Trade Liberalization, Market Integration and Industrial Concentration: The Spanish Economy During the 19th Century

(1)

EUI WORKING PAPERS

EUROPEAN UNIVERSITY INSTITUTE

(2)

European University Institute 3 0001 0034 1923 3 © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(3)

EU RO PEAN U NIVERSITY IN ST IT U T E , FLO R EN C E

DEPARTMENT OF HISTORY AND CIVILIZATION

EUI Working Paper HEC No. 2000/3

Trade Liberalization, Market Integration

and Industrial Concentration:

The Spanish Economy During the 19th Century

El is e n d ap a l u z i e, j o r d ip o n s and

Da n ie l A. Tir a d o

BADIA FIESO LA N A , SAN D O M EN IC O (FI)

(4)

N< 5V V T U 7 o

WP 9 4 0

EUR

'O "Si

$

No part o f this paper may be reproduced in any form without permission of the authors.

European University Institute Badia Fiesolana I - 50016 San Domenico (FI)

Italy © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(5)

Trade liberalization, market integration and industrial

concentration: the Spanish economy during the 19th century

Elisenda Paluzie1, Jordi Pons2 and Daniel A. Tirado3

1 Departament de Teoria Econòmica, Universitat de Barcelona

2 Departament d’Econometria, Estadistica i Economia Espanyola, Universitat de Barcelona

3 Departament d'Història i Institucions Economiques, Universitat de Barcelona and European University Institute, Florence

JEL classification: F12, F15, R12, N73

Key words: trade liberalization, market integration, industrial concentration, economic geography, economic history of Spain

Corresponding author:

Elisenda Paluzie

Departament de Teoria Econòmica Universitat de Barcelona Av. Diagonal 690 08034 Barcelona Catalonia (Spain) (email:paluzie@eco.ub.es) © The Author(s). European University Institute. Digitised version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(6)

1. Introduction

Current trade liberalization in Europe has stimulated the theoretical and empirical analysis of the effects of economic integration on industrial location. From a theoretical point of view, a new field, the so-called “new economic geography”, has made an important contribution to deal with this issue.1 New economic geography models show that there might exist a strong relationship between economic integration and geographical concentration of industries. These models also identify the mechanisms through which such a relationship may arise. In this regard, increasing returns in presence of trade costs and labor migration between regions (Krugman 1991), vertical linkages between upstream and downstream industries when both are imperfectly competitive (Venables 1996), or simply factor accumulation in presence of external economies (Baldwin 1997), can sustain the formation of huge productive differences across regions when a process of economic integration starts.

In contrast, empirical analysis on this subject is still patchy and inconclusive. Scarce empirical work has followed two main lines of work. On the one hand, Hanson (1997) has tried to estimate a structural equation derived from this kind of models, that is the market potential function. On the other, papers as those by BrUlhart (1996) or Briilhart and Torstensson (1996) have tried to test for the empirical validity of these models by analyzing the accomplishment of some of their theoretical predictions.2

Within this framework, the aim of this paper is to analyze, both theoretically and empirically, the effects of trade liberalization and market integration in 19th century Spain on the geographical concentration of industrial activity. The study of historical cases of market integration brings us the opportunity of implementing theoretical models matching well-known economic experiences. Besides, it gives us the chance of carrying out empirical exercises. And last but not least, these kind of studies may suppose a step forward in our understanding of some episodes of economic history.

The paper is organised as follows. In section 2, we present the main facts characterizing the Spanish process of regional and international integration. In section 3 we develop an economic geography model that tries to fit 19th century Spanish experience. In section 4, we test if Spanish empirical evidence on industrial

1 A good survey on this new theoretical approach in Ottaviano and Puga (1998). It is also worth mentioning the handbook on this topic by Fujita, Krugman and Venables (1999).

2 In this paper we will follow the line o f work opened by BrUlhart and Torstensson (1996).

(7)

concentration and its determinants lends support to the predictions of our model. Finally, section 5 summarizes the main findings and offers some concluding remarks.

2. Internal and external market integration in Spain during the 19lh century: the facts

During the second half of the 19th century, the Spanish economy experienced a twofold process of economic integration. On the one hand, the construction of the railway network started in 1848. From then on, a huge amount of financial resources were invested in this business and, as a consequence, in a few years Spain completed its basic railway network. Table 1 shows the evolution of the railway lines operating during this period. It is easy to verify that the basic network was constructed between the 50s and the 90s. Afterwards, amounts invested in the railway network became scarce.

Table 1

Railway network operating in Spain (in km.)

1850 28 1855 440 1860 1880 1865 4756 1870 5316 1875 5840 1880 7086 1885 8300 1890 9083 1895 10526 1900 11040 1905 11309 1910 11362 1915 11424 1920 11445 1925 11543 1930 12030 1935 12254

Source.- Cordero and Menéndez (1978).

3 © The Author(s). European University Institute. Digitised version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(8)

The efficiency of this railway network has been a workhorse among Spanish economic historians for years.3 Nevertheless, it seems rather clear that the construction favored the reduction in transport costs between Spanish regions and it stimulated

I

economic integration all over the country. It has been argued that the substitution of old means of transportation for railways generated a social saving between 7 and 12% of Spanish GDP.4 In fact, as we can see in Table 2, the amount of merchandise transported by rail increased enormously during these years and Spanish regions received a strong stimulus to increase their productive specialization.

