The microeconomic analysis of the household and the labour market, 1880-1939
Symposium organized by Paul Johnson from April 23 to 29, 1997
Baines Dudley (London School of Economics), Burton Valérie (Memorial University of Newfoundland St John’s), Camps Enriqueta (University of Pompeu Fabra, Barcelona), Eeckhout Patricia (van den) (Universiteit Brussel), Engelen Theo (University of Nijmegen, the Netherlands), Federico Giovanni (University of Pisа), Green Alan (Queens University, Kingston, Ontari), Guinnane Timothy (Yale University, USA), Hatton Tim (University of Essex), Horrell Sarah (University of Cambridge), James John (University of Virginia), Johnson Paul (London School of Economics), Kiep Paul (Vakgroep Geschidenis, Nijmegen, the Netherlands), MacKinnon Mary (McGill University, Montreal), Morris Susannah (University of Oxford, UK), Oxley Deborah (University of New South Wales, Sydney), Scholliers Peter (Universiteit Brussel), Sundström William, (Santa Clara University, USA), Thomas Mark (University of Virginia), Treble John (University of Wales, Bangor), Wardley Peter (University of the West of England, Bristol), Wood Jenny (Hull University, Great Britain).
Economists, sociologies and policy-makers make frequent reference to historical experience in order to justify thеir current theories aid actions. Sometimes the past is presented in a positive light – praise for the ‘Victorian values’ of independence and thrift, or the social solidarity of industrial working-сlass communities – and sometimes it is disparaged for its harshness, poverty, incivility and aggressive individualism. In either case, modern interpretations of the behaviour of people in the past owe more to historical imagination than to empirical investigation. Very little detailed research has been undertaken into the historical behaviour of individuals and households, and how they reacted to the opportunities and threats of economic aid social circumstances in the past. In consequence, our beliefs about past behaviour are derived primarily from exceptional sources such as personal diaries and autobiographies, from contemporary social commentaries, and from the analysis of aggregate time-series data sets.
This colloquium, organized by Paul Johnson (London School of Economics, LSE) and Tir Guinnane (Yale) provided a forum for comparison of the results of dеtailed empirical analyses of individual and household behaviour in the past. The source material is data on the economic and demographic circumstances of many tens of thousands of households which was collected from the late nineteenth century by the first generation of social scientists. The technique of small-scale intensive household surveys was established in Belgium by Henri Le Play, and subsequently developed and applied to large scale surveys most notably by Charles Booth in England and Carroll Wright in the United States.
From the 1880’s detailed household surveys and censuses were conducted in all the industrial economies to investigate issues of employment, income, expenditure, housing conditions, fertility, education and migration. Despite the great investment of time in the collection of these data about individual and household circumstances, the analysis undertaken by the original investigators was cursory. They were overwhelmed by the volume and complexity of their datasets, and frequently did little more than sum the totals of the main characteristics and calculate averages.
In many cases the information collected was similar in detail and richness to that gathered in modern labour force aid household surveys. These early survey data provide insights into the interaction of household structure and labour market activity in past societies similar to those derived by government departments aid social scientists today. Recent advances in micro-computing have enabled many of these comprehensive historical datasets to be put into machine readable form, allowing economic historians directly to investigate the micro-economic behaviour of households in the past.
The papers presented at this colloquium investigated both the period around the end of the nineteenth century aid the years of the inter-war economic depression, using data from Belgium, the Netherlands, Germany, England, Canada and the USA. Ail twelve papers shared a common approach of using microdata sets to address questions which could not be аdе- quately answered using aggregate data. These questions all fell within three areas of inquiry concerning : the extent and distribution of resources both within and between households; reproductive behaviour and household circumstances; employment in the labour market and in the domestic economy.
