Forecasting Trade for Port Projects in the Developing World

By Terence D. Smyth,

Introduction

This paper is an extension of a presentation of the same name to the Transport Research Board 23^rd Annual Summer Ports, Waterways & International Trade Conference, July 19-22, Seattle, Washington, U.S.A., The paper deals with port traffic forecasting in developing countries, derived from the experience of Seaport Consultants Canada Inc. (Seaport) in East and Southeast Asia. It begins with a summary of the projects on which it is based and then focuses on sources of data for forecasts and data interpretation. The emphasis is on container traffic.

The work behind this presentation was in mid-1992 to mid-1997. It does not incorporate the financial crisis that hit Asia, and especially Indonesia, in the fall of 1997. The crisis is discussed briefly at the end of this paper.

Cases

Seaport has developed port traffic forecasts as part of the following four projects:

• Port of Panjang, Indonesia (1992)
• Port of Tanjung Priok (Jakarta) and Western Java, Indonesia (1995)
• Port of Nanjing, China (1996)
• Port of Tanjung Perak (Surabaya), Indonesia (1997)

The following sections characterize these projects.

Panjang (1992)

Panjang is a regional port on Lampung Bay near the southern tip of Sumatra. The nearest city is Bandar Lampung, a provincial capital with a population of about 0.5 million. Panjang has a natural deep harbor, the only one serving eastern and southern Sumatra. Lampung Bay contains many potential port sites as well as the existing Panjang port.

The port, which handles mix of bulk, break-bulk and containerized cargoes, serves Lampung Province and southern Sumatra in general. It is dominantly an export port, especially for container cargoes. Most import general cargoes come into the Port of Tanjung Priok and are transported by truck to southern Sumatra via the Merak – Bakauheni ferry. In the case of general goods inbound to Sumatra, the ferry traffic volume to Sumatra exceeds the import volume of the port by many times.

Western Java (1995)

Nanjing, China (1996)

Surabaya, Indonesia (1997)

Data for Port Traffic Forecasting

This review is based on the Indonesian projects and a top-down forecast approach. It addresses what data can one find from various sources and how one can use it in a port traffic forecast.

International Data

International economic and trade data can be useful to guide model development, and provide a basis for some projections and a check for the reasonableness the results of forecasts for particular ports. Some data sources are discussed below.

The World Bank’s World Development Indicators, especially the CD-ROM version, is a good source of a wide variety of international data. Because it is broadly consistent, it is valuable for inter-country comparisons. It is also timely: the 1998 edition has 1996 data for many countries. It provides time series back to 1970. Some examples of data from this source are discussed below.

Table 1 provides real gross domestic product (GDP) growth rates for a number of countries in East and Southeast Asia. It shows the high overall growth rates in China and Southeast Asia, and the changing structure of world output, in particular the expansion of manufacturing output in Southeast Asia. Manufacturing output is often a strong indicator of container traffic growth.

Figure 1 shows the changing share of manufacturing output in GDP in Japan (a mature economy), South Korea (a newly industrializing country) and Indonesia (a developing country). Manufacturing in Japan has been in relative decline since 1970, while Korea's peaked in the mid 1980s. The share of manufacturing in Indonesia's GDP has increased steadily since 1970, and particularly rapidly since the early 1980s. As shown below, the Southeast Asian countries with rapid expansion of manufacturing output also have had high growth rates in their container trades.

If one is developing a container traffic forecast for a country with a rapidly-growing manufacturing sector, a question arises about the limits to the relative size of this sector. Figure 2 and Figure 3 show that there is a general relationship between industrial structure and per capita GDP. Industry's share of GDP tends to peak around 45% of total output at per capita incomes in the $3,000 to $5,000 range (1987 dollars), and manufacturing peaks around 30% at per capita incomes in the $2,000 to $4,000 range. While the relationships are quite loose, they indicate that one needs to consider limits to manufacturing growth and to expect that high container growth rates will taper off in future.

Table 2 provides a measure of world trade growth in dollar terms. Even the constant-dollar figures are higher than physical trade growth in most periods. For example, the volume of world trade (i.e., in constant dollar terms) grew at 6% a year between 1990 and 1995. The growth of seaborne trade in physical terms was half that rate for this period (see below).