Technical improvements in coastal transportation came later than in other countries and their impact was less marked; nonetheless, if we look at the data reported in Table 2, we can affirm that they also contributed to the reduction of the economic distance between regions. Furthermore, institutional changes in the money and banking sectors may also have contributed to the reduction in the transaction costs even though historiography has not emphasized the effects of these changes in market integration. In this regard, the unification of the monetary system around the peseta in 1869 and the establishment of the Bank of Spain in all provincial capitals, together with the transfer system established by this central bank after 1885, should also be considered in the analysis of Spanish market integration.5

Table 2 Merchandise traded (1000s Tm) Railway Coastal Shipping 1870 3.2 0.79 1875 4.01 1.01 1880 6.08 1.04 1885 8 4 4 1.15 1890 9.28 1.18 1895 12.22 1.56 1900 15.34 2.04 1905 17.96 2.35 1910 20.98 2.96 Sources - Gômez Mendoza ( 1982) and Frax ( 1981 ).

3 As an example see Nadal (1975) or Tortella (1981). 4 G6mez Mendoza (1982).

5 See Tortella (1970) and Castafteda and Tafunell (1993).

(9)

This process of market integration also had its international scope. On the one hand, from the 70s on, the worldwide reduction in freight rates generated a new upsurge of international trade. On the other, liberal trade reforms in the European continent collaborated in stimulating this process as well. Trade liberalization in Spain started with the 1848 tariff act. In this year, Spain gave up its last mercantilist prohibitions. However, Spanish tariffs reached their minimum values in the context of the 1868 Tariff act and with the development of some commercial agreements during the 80s. This process finished abruptly in the first 90s, when the main commercial treaties were refused and the Spanish government passed a new and highly protectionist tariff act.

Figure 1.

Nominal Average Tariff Rates

Source: Tirado (1996) and Tena (1999)

(10)

Figure 2.

Tariff rates by sectors

I

Ions iiy

Source: Tirado (1996)

Looking at the data plotted in figures 1 and 2 it is easy to conclude the existence of two main periods in Spanish commercial policy between 1870 and 1913. From 1868 to 1890, liberalization was the common fact in Spanish trade policy. From then to WW1, Spain returned to the practice of protectionist policies that hampered its integration in the international markets. Nevertheless, as we show in figure 3, during the period 1870- 1890, the Spanish economy benefited from these years of liberalization. Both the volume of merchandise traded and the openness rates increased radically until the new upsurge of protectionist practices at the turn of the 90s.

(11)

Figure 3.

Imports and exports Volume Indices (1913=100)

Our objective in the remaining of this paper is to analyze the effects of this twofold integration process on Spanish industrial concentration. To this aim, we will first develop a theoretical framework that allows us to offer some theoretical predictions.

3. T he model

We use an economic geography model with three regions to illustrate the effects of economic integration on industrial concentration. We follow closely Fujita, Krugman and Venables (1999, chapter 4) general specification of economic geography models.

(12)

We consider a world economy consisting of three regions: 1, 2 and 0, the latter representing the external economy.

There are two sectors, agriculture and manufacturing.

Agriculture is perfectly competitive and produces a homogeneous good using a specific factor called “peasants”. It is a constant-retum sector tied to the land. The peasant population is assumed as being completely immobile between regions.

Manufacturing is a monopolistically competitive sector and produces a variety of differentiated goods, using a specific factor called “workers”.

All three regions can trade with each other. Workers are mobile only between the “domestic” regions 1 and 2 but not with the external region 0.

We shall choose units such that the world manufacturing labor force is LM = p and the world agricultural labor force is LA = 1 - p

The share of labor force devoted to manufacturing in region 0 is Xo, X| in region 1 and A.2 in region 2. The size of the labor force in each region is therefore LSM = p ,XS with s= 0, 1 or 2. Workers are not allowed to move to nor from the external region 0 so we assume Xo is constant at any point in time.

Agriculture is evenly divided between the three regions so that:

L0a=L,a = Lt* = (1 -fi)/3

Agricultural goods can be freely transported and are produced under constant returns so agricultural workers will have the same wage rate in all regions.6 We use this wage rate as the numeraire and set it equal to one.

In sections 3.1, 3.2 and 3.3 we describe the consumer and producer behavior and state the principal results of the consumer and producer optimization problems introducing transportation costs between the three regions. In section 3.4 some normalizations are performed and finally, section 3.5 describes the dynamics of the model.7

3.1 Consumer behavior

All individuals in this economy share a utility function of the form:

6 This assumption is highly unrealistic. Pujita, Krugman and Venables, chapter 7 (1999), analyze the implications o f breaking it up. However, the main conclusions o f new economic geography are not refuted.

7 A complete presentation o f the model, including detailed derivations, can be found in Paluzie (1999).

(13)

U = M ’, A'->‘ (1) where M represents the quantity index of the consumption of manufactured goods and

A is the consumption of the agricultural good. // is the expenditure share of

manufactured goods.

