LIVING STANDARDS IN BRITAIN 1900-2000: WOMEN'S CENTURY?
Two composite measures are calculated to map improvements in living standards over the 20th century: the Dasgupta- Weale index and the Human Development Index. A gendered version of the latter is also considered. Indicators of income, leisure, inequality, wealth, health, education and political rights are included. The indices reveal a century of progress. But progress has been neither continuous nor uniformly shared. Downturns are evident in some of the indicators since 1980, demonstrating that the gains are not immutable and need to be protected. Women's position has improved if the end of the century is compared to its beginning, but there has been little change in women's position relative to men's over the last few decades on the dimensions considered here.
Few people would dispute that the standard of living of the average Briton has improved over the past hundred years. National income per capita has grown from [pounds]44.65 in 1900 to [pounds]252.58 by 1998, in constant prices. Boys born today can expect to reach the age of 75, compared with a life expectancy of only 46 for their predecessors in 1900, maternal mortality has been reduced by a factor of 79, and a university education is no longer the privilege of a small minority. In addition, we have access to a wider range of goods, can purchase more exotic foods, regularly travel to foreign countries, and more of us own our homes. But progress has not been continuous throughout nor uniformly shared.
Two composite measures of living standards are calculated to map the improvements in welfare observed over the 20th century: the Dasgupta-Weale index, which ranks changes in various indicators of welfare, and the Human Development Index, which describes the distance to go to achieve some desirable level of welfare. The gendered version of this index examines differences in male and female attainment and share in general progress (United Nations, 1995). These measures can incorporate a wide range of factors argued to affect well-being and we begin by considering the indicators to be included.
Progress has most often been proxied by growth in real income per capita which captures increases in consumption and command over resources. But other factors which affect welfare and which can be related to the national income approach have received attention, for instance, leisure and inequality (Nordhaus and Tobin, 1972; Beckerman, 1980). Leisure is incorporated by taking explicit account of the hours worked to generate income and the income level can be adjusted for changes in its distribution. In addition, the certainty with which income is received may be important. The unemployment rate and coverage of unemployment insurance can be used as indicators of this certainty. Wealth also enhances access to resources and welfare. The level of real wealth per capita will capture this but will ignore the tendency for wealth to accrue disproportionately to a small minority. Owner occupation of housing, eligibility for state pensions and membership of building societies provide proxies for the distribution of weal th through the various sections of society.
These measures focus on the material aspects of wellbeing. The World Bank has highlighted other factors that affect people's enjoyment of and chances in life, such as health, longevity and education, which has inspired the construction of the Human Development Index (MDI). Based on Sen's (1973) notion of capabilities and opportunities, this measure seeks to incorporate other aspects which contribute to progress in the lives people lead but will not always result from or be highly correlated with increased income. Thus Human Development Reports examine the progress of countries towards a developed state on three fronts: income, health and education. A similar approach is adopted here. Dramatic improvements in life expectancy, infant mortality rates, the incidence of diseases such as tuberculosis and smallpox as well as the proportion of deaths occurring through these diseases, and from maternal mortality and childhood diseases have been achieved in Britain over the 20th century. These variables are incorporate d into an indicator of health. Education too has shown significant improvements. Compulsory primary education was introduced under Forster's Education Act in 1870, but compulsory secondary education has been a feature of this century. Increasing numbers staying at school beyond the minimum leaving age has also fuelled increased ability to undertake further and higher education. Changes in the proportion of the relevant age groups in education are used as a measure of this progress.
However, capabilities and opportunities are not just enhanced by health and education. The ability to influence the determination of policy in one's country has been recognised as important. This aspect of welfare has been incorporated into comparisons of living standards by Dasgupta and Weale (1992) who consider political and civil rights alongside the previously mentioned indicators. The electorate was widened in 20th century Britain and the proportion of the adult population with the right to vote in elections measures this aspect of political rights. But some influence on one's destiny also may be afforded by direct involvement in organisations, for instance, trade unions can exercise influence on terms and conditions of employment. Thus membership of a trade union may afford greater determination of the day-today welfare experienced at work and this is included as a second measure.
Finally, accounts of 19th-century living standards have often emphasised the less tangible aspects of change that impacted on welfare. They highlight the disamenities associated with urbanisation; overcrowding in poor accommodation, insanitary conditions, increased crime and high levels of pollution (Williamson, 1981; Lindert, 1994; Mokyr 1988). These aspects of living conditions are considered in this 20th century overview. In addition, some indicators reflecting beneficial changes are incorporated. More travel and communication is assumed to broaden horizons and enable contact with others, and the increased use of energy in households taken to signify greater comfort and the use of machinery to mitigate the arduousness of some household tasks.
The aim here is to take an eclectic view of living standards and evaluate the evolution of the six components; income and leisure, wealth and income security, health, education, political and workplace rights, and living conditions; and the aggregate over the 20th century.
The Dasgupta-Weale index
The Dasgupta-Weale index (1992) was developed to allow comparisons across countries by using the Borda rule to aggregate component variables, but it has also been used to consider the temporal development of welfare (see Crafts, 1997a, 1997c). The data here have been collected for five-yearly intervals and the method gives a rank to the value for each part of the index in each year.  The year with the worst performance of an indicator is assigned the rank 1 and the value of the variable in the other years is ranked relative to this, with the best performance ranked 21. The ranks are summed across all elements in each year to give a final score on which the component can be ranked. For instance, the ranking of each year on the income component will be based on the sum of the ranks of the income, leisure and inequality elements which make up this component. Obtaining an overall index of welfare requires aggregating the ranks for the six components and ranking this final score. Thus the progress of the aggreg ate index is calibrated by position relative to other years. The Dasgupta-Weale index has the advantage that ranking makes it easier to assign a plausible value when observations are unavailable. Also it can easily incorporate a heterodox collection of indicators of the determinants of well-being.
Before considering the movement of the components and their contribution to welfare, the background to some of the individual elements is discussed. The income component is derived from a composite measure of real national income per capita, the distribution of income as measured by the Gini coefficient in each year, and hours worked in the labour market per capita to reflect changes in amount of leisure enjoyed. A simple index is computed: the three elements are ranked individually then combined in an average where each is weighted equally. A more comprehensive treatment of the trade-off between income, leisure and inequality is computed for the construction of the HDI in the next section.
Real income has grown markedly over the century, with particularly rapid improvements observable after World War II (Chart 1). This was accompanied by decreasing inequality for the first three quarters of the century. However, the dispersion of income widened between 1975 and 1990, reaching levels last seen before the war, and has shown little tendency to narrow since. Adjusting income gains for the associated level of inequality enhances the earlier gains and reduces later ones, indicating that progress has not been as continuous as increases in income alone would imply.
