The paper is structured as follows. Section 2 outlines the literature focusing mainly on the relationship between the ICOR and both of per capita income and growth rate. Section 3 provides empirical evidence on the levels and trends of ICOR and their relationship with growth and income factors, in a number of Asian economies and Thailand. Section 4 summarizes the analytical results and concludes. 2. Literature There have been a number of the studies to investigate the international variations in the ICOR as we stated in the introduction, and we herein summarize main literature on these studies. It seemed to be a study by the U.N. Economic Commission for Latin America in 1955* that the ICOR was utilized for the first time mainly to obtain estimates of the productivity of capital and of the investment required to attain a targeted level of income in the economy. Since then, most of the efforts have been concentrated on the investigation of the relationship between the ICOR and both of per capita income and growth rate. Kuznets (1960) verified the existence of a positive association between the ICOR and the level of per capita income, as a part of his intensive works on \"quantitative aspects of the economic growth of nations\". Based on the observation on international variations in the ICOR in the period of 1951-57, it found that the gross ICOR tended to be lower in the lower-income group of countries and higher in the higher-income group ranging from 2.5 to 7.3; the net ICOR showed the same tendency ranging from 1.8 to 5.3; and the inter-income-group differences in the ICOR was not coming from the intergroup differences in their industrial weights, but from the intergroup differences in each industrial ICOR, i.e., the ICOR of agriculture, manufacturing and services. Kuznets (1961) also observed a general tendency of the gross and net ICOR to rise over the changes of time. These findings imply that the ICOR is an increasing function of the level of per capita income across counties and times. With regard to the relationship between the ICOR and growth rate, several studies provided the evidence on their inverse relationship. Ohkawa and Rosovsky (1962) identified their inverse association by showing a series of graphs on the annual growth rate of GDP and the ICOR for Japan in the period from 1931 to 1980 with seven-year moving average. Leibenstein (1966) also confirmed the existence of an inverse relationship between the ICOR and growth rate as a general tendency across 18 countries, while Beckerman (1965) identified this inverse association in 10 European countries and the U.S.A. during the period from 1956 to 1962. The relationship between the ICOR and both of per capita income and growth rate were examined systematically in the theoretical and empirical senses by Vanek and Studenmund (1968) and Sato (1971). Vanek and Studenmund (1968) constructed the simple theory on 3 stationarity of all the series by using the unit root test as a usual procedure. In this study, we conduct the Ng and Perron test' on the null hypothesis that the level and the first difference of each variable has a unit root by choosing to include only an intercept in the test equation. This test constructs four test statistics that are based upon the de-trended data. These test statistics are modified forms of Phillips and Perron statistics (MZa, MZt), the Bhargava (1986) statistic (MSB), and the ERS Point Optimal statistic (MPT). Table 7 reports that the null hypothesis on a unit root was rejected at more than 95 % significance level in the level and first difference of the net ICOR, the first difference of per capita GDP and the level of GDP share of service sector (sv). Thus, to make a regression valid, we modify the (N.1) and (N.2) for their estimation into the following ones by using only stationary time-series data. d (The net ICOR) = const. + o * d (logy), o>0 (N.1) d (The net ICOR) = const. + o *d (log y) +v *sv, >0 (N.2") Estimation Outcomes with Discussions Table 8 indicates the results of estimating the equation (N.1) and (N.2"). We insert the dummy variable for the period of 1997-2002, in which the impacts of the 1997 currency crisis entailed the abnormally high level of the net ICOR, more than 4.0. Main findings are as follows. First, the coefficient of the first difference of per capita GDP, d (log(y)) is significantly