1 Second Industrial Revolution in Italy (1860-1913) Abstract, Lecture notes of Economic history

Download 1 Second Industrial Revolution in Italy (1860-1913) Abstract and more Lecture notes Economic history in PDF only on Docsity! 1 Second Industrial Revolution in Italy (1860-1913) Renato Giannetti University of Florence Department of Historical and Geographical Studies Margherita Velucchi University of Florence Department of Statistics Abstract The paper deals with two topics largely debated in economic history of the Italian “catch up” in the last quarter of 19th century. Some economic historians emphasize the discontinuity (Mori, 1992; Giannetti, 1998; Vasta, 1999) analyzing the Italian case in the context of the Second Industrial Revolution, when new “science based” sectors, like chemicals and electricity, drove the pattern of economic growth. Others (Bonelli, 1979; Cafagna, 1989; Federico, 1995) emphasize the continuity of the Italian growth, which was due essentially to the comparative advantage of a late comer in manufacturing sectors like textile (silk). We use the time series approach of cointegration and common trends to give a new perspective about these topics searching which were the leading sectors and the extent of interindustry linkages. We find that the new sectors – chemicals and electricity- representing the new “technological regime” of the Second Industrial Revolution, were the leading ones even in Italy, showing a common trend with the aggregate industrial production. A Granger causality approach, both among the cointegrated sectors and between them and all the others, confirms finally that electricity and chemicals where the leading sectors, suggesting that the new “science-based” industries were at the origins of the Italian “catch up” of the Giolitti’s Age (1890-1913). 2 Introduction There is little agreement on the industrialization of Italian economy in the second half of the 19 th century. Some scholars (Romeo,1961; Mori, 1992; Sereni, 1966) assess that was essentially the public policy effort to establish an iron and steel industry to push the Italian “early start” between the 1880s and the 1910s. Others ( Gershenkron, 1962; Fenoaltea, 1973) claim that iron and steel represented a too narrow industrial base to ensure a self sustained industrial growth, lacking an appropriate integration with machinery and equipment sector, which moreover presented a production function more adequate to the Italian comparative advantage. Others, (Bonelli, 1979;Cafagna, 1989) propose an alternative approach, based essentially upon trade data, to assess the crucial role of the traditional sectors ( food, textiles, especially silk) in sustaining the industrial growth of Italy along the entire 19 th century, and later too. Few scholars, finally, (Giannetti, 1998; Vasta, 1999), suggest that the economic boom of the Giolitti’s Age can be described in the context of the Second Industrial Revolution regime, characterized by the rise of new industrial sectors: chemicals and electricity. Here we investigate the matter by utilizing modern time series methods to explain which were the industrial sectors driving the Italian growth from the Unification (1860) up to the early 20 th century, and the interrelationship among them (Fenoaltea, 2003). The basic idea of time series methods is that individual industries output movements give information on the extent that either common or industry specific forces drive aggregate industrial growth. These methods identify the extent to which common features are present in individual data and whether a single or small number of stochastic trends represent the data. The smaller the number of common stochastic trends (the greater the existence of cointegration) in disaggregated industrial production, the more pervasive are the effects of one or more industry specific productivity shocks. On the contrary, if the effects of industry specific productivity shocks were localized, any industry would have distinct output trends. The methods of cointegration and common trends help to answer also to other two largely debated questions of the Italian economic growth between 1860 and 1913, i.e. which were the leading sectors and the extent of interindustry linkages. The most part of economic historians underline the reduced interrelatedness of Italian industrial matrix in this period, even emphasizing different causes for it: a low aggregate level of demand (Bonelli, 1978), an inadequate tariff policies for machinery and equipment industry (Gershenkron, 1962), etc. Extension of the analysis to consider causal relationship among industries follows from the 5 The next paragraphs concerns, firstly, how many sectors show stochastic common trends and, secondly, which of them represent the key sources of growth of Italian industrial production 4 . 