Historical studies of economic development often provide researchers with long time series of data with which to test their models and hypotheses, but they also often pose particular problems of missing, scarce or unrepresentative data. This paucity of evidence increases the potential value of each new observation and makes economic historians conscious of the need to use information drawn from all levels of economic activity, including individuals and families, regions and the nation itself. There are, however, obvious problems in trying to use cross-section data drawn from a single point in time to answer dynamic questions, and likewise in attempting to analyse the complexity of economic development with often schematic aggregate time-series data. Much of the discussion at the 1986 Anglo- French Colloquium, devoted to quantitative economic history, centred on two issues. First, which time period provided the most appropriate focus for researchers attempting comprehend the pattern of economic growth and change, and second, was the most revealing unit of analysis the nation, the region or the household?

This, the eighth in the series of Anglo-French Colloquia originating under the auspices of the Economic and Social Research Council and the Maison des Sciences de l'Homme, allowed the participants both to discuss the details of their current research and to consider some of the broader problems of applied economic history. The Colloquium was sponsored jointly by CEPR and the Ecole des Hautes Etudes en Sciences Sociales, Paris, with financial support from the ESRC and the Centre National de la Recherche Scientifique (CNRS). It was held on 22-23 September and was organized by Francois Bourguignon of the EHESS, Roderick Floud, Co-Director of the Centre's programme in Human Resources since 1900, and CEPR Director Richard Portes.

The advantages of a macroeconomic approach to applied economic history were illustrated by the work of Maurice Levy-Leboyer (Ecole Normale Superieure, Paris) and Francois Bourguignon (EHESS) in building 'An Econometric Model of France During the Nineteenth Century'. The absence of consistent sets of fiscal data has hitherto prevented the construction of reliable national income estimates for nineteenth century France. Levy-Leboyer and Bourguignon therefore turned to more detailed data on agriculture, industrial production, and international trade to model these three sectors separately. The sectoral models indicated that agricultural prices were the most important influence on the early nineteenth century economy, but that after 1870 the dominant forces were competition in export markets from UK and German goods, and crowding-out of new capital formation by the state. Their macroeconomic model also suggested that after 1860 France experienced a rise in the marginal propensity to consume food, which hampered industrial development by switching domestic demand towards the agricultural sector.
A number of participants were puzzled by this rise in the elasticity of demand for foodstuffs between 1860 and 1880. Levy- Leboyer suggested that this finding was corroborated by 'microstudies' of food consumption in Paris, though he also noted that there were major regional differences in patterns of demand for food that were not captured by their model.

Tony Wrigley (LSE) also argued that agriculture was a dominant influence on the course of economic growth in the early stages of industrialization. In his paper, 'Population Growth and Economic Change in Early Modern England', Wrigley noted that the UK population grew by more than 250% between 1541 and 1871. This increase in the number of mouths to feed did not lead to famine, nor did the increase in labour supply lead to any secular decline in real wages, despite the fact that the absolute size of the workforce employed in agriculture remained constant. The output/labour ratio in agriculture therefore roughly doubled over the period. In the early nineteenth century, Wrigley concluded, Britain had a higher per capita income than other European countries not because of its earlier industrialization, but because of higher labour productivity in agriculture.

Levy-Leboyer, in commenting on Wrigley's paper, suggested that larger harvests and the enclosure of open fields must have required an increase in labour inputs. Wrigley agreed, but noted that labour input could have risen without an increase in the numbers employed in agriculture. Labour was employed more effectively during winter months on land improvement and maintenance, and there was an important substitution of animal power for human labour. Solomos Solomou (Peterhouse College, Cambridge) questioned the length of the period studied by Wrigley, arguing that the long-run rise in agricultural production over three centuries concealed periods, sometimes several decades in length, when agricultural productivity was stagnant or population growth was minimal. Wrigley conceded that such short-run variations helped illuminate relationships such as those between grain prices and human fertility, but suggested that they had little effect on the process of economic growth and development in the long run.