positive in both equation (N.1") and (N.2') as expected in the theoretical model. Second, the coefficient of GDP shares of service sector, sv, is insignificant in equation (N.2'). Though it may include multicollinearity problem like the Asian analysis, it may mean that the difference in service-sector share does not give any impacts on the difference in the level of the net ICOR, which is also consistent with the outcome of Kuznets (1960). We herein investigate further the impacts of sector's shares on the net ICOR by the more direct way that Kuznets (1960) adopted, since Thailand has the data of sector's capital stocks'\1. Introduction This paper aims to examine the trend in the incremental capital-output ratio (hereafter ICOR) and its relationship with per capita GDP and GDP growth rate by utilizing the panel data of a number of Asian economies and the historical time-series data of Thailand. The ICOR is one of the most important concepts and analytical instruments of both economic growth theory in academic circles and development planning in policy makers. The ICOR is defined as a ratio between incremental changes in capital over incremental changes in output. As a traditional literature, the Harrod-Domar Growth model , in which the rate of growth of output is determined by the rate of saving and the ICOR, suggested that the ICOR could be a key variable to link investment requirements with targeted rates of economic growth. Kuznets (1960) also presumed an association between the two variables in such a way that "the additions to output require the additions to reproducible capital stock, or the latter permit (not necessarily guaranteeing) the additions to output", and "the association between reproducible capital and output may be sufficiently strong to warrant an interest in and examination of the ratio between the two". Thus, a number of studies on the regularities in variations of the ICOR have been conducted as described in the following section. The use of the ICOR appears to be still valid in present times, in particular, in less-developed countries. It is because less-developed countries regard capital stock as an important factor in explaining economic growth due to its scarcity in their economies and also because they often suffer the lack of data appropriate for sophisticated model construction for economic projections. This paper revisits the ICOR issue by focusing on the case of a number of Asian economies and Thailand historical trends. Asia has recently been a growth center in the world. For its future, the Asian Development Bank (ADB) presented the scenario called the "Asian century", in which Asian share of global GDP will nearly double from 27 percent in 2010 to 51 percent by 2050. One of the driving forces for Asian economic growth has been the development of international production and distribution networks in manufacturing sectors (see e.g. Kimura, 2006), and the formation of the networks has required intensive investments in terms of infrastructure and production facilities. The linkage between growth and investment, therefore, has been of great concerns for policy makers as well as academic circles in recent times.' In particular, such latecomer's economies as Cambodia, Lao PDR and Myanmar in Mekong region have faced serious needs for heavy investments under their conditions of scarce capital stocks to attain a targeted economic growth. Here comes the necessity to explore and signify the ICOR levels whose variations are standardized by such theoretically-related variables as per capita GDP and GDP growth rate among Asian economies. Knowing the standard ICOR levels may also contribute to assessing the investment efficiency and the productivity of capital stock in specific economies. In this sense, the ICOR would be an old but still new issue for economic growth theory andprices\".