Local Multipliers and the Growth of Services: Evidence from late 19th Century USA, Great Britain, and Sweden

by Vinzent Ostermeyer, Lund University, Department of Economic History. 

Read more about Vinzent's reserach here.
Read the full paper (open access) here. 

A common periodization of economic development is that first labor shifts out of agriculture into industry and only then the service sector grows. However, such views disregard that already during the late 19^th century — a period commonly associated with rapid industrialization — services were an important part of the economy growing at rates comparable to or even faster than industry (Weiss, 1967, 1971; Hartwell, 1973; Gemmell and Wardley, 1990; Broadberry et al., 2010, 2018; Rosés and Wolf, 2019). Why did such a service economy evolve earlier than conventionally thought?

While economic historians traditionally focus on explaining the emergence of industry, less work has been devoted to studying the emergence of services (Broadberry et al., 2018). Given that the historic origins of the service sector are largely unclear, it can be useful to look at contemporary periods for possible explanations. The employment multiplier framework formulated by Moretti (2010) predicts that industrial growth should have been a key contributor to the growth of services as each new industrial jobs stimulates through increased demand the local service sector. Moretti (2010) shows that in the contemporary USA 1.6 service jobs are created for each new industrial job. This number increases to 2.5 for the creation of skilled industrial employment. The multiplier is larger for skilled industrial jobs because they require higher levels of human capital, which translates into higher productivity and wages. Additionally, relatively wealthier households tend to spend more of their income on services.

I contribute to the literature on the historical emergence of services by testing whether and if so by how much multiplier effects can explain the emergence of the service sector. Specifically, I use full-count census data across a set of countries to estimate employment multipliers in a consistent way for the late 19^th century. Using data from the IPUMS project published by the Minnesota Population Center (2019) and Ruggles et al. (2021), I build a detailed panel measuring the size of the industrial and service sectors at the regional level in the USA, Great Britain, and Sweden. In my approach, I classify the occupation, sector, and skill levels of individuals according to HISCO and HISCLASS codes.

I then follow the regression approach by Moretti (2010). For each country, I regress the change in the number of jobs (E) in the non-tradable (NT) – i.e., service – sector on the change in log total employment in the tradable (T) – i.e., industrial – sector for region r between two census years t. This approach controls for the potential bias of time-invariant factors and common shocks across regions in a given year. I use cluster-robust standard errors at the regional level in all regressions to account for serial correlation and possible heteroscedasticity in the standard errors.

β₁ measures the employment multiplier as an elasticity. To give it a more intuitive interpretation, I multiply it with the ratio of non-tradable to tradable jobs. The employment multiplier then measures the number of jobs created in the non-tradable sector given an additional job in the tradable sector. To also isolate arguably exogenous variation in the demand for labor in the tradable sector, I follow Moretti (2010) by using a Bartik shift-share instrumental variables approach where I instrument the change in tradable employment with a region-specific weighted average of the national growth in tradable employment. Intuitively, the instrument relies on the notion that cities with a higher initial share of tradable employment in industry i should experience a larger positive shock if this industry grows at the national level.

My main finding is that across countries, the addition of one tradable job in a local labor market led to the creation of 0.5 to 1 additional non-tradable job(s). Next, I divide tradable employment into a skilled and non-skilled part and use both as outcome variable. As predicted, I find that the multiplier effect is entirely driven by the creation of skilled tradable employment. The effect was largest in the USA where adding one skilled tradable job increased local service employment by about 2.5 jobs. In Great Britain and Sweden a little less than one service job was created.

One concern against utilizing multiplier effects for public policy is that they could mainly lead to the expansion of unskilled service employment. I argue that such concerns are unwarranted for the late 19^th century. By splitting the service sector into a skilled and non-skilled part and estimating separate multiplier effects for both, I show that multiplier effects contributed to the growth of both types of services.  I also show that employment multiplier effects increased employment across a range of different services and most notably personal and business services. Especially the multiplier effect for business services is noteworthy as they were key for industrialization and required relatively more skills.