The quantity index M, is a sub-utility function that depends on the consumption of all varieties of manufactured goods. M is defined by a constant elasticity of substitution function:

M =

_{[ r - H *}

(2)

where m(i) denotes the consumption of each available variety, n is the number of varieties produced and the parameter p represents the intensity of the preference for the variety in manufactured goods. The number of varieties produced, n, is assumed to be large, although smaller than the potential range of products.

a s —!— represents the elasticity of substitution between any two varieties. 1 ~ P

Utility optimization yields the following compensated demand function for the y'th variety of the manufacturing product:

m (j) = P ù V k -o

£ p { i ) ^ 'd i

■M (3)

The expression for the minimum cost of attaining M is then,

p/ £ P (j)m(j)dj = M £ p(i) r 'di p- V where -yPl\ [ p a y - 'd i = c (4) (5)

G is the price index for manufactured goods which measures the minimum cost of

purchasing a unit of the composite index M of manufacturing goods. Demand for j can be written as

m ( j) = ■M _{(6 )} 9 © The Author(s). European University Institute. Digitised version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(14)

g j ^ R / O £ ,

A - Q- M) Y

A ~ _pD* (7)

for j e [0,n] (8)

Holding G constant, the price elasticity of demand for every available variety is constant and equal to 0.

3.2 Introducing transportation costs between the three regions

In our model, we have three different locations for the production and consumption of goods. For the moment, let’s assume that each variety is produced in only one location and that all varieties produced in a particular location are symmetric, having the same technology and price. We denote the number of varieties produced in location r by n„ and the f.o.b price of one of these varieties by p rM.

It is costly to ship goods in all directions. We assume Iceberg transport costs. If a good is shipped between either of the two domestic locations only a fraction I/T arrives. If a good is shipped between either domestic location and the outside world, only a fraction I/T a arrives.

If a manufacturing variety produced at location r is sold at price p,M then the delivered price p rsM of that variety at consumption location s is given by,

The manufacturing price index may take a different value at each location. Assuming that price is constant across varieties in each location ( p(i) = p, ) and Iceberg transport costs, the price index in location 5 is given by,

for r = 0,1,2 and s = 0,1,2 when r = 1 and s = 2 or r =2 and s = 1 when s =0 and r = 1,2 or r = 0 and s - 1,2 © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(15)

(9)

And now consumer demand in location s for a good produced in r is given by,

( 10)

To supply this level of consumption Trfx m ,( j) units have to be shipped. Summing across locations in which the product is sold, total sales of a single location r variety are given by,

3.3 Producer behavior

We can now turn to the producer behavior.

The production of any variety of manufactured good involves economies of scale. Technology is the same for all varieties and in all locations and involves a Fixed

Because of increasing returns to scale, the preference for variety by consumers and the unlimited number of potential varieties of manufactured goods; each variety will be produced by a single, specialized firm, in only one location. This means that the number of manufacturing firms is the same as the number of available varieties.

3.3.1 Profit maximization

The profit of a particular firm producing a specific variety at location r and facing a given wage rate for manufacturing workers, wrM, is given by,

(ID

input F and a marginal input requirement cM. The only input is labor; thus, the production of a quantity qM of any variety in any given location requires labor input /M, given by

*—« M M M / p A/ W v

n , = p r qr - w , (F + c qr ) (13) Profit maximization implies that

(14) l l © The Author(s). European University Institute. Digitised version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(16)

Therefore, firms mark up price over marginal cost by a factor 1/p.

If there is free entry of firms into manufacturing, profits must be driven to zero. Given the pricing rule, the profits of a firm at location r are:

F ( a - \ Ÿ

n. =

a -1 Hr (15)

The zero-profit condition implies that the equilibrium output of any firm is: < ,« • = 4 ( ^ - 0 (16)

c

Output per firm is the same in each region.

The associated equilibrium labor input is also constant and given by,

/• = F + c u q ' = Fo (17)

(18) Then the number of manufacturing firms at location r is:

^ W j M ~ ~ T ~~ Fa

We observe that market size affects neither the mark-up of price over marginal cost (14) nor the scale at which individual goods are produced (16) but it does affect the number of manufacturing firms (or number of varieties produced) (18). Therefore, in our model all scale effects work through changes in the variety of goods available.

3.3.2 The manufacturing wage equation

Mr

H r

The equilibrium output of any firm has to be equal to the demand for it, so:

= M Y J Y A P , U r a ( T r, M ) ' a G 1° - ' (19)

sex

By applying some rules of algebra, we can obtain the manufacturing wage equation:

M

Y . Y^ Tr,U G, (20)

This equation gives the manufacturing wage at which firms in each location break even. To obtain the real wage equation of location r manufacturing workers, wrM, we have to deflate the nominal wage by the consumer price level G,M(p,A) l M. We obtain:

co M = w rMG f , ( p rAy i'-‘‘> (21) 3.4 Normalizations © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(17)

We can simplify the manufacturing price index and the wage equation by choosing units of measurement.

First, we are free to choose units of measurement for output (be it units, tens of units, kilos...). We choose units such that the marginal labor requirement satisfies the following equation:

a - 1

Then the pricing equation, (14), becomes M M

P r = W r

(22)

(23) Second, as the number of firms is simply an interval of the real line, (0, n], without loss of generality we can choose units of measurement for this range. We choose convenient units by setting the fixed input requirement F to satisfy the following equation:

(24) The equilibrium labor input, (17), now becomes:

/ ' = F a = fi (25)

And therefore the number of manufacturing firms in each location given by equation (18) is now:

L,u

We can then simplify the price index and wage equations. The price index equation, (9), becomes:

G. =

And the wage equation, (20), can now be written as: i

(26)

(27)

(28)

3.5 Instantaneous equilibrium

Equilibrium in this model is given by the simultaneous solution of 12 equations (4 for each region), which determine the income of each region, the price index of

(18)

manufactures consumed in that region, the wage rate of workers in that region, and the real wage rate in that region.