The Gini coefficient will register a reduction of inequality wherever in the distribution redistribution from richer to poorer occurs, and so provides an unsatisfactory measure if eradication of poverty is the prime concern. It would have been desirable to include the proportion of people in poverty as an additional element in the income component, but the paucity of surveys and different bases on which poverty has been defined render this an unreliable measure. But some comment on the extent of poverty over the century can be made. Rowntree's (1901, 1951) surveys of York indicate high levels of poverty in 1899 and during the inter-war period - some 30 per cent. This was significantly reduced to only 3 per cent of the survey population living below an absolute poverty line by 1951. However, subsequent studies have demonstrated that poverty has not been eradicated in Britain. Large proportions of pensioners and children of the long-term unemployed, low paid and single parents continue to suffer relative povert y. Indeed 7 million people received means-tested benefits in 1985 and 16.3 million people were estimated as living in households on or just above the poverty line in 1983; that is, nearly one in three of the UK's population (Johnson, 1994, pp. 301 and 316). Statistics like these indicate that the benefits of a century of increasing incomes have left the relative position for many unchanged.
Working time per worker has declined from 2834 hours per annum in 1900 to 1604 hours in 1998.2 Decreases in work time per capita have been less rapid, in part because of women's increased labour force participation which has increased the proportion of the population in the labour force. The large reduction in hours in 1919 is clearly visible and there has been a trend decline since. Although gains in income have occurred alongside increases in leisure, cycles in the income component are discernible (Chart 2a and Table 1).
Wealth has grown nearly as fast as income over the century but downturns occurred during World War II, in the 1950s, and also in 1990-95, possibly as a consequence of declining house prices.  How evenly distributed were the benefits of wealth? The proxy measures used are intended to capture assets which may be owned by the middle and lower parts of the distribution. Owner-occupation of housing increased from 10 per cent in 1900 to 67 per cent by 1998. This largely occurred from the 1950s and was encouraged by tax relief on mortgage interest payments, the exemption of housing from capital gains tax and the Thatcher government's sale of council housing to tenants. Building society membership stood at i per cent of the population in 1900 but had risen to 77 per cent by 1995, before many of the demutualisations occurred. Pensions, too, may be considered an element of wealth. Although the old had been entitled to support in case of need from the days of the Elizabethan Poor Law, state provision and entitlement were extended under the Liberal government's reforms of 1906 to 1914. Non-contributory pensions of 5 shillings a week paid to the poor over the age of 70 were introduced in 1909. The age limit was reduced to 65 in 1928. Pensions were made universal for workers by the National Insurance Act of 1946. From just over 2 per cent of the population in receipt of state pensions in 1915, nearly 19 per cent were pensioners in 1998. These measures indicate that growing wealth, even if not more equally distributed, has been accompanied by greater ownership of assets by many.
Income security has been captured by two measures. The unemployment rate shows clear cycles with peaks in the first decade of the century, the 1920s and 1930s, the early 1980s and early 1990s, although a persistently high level of unemployment is evident from the 1970s onward. But the misery of unemployment may, to a limited extent, be offset by the increasing number of workers who have insurance against this eventuality. Want occasioned by lack of work was also provided for by the Poor Law but was available only in the event of demonstrable financial distress, and at the discretion of the Poor Law guardians. The National Insurance Acts of 1911 extended coverage to largely skilled workers who paid contributions and received benefits tied to these contributions in the event of unemployment. These provisions were promptly put to the test with the rising unemployment of the interwar years. The social reforms that led to the post-World War II creation of the welfare state incorporated the contributory system of N ational Insurance. It made payments compulsory for those in work and benefits an entitlement for a fixed maximum duration if an adverse contingency arose, rather than a means-tested discretionary payment. The proportion of the working population with coverage has risen from 34 per cent in 1920 to 60 per cent by 1975. Aggregation of these six elements into a wealth and income security ranking shows a general upward trend with cycles mainly generated by the unemployment rate (Chart 2a).
A more significant break with the past occurred with the creation of the National Health Service. A contributory health insurance scheme first paid illness, disability and maternity benefits in 1913. This scheme covered most employees and entitled insured workers to out-patient care but did not provide hospital treatment, nor provide for the ill health of dependents, leaving many reliant on Poor Law and charitable assistance (Mckinnon, 1994). The 'Beveridge Report', Social Insurance and Allied Services (UK 1942) identified five evils - want, disease, ignorance, squalor and idleness - that comprehensive insurance and provision of services were designed to eliminate. A comprehensive health care system which directly provided health services and was centrally funded was established under the National Health Service Act of 1946. This made medical attention available free of charge to everyone who needed it and came into operation in July 1948. These developments, along with improved public health, nutrition and a ccommodation, have led to an unambiguous improvement in the health environment over the century. The improvements are evident: there has been a fairly continuous increase in life expectancy and declines in infant mortality, deaths caused by pregnancy-related illnesses, tuberculosis and childhood diseases (such as measles, whooping cough, scarlet fever and diphtheria), and typhoid and paratyphoid fever and a decline in notifications of respiratory tuberculosis and smallpox. However, some of the increase in expenditure associated with these improvements should be deducted to avoid double counting with national income. Real expenditure on health (hospitals and public health) has increased from 46p per capita in 1900 to some [pounds]189.67 per capita (in 1900 prices) by 1998. Offsetting the health gains with an element reflecting expenditure leaves a health ranking showing improvement through the 20th century, although declines do occur during the two World Wars (Chart 2a).
Education too was affected by the creation of the welfare state. Although primary education was compulsory from 1870 and the Education Act of 1918 raised the school leaving age to fourteen, secondary education remained the preserve of middle-class children. Compulsory and free education was provided for all school age children in state schools by the Education Act of 1944. In 1947 the school leaving age was set at fifteen but was raised to sixteen in 1972. However, class inequalities in access to higher education remain, with working-class children still being less likely than middle-class ones to continue school beyond the minimum leaving age and to go on to university (Johnson, 1994, p. 310). Two components have been included in the education index, the proportion of 5-19 year olds in primary and secondary education and the proportion of 15-24 year olds in tertiary education, covering both further and higher education. The information is readily available only for state schools, so it has been assumed that private education constitutes a constant proportion of the total over time, which is doubtless unrealistic. There were also problems obtaining comparable figures for various types of tertiary education over time (see appendix). The effect of the Education Act of 1944 is apparent. There is a large rise in the proportion of children in school with the advent of compulsory secondary education. Tertiary education shows greater variability. Mound 22-26 per cent of the relevant population were receiving tertiary education in the first quarter of the century. This rose in the 1930s, dropped back to half its previous level in World War II, but has shown a fairly continuous rise since. Unfortunately the ranking method means that a very small change in actual numbers can generate cycles and these seem to drive much of the component index (Chart 2b).
