Banka Şubeleri İçin Uygun Yer Seçiminin Belirlenmesi

Abstract

Teknolojik gelişmelere bağlı olarak banka şubeleri tarafından sunulan hizmetin karşılanabileceği alternatif kanallar olarak kredi kartları, telefon-internet bankacılığı, operasyon merkezi, otomatik vezne makineleri (ATM – “Automated Teller Machine”), satış noktası terminalleri (POS – “Point of Sale”) vb. kullanımında artış olmasına rağmen; mevcut müşterilerin bankaya sadakatini arttırmak, yeni müşteri elde etmek ve tüm müşterilerle iletişimi sürdürebilmek açısından şubeleşme, bankalar açısından önemini korumaktadır. Türkiye’deki nüfus ve şehirleşme oranındaki artış ile birlikte bankalar, müşteri sayısını arttırmak için şubeleşme konusundaki çalışmalarına hız vermiştir. Türkiye Bankalar Birliği tarafından sunulan istatistiki bilgilere göre, Türkiye’de hizmet veren banka şubelerinin sayısı 2012 yılında 10.159 iken 2013 yılında %7,8 artış ile 10.952 sayısına ulaşmıştır. Bu sebeple, şubelerin en uygun yerleşim yerinin belirlenmesi, bankaların stratejik hedeflerine ulaşabilmesi için önemli bir husustur. Şube yerleşimi için belirlenecek yer; başta bankanın strateji ve vizyonu, aday noktadaki müşteri profili ile şubenin yerleşeceği bölgenin özellikleri olmak üzere çok sayıda faktöre bağlı olarak değişkenlik gösterebilmektedir. Yazında konuyla ilgili birkaç çalışma bulunmakta olup bu çalışmalar detaylı incelendiğinde ele alınan kriter ve matematiksel modellerin farklılık gösterebildiği, ortak bir kriter seti ve modelin olmadığı anlaşılmaktadır. Bu doktora tezinde, banka şubeleri için en uygun yerleşim yerinin bulunması ile ilgili özgün bir yöntem sunulmaktadır. Bu amaçla öncelikle, detaylı yazın araştırması yapılarak ve uzman görüşünden yararlanarak banka şubeleri için en uygun yerlerin belirlenmesini etkileyecek kriterler tespit edilmiştir. İkinci aşamada, bankacılık sektöründe deneyim sahibi uzmanların görüşlerinden yararlanarak ve ikili karşılaştırma yönteminin yardımıyla hiyerarşik yapıdaki kriterlerin dört farklı şube tipi için önem seviyesi belirlenmiştir. Önceliklendirilen kriterler, banka şubelerinin tek ve çok dönemli yerleşimi için uygulanmak üzere önerilen iki farklı matematiksel modelde kullanılmıştır. Önerilen modellerin büyük ölçekli problemler için en iyi çözümünün bulunamaması nedeniyle, Tabu Arama sezgisel yaklaşımı geliştirilmiş ve Tabu Arama ile elde edilen sonuçlar CPLEX 12.2 ile kıyaslanmıştır. Sonuç olarak, önerilen yöntem Türkiye’de hizmet veren büyük bir bankanın İstanbul’daki şube yeri seçim problemi için uygulanmıştır.

