Economic development according to Friedrich List

Transcription

1 Department of Eonomis Armando J. Garia Pires & José Pedro Pontes Eonomi development aording to Friedrih List WP03/2015/DE WORKING PAPERS ISSN

2 Eonomi development aording to Friedrih List 1 By Armando J. Garia Pires 1 and José Pedro Pontes 2 Date: 20/02/2015 Abstrat: In this paper, we develop a Listian model of eonomi development. The eonomy onsists of a primary setor and a potential industrial setor that an arise via industrialization. Industrialization however depends on if the primary setor speializes on the primary produt, whih an lead to a division of labor between the primary and the industrial setor. In this ase, the industrial setor will use a modern tehnology to produe industrial goods. If suh does not our, the primary setor ontinues to produe all goods with a traditional tehnology of prodution. In addition, the industrial setor has to deide if it onentrates prodution in one loation or if it disperses prodution in two loations. We show that the level of transport osts matters for division of labor and for the degree of manufaturing agglomeration if and only if the refinement of the primary input is strong, i.e., if the raw material loses a lot of weight through industrial transformation. Otherwise, if the industrial proess is not so muh weight-losing, industrialization an begin with a deentralized symmetri spatial pattern independently of the transport osts level. Keywords: Friedrih List; Eonomi Development; Industrial Agglomeration; Division of Labor. JEL Classifiation: O14, R11, R30 1 Centre for Applied Researh at NHH (SNF), Norwegian Shool of Eonomis (NHH), Helleveien 30, 5045 Bergen, Norway. Tel. +(47) ; armando.pires@snf.no 2 Shool of Eonomis and Management, Universidade de Lisboa, ISEG, Rua Miguel Lupi 20, Lisboa, Portugal. Tel.: +(351) ; ppontes@iseg.ulisboa.pt

3 2 Introdution Friedrih List is mostly known for the infant industry argument of protetion 3. However, everyone that reads his main opus The National System of Politial Eonomy (List, 1841), will easily realize that the main fous in List s philosophy is how to develop eonomially a nation. In partiular, List main onern is how a ountry an transform its eonomy from an agriultural one to an industrial one. In this paper, we develop a Listian model of eonomi development. The eonomy onsists of a primary setor and an industrial setor that an potentially arise via industrialization. Industrialization however depends on if the primary setor speializes on the primary produt, whih an lead to a division of labor between the primary and the industrial setor. In this ase, the industrial setor will use a modern tehnology to produe industrial goods. If suh does not our, the primary setor ontinues to produe all goods with a traditional tehnology of prodution. In addition, the industrial setor has to deide if it onentrates prodution in one loation or if it disperses prodution in two loations. In this sense, this paper ombines the ideas of List on eonomi development with those of Adam Smith (1786) on the division of labor, and of Krugman (1991) on eonomi agglomeration 4. Although List reognizes the importane of Adam Smith s onept of division of labor, he defends that Adam Smith ignores the idea of produtive power, i.e.: the potential that a ountry has to develop and industrialize. In other words, a ountry ould for instane abdiate of some gains in the present aruing from trade in order to reap the future gains brought by eonomi development and industrialization. In the same, way List seems to antiipate the ideas latter developed in spatial eonomis and eonomi geography. He for example states that: Compare, on the one hand, the value of landed property and rent in a distrit where a mill is not within reah of the agriulturist, with their value in those distrits where this industry is arried on in their very midst, and we shall find that already this single industry has a onsiderable effet on the value of land and on rent; that there, under similar onditions of 3 See for instane Harlen (1999); Levi-Faur (1997a,b); and Sai-wing Ho (2005). Modern views on protetion an be found in Krugman (1984, 1993), and Franois and van Ypersele (2002). 4 On the onept of division of labor see Pontes (1992), Stigler (1951), Williamson (1981), Young, (1928), and Beker and Murphy (1992). On the eonomis of agglomeration see Pais and Pontes (2008), Sott (1983), Sott (1986), Glaeser (1989), and Glaeser and Gottlieb (2009).