Political rights are measured by the number of electors on the register at elections as a percentage of the adult population at censuses. The lack of comparability of dates and the exclusion of 19 year olds in the figures for the adult population encouraged the use of ranges to avoid giving too much weight to insignificant changes.  A large jump in rank occurs after 1918 when women aged 30 and over were allowed the vote; men had to be only 21. In 1928 another rise occurred when the voting age for women was lowered to be equal to that for men. In 1969 a further small increase in the electorate occurred when the vote was extended to people aged 18 and over. The indicator for workplace rights, trade union membership as percentage of those working, shows peaks in 1920 and 1930 with falls in between and in the 1930s. Subsequently, there was an increase followed by a plateau through to the 1970s, a further increase followed by a steady decline after 1980. The cycles in the overall index are driven by trade union membership, particularly in the later period.
The final component, living conditions, contains various elements. On the positive side, continuous declines in household size are seen as an improvement indicating less overcrowding. This was possibly aided by the public sector house building programme started after World War II. The increase in travel and communication after setbacks in the 1920s and 1940s is also seen as an improvement. There has been a steady increase in fuel used for domestic purposes, although a decline is observable during the war. The hard labour associated with domestic tasks has been reduced since the advent of gas as a domestic fuel for cooking and lighting in the late 19th century, but fetching water and the use of communal privies continued for many until World War I. Electricity was relatively expensive initially and prices started to fall only after the creation of the Central Electricity Board in 1926 began to improve distribution. In 1931 only 32 per cent of English households had electricity; by 1949 this had risen to 86 per cent (D'Cruze, 1995). The negative aspects of 20th century living conditions have been the rise in reported crimes relative to population size, and pollution, although recently the upward trend in pollution has been reversed. Overall, the two World Wars, the 1960s and the latter part of the 1980s show downturns in this composite living conditions index (Chart 2b).
Although these components of the Dasgupta--Weale index show Spearman's rank correlation coefficients greater than 0.70, comparison of the correlation in year-to-year growth rates reveals the extent to which this is a result of trend growth (table 2). Political rights is the only component whose movements are highly correlated with those of the other components, which is consistent with the arguments for including this measure put forward by Dasgupta and Weale. Movements in the health and education and wealth and income indicators are also correlated.
Aggregating the components into the Dasgupta-Weale index reveals improvement over the century with cyclical movements around the trend (table 1 and chart 4). In particular declines occur in the early 1920s, 1935-45 (WWII), 1955-60 and 1980-90. The 1950s decline occurs because of downturns in the indices for education, political rights and living conditions. Note that World War I is not well identified. This is because data limitations are quite severe before 1920. While the index generates some predictable cycles, it does suffer from an inability to capture the speed of change in components; rapid change may be reduced to a movement of one rank, while very small changes are given equal weight and can generate large cycles. Thus the index appears volatile and may be more accurate in its representation of the trend than cycles.
The augmented HDI
The Human Development Index is comprised of components which measure how far each indicator has moved from the minimum value towards a desirable level or the maximum deemed attainable.  This constitutes a distance measure of welfare ranging between 0 and 1 for each indicator. A simple average of the component indices provides the HDI. The HDI is usually comprised of income, life expectancy and education (years of schooling or literacy), the three measures most commonly identified and correlated with the development of human capabilities and achievements. But the methodology can be extended to incorporate the other components identified as important to living standards in Britain over the course of the 20th century. However, the living conditions component is excluded due to the difficulties of forming priors about the maximum attainable values for household size, domestic fuel use, pollution and crimes. Therefore the augmented HDI is computed on the basis of five equally-weighted components: income, wealth , health, education and political rights.
Most of the elements remain as used for the Dasgupta- Weale index, although a few alterations are made. First, trade union membership is given only a one-third weight and voting a two-thirds weight in the political rights index. Second, education and health indices are computed and then weighted by one minus the percentage expenditure on each in GDP. This attempts to remove the double counting of expenditure on these components resulting from increases in income, so leaving a measure of exogenous improvements in these areas. In addition, expenditure on education as a percentage of GDP is adjusted for the proportion of children in the population before it is applied to the index, so avoiding increases in expenditure arising from increased population appearing as increases in expenditure per capita.  Third, the treatment of income, leisure and inequality is altered. Real income per capita is adjusted for hours worked per capita, giving income per hour worked.  The importance of adjusting gains in income f or leisure gained or lost as an indicator of welfare has already been noted and methods have been devised (Nordhaus and Tobin, 1972; Beckerman, 1980) and have been utilised in living standards comparisons (see Crafts, 1997b; 1997c). Using real income per hour worked values extra hours of leisure at the average wage rate, and assumes there has been no change in the productivity of leisure relative to labour market time. However, in the absence of accurate information on productivity in leisure activities this equivalent valuation is used.  Income per hour worked is adjusted for inequality through multiplication by (1 - Gini coefficient of inequality). This composite measure of income and leisure adjusted for inequality is used as the income component of the augmented HDI.
Before the indices of the individual components can be computed the minimum and maximum attainable values need to be decided. The maximum values are described as the Achievable Millennium Standard (AMS). Table 3 details this standard and the information on which it was based. In essence, the AMS combines the highest levels observed worldwide for each of the variables. Thus it sets a standard of the level of real income per hour worked found in France and Norway, the wealth of the USA, the life expectancy of Japan, the infant mortality rate of the Nordic countries and the tertiary education enrolment rates of the Netherlands as representing the best that might have been achievable in Britain at the end of this century. Of course, by taking 'best practice' for each individual component, no consideration is made of possible trade-offs, so it does not address whether all objectives are, in fact, simultaneously achievable. For instance, more education might entail reduced income per capita, or increased wealth may also necessitate increased inequality. Instead the AMS provides a hypothetical standard against which to measure Britain's achievements over the past hundred years.
Consideration of the individual components shows fairly continuous improvement in most aspects of the index (chart 3). Education again shows downturns in the 1920s, 1940s, 1965 (which may be a product of inconsistencies in the series) and the 1980s. Health suffers a downturn between 1990 and 1995 as a result of the surprising increase in notifications of respiratory tuberculosis; a poverty-related disease which, until recently, was thought to be eliminated. Political rights show the massive jump that occurs in 1918 when women get the vote, but the measure has a brief downturn in the 1920s and a considerable decline from 1980 onwards due to declining trade union membership. The wealth and income security index indicates declines in the early years of the century, during the interwar period and around 1995. Income remained fairly static over the first two decades of the century, but showed a continuous improvement thereafter. 
Overall the patterns of movement in these components balance out to give the impression of steady progress over the course of the 20th century when they are aggregated into the augmented HDI (chart 4). The exception is a brief fall during 1920-25 which was driven by declining trade union membership augmented by less education and declining personal sector wealth. Within this picture of continuous postwar progress the poor performance of some of the components since 1980 should not be ignored. Falling trade union membership may be regarded with some ambiguity, but the decline in wealth after 1990 must occasion some concern and the decline in the health index offers serious commentary on the inequality resulting from the 'income-generating' policies adopted around this time. Furthermore, the Achievable Millennium Standard may be hypothetical, but Britain at the end of this century has some distance to travel to reach the 'best practice' standards achieved elsewhere. It has attained less than 80 per cent of the distance between the assigned minimum and maximum values.