The question of the most appropriate time period for the analysis of macrodata was taken up by Solomos Solomou in his own paper, on 'Kondratieff Long Waves in Economic Growth, 1850-1913'. In 1925, Kondratieff suggested that economic activity was characterized by cyclical fluctuations of about 50 years duration, but this contention has proved controversial. Solomou compared output data for the United Kingdom, France, Germany, and the United States between 1850 and 1913, but found little evidence of Kondratieff waves at either national or world level. His analysis did indicate a number of irregular shocks of different durations which affected individual economies, and the interactions of these national shocks often produced shorter-run cycles. Solomou concluded that economic growth is best understood by examining the specific historical factors that underpin each national cycle, rather than by attempting to fit a long-run model with a rigid periodicity to data from a variety of countries.
Some participants welcomed this conclusion, agreeing that the value of historical cross-country comparisons lies in the distinctive nature of each nation's development path. Others thought that broad development models could still be useful to explain such phenomena as long swings in the pattern of international migration.

Dudley Baines (LSE) disagreed, citing his own 'Recent Quantitative Research on British Emigration and Internal Migration in the Nineteenth Century'. Baines had used census data and vital statistics to estimate internal and external migration from each county in Britain between 1861 and 1900. There was, he found, considerable variation in emigration rates across counties. This was not, however, correlated with the degree of urbanization, real incomes or the proportion of the population working in agriculture, factors which have been suggested by those seeking to explain long swings in transatlantic flows of labour and capital. In the discussion which followed his presentation, Baines noted that the absence of comprehensive time-series data prevented a detailed assessment of the relative importance of 'push' and 'pull' factors in emigration flows, but he had found that information from previous waves of emigrants seemed to be one of the most important factors influencing emigration rates.

Most economic historians regard the interwar years as an important and reasonably self-contained period for analysis. Several papers focused on the macroeconomics of this period and its lessons for more recent experience. Jacques Mistral (Ecole Nationale de la Statistique et de l'Administration Economique (ENSAE)) presented a paper entitled 'Trois Devaluations du Franc: Un Repli Strategique?', in which he investigated the background to and economic impacts of the franc devaluations of 1924-6, 1957-8 and 1982-3. Mistral argued that each of these parity changes formed part of a broader policy for the franc, designed to avoid abrupt, unplanned realignments like those of sterling in 1931, 1948 and 1967. French policy has been more forward-looking, Mistral argued, and the improved competitiveness that resulted from each devaluation was exploited effectively through a mixture of complementary public initiatives and market responses.

In 'The Anatomy of Financial Crises: The 1930s and the 1980s', Richard Portes (CEPR and Birkbeck College, London) outlined the results of joint research with Barry Eichengreen on the international debt crisis of the 1930s, reported in CEPR Discussion Paper Nos. 75 and 130. Portes pointed out that the many apparent similarities between international and domestic economic problems in the 1930s and 1980s have made the interwar experience of growing interest. But such comparisons are not always appropriate, he noted. Although the debt-servicing problems of today resemble those of the 1930s, institutional arrangements differed significantly in the two periods. These changes and more astute economic policies have altered the linkages among exchange market disturbances, debt defaults and bank failures so as to diminish the likelihood of financial collapse in the 1980s.

Pierre Villa (Centre d'Etudes Prospectives d'Economie Mathematique Appliquees a la Planification, (CEPREMAP)) reported his work on 'The Wage-Price Spiral in Interwar and Postwar France'. The French economy, Villa argued, had also been characterized by major institutional changes between the interwar and postwar periods. There had been a shift from a more competitive regime of flexible prices towards an oligopolistic regime in which wage costs are more rigid. These changes were fostered by the development of 'implicit' wage contracts and by the social legislation of postwar governments, which increased the costs of hiring and firing. Villa attempted various econometric tests of the wage-price relationship for different periods, but the results were not always robust. Confirmation of these results, he argued, required the incorporation of his wage-price hypothesis into a more comprehensive econometric model of the French economy.