Another question to ask is if exports tend to level off as per capita GDP increases. Figure 4 indicates that this is the case.

It is useful to have measures of overall world seaborne trade and that in other ports to check port forecasts for consistency and reasonableness. Some examples are:

• World seaborne trade (Table 3). The data in this table show that overall seaborne trade has grown about 3% a year in recent years, and that it has outpaced world economic growth in most periods. In general, one should expect that the growth in general cargo throughput of a port in an industrializing country will exceed that of the country's GDP.

• Table 4 presents container port throughput in selected East and Southeast Asian countries. It shows that the countries with rapidly expanding manufacturing industries such as Indonesia have also had extremely rapid container traffic growth over many years. The rate of container traffic growth in the more mature economies such as Korea and Taiwan has been much lower in recent years. Hong Kong is a special case because it serves much of southern China.

Hong Kong is of particular interest because it is within Asia and in several recent years its port has had the world’s greatest container volume (Table 5). It provides some useful guidelines for other ports. Some observations are:

• It is a huge port. In many recent years, the incremental container throughput in Hong Kong has exceeded the total container traffic of Canada.
• Its containerization rate has continued to climb to reach 60% of total cargo, over 80% of total general cargo and 98% of deep-sea transshipment general cargo.

National Economic Data

The availability of data varies greatly with country. Indonesia has quite extensive collection of statistics and a professional central statistics bureau (Biro Pusat Statistik). Some examples of national economic and trade statistics follow.

Growth in gross domestic product is a key factor in many forecasting procedures. Figure 5 shows the growth in total Indonesian GDP from 1960 to 1996. Following the economic chaos in the last years of the Sukarno regime (1960 to 1966), Indonesia had consistently high GDP growth in most years. The early years (1969 to 1981) reflected the development of the oil and gas industry and high energy prices. The volatility in the early to mid 1980s was a consequence of the decline in energy prices. Beginning in the early 1980s, the country greatly freed up its private sector and manufacturing started to boom. The Indonesian government reports GDP with and without the oil & gas sector.

Indonesia has an investment coordinating board whose role is to approve and report on foreign and domestic investment (Figure 6). While not all approved investment is implemented, the approvals (especially of manufacturing) provide an indication of future traffic growth and in some cases port traffic does correlate with lagged investment approvals.

In nominal dollar terms, the volume of international trade has expanded greatly since 1970 (Figure 7). From the early 1970s to the mid 1980s, oil and gas were the dominant exports. Since the mid 1980s, oil and gas have remained relatively constant while other exports - primarily manufactures - increased rapidly.

The characteristics of Indonesian imports (Table 6) are not what one would expect. They are dominantly raw and intermediate goods for further manufacture and in some cases re-export. Consumer goods were a relatively high proportion of total exports in the early 1980s due to the purchasing power created by high energy prices. Since then they have been only 2% to 5% of total imports, although the proportion was creeping up in the 1990s.

Regional Economic Data

A fair amount of data is available on a provincial basis, although it is not as comprehensive, accurate or timely as the national data. There are provincial counterparts of the national statistics bureau. Some examples with a focus on East Java Province (Surabaya) follow.

The provincial comparison (Table 7) shows that the Jakarta region is by far the most prosperous with a per capita gross regional domestic product (GRDP) over 3 times that of Indonesia as a whole. In terms of population and GRDP, West Java Province and East Java Province are quite equal. The major concentration of economic activity in the country is in the contiguous areas of DKI Jakarta and West Java Province.

The structure of the economy of East Java Province (Figure 8) shows that manufacturing and services have been steadily increasing their shares of GRDP.

Table 8 shows the kinds of manufacturing industries in East Java Province. Most of these industries are relatively simple; it is not (like Nanjing, for example) an area of high value-added industries such as electronics.

Investment approval data is also available at the provincial level. Figure 9 shows that there was a surge of interest in the chemical industry (of which a considerable amount has been implemented) and Figure 10 shows that virtually all of this investment is within 100 km of Surabaya and its port.