The gender-related development index
The HDI can be broken down into gender components to reflect male and female achievement of capabilities and share in general progress. The aim is to construct an index which will reflect a worse state of affairs as gender inequality increases, given the same overall mean valuation of the component. The method extends Atkinson's (1970) approach to measuring income inequality and constructs an 'equally-distributed equivalent achievement' variable, [x.sub.ede] for any pair of female and male achievements ([x.sub.f],[x.sub.m]) according to a social valuation function which reflects a preference for equality.  [x.sub.ede] is the level of achievement that, if attained equally by men and women, would be judged as socially valuable as the actually observed achievements ([x.sub.f],[x.sub.m]). [ll] The parameter representing aversion to inequality, [varepsilon], is given a value greater than 0. If [varepsilon] = 0, there is no concern for equality and the formula collapses to the simple arithmetic average, equivale nt to the component index calculated earlier. The larger [varepsilon] the lower the valuation given to [x.sub.ede] and the more it falls below the simple component index. Where [varepsilon] = [infty] social achievement is judged purely by the achievement of the worse off group. [varepsilon] is typically chosen to be 2, representing a moderate degree of aversion to inequality and this valuation is used here.  The index provides a measure of overall achievement taking note of inequality. The gender-related development index is calculated from a simple average of these indices, [x.sub.ede], for each component as previously.
In what areas of living standards might a differential impact between men and women have been observed over the 20th century? The equalisation of political rights for women and men has already been noted. Other dimensions of women's lives have also changed and improvement may be expected in the areas of educational attainment, women's health relative to men's as maternal mortality has been reduced, and changes in income shares and hours worked resulting from the rapid increase in women's labour force participation observed in the second half of the century. Some 31 per cent of working age women were employed in 1900 compared with 52 per cent in the labour force in 1998, not so far from the 68 per cent observed for men in that year. Equal opportunities legislation -- such as the Equal Pay Act 1970, Equal Value Amendment 1984, and Sex Discrimination Act 1975 -- as well as women's improved educational attainment and greater labour force attachment, would be expected to have a positive impact on their relative ho urly pay.  These changes would anticipate considerable improvement in women's development index relative to men's over the course of the 20th century. Consideration of the trends in these four variables, used to calculate the gender-related development index, provides partial confirmation of this view. 
Women have always been likely to live longer than men, but improvements in nutrition, medicine and health have been of greater benefit to men through most of the century. Although some relative improvement occurs for women between 1950 and 1970, subsequent decline leaves women in a lower relative position than previously (chart 5). It is difficult to identify exactly what underpins these changes. Data on relative infant mortality suggests that girls benefitted disproportionately from improvements affecting early health until 1970. Subsequently advanced medical techniques for correcting perinatal problems and inherited disorders may have particularly benefitted boys. The decline of employment in heavy industry, mining and agriculture may have reduced the incidence of industrial injuries and employment-related diseases for men. Stress induced by the double burden of working and running homes may have militated against women's health, or women may have become more likely to do life threatening things, such as sm oke, drink and drive.
The gains in education are much clearer (chart 5). While girls suffered no disadvantage in primary education through this century, disadvantage was evident in their secondary education and this was eliminated only after 1945. Disadvantage continued longer in tertiary education. Women were less likely than men to be enrolled in further education until 1955 and in universities until 1995. However, after these dates the balance has not just been redressed but exceeded. Women are now more likely than men to be engaged in tertiary education. Taking all four stages of education together, women's disadvantage in access to education was eroded around 1960 and exceeded male education after 1985.
Women's share of total income has also shown significant gains (chart 6). The relative share of labour income going to women is based on average wages, hours worked, and numbers working full- and part-time. From a reasonably constant 17 per cent of all income from work accruing to women in the first three decades of the century, the 1930s saw women suffering a decline. This situation improved as they were drawn into the labour force in World War II, although subsequent retrenchment caused a brief decline before significant growth occurred from 1965, reaching some 28 per cent in 1998. Women's share of labour income can be applied to national income to obtain per capita incomes for men and women.  As before it is desirable to adjust per capita income for hours of work to incorporate leisure into the measure. However, when looking at work by gender it is important to include not just market work but also non-market work, or housework. Indeed, given women's involvement in unpaid domestic labour and the constr aint this places on their market work it is unreasonable to apply a simple income/leisure trade-off. Thus hours of paid work and hours of unpaid domestic work by both men and women are incorporated into the computation of pay per hour by gender. Some historical data on housework time are available (Gershuny, 1983; Bowden and Offer, 1994) and use is made of these to calculate women's labour time. Men's involvement in domestic labour has stood at about 10 per cent of women's time; thus a value of 0.7 hours of housework per day is assigned for men throughout. This will show men doing an increasing amount of total housework in recent years as domestic labour time has declined. Paid work is only done by those employed so the occupied labour force figures are used to get the total hours of paid work done by women (full-time and part-time) and by men (full-time). Domestic work is done by householders, so the number of men and women that might be assumed to be householders is used to calculate total domestic labour. Adding these two types of work together results in total labour time which is expressed as a per capita figure for men and women in the population (chart 6).  It can be seen that for both men and women there has been a significant decline in average hours worked per week over the century. The decline has been fairly continuous for men, but an increase in hours worked occurred for women between 1935 and 1960 as more women joined the labour force. Women work considerably more hours than men and the gap has widened over time, notably in the 1990s. The income per hour received by men and women is obtained from the income share and hours worked data.
Before the gendered index can be ascertained, achievable targets and minimum valuations for the variables need to be decided. Gendered income per hour is the most difficult to acquire. However the Human Development Report provides information on male and female income per capita and male and female hours of market and non-market work for selected countries (UNDP, 1999, pp.138 and 237). The figures show income per hour worked to be greatest for both men and women in the USA; Norway is the next highest. Norway shows the greater equality in male/female pay ratios: 0.69 compared with the USA's 0.64 and the UK's 0.60. However, it would be incorrect to use the figures for men and women in the USA as comparators as this would implicitly be accepting the USA's degree of gender inequality in pay rates and amounts of paid and unpaid work performed as the desirable level. Instead the maximum achievable income per capita (male and female together) for the least hours worked per capita (male and female together) are consi dered. This assumes that the same total amount of income, leisure and housework could be achieved if all work and pay were distributed evenly between the sexes. Again the USA comes out highest at $10.88 per hour worked. This compares with $9.42 in Norway and $8.33 in the UK. On this basis it is assumed that 31 per cent higher income per hour could have been achieved in the UK in 1998, that is [pounds]0.196 in 1900 prices, and that this would have been the maximum achievable for both men and women if work and pay were equally distributed. In fact, male pay per hour in the 1990s exceeds this figure, rising to [pounds]0.264 in 1998. If the view is taken that the higher income for one sex can only be achieved through unequal division, it seems reasonable not to give the index a higher value ([greater than]1.00) beyond this point. Thus the 1990 male incomes are capped at [pounds]0.196 to avoid undue weight in the income index.