2. Common Trends among sectors In the previous paragraph, we showed that a “small” number of sectors shaped the growth of the industrial production in Italy during 1861-1913. Now we consider whether the stochastic trends driving overall industrial output can be associated with specific industrial sectors. We apply the same procedure used in the previous paragraph for all sectors to identify what sectors have the same common trend. In other words, we look for the sub - groups of sectors showing the same stochastic trend and therefore driving the growth of the industrial production. Table 3 presents the results of the Johansen Cointegration Test for the sub - group involving the Machinery and equipment, Electrical and Chemical sectors 5 , showing the existence of two significant cointegrating vectors at 1% significant level; this implies that the output of all the sectors in the group were shaped by a single stochastic trend. Starting from this result, we added other sectors to this group, searching for an enlargement of the initial sub group. Many alternative sub-groups have been tested and most of them show that the stochastic common trend among the sectors involved completely disappears 6 . However, in Table 4 we proved the existence of a common stochastic trend among a wider sub group of sectors at a lower significant level (5%). To summarize, using the Johansen procedure, we identified a core group of sectors (machinery and equipment, chemicals and electricity) showing a very strong relation (a significant common trend at 1% level) and a wider group ( the previous sectors plus Food, Leather, Printing and Paper, Clothes, Iron and Steel), showing the existence of a common trend but with a lower significant level (5%). Finally, we can exclude at 1% significant level that Textiles, Non metallic minerals, Mining, Miscellaneous and Wood sectors share a common trend both with the core group and the wider one. Moreover, they do not exhibit a common trend among them too. 4 The procedure follows Greasley and Oxley (2000) even if it is slightly different due to the structure of data. They have industry groups hence, they, firstly, search for the common trend within the industry, then look for a common trend between the industries. 5The historians discussions focused on these as key sectors to explain the effect of the Second Industrial Revolution, hence we started from them exploring the cointegrating relations among the sectors. 6 The results of the tests concerning different sub groups are not included here but they are disposable on request to the authors. 6 The method of common trends clarifies which are the sources of industrial growth by reducing the range of sector-specific forces shaping the aggregate industrial production; moreover, it highlights which were the sectors whose output shaped the Second Industrial Revolution in Italy. The up-trend of the 1890s, “Giolitti’s age” arose from a small number of sectors: Chemicals, Electricity and Machinery and equipment, which show a single common stochastic trend at the highest significant level, were the sources of growth for the permanent components of the industrial production. The wider group sharing the common trend with them at a lower significant level represent a weaker (in probabilistic terms) source for the industrial production growth in the same period; they share the technological shocks which pushed Chemicals, Electricity and Machinery and equipment, but they do not influence the Italian industrial production as well. The next paragraph extends the previous results analysing the causality relation ( Granger Causality) both among the cointegrated sectors and between them and all the others to investigate the interdependences among sectors and the eventual existence of leading sectors. 3. Causality in Industrial Production The previous paragraph shows the existence of a group of cointegrated sectors driving the overall industrial production in Italy in 1861-1913. The core group involves three technologically advanced sectors (Machinery and equipment, Electricity, Chemicals) and their relation is significant at 1% level, while a wider group involving the previous plus other five sectors is detected at a lower significant level (5%). Now the central issue is to verify whether particular industries within the cointegrated group defined above had causal linkages which spilled across the common trend groupings. In fact cointegration and common trends provide an alternative explanation to the expenditure-based input-output approach to measure interrelations among sectors. Extension of the analysis to consider possible causal relationships among sectors growth follows from a discussion of common trends. For those sectors driven by common forces, causality tests within and across cointegrated groups help to describe the sources of growth. Even if technological progress spilled across the economy, the sources of innovation may have