Another useful data set at the provincial level is tonnages of imports and exports for the province as a whole. These are not shown here because there is too much detail to summarize in a table. While such data rarely corresponds with port traffic, it is useful to develop the characteristics of imports and exports (such as the overall degree of containerizablity) since the ports as a rule do not keep track of container cargo by commodity.

Port Data

All port traffic data is collected in the first instance by the ports and the port corporations. It eventually finds its way into national summaries.

Figure 11 provides total Indonesian port throughput from the early 1950s to 1996. As with the value of international trade above, the first major cargo in tonnage terms was oil. From the mid 1980s, however, exports of other cargoes and domestic (loading and unloading in the graph) trade grew rapidly.

The Port of Tanjung Priok (Jakarta) is Indonesia's major general cargo port. Some trends in this port were:

• Rapid container traffic growth (Figure 12), with a closer balance between imports and exports in recent years. In the early years, the tonnages of imports greatly exceeded those of imports.
• Generally high growth rates in container traffic since 1980 (Figure 13). Growth was relatively low and volatile in the early to mid 1980s due to oil price fluctuations and economic deregulation. More recently, growth has been tapering off since 1990.
• The port's share of regional and world port throughput (Figure 14). The port's share has been increasing in all cases.
• Changes in the characteristics of the container trade (Figure 15). The percentage of empty containers has been declining as export volumes have become closer to those of imports, and there has been a move toward larger vessels calling at the port (mostly feeder regional trade ships).

Some trends in the Port of Tanjung Perak were:

• Overall port traffic (Figure 16). Note that these are in revenue tons - the way the port records its traffic.
• Growth in international and domestic container tonnages (Figure 17). The growth rates have been high but volatile, especially those of domestic traffic.

• The port's share of regional and world port throughput (Figure 18). As with Tanjung Priok, the port's share has been increasing.
• The empty container balance (Figure 19). The opposite of Tanjung Priok, Tanjung Perak has an excess of export container cargoes and the shipping lines must bring empty containers to the port.

Approach to Forecasts

This paper does not address particular forecasting techniques as they vary greatly from project to project. One must be opportunistic in the overall approach to take advantage of whatever information one can assemble in the time available. The forecasting procedure may be subjective in the absence of adequate data (such as assumptions of growth rates for particular categories of cargoes) or statistical in cases where there is adequate data for formal models.

Some ways to evaluate the results of a container traffic forecast include:

• Is the forecast broadly consistent with the industrial development patterns in its hinterland? These include broad measures of existing industry, investment intentions, development and occupancy rates in industrial estates, and creation of particular industries. Container traffic sources are usually too diversified for an aggregation of output of individual industries to be meaningful.
• Is the forecast consistent with the outlook and plans of the container shipping industry? While shipping companies have a short-term focus, they do know their markets.
• Does it incorporate all the structural elements of container traffic? These include tons of cargo per TEU and percent empty containers for import and export cargoes, recognition of the driving force for container traffic (imports, exports or transshipment), the TEU/box ratio, etc. A number of forecasts by others have not done this.
• Is the overall forecast growth path, including upper and lower bands, consistent with historical trends?
• Are the year-to-year percentage increases of the forecast in line with past trends? There is often more consistency over time among growth rates that of overall traffic volumes. Figure 20 provides an example for international container traffic in western Java .

• What are the port's regional and world market shares? These of course require projections of regional and world markets, which are also uncertain, but total world trade should be more predictable than that of an individual port. A major port in a rapidly industrializing area should capture an increasing share of world container traffic and possibly of its region. Figure 21 provides an example, also for international container traffic in western Java .

Lessons from the Asian Crisis

The data in this paper predates the Asian crisis that began in mid-1997, and the Indonesian forecasts developed from this material for the various projects were based on a continuation of the trends apparent in Indonesia since the mid 1980s. These in turn had a strong underlying basis: continuing deregulation of the real and financial sectors, development of financial markets, considerable foreign direct investment in the country and access to international capital by Indonesian companies.

