much “surplus labor” in rural areas, as farmers constantly “opened up” new cultivated land and made labor a scarcity in comparison to land, while the Lewis model predicts the opposite. Second, technology change in agriculture was arguably higher than in other sectors, which is not what is predicted by Lewis model either. Table 1 show that total factor productivity (TFP) in agriculture was higher than those in industry, manufacturing and service sectors, during 1981–1990. Warr (2007, 2008) also came to the same conclusion. Consequently, the dualism in Thai economy before late 1980s did not cause major rural– urban migration as predicted in the Lewis model. Although return in agricultural sector was lower than that in the non-agricultural sector (both in urban and rural areas), the difference was not high enough to overcome the non-monetary factors discouraging migration such as living and housing conditions that were perceived better in rural areas.
Large migration flow was only evident in late 1980s toward early 2000s, when export and investment booms in urban areas caused rapid increase in real wage (see Figure 7), and hence widened the rural–urban earning differential. Note that most migrants during this period were relatively young.
As for employment in urban areas of Thailand, like many other developing countries, it does have a large proportion of its urban labor force working as self-employed rather being employees, as shown in Figure 17. It can then be said that more migrants now become capitalists, rather than workers seeking higher wage as predicted in the early stage of Lewis model. In other words, the “two sectors” seem to collapse into one commercialized sector, which is one characterization of the Lewis’ turning point (Ranis, 2004), without staying too