The incomes of the three regions can be written:

(29)

(30)

(31)

And the price index equations,

= l V ( w . -T.)' ° + X , - ( w , ) " + X, ( « , T)' - (33) G, = [x, (w, T,)1 ° + X,(w, T)' ° + X, •(wI ),'*}-« (34)

These price index equations have a crucial property. The price index in either of the domestic regions, 1 or 2, will tend to be lower, the higher the share of manufacturing that is concentrated in this region, the lower the share of manufacturing in the other domestic region (associated derivatives are presented in Section A. 1 of the Appendix). So a shift of manufacturing into one of the regions will tend, other things being equal, to lower the price index in that region and thus make the region a more attractive place for manufacturing workers to locate. This is a form of “forward linkage” or cost effect that tends to reinforce an unequal geography. This effect is entirely due to internal transport costs. As we are assuming 2o constant (non-labor mobility with the external region), it is the existence of transport costs between the two domestic regions that produce this forward linkage.

The wage equations for the three regions are:

(35) (36) T (3 7) © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(19)

I.ike the price index equations, the wage equations also exhibit an important property. Assuming price indexes in all regions remain constant, then nominal wage rate in a region will tend to be higher if income in this region is high (associated derivatives are presented in Section A.l of the appendix). Nominal wage rate in a domestic region would be higher, the higher income is in the world economy as a whole, but especially the higher income is in this particular region. The reason is that firms can afford to pay higher wages if they have good access to a larger market. This is thus a form of “backward linkage” or “home market” effect that reinforces the forward linkage analyzed before.

These two properties allow us to formulate a first hypothesis. Industry will tend to concentrate in regions closer to larger markets.

And finally, we obtain the real wage equations by deflating the nominal wage by the cost of living index, as in (21), but with the price of agriculture equal to one everywhere:

a>0 = wt G ,- - >—- OC

(U, = vt^G, 11 (3 9 )

a>2 - w 2G 2“ (4 0 )

In this model the distribution of manufacturing across regions is given at any point in time by the simultaneous solution of these 12 equations ((29) to (40)).

Over time, the distribution of manufacturing across regions evolves because workers can move between the two domestic regions. We assume that Xi, the regional allocation of manufacturing labor, adjusts according to the real wage ratio <B| / to2 in the following fashion:

A,

= y(ru, la>2) (41)

We will now look at some numerical examples to see the possible results of the dynamics over time.

We plot a)| /e>2, the ratio between the two domestic regions’ real wage rates in manufacturing, against region 1 ’s share of manufacturing. Any point where the wage differential is 1 is an equilibrium. Such an equilibrium is stable if the schedule is downward sloping, because where that is the case, whenever one region has more

(20)

workers than the other, workers will tend to migrate out of that region. In contrast, if the schedule is upward sloping, that is if coi /ti>2 increases with Xi, then the equilibrium is unstable because workers will tend to migrate into the region that already has more workers. There may also be corner equilibria: if all labor is concentrated in location 1, it will stay there if coi > a>2 and conversely.

In the numerical example we assume o = 6, p = 0.4, T = 1.75 and Xo=l/3. But we let the external transport cost To take three different values corresponding to three different cases, the high-transport case, the intermediate-transport case and the low- transport case. This will allow us to analyze how the integration of the domestic economy with the outside world, as measured by the cost T„, will affect the equilibrium allocation of labor between the two domestic regions. The external trade cost is a measure of the barriers to free trade, both natural, due to the costs of distance, and artificial, due to tariffs.

These three different cases are shown in figures 4 to 6.

In figure 4, we have To=1.9. If a unit of the good is shipped from the external region to one of the domestic regions, only 0.52 units eventually reach their destination. Roughly, this transport cost is equivalent to a loss of half of the goods transported en route. We consider this value to correspond to a high transport cost. In this case the wage ratio is greater than one if X| is less than 1/3, less than one if X| is greater than 1/3. If a region has more than a third of the world manufacturing labor force it is less attractive to workers than the other region. In this case the economy converges to a long run symmetric equilibrium in which manufacturing is equally divided between the two domestic regions. © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(21)

Figure 4

Real wage ratio t»i /o) 2 against labour force in Region 1, To=1.9

h i g h e x t e r n a l t r a n s p o r t c o s t c a s e

^

00 0 1 0 . 2 0 . 3 0 . 4 0 . 5 0 6 0 7

r e g i o n 1 l a b o r f o r c e

In figure 5, we show what happens when the economy is opened slightly, To=1.4. (In this case, this transport cost corresponds to an arrival fraction of 71%). The equilibrium at which manufacturing population is evenly distributed between the two domestic regions, each of them having one third of world manufacturing population, is still stable. Concentration of population in either region is, however, stable, as well. There are two other unstable equilibria that lie between the stable equilibria. If Xi starts from either a sufficiently high or a sufficiently low initial value, the economy will converge not to the symmetric equilibrium, but to a core-periphery pattern with all manufacturing in only one region. There are five equilibria, three stable (the symmetric equilibrium and manufacturing concentration in either region) and two unstable.