4 3 natural fertility, the total value of the land has not merely inreased to double, but to ten or twenty times more than the ost of ereting the mill amounted to; and that the landed proprietors would have obtained onsiderable advantage by the eretion of the mill, even if they had built it at their ommon expense and presented it to the miller. [ ] As it is in the ase of the orn mill, so is it in those of saw, oil, and plaster mills, so is it in that of iron works; everywhere it an be proved that the rent and the value of landed property rise in proportion as the property lies nearer to these industries, and espeially aording as they are in loser or less lose ommerial relations with agriulture. And why should this not be the ase with woolen, flax, hemp, paper, and otton mills? Why not with all manufaturing industries? We see, at least, everywhere that rent and value of landed property rise in exatly the same proportion with the proximity of that property to the town, and with the degree in whih the town is populous and industrious. Furthermore, List defends that the gains from o-loation are speifi to the industrial setor. The same does not our in the agriultural setor: Agriulture an only progress, the rent and value of land an only inrease, in the ratio in whih manufatures and ommere flourish; and manufatures annot flourish if the importation of raw materials and provisions is restrited. In this sense, while List learly argues against protetion in agriulture, he seems to defend that tariffs on the industrial goods an either be dispensed in the starting stages of industrialization or not, depending on if the region has already a suffiient mass of industry to promote further industrial development. Hene, List takes as a departure point the Adam Smith s (1776) notion of oupational division of labor. Let us assume that a produtive proess, whih was formerly ahieved by undifferentiated workers, is split in heterogeneous tasks, eah task being assigned to a different worker. Then, it an be shown that on aount of several fators, a strong inrease of labor produtivity follows. This oupational speialization is, in Smith s own words, limited by the extent of the market, by the number of workers that an be engaged in the speialization proess. This set of potential partiipants is bounded from above by the population density and by the state of transport and ommuniation tehnology. By omparison with Adam Smith (1776), Friedrih List fouses the geographial dimension of the division of labor. Not only omplementary tasks are performed by different workers, but

5 4 also they an be done in different regions/ountries, whih then speialize in heterogeneous sets of produtive tasks. List regards the geographial division of labor as a soially negative phenomenon for two kinds of reasons. First, it makes the onstraint that market size exerts upon industrialization more binding, sine upstream and downstream tasks are now performed over long distanes in different regions or ountries. Seond, he onsiders that geographial speialization leads inevitably to umulative development asymmetries aross regions and ountries. List s argument is the following. The separate tasks that result from the division of labor have different degrees of apitalization, i.e., intensity in the use of apital, both physial and human. 5 Hene, national or regional speialization in subsets of tasks leads to different degrees of apital intensity aross spae. As market fores tend to reward basially physial and human apital rather than raw labor fore, spatial inequalities in apitalization will reinfore themselves in a umulative way. List then proeeds to enumerate the poliy tools that an prevent the geographial division of labor to arise, the most obvious one being high tariffs. However, List s disussion goes far beyond an argument in favor of trade protetionism as we shall see ahead. The rest of the paper is organized as follows. In the next setion, we desribe the basi assumptions of the model. Then we introdue the rules of the game of the spatial eonomy. We then derive the equilibrium of the game. We onlude by disussing the main findings. 5 Friedrih List seems to be the first eonomist to use the onept of human apital, whih he labels as mental apital.

6 5 Assumptions of the model We assume a spatial eonomy that is desribed by the following assumptions: There are two ountries, Home (H) and Foreign (F), that are ompletely symmetri in number of onsumers/workers, land endowment and available tehnologies of prodution. The eonomy entails the prodution of an agriultural (or primary ) good, whih is transformed to give a manufatured omposite onsumer produt. The two goods are stritly omplementary in prodution, in the sense that there is a fixed proportion between the rate of output of agriulture and manufature. W. l. g., we assume that this proportion is one to one. The tehnology of agriultural prodution is onstant returns to sale, and we an model it through a Cobb-Douglas prodution funtion: Q 1 AL = S (1.1) where Q Agriultural output A Produtivity parameter L Labor input S Land input is a parameter suh that 0 < < 1 We assume that eah ountry has the same fixed amount of agriultural land that is used in prodution. For instane, we may assume that eah ountry ontains nfarms, eah having one unit of extent. Hene, eah ountry uses an amount of land S = n. By hoosing adequate unit measures for the agriultural output, Aredues to unity. Furthermore, if the prodution funtion (1.1) is divided by S, it beomes q = l (1.2)