been located in particular sectors. To test the existence and the direction of causal relations we used the Granger- type causality tests. The Granger (1969) approach to the question of whether x causes y is to see how much of the current y can be explained by past values of y and then to see whether adding lagged values of x can improve the explanation of y: y is said to be Granger-caused by x if x helps in the prediction of y, or equivalently if the 7 coefficients on the lagged x's are statistically significant. It is important to note that the statement " x Granger causes y" does not imply that y is the effect or the result of x. Granger causality measures precedence and information content but does not by itself indicate causality in the more common use of the term (Greene, 2000). Various tests of Granger-type causality have been derived, in this work we run bivariate regressions of the form: ),(, , 1:1: 0 1:1: 0 yxuyxx xyy t l i iti l i itit t l i iti l i itit ∀+++= +++= ∑∑ ∑∑ −− −− βαα εβαα The reported F-statistics are the Wald statistics for the joint hypothesis: 01 100 : 0...: HH H l ==== βββ If the test does not reject the null hypothesis, this means that “x does not Granger Cause y” 7 . We concentrated on the non-stationary, cointegrated sectors, specifically focusing on those also ascribed in the historiography as having a key role in leading the Second Industrial Revolution. In particular, we consider Chemicals, Machinery and equipment, Electricity both for their role in the Second Industrial Revolution and for their strong common trend highlighted in the previous paragraph; then we analyse also the causal relations of Iron and Steel, Food and Clothes, even sharing a weaker common trend with the previous group. Finally we consider Textiles, Mining and Miscellaneous (shipbuilding), even if they do not show any common trend, because they represent crucial sectors – especially textiles - in the Italian industrial production between 1861 and 1913. We ran the bivariate regressions for all these sectors to discover the existence of causal relations between them and among them and all the sectors 8 . The results are shown in Table 5. According to them, the sectors with the most pervasive links to other sectors (at 1% significant level) are Chemicals (6 links), Electricity and Miscellaneous (4 links), then Textiles (3 links) and Machinery and equipment (2 links). At a first sight, the results from the causality tests confirm the strength of the cointegrating relations described above: Chemical, Electrical and machinery and equipment sectors share a common trend and are the leading industrial sectors in Italy between 1861-1913. Moreover, 7 The number of significant lags, l, is chosen according the Akaike’s Information Criterion, augmented by extra lags depending on the order of integration of the series. The I(1) series are added by one extra lag to each variable in the equation. 8 Notice that any causal link among the non-stationary sectors may be long term, since their output movements have permanent effects. 10 Tables and Graphs: Figure I: Sectors Series (1861-1913) 80 120 160 200 240 280 65 70 75 80 85 90 95 00 05 10 CLOTHES 400 500 600 700 800 900 1000 65 70 75 80 85 90 95 00 05 10 FOOD 0 50 100 150 200 250 300 65 70 75 80 85 90 95 00 05 10 PAPER 0 40 80 120 160 200 65 70 75 80 85 90 95 00 05 10 CHEMICALS 0 40 80 120 160 200 240 65 70 75 80 85 90 95 00 05 10 ELECTRICITY 20 40 60 80 100 120 140 160 65 70 75 80 85 90 95 00 05 10 MINING 100 150 200 250 300 350 400 450 65 70 75 80 85 90 95 00 05 10 WOOD 40 80 120 160 200 240 280 65 70 75 80 85 90 95 00 05 10 NON_METALLIC_MINERALS 100 200 300 400 500 600 700 800 900 65 70 75 80 85 90 95 00 05 10 MACHINERY_AND_EQUIPMENT 0 20 40 60 80 100 120 65 70 75 80 85 90 95 00 05 10 IRON 80 120 160 200 240 280 320 65 70 75 80 85 90 95 00 05 10 LEATHER 100 200 300 400 500 65 70 75 80 85 90 95 00 05 10 TEXTILES 18 20 22 24 26 28 30 65 70 75 80 85 90 95 00 05 10 TOBACCO 4 8 12 16 20 24 28 32 65 70 75 80 85 90 95 00 05 10 MISCELLANEOUS 11 Figure II: Industrial Production Series (1861-1913): 1000 2000 3000 4000 5000 6000 65 70 75 80 85 90 95 00 05 10 INDUSTRIAL_PRODUCTION Table I: KPSS Unit Root Test Sectors Levels 1st Difference Clothes 0,43 0,04 Food 1,16 0,20 Printing and Paper 64,70 0,10 Chemicals 0,57 0,20 Electricity 0,35 0,58 Mining 2,07 0,09 Non metallic minerals 0,82 0,12 Wood 0,25 0,05 Machinery and equipment 99,50 0,08 Iron and Steel 33,90 0,10 Leather 2,08 0,14 Tobacco 0,10 0,05 Textiles 0,37 0,02 Miscellaneous 1,55 0,21 Spectral Estimation Method: Parzen Kernel, Andrews Bandwidth; Trend and Intercept. Asymptotic Critical Values: 1%: 0,216; 5%: 0,146; 10%: 0,119. *: Spectral Estimation Method: Parzen Kernel, Andrews Bandwidth; Intercept. Asymptotic Critical Values: 1%: 0,739; 5%: 0,463; 10%: 0,347 12 Table II: Cointegration Test (Johansen, 1987): All Sectors Hypothesized