Forging Ahead, Falling Behind and Fighting Back Page 5
American overtaking was based on superior productivity performance. The estimates reported in Table 3.2 show that labour productivity growth in the United States was superior to that in the United Kingdom across the whole of the private sector during the forty years after 1870. Although much discussion has concentrated on comparisons of manufacturing, we see that the productivity growth gap was much larger in a number of other sectors and superior American performance in services was more important in the overall picture. It should also be noted that, over the period 1871–1911 as a whole, British labour productivity growth was a bit faster than the rate achieved in the first half of the nineteenth century and American outperformance owes much more to acceleration in the United States than slowdown in the United Kingdom.
Table 3.2 Sectoral labour productivity growth before the First World War (% per year)
UK, 1871–1911 USA, 1869–1909
Agriculture 0.6 1.0
Industry 0.9 1.5
Manufacturing 1.1 1.3
Construction 0.1 2.0
Utilities 1.4 4.0
Services 0.4 1.1
Transport and communications 0.7 2.5
Distribution −0.3 1.3
Finance and services 0.9 1.4
Government 0.5 0.0
Whole economy 0.8 1.5
Source: Broadberry (2006).
This point is reinforced by the growth accounting estimates reported in Table 3.3. These show that after the Civil War productivity growth in the United States surpassed anything that Britain had achieved during the Industrial Revolution and was far in advance of its own pre-Civil War performance. Initially, this represented post-war recovery but, by the end of the nineteenth century, had moved to a new level in the context of the so-called Second Industrial Revolution. During the early decades of the twentieth century, the United States would be in the forefront of the development of the most important new technologies, including aviation, the internal combustion engine, mass production, electricity and petrochemicals (Mowery and Rosenberg, 2000). This American prowess in technological progress is reflected in Table 3.3 in a significant acceleration in TFP growth to a pace far above that seen in the First Industrial Revolution.1 This acceleration was not matched by the United Kingdom and, indeed, TFP growth seems to have been very disappointing in the immediate pre-First World War period, as was stressed by Feinstein et al. (1982), who identified this as a period that could be described as a ‘climacteric’.
Table 3.3 Contributions to labour productivity growth (% per year)
Education Capital per hour worked TFP Labour productivity growth
United Kingdom
1856–1873 0.2 0.8 1.2 2.2
1873–1899 0.3 0.4 0.5 1.2
1899–1913 0.3 0.4 −0.2 0.5
1913–1924 0.3 1.3 0.6 2.2
United States
1800–1855 0.0 0.2 0.2 0.4
1855–1871 0.0 0.5 −0.4 0.1
1871–1890 0.0 0.8 1.0 1.8
1889–1909 0.3 0.5 1.3 2.1
Note: For United Kingdom estimates are for the whole economy and for the USA are for the market sector. Labour productivity growth is measured in terms of GDP per hour worked.
Sources: United Kingdom: Feinstein et al. (1982), Matthews et al. (1982); United States: Abramovitz and David (2001) and Kendrick (1961) with education derived using Morrisson and Murtin (2009) for 1889–1909.
The notion of a climacteric is of a sharp reduction in trend growth and, as proposed by Feinstein et al. (1982), a cessation of TFP growth between 1899 and 1913. It should be accepted that the existence of a climacteric is not decisive with regard to the growth-failure hypothesis. For example, it has been suggested that a hiatus in British productivity growth resulted from a waning of technological opportunity in a pause between general-purpose technologies with the steam age petering out before the electricity era took over (Lipsey et al., 1998). On the other hand, a constant trend rate of growth could represent a failure if acceleration was possible, as was underlined by Crafts et al. (1989). On balance, however, establishing that there was a climacteric in TFP growth would strengthen the hand of those arguing for a growth failure and would imply that American overtaking owed a good deal to a slowdown in British as well as a speeding up of American growth. So was there a late Victorian/Edwardian climacteric?