Distribution of resoиrces
Sara Horrell (Cambridge) and Debbie Oxley (University of New South Wales) investigated the distribution of resources and the adequacy of male earnings to support family dependants using British household budgets taken from the US Bureau of Labor survey of 1890/1. Overall, husbands earned 75 % of household income but this percentage varied according to the industry in which the head of household was еmрlоуed and household characteristics such as the number and ages of children. In manу cases men’s earnings were inadequate as a sole source of family income and would have placed households below contemporary poverty lines whist in most cases these earnings were insufficient to maintain proper nutritional standards. Horrell and Oxley presented a picture of dynamic adaptation to circumstances within the household with income from lodgers, female earnings and in particular children’s employment providing important and varying additional sources of income according to need. In industries where there were few opportunities for local employment, an alternative strategy was for older children to leave home thereby facilitating a better match of resources to income within the household. They conclude that far from being a rational and utility maxi- mising choice brought about by higher earnings for men, the reliance upon a sole male earner was often the result of a lack of job opportunities for other household members which carried a welfare cost for the rest of the household. The interface between the male worker and the household was explored further in a paper by Valerie Burton (Memorial University of Newfoundland). Using data taken from the Board of Trade’s crew agreements for merchant seafarers, coupled with nominal linkage to household census enumerations, Burton examined the pattern of employment of merchant seafarers in late nineteenth century English ports, thеir on-shore household circumstances aid their ability to contribute resources to these households. Burton found that the majority of seafarers were married household heads, whilst a sizeable majority were sons of the household in which they were enumerated. Whilst these onshore household bases were important for employment opportunities due to the local nature of the labour market for seafarers, these men did not provide the sole source of income within their households in part due to the periods they spent away at sea when wages could not be sent home.
Acting as discussant Peter Wardley (University of the West of England) suggested further information about the strategies ofseafaring households could be gained by using the records of other local institutIons such as voluntary sodeties, poor law authorities etc. to sec how and when members of seafaring household’s may have sought extra-familial help through periods of hardship. In the discussion of Horrell and Oxley’s paper Wardley warned about the dangers of inferring region from industry in the Bureau of Labor survey by suggesting that those households with differently occupied household heads necessarily came from different localities with different local labour markets.
The microdata itself аgain came to the fore with the two papers by Тim laiton and Roy Bailey (University of Essex) and Dudley Вaines and Paul Johnson (LSE) which both used the newly computerised New Survey of London Life and Labour (NSLLL). Much was made of the strengths of their source against several comparators : over earnings surveys – as a poverty survey the NSLLL includes those with zero earnings; over censuses for Britain – it includes income and expenditure date; and finallу over other poverty surveys due to its scale and sampling frame.
Hatton and Bailey used this survey of the living standards of 28,000 metropolitan households to examine, firstlу, how the use of different poverty fines affects the number of households found to be in poverty and secondly the effects of the inter-war social security system in relieving poverty. Thеу also considered how these effects compared with the post-Beveridge social security system if it had been applied in the inter-war period. They demonstrated that the inter-war social security system played a much greater role in alleviating poverty than previously thought but that its impact was many times greater for older households. Whilst the introduction of 1950’s system would have reduced the total number of households in poverty by a further 50 %, 17 % of household’s would still have been in poverty in interwar London.
Вaines and Johnson examined whether the withdrawal of older workers from the labour force was the result of the voluntary choice of retirement or enforced consignment to the ‘industrial scrap-heap’. They found that whilst employed male household heads aged over 60 were producing 58 % of household income, when non-earning heads over 60 are included this figure reduces to 39 % and for aIl males the figure is 36 %. They concluded on the basis of this and other findings that the consideration of only household heads exaggerates the income of older men and hence their ability to finance retirement. Those older men with reasonable incomes who could have chosen to retire were those who remained еmрlоуed thereby demonstrating little evidence to substantiate the hypothesis of voluntary retirement. Bill Sundström (Santa Clara University) questioned the ability of Вaines and Johnson to identify retirement in terms of permanent withdrawal from the labour force as opposed to temporary involuntary unemployment.
Fertility and the household
The second major area of enquiry concerned fertility and the household. Tim Guinnane (Yale) and John Brown (Clark University) used district and individual level data to examine the demоgraрhiс, social and economic correlates of the fertility transition in Bavaria between 1850 and 1914. By using more social and economic data in a less aggregated form than the Princeton European Fertility Project, Guinnane and Brown challenged the previously accepted timing and causes of the fertility transition. They did not find evidence to support an innovation-diffusion interpretation of the fertility transition but rather point to the importance of urbanisation and the structure of the rural economy in influencing changes in fertility. Being catholic, belonging to a lower occupational class or being bonn in a rural area, even if followed by migration to an urban area, were all found to positively influence fertility.