The lowest GDP per capita for both men and women is found in Sierra Leone and the greatest time worked per day in those countries with survey material occurs in Nepal. Combining observations for these two countries gives the lowest level of income per hour worked for both men and women with an overall gender pay ratio of 0.33. These levels are compared with the UK figures available in the same report and the ratio used to adjust our hourly income figures for the UK in 1998 to ascertain minimum values. This results in a minimum standard of 0.05p for women and 0.42p for men, considerably below the figures actually observed in 1900. For the other variables minimum and maximum standards are more straightforward. Voting and all levels of education are quoted as a percentage of the relevant population and are given a minimum value of 0 per cent and a maximum of 100 per cent. Life expectancy values are again taken from the highest and lowest observed in the Human Development Report. The lowest life expectancy for bo th men and women occurs in Sierra Leone at 35.8 and 38.7 respectively. Japan again represents the highest at 82.9 for women and 76.8 for men. Iceland too achieves a life expectancy of 76.8 for men.
To what extent do the gendered components differ (table 4)? There are clear differences in the political index until equality is achieved. Boys' education index is higher until 1960; then girls are seen to overtake. However the difference is marginal as most of the effect occurs at tertiary level and relatively small numbers are involved initially. The life expectancy index is similar for men and women, with only small differences after 1920 until men steal a lead in the 1990s. Large differences are observable in the income and leisure index. Women's relative position has improved since 1900 but they still lag way behind men in 1998, standing at only just over a third of men's hourly pay even in the late 1990s when men's pay has been capped. Hourly pay from labour market work for female full-timers stands at 0.64 of men's in 1998 and women are still spending more time doing unpaid domestic work than they are in the labour market, despite faster reductions in time spent at the former than the latter since 1965 .
What effect do these individual components have on the indices calculated for men and women separately (chart 7)?  Women's gain on men over the first decades of the century was largely a result of their improved political rights. The gap between the indices narrowed in the 1930s but then they pulled apart. However the gender gap has narrowed over the century. The women's index has moved from 48 per cent of men's index in 1900 to 86 per cent in 1998. But improvements have been discontinuous. The maximum level achieved was 95 per cent in the 1930s. In 1965 women's index stood at 86 per cent of men's and there has been little change since. The comparison of the two indices offers an interesting perspective. Getting the vote was the major improvement to women's relative living standards this century. Despite subsequent labour market gains, equal opportunities legislation, improved access to education and recognition of women's issues - such as women's health - the differential has not narrowed since 1965. The extra 10 per cent of total income that women have gained since 1965 has been acquired both through gains in relative pay brought about by equal opportunities legislation and more hours devoted to the labour market. Despite declines in housework time this has led to a relative increase in the total number of hours worked by women compared with men, thus negating much of the relative benefit of the increased income share. 
Calculation of the gendered HDI index and comparison with a reduced standard version based on only these four components still shows the gains achieved over the 20th century (chart 8). The gendered index gives more weight to voting issues in the interwar period but the detrimental effect of persistent gender inequality on living standards is present throughout the second half of century. The gender-related development index implicitly incorporates a gender equality index which can be realised if the gender-related index is compared with the standard index.  The resulting index can vary from 0 to 1 as equality is increased. Consideration of this measure shows a starting point of 0.67 in 1900 rising to 0.94 in 1920. The 1930s are associated with measures of 0.96 and 0.97 and the postwar period characterised by a value of 0.95 until 1985. After this the index slides to 0.93, below the level achieved in 1920.
The progress made over the 20th century should not be understated. Areas of health, education, income inequality and gender inequality have shown significant improvements along with growth in national income and wealth. The Conception and development of the welfare state has been behind much of the improvement. But there is no room for complacency, some of the reversals seen in the 1980s and 1990s demonstrate that the gains are not immutable and need to be protected. Furthermore, women's position has improved if the end of the century is compared with the beginning, but little progress has been made in changing women's position relative to men's over the last few decades on the dimensions considered here. A century of progress? Overall, yes. The women's century? It is more debatable whether the last hundred years warrant such a title.
(*.) Faculty of Economics and Politics, University of Cambridge.
(1.) Most of the data covers the UK, so adjustments have been made to Irish data prior to 1922 on the basis of the situation in subsequent years to incorporate an estimate for Northern Ireland in the early decades of the century. Often data were not available for every observation and interpolation between years has been used for these points. Furthermore, few data were collected during the two World Wars, 1914-18 and 1939-45. In these years 1913 is often used to represent the situation in 1915 and 1938 for 1940. This smooths progress over the wartime periods and obviously means that judgements cannot be made about the effect of war on welfare. Whilst this is undesirable, the absence of adequate data precludes any alternative solution. A detailed description of the data sources and the construction of the variables is given in the appendix.
(2.) The hours worked calculated here are approximately 100-150 per annum more than those found by Matthews, Feinstein and Odling-Smee (1982) for the earlier time periods. Part of this difference may be accounted for by their deduction of sickness and strikes, in addition to holiday, from work time.
(3.) Note that figures are not yet available for the level of wealth in 1998.
(4.) The ranges used are [less than]99 per cent, 90-98, 80-89, 70-79 etc.
(5.) The index for each component in each year is constructed from the value of the indicator, x, as follows: ([X.sub.t] - [X.sub.min])/([X.sub.max] - [X.sub.min]).
(6.) The assumption that a constant proportion is spent on private education through time is maintained. Even if the proportion does vary over time the resultant over/understatement of expenditure relative to GDP will partially offset the under/overstatement of the proportion of children in education.
(7.) The UN has tended to see income per capita as offering diminishing returns for human development over the threshold level necessary to achieve a desirable level of development and so discounts income above this level. This treatment has proved controversial and there have been recommendations that income should be measured relative to an achievable maximum like the other HDI components. Certainly it seems more appropriate to adopt this latter stance for 20th century Britain, where comparisons are to be made against the best achievable rather than whether some standard of development has been reached.
(8.) For additional discussion of the imputation of changes in work hours see Crafts (1997c).
(9.) The only significant correlation in growth rates of the indices is between political rights and health care, otherwise the movements of the components are uncorrelated.
(10.) [x.sub.ede] = ([p.sub.f][[x.sub.f].sup.1-[varepsilon]] where [p.sub.f] and [p.sub.m], represent the proportions of females and males respectively in the population. [varepsilon] reflects the degree of preference for equality. See UNDP (1995), Technical notes, pp. 125-35.