(22)

r e a l wcg e re tie 0 .9 9 4 C .9 9 6 0 9 9 5 1 .0 0 0 ' 0 0 2 '. 0 0 4 ^ .0 0 5 1 0 0 5 Figure 5

Real wage ralio <Oi /to2 against labor force in Region 1, To=1.4

i n t e r m e d i a t e e x t e r n a l t r a n s p o r t c o s t c a s e

r e g i o n t l a b o r f o r c e

Finally, when the economy is opened further, To=1.3 (arrival fraction of 76%), the symmetric equilibrium becomes unstable and the only stable allocations are concentration in one region or the other. ©

The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(23)

Figure 6.Real wage ratio toi /u>2 against labor force in Region 1, T0 = 1.3

l ow e x t e r n a l t r a n s p o r t c o s t c o s e

r e g i o n 1 l a b o r f o r c e

Figure 7 shows how the types of equilibria vary with external transport costs. Solid lines indicate stable equilibria, broken lines unstable.

Figure 7. Equilibria and external transport costs

region 1 _©The Author(s). European University Institute. Digitised version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(24)

At high external transport costs, there is a unique stable equilibrium in which manufacturing is evenly divided between the two domestic regions. When external transport costs fall below some critical level (T(S), the sustain point), a core-periphery equilibrium in which all manufacturing is concentrated in one of the two domestic regions becomes possible although the symmetric equilibrium is still stable. When the economy is opened further,.and external transport costs fall below a second critical level (T(B), the break point), the symmetric equilibrium becomes unstable and so the domestic economy must necessarily show a core-periphery pattern with all manufacturing concentrated in one region.

A second hypothesis that can he formulated is the existence o f a monotonic relationship between external trade costs and internal industrial concentration Trade liberalization stimulates the settlement o f core-periphery patterns. The reduction in external trade costs amplifies concentration outcomes o f internal market integration

Break and sustain points depend on some of the parameters in the model. This dependence is summarized in table 3 which reports the break point and the sustain point at different values of p and a.

Table 3. Break and sustain points at different values of p and o

p=0.38 p=0.40 p=0.42

cx=5 T(B)=1.7 T(B)=1.9 T(B)=2.2

No sustain point No sustain point No sustain point <7=6 T(B)=1.28 T(B)=1.35 T(B)=1.4

T(S)=1.35 T(S)=1.45 T(S)=1.6

<7=7 I(B)=1.162 T(B)=1.20 T(B)=1.24

T(S)=1.165 T(S)=1.22 T(S)=1.28

The sustain point always occurs at a higher value of To than does the break point. Both critical values are increasing in p, so the range of transport costs in which the core-periphery pattern occurs is greater, the larger the share of manufactures in the

(25)

economy. The manufacturing sector can then generate forward linkages via supply and backward linkages via demand which constitute centripetal forces that allow a core periphery equilibrium to be sustained over a wide range of transport costs. Both critical values are decreasing in a, so the range of transport costs in which the core-periphery pattern occurs is greater, the smaller the elasticity of substitution among products (the more differentiated are the products). If a decreases, then the number of varieties produced at a location increases and the firm’s price cost mark-ups decrease. Scale effects are thus stronger, the smaller the elasticity of substitution between varieties and, therefore, the range of transport costs at which the core-periphery equilibrium occurs is larger. Centripetal forces in the model are a combination of economies of scale, market size and transport costs. By decreasing c, we are increasing the magnitude of scale economies.

A third hypothesis that can be formulated is the existence o f a positive relationship between the degree o f scale economies and industrial concentration Industries with a high degree o f scale economies will be the most localized.

4. Empirical analysis: scale economies, centrality and industrial concentration

The first step in our empirical analysis is to establish the evolution of geographical concentration of industry during the period analyzed. In order to capture the degree of concentration of Spanish industrial sectors, we have calculated locational Gini indices that measure the locational structure of industrial production in three points in time.8 1856, previous to the integration process; 1893, when external integration is at its maximum and internal integration seems to have concluded its first wave; and 1907, when Spanish economy reduces its international integration and internal integration does not undergo significant changes with respect to the situation achieved in 1893.

1 These indices have been constructed proxying the contribution o f each sector and province to the industrial product by the data on the tax payments o f the different sectors and provinces to the

Contribución Industrial y de Comercio (Industrial and Commercial Contribution) in 1856, 1893 and

1907. For more details about sector aggregation and indices construction see Tirado, Paluzie and Pons

(2000) 21 © The Author(s). European University Institute. Digitised version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(26)

Table 4

Gini indices of geographical concentration of industry

1856 1893 1907 Food 0.34 0.43 0.44 Textiles 0.80 0.91 0.92 Metallurgy 0.71 0.79 0.78 Chemistry 0.63 0.66 0.69 Paper 0.76 0.71 0.69

China, glass and ceramics 0.48 0.54 0.58

Wood and cork 0.86 0.72 0.67

Tanning and leather 0.61 0.70 0.70

Others 0.71 0.79 0.78

GLOBAL 0.44 0.61 0.61

Source- See Tirado, Paluzie and Pons (2000).

Table 4 reports locational Gini indices of the distribution of nine industries across 45 Spanish provinces in these three dates. The main conclusions are straightforward. Spanish industry increased its concentration between 1856 and 1893. That is to say, meanwhile the first steps in internal market integration were given and when Spanish economy was suffering an accelerated process of integration into the international market. In contrast, this process of geographical concentration arrived to a halt when Spain put a brake on its integration into the international economy and the internal market integration process decelerated.

In our model, we show how international market integration might amplify concentration outcomes of internal market integration. The evidence seems to lend support to this hypothesis. As we can see in Table 5, agglomeration and trade barriers follow an inverse relationship in 6 out of 8 sectors analyzed.