Unlike Guinnane and Brown, Theo Engelen and Paul Klep (University of Nijmegen) started from the premise of an overt- 1у economic explanation of fertility transition. In their paper examining fertility in varions agricukural communities in the Netherlands they used fertility decline as a direct proxy for the declining economic utility of children. They argued that differential fertility rates between areas could be explained by the extent to which parents could exercise control over their children thereby realising actual benefits from higher levels of fertility, rather than being the result of simple changes in economic circumstances and the potentiel changes in the costs and benefits of children which these would create.
Enriqueta Camps (University of Pornpeu Fabra) was discussant on both fertility papers. Camps challenged Engelen and KIep’s use of changes in fertility as a measure of changes in the economic utility of children arguing that their model implausiblу suggested parents were capable of making long run predictions about the economic circumstances which would be facing their off-spring and hence their ability to control their resources within the household.
Sara Horrell suggested that the authors should analyse the actual costs and benefits of children, alternative labour sources in the household such as servants and the nature of local labour contracts before they could conclude that the economic utility of children declined.
Methodological considerations about the use of micro data were the substantive topics of two of the papers given at the colloquium. In a paper given by Jenny Wood (University of Hull), arising out of some joint work undertaken with Geurt Collenteur (University of Groningen), detailed consideration was given to the dеfinition of the « household » and its nature as an economic unit with the inclusion or exclusion of family members, servants and lodgers. In their research, Collenteur and Wood compared household structures in five villages in Groningen province in the Netherlands finding differences in the number and ages of children present in different areas : a finding consistent with the different levels of fertility in various communities as found by Engelen and Klep.
Patricia van den Eeckhout and Peter Scholliers (Free University, Brussels) presented a paper which went to the heart of the methodological problems of historians’ practice of using microdata collected for one purpose in the past to investigate issues of concern to us today. By examining a number of surveys of women’s paid labour force participation in Belgium in the nineteenth century they argue that the findings may tell us more about the underlying ideology of the survey takers than they do about any empirical ‘reality’. Van den Eeckhout and Scholliers suggested that historians should use these valuable microdata sources to illuminate the social and political circumstances in which they were constructed rather than just undertaking a ‘traditional’ critique of the survey with regards to its sampling frame before rushing head long for the ‘facts’. Camps, as discussant on this paper, asked whether the large number of surveys which аll demonstrated a relationship between women’s participation and household income could all be suffering from a consistent bias. In a lively and full debate much consideration was given to whether these ‘biases’ could be redressed with the use of statistical techniques.
The next group of papers all analysed questions concerning labour markets using microdata. Alan Green (Queen’s University) and Mary Mackinnon (МcGill University) used the 1901 census manuscripts for Toronto to consider how well British born migrants assimilated into the labour market and whether they fared better than migrants from continental Europe. They did not find any assimilation effects for adult migrants although for young migrants their eamings profiles approximated these of the native born. Unlike in the work of Hatton (who along with Horrell were the two most cited participants of the colloquium) Green and Mackinnin concluded that as long as migrants were white and literate their earnings profiles were not affected by where outside North America they carne from. It was illiteracy not ethnicity or religion, which explained the lower eamings of some migrants.
Giovanni Federico (University of Pisa), who acted as discussant on all the labour market papers, suggested that Green and MacKinnon should reframe their issue to consider all non-native born workers in comparison with Canadians as, unlike in the United States, very few migrants to Canada were not British. Mark Thomas (University of Virginia) suggested that migrants were more likеlу to move again than those who were native born and therefore those migrants who had fully assimilated may have moved on and were not being picked up in the census.