7 6 where Q q is the produtivity per unit of land S L l is the intensity of ultivation, i.e. the amount of labor used per unit of land S It is then lear that the marginal returns of applying labor to land are dereasing. We also assume that the number of farmers, n, is high enough so that they work under onditions of perfet ompetition. The ompetitive prie of the omposite onsumer good is labeled as p and is taken by all farmers. The tehnology of transforming the agriultural good in order to produe a onsumer good depends on the existene of vertial integration between the suessive prodution stages. If the agriultural good is proessed in-house by the farmers themselves, the prodution entails only variable osts, the average labor produtivity in refining being labeled θ. Given the omplementarity between the two tasks, agriulture and manufature, the intensive (per unit of land) prodution funtion is given by: l {, θ} q = Min l l = l (1.3) By solving the inequality θ l > It is easy to realize that prodution funtion (1.3) holds if l ( 1) l θ > (1.4) ( ) 1 Otherwise, if θ l < holds, the labor produtivity in agriulture exeeds the produtivity in the task of refining the input. Hene, less than the whole agriultural output an be transformed, thus preventing the fixed proportion between the output rates of the two tasks to hold. Condition (1.4) is more binding if the produtivity of the manufaturing ativity θ is low. By ontrast, if input transformation is vertially disintegrated, it will be performed by a monopolist firm with a fixed equipment (a mahine) embodying F units of labor in its

8 7 onstrution. It is assumed for simpliity that the working of the mahine requires no variable osts. Hene, the intensive prodution funtion that onerns the ollaboration of independent firms (farmers and a manufaturing fatory) is, if we assume that the fixed ost is already sunk: l {, } q = Min l + = l (1.5) Transport osts of goods within eah ountry are zero. Both the primary input and the onsumer goods have an ieberg transport ost τ > 1. It is neessary to dispath an amount τ > 1 from the origin ountry in order that one unit amount arrives in the host ountry. We also assume that the transport ost is paid by the purhaser, so that a fob mill priing regime holds in this spatial eonomy. The rules of the game In Figure 1, we plot the extensive form of the game that models our spatial eonomy. Figure 1: Game in extensive form

9 8 In words, eah ompetitive farmer deides first whether to transform the agriultural produt in-house or sell it in the market thus letting another firm to perform the transformation into a finished onsumer good. All farmers take the same deision onerning vertial integration, sine they are symmetri. If the farmers deide to speialize in agriulture, a manufaturing firm makes the proessing. This firm makes first a loation deision: either to onentrate geographially prodution by setting up only one plant in a ountry (Country F, w.l.g.) and serving the other market through exports; or setting up two plants, one in eah ountry in order to serve the loal demand thus ahieving proximity between prodution and ustomers. The former strategy entails positive transport osts of the onsumer good, while saving on fixed osts if it is ompared with the latter strategy. If ondition (1.4) holds, the profit funtion of the vertially integrated farmer is: ( θ ) π = p l w + l R (2.1) Profit funtion (2.1) impliitly assumes that labor is not divided. The same workers perform both produtive tasks: land ultivation, with wage rate w; and agriultural good proessing, with wage rate equal to θ, the labor produtivity in this task. Ris the land rent whih we assume to be earned by the farmers themselves, so that they are also land owners. Sine farmers work under ompetitive onditions, the profit is kept at a zero level and land rent an be written from (2.1) as ( θ ) R = p l w + l (2.2) Heneforth, we will assume that the land rent Rin (2.2) is the farmer s payoff in the game depited in Figure 1. Let us onsider now the outome of vertial integration by the farmers. By maximizing the rent funtion (2.2) in relation to l, we obtain the optimal intensity of ultivation * l : l p = w + θ * 1 1 (2.3)

10 9 The optimal ultivation intensity inreases with the ompetitive prie of the onsumer good and dereases with labor rewards both in ultivation and in proessing. Combining (1.3) and (2.3), we obtain the produtivity per unit of area (or the output of eah farm): q p = ( l ) = w + θ * * 1 (2.4) The output of eah farm varies similarly to the optimal intensity of ultivation. It is also lear that the transformation of the agriultural input is a refinement, in the sense that it brings about a loose of volume/weight by the raw material. Hene, if the farmer speializes in primary prodution and is paid a prie k for the raw produt, the optimal intensity of ultivation is l 0 k 1 ( 1 ) = w (2.5) In turn, the produtivity per unit of area beomes q 0 k ( 1 ) = w (2.6) From (2.4) and (2.6), we simplify inequality q < q * 0 To get the ondition p θ > 1 w k (2.7) Usually, we have p > k beause the former prie overs both raw food prodution and transformation, whereas the latter only pays for agriultural osts. Hene, transformation will be weight-losing if the labor ost of proessing is a signifiant share of total wage osts. We an assume this ondition to hold.

