Elen Twrdy, Ph. D. Mlan Batsta, Ph. D. Unversty of Ljubljana Faculty of Martme Studes and Transportaton Pot pomorščakov 4 632 Portorož Slovena Prelmnary communcaton Receved: 2 th November 213 Accepted: th December 213 A Dynamcal Model of Contaner Throughput of the North Adratc Multport Gateway Regon SUMMARY The ports Rjeka, Koper, Treste, Vence and Ravenna are ports of the North Adratc multport gateway regon (NA). They have a good geographcal locaton and modern contaner termnals, but ths multport gateway regon has been facng lower rates of contaner throughput than the rest of the multport gateway regons n Europe. The paper ams at provdng an overvew of the dynamcs of competton among the NA ports Koper, Treste, Vence, Ravenna, Rjeka and at analysng the throughput n ths ports n the last twenty years. Based on these numbers, a smple but effcent model to forecast the possblty of growth or fall of traffc n the next year has been found. Key words: Multport gateway regon of the North Adratc, competton, contaner throughput POMORSTVO Scentfc Journal of Martme Research 27/2(213) str./pp. 361-367 361
1 INTRODUCTION The North Adratc multport gateway regon (NA) conssts of the ports of Ravenna, Vence, Treste, Rjeka and Koper [1]. Due to ther geographc proxmty they share the same hnterland and a lmted number of customers. In ths vew they can be treated as a sngle port system although they are located n three dfferent countres, wth a dfferent economc potental and dfferent development plans. In general, we can say that the NA ports have, wth respect to potental markets (Swtzerland, Germany, Austra, Hungary, the Czech Republc, Slovaka), a good geographcal locaton and modern contaner termnals. However, ths multport gateway regon has been facng a stagnaton of the contaner throughput f we compare t wth other multport gateway regons n Europe [1]. The purpose of ths paper s to present some smple analytcal methods whch we have used to detect some nternal connecton among the ports n the North Adratc. In partcular, we shall present a market share analyss, shftshare analyss and a smple Markov chan method to predct the behavour of the NA port system wth respect to the growth of or decrease n the contaner traffc. 2 Contaner Throughput Data n the North Adratc Ports The startng ponts of the analyss are the data of the total contaner traffc n the NA ports n the perod from 199 to 212. The data have been collected from the ports annual reports and are shown as graphs n Fgure 1. A quck regresson analyss has shown that n these twenty-two years the total contaner traffc ncreased almost exponentally, on an average between 6 % and 7 % per year, but the rate has vared among ports. It can be notced from Fgure 1 that an almost exponental growth has been realzed by the ports of Koper and Vence, where the overall fastest throughput has been recorded at the port of Koper, at an average of 11 % per year, whle at the port of Vence the growth has been approxmately 7 % per year. The throughput n the rest of the ports has been more rregular. It can also be notced that, at the port of Ravenna, the traffc has barely ncreased at all, whle the mnmum throughput has been realzed and remaned at the port of Rjeka. The port of Rjeka lost a great deal of traffc n the perod from 1997 to about 23, when the ncrease n the Rjeka s Rjeka Koper Treste Veneca Ravena 6. 5. 4. 3. 2... Fgure 1 Evaluaton of the contaner throughput for the NA ports n the 199-212 perods 362 POMORSTVO Scentfc Journal of Martme Research 27/2(213) str./pp. 361-367
contaner throughput was more n lne wth that of Koper, Treste, and Vence. In the year 28 and 29, when the fnancal and economc crses started, the throughput n the North Adratc ports decreased for about 15 %, but n such a stuaton only the port of Vence stll had a slow growth (5 %). The largest drop n traffc was recorded n Treste, a decrease of more than 58, s (17.5 %), though by percentage Rjeka fared worse, declnng at the rate of 22.5 % (38, s fewer). 