Table 3.4 displays the estimates from which Feinstein et al. (1982) inferred a climacteric. They based their analysis on endpoint calculations of growth rates between business-cycle peaks using the ‘compromise’ measure of GDP, which is a geometric mean of the expenditure, income and output measures.2 Solomou and Weale (1991) suggested the ‘balanced’ measure, namely, weighting these alternatives according to their reliability rather than equally, which dilutes the slowdown a bit. The more important point is to note that what Table 3.4 reveals is one business cycle with strong growth in the 1890s followed by a very weak growth cycle between 1899 and 1907 and then a bounceback. This suggests investigating whether the differences between the various periods are statistically significant, and it turns out that generally they are not. A more sophisticated analysis based on time-series econometrics does show a decrease in trend growth after 1899 but of only about 0.1 percentage points per year (Crafts et al., 1989).
Table 3.4 Growth of real GDP and TFP, 1856–1937 (% per year)
GDP TFP
Compromise Balanced Compromise Balanced
1856–1873 2.2 0.8
1873–1882 1.9 1.7 0.6 0.4
1882–1889 2.2 1.6 0.9 0.2
1889–1899 2.2 2.2 0.8 0.8
1899–1907 1.2 1.4 −0.3 −0.1
1907–1913 1.6 1.7 0.4 0.5
1924–1929 2.6 2.4 1.2 1.0
1929–1937 2.0 2.0 0.6 0.6
Note: Education included in TFP growth.
Sources: 1856–1913: Compromise estimate from Feinstein et al. (1982) and balanced estimate from Solomou and Weale (1991); 1924–1937: Compromise estimate from Matthews et al. (1982) and balanced estimate from Sefton and Weale (1995).
Those who have supported the idea of a climacteric have sometimes supposed that manufacturing productivity growth faltered as the impetus from steam power waned and old industries, with weak productivity performance, retained too high a weight in the structure of the economy. These ideas are misleading, although it is, of course, true that some steam-intensive sectors such as coal and cotton had experienced much higher productivity growth earlier in the nineteenth century. Investment in steam power remained very strong as technological progress continued to reduce its cost, and its contribution to industrial productivity growth peaked in the decades before the First World War, as can be seen in Table 3.5. Overall, the hypothesis of a serious climacteric is not convincing.
Table 3.5 Steam power growth and British industrial output and labour productivity growth (% per year)
1800–1830 1830–1850 1850–1870 1870–1910
Rates of growth
Steam HP 5.2 5.7 6.3 4.8
Steam HP/worker 2.6 4.1 5.0 3.5
TFP in steam power 0.06 1.2 3.5 1.7
Contributions to output growth
Steam capital 0.17 0.18 0.35 0.53
TFP in steam power 0.03 0.10 0.13 0.12
Total 0.20 0.28 0.48 0.65
Contributions to labour productivity growth
Steam capital deepening 0.09 0.13 0.28 0.39
TFP in steam power 0.03 0.10 0.13 0.12
Total 0.12 0.23 0.41 0.51
Memorandum item (%)
Steam engine income share 3.3 3.2 5.6 11.1
Source: Crafts and Mills (2004).
3.2 Improved Growth Potential
The late nineteenth-century British economy clearly had a higher growth potential than had been the case at the time of the Industrial Revolution, and this is reflected in the stronger rate of labour productivity growth – 1.2 per cent per year in 1873–1899 compared with 0.3 per cent per year in 1800–1830 (Tables 2.5 and 3.3). By now, the economy was achieving a higher rate of capital deepening and e
ducational standards were rising quickly. Innovative effort had intensified in an economy more conducive to endogenous innovation and TFP growth was more broadly based than during the early nineteenth century. Even so, this was not enough to match the advance in growth potential in the United States, which had moved to a level far ahead of the leader in the First Industrial Revolution.