Andre Straus (CNRS) highlighted two severe problems involved in building such a model of the interwar French economy. The first was the choice of an appropriate theoretical framework; the second was the availability of appropriate economic indicators. For Straus, a key question was whether the interwar French economy could be characterized as a disequilibrium economy in which prices were inflexible, or 'sticky', and in which markets adjusted through changes in quantities transacted, not through prices. Data on different markets were seldom collected and analysed in a uniform fashion, often preventing effective testing of competing hypotheses. Straus suggested that macroeconometric models did not provide enough sectoral and spatial disaggregation of market data. As a result, they gave a misleading picture of homogeneous national markets, which biased any assessment of the degree of disequilibrium in the economy.

Levy-Leboyer agreed that macromodels which ignore the heterogeneity of different sectors and regions may produce biased results, but argued that a simple model capable of elaboration was a more fruitful starting point than a very complex model arbitrarily constrained because of deficient or non-existent data.

Sectoral and regional diversity was not seen as a problem by all the participants; some viewed it instead as an opportunity to increase our understanding of the course of national economic and social development. Herve Lebras (Institut d'Etudes Demographiques), in 'Deux Siecles de Disparites Regionales en France', argued that much historical analysis of French development over the last 200 years had erred in ignoring the extensive regional variation in many social, political and economic indicators. To counter this overemphasis on aggregate data, Lebras had gathered a large body of demographic, political, cultural and economic data on each departement in France in an attempt to identify the foundations of regional variation. Presenting the data in map form clearly revealed differences between the North, South-East and South-West of France in religious and political affiliation, in patterns of family organization and inheritance, in employment status and in migration. Cartographic representation also illustrated how these regional differences varied over the nineteenth and twentieth centuries. In response to questions, Lebras argued that maps of descriptive statistics offered better insights into the spatial nature of French diversity than did multivariate statistical analysis.

Jean Heffer (EHESS) and Jacques Mairesse (EHESS and ENSAE) used even more finely disaggregated data in their paper written with Jean-Marie Chanut on 'La Culture du Ble en France au Milieu du 19eme Siecle: Rendements, Couts, Prix'. The authors used data on wheat production in 1852 from a cross-section of over 360 arrondissements to estimate yields, prices and wages throughout seven regions of France. The detailed nature of the production data revealed that the sizeable inter- and intraregional variations in yields and production costs were not matched by variations in the price of wheat. Whereas the market for wheat in mid-nineteenth century France appeared to be strongly integrated at both national and regional level, the market for labour and other factors of production was still very localized. Heffer and Mairesse suggested that the disaggregated analysis of this single, but very important, agricultural product could give more insight into the modernization and integration of the French economy than could often be gained from macroeconomic analysis, and they hoped to be able to replicate their study using cross-section data for later years.

Cormac O'Grada (University College Dublin and CEPR) pursued the theme of diversity in agricultural practice and performance in his paper, 'Arthur Young's Agricultural Statistics: A Reassessment'. Young had surveyed agriculture in France, Ireland and England during the late eighteenth century, and had identified high rents and the introduction of new crops as indicators of successful farming. His writings convinced contemporaries that land improvement, particularly enclosure, was the key to the advances in English agriculture and to the relative backwardness of the peasant farming of Ireland and Northern France. During the two centuries which followed Young's work, interpretations of European agricultural development were based on his conclusions about the efficiency of English capitalist agriculture and the backwardness of other national types of agricultural organization. O'Grada used Young's own data on yields to show that, when adjusted for variation in soil type, grain yields in England were very similar to those in Ireland and France. This suggests, according to O'Grada, that careful regional and cross-country comparisons show historical experience to have been much more complex than Young's national stereotypes would suggest.