This paper has several contributions to and implications for the literature. In a comparison to a step-wise understanding of economic development where industrialization occurs first and only afterwards there is an emergence of services, I show that both sectors are intimately related as a substantial part in the growth of services is due to simultaneous industrial growth. Because the industrialization during the late 19^th century created growth in other sectors as well, its role for economic development is arguably broader than often recognized.

My paper also contributes to the current literature on employment multipliers. While this literature focuses on single countries in contemporary settings (Moretti, 2010; Moretti and Thulin, 2013), I employ a unified methodology and use for the first time data harmonized across a set of countries. This enables a plethora of comparisons. Given that I follow Moretti (2010), I can compare the historic multiplier effects in the USA to the contemporary ones. Overall, both multipliers are of very similar size. Second, I can compare the employment multiplier effects across countries historically. Consistent with the theoretical framework, I find larger multiplier effects in countries that are less technologically advanced and have a higher geographic internal mobility, e.g., Sweden compared to Great Britain.

 

References: 

Broadberry, S., Federico, G., and Klein, A. (2010). Sectoral developments, 1870–1914. In Broadberry, S. and O’Rourke, K. H., editors, The Cambridge Economic History of Modern Europe, pages 59–83. Cambridge University Press, Cambridge.

Broadberry, S. N., Cain, L. P., and Weiss, T. (2018). Services in American Economic History. In Cain, L. P., Fishback, P. V., and Rhode, P. W., editors, The Oxford Handbook of American Economic History, Vol. 1, pages 234–260. Oxford University Press, Oxford.

Crouzet, F. (1982). The Victorian Economy. Methuen, London.

Gallman, R. E. and Weiss, T. J. (1969). The Service Industries in the Nineteenth Century. In Fuchs, V. R., editor, Production and Productivity in the Service Industries, pages 287–381. NBER, Cambridge, Massachusetts.

Gemmell, N. and Wardley, P. (1990). The contribution of services to British economic growth, 1856–1913. Explorations in Economic History, 27(3):299–321.

Hartwell, R. M. (1973). The Service Revolution: The Growth of Services in Modern Economy. In Cipolla, C. M., editor, The Fontana Economic History of Europe, number 3 in The Industrial Revolution, pages 358–396. Wiliam Collins Sons & Co. Ltd, Glasgow.

Klein, A. (2019). Regional inequality in the United States: Long-term patterns, 18802010. In Rosés, J. R. and Wolf, N., editors, The Economic Development of Europe’s Regions: A Quantitative History since 1900, pages 363–386. Routledge, New York.

Lee, C. H. (1979). British Regional Employment Statistics 1841-1971. Cambridge University Press, Cambridge.

Lee, C. H. (1984). The service sector, regional specialization, and economic growth in the Victorian economy. Journal of Historical Geography, 10(2):139–155.

Mathias, P. (1969). The First Industrial Nation: An Economic History of Britain, 17001914. University Paperbacks. Butler & Tanner Ltd., London.

Moretti, E. (2010). Local Multipliers. American Economic Review, 100(2):373–377.

Moretti, E. and Thulin, P. (2013). Local multipliers and human capital in the United States and Sweden. Industrial and Corporate Change, 22(1):339–362.

Minnesota Population Center (2019). Integrated Public Use Microdata Series, International: Version 7.2 [dataset], Minneapolis, MN, IPUMS, 2019. https://doi.org/10.18128/D020.V7.2.

Rosés, J. R. and Wolf, N. (2019). Regional economic development in Europe, 1900-2010. In Rosés, J. R. and Wolf, N., editors, The Economic Development of Europe’s Regions, Routledge Explorations in Economic History, pages 3–41. Routledge, New York.

Ruggles, S., Flood, S., Foster, S., Goeken, R., Pacas, J., Schouweiler, M., and Sobek, M. (2021). IPUMS USA: Version 11.0 [dataset]. Minneapolis, MN. https://doi.org/10.18128/D010.V11.0.

Schön, L. (2010). Sweden’s Road to Modernity: An Economic History. Studentlitteratur, Lund.

TombstoneWeb.com (2011). Tombstone Arizona History. https://web.archive.org/web /20110625071235/http://www.tombstoneweb.com/history.html.

Weiss, T. (1967). The Service Sector in the United States, 1839 to 1899. The Journal of Economic History, 27(4):625–628.