The simplest measure of the quantity of formal education of the labour force shows a considerable increase so that average years of schooling had advanced from about two years in 1820 to 4.2 in 1870 and 6.75 in 1911 (Morrisson and Murtin, 2009). The quality of education had also improved, notably in terms of a rapid expansion of scientific and technical education beginning in the late nineteenth century with significant investments in municipal technical colleges, polytechnics and red-brick universities (Sanderson, 1988). Apprenticeship flourished with about 4 per cent of the industrial workforce in this category compared with about 1 per cent a century earlier (Wallis, 2014). Overall, on the eve of the First World War, the British labour force more than matched the United States in terms of skills per worker used in production (Broadberry, 2003).
In the late nineteenth century, capital deepening contributed 0.4 percentage points per year to labour productivity growth. This was higher than at the start of the century partly because the investment rate had risen somewhat (Tables 2.8 and 3.6) and partly because population growth had slowed down. If all savings had been used at home rather than finding higher returns abroad, the share of investment in GDP would have been some 5 percentage points higher. The capital market had become much more sophisticated; the market value of corporations was 256 per cent of GDP in 1910 (Hannah, 2015) and the stock of corporate bonds had grown to 22 per cent of GDP in 1909 compared with only 0.5 per cent fifty years earlier (Coyle and Turner, 2013). Clearly, further improvements were necessary. For example, company law was inadequate in various respects, which meant company accounts did not present adequate information (Edwards, 1989) and that the market for new issues was inefficient (Foreman-Peck, 1990). Accordingly, an adequate market for corporate control through takeovers did not exist (Hannah, 1974).
TFP growth in the last quarter of the nineteenth century at 0.5 per cent per year was just a little higher than in the first quarter. Nevertheless, it is probably fair to say that the potential for growth based on innovation was stronger than at the time of the Industrial Revolution. In part, this is because TFP growth was less reliant on one star sector, cotton, whose contribution was now much smaller with its TFP growth at about 0.5 per cent per year rather than around 3.5 per cent.3 In part, incentive structures for endogenous innovation were somewhat more favourable as was reflected in the strength of independent invention under an improved patent system (Nicholas, 2011) and the beginnings of corporate R & D, albeit still on a small scale (Edgerton and Horrocks, 1994). Notably, the greater availability of useful knowledge and trained research personnel allowed science-based technological progress and reduced the costs of innovation (Mokyr, 2002), while the allocation of talent was less distorted as the relative returns to the rent-seeking activities of the early nineteenth century declined.4
Nevertheless, in the era of the Second Industrial Revolution, technological leadership moved inexorably to the United States, which exhibited a higher growth potential than contemporaneous or industrial-revolution Britain. Ultimately, the United States made larger investments in advanced human capital and the knowledge economy, which would become central to technological progress in the twentieth century, as is reflected in Table 3.6.
Table 3.6 The environment for endogenous innovation
UK, 1913 USA, 1920
GDP ($1990GK, mn.) 224.6 598.4
Non-residential capital stock ($1990GK, mn.) 301.0 1182.2
Population (mn.) 45.6 106.9
Non-residential investment/GDP (%) 7.4 12.5
R & D expenditure/GDP (%) 0.02 0.25
University students (% population) 0.07 0.56
Civil engineers/employment (%) 0.05 0.13
Traditional professions/employment (%) 0.53 0.95
Years of schooling of labour force 6.47 7.84
TFP growth 0.5 1.4
Note: United Kingdom TFP growth is for 1873–1899 which is regarded as a better reflection of potential than the blip in 1899–1913.
Sources: Crafts (1998) updated, dataset for Broadberry (2006) and Morrisson and Murtin (2009).
3.3 Did Late Victorian Britain Fail?
The heading of this section is also the title of a famous article written by McCloskey. In it he claimed that in the pre-First World War period the British economy was ‘growing as rapidly as permitted by the growth of its resources and the effective exploitation of the available technology’ (1970, p. 451). This conclusion was based on three very neoclassical arguments. First, using the insights of the traditional growth model, it was argued that devoting more resources to home investment would have run into diminishing returns. Second, it was claimed that the technical choices made by British firms were efficient and that the highly competitive market environment ensured that there would be no serious and persistent errors at the industry level while the capital market operated to equalize returns at the margin to different types of investment. Third, it was maintained that British productivity growth could not have been any higher. This, in effect, rules out the possibility that the United Kingdom could have anticipated the American move to faster technological change reflected in the TFP growth estimates in Table 3.3.
