model or only for parameters considered to have a significant impact on the results. Useful in one-way sensitivity analysis is the choice of the values of the smallest and highest of the parameters chosen to be analysed. The choice of these parameters was determined by their uncertainty or their dependence on factors that are not related to the incidence of decision-makers (population dynamics, the degree of promotion of students in the previous cycle, or market dynamics work). For the application of the LCC method and the sensitivity analysis was prepared Table 2, which shows the costs related to the introduction of a license program.
According to Brown and Yanuck (1980), the LCC method is used when it is necessary to draw up a decision related to the purchase of an asset that requires substantial maintenance and which induces considerable operational costs over its life (Brown & Yanuck, 1980). Dhillon (2009) considers that the primary usage of the LCC method lies in the possibility of comparing competitive projects, long-term planning, and providing support to budgeting and control processes, selecting the best bid or elaboration of decisions on replacement of equipment (Dhillon, 2009). More recent approaches regard the LCC method as a useful engineering tool that can be used in the design phase and when product or service development purchases are made, as well as as an instrument that can be applied in a pro-active manner in management and management accounting, as well as in the management of environmental issues (Asiedu & Gu, 1998;Ashworth, 1993;Emblemsvåg, 2003;Gluch & Baumann, 2004;Karim, 2006;Spickova & Myskova, 2015). Optimal total number of students to reach the break-even point 425 The average annual number of students 85

Results for Objective 1
Making a summary of the descriptive statistics, using the data in Table 1, the following results are obtained and presented in Table 3. The smallest mean from the four series of data is for 2018, and the biggest is 11.41 for 2017, and the median is between 9 and 13. The mode is between 1 (2019) and 13 (for 2017). The minimum for this series of data is 0 (for three series), and 4 (for 2018), and the maximum is 27 (for 2016), 24 (for 2019), 22 (for 2017), and 18 (for (2018). The range for the difference between the maximum value and the minimum value is the biggest for 2016 (27), Table 2 and the smallest is for 2018 (14). The sample variance is between 21.56 (for 2018) and 75.26 (for 2019), and the standard deviation is between 4.643 (for 2018) and 8.676 (for 2019), having small values for all the variables. Therefore, it can be considered that the series is relatively homogeneous. The standard error is between 1.126 and 2.104. Skewness has negative value only for 2017 (-0.33), the series being negatively asymmetric, but weakly, the curve is in the left. For the other three series of data, the skewness is positive between 0.254 and 0.409, meaning that for 2017 the series of data is skewed to the left, and for other three are skewed to the right. The kurtosis is negative for 2019, and 2018, but for 2017 and 2016 is positive (0.134 and 0.896), being much below the benchmark for a normal distribution of 3, which is positioned near normality, meaning that the curve is not so sharpened, having a platykurtic curve.

Monte Carlo simulation
In order to simulate the future number of enrolled students, it will be determined the daily probability using relative frequency (Table 4). Table 4. Data regarding the enrolled students and the probability of enrolment Using the Monte Carlo method involves applying the following procedure (Luban, 2005): in step 1, are determined the probabilities P(X = x i ) = P(x i ) and the cumulative distribution function F(x i ) = P(X ≤ x i ) = ΣP(v), for x i ∈ {x 1 , x 2 , ..., x m }; in step 2, there are associated intervals of random numbers for each discrete variable (graphically, F(x) has the form of a step, its height being equal to the probability P(x i ), and the interval of random number associated with the value x i as Figure 1). To each value xi is associated the interval (F(x i -1), F(x i )) with F(x 0 ) = 0 (Table 5); in step 3, it is generated a random number u uniformly reported in the interval (0, 1) using a generator (Table 5); in step 4, the same procedure is made as in the step 3 until it is obtained the desired simulated selection.  According to the calculations, the number of registered candidates will vary between 16.48 and 21.71.

Gauss distribution
Calculating the distribution of the analysed series according to the pre-established coefficients of Gauss and the original series is obtained the values from Table 6. They are graphically represented in order to observe the difference between Gauss representation and the real situation for the four analysed years (Figure 2).  Analysing the graphics above (according to both sets of data) are observed (Figure 3): In the first graphic, for 2019, compared with the Gauss representation, there are two kurtoses, the smallest is on the left and is very reduced compared to the other four graphics, the highest being on the right, so the skewness is positive and the series is leptokurtic; In the second graphic, for 2018, there are also two kurtoses, in the centre and the right, almost equal, so the series is platykurtic; In the third and the fourth graphic, for 2017 and respectively, 2016, they are also two kurtoses, the highest in the centre and the smallest in the right, but high enough, the series being leptokurtic. All the four series are not following a perfect Gauss distribution, but one made in the right and the centre, showing that the highest number of candidates is concentrated in the right, and respectively in the centre (for very good and medium values).