Weiss, T. (1971). Urbanization and the growth of the service workforce. Explorations in Economic History, 8(3):241–258.

"Wealth invested in beauty": reinterpreting Renaissance Florence and the Little Divergence from GDP estimates of 1427 Tuscany

 Jan Luiten van Zanden and Emanuele Felice

The full paper can be read here

How wealthy, and how unequal, was pre-industrial Europe? And how rich was the South of Europe compared to the North Sea area: did the Little Divergence already start in the late Medieval Period? And if this was the case, what are the reasons for the decline, perhaps starting already in the XV century, of Italy?

These are the main questions we address in this reconstruction of the historical national accounts of Tuscany in 1427. It is based on one of the most detailed, extensive and probably reliable quantitative source available for Medieval Europe, the Florentine Catasto of 1427, which has accurate information on the composition, the occupations and assets of 61,123 households in Tuscany in that year.

According to our estimates, by the early XV century Tuscany was in per capita GDP, in real terms, only slightly above England (maybe less than 20%), and slightly less above Holland (maybe around 13%); this gap is much smaller than the one resulting from the Maddison project (in 1427 Centre-North Italy has a GDP per capita between 70 and 100% higher than Holland and England, in the same period).

In addition, in the process of creating a benchmark estimate of Tuscany’s GDP in 1427, we also learn a lot about the structure of its economy. In fact, our results point at a fundamental institutional difference, between Tuscany on the one side, and England and Holland on the other: the productivity gap between industry and services on the one hand and agriculture on the other hand in Tuscany was much larger than in England and Holland. Furthermore, in Tuscany contrasts between city and countryside are exacerbated by the large income streams from agriculture to the cities (and in particular Florence). This increases the income of the urban elite – which spends it on the conspicuous consumption that gives rise to the Renaissance – and depresses rural incomes.

These findings confirm an institutional explanation of Tuscan economic decline first put forward, in pioneering works, by Stephan Epstein, in the early 1990s, and later corroborated by several studies on the basis of different sources (such as those by Van Bavel, Cohn, Alfani and Ryckbosch, Alfani and Ammannati): Tuscany was characterized by high extractive rates in favor of the elite of capital city, to the detriment of the subdued cities and, most of all, of the countryside. This led to underdeveloped markets for labor and capital in the countryside, which sharply contrasts with what we know about Holland and England, where mobility of labor and capital was much higher. In Tuscany, these blockades to market functioning in turn resulted into very low productivity in agriculture, a high share of labor in agriculture and, arguably, lower economic growth. The high extractive rate by the capital elite also brought about a very high income per head in industry and above all in services, as compared to agriculture: this may also explain why, in spite of a small difference in per capita GDP, Tuscany boasted a much richer material culture with respect to England and Holland, as testified by the florescence of the arts in the capital city at that time: the surplus income was invested in beauty.

The peculiarity of Tuscany’s institutional pattern (with respect also to other Italian territories, such as Lombardy and Sicily) helps to understand the limitation of GDP estimates based on the indirect approach, as are the series available for pre-industrial Centre-North of Italy: these tend to be heavily based on a few sources and for specific sectors and territories. In this case, the real wages of the construction workers from Florence may not be a good proxy of the entire urban sector, for a fragmented economy where labor and capital were immobile as Tuscany was; let alone for the entire Centre-North of Italy. Moreover, when it comes to international comparisons and long-run historical series differences in purchasing power parities must be properly considered, as well as their changes through time.

Our conclusions may be a useful result for other ‘fragmented’ economies as well, and for historical estimates based on a few hypotheses over-extended through time and space: cautiousness is warranted, and should always be complemented by direct evidence when available. But we also point – very indirectly – to the paradox that the beauty of the Renaissance, the unmatched brilliance of its arts, was made possible by the highly uneven distribution of wealth and income that in the long run undermined the vitality of Tuscan economy and society.  

The Panopticon of Germany’s Foreign Trade. New Facts on the First Globalization, 1880-1913.

Wolf-Fabian HUNGERLAND and Nikolaus WOLF. 