(11.) [x.sub.f] and [x.sub.m] are the distance values for each component calculated as previously: [x.sub.i] = ([x.sub.it] - [x.sub.i min])/([x.sub.i max] - [x.sub.i min] where i = f, m.
(12.) [varepsilon] = 2 represents the harmonic mean of [x.sub.f] and [x.sub.m]. For a ratio [x.sub.m]/[x.sub.f] = 1.5 with equal population shares and [varepsilon] = 2 a unit increase in female achievement will contribute 2.3 times as much to [x.sub.ede] as a unit increase in male achievement.
(13.) The World Bank points out that relative earnings sit uneasily within a measure that is intended to capture relative use of and access to resources, as women may have use of more income than they earn. However, relative earnings capture important asymmetries in amounts of paid and unpaid work performed by men and women and earning may convey more power in decision making and over access to resources within the family, thus rendering it a useful indicator.
(14.) The gender-related development index is usually calculated for income, education and health. Political rights are incorporated into a separate index by the World Bank, the gender empowerment measure, based on income, political rights and men's and women's shares of administrative and managerial positions. Here political rights are added with equal weight into the original set of three, so constructing an index in accord with Dasgupta and Weale's identification of the importance of political rights in determining living standards.
(15.) This makes the, probably unrealistic, assumption that other forms of income are divided in the same gender ratio as labour income. Men might be expected to have a larger share of wealth although this may be partly offset by women's share of pension income resulting from their greater longevity.
(16.) Note that this computation of labour time will tend to under-state men's contribution for those men who are widowed, divorced or single householders. However, women's per capita income is based on pay per hour for full-timers; as women working part-time tend to receive lower hourly pay, this will overstate women's share of total income. Both effects might be expected to increase over time but it is hoped they will counterbalance one another.
(17.) Capping is consistent with the World Bank use of the HDI where GDP per capita is capped at levels consistent with a satisfactory stage of development so avoiding distance to be achieved being a maximum which continually grows along with the expansion of the industrialised nations.
(18.) This calculates an index as a simple average of the four components for men and women separately.
(19.) The lack of improvement in women's relative welfare despite the increase in income share might be taken to imply that leisure has been overvalued. Leisure has been valued at the average wage per hour worked and takes no account of the possibility that the productivity of time spent in leisure may have risen faster than productivity at work. If this is the case leisure will have been undervalued.
(20.) Gender equality index = gender-related development index/standard HDI index.
Atkinson, A.B. (1970), 'On the measurement of inequality', Journal of Economic Theory, 2, pp. 244-63.
Beckerman, W. (1980), 'Comparative growth rates of "measurable economic welfare", some experimental calculations', in Matthews, R.C.O. (ed.), Economic Growth and Resources, 2, London, Macmillan, pp. 36-59.
Bienefeld, M.A. (1972), Working Hours in British Industry: An Economic History, London, Weidenfeld and Nicolson.
Bowden, S. and Offer, A. (1994), 'Household appliances and the use of time: the United States and Britain since the I 920s', Economic History Review, XLVII, pp. 725-48.
Cleary, E.J. (1965), The Building Society Movement, London, Elek Books.
Crafts, N.F.R. (1997a), 'Some dimensions of the "Quality of Life" during the British industrial revolution', Centre for Economic Performance, LSE, discussion paper no. 339.
(1997b), 'The Human Development Index and changes in standards of living: some historical comparisons', Department of Economic History, LSE, mimeo.
(1997c), 'Economic growth in East Asia and western Europe since 1950: implications for living standards', Notional Institute Economic Review, October, pp. 75-84.
Dasgupta, P. and Weale, M. (1992), 'On measuring the quality of life', World Development, 20, pp. 119-31.
D'Cruze, S. (1995), 'Women and the family', in Purvis, J. (ed.), Women's History: Britain 1850--1945, London, UCL Press, pp. 51-84.
Gershuny, J. (1983), Social Innovation and the Division of Labour, Oxford, Oxford University Press.
Johnson, P. (1994), 'The welfare state', in Floud, R. and McCloskey, D.N. (eds), The Economic History of Britain since 1700, vol. 3, Cambridge, Cambridge University Press, pp. 284-317.
Lindert, P.H. (1994), 'Unequal living standards', in Floud, R. and McCloskey, D.N. (eds), The Economic History of Britain since 1700, 1, Cambridge University Press, pp. 357-86.
Maclennan, D. (1982), Housing Economics, London, Longman.
Matthews, R.C.O., Feinstein, C.H. and Odling-Smee, J.C. (1982), British Economic Growth 1856-73, Oxford, Clarendon Press.
Mckinnon, M. (1994), 'Living standards 1879-1914', in Floud, R. and McCloskey, D.N. (eds), The Economic History of Britain since 1700, vol. 2, Cambridge, Cambridge University Press, pp.265-90.
Mitchell, B.R. (1988), British Historical Statistics, Cambridge, Cambridge University Press.
Mokyr, J. (1988), 'Is there still life in the pessimist case? Consumption during the industrial revolution 1790-1850',Journal of Economic History, pp. 69-92.
Nordhaus, W.D. and Tobin, J. (1972), Is Growth Obsolete?, New York, Columbia University Press.
Pollard, S. (1989), Britain's Prime and Britain's Decline, London, Edward Arnold.
Rowntree, B.S. (1901), Poverty: A Study of Town Life.
Rowntree, B.S. and Layers, G.R. (1951) Poverty and the Welfare State.
Sanderson, M. (ed.) (1975), The Universities in the Nineteenth Century, London, Routledge Kegan Paul.
Sen, A.K. (1973), On Economic Inequality, Oxford, Clarendon Press.
Solomou, S. and Weale, M. (1996), 'Personal sector wealth in the United Kingdom 1920-1956', NIESR discussion paper no. 99, September.
United Nations (1995), Human Development Report, Oxford, Oxford University Press. (1999), Human Development Report, Oxford, Oxford University Press.
Vanek, J. (1974), 'Time spent in housework', Scientific American, 231, pp. 116-20.
Williamson, J.G. (1981), 'Urban disamenities, dark satanic mills and the standard of living debate', Journal of Economic History, pp. 75-84. (1985), Did British Capitalism Breed Inequality? Boston, Allen and Unwin.