Markov chains
All the values per year are presented in Table 7, and then, using stochastic calculations, are transformed into probabilities. From the table above, data are put on four categories, in order to form an equal number of columns and rows to make probability calculations (Table 8). Making the calculations specific to probability, are obtained the numbers from Table 9.  By calculating the probability P i of each year in total candidates on analysed years (P i = T i /T), it is obtained the line vector: (0.26 0.27 0.23 0.24). Thus, P 1 = 189/732 = 0.26; P 2 = 194/732 = 0.27; P 3 = 171/732 = 0.23; P 4 = 178/732 = 0.24. Then, it is made the calculations, according to the Markov chains method (Table 10). Putting this data in a graphical representation (Figure 3) could be perceived as a forecasted trend for the future years. Can be observed that: i) for the first period between 1-4 days the forecasted number of enrolled students will follow a decreasing trend for 2021 (0.25) and 2023 (0.24), and after 2024 it will stabilize to 0.25 for a long time; ii) for the second period of enrolment between 5-8 days the trend will be rising for 2021, 2022 and 2023 (0.26), after this from 2024 also tend to stabilize to 0.25, as the first period; iii) For the third period between 9-13 days the trend is decreasing, having lower values as 0.24 for 2021 and 2022, is increasing for 2023 at 0.26, then stabilizing at the same values as the first periods at 0.25 for an extended period; iv) for the last analysed period, the trend is entirely stagnated at 0.24, being the lowest, with a pick of 0.25 in 2021, then stabilizing after 2024 at 0.25, like all other analysed periods. The value where all four periods may be probably stabilized is 0.25. Thus the ideal number of enrolled students may run between 180 and 186 (by approximation in minus or plus). According to the forecasted values per year, measures could be taken from time for the last two periods when the values of enrolled students were the lowest, even if they will be stabilized at the same values as the others, meaning for the periods between 9-13 days and 14-17 days. That means that for the first two periods, in the first eight days, students are coming more to enrol. Thus, a plan for a better enrolment could be implemented in order to diminish this issue. It means that the website information, the mouth-to-mouth methods from professors and students, the new technologies (Facebook, WhatsApp, and telephones), the campaigns taken from the time of the analysed university had results, and the interested persons will come in the first days to enrol.
Based on the results obtained for objective 1, Hypothesis 1 is validated as follows: i) The Monte Carlo method allows the simulation of the future number of enrolled students and the determination of the daily probability, by using the relative frequency obtaining an average interval of the daily enrollments. This can significantly contribute to ensuring the efficiency of the analyzed institution; ii) The Gaussian distribution highlights the difference between Gauss's predetermined coefficients and the real situation for the analyzed period, so the method can be applied to manage student enrolment; iii) Since the prediction of enrollment of future students can be done by the Markov chains method, it is considered that this method can be useful for implementing future university management strategies, which will contribute to improving the relationship with students, from the enrollment stage.

Results for Objective 2
For the product lifecycle cost assessment, it was performed the first two forms of sensitivity analysis. In both one-way sensitivity analysis and multi-directional sensitivity analysis, the parameters whose values were tested were the number of students and the amount of the annual fee (Table 11-13).  191 53,191 53,191 53,191 53,191 265,957 Annual income (Eur) -6,176 -787 -1, 119 -574 -4,612 -13,268 From the sensitivity analysis of the LCC method model we can draw the following conclusions: for a constant number of 50 students annually it is impossible to reach the break-even point, the licensed program being ineffective for fee values between €638 and €1,064; in the variant of attracting a maximum number of students for each year of study, profitability can be obtained from year 2 of the program to year 4; in the case of average values of the number of students generated from the analysis of other master programs, the master program is profitable in years 2 and 3, and its overall profitability is achieved. However, the same result cannot be achieved in the variant of a total number of students over the life cycle below the minimum allowable values of the average break-even threshold (438 students). Table 11  Table 13. Sensitivity analysis by the number of students and the amount of annual fee -multi-directional analysis