(27)

Table S

Sign of Changes in Protection Levels and in Gini indices between 1865 and 1893

Protection Concentration Food - + Textiles - + Metallurgy + + Chemistry - + Paper +

-China, glass and ceramics - +

Tanning and leather + +

GLOBAL - +

Source.- See text.

Nevertheless, this kind of evidence does not secure that centripetal forces generated by increasing returns or forward and backward linkages were in the base on industrial agglomeration. To prove that, we are going to empirically analyse the other predictions derived from our model.

If increasing returns were the mechanism favoring cluster formation when regions integrate, we should verify the existence of a relationship between each sector’s scale of production and its geographical concentration. We have carried out this kind of analysis for the higher industrial concentration year in our sample, 1893.We constructed scale economies indicators for each of the nine industries studied.9 Table 6 shows the calculated values of production scale per plant and the rank occupied by each industrial sector according to this indicator.

If it was the case that forward and backward linkages were the mechanism causing industrial agglomeration we should expect that industrial location should be biased to the economic center of the country. The higher the centrality location bias of an industry, the higher should be the geographical concentration of this industry. To test this prediction of the model we have elaborated a centrality index for each one of the Spanish provinces. From this index, we have constructed a centrality location index for

’ Details about construction o f this index in Tirado. Paluzie and Pons (2000).

(28)

each of the nine industries (see Table 6). This index enables us to analyze the centrality location bias of each industry.

Table 6

Gini, scale Economies and location bias towards center in 1893

Gini Rank Scale

Economies Rank Centrality bias Rank Food 0.43 9 78 14 8 1.95 9 Textiles 0.91 1 182.89 4 10.83 1 Metallurgy 0.79 2 433.21 1 5.94 2 Chemistry 0.66 7 226.13 3 4.26 5 Paper 0.71 5 231.91 2 4.05 6 China, glass and ceramics 0.54 8 53.48 9 2.79 8 Wood and cork 0.72 4 87.36 6 3.84 7 Tanning and leather 0.7 6 82.68 7 4.71 4 Others 0.79 3 176.03 5 5.02 3 Spearman p Kendall x 0.62 0.33 Spearman p Kendall x 0.89 0.72

Source.- See text.

In figures 8 and 9 we have plotted ranks in sectorial scale economies and in centrality bias on Gini's coefficients for each of our nine industries. In order to test the existence of a relationship between the Gini index for the different sectors and the scale economies and the centrality bias variables, we used two coefficients of rank correlation: the non parametric tests Kendall x and Spearman p. These rank correlation coefficients can be obtained from the following expressions (see Lehmann (1975) and Hettmansperger (1984)): x_{= £ £} 1-1 j=i U .i-Vu n ( n - l ) © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(29)

where Uy = sgn(Xj-Xj), V,j =sgn(Yj-Y|), n is the number of observations and finally, X and Y are the variables for which we want to test the existence of association; and:

p = l -6 E d ?

where d, = rank (X,)- rank (Yj).

As our sample has less than 10 observations, we have not been able to use the normal approximation to test the existence of association between the variables using these two statistics. So, we have been forced to use Kendall (1970) tabulated values for the x statistic.

The empirical evidence on the relationship between Gini indices and scale economies indicates the existence of a positive and statistically significant lelationship between these two variables (we can reject with a significance level a = 0.13, the null hypothesis of absence of relation between them). That is to say, sectors with high scale of production tended to be the most geographically concentrated.

(30)

Scale economies ranks vs Gini ranks

Figure 8

SCALE

In analyzing the relationship between the centrality index and the Gini coefficients, we can reject with a significance level a =0.01 the null hypothesis of absence of relationship between these two variables. The value calculated for the Spearman rank correlation coefficient also points to the existence of a strong and positive relation

between these two variables. ©

The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(31)

Centrality ranks vs Gini ranks

Figure 9

0 2 4 6 8 10

CENTRALITY

Hence, empirical analysis lends support to the theoretical predictions derived from the model. Spanish market integration was in the origin of the increase of industrial geographical concentration during the second half of the 19th century. Contrary to the opinion of some Spanish economic historians, we have shown that industrial concentration came along with international trade liberalization and did not increase when protectionist practices came back after the 90s.10 Besides, this locational concentration tended to be higher in those industries with a higher degree of scale economies and also in those with a higher locational bias to the economic center. So, increasing returns and forward and backward linkages would have acted as the centripetal forces favoring industrial concentration during this episode of external and internal market integration in Spain.

l0This view appears in some classical works in Spanish Economic History such as Sanchez-Albomoz (1987). 27 © The Author(s). European University Institute. Digitised version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(32)

5. Concluding remarks

During the second half of the 19lh century, the Spanish economy underwent a major process of internal market integration while the commercial policy of the period had two different phases, a first phase of trade liberalization and a second characterized by the return to protectionism. This period corresponds also to Spanish initial industrialization, a process that was characterized by a striking geographical concentration of industrial activity. In order to match this historical experience and derive a relationship between trade policy, market integration and regional inequalities, we set an economic geography model with three regions, two domestic and one external. The results of our theoretical exercise show that internal market integration combined with the opening up of a closed economy brings regional polarization when scale economies and forward and backward linkages are present.

Our empirical analysis, built on a historical case study, has verified the existence of a positive relationship between the degree of scale economies and industrial concentration and also between the degree of closeness to the economic center and industrial concentration. Though less conclusive, our empirical data on trade policy also seems to point in the direction of our theoretical predictions.