Our "base case" forecasts incorporated short-term GDP growth rates comparable to those published by the international financial institutions, long-term GDP growth rates somewhat below the official Indonesian forecasts (such as 6% a year versus 7% a year), and a slowly evolving industrial structure to a higher level of manufacturing. They also incorporated the likely trends in container characteristics such as the ultimate degree of containerization and the path toward this value, the tons per TEU and the TEU/box ratio. Variants from the base case involved higher and lower GDP growth rates, different container characteristics and in some cases particular assumptions about cargo tonnage trends. None of the cases involved a meltdown of the Indonesian economy.

Total container traffic (international and domestic) through the ports of Tanjung Priok and Tanjung Perak grew by about 20% in TEU terms in 1997. Domestic growth was considerably greater in percentage terms than international traffic. In the first quarter of 1998, Indonesian container traffic was essentially flat, and the outlook for this year and next is limited to no growth.

Until mid 1997, the real Indonesian economy appeared sound. Saving and investment were high, the country had a decent current account balance, government fiscal policies were conservative (among other things, limited and declining government debt), and the country seemed to be riding a globalization trend. There were few signs of impending financial weakness beyond some cautionary words in the 1997 World Bank country report about "insolvent banks." There was a small property bubble in the Jakarta area, but it was of minor consequence.

What happened?

• First, there was contagion from Thailand. Speculation against the Thai baht flowed over to speculation against the Indonesian Rupiah. Proving susceptible to speculation, the Rupiah fell against the US dollar from some US$1.00 = Rp.2,450 in mid 1997 to a brief low of Rp.17,000 in January 1998 (it was generally in the Rp.13-15,000 in July 1998).
• There really were a number of insolvent banks. Many of these were controlled by the major Indonesian conglomerates and served mainly as pipelines for the flow of funds to the conglomerates.
• Although Indonesia had generally good banking laws, they were not enforced.
• Private Indonesian firms had borrowed some US$80 billion in the international markets denominated in foreign currency. Since much of this was neither hedged nor swapped, many of these borrowers became insolvent as the Rupiah declined. In early 1997, borrowers had a choice of borrowing in Rupiah (20% interest rate), in dollars (10% interest plus 5% historical depreciation in the Rupiah, or about 15% effective interest cost), or in dollars with swapping (15% to 20%). Many simply chose dollars.
• "Crony capitalism" was a major form of business organization in Indonesia with then President Suharto's family and a small number of tycoons in control of much of the economy. Government policy and rulings often favored such companies. Although this was well known, in business terms it did not seem to matter much because many of these companies were quite effective. But they were hardly fair to Indonesians as a whole.
• There was no independent legal system of any consequence. In particular, there were no bankruptcy laws, as lenders discovered late to their chagrin.
• There was a pending transition of power in Indonesia from Suharto. He seemed to be following the Javanese king approach of dying with his boots on.
• Finally, the increasing difficulty of doing business in Indonesia even if one had a real business (such as a manufacturer with a world export market). Banking procedures became onerous, foreign banks refused to recognize letters of credit from Indonesian banks, and empty containers were scarce.

The only way of foreseeing such a crisis would have been to develop alternative scenarios for the Indonesian economy that were far from mainstream thought. It would have been difficult to find any writings on the subject prior to mid 1997. Even the most far-out prognostications for the economy would not have predicted what has transpired.

Port traffic forecasts are often treated as if they have lives of their own. An economist or a study team prepares a forecast and others (a port authority, an investor or a port planner) use the forecast for their own evaluations. Two concepts tend to get lost in the process: that forecasts contain considerable uncertainty, and that there is usually a decision hinging on the forecast. If the decision is to add a berth to an existing terminal, the consequences of forecast error are not great; the berth may be a year or two premature. If the decision involves hundreds of millions of dollars to build a new port, then forecasting results are very important.

In mid 1997, an investor could really have asked only two broad questions about Indonesian traffic forecasts:

1. Will the transition of power in the country have any major effect on its economy? The answer would probably have been that it could, but a reasonably smooth transition is likely. It would not have stopped an investor.
2. Is there anything that would change the underlying factor behind Indonesian container trade growth: globalization and the shift of basic manufacturing to the developing world. The answer would probably have been that there will be limits on this trend but that it will continue for a number of years and then taper off. Competition from China or Vietnam and saturation of the world market for basic manufactures could place such limits.