Trade, development, and poverty-induced comparative advantage

Abstract

This paper deals with the relation between trade and development when poverty affects individual decision making. We develop a two-sector model that links production and schooling decisions under poverty with standard neo-classical trade analyses. The decision to either work or acquire skills depends on households having reached subsistence levels of income, implying that the income level of a country becomes important in establishing comparative advantages and trade patterns. Trade liberalisation is always allocative efficient, but its timing is important for the speed by which countries industrialise. Our analysis supports the idea that there are instances that stalling trade liberalisation may serve industrial development.

Keywords:

• comparative advantage
• poverty
• economic development

JEL Classifications:

• F00
• O11

Acknowledgement

The authors would like to thank participants of the DEGIT X conference in Mexico City and two anonymous referees for their helpful comments and suggestions.

Notes

 1. A different way to put this is to say that work is a Giffen good. Giffen behaviour can be a consequence of minimum consumption requirements (Marrewijk and van Bergeijk 1990). Indeed, the work/training decision meets the three conditions under which Giffen behaviour can be observed (Jensen and Miller 2008): households face subsistence concerns, they consume a basic (work) and a fancy good (training), and the basic good has no ready subsitute.

 2. That is, we implicitly assume that the wages just above subsistence are not high enough to lead to a situation in which higher wages lead to a lowering of the number of hours worked. The latter phenomenon is a well-known possibility in the literature on labour markets, where it leads to the backward bending part of the individual labour supply curve. Our assumption seems reasonable in light of the fact that backward bending supply curves are usually considered to occur in situations of relative affluence.

 3. This pattern has clear historical antecedents. Industrial and commercial centres typically emerged either at locations where the soil was fertile enough to boost large populations not directly engaged in food production, or where particularly favorable water routes made it possible to import food from other regions, such as the cases of Venice, the Black Sea and Holland (Cipolla 1980, 75–6).

 4. Non-restrictive, since it also leaves room for other interpretations than random choices. For instance, untrained individuals may see the importance of people becoming trained and are willing to sacrifice income in order for others to become trained. The amount of time devoted to training is thus actually what each individual decides to ‘chip in’ for the common good of getting a trained labour force.

 5. The expression features nominal wage rates, where in equation (1) real wages would be due. However, the price index drops out.

 6. The training decision above subsistence implies that individuals are myopic, perceiving current wage differentials to persist forever. It can be shown, however, that equation (1) is also consistent with static forward-looking expectations (see Baldwin et al. 2003, Section 2.B.4). Moreover, it is intuitively plausible: untrained labour bases its decision to become trained or not on the profitability of doing so.

 7. We chose a general formulation because training-while-working may affect labour input disproportionally. Training could affect the efficacy of labour input positively, also when it does not directly apply to food production, but it may also require more effort and energy than the hours devoted to it. In any case, we assume that the individual takes the value of ζ as given.

 8. In modelling the manufacturing sector and the demand side of the economy, we follow standard practice in international trade modelling and the new economic geography literature. See, for example, Brakman et al. (2001) and Baldwin et al. (2003).

 9. We ignore subscripts to distinguish between varieties as each variety enters consumer demand symmetrically. Hence, equilibrium output, price and labour requirements will be the same across varieties.

10. Note therefore that the effect of subsistence levels of income on consumption choices works through the budget constraint. In our framework, utility maximisation can be seen as a three-tier process, where consumers first decide on their current and future income levels, and then divide the remaining income optimally over consumption possibilities.

11. That is, we assume identical and homothetic preferences. We apply it to maintain focus on the relation between poverty, training decisions and comparative advantage. Moreover, it keeps results tractable, for which reason homothetic preferences are also the standard assumption in trade theory. See, however, Matsuyama (2000) and Stibora and de Vaal (2007) for a treatment of non-homothetic preferences in a trade theoretic framework.

12. Land rents are equal to food production minus what is paid to untrained labour, hence (1−β)w£/β. Our assumption that land rents are redistributed to the whole population implies that initially, when all individuals are untrained, each untrained labourer is also owner of land and that this does not change when individuals become trained.

13. By equation (3), H is also the share of trained labour in society.

14. Though technically feasible, it makes no sense to allow for w = E_min at positive levels of H as then positive H could never have been reached.

15. We note that a second steady state equilibrium exists, which occurs when even at H = 0 wages are at subsistence. Then w−E_min  = 0 also implies a positive steady state level of H. It is, however, immediately clear that this is a theoretical possibility only, as supposedly any economy has started at some point in time without any amount of trained labour (hence positive H could never have been reached). The limiting case when w−E_min  = 0 at H = 0 is possible though, but highly unstable. Any rise in H leads to an upward spiral until income is above subsistence level and, subsequently, the stable steady-state equilibrium is reached.

16. Extensive sensitivity analysis shows that the basic form of the figure is invariant to alterations in any of these parameters, provided of course that training levels are positive at H = 0.

17. In contrast to Figure 3, the horizontal axis exhibits constant increments in time and not in H. All values of [Hdot] were calculated by using the actual amount of H in the previous period.

18. The smoothness of [Hdot] also disappears if the effectivity C by which individual training transforms into trained labour approaches one. The reduction in untrained labour as well as the reduction in desirability of getting trained during the first period is so large that the economy immediately swithches to the above subsistence state and remains there forever.

19. This does not imply that our definition of steady state should be that training levels are zero for two consecutive periods. In our numerical calculations, we made training levels depend on past period's wages. Such lagged operationalisation of the simulateity of decision making of individual agents is not warranted when calculating the steady state mathematically.

20. During period 1 training levels are [1−E_min /βA] for North and [1−E_min /βA*] for South, amounting to H = C[1−E_min /βA] and H* = C[1−E_min /βA*] at the beginning of period 2. Using this in equation (21) and assuming ζ(T) = 1−T and ζ(T*) = 1−T*, shows that the relative price of manufactures in period 2 is smaller than A/A* (which was the initial relative price). Taking the derivative of the relative price with respect to A and evaluating it for A = A*, yields d(p/p*)/dA < 0. Hence, during period 1, p/p* falls and becomes lower than one.

21. Strictly speaking, we cannot speak of initial comparative advantages, since there is no manufacturing sector initially. Were we to assume starting points with epsilon small, equal shares of trained labour, South's initial comparative advantage would be in manufacturing however.

22. This clearly contrasts to standard treatments of comparative advantage, where comparative advantage is typically taken as given. In comparison to the dynamic comparative advantage literature, the mechanism we offer is completely different. The literature on endogenous comparative advantage explains how comparative advantages evolve when there are no inherent differences between agents, see Yang and Ng (1998) for an overview. In addition, the relation between exogenous comparative advantage and endogenous comparative advantage has received attention, for instance by establishing conditions under which initial exogenous comparative advantages may change. An early and seminal contribution is Yang (1994). As of yet, comparative advantage has not been linked to decision making under poverty at all.

23. In the end, this must imply that the room for training becomes negative, so that untrained labour gets under subsistence. This opens the door to endogenous population growth – some people will starve to death – but we will not consider this option.

24. This holds when the population growth rates in both countries are not too far apart. Only when the more fertile North also reaches steady state at subsistence levels of income, while having a population growth rate exceeding that of South considerably, will the upward sloping part reappear.

25. Specifically, the relative consumption shares of agricultural and manufactures products and the degree of economies of scale in the agricultural sector.