This study allows us to derive some conclusions that could be of interest both in a controversial debate in the economic history of Spain and in the current debate on the effects of the European integration process. First, contrary to the view of some classical works on Spanish economic history, our analysis shows that industrial concentration occurred in parallel to trade liberalization and did not increase when protectionist practices returned at the end of the 19lh century. That is to say, protectionist policies did not foster geographical industrial concentration in 19th century Spanish economy.

Second, dealing with the effects of the current European integration process, it is worth noting that if barriers to labor mobility are maintained between member states while absent regionally, then one of the first effects of European integration could be the increase in within states regional concentration of industries.

(33)

References

Baldwin, R. E. (1997). Agglomeration and endogenous capital. Graduate Institute of International Studies: University of Geneva.

Brülhart, M. (1996). Commerce et socialization géographique dans l'Union Européene.

Economie Internationale 65, 169-202.

Brülhart, M. and Torstensson, J. (1996). Regional integration, Scale Economies and

Industry Location in the European Union. CEPR Discussion Paper N° 1435.

Castaneda, LL. and Tafunell, X. (1993). Un nuevo indicador para la historia financiera espanola: la cotización de las letras de cambio a corto plazo. Revista de Historia

Econòmica 11, 367-383.

Cordero, R. and Menéndez, F. (1978). El sistema ferroviario espanol. In Artola, M. (dir.). Los ferrocarriles en Espaha. I El Estado y los ferrocarriles. II. Los ferrocarriles

y la economia. Banco de Espana, Servicio de Estudios, Madrid.

Frax, E. (1981). Puertos y comercio de cabotaje en Espaha, 1857-1934. Banco de Espana, Servicio de Estudios, Estudios de Historia Econòmica, Madrid.

Fujita, M., Krugman, P. and Venables, A. J. (1999). The spatial economy. Cities,

Regions and International Trade. Cambridge: MIT Press.

Gômez Mendoza, A. (1982). Ferrocarriles y cambio econòmico en Espaha, 1855-1913. Madrid: Alianza.

Hanson, G. H. (1997). Increasing returns, trade, and the regional structure of wages.

Economic Journal 107, 113-133.

Krugman, P. (1991). Increasing returns and economic geography. Journal o f Political

Economy 99, 484-499.

Nadal, J. (1975). El fracaso de la revolución industrial en Espaha, 1814-1913. Barcelona: Ariel.

Ottaviano, G. and Puga, D. (1998). Agglomeration in the Global Economy: A survey of the New Economic Geography. The World Economy 21, 707-731.

Paluzie, E. (1999). Trade policy and regional inequalities. Forthcoming in Papers in

Regional Science.

Prados de la Escosura, L. and Tena, A. (1994). Protectionism in Spain. In Lindert, P. H., Nye, J. V. and Chevet, J. M. (eds.). Political economy o f protectionism and commerce,

eighteenth and nineteenth centuries. B7 Proceedings of the Xlth International Economic

History Congress. Milan: Università Bocconi, 63-76.

(34)

Sanchez-Albomoz, N. (1987). The Economic Modernization o f Spain, 1830-1930. New York: New York University Press.

Tena, A. (1989). Sector exterior. In Carreras, A. (comp.). Estadisitiças Histôricas de

Espana Siglos XIXy XX. Madrid: Banco Exterior.

Tirado, D. A. (1994). La Protection Arancelaria en la Restauration. Nuevos indicadores. Revista de Economia Aplicada 2, 183-203.

Tirado, D. A. (1996). La Protecciôn Arancelaria en la Restauracion: impactos de corto

y largo plazo. Espana, 1870-1913. PhD dissertation. Universitat de Barcelona:

Microforma.

Tirado, D.A., Paluzie, E. and Pons, J. (2000). Economic integration and industrial location. The case of Spain before WWI. Universitat de Barcelona: Documents de Treball de la Divisiô de Ciències Economiques, Juridiques i Empresarials, E00/63, february.

Tortella, G. (1970). El banco de Espana entre 1829 y 1929. La formaciôn de un banco central. In El banco de Espana. Una historia econômica. Madrid: Banco de Espafia. Tortella, G. (1981). La economia espaflola, 1830-1900. In Tufiôn de Lara, M. (dir ),

Historia de Espana. Vol. VIII. Revoluciôn burguesa, oligarqula y constitucionalismo (1824-1923), 9-167. Barcelona: Labor.

Venables, A. J. (1996). Equilibrium locations of vertically linked industries.

International Economic Review 37, 341-359.

(35)

Appendix

A.l Forward and backward linkages

We want to examine the direct effect of a change in the location of manufacturing on the price index of manufactured goods in the domestic regions.

G, =[v (W8-T,)‘-” + V (w1)'-° + Mwj-T)1 (33) G, = [x, (w„ T,)'-” + X,(w, T)' ” + X., ■(»,)' ")'■ (34)

The price index equations, (33) and (34), are symmetric and so have a symmetric solution: if X| = Xi and Y |= Y2, then there is a solution with G| = G2 and <D| = 0)2. The symmetric solution satisfies the following relationship:

G' ° = A0 .(®0 • r0)‘" + X co'° (1 + T '-°), (A .l)

where the absence of subscripts denotes that these are symmetric equilibrium values. We then linearize the price indices around the symmetric equilibrium. Around this point, an increase in a variable in one location is always associated with a change, of opposite sign but of equal absolute magnitude, in the corresponding variable in the other location. So letting dG = dGi = -dG2 and so on, we derive, by differentiating the price indices, ( l - < r ) ~ = 4 0 G ( C ) , „ ( G ) ( <a d a i\ (1-CT) — ---~ + (I + 7 ) I I -A + ( l - e r ) ---\û)0 ) (o0 \(0 J U O) ) (A.2)

If dw =0 and dwo = 0 (the supply of labor to manufacturing is perfectly elastic), then remembering that (1-cr) < 0 and T > 1, equation (A.2) implies that a change dX/X in manufacturing employment has a negative effect on the price index, dG/G. This is a form of “forward linkage” or price index effect. It means that the location with a larger

(36)

manufacturing sector has a lower price index for manufactured goods, because a smaller proportion of this region’s manufacturing consumption bears transport costs.