This assessment has, of course, proved highly controversial and allegations that a number of serious failures, either market failures or government failures, inhibited economic growth have continued. One of the most celebrated of these claims has been ‘entrepreneurial failure’ (Landes, 1998). Another well-known hypothesis is that the capital market unduly favoured foreign investment and exhibited institutional failures that undermined the flotation of new business and slowed down structural change (Kennedy, 1987). Yet another criticism is that Britain failed to develop the type of national innovation system which would be crucial to twentieth-century economic growth (Goldin and Katz, 2008). Finally, over-reliance on ‘self-regulating’ markets and a regrettable lack of state intervention was the charge levelled by Elbaum and Lazonick (1986). A notable version of this last argument was made by Richardson (1965), who suggested that tariff protection of new industries was required to prevent an over-reliance on old industries with limited potential for productivity growth.
Some parts of McCloskey’s defence of British growth performance stand up well to scrutiny. As Broadberry says, ‘In most industries competitive forces acted as a spur to efficiency, with existing rivals or new entrants ready to take up opportunities neglected by incumbent producers’ (1997, p. 157). It is a staple of the literature that the only well-established failure to adopt cost-effective new technology, namely, not to switch from the Leblanc to the Solvay process in soda manufacture, was in a cartelized activity, and this is seen as underlining the point that competition was an antidote to entrepreneurial failure (Magee, 2004). In the most-studied choice of technology, that between ring spinning and mule spinning in cotton textiles, the evidence seems clear-cut that the British industry was rational to stick with mule spinning for the vast majority of its production (Ciliberto, 2010; Leunig, 2001). American technology was often not appropriate or lacked technological congruence in British conditions where relative factor prices and market size were different, as Henry Ford found out when trying to use mass-production methods to make cars in Britain.
Although the new economic history has largely succeeded in rejecting claims of managerial failure in the pre-1914 British economy, it is important to recognize that complete exoneration would be going too far. For example, railways was a major sector whose performance was clearly inadequate; Crafts et al. (2008) quantified the excess of actual over minimum feasible costs for a sample of fourteen major railway companies and concluded that median cost inefficiency was 10.2 per cent in 1900, equivalen
t to about 1 per cent of GDP. Two salient features of the railway sector were that competition was weak and so were shareholders in companies that were notable for the separation of ownership and control. The key point is that this entailed significant principal–agent problems as railway managers had considerable scope to pursue their own objectives and to fail to minimize costs at least while profits remained ‘acceptable’ (Cain, 1988).5
This example should not be taken as typical of the pre-1914 economy; on the contrary, railways were something of an outlier in terms of both barriers to entry and the degree of separation of ownership and control (Cheffins, 2008). However, while railways were the exception in 1900, cases of weak competition together with weak shareholders would become all too common after 1950. So, we may see railways as a harbinger of problems that would impair British economic performance in the decades of acute relative economic decline during the long post-war boom.
It has also become clear that there was no major capital market failure. Foreign investment accounted for about a third of all British savings, but this was justified by the returns available and the diversification of risk that was achieved.6 British investors would not have been well served by switching out of foreign assets and into new domestic industries (Chabot and Kurz, 2010; Edelstein, 1982). The banking system was not markedly inferior to those of Germany or the United States despite oft-repeated claims to that effect. The British financial system was specialized and the clearing banks provided valuable support to industry through financing working capital while longer term finance could be obtained through corporate bonds and equities and, of course, through retained earnings (Chambers, 2014). Universal banks in Germany did not make a significant difference to the performance of firms with which they had close relationships or supply major amounts of finance for industrialization (Edwards and Ogilvie, 1996; Fohlin, 2012). J. P. Morgan in the United States added value for investors by improving company management and achieving market power through mergers (de Long, 1991) but can hardly be seen as the key underpinning of the acceleration in American TFP growth.