3 Competton/Cooperaton between the North Adratc Ports As NA ports operate n a much closed system, they have to cooperate and compete among each other at the same tme and a phenomenon of co-opetton s a natural way for survvng. Branderburger et al [2] defned coopetton as a mx of the verbs cooperaton and competton. It s a synonym for the wn-wn strategy for ports whch are very close one to another and have to cooperate, but at the same tme are n competton for the market share. In what follows, we have defned the market share of a port n a partcular year: s = 1,..., N N j= 1 j ( ) (1.1) where s s the market share of -the port, N s the number of ports and s the contaners throughput n. In Fgure 2, the dynamcs of the market share for the NA ports n the last twenty-three years have been presented. There s a very nterestng stuaton between the ports of Vence and Treste they have almost mrrored shares and the same thngs may be observed f we compare the port of Koper and Ravenna. However, as opposed to the ports of Vence and Treste, where when one port loses a contaner throughput the other one takes t over, the share of the port of Koper has practcally ncreased over the years, whle, n the same perod of tme, the share of the port of Ravenna has decreased. Over these years, the port of Ravenna lost ts leadng poston n the contaner traffc and was forth-placed among the NA ports. Unfortunately, t seems that the port out Rjeka s out of ths game. Rjeka Koper Treste Veneca Ravena 4. 35. 3. Market share % 25. 2. 15.. 5.. Fgure 2 The evaluaton of the market share n the contaners throughput for the North Adratc ports (199-212) POMORSTVO Scentfc Journal of Martme Research 27/2(213) str./pp. 361-367 363
5. 4.75 4.5 1/HH 4.25 4. 3.75 3.5 Fgure 3 The evaluaton of the recprocal HH ndex for North Adratc ports (199-212) To enhance these observatons we have used the HH ndex whch s defned as [3] N HH= s (1.1) = 1 2 It can be seen that the recprocal of the HH ndex measures the number of effectve ports n the regon. The evaluaton of ths ndex over the past two decades s shown on Fgure 3. As revealed n the graph, on an average, the port system acts as there have been only four ports nvolved. 4 Shft-share Analyss In ths chapter, the shft-share analyss s presented accordng to the methodology proposed by Notteboom [4]. In essence, the shft-share analyss represents the analyss of the absolute growth of the contaner traffc (ABSGR), the share effect of the port (SHARE) and the total shft of the port (SHIFT) over the years. These quanttes are defned as follows ABSGR = (t) (t ) (1.2) SHARE ( t ) ( t ) 1 ( t ) (1.3) TOTAL = TOTAL SHIFT = ABSGR SHARE (1.4) where ndex stands for -th port, TOTAL (t) s the total port throughputs n the perod t t, t s the current tme (year) and t stands for the ntal tme (year). Whle ABSGR s a self evdent varable, SHARE and SHIFT varables deserve a bt of explanaton. Accordng to Notteboom [4] share reflects the expected growth of contaner traffc as f t would smply mantan ts market share whle shft reflects the total number of contaners a port has actually lost.e. the dfference between the expected throughput and the actual throughput. In contrast wth Notteboom, who has used a four-year perod n hs analyss of the European ports [1], we shall present the analyss of the NA ports competton dynamcs n the course of one year usng the SHIFT varable. On Fgure 4, the evaluaton of the shft of contaners to the NA ports durng the last twenty-three years has been presented. All graphs have shown a knd of an oscllatory behavour out of whch we can see a dynamc of the competton among ports for the contaner market. There are also several peeks that can be seen on the graphs. Thus, n 1998, the port of Rjeka suffered a relatvely great shft of the expected contaners to other ports, but the stuaton recovered n 23 and 27. Italan ports have shown an oscllatory shft of contaners approxmately ranged between ± 3.. A bt dfferent stuaton has been for the port of Koper where we can see that, for a very long perod, namely from 1993 to 211, the port ddn t lose any expected contaners. However, t may be seen from the graphs that the shft 364 POMORSTVO Scentfc Journal of Martme Research 27/2(213) str./pp. 361-367
5 Port Ravena 5 Port Veneca 4 3 3 2 - -2 - -3-5 -3-7 5 Port Treste 3 - -3-5 -7 Port Koper 3. Port Rjeka 8 2. 6. 4 2. -. -2. -2-4 -3. -4. Fgure 4 Dynamc (one year) shft of contaners n dfferent NA ports over the 1991-212 perod of tme process has recently becomes a bt unstable. Namely, n 212, the port of Ravenna accommodated more contaners than ever n observed perod, whle, for the frst tme over a long perod of tme, the port of Koper lost the expected number of contaners. 5 The Markov Chan Model of the Contaner Throughput The queston now arses how to predct the behavour of the contaner traffc n NA ports. We want to know, n partcular, f the market wll grow or wll t fall n, for example, the next few years. The usual approach to answer such questons n terms of quantty s to analyse the data by sophstcated tme seres econometrc methods. However, ths paper wll show use the Markov-chan model by whch we wll estmate transton probabltes between two possble states: the state when the total throughput s growng and the state when the total throughput s fallng. Thus the contaner traffc growth rate ndex (CTR) can be defned as ( + 1) - ( ) ( t ) TOTAL t TOTAL t CTR = TOTAL (1.2) POMORSTVO Scentfc Journal of Martme Research 27/2(213) str./pp. 361-367 365
