Colonialism, Learning and Convergence: A Comparison of India and Taiwan

ABSTRACT

This article combines the traditional learning model with recent theory of economic growth using Maddison's long-run real GDP per capita data. We first discuss the performance of India and Taiwan in the world economy. While economic theory suggests that India could have grown faster than Taiwan, this was not the case after the Second World War. We identify differences in terms of learning in the development experience of the two countries under colonial rule. The colonial distortion in industrial and agricultural development appeared to be more important in India than in Taiwan. The most important factors, however, were health and education. We then derive learning coefficients of India with the United Kingdom, Taiwan with Japan, and Japan with the United States through the pre-war and post-war periods. Long-run coefficients of learning are estimated. Using the conventional unit root-based test of convergence, we show that the model of learning leads to a logistic model of economic growth. We then confirm that real GDP per capita of Taiwan was converging to that of Japan, whereas that of India showed only slight convergence to the United Kingdom or Taiwan. Our results underscore the importance of studying colonial experiences of the two countries.

Keywords:

• Colonialism
• learning
• convergence
• economic development
• India
• Taiwan

JEL CLASSIFICATIONS:

• 011
• 053
• 057
• N25
• C32

Acknowledgements

We are grateful to Professor Thomas Lindblad, Professor Angus Maddison and other participants at the Groningen conference in 2003 for comments and suggestions. Earlier versions of this article were presented at a joint session of the 2002 Annual Convention of the Allied Social Science Associations (ASSA)/Association of Indian Economic Studies (AIES) in Atlanta, GA, and also at Kansai University in Osaka and the International Centre for the Study of East Asian Development (ICSEAD) in Kitakyushu, Japan. We are indebted to participants in these seminars and also to Professors Ann Carlos, Shinichi Ichimura, Teng-hui Lee, Hiro Lee, Robert McNown and Naci Mocan, Eric Ramstetter and Ron Smith for help in reference search, discussions and suggestions. As usual, all errors of omission and commission are ours.

Notes

**Denotes significant at the 5% level.

***

**

*

***

**

*

***

1. The original source is the Market Intelligence Center of the Institute for Information Industries in Taiwan. According to this institution, ‘Taiwan IT companies produced more than $19 billion in hardware products in 1998. Adding offshore production, the total reached $32 billion, an increase of 8.4% over 1997’ (Underwood, 1999, p. 22).

2. For a discussion of the Geary-Khamis method and references, see Kravis et al. (1982, pp. 89–93). During the 2003 conference at Groningen, Angus Maddison pointed out that updated data series are available up to 2002. However, in order to avoid possible distortions due to the Asian financial crisis and since this paper is concerned with long-run convergence, we kept the original series running from 1911 to 1992 or 1994.

3. Lee was a consultant of the Sino-American Joint Commission on Rural Reconstruction. The scholar-cum-politician became president of Taiwan in 1988 and was re-elected for another four years in 1990.

4. The growth rate here is defined as the rate compounded at the end of each year, g = (X _t − X _{t − 1})/X _{t − 1.}

5. Compared with Korea and Japan, the illiteracy rate in Taiwan in the early 1950s was unusually much higher. It is not clear whether this is due to the fact that the new Chinese émigré regime after the war considered those Taiwanese who could speak only Japanese but not Chinese as illiterate. More study is called for.

6. Ratios calculated by dividing the original date by the mid-year population of 1950 (UNESO, 1963, p. 33).

7. In 1959/60 the former President of Taiwan, T. H. Lee, then a member of the Joint Committee of Rural Reconstruction (JCRR), a US aid organization, remarked to a group of graduate students at the National Taiwan University (including the first author of this paper) that Taiwanese farmers read Japanese magazines even before the JCRR experts, learning how to diversify and improve their methods of farming (Hsiao & Hsiao, 1996, p. 240, n. 32; on the use of the Japanese language in rural Taiwan, see Tsurumi, 1977, p. 154). This is in contrast to reports that the Japanese used the police force to compel Taiwanese farmers to shift to new rice seeds (Ponlai rice) during the initial stage of the Green Revolution in the 1920s. This policy at any rate did alter attitudes among Taiwanese farmers. In India, however, ‘the village was the great fortress of conservatism. … Without benefit of education, … suspicious of new methods of agriculture, the Indian ryot, or peasant, dimly appreciated the fact that his standard of living was tragically low, but at the same time he was often the despair of those who tried to improve his lot’ (Wallbank, 1958). Around 1969, the following remark was made by a Chinese mainland resident of Taiwan in his 50s: ‘Those stupid Taiwanese. All they learned under the Japanese was to line up in queues’ (Tsurumi, 1977, p. 156).

8. Technology may also be transferred through foreign direct investment and cost-free diffusion (Gomulka, 1990, p. 161). Our analysis here is consistent with the emphasis on the role of technological progress and human capital development in achieving productivity growth and a process of catch-up as emphasized by Ambramovitz in an interview with this scholar (Abramovitz, 1999).

9. In this interpretation, the constancy of r is unrealistic as it implies that the least developed country is the most innovative (Gomulka, 1990, p. 160). However, Hsiao & Hsiao (2001b) show that Taiwan and Korea surpassed the threshold of development already during the colonial period, both possessing the ability to absorb foreign technology during the early stage of post-war development.

10. In the estimation process, we apply the parsimony principle by first assuming that β _j is a decreasing linear function (p = 1) to estimate the model. If this assumption does not hold, then we re-estimate the model by assuming that β _j is a concave quadratic function (p = 2). We have estimated the model from a lag length (k) varying from 1 to 8 with the optimal lag length chosen by the minimum Akaike Information Criterion (AIC).

11. During the Groningen conference in 2003, Thomas Lindblad questioned whether the relationship between Japan and Taiwan was similar to that between the United Kingdom and Singapore or Hong Kong). Offhand, we think that Singapore and Hong Kong are city-states with much smaller economies than Taiwan and probably closer to the United Kingdom/India than to Japan/Taiwan in terms of human capital development. More research is called for in this area.