The full paper in the EREH can be read here

We are used to distinguishing between the “first” and the “second” globalization, separated not only by two world wars, but also by changes in technology and institutions, and hence their basic economic logic. The first globalization is typically described in terms of “classical” trade models of comparative advantage, where countries trade to take advantage of their differences. In contrast, the second globalization is largely described in terms of “new” trade models based on monopolistic competition and firm heterogeneity. Here, similar countries trade because they are all populated by firms exploiting economies of scale and differences in productivity.

The similarities and differences between these two globalizations are subject to a large and growing literature (Baldwin 2016, Jacks and Stürmer 2020). Given the rise in trade between very different countries like the USA and China, Paul Krugman asked during his Nobel prize lecture of 2008: “is the world becoming more classical?” (Krugman 2008). In a new paper (Hungerland and Wolf, EREH forthcoming), we describe Germany’s foreign trade 1880-1913 with new and very detailed evidence. To us, this evidence begs the question of how “classical” has the world ever been in the first place? Put differently, to what extent can “new” trade theory help us to understand the first globalization?

We have three main findings. First, and least surprising, Germany got increasingly specialized in manufacturing, notably chemicals, machinery and transport equipment. This is fully in line with predictions of “classical” trade models, and Germany having a comparative advantage in industries that use physical and human capital intensively. Second, however, we find that nearly all growth in exports and most growth in imports took place along the extensive margin, mostly driven by new products traded with old trade partners. Third, we find that between 25-30% of trade at our finest level of disaggregation is intra-industry trade, i.e. trade in the same product category. The latter two findings imply that we cannot understand the first globalization unless allowing for very substantial heterogeneity within countries and industries.

To create our data, we first digitalized all historical statistics on the foreign trade of the German customs union and the two major port cities Bremen and Hamburg that stayed outside of that customs union before 1889. Our data covers all imports and exports from 1880 to 1913 of all products, with all trade partners, captured in values and quantities. Next, we reclassified all data to the SITC system, and used a quota method to merge the Bremen and Hamburg data with that of the German customs union to create one consistent dataset. Using the SITC, we can compare this to historical trade data for other countries (e.g. Italy, see Federico and Wolf, 2012) and modern trade data. In a related paper, Hungerland and Altmeppen (2021) provide details on this and discuss, which revision of SITC is best suited to create comparable, historical and long-run trade data.

Figure 1 shows the growth of imports and exports of the German Empire, 1880-1913. This trade growth was much faster than GDP, resulting in a rising openness ratio. Moreover, Germany was catching up to the UK, to become the second largest trading nation in the world by 1913.

FIGURE 1
AGGREGATE IMPORTS, EXPORTS, TRADE BALANCE

In 1913-marks. Statistical items excluded. Source: Own calculations.

  

Table 1 is a first cut through the aggregate data: we see how imports and exports grew between 1880 and 1913 at the level of 1-digit SITC sectors. The pattern is roughly in line with “classical” trade models, where exports of manufacturing products grow more rapidly than imports, while the opposite holds for agricultural products and raw materials. However, the growth of manufacturing exports is accompanied by very strong growth in manufacturing imports, and Germany continues to export agricultural products and especially raw materials (such as coal).

 

TABLE 1
SECTORAL TRADE GROWTH

Our data allows us to dissect aggregate trade growth much further, down to the level of 5-digit SITC (the product-level) for each trade partner of Germany. The number of products traded increased from 334 in 1880 to 834 in 1913, while the number of trade partners grew from 34 (1880) to 86 (1913). Let us define each product-country combination as a variety. In 1880 we observe 971 import varieties and 1,482 export varieties. A generation later in 1913 we observe 10,145 import and 29,263 export varieties. Following Amiti and Freund (2010), we can decompose aggregate trade growth over all varieties into three margins. First, growth can occur along the intensive margin, where trade in existing varieties is expanding (“more of the same”). Second, there can be growth along the extensive margin, where new varieties enter (either old products are traded with new trade partners, new products traded with old trade partners, or new products with new trade partners). Finally, growth can occur along the extensive margin, where old varieties disappear. Figure 2 shows the relative contribution of each of these margins to Germany’s trade expansion before World War I.

 

FIGURE 2: MARGIN DECOMPOSITION, 1880-1913

Margins in percent of total trade according to eq. 1. Source: own calculations.