Now, we are interested in finding the direct effect of a change in a region’s income on each region’s nominal wage.

" , = [KoT ^ ' G / - ' + Y,G’ -' + Y2T ' ° G 2‘

w 2 = [yX - ’g ° ' + y]t' "g ° ' + y2G2° ' f

(36)

(37)

The wage equations for the domestic regions (36) and (37) are also symmetric and so have a symmetric solution. The symmetric solution satisfies the following relationship:

a>° =Y0 -T0' ° G0° 1 + (1 + T ' ° ) - Y G° ' (A.3)

where the absence of subscripts denote that these are symmetric equilibrium values. We then linearize the wage equations around the symmetric equilibrium, as we did with the price index equation. We obtain,

Assuming that the price indices remain constant (dG = 0 and dGg= 0), equation (A.4) implies that a change dY/Y on a domestic region income has a positive effect on the nominal wage of this region, dw/w. This is a form of “backward linkage” or home market effect. It means that firms in the region with a larger income can afford to pay higher wages because these firms have better access to the larger market.

A change dYiYYo in the external region income has a positive effect on the nominal wage of both domestic regions. This result arises because both regions have the same access to the external market.

(37)

EUI

WORKING

PAPERS

EUI Working Papers are published and distributed by the European University Institute, Florence

Copies can be obtained free o f charge - depending on the availability o f stocks - from:

The Publications Officer European University Instiute

Badia Fiesolana

1-50016 San Domenico di Fiesole (FI) Italy

Please use order form overleaf

(38)

Publications of the European University Institute

To The Publications Officer

European University Institute Badia Fiesolana

T50016 San Domenico di Fiesole (FI) - Italy Fax: +39-055-4685 636

e-mail: publish@datacomm.iue.it http://www.iue.it

From N a m e... Address...

□ Please send me a complete list o f EUI Working Papers □ Please send me a complete list o f EUI book publications □ Please send me the EUI brochure Academic Year 2000/2001

Please send me the following EUI Working Paper(s):

Dept, n°, author ... Title: ... Dept, n°, author ... Title: ... Dept, n°, author ... Title: ... Dept, n°, author ... Title: ... D a te ... Signature

X

(39)

Working Papers in History

Published since 1998

HEC No. 98/1

Luisa PASSER1N1 (ed.)

The Question of European Identity: A Cultural Historical Approach.

Papers by Liitzeler, Niethammer, Liakos

HEC No. 98/2

Christophe DUHAMELLE

La noblesse du Saint-Empire et le crédit: autour du Von den Reichsstandischen

Schuldenwesen de Johann Jacob Moser

(1774)

HEC No. 98/3

René LEBOUTTE

Du registre de population au registre national. Deux siècles de pratique administrative en Belgique, 1791-1991

HEC No. 98/4

Jean-Pierre CAVAILLÉ Louis Machon: une apologie pour Machiavel

HEC No. 98/5

Natacha COQUERY

Les courtisans et le crédit à Paris au XVIIIe siècle

HEC No. 98/6

Mark HÀBERLEIN

Kinship, Gender, and Business Fuilure: Merchants’ Bankruptcies and Social Relations in Upper Germany, 1520-1620

HEC No. 98/7

Roberto DI QUIR1CO

Italian International Banking, 1900-1950

HEC No. 98/8

Frances M.B. LYNCH

De Gaulle’s First Veto: France, the Rueff Plan and the Free Trade Area

HEC No. 99/1

Johannes U. MÜLLER/ Bo STRÂTH (eds) Nationalism and Modernity

HEC 99/2

Luisa PASSERINl/Polymeris VOGL1S (eds)

Gender in the Production of History

HEC 99/3

Pegerlo SAAVEDRA

Crédit et relations sociales dans la Galice rurale à la fin de l'Ancien Régime

HEC 99/4

Concepciôn GARCIA-IGLESIAS The Transmission of International Business Cycles Shocks: Evidence from the European Periphery, 1861-1913

HEC 99/5

Concepcion GARCIA-IGLESIAS Interest Rate Risk and Monetary Union in the European Periphery: Lessons from the Gold Standard, 1880-1914

HEC 2000/1

Concepciôn GARCIA-IGLESIAS The Gold Standard and its Influence on Growth Rates for the European Periphery, 1865-1913

HEC 2000/2

Daniel A. TIRADO/Elisenda PALUZIE/ Jordi PONS

Economic Integration and Industrial Location *out of print © The Author(s). European University Institute. Digitised version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(40)

HKC 2000/3

Elisenda PALUZIE/ Jordi PONS/ Daniel A. TIRADO

Trade Liberalization, Market Integration and Industrial Concentration: The Spanish Economy During the 19lh Century © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(41)

(42)

■ © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(43)

(44)