Asian Historical Statistics Project: Difficulties and Expectations

Shinichi Ichimura


A few years ago, when I first heard that a research project called the Asian Historical Statistics was going to be organized, I certainly expected outstanding results from the project. This was because I was well aware of the great impact on the studies of long-term economic growth not only in Japan but also abroad that the previous Hitotsubashi project, the Long-term Economic Statistics of Japan had made. At the same time, as someone who has applied himself to quantitative analysis of the economies in Southeast Asia for 40 years, I foresaw that there would be many difficulties faced by the project. This feeling was similar to Prof. Simon Kuznets' assessments described by Prof. Harry Oshima in the Project Newsletter No. 10 "Realizing Simon Kuznets' Dream" (October 1997). Upon the request of the Ford Foundation, Kuznets embarked on a trip to explore the possibilities of studying prewar Asian economic conditions soon after the World War II. However, his conclusion was that with the exception of Japan, this would not be a feasible project. He suggested to the Ford Foundation that before undertaking any economic analysis, it would be necessary to start with training statistical survey specialists in the various Asian countries. Prof. Oshima evidently viewed the Hitotsubashi project as a realization of the dream that Kuznets had abandoned over 40 years ago. How did the results of the Asian Historical Statistics Project turn out? Below I would like to outline my experiences in working with Southeast Asian statistics in the hopes that it may be of some use in answering this question.

1. Statistical Conditions in Prewar and the Immediate Postwar Periods

In 1969, soon after I assumed the directorship of the Southeast Asian Research Center at Kyoto University, I visited several countries in Southeast Asia and thoroughly assessed the possibilities for empirical economic analysis. At the time, my conclusions were not considerably different from those reached Kuznets a couple of decades before. Consequently, I decided to begin by undertaking the training of specialists and conducting statistical projects myself. The first fruit of these efforts was Socio-Economic Survey of South Sumatra (1975), and the second was the construction of an Interindustrial Input-Output Table for Indonesia. The biggest supporter and collaborator in the first project was Mizuno Hiroshi of the Statistical Bureau of the Prime Minister's Office, while the late Prof. Kaneko Takao was the most active collaborator in the second project. On the Indonesian side, I was fortunately associated with the late Dr. Harsja Bachtiar (Director of the Social and Economic Research Institute at the time) and his staff as well as a number of other talented specialists from the Central Bureau of Statistical. (I had started the training of these people during my days at Osaka University at the request of Minister, Dr. Widjojo). From these two projects, we obtained a detailed understanding of the state of Indonesian statistics. Conditions have improved since then, but for long-term statistics, the problems from the past periods still remain; therefore, it is not difficult to imagine that difficulties would arise.

In the early 1970s, while calculating Indonesia's national income and manufacturing production statistics, I realized a few points. One of them was that Indonesian statistics at the time retained formats used during Dutch colonial rule and had not changed significantly since independence. This point has been highlighted recently in the 16 volume compilation by a group of Dutch scholars, A Selection of Statistical Source Material from the Early 19th Century up to 1940(1975-1995). Thus, to analyze prewar and postwar (basically up to around 1975) statistics, the cooperation of Dutch scholars is essential. Just as it is necessary to have the cooperation of Japanese experts to analyze Taiwanese and Korean statistics, mobilizing specialists from former colonial metropoles is required in many instances for Southeast Asian countries.

2. National Income and Input-Output Tables

In Indonesia at the time, there were hardly any statistical estimates of consumption in national income statistics or statistical estimates for value-added in the tertiary industries. Around 1975, a household budget survey (SUSENAS) was conducted and as a result, consumption became no longer a category assumed from leftover amounts. At the same time, with the completion of a commercial census, at least the category of errors and omissions in national income statistics came to have some meaning. Therefore, the national income statistics for Indonesia up to the 1980s must be analyzed with great care. For the South Sumatra survey, consumption and commercial activity value-added was estimated based on a sampling system entirely designed by Mr. H. Mizuno. I believe that this outstanding design made at least some contribution to the development of Indonesia statistics. Among the people who shared the pains and tribulations with us back then were, Mr. Sugito and his staff at the Central Bureau of Statistics, as well as Dr. Thee Kian Wie and Dr. Jonker Tamba and several others at the Social and Economic Research Institute (LEKNAS). These people are still in good health. Ideally, statistics from the 1970s and the 1980s can be organized while these and others who were actually involved in the surveys are still in good health. The same can be said for statistics in other countries as well.