 

Clearly, the extensive margin dominates the picture. Interestingly, this is true for imports and exports alike. In our paper, we show that the extensive margin dominates growth in all types of trade, manufacturing and non-manufacturing, and trade within and outside of Europe. Even trade growth between Germany and the USA was dominated by the extensive margin. Within the extensive margin, the most important element is the entry of new products in trade with existing trade partners. Moreover, we show that this is unlikely to be a statistical artefact stemming from an increasing level of detail in the historical classification system. If we restrict our attention to only those products and countries that were already recorded in the first year of our sample (1880), the picture remains largely unchanged (although this certainly underestimates the extensive margin).

A related question is whether and to what extent there was intra-industry trade, hence exports and imports of the same products in a given year. In figure 3 we show, separately for non-manufacturing and manufacturing trade, what share of trade is intra-industry, varying the level of aggregation from sectors (1-digit SITC) to products (5-digit SITC).  

 

FIGURE 3
INTRA-INDUSTRY TRADE

Intra-industry trade in percent of total trade, Figure 3A: SITC sectors 0 to 4. Figure 3B: SITC sectors 5 to 8. Source: Own calculations.

We find that even at the product-level, intra-industry trade accounts for between a quarter and a third of all trade, in manufacturing and non-manufacturing. Zooming into this more closely, we find that intra-industry trade was more prominent with rich economies—i.e. mostly with European trade partners.

Overall, our evidence suggests that simple “classical” trade models that focus on differences between countries miss crucial aspects of the first globalization. Once disaggregated trade data becomes available, we see that most of the growth in imports and exports was due not to increase in the value of trade in specific products between trade partners but growth in the number of products and trade partners. Our new evidence for Germany is in line with similar findings for Belgium before 1914 (Huberman et al., 2017) and Japan before 1914 (Meissner and Tang, 2017, 2018). Such growth along the extensive margin may have been driven by changes in trade costs due to improved transportation technology, by rising incomes, or also reflect strategies to differentiate products by branding and quality. But together with the evidence on large-scale intra-industry trade this suggests that heterogeneity within countries and industries should be systematically taken into account.

Yet, “classical” trade models are obviously not dead. At a very broad level they seem to capture, how countries specialized before 1914 – after all, they were invented to capture exactly this. Germany did specialize in manufacturing products, while agriculture in Germany and elsewhere in Europe was increasingly exposed to import competition: the “European grain invasion” (O’Rourke 1997) was very real. Farmers responded either by giving up and leaving agriculture, lobbying for protection, or shifting to different products (Suesse and Wolf 2020).

To us this suggests thinking about globalization along the lines of a hybrid model such as Bernard et al. (2007) that combines comparative advantages at the country-level with heterogeneity at the firm-level. Doing so might have far-reaching implications for our interpretation of the costs and benefits, the winners and losers, and hence for the political economy of globalization. With more disaggregated trade data available for more countries – hopefully in standardised and comparable form -, this is opening the door for a new understanding of the history of globalization.

 

 

 

References

Amiti, M. and C. Freund (2010), "The Anatomy of China's Export Growth," NBER Chapters, in: China's Growing Role in World Trade, pages 35-56, National Bureau of Economic Research, Inc.

Baldwin, Richard (2016), The Great Convergence. Information, Technology and the New Globalization. The Belknap Press of Harvard University Press, Cambridge.

Bernard, Andrew B., Stephen J. Redding and Peter K. Schott (2007), “Comparative Advantage and Heterogeneous Firms”, Review of Economic Studies, 74, 31–66.

Federico, G. and Wolf, N. (2012). “A Long-run Perspective on Comparative Advantage”, In: G. Toniolo (Ed.), The Oxford Handbook of the Italian Economy since Unification, Oxford: Oxford University Press, chap. 12. 327–350.

Huberman, M., C. Meissner and K. Oosterlinck (2017), “Technology and Geogrpahy in the Second Industrial Revolution: New Evidence from the Margins of Trade“, The Journal of Economic History 77 (1), 39-88.

Hungerland, Wolf-Fabian and C. Altmeppen (2021), “What is a product anyway? Applying the Standard International Trade Classification (SITC) to historical data”, Historical Methods: A Journal of Quantitative and Interdisciplinary History.

