Thailand’s Overlooked Growth Story: Labour Reallocation and Domestic Investment

In a famous 1994 article in Foreign Affairs called “The Myth of Asia’s Miracle”, Paul Krugman applied the ideas of growth accounting to Asia. He argued that for Asian countries growth in inputs — the labour force, and especially the capital stock such as infrastructure, equipment and factories — accounted for all of Asia’s rapid growth being experienced at that time.

“Technological change”, a term denoting productivity increases, explained almost none of the growth. Krugman also said that Asia’s growth was unsustainable because the growth was due to rapid expansion of the capital stock which would eventually face diminishing returns. As a country builds factories, the new facilities make a big difference to output but over time they will make less and less of a difference. Hence, growth will slow down.

Krugman’s argument was provocative and seemingly confirmed by the Asian Financial Crisis that halted the preceding decade of rapid growth in Southeast Asia. But his analysis was problematic at two levels. He drew upon growth accounting studies conducted by other scholars for his empirical evidence, as his article made clear. But these studies were mainly based on data from Singapore and Hong Kong. These two city-states differ from nearly all of the rest of Asia. They lack low-productivity traditional agriculture.

Krugman also overlooked the effect of people’s movement out of low-productivity agriculture to higher-productivity industry and services. This economic shift is a major source of Asian growth. Krugman’s methodology, which only considered overall workforce growth and did not account for workforce reallocation among sectors, missed the latter effect. He did not realise the omission possibly because he was focused on economies that are not typical of Asia.

The 1997-99 Asian Financial Crisis was a turning point for Thailand, but there is more to the story than a slowdown in the effect of physical capital accumulation, as Krugman’s article would have it.

Extending Krugman’s framework, productivity growth per worker is better understood by distinguishing three sources, not just two. First, growth of factor inputs (physical capital and human capital) relative to raw labour; second, growth of the productivity of these factors, through technical change; and third, growth of aggregate output per worker due to the reallocation of labour from lower-productivity sectors (mainly agriculture) to higher-productivity sectors (mainly industry and some services). Krugman focused on the first two. These are within-sector sources of productivity growth. He missed the third, which is a between-sector source of growth in productivity per worker.

Take the case of Thailand, a country that is far more typical of Asia than Singapore or Hong Kong. Like most of Asia, it has a large agricultural sector. But average productivity (output per worker) is low in that sector. Figure 1 summarises the changing sectoral composition of employment in Thailand from 1960 to 2023. As a share of the total workforce, employment in agriculture has declined throughout this long period. The absolute number of workers employed in agriculture has declined since about 1989. About two-thirds of these workers leaving agriculture have gone into the services sector and about one-third into industry.

Figure 2 now combines these data on employment with data on the composition of GDP in Thailand over the same period. The data show that value-added per worker in agriculture is much lower than in services or industry. The movement of labour out of low-productivity agriculture into higher-productivity services and industry is a source of output growth that Krugman overlooked.

Figure 1. Employment in Thailand, 1960 to 2023

Figure 2. Productivity per worker in Thailand, 1960 to 2023

The crucial question is how much of Thailand’s productivity growth derives from input growth, technological improvement, and reallocation across sectors. Table 1 reports the results of a three-component decomposition exercise.

Table 1. Thailand: Sources of aggregate productivity growth per worker per year, 1981–2021

These estimates confirm that the 1997-99 Asian Financial Crisis was a turning point for Thailand, but there is more to the story than a slowdown in the effect of physical capital accumulation, as Krugman’s article would have it. Since the crisis, annual productivity growth has been less than half of that before the crisis. The decomposition helps in understanding the reason for that. The largest change is the decline in the within-sector input effect. This consists mainly of a decline in the rate of growth of the physical capital stock per worker. Investment in physical capital has three components: domestic private business investment, public investment and foreign direct investment. The first is by far the largest and the decline in this component is the main reason for the large drop in the within-sector input effect.

The results also show that the between-sector reallocation effect has been important in both periods, explaining almost half of the productivity growth per worker. This contributor to productivity growth also declined, but the input effect declined even more. To some extent, the decline in the between-sector effect must be considered a consequence of the decline in investment. As agriculture continues to decline as a share of total employment, this inter-sectoral source of aggregate productivity growth will presumably become progressively less important. Intra-sectoral specialisation, reflecting Thailand’s changing comparative advantage, will become more important.

To raise long-term productivity growth significantly, an increase in the level of domestic private sector investment will be required. It must be supported by well-designed public infrastructure investment. The form this private investment will take is, of course, best determined by the firms themselves, not the government. But the policy measures that might achieve an increased level of private investment deserve close attention.

2024/288