Initially, when we decided to make an input-output table for Indonesia, the whole undertaking looked like an impossible task. However, we believed that if we just went ahead with everything aggressively, we could discover the "holes" in Indonesian statistics. Thus, Prof. Kaneko and I went ahead with the project. Our project was based on the assumption that we would be able to apply information from the Philippines IO table, which had been completed just a little earlier. The IO table was completed by the staff of the government statistical office whom I had trained at Osaka University at the request of the Philippines government a few years before. From them, we obtained the most detailed Philippines IO table and used the input coefficients to estimate input amounts for each industry in Indonesia. Then, we checked its correspondence with industry-specific production levels, and made efforts to adjust them. The difference was at first as much as around 20-30% of each industry's production amount, but we were not surprised, because we already had some precedents in Japan's first IO table for 1951. In the process of preparing the first table, rice output was reported as 60 million koku (1 koku=4.96 bushels), and the demand was estimated as 70 million koku. We recalled the Agriculture and Forestry Ministry officially stating that the difference of 10 million koku was accounted for as "broken rice." Our IO table surprised Prof. Alan Strout, who was a visiting professor at the University of Indonesia from the Unites States, and through him, our project became well known among government officials and foreign embassies in Jakarta. This praise circulated and augmented the trust of the Japanese team operating under Indonesian Bureau for National Development, whose minister was Dr. Widjojo. Shortly after, in 1975, the Indonesian government decided to compile Industrial Census as a part of official statistics, and this made possible the calculation of a full-scale IO table later. The person who supervised this project was Mr. Mihara of the Institute of Developing Economies.

As we can see from above, while there are differences in conditions by country, prewar statistics and those for 1945-1975 in Asian countries are not plentiful, and the numbers must be ascertained and analyzed with great care. Moreover, such an undertaking should be done quickly, before people who were involved at the time pass away. This Hitotsubashi project is probably the last opportunity in terms of timing to check with such individuals. I understand that this project is soon to be closed down, but my hope is that it can be continued as several smaller projects, each focusing on the data of individual countries.

3. Problems in Trade and Investment Statistics

After the mid-1970s, the quality of economic statistics in various countries began to improve rapidly. There were problems with reliability of surveys, and survey coverage expanded only step by step, but overall, by now the statistics have neared levels where they were usable. However, as is well known, Southeast Asian trade statistics contain serious problems. In particular, the problems of entrepot trade and black market trade are important. In Southeast Asia, with two large entrepot trade ports in Singapore and Hong Kong, trade cannot be discussed without taking into account the role of these two ports. Singapore does not publish the contents of its trade with Indonesia. From what I have heard, the Singapore government actually keeps track of the trade statistics. The Indonesian government has made its trade statistics with Singapore public, but the numbers seem too low. Most likely, the scale of the black market trade, which is not included in the official statistics, is quite large. Hong Kong's data does not have such problems, and import-export and re-import-export are listed in separate categories. But the reliability of these statistics for previous years needs to be checked. I have not heard of any such projects being started. There is considerable black market trading in other countries as well. In all likelihood, the farther back into the past we go, the more such trade must exist. The world trade matrix compiled by the Institute of Developing Economies in cooperation with the United Nations, I have heard, includes estimates for trade between Singapore and Indonesia, but I wonder how far into the past these numbers can be generated.

Furthermore, with investment statistics, there are large amounts that have been omitted from official statistics. Even in countries such as Taiwan which has highly reliable statistics overall, when it comes to investment, there are hardly any reliable statistics. Since this is the case for the present, quality of statistics are probably worse in the past. In addition, for Indonesian investment statistics, oil development-related statistics are completely missing. As a result, the amounts for US investment into Indonesia turn out extremely small. The World Bank and the International Monetary Fund often call for greater transparency in banking and financial practices, but pretend not to see the absence of transparency in oil-related statistics. To study Indonesian economic growth without oil-related data is to overlook half of its long-term development story. With such areas as trade and investment, there are many limitations in estimating back into the past. Nevertheless, some effort is necessary, as these are important points that shed light on the inner dynamics of economic development.

4. Reliability of Public Finance Statistics

Finally, I would like to make a few comments on the low level of reliability of government public finance statistics. For example, with the help of experts from the Philippines, I once attempted to estimate how many dollars had been taken out of the country through the corrupt practices of the Marcos group in the last years of the Marcos era. The conclusion was that some US $5-7 billion had been confiscated out, but what we also realized was that many cases which were recorded as government investment in the budget turned out to be government consumption in actuality. In other words, we had numbers only for the budget but not for the reliable settlement of accounts. In consequence, the marginal capital coefficient ICOR grows very rapidly. This clearly occurred in the late-Marcos period. Of course, estimations of national income statistics should be based on settlement of accounts, but it is not possible to obtain accurate statistics on this point. Within categories of aid money and extra-budgetary expenditures in the governments of many countries, large amounts are linked with corruption. However, it is very difficult to generate a clear picture of the exact amounts of expenditures. Nevertheless, without such numbers, accurate and meaningful estimation of GNP cannot be conducted. For the period of 1945-1975, and even afterwards, the principal wellspring of corruption can be said to lie in this opaqueness in the statistics.

(Director, International Center for the Study of East Asian Development)