End of
Year Based on the results provided by the multi-directional analysis it can be concluded that a positive return on the overall life cycle study program can be achieved in the context of the annual increase in both the number of students and the tuition fee. Another tool offered by econometric modelling is the analysis of correlations between variables (Table 14). There is a strong inverse correlation between R&D and operational cost variables, as well as between R&D and closing costs. The linear regression method studies the link between the LCC cost dependent variable and several independent variables associated with the various stages of the study program's life cycle -R&D costs, operational costs, and support costs and withdrawal costs. Independent variables have been chosen as the main cost elements, components of the LCC method, which define its structure. The cost values associated with the LCC method were previously calculated and are presented in Table 15, and used cost values at the level of 2019. The general form of the simple regression equation is Y = c(1) + c(2) × X, X defining the independent variable and Y defining the dependent variable. The probability associated with parameters c(1) and c(2), (which can take values between 0 and 1) gives the measure of the significance of the parameters generated (obtained from the generation of the regression equation), i.e., the closer the value obtained is closer to 0, the higher the significance of the parameter. If the probability value is close to 1, by studying the results of the t-test or student test, the insignificance of the parameter can be concluded.
In order to interpret the results presented above, related to simple regression equations, it is necessary to define the main elements generated by the modelling program theoretically. Based on the estimates of the regression parameters for the proposed equations obtained using the SPSS program (Supplementary Material, Annexes 1-4), the information in Table 16 was obtained.
By observing the data in Table 16, it can be concluded that there is a high significance of the parameters for Model 1 and Model 4, while Model 2 and 3 have low meanings of the parameters. In the case of the first model, R-square = 0.591 means that 59.10% of the variance of variable y can be explained using the regression equation, the difference up to 100% not explained. In the case of the second econometric model, R-square = 0.022 means that 2.20% of the variance of the dependent variable is explained by the econometric model, which allows us to conclude that this model is less relevant than the first. For the third and fourth equations, the values of R-squared are 0.01 and 0.1815, so 1.01% in the case of model three and 18.15% in the case of the model appeared from the variance of the dependent variable is explained by the chosen model. The decreasing order of relevance of the chosen models is, therefore, Model 1, Model 3, Model 4, and Model 2. Thus, we can conclude that the value of the cost of the LCC is strongly dependent on the value of the R&D costs as well as the operational ones, with the total and closing costs having a lower impact. The t-test is utilized to test the statistical significance of the linear relationship between the two variables : t = 0.815 with 3 degrees of freedom and associated probability p = 0.474. It can also be observed in this analysis the increased significance of models 1 and 4 compared to the other two models. Once again, the strong influence of R&D and operational costs in determining the cost associated with the life cycle compared to the closing costs or baskets of service is highlighted. Based on the results obtained for objective 2, Hypothesis 2 is validated as follows: i) the sensitivity analysis of the LCC method highlights the positive profitability of the study program throughout the life cycle resulting in the context of the annual increase in both the number of students and the tuition fee or in the case of values averages of the number of students the study program is profitable in years 2 and 3; ii) The LCC allows the identification of those cost areas or areas that require higher growth or deeper control in order to maintain them at specific rates or to reduce them, where possible; iii) the statistical methods applied highlight the strong influence of research and development costs and operational costs in determining the cost associated with the life cycle compared to closure costs.

Conclusions
The education market is dynamic and continuously changing, and economic-social factors and internationalization require universities to rethink their missions and have greater flexibility concerning today's developments. Universities need to restructure their curricula so that they quickly meet labour market requirements. To attract more valuable students and maintain them, the institution could follow the steps developed above and use the Monte Carlo simulation in order to determine possible future problems and to determine the right number of necessary students, which may ensure the future performance of the university. As it is observed, the simulation method used above helps the institution to determine the future risks and to predict when and what relationship marketing programs to use in order to attract, maintain and grow the number of valuable students.
The introduction of a new study program should start from a thorough analysis of market requirements, customer characteristics, and competitors' offers, which often makes the price at which it can be offered on the Market. Furthermore, an analysis of the profitability of a new study program is not complete and relevant if it does not cover the entire life cycle of the program, thus it is essential to include the LCC method in the analysis. One of the main problems in reaching and maintaining a specific rate of profitability is estimating the majority of costs in the initial phase, and managing them in the growth phase and after the time of their appearance. The main problem with cost management is thus the failure to include all costs over the life cycle of the services at the time of the initial estimate. Thus, there is a high risk that there will be a cost of study programs that are not accurately estimated, and that is not properly managed in terms of strategic management. In this context, the financial target at a university should aim to maximize income while maintaining or even reducing expenditure. The solution of creating their funding resources and avoiding dependence on a single funder seem to be fundamental needs. The calculation of standard costs per student could also ensure the economic basis of the fees charged for university services. At the same time, creating an organizational culture based on saving could ensure financial performance. Also, as a research limitation, it is considered that the application of LCC requires considerable resources of time and the data necessary for its application may be challenging to obtain or that its accuracy may suffer due to uncertainties that may arise in the future.
Universities, both state and private, need to be aware of the inadequacy of the funding from study fees and must become much more cautious in calculating costs. Often, management structures, as well as students, need detailed justification of how funds are managed. As a result, the accounting tool is best suited to highlight the costs associated with the educational process, only thus succeeding in calibrating tuition fees, which meet both the expectations of potential students and the need for survival and/or institutional prosperity.
The paper contributes to the development of scientific literature in the field of university management and managerial accounting, by the fact that the study proposes a reorganization of the management of universities by making calculations on the viability of developing new undergraduate study programs throughout their life cycle. The research limitation results from the fact that the proposed model is implemented only on a single license program. It is obvious that each university program (bachelor, master, doctorate) has its own requirements (number of teachers, university infrastructure, methodological requirements, number of students, etc.) that must be analyzed separately. The model can be extended to other university study programs, so as to contribute sustainably to increasing university performance and cost management. As a future research direction, it can be considered, the application of the bidirectional model on other university specializations, which have different characteristics for the study programs and different student admissions.