25 2 15 Growth Rate Index n % 5-5 - -15-2 Fgure 5 Contaner traffc growth ndex for NA ports n the 1991-212 perod of tme where TOTAL (t ) stands for the total contaner throughput n a year t. The evaluaton of the ndex over the observed perod s shown on Fgure 5. A partcular state s now smply detected by comparng two successve values of the ndces. If CTR 1 < CTR then the state at tme s the state of a traffc growth and when CTR 1 > CTR the state at tme s the state of a traffc fall. Out of these graph, we can now count the number of transtons between the states: we have 5 transtons from the state of traffc growth to the state of traffc growth; we have 7 transtons from the state of traffc growth to the state of traffc fall; we have 6 transtons from the state of traffc fall to the state of traffc growth; we have 2 transtons from the state of traffc fall to the state of traffc fall. Hence t follows that the calculated probablty to reman n the state of the marked growth s 42 % and the one to transt to the state of the market fall s 58 %. If we are n the state of the market fall then the probablty s 25 % to stay n ths state over the next year and 75 % refers to the probablty to go back to the state of the market growth. The Markov chan model wth these transton probabltes and the Markov chan model wth the transton probabltes obtaned from 1991 to 22 are shown on Fgure 6. Wth ths model we can do some future 1991-22 1991-212.5 Growth.5 Fall.25.42 Growth.58 Fall.25.75.75 Fgure 6 The Markov chan model for the growth or fall of the relatve rate of the total contaner throughputs n the NA ports. 366 POMORSTVO Scentfc Journal of Martme Research 27/2(213) str./pp. 361-367
assessments. For example, the probablty that the market fall for two successve years s around 6 % and that the fall remans successvely three years the probablty s around 2 %. The probablty that the market grows n the two successve years s about 18 % and about 7 % f the market grows n the three successve years. It s obvous from ths that the NA ports system wll most probably oscllate between the two states every year. 6 Conclusons The NA ports of the multport gateway regon of the North Adratc have a very good locaton especally for contaners arrvng from the Far East and ntended for the market of Central and South-East Europe. Even f these ports have modern contaner termnals, they are facng lower rates of contaner throughput than the rest of the multport gateway regons n Europe. The market potental for the NA ports n the contaner market n 23 appears to be ambtous n terms of the absolute growth as t mples +348 % traffc growth out of the year 2 as compared to the 73 % growth n the market as a whole and, n terms of the market share, growng from the current 4.3 % to reach 11.3 % n 23 [5]. Every port n the NA port regon s tryng to ncrease ts throughput but not all are successful n dong so. Co-opetton n ths multport gateway regon s very much present as these ports share the same hnterland. It s also evdent that the shft of the contaner throughput and of the market share s strongly emphaszed between them. The paper has revealed a smple model that wll be of great help to forecast the state of the NA ports for the next year. Among others, t s shown that the most probable state of the growth of throughput s annually alternatng between grow and fall, whle, wth a relatvely hgh probablty, these ports wll have a contaner growth. The authors are aware that more data should be avalable to buld more relable models. References [1] Notteboom T., Concentraton and the formaton of mult-port gateway regons n the European contaner port system: an update, Journal of Transport Geography 18 (2), 567-583 [2] Branderburger A., Nalebuff B., Co-opetton, 1996, Doubleday, New York [3] Martn, S., 22., Advanced Industral Economcs, Blackwell Publshers Ltd, Oxford, UK [4] Notteboom T., Concentraton and load centre development n the European contaner port system, Journal of Transport Geography 5 (1997), 99-115 [5] www.portsofnapa.com POMORSTVO Scentfc Journal of Martme Research 27/2(213) str./pp. 361-367 367