11 10 A monopolist firm now makes the transformation of the input. This hange goes together with a tehnologial shift in transformation. The unit variable ost of manufaturing, whih was 1 > 0 in the former prodution regime swithes to zero, while a fixed ost 0 θ F > emerges. This latter ost was absent before and its arise means that manufaturing eases to done in the ontext of a ottage and it beomes performed in a large fatory. As it is depited in Figure 1, the monopolist manufaturing firm has two available loational strategies. Consider first the strategy that onsists in setting up a single plant in the Foreign ountry in order to serve the loal market and export bak to the Home ountry. This geographi onentration strategy allows the firm to save on fixed osts but inurs on higher transport (trade) osts related with moving the onsumer goods aross the ountries. The profit funtion of this onentration strategy is: ( τ ) ( ) π = p k Q + p k Q F (2.8) 1 m m where p m is the fob mill prie of the onsumer good k is the fob mill prie of the agriultural input Q is the total output of the onsumer good in eah ountry Sine we assume that the two goods (agriultural and manufatured) are omplementary, the proportion between their rates of output is fixed. Hene, Q stands for the output of both goods. Differene in weight between input and output is refleted in the respetive pries, k and p m. In profit funtion (2.8), τ stands for the ieberg transport (trade) ost between the two ountries. It is neessary to dispath exatly τ > 1units of a good from the home ountry so τ 1 that exatly one unit arrives to destination. A share of the sent good is lost in transit. τ We also define the following pries, p f d fob mill prie p delivered (full) prie

12 11 The relation between p f and p is given by d p d pd amount dispathed from origin = τ p f = = τ > 1 (2.9) p amount arriving to destination f In addition, it is assumed that eah seller uses a spatial prie poliy onsisting in harging the same fob mill prie to all its ustomers. If the industrial firm deentralizes prodution by setting up two plants, one in eah ountry, his profit funtion is ( ) π 2 = 2 Q pm k F (2.10) Where k, pm and Q have the same meaning as in (2.8). The following stage in the game whose extensive form is plotted in Figure 1 entails the seletion of pries by the manufaturer both for the intermediate good, k, and for the onsumer good p m. We suppose here that the proessing firm has two available ations, namely, either to behave as a monopolist or exiting the market. If the manufaturer intends to be a monopolist, it is aware that it faes elasti demands, in the form of a lower bound for the agriultural input and a higher bound for the prie of the final onsumer good. Were these bounds not respeted, the ompetitive farmers would have an inentive to swith to vertial integration of primary prodution and manufaturing, refining the input in-house and selling the onsumer good diretly to the onsumers. Nevertheless, if the manufaturer sets these limiting pries, it may happen that its profit is negative. Then, the refiner has an inentive to exit the market. A non-oordination situation (a sort of poverty trap ) arises, the output of the onsumer good being redued to zero. Let us onsider first that the manufaturer has previously deided to set up a unique plant in the foreign ountry, thus onentrating geographially the prodution. Its profit funtion is expressed in (2.8). We have to alulate the pries of the input and the final good, k and p m respetively that maximize its profit, while deterring farmers to return to vertial integration. Clearly, eah onsumer should be able to purhase the manufatured good at a prie p m not higher than the ompetitive prie p. Hene, p m fulfills the ondition

13 12 p p τ m (2.11) It appears as self-evident that the monopolist maximizes profits in (2.8) by quoting output prie as p pm = ε, where ε > 0, arbitrarily small (2.12) τ Derivation of the limit prie for the agriultural produt is more involved. Let R (.) be the land rent as a funtion of the prie that eah farmer reeives for his output. Then, intermediate good prie k should lead eah farmer to have a rent level at least as high as the level that he would get by integrating vertially in the ontext of a ompetitive market, i.e. ( ) R( k ) R p (2.13) ( ) ( θ ) R p p q w l R( p ) is the land rent under vertial integration. * * = + (2.14) q and l are the optimal output and input * * under vertial integration, being given by (2.4) and (2.3), respetively. It is not diffiult to onlude that the land rent is given by: ( ) R p 1 ( 1 ) ( 1 ) ( ) = p 1 > 0 w + θ (2.15) Moreover, the land rent of a farmer that only ultivates the land and reeives a prie k for the raw agriultural produt an be shown to be: ( ) R k 1 ( 1 ) ( 1 ) ( ) = k 1 > 0 (2.16) Equating (2.15) and (2.16), we have: ( ) R( k ) R p = (2.17)