Ranks of components of Dasgupta-Weale index Year Income Wealth Education Health Political Living DW & work conditions index place 1900 4 3 1 1 3 1 2 1905 1 1 2 2 1 2 1 1910 2 2 6 4 4 3 3 1915 3 4 9 3 3 7 4 1920 5 7 8 5 9 5 6 1925 6 5 4 6 5 10 5 1930 6 6 12 7 7 12 9 1935 8 8 14 9 6 14 10 1940 9 9 3 8 8 11 8 1945 10 10 5 9 10 3 7 1950 12 11 10 11 15 8 11 1955 11 13 12 12 17 12 14 1960 13 13 7 13 13 8 11 1965 14 15 13 14 11 5 13 1970 15 12 15 15 18 14 15 1975 21 16 17 15 20 18 18 1980 18 20 19 17 21 17 20 1985 16 17 16 18 19 19 17 1990 16 18 19 19 17 14 16 1995 19 18 20 20 15 19 19 1998 20 21 21 21 13 21 21 Pearson's correlation coefficients between growth rates in components of the index Political Education Health Income Living conditions Education 0.40 [*] Health 0.53 [*] 0.48 [*] Income 0.71 [*] 0.11 0.07 Living conditions -0.02 0.18 0.08 -0.03 Wealth 0.63 [*] 0.26 0.02 0.83 [*] -0.12 Note: (*.)denotes coefficient significant at 10% level The achievable millennium standard Minimum Maximum Real income per hour worked adjusted for inequality 0.46 30.26 Real wealth per capita 100 780 Proportion of population with state pensions 0 20 Percentage of houses owner occupied 0 70 Building society members as proportion of adult poulation 0 80 Unemployment rate 10 0 Insured as percentage of workforce 0 70 Life expectancy at birth 37 80 Infant mortality rate per thousand live births 191 4 Deaths form pregnancy, tuberculosis, typhoid and childhood diseases as percentage of all deaths 20 0 Notifications of smallpox and tuberculosis per annum 80,000 0 Percentage of population 5-19 in state primary and secondary education 14 84 Percentage of population 15-24 in tertiary education 0 63.3 Percentage of population over 18 with votting rights 0 100 Trade union members as percentage of labour force 0 50
Notes: Income per hour worked: early 20th century studies indicate [pounds]1 per week as low pay for males. Assuming a family of four with approximately half the population in families of four, the other half in families of two, suggests a minimum income of [pounds]17 per annum per capita. Maximum work time per capita calculated at ten hours per day, six days per week for 50 weeks of the year multiplied by the ratio of actual hours worked per worker to hours worked per capita in 1900. A maximum degree of inequality of 0.65 is assumed. This gives a minimum value of pay per hour worked of 0.46p. For 1998 UNDP Human Development Report 1999 gives income per capita for various countries. OECD Employment Outlook June 1998 provides hours per worker and employment/ population ratio. From these, income per hour worked for various countries can be calculated. Norway and France are the highest at $22.62, 34 per cent above the UK at $16.91. Adjusting by an assumed Gini coefficient for inequality of 0.30 for Norway, compa red with 0.373 in the UK in 1998, suggests that inequality-adjusted income per hour worked could have been raised by 49.3 per cent in the UK in 1998, giving a maximum value of 30.26p in 1900 prices.
Real wealth per capita: minimum value assumed lower than the [pounds]150 found in 1920, say [pounds]100. Maximum is taken as private wealth per capita in the USA in 1996 converted to [pounds], 1900 (Statistical Abstract of the United States, 1998, table 770).
Pension based on percentage of population over 60 (Annual Abstract of Statistics, 1999, table 5.3)
Owner occupation of housing: compared with USA where ownership is 66 per cent so the UK seems to be approaching a maximum, say 70 per cent (Statistical Abstract of the United States, 1998, table 469).
Building society membership: determined by 77 per cent adults in population with few added to account for children's accounts.
Unemployment: OECD Main Economic Indicators, October 1999, p. 32. Lowest Luxembourg 2.8 per cent, highest Spain 18.8 per cent, assumed a lower range for the UK.
Causes of death: based on highest UK figures and assumption that deaths from these causes can largely be prevented.
Insured labour force: labour force as per cent of working population 63.7 per cent, maximum of 70 per cent assumed to allow for a small future increase in female participation.
Life expectancy at birth: lowest Sierra Leone 37.2, highest Japan 80.0 in 1997 (UNDP Human Development Report 1999, p.134) Infant mortality: Norway, Japan, Sweden, Finland, Singapore equal lowest at 4, highest Niger at 191 (UNDP HDR 1999, p.1 68).
Disease: Tuberculosis cases per 100,000 people, zero Canada and Australia, 503.5 Dijbouti (UNDP HDR 1999), for UK minimum highest number of cases used, and assumed can eradicate for maximum.
Primary and secondary education: minimum Niger 9.4 per cent primary, 24.4 per cent secondary (UNDP HDR 1999). Averaging both levels and adjusting by ratio of UK figures in UNDP report to percentage found in 1998, as a proxy for the proportion found in state schools, gives a minimum of 14. Maximum: primary 99.9 per cent enrolment in most developed countries, secondary 99.9 per cent in Japan, Belgium, Sweden, Netherlands, Ireland, Korea, Czech Republic. UK 91.8 per cent. Again adjusting to match state school figures suggests maximum of 84.
Tertiary education: Female tertiary students per thousand women and male/ female ratio (UNDP HDR 1999). Adjusted figures suggest 23 per cent 16-24 year olds in tertiary education in the UK compared with maximum percentages of 44 per cent and 47 per cent in the USA and Canada respectively and 0.7 per cent in Chad. European Commission, Employment in Europe, 1996 gives 15-19 and 20-24 year olds in education and training in 1994. This suggests 47 per cent 15-24 year olds in education or training in the UK, 68 per cent in France and 69 per cent in the Netherlands. This is more comparable with the 44 per cent found for the UK in 1998. If the ratio of Netherlands to the UK is applied to this percentage it gives a maximum percentage in tertiary education in the UK of 63.3 per cent.
Voting minimum 0 per cent, maximum 100 per cent of adult population
Membership of trade unions as a percentage of the labour force: extent of unionisation is highly variable across countries, Finland and Denmark have 80-90 per cent of the labour force unionised, but combinations of low unionisation along with considerable influence in bargaining occur; see, for example, Germany. Assume minimum of 0 per cent and maximum of 50 per cent.