Just as regional and sectoral disaggregation can complement the analysis of economic change at national level, so the study of individual or household behaviour can yield greater understanding of economic relationships. Paul Johnson (LSE and CEPR) and Leslie Hannah (LSE and CEPR) both examined the economic impact of the ways in which individuals make financial provision for their old age. The 'life-cycle' hypothesis suggests that individuals borrow against their future income when young and run down their savings when old, in order to ensure a more even level of consumption over their lifetime. In 'Did Workers Save for Old Age in Victorian and Edwardian Britain?', Johnson investigated the extent to which workers did adopt this approach. He found that despite the existence of appropriate financial institutions, the majority of workers failed to accumulate sufficient funds to avoid pauperism in old age. He suggested that this lack of thrift was not due to any moral failing on the part of the Victorian working class, as was claimed by many contemporary observers, but rather to the severe income and liquidity constraints that workers faced. Rising incomes in the twentieth century and the provision of some economic security in old age through the state pension scheme had gradually induced people to engage in more deliberate 'life-cycle' saving.

Hannah explored the role of occupational pensions in his paper, 'Occupational Pension Funds: Getting the Long-run Answers Right', also available as Discussion Paper No. 99 and discussed at a lunchtime meeting reported in CEPR Bulletin No. 13. The elderly in the United Kingdom now receive less direct financial support from the state than in any other industrialized country except Japan. The United Kingdom has instead a much larger occupational pension system, which receives very generous tax treatment. Hannah argued that this pattern of pension provision almost certainly had an adverse impact on labour mobility. In the discussion which followed his paper, Hannah emphasized that a society's arrangements for financial provision in old age may well have a far greater long-run impact on economic growth and welfare than actuArial,Helvetica,Sans-Serif considerations may suggest.

One of the themes of the Colloquium was our understanding of the process of economic growth and the welfare changes that result from it. It was particularly appropriate that the measurement of individual or household welfare was considered directly in two presentations. Jay Winter (Pembroke College, Cambridge, and CEPR), in his paper 'Some Paradoxes of the Great War', examined the effects of the 1914-18 war on the welfare of the civilian populations of Britain, France and Germany. Data on incomes or real wages are an unreliable guide to welfare levels in this period, Winter argued, because of rationing, controlled markets and high inflation. He chose instead to use mortality data to see whether the war affected the prewar trend towards increased life expectancy. Despite the enormous resource cost of the war effort, life expectancy improved in both Britain and France, but not in Germany. Winter argued that in Britain and France, effective political control of the wartime economy allowed a general 'levelling-up' of economic standards. This prevented the emergence of an 'entitlement' failure, in which individuals lose their ability to secure a minimum level of resources.

The idea of using demographic data to indicate a population's welfare and the level of resources available to it was also adopted by Roderick Floud (Birkbeck College, London, and CEPR), in his work on 'Measuring European Inequality: The Use of Height Data'. The problems of constructing consistent measures of per capita real income, appropriately adjusted for environmental, social and cultural differences are well known. They can be circumvented by looking at the mean height of the population and the variance around this mean, since height is the best composite measure of net nutritional status and hence a good measure of welfare.

Floud produced evidence from a number of European countries since 1860 which showed considerable and systematic variation in height across socioeconomic classes, across regions and over time. Most of the increase in average height has occurred this century, and nineteenth century economic development had little effect on the bulk of the European populations. Indeed, in several cases regional differentials in height widened in the earlier part of economic development, and only converged again much later. This is consistent with the 'Kuznets curve', Floud observed, whereby economic development can lead initially to increased inequality.

The use by Winter and Floud of demographic data to answer welfare questions that are commonly analysed in strictly economic terms was a clear illustration of how the paucity of data has obliged economic historians to seek new analytical approaches. Even in papers where more conventional economic evidence was available, an awareness of the diversity of historical experience led to a very cautious interpretation of the models. No single form of data, no single model or method of analysis seemed sufficient to comprehend the dynamics of growth and change. The variety of approaches adopted by economic historians in their study of the past is a realistic reflection of the complexity of historical development.