14 13 We solve this equality for prie of the input k aording to the following steps: 1. Substitute (2.15) and (2.16) into (2.17), and ut the ommon term ( 1 ), yielding: p 1 1 ( 1 ) ( 1 ) ( 1 ) ( 1 ) = k w + θ 2. Take logs and simplifying gives: θ ln k = ln p ln We assume that the produtivity of the transformation task is not muh higher than θ the reward of the land ultivation task, so that, > 1 but it is small. Then, we an use w the MaLaurin approximation, i.e., ln ( 1 ) + x x, for a small x, and obtain: θ ln k ln p w 4. Taking exponentials, simplifying and solving for k, we obtain the input prie k that solves equation (2.17), p k (2.18) e θ θ w Sine e > 1, then k < p. The manufaturing firm quotes a prie k + ε, ε > 0, arbitrarily small, for the intermediate good. This prie is lower than the ompetitive prie for vertially integrated farms. The output of a farm that only performs agriulture and reeives a prie slightly higher than (2.18) for its output is, by analogy with (2.6), lose to 1 k q = w (2.19) Substituting (2.18) in (2.19), the farm output beomes

15 14 p q = θ e w ( 1 ) (2.20) Sine eah ountry has n farms, the output of the onsumer good in both the Home and Foreign ountries will be p Q = nq = n θ e w ( 1 ) (2.21) Then, we have seen before in (2.8) that the profit funtion of a monopolist manufaturer that sets a single plant in the Foreign ountry is: ( τ ) ( ) π = p k Q + p k Q F 1 m m Where the first term stands for operating profits in the Home ountry, the seond term represents profits in the Foreign ountry (where the single plant is loated) and the third term is minus the fixed ost. This profit funtion may also be written as: ( τ ) π1 = Q 2 pm k 1+ F (2.22) Plugging expressions (2.12), (2.18) and (2.21) into the profit funtion (2.22), the latter beomes ( 1 ) p 2 p p π1 = n ( 1+ τ ) F θ θ τ e w e Dividing by 2n, we obtain the profit per inhabitant/farmer: ( 1 ) π1 1 p 2 p p F π 1 = ( 1+ τ θ θ ) 2n 2 τ 2n e w e (2.23) Heneforth, we adopt the notation, π i π i, i = 1, 2 (2.24) 2n

16 15 Hene, π i stands for the per apita profit generated by a monopolist that sets up a number i (either one or two) of plants. Consider now the subgame where the manufaturer has previously deide to establish two plants, so that its profit funtion is given by (2.10), In this ase, Q and ( ) π 2 = 2 Q pm k F k are still given by (2.21) and (2.18), while the fob mill prie of the final good is now equal to p sine ustomers live in the neighborhood of a plant. Thus, the aggregate profit of a manufaturer that ahieves proximity to onsumers is ( 1 ) p p π 2 = 2n p 2F θ θ e e Dividing π 2 by 2n, we obtain the profit per inhabitant as: ( 1 ) π 2 p p F π 2 = p θ θ 2n n e e (2.25) Finding the equilibrium of the game The lass of games depited in Figure 1 ontains purely sequential, perfet information games, whose subgame perfet equilibrium (SGPE heneforth) an be found very simply by means of bakward indution. We will redue this lass of games to one defined over three parameters: θ labor produtivity in the refinement of the agriultural input when the proessing is made by the farmers themselves, under vertial integration. F F 2n per apita ( per farmer ) fixed ost whenever transformation of the input ours under vertial disintegration. τ > 1, ieberg transport (trade) osts aross the two ountries.