Hungerland, Wolf-Fabian and Nikolaus Wolf (2021), “The Panopticon of Germany’s Foreign Trade, 1880-1913. New facts on the First Globalization”, European Review of Economic History (forthcoming).

Jacks, D. and  M. Stuermer (2020), “What drives commodity price booms and busts?”, Energy Economics, 85, 104035.

Krugman, P. (2008), “The Increasing Returns Revolution in Trade and Geography”, The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2008, Prize Lecture.

Meissner, C., & Tang, J. (2017). “New goods and markets versus more of the same: Japan's entry to world markets during the first age of globalisation”, VOXEU.org, 16 June 2017.

Meissner, C., & Tang, J. (2018), “Upstart Industrialization and Exports: Evidence from Japan, 1880–1910”, The Journal of Economic History, 78(4), 1068-1102.

O’Rourke, K. H. (1997), “The European Grain Invasion, 1870-1913”, The Journal of Economic History, 57 (4), 775-801.

Suesse, M. and N. Wolf (2020), “Rural transformation, inequality, and the origins of microfinance”,

Journal of Development Economics, 143, 102429.

 

 

 

The paradox of ‘Malthusian urbanization’: Urbanization without growth in the Republic of Genoa, 1300 to 1800

By Luigi Oddo (University of Genoa, Department of Political Science) and Andrea Zanini (University of Genoa, Department of Economics).

From the second half of the 20^th century, the role of urbanization in the development process has become an extensively investigated topic. In standard urbanization models, from the pioneering studies by Lewis (1954) on urban pull factors, then passing on the great classics of the subject by de Vries (1984) and Bairoch (1988), who stressed rural push factors, cities have almost always been represented as a factor of economic development in the pre-industrial world. Urbanization levels and city sizes have often been used as empirical proxies for the level of income per capita (De Long and Shleifer 1993; Acemoglu et al. 2002, 2005; Malanima 2005; Maddison 2008; Dittmar 2011).

The role of urbanization as a factor of economic growth has also been included in the literature concerning Malthusian population theory. Clark (2007) and Voigtländer and Voth (2009, 2013) suggested that high urbanization rates helped keep down fertility and to drive up death rates, allowing living standards to rise, but through purely Malthusian mechanisms. Overall, both urbanization models and Malthusian population theory basically support the principle, even if through different mechanisms, city growth was a factor of economic advancement in the pre-industrial world.

However, in the last 20 years, the idea that urbanization level is closely correlated with levels of income and growth has started to be questioned, especially in the literature concerning developing countries (Fay and Opal 2000, Henderson et al. 2013, Gollin et al. 2016). The experience of the Third World in the second half of the 20^th century clearly shows that changes in income do not explain shifts in urbanization. Urbanization continues even during periods of negative growth.

On this basis, we aimed to shed light on the relationships between urbanization and economic growth in the pre-industrial era, analyzing the case of an Italian pre-unification state, the Republic of Genoa, from 1300 to 1800 with a novel dataset of cities and rural populations. Genoa was one of the most powerful Italian maritime republics, probably exhibiting one of the highest degrees of urbanization in Europe in the Late Middle Ages. However, the history of the Republic of Genoa was described by cyclical Malthusian stagnations, which were characterized by almost flat-lined growth at the population level (Figure 1). 

Figure 1: urbanization (upper) and population (lower) of Genoa 1300-1800

The coexistence of these contradictory elements raises the following question: is a high degree of urbanization always a sign of economic advancement in the pre-industrial world?

To answer this question, our paper introduces elements of rural-urban migration models within the Malthusian population theory to provide a different perspective on the interactions between urbanization levels and demographic dynamics. Focusing on the Republic of Genoa, this approach brings out that if high levels of urbanization do not reflect substantial increasing productivity in agriculture and growing urban labor demand, urbanization per se is not sufficient to take-off from Malthusian stagnation. Therefore, the rise in urbanization cannot match sustained economic growth. The natural consequence is urbanization growth that follows a stop-and-go trend, where city populations inflate and deflate cyclically. To describe this phenomenon, we coined the term ‘Malthusian urbanization’.

Read the full paper here