![Sample our Global Development journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5929/TOC-Subject-Sample-2021-Global-Development.jpg"})
Abstract
This article presents an economic geography model with two differentiated sectors that exhibit weaker inter and stronger intra-industry input-output linkages. Labour is also differentiated according to skills in a hierarchy of tasks they can perform. Globalisation occurs in two distinct phases, leading to the agglomeration of an industry (manufacturing) in the first wave, which is subsequently displaced by the other industry (services) when the second wave of globalisation takes place. Because of agglomeration effects, the increase in relative endowment of a factor may increase its relative wages, leading to more inequality. Within and between nations inequality can result.
Acknowledgements
I am grateful to Stephen Redding, Frederic Robert-Nicoud, Daniel Sturm, Emanuel Ornelas, Rachel Ngai, David Greenaway and other participants at the CEP International Trade Seminar, and also Davin Chor of Singapore Management University for their comments. I acknowledge the scholarships of Ministry of Trade and Industry (Singapore) that made this research possible.
Notes
1. See Bairoch (1982).
2. Source: Confederation of British Industries (CBI).
3. David Kucera and William Milberg (2002) estimate that trade expansion between OCED and non-OCED countries has resulted in the loss of 3.4 million manufacturing jobs in 10 OECD countries between 1978 and 1990.
4. The historical stylised facts of the two waves of globalisation are described in detail in Baldwin and Martin (Citation1999).
5. Bairoch (Citation1989) estimates that the 800-km shipment transport cost for iron goods as a percentage of production cost to be 27% in 1830, 21% in 1850, 10% in 1880 and 6% in 1910. Shipping costs therefore had already fallen significantly prior to the First World War.
6. The presence of Balassa-Samuelson type services in our economy, which are subjected to the ‘tyranny of distance’, cannot explain international patterns of specialisation since they cannot be exported. Hence, services in this article strictly refer to exportable services.
7. Except for the distinction between homogeneous agriculture (or the numeraire sector) and the differentiated sector.
8. Henceforth known as FKV.
9. A fore-runner to this article is by Baldwin, Martin and Ottaviano (Citation2001) – henceforth known as BMO – which shows that intermediate levels of trade costs may cause industries to agglomerate in the North, a result that is consistent with standard NEG predictions. But when the ‘cost of trading ideas’ falls (that is, greater global spillover of knowledge), it can result in the relative ‘deindustrialisation’ of the North, which in that exposition is simply the loss of industries (without being precise on what kind of industries are lost). Unlike Baldwin, Martin and Ottaviano (Citation2001), there is no accumulation of capital, notechnology spillovers, no learning effects and no ‘cost of trading ideas’. Furthermore, this article shows that the loss of manufacturing is not merely relative, but absolute. The second and distinct wave of globalisation can result in a shift in the endogenously determined comparative advantage – away from manufacturing and to services. As the North gains services firms, it can lose manufacturing firms in absolute terms.
10. Neary (Citation2001) provides an interesting exposition on the shortcoming of various NEG models.
11. An alternate way of putting this is that the agricultural sector continues to operate in both regions to equalise wages. The monopolistic but atomistic firms in the industrial sector effectively face an infinite supply of labour at the given wage rate.
12. The labour market need not be quite so dichotomous. For example, one can allow both manufacturing and services to use both skilled and semi-skilled workers, with services having a higher skill intensity compared to manufacturing. It is possible to create the same qualitative result. However, the choice of endowment will become more important in this alternative setup. For example, in the first wave of globalisation when manufacturing agglomerates, the wages for semi-skilled workers might rise above skilled workers since they are used more intensively, thereby presenting interpretational difficulties. With the setup presented in this article, this anomaly will never arise since K-type workers are a perfect substitute for L-types but not the other way around. Less skilled workers will never receive more than their more skilled counterparts in any equilibrium configuration.