17 16 The following speifiations of other parameters are made: We also give another form to parameter θ, through defining 1 = (3.1) 2 p = 1 (3.2) w = 1 (3.3) θ θ 2 δ e = e (3.4) It an be seen very easily that δ has the following properties: it is greater than 1, it is stritly inreasing and onvex in θ. δ an be diretly interpreted as the refining rate, thus given the number of physial units (e.g., weight units) of the input should be onsumed in proessing on order to yield one unit of the final good. This an be understood easily if we bear in mind the definition of the prie of the intermediate good, k, in (2.18), together with definition (3.2), as we have than p k θ = e = δ = amount of input onsumed amount of onsumer good manufatured With the numerial speifiations (3.1), (3.2), (3.3) and the definition made in (3.4), it is lear that the profit funtion of a manufaturer that sets only one plant (written in (2.23) simplifies to τ π 1 = F 4δ τ δ (3.5) If the manufaturer opts by settling two plants, then its profit funtion (written in (2.25)) simplifies to π = 1 2F 2δ δ (3.6)

18 17 Sine we have a lass of games defined in three parameters and two-dimensional plots allow only to examine game situations in two parameters, we have hosen to assign two different values for the ieberg transport ost τ. In empirial studies, it is usual to estimate trade osts as a 20% share of total exports or imports value. Hene, we will assume first that τ = τ = (3.7) τ τ = will be taken as the normal or low level for transport (trade) osts. By ontrast, if 4 trade relations between the ountries deteriorate, the trade ost will be raised to the double of the level shown in (3.7), i.e. The ase with low transport (trade) osts. τ = τ = (3.8) τ 5 3 Substituting 5 τ = in profit funtion (3.5), ondition π 1 > 0 is regarded to mean the same as 4 32δ 45 F < (3.9) 2 80δ Furthermore, onsidering profit funtion (3.6), ondition π 2 > 0 is equivalent to 1 F δ < (3.10) 2 4δ Let us assume now that π 1 and π 2 are both positive. Bearing in mind (3.5) and (3.6), with 5 τ =, ondition 4 π 2 > π 1 an be seen to mean that 8δ + 5 F < (3.11) 2 80δ

19 18 Figure 2 depits the regions where inequalities (3.9), (3.10) and (3.11) are satisfied in parameter spae 5 ( δ, F ) for low transport osts τ = 4.

20 19 The ase with high transport (trade) osts We now deal with the ase with high transport (trade) osts, where 5 τ =. Substituting this 3 value of τ in profit funtion (3.5), the per apita profit of a firm setting up a unique plant is 3 2 π 1 = F 2 10δ 3δ (3.12) The profit of a manufaturer whih establishes two plants is still given by (3.6), sine no transport osts are inurred here. The ondition π 1 > 0 is seen to be equivalent to 9δ 20 F < (3.13) 2 30δ Condition π 2 > 0 is still equivalent to (3.10). Moreover, ondition meaning as π 1 < π 2 has the same 6δ + 5 F < (3.14) 2 30δ

21 20 Figure 3 shows the regions where inequalities (3.10), (3.13) and (3.14) are satisfied in parameter spae 5 ( δ, F ) for high transport osts τ = 3. Transport osts and the degree of industrial transformation as auses of industrial agglomeration In this subsetion, we assess the impat of transport osts and the degree of input refinement on the spatial onentration of manufaturing. Taking into aount the profit funtion in (3.5) onerning the ase where a single plant is set up by the industrial firm, we have τ π 1 = F 4δ τ δ (3.15) The partial first derivative of π 1 in relation to the transport ost ieberg rate,τ, is negative τ 4δ τ δ π = + < 0 (3.16) This result is natural sine dereasing transport osts raises the profit of onentrating prodution in a single point in spae. If we ompute the ross partial derivative in relation to δ, we obtain π τ δ δ τ δ = + > 0 (3.17) Consequently, inreasing the refining rate δ dampens the negative effet that inreasing transport osts has upon the profit of a firm, whih transats its input and output aross ountries. We an rationalize (3.17) by realling that, under the ieberg tehnology, transport osts are amounts of goods (both manufatured goods and raw materials) that are lost in transit between prodution and onsumption sites. Consequently, inreasing the refining rate, δ, dereases the amount of input that must be arried between the farm and the fatory in

22 21 order to produe one unit of manufatured good. For a given ieberg rate, τ, it also dereases the amount that is lost in transit Another way of expressing this is to say that the transport ost of the input outweighs the freight ost of the manufatured produt, sine the industrial proess onsists in refining a raw material, onverting it in a lighter produt. As we have seen before, the transport ost of the input is k ( τ 1). Sine k, the fob mill prie of the input, dereases with δ (see (2.18) and (3.4)), the total transport ost dereases with the refining rate, δ. Hene, rising the degree of industrial transformation also favors the agglomeration of manufaturing in a single plant, in a single ountry, thus making a geographial division of labor to emerge.