The position of particular groups within the labour market was also the issue for Вill Sundström (Santa Clara University). Sundström tried to explain why during the 1930’s in the US the labour force participation of black married women declined whilst for white women it increased. Using microdata drawn from the 1940 census he tested two hypotheses : flrstly the discouraged worker effect and secondly, the relief program incentive effect. Sundström argued that it was the discouraged worker effect which could explain the divergent changes in participation experienced by black and white married women during the depression with unemployment rates in local urban markets having a dramatic effect on black women’s participation probabilities in 1940.
Tim Hatton suggested that regional variation in industrial structure should be included within Sundström’s analysis. Susannah Morris (Oxford) considered that it was also necessary to include the relative proportion of black and white workers within that industrial structure as black and white workers mау have been fundamentally different in terms of both their skill levels and the industries in which they work thereby facing very different local unemployment rates to their white counterparts. It was generally thought that wage levels should also be included in the analysis.
John James and Mark Thomas (University of Virginia) considered another issue relating to the US labor market, in this caseasking whether US manufacturing firms practised labour hoarding at the end of the nineteenth century and if so which types of workers were kept on and why? The data used are taken from local labour surveys compiled by various State Bureau of Labor Statistics. Analysing the data in both cross sectional form and as a pooled time series they conclude that there was labour hoarding : if output changed 1 % employrnent changed by 0.5 %. This hoarding was concentrated in the higher end of the wage distribution. Federico considered the possible causes of labour hoarding as either being due to a need for firm specific skills or because firms sought to minimise the transaction costs of hiring and firing. Hе suggested that both of these factors would have changed over time.
John Тreble (University of Wales, Bangor) then turned attention away from the US to consider the productivity of British coal mines using microdata from County Durham in the 1890’s. Тhis data set known as the Victorien Household Panel Study has been compiled by Тreble using wages books, census material, and a large amount of other company specific information and provides him with dеtailеd data about worker productivity and earnings under different wage rate and employment regimes. Hе conduded that there were significant positive wage effects on productivity but that these wage еffects were not closely associated with time series variations in productivity which were related to simultaneous changes in manning levels and deployment of the workforce. Federico suggested that Тreble could more fully exploit the data that he had and analyse worker effort in relation to workers’ household characteristics : a task which we were informed Тreble had already done, but, elsewhere.
All participants at the conference were grateful on two interrelated fronts, firstly intellectually and secondly physically (although not necessarily in that order of priority) . Intellectual thanks were given for both the invention and advancement of the computer hardware and software that enabled the detailed use of microdata to address historical questions and for the fast that these valuable datasets had survived. Much concern, however, was expressed at the new policies under consideration by many national record offices which may threaten the preservation of such data in the future and all historians were urged to act to inform archivists and civil servants of the value of microdata. Secondly, participants were overwhelmed by the sumptuous beauty of their surroundings proffered by the ever hospitable Fondation, which provided a small taste of the ‘good life’ that cannot be sustained on academic salaries!
Dudley Gaines – Labour force participation and the economic well-being of older men in London, 1929-1931
Volerie Burton – Seafaring households and the market for seafaring labour in English ports in the late nineteenth century
Patricia van den Eeckhout and Peter Scholliers – The construction of women’s paid labour – The Belgium inquiry of 1891 into working class budgets
Theo Engelen – Declining economic utility of children in varions agricultural worlds in the Netherlands, 1889-1960
Alan Green and Mary Mackinnon – Were the British better? A comparison of immigrant and nativeborn male employees in Toronto, 1901
Тimоthу Guinnane – The fertility transition in Bavaria, 1850-1914 : a view from aggregate data
Tim Hatton – Poverty and the welfare state in inter-war London
Sara Horrell and Deborah Oxley- Breadwinning, poverty and ressource allocation in late nineteenth century Britain
John James and Mark Thomas – Labor hoarding, skills and productivity during depressions : microeconomic evidence from the US in the 1890’s
Paul Johnson – Labour force participation and economic well-being of older men in London, 1929-1931
William Sundström – The decline of labour force participation among married black women, 1930-1940
John Treble – Productivity in British coalmining : evidence from microdata, county Durham, 1890-1893
Jenny Wood – Household aid family structure in а changing economy in the province of Groningen, 1870-1910