\" For \"output ()\A Revisit to the Incremental Capital-Output Ratio: The Case of Asian Economies and Thailand Hiroyuki Taguchi " Suphannada Lowhachai B Abstract This paper aims to examine the trend in the incremental capital-output ratio (ICOR) and its relationship with per capita GDP and GDP growth rate by utilizing the panel data of a number of Asian economies and the historical time-series data of Thailand. It might be significant to know the linkage between growth and investment through the ICOR level, since Asian economies have faced serious needs for heavy investments to attain a targeted growth. The panel-data analysis confirmed that the gross ICOR had a positive correlation with per capita GDP and a negative association with GDP growth rate as expected in a theoretical model. The time-series analysis verified that the net ICOR was positively correlated with per capita GDP. Both analyses showed that industrial shares did not affect the level of the gross and net ICORs. Keywords: Incremental capital-output ratio (ICOR), Asian economies, Thailand, investment requirement, per capita GDP, GDP growth rate, panel data, historical time-series data, gross ICOR and net ICOR. " Hiroyuki Taguchi is an advisor (an expert of Japan International Cooperation Agency) belonging to Competitiveness Development Office, Office of the National Economic and Social Development Board, Government of Thailand (962 Krung Kasem Road Pomprab District, Bangkok 10100, Thailand, E-mail: taguchi@nesdb.go.th, Fax: +66 2281 1821); and a deputy director-general, Cabinet Office, Government of Japan; Ph.D. in Social Sciences from Waseda University. His research focuses on Asian economies and international economics. Suphannada Lowhachai is an official in National Accounts Office, Office of the National Economic and Social Development Board, Government of Thailand (962 Krung Kasem Road Pomprab District, Bangkok 10100, Thailand, E-mail: Suphannada@nesdb.go.th, Fax: +66 2281 1821). She is in charge of the Flow-of-Funds Accounts in National Accounts Office; PhD in Economics from School of Oriental and African Studies, University of London. Her research focuses on development economics of ThailandTable 5 Gross ICOR Standardized by per capita Income and Growth Rate (G.1') Growth Rate (%) 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 per capita Income (US dollar) 500 4.90 4.48 4.07 3.65 3.23 2.81 2.39 1.98 1.56 1,000 1.14 5.01 4.59 4.17 3.75 3.34 2.92 2.50 2.08 1.66 1,500 1.25 5.07 4.65 4.23 3.82 3.40 2.98 2.56 2.14 .73 2,000 1.31 5.11 4.70 4.28 3.86 3.44 3.02 2.61 2.19 1.77 2,500 1.35 5.15 4.73 4.31 3.89 3.48 3.06 2.64 2.22 1.80 1.39 3.000 5.18 4.76 4.34 3.92 3.50 3.09 2.67 2.25 1.83 3,500 1.41 5.20 4.78 4.36 3.95 3.53 3.11 2.69 2.27 1.86 1.44 4,000 5.22 4.80 4.38 3.97 3.55 3.13 2.71 2.29 1.88 4,500 1.46 5.24 4.82 4.40 3.98 3.57 3.15 2.73 2.31 5,000 5.26 1.89 1.48 4.84 4.42 4.00 3.58 3.17 2.75 2.33 1.91 1.49 5,500 5.27 4.85 4.43 4.02 3.60 3.18 2.76 2.34 1.93 6,000 5.28 1.51 4.87 4.45 4.03 3.61 3.19 2.78 2.36 1.94 6,500 1.52 5.30 4.88 4.46 4.04 3.62 3.21 2. 79 1.95 7,000 2.37 1.53 5.31 4.89 4.47 4.05 3.63 3.22 2.80 2.38 1.96 7,500 1.54 5.32 4.90 4.48 4.06 3.65 3.23 2.81 2.39 5.33 1.97 8,000 1.56 4.91 4.49 4.07 3.66 3.24 2.82 2.40 1.98 8,500 1.57 5.34 4.92 4.50 4.08 3.66 3.25 2.83 2.41 9,000 5.35 1.99 4.93 4.51 1.57 4.09 3.67 3.26 2.84 2.42 2.00 9,500 1.58 5.35 4.94 4.52 4.10 3.68 3.26 2.85 2.43 2.01 10,000 5.36 4.94 1.59 4.53 4.11 3.69 3.27 2.85 2.44 2.02 1.60 21When we then simply combine the equation (1.1) and (N.1) or (N.2) for the gross-ICOR estimation under the assumption that the depreciation ratio o and capital-output ratio K/Y are stable in the long run among the sample economies, we can get the following equations. "GOR = const. +a * logy + B * g, a>0, B0, B0 (N.1) Alternatively, we attempt to add such explanatory variables as output shares of industrial sectors in the net-ICOR equation. The previous studies have represented ambiguous impacts of industrial shares on the ICOR; Kuznets (1960) and Vanek and Studenmund (1968) did not extracted clear-cut or expected effects of industrial shares on the ICOR, while Sato (1971) emphasized that changes in industrial structure are more important than changes in income level as the factor to explain the ICOR. Thus, it seems to be significant in this study to re-confirm the industrial-share impacts on the ICOR. When we let sh the industrial sector's share in output, the alternative equation will be as follows. The net ICOR = const. +a * logy ty * sh, a>0 (N.2) When we then simply combine the equation (1.1) and (N.1) or (N.2) for the gross-ICOR estimation under the assumption that the depreciation ratio S and capital-output ratio K/Y are stable in the long run among the sample economies, we can get the following equations. The gross ICOR = const. + a * logy + B * g, a>0, B0, B