23 22 Main onlusions Four main onlusions arise: a) Division of labor, both oupational and geographial, is aused by low transport osts (low τ ) and high population density (high n). Together these two fators bound the extent of the market from above, i.e., they limit the number of workers that are potentially available to engage in a speialization proess 6. b) Geographial division of labor, i. e., onentration of manufaturing in a single plant in a single ountry, implies both low transport osts (low τ ) and intermediate population density (intermediate n). If density is too small, there will be no vertial disintegration. By ontrast, if the number of inhabitants in eah ountry is too high, the industrial output will be produed in both ountries. ) The size of ieberg transport osts depends on two fators: the rate of the intermediate good lost in transit, τ 1, and its prie k. This prie is inversely related with the refining rate δ (see (2.18) and (3.4)), beause this is the number of input units that must be used to produe one unit of onsumer good with a fixed parametri prie. If δ is low, then k will be high, so that transport osts will be high independently of the transport rate τ. Consequently, if the refinement rate is low, manufaturing an start in eah ountry even if transport osts are low. d) A normative onlusion, stressed by Friedrih List himself, is that geographial division of labor is soially negative. Indeed Proximity yields a higher agriultural output for eah farm and ountry than Conentration. The reason is that Conentration is eonomial if and only if transport osts are low, as it implies moving inputs and outputs aross ountries. Conentration is more profitable than Proximity as a loation strategy for the monopolist provided that transport osts and the prie k of the raw material are low. The latter ondition implies a low intensity of land ultivation and a small agriultural output in eah farm and ountry. Furthermore, the land rent expressed in (2.16) dereases with k and beomes lower if the manufaturer installs only one plant. 6 Amador and Caldeira Cabral (2014) observe that after 2003 Portugal had a sustained surge in exports, whih annot be easily explained, sine the Euro was strong and the oil pries were on the rise. Our result that a redution in trade osts ondues to a division of labor between the primary setor and the industrial setor, an help to explain this ourrene. In fat, in 2003, the EU s Common Agriultural Poliy (CAP) shifted from produt support to produer support, whih in pratie ondue to a redution in the level of protetion in the agriultural setor (European Commission, 2012).

24 23 Disussion In this paper, we have developed a Listian model of eonomi development, where the eonomy is made up of an primary setor that produes good using a traditional tehnology and an industrial setor that an arise if the primary setor speializes in the primary good, i.e.: if division of labor arises in equilibrium. If division of labor does not our, the eonomy is stuk in an underdevelopment trap. If division of labor does our, whih is limited as in Adam Smith by the extent of the market, industrialization an arise with either industrial dispersion or agglomeration, i.e., either with or without geographial division of labor. Geographial labor division is seen to arise if transport osts are low and the refining rate in manufaturing is high. The latter ondition implies that the prie of the agriultural input is low, thus diminishing the intensity of its ultivation. In this ase, both setors in the eonomy agriulture and manufaturing will produe smaller outputs than those that would prevail if industrial prodution were deentralized between the two ountries. Otherwise, if the industrial proess is not so muh weight-losing, industrialization an begin with a deentralized symmetri spatial pattern independently of the transport osts level. In future work, we wish to look at one of the most innovative ideas by List: eonomi agents have inomplete information about trade osts, espeially in what onerns trade osts in international trade. The randomness of trade osts an arise beause of disruptions in international trade. These disruptions an be aused for example in extreme ases by wars or environmental disasters (suh as the one in the nulear plant in Fukushima that putted at risk the just in time system of prodution of Japanese manufaturers). But the disruptions in international trade an also be due to more normal auses suh as different legal systems, different institutions, or even different ulture and language that makes international trade rules more unertain. This line of researh is in our opinion very promising, in partiular for two reasons. First, to our knowledge there is very little researh on the effets of unertain trade osts on eonomi ativity, trade, and loation of prodution. Seond, the randomness on trade osts an in our

25 24 view help to partially explain one of the most important puzzles in international and maroeonomis: the border puzzle 7. The border puzzle refers to the empirial evidene that equally distant regions, trade muh more with eah other, even after orreting for trade barriers, if they are loated in the same ountry than if they are loated in different ountries. For instane, MCallum (1995) finds that in 1988, trade between Canadian provines was 2,200 perent larger than the trade between U.S. states and Canadian provines. In this way, a border between two ountries might just introdue an unertainty that redues international trade aross borders. In future work, we will try to verify this assertion. 7 Obstfeld and Rogoff (2001) defend that the border puzzle is one of the six puzzles of open eonomy maroeonomis.