13. This will be relaxed later to highlight the effect of endowment on changes in industrial specialisation and wages.
14. The model works with a Cobb–Douglas utility function as well. In equilibrium, different worker types will be paid different wage rates. With Cobb–Douglas preferences, this implies that expenditures on each class of good is different for each type of worker. The Quasi-linear preferences allow us to ignore these income effect considerations. All workers, regardless of how much they are paid in equilibrium, will demand the same amount of XM and XS . Higher income translates to higher demand for the A good. The upshot of this is that there will not be any ‘Home Market Effect’ arising from consumer expenditure for the differentiated industries. All agglomeration is due to forward and backward linkages on the production side.
15. This assumption has no bearing on the qualitative pattern of results except tonote that if the intensity of demand for services is too high, the equilibrium will show dispersion of the services industry even at low levels of services trade costs since skilled workers are assumed to be evenly spread between the two locations.
16. It is not necessary to assume that the linkages are symmetric between the two differentiated industries. The input-output table (see Appendix) suggests they are not. However, in the numerical simulations later, the inter and intra-industry linkages are nevertheless assumed to be symmetric as this has no relevance for the qualitative result.
17. Off equilibrium, if conditional demand is larger than supply of factor , wage wK
has to increase until the equilibrium value [wbar]K
solves for
. If conditional demand is not larger than the supply at that location, [wbar]K
must fall to that of the next skill tier wL
.
18. Again, in an off equilibrium situation where , then wL
must increase until the equilibrium value [wbar]L
solves for
. If
, then wL
will take the value of 1 since some semi-skilled workers will work in the agricultural sector.
19. The wage functions are also non-differentiable.
20. The adjustment parameters ϵ F and ϵ W represent a search increment of a magnitude that is one-hundredth of a per cent of the underlying variables.
21. In the numerical simulations, this paper assumes the intra- and inter-industry linkages of both sectors to be symmetric. Though the IO table suggests that they are slightly different, it does not change the results qualitatively, so long as intra-industry linkages are stronger.
22. In the standard NEG literature, the break point is the value of trade costs at which the symmetric equilibrium with equal shares of industry between two locations becomes unstable. The sustain point is the value of trade costs at which the asymmetric equilibrium (with agglomeration) can be sustained.
23. As standard in NEG models, the sustain point comes before the break point since the value of trade costs at which the symmetric equilibrium becomes unstable is lower (or phi-ness higher) than the value of trade costs at which the asymmetric equilibrium (agglomeration) can be sustained. There is a region of overlap between the two (see and
in ) where both the agglomeration or symmetric equilibrium are stable, depending on the initial conditions. Because of the possibility of multiple equilibria in this region of overlap, expectations become important since some firms may shift location in anticipation that this will change the initial conditions enough to trigger agglomeration. To keep exposition simple, this paper ignores the role of expectation.
24. In equilibrium, there will still be some manufacturing firms in the North. The presence of a large number of services firms means that there will still be demand for manufacturing intermediates.
25. One can think of the increase in K-types workers to arise from a process of skillacquisition, where the L-type workers acquire skills in response to the higher wages (or agglomeration rent) in the services sector. This article does not model the dynamics of skills acquisition. Instead, the initial endowment is changed to reflect the acquisition of skills and a comparative static analysis is carried out.
26. A migration of skilled K workers from South to North will also have to same effect of allowing greater services agglomeration in the North. However, this will result in a larger North market and bring the Home Market Effect into operation.
27. Obviously, the result is dependent on the parameters chosen. If the number of K-type skilled workers is so abundant in the North that the full agglomeration of services there does not exhaust the pool of K workers, then wK = wL . In other words, even after allowing for the increase in K type workers, the employment constraint must bind (all K workers employed in services).
28. Some economists have noted that inequality in developing countries can arise since the production that is offshored from developed countries allows the skilled workers in the developing countries to earn a wage premium above the rest of the population.
29. Some economists have noted that inequality in developing countries can arise since the production that is offshored from developed countries allows the skilled workers in the developing countries to earn a wage premium above the rest of the population.