Although technology has improved and distribution channels such as credit cards, mobile-internet banking, operation centers, automated teller machines, point of sales etc. have become alternative opportunities for reaching bank services, the branch offices are still important for the banks to gain new customers and keep in touch with them. Growth in population and urbanization in Turkey forces the banks to increase the number of their branch offices in service to reach new customers. According to the statistics of The Banks Association of Turkey, the number of total domestic bank branches has increased by 7.8% from 10,159 to 10,952 in the 2013. On the other hand, JPMorgan & Chase opened 89 new branches in June 2013, so that increased number of its branches from 5,608 to 5,697. In the same period, BB&T increased number of its branches from 1,775 to 1,851 (Retail Banker International, 2013). This shows that, by the effects of increase in total population, population per bank branch and individual earnings, banks try to increase the number of their branches, by locating them in the right places. Therefore, locating the branch offices in the best place is one of the most important decision problems for Turkish banks, as well as for the banks in other countries. Finding the best location for bank branches depends on a number of distinct measures that differ according to the banks’ strategies and vision, customer profile in the potential location and features of the place where the branches will be located. In order to open new branches, banks can focus on the places where they or their competitors have no branches, often areas involving industrial and commercial activities, organized industrial zones, shopping centers, collective housing areas, touristic regions, universities etc. Depending on the bank’s strategies; economic development level, population and demographic characteristics, distribution network, latent customers and their proximity to the potential markets, physical location, credit and deposit potential of the candidate regions may be important factors in terms of branch location. Meanwhile, due to the different characteristics of the customers effective on transaction volume, banks prefer to open varied types of branches (individual, commercial, corporate, private, entrepreneurial etc.) to minimize their costs and maximize their business process efficiency. Thus, importance and effect of the characteristics of the potential locations can differ depending on the branch type. For example, in regions where commercial and industrial activities are high and big investments are made, commercial and corporate branches are generally selected, while places where population and/or level of collective housing is high are chosen for individual banking. Also, branches providing private banking services are mostly located where average household income is high. There are several studies in the literature about finding the best place for bank branches. Detailed literature review shows that different criteria and mathematical models are used for the problem of bank branch location where common criteria do not exist. In most studies, different statistical techniques and criteria are discussed for the decision-making problem of finding the best places for bank branches. Clawson (1974) proposed Stepwise Linear Regression to solve bank branch location problem, setting realistic performance standards for different potential sites, and specifying remedial actions for poorly performing branches from a sample of 26 branches. Boufounou (1995) used Regression Analysis and conducted some statistical tests to determine the priority and significance level of related criteria for a Greek bank in order to evaluate existing branches’ performance, assess performance of potential sites and place branches in more efficient locations. Ravallion and Wodon (2000) also used Regression Analysis to explain the relationship between economic indicators and the location of Bangladesh’s Grameen Bank’s branches to measure potential gains of switching out of farming. Since it considers multiple criteria, Multi Criteria Decision Making (MCDM) models are very common for bank branch location problems. Because of the uncertainties in the comparisons of the criteria and the alternatives, fuzzy set theory is used very frequently. For instance, Min (1989) proposed Fuzzy goal programming method to find the most appropriate places for commercial bank branches in Ohio. In order to select one among six cities in South Eastern Anatolia for opening a new branch, Cinar (2009) used Fuzzy Analytic Hierarchy Process (AHP) to find the priorities of the related criteria and TOPSIS to rank the cities. Rahgan and Mirzazadeh (2012) used Fuzzy AHP to specify the weights of criteria and evidential reasoning to order the alternatives. As a result of the improvements in information technologies, Morrison and O’Brien (2001) used Geographical Information Systems (GIS) by using a spatial interaction model (Huff, 1963) in four stages: The probability of customers visiting a branch is estimated, the expected distribution of customers is distributed, the expected number of transactions at a given branch is computed and the impact of removing one or more branches from the network is analyzed. Although the mathematical programming literature related to facility location problems is rich, studies specific to bank branch location problems are scarce. In one of the earliest methods, Min and Melachrinoudis (2001) proposed a three-hierarchical location-allocation model for bank branches by considering risk and uncertainty, where a chance constrained goal programming model was developed by using forecasting techniques. The methodology was applied in a province in the USA. Miliotis et al. (2002) introduced a two-step methodology in which firstly the minimum number of branches to meet coverage requirement was found and then locations of branches to maximize the coverage was determined. Wang et al. (2002) considered the problem of locating automated teller machines (ATMs), internet mirror sites, or other immobile (permanent location) service systems of limited service capacity. They model these service facilities as simple M/M/1 queuing systems and solve the model using three different heuristic approaches. Wang et al. (2003) developed a mathematical model for the branch location in Amherst, New York. Unlike the P-Median model, the model consists of a budget constraint related to opening and closing branches. Zhang and Rushton (2008) proposed a model maximizing total benefit with the budget constraints related to opening branches and the waiting time of the customers. Alexandris and Giannikos (2010) proposed a new model for maximal covering location problem and illustrated the applicability of the proposed model by means of a case study concerning the location of bank branches. The aim of the model is to maximize the total population covered by the selected branches. Xia (2010) formulated a mathematical programming model considering operations and rental costs, demand and distance between branches. On the other hand, different solution methods are used for bank branch location problem in the literature. Miliotis et al. (2002) used GIS and applied their methodology to a bank in Greece. Wang et al. (2003) developed Greedy Interchange, TS and Lagrangean Relaxation to apply this NP-Hard model for the branch location in New York by using 270 generated problems. Zhang and Rushton (2008) used Genetic Algorithm to solve their proposed problem. Xia (2010) proposed a hybrid nested partitions algorithm to solve the large scale problem. As a result of the literature review, we concluded that the bank branch location problem is studied mostly as an MCDM. The mathematical programming models in the literature, on the other hand, concentrate on modeling and solving the pre-defined problem. They do not provide any information regarding specifying the criteria and do not consider the multiple criteria nature of the problem. To the best of our knowledge, there is no integrated methodology that combines problem structuring phase (i.e., as in MCDM models) and mathematical programming models for the bank branch location problem. This thesis presents a methodology to find the best location of bank branches. For this aim, firstly, a number of criteria are selected by the help of a detailed literature review and expert judgments. According to the experts’ judgments and detailed literature review, average transaction volume is the most effective factor for bank branch profit. Moreover, the distance between all the potential points is another main criterion, especially to avoid opening multiple branches close to each other. Thus, it is determined that the opening of new branches in close proximity should be penalized. Moreover, experts indicate that costs of opening a new branch as well as closing an existing one are other important criteria for bank branch location problems. Also, sub criteria affecting the transaction volume of bank branches are determined. Subsequently priorities of these criteria for four different types of bank branches are identified based on expert judgments using pairwise comparisons. The priorities are used in two new mathematical models developed to decide the best branch locations of bank branches in Turkey for single and multi-period planning. The objective function of the models maximizes the total net profit, as the difference between the total benefits of expected average monthly transaction volume minus the penalty of opening branches nearby to each other, cost of opening new and closing available branches. The number of branches to be opened for each type is limited depending on the budget constraint. Also, some branches cannot be closed by the strategy (referring branches opened in the last three years in this thesis, according to the expert opinion). Since optimal solution cannot be easily found for big regions, a Tabu Search heuristic approach is developed for both models and the results are benchmarked with CPLEX 12.2. Finally, the proposed methodology is applied for a large Turkish national bank’s branch location problem in Istanbul.