26 25 Referenes Amador, J. and Caldeira Cabral, M. (2014), A Eonomia Portuguesa no Contexto Global, in Alexandre, F. (ed.), A Eonomia Portuguesa na União Europeia , Coimbra, Conjuntura Atual Editora. Beker, G. and Murphy, K. (1992), The Division of Labor, Coordination Costs, and Knowledge, Quarterly Journal of Eonomis, 107, European Commission (2012), The Common Agriultural Poliy: A Story to be Continued, Luxembourg: Publiations Offie of the European Union. Franois, P. and van Ypersele, T. (2002), On the Protetion of Cultural Goods, Journal of International Eonomis, 56, Glaeser, E. and Gottlieb, J. (2009), The Wealth of Cities: Agglomeration Eonomies and Spatial Equilibrium in the United States, Journal of Eonomi Literature, 47, Glaeser; E. (1989), Are Cities Dying?, Journal of Eonomi Perspetives, 12, pp Harlen, C. (1999), A Reappraisal of Classial Eonomi Nationalism and Eonomi Liberalism, International Studies Quarterly, 43, Krugman, P. (1984), Import Protetion as Export Promotion: International ompetition in the Presene of Oligopoly and Eonomies of Sale, in Kierzkowski, H. (ed.), Monopolisti Competition in International Trade, Oxford, Clarendon Press. Krugman, P. (1991), Inreasing Returns and Eonomi Geography, Journal of Politial Eonomy, 99, Krugman, P. (1993), The Hub Effet: or, Threeness in Interregional Trade, in Ethier, W.; Helpman, E. and Neary, P. (eds.), Theory, Poliy and Dynamis in International Trade: Essays in honor of Ronald W. Jones, Cambridge: Cambridge University Press. Levi-Faur, D. (1997a), Friedrih List and the Politial Eonomy of the Nation-State, Review of International Politial Eonomy, 4, Levi-Faur, D. (1997b), Eonomi Nationalism: From Friedrih List to Robert Reih, Review of International Studies, 23,

27 26 List, F. (1841), The National System of Politial Eonomy, London, Frank Cass. MCallum, J. (1995), National Borders Matter: Canada-U.S. Regional Trade Patterns, Amerian Eonomi Review, 85, pp Obstfeld; M. and Rogoff, K. (2001), The Six Major Puzzles in International Maroeonomis: Is There a Common Cause? In Bernanke, B. and Rogoff, K. (eds.), NBER Maroeonomis Annual 2000, Volume 15. Pais, J. and Pontes, J. (2008), Fragmentation and Clustering in Vertially-Linked Industries, Journal of Regional Siene, 48, Pontes, J. (1992), Division of Labor and Agglomeration Eonomies, Estudos de Eonomia, 12, Sai-wing Ho, P. (2005), Distortions in the Trade Poliy for Development Debate: A Re- Examination of Friedrih List, Cambridge Journal of Eonomis, 29, Sott, A. (1983), Industrial Organization and the Logi of Intra-Metropolitan Loation: I. Theoretial Considerations, Eonomi Geography, 59, Sott, A. (1986), Industrial Organization and Loation: Division of Labor, the Firm, and Spatial Proess, Eonomi Geography, 62, Smith, A. (1776), An Inquiry into the Nature and Causes of the Wealth of Nations, Oxford, OUP. Stigler, G. (1951), The Division of Labor is Limited by the Extent of the Market, Journal of Politial Eonomy, 59, Williamson, O. (1981), The Modern Corporation: Origins, Evolution, Attributes, Journal of Eonomi Literature, 19, Young, A. (1928) Inreasing Returns and Eonomi Progress, Eonomi Journal, 38,

28 27