No. of CE(s) Trace Stat. Max Eigenvalue stat. None 829,96** 172,22** At most 1 657,73** 148,20** At most 2 509,52** 111,86** At most 3 397,66** 87,74** At most 4 309,91** 69,68** At most 5 240,23** 81,50** At most 6 178,73** 51,71** At most 7 127,01** 36,10 At most 8 90,91** 31,11 At most 9 59,79** 29,32* At most 10 30,46* 18,23 At most 11 12,23 12,19 At most 12 0,03 0,03 *: significant at 5% level, **: significant at 1% level 15 Textiles Clothes 0,09 0,64 Food 0,000** 0,81 Paper 0,14 0,91 Chemicals 0,26 0,05 Electricity 0,89 0,91 Mining 0,03 0,12 Non metallic minerals 0,001** 0,87 Wood 0,000** 0,50 Machinery and equipment 0,06 0,86 Iron and Steel 0,03 0,69 Leather 0,26 0,03 Miscellaneous 0,34 0,85 Food Clothes 0,09 0,000** Paper 0,000** 0,23 Chemicals 0,29 0,000** Electricity 0,10 0,03 Mining 0,005** 0,18 Non metallic minerals 0,000** 0,88 Wood 0,000** 0,21 Machinery and equipment 0,005** 0,59 Iron and Steel 0,000** 0,97 Leather 0,27 0,80 Textiles 0,81 0,69 Miscellaneous 0,03 0,000** Mining Clothes 0,02 0,03 Food 0,45 0,005** Paper 0,18 0,12 Chemicals 0,56 0,10 Electricity 0,49 0,24 Non metallic minerals 0,20 0,63 Wood 0,03 0,09 Machinery and equipment 0,03 0,23 Iron and Steel 0,13 0,33 Leather 0,87 0,27 Textiles 0,12 0,23 Miscellaneous 0,46 0,03 Clothes Food 0,000** 0,09 Paper 0,46 0,10 Chemicals 0,04 0,10 Electricity 0,10 0,17 Mining 0,03 0,01** Non metallic minerals 0,001** 0,95 Wood 0,000** 0,63 Machinery and equipment 0,004** 0,34 Iron and Steel 0,005** 0,91 Leather 0,57 0,004** Textiles 0,64 0,61 Miscellaneous 0,87 0,08 Miscellaneous Clothes 0,19 0,87 Food 0,07 0,03 Paper 0,000** 0,79 Chemical 0,15 0,07 16 Electricity 0,91 0,77 Mining 0,05 0,46 Non metallic minerals 0,000** 0,000** Wood 0,003** 0,08 Machinery and equipment 0,03 0,95 Iron and Steel 0,002** 0,78 Leather 0,13 0,15 Textiles 0,74 0,34 *:Significant at 5%, **: Significant at 1% 17 References Bernard, A. B., Durlauf, S. N. “Convergence in International Output”, Journal of Applied Econometrics, 10, 1995, pp. 97-108 Bonelli, F. “Il capitalismo italiano. Linee generali di interpretazione”, in R. Romano e C. Vivanti (a cura di) Storia d’Italia. Annali, vol. I. Dal feudalesimo al capitalismo, Torino, Einaudi, 1979, pp. 1195-1255. Cafagna, L. Dualismo e sviluppo nella storia d’Italia, Venezia, Marsilio, 1979. Engle R., Granger, C.W.J., “Cointegration and Error Correction Representation, Estimation and Testing”, Econometrica, 55, 1987, pp.251-276, Federico, G. (a cura di) The Economic development of Italy since 1870, Aldershot, Elgar. 1994 Fenoaltea S. “ Lo sviluppo dell’industria dall’Unità alla Grande Guerra: una sintesi provvisoria”, in Toniolo G., Ciocca P. ( a cura di), Storia economica d’Italia, vol. III, : Industrie, mercati, istituzioni, t.1, Le strutture dell’economia, Roma, Bari, Laterza 2003, pp. 137-193. Fenoaltea, S.“Riflessioni sull’esperienza industriale italiana dal Risorgimento alla prima Guerra Mondiale”, in G. Toniolo (a cura di), Lo sviluppo economico italiano 1861-1940, Bari, Laterza, 1973, pp. 121-156. Gerschenkron, A. Economic backwardness in historical perspective, Cambridge (Mass.), Harvard University Press; 1962; trad.it. Il problema storico dell’arretratezza economica, Torino, Einaudi, 1965. Giannetti, R., Tecnologia e sviluppo economico italiano 1870-1990, Bologna, Il Mulino, 1998 Granger, C.W.J., “Investigating Causal Relations by Econometric Models and Cross-Spectral Methods”, Econometrica, 37, 1969, pp.424-438. Greasley, D., Oxley, L., “British Industrialization, 1815-1860: A Disaggregate Time-Series Perspective”, Explorations in Economic History, 37, 2000, pp.98-119. Greene, W.H., Econometric Analysis, Prentice Hall International Inc. 2000 Johansen, S. “Estimation and Hypothesis Testing of Cointegration Vectors in Gaussian Vector Autoregressive Models”, Econometrica, 59, 1991, pp. 1551-1580. Johansen, S., “Statistical Analysis of Cointegration Vectors”, Journal of Dynamics and Control, 12, 1988, pp. 231-254. Johansen, S., Likelihood based Inference on Cointegration in the Vector Autoregressive Model, Oxford:Oxford University Press, 1995 Kwiatkoowski,D., Phillips P.C.B., Schmidt,P., Shin, Y.,“Testing the Null Hypothesis of Stationarity against the Alternative of a Unit Root: How Sure Are We that Economic Time Series Have a Unit Root?”, Journal of Econometrics, 54, 1995, pp. 159-178 Mori, G. “ L’economia italiana dagli anni ottanta alla prima guerra mondiale”, in G. Mori (a cura di), Storia dell’industria elettrica in Italia, vol. I: Le origini, Roma-Bari, Laterza.1994 , pp. 1-106 Sereni, E. Capitalismo e mercato nazionale, Roma, Editori Riuniti, 1966 Stock, J., Watson, M., “Testing for Common Trends”, Journal of American Statistical Association, 83, 1988, pp.1097-1107. Vasta M. Innovazione tecnologica e capitale umano in Italia (1880-1914). Le traiettorie tecnologiche della seconda rivoluzione industriale, Bologna, Il Mulino, 1999.