Indices of gender-related development index components Year Income per Voting Life Education hour worked expectancy Males 1900 0.17 0.60 0.25 0.46 1905 0.16 0.60 0.31 0.47 1910 0.16 0.59 0.38 0.49 1915 0.16 0.56 0.43 0.50 1920 0.18 0.95 0.48 0.49 1925 0.19 0.87 0.52 0.50 1930 0.21 0.90 0.56 0.53 1935 0.22 0.94 0.60 0.55 1940 0.25 0.91 0.65 0.45 1945 0.31 0.91 0.69 0.49 1950 0.38 0.89 0.74 0.58 1955 0.39 0.93 0.75 0.58 1960 0.46 0.93 0.79 0.57 1965 0.51 0.90 0.80 0.61 1970 0.58 0.97 0.80 0.66 1975 0.72 1.00 0.82 0.74 1980 0.74 0.96 0.85 0.74 1985 0.79 0.96 0.87 0.72 1990 1.00 0.95 0.92 0.76 1995 1.00 0.93 0.94 0.79 1998 1.00 0.92 0.96 0.79 Females 1900 0.03 0.00 0.25 0.42 1905 0.03 0.00 0.31 0.44 1910 0.03 0.00 0.38 0.46 1915 0.03 0.00 0.43 0.46 1920 0.03 0.56 0.47 0.46 1925 0.03 0.54 0.51 0.48 1930 0.04 0.88 0.55 0.51 1935 0.04 0.93 0.59 0.53 1940 0.04 0.92 0.64 0.42 1945 0.05 0.91 0.68 0.46 1950 0.06 0.91 0.72 0.54 1955 0.06 0.92 0.74 0.57 1960 0.07 0.92 0.79 0.58 1965 0.08 0.90 0.81 0.62 1970 0.10 0.97 0.82 0.69 1975 0.14 0.97 0.84 0.78 1980 0.17 0.98 0.86 0.76 1985 0.20 0.98 0.88 0.75 1990 0.29 0.98 0.91 0.82 1995 0.33 0.96 0.93 0.88 1998 0.39 0.96 0.93 0.89
Appendix: data sources and variable computation
(Sources: Mitchell (1988), British Historical Statistics, abbreviated (M); supplemented with later data from Statistical Abstract of the United Kingdom/Annual Abstract of Statistics, various years, abbreviated (C); unless stated otherwise).
Income. Real income per capita, [pounds] 1900. Population: male, female, by age (M, pp.11-13) Interpolated between census years pre-1980. Prices: Ministry of Labour, Department of Employment index of retail prices, UK (M, p. 739). Rebased to get a series from 1900 to 1980, food items pre-1914, all items subsequently. National income: current prices, UK (M, p. 828)
Hours worked and hourly earnings. Average weekly earnings and hours of manual workers UK 1938-80 (M, pp.173-8). Indices of normal weekly hours of manual workers, UK, 1920-80 (M, p. 147). Money wages UK 1880-1936 (M, p.151) used for 1900-15. Weekly wage rates of manual workers UK, 1900-80 (M, p.151). Average normal hours in certain industries (Bienefeld, 1972, p.122). Series combined to get consistent series over the century. Women's wage rates to 1935 assumed to be 50 per cent of those of male workers. Hours calculated from various series. In the early years we classify industries as male or female, suggesting women work 1.5 hours less, 1920--35 assumed females 0.3 labour force, males 0.7 and average hours difference of 4.2 in 1938 applied to get average full-time hours for men and women. Part-time hours from 1950 onwards (British Labour Statistics, Historical Abstract).
Labour force. Occupied, unoccupied, unemployed, male and female (M). Female part-time workers, 1950 onwards (Employment Gazette, various issues).
Holidays. Annual paid holiday entitlement (Social Trends, 1995, table 4.17) Weekly hours of work and paid holidays: manual workers UK, for 1951-95 (Social Trends, various issues). Paid holiday starts in 1930s, assumed one week.
Housework. Housework hours per day 1920, 1935, 1960, 1975, 1985 (Bowden and Offer, 1994; Gershuny, 1983, p. 151) Same assumed pre-1920 on basis of evidence in Vanek (1974). Interpolated between years and 5 hours per day assumed from 1990 onwards.
Householders. Female householders married, widowed and divorced women per thousand women aged 15+. (M). Same assumed for men.
Inequality of income. Gini coefficient calculated from distribution of personal incomes before tax 1915, 1940, 1950 and at five-yearly intervals subsequently (C). 1880-1913 figures based on Bowley and Stamp data for all income recipients (Williamson, 1985, p.68, table 4.5). Interpolation used between earlier years.
Female share of total income. Calculated as numbers working times weekly wages times hours for female full-timers, and numbers working part-time times weekly wages times relative part-time/full-time hours for women working part time, given as percentage of female income plus male labour income (numbers employed and wages).
Working time -- annual total. Calculated as product of weekly hours worked in the labour force, 52 weeks a year, and numbers employed for male and female, full-and part-timers; less holiday entitlement.
Non-labour market and labour market working time -- annual total. Males: hours in labour market worked by those occupied plus 4.9 hours domestic work per week done by householders. Females: hours in labour market worked by those occupied (full- and part-time separately) plus housework time per week times number of female householders.
Unemployed. Total numbers registered as unemployed, GB (M. pp.127-8) calculated as a percentage of the active labour force.
Unemployment insurance. Numbers with unemployment insurance (C), calculated as a percentage of the working population.
Wealth. Personal sector wealth 1920-56 (Solomou and Weale, 1996), 1961-95 (Social Trends, various years), assumed 150 in [pounds], 1900 for earlier years and assume highest level achieved (1990) also applies to 1998. Given in real values per capita.
Pensions. Pensions payable under Old Age Pensions Act, etc. 1915 onwards, male and female (C), given as percentage of population
Building Society membership. Number of members/share investors and depositors of Building Societies, GB, 1913 onwards (C). Earlier information on numbers of members for 1869, interpolated (Cleary, 1965, p. 288). Last figure adjusted to be equal to highest to account for demutualisations. Given as a percentage of population.
Owner-occupied housing. Percentage of houses owner-occupied 1914, 1938, 1950 (C), supplemented by information from Maclennan (1982, p. 172).
Expectation of life at birth. Male and female, England and Wales (C).
Infant mortality. Deaths of infants under one year per thousand live births (M, p. 57 ff, C).
Deaths analysed by cause. Deaths from typhoid and paratyphoid fever, tuberculosis, puerperal septicaemia and other pregnancy-related, measles, scarlet fever, whooping cough and diphtheria (C). Given as a percentage of total deaths. 1923 used for 1920, assumed to be the same percentage pre-1920.
Notifications of infectious diseases. Respiratory tuberculosis and smallpox, numbers 1920 onwards (C), assumed same pre-1920.
Health expenditure. Public expenditure of local authorities, UK, other than out of loans till 1929, other than capital works after 1929, on sewerage, water supply and hospitals, health, welfare and local authority services (M. pp. 612 ff). Expenditure by central government on health, 1950 onwards (C). Given as a percentage of national income.
Education. Number of children attending inspected day schools, primary. Number of pupils at public and grant aided secondary schools, UK (M, p. 798 ff). University education, number taking courses, male and female, GB, 1920 onwards (M, pp. 811-12). Earlier information from Sanderson (1975) and Pollard (1989, p.182 ff). Public education. Number on school registers, primary, secondary, UK, boys and girls (C). Training, evening and similar schools, day technical classes, art classes, continuation classes, specialised education for industry and commerce, male and female. Further education establishments, part-time, full-time, sandwich courses, technical instruction. Universities includes polytechnics from 1992. Male and female (C). The inconsistent bases on which these series were collected across institutions, countries and gender meant that various manipulations and assumptions were required to attempt to get a series for the whole century. From 1945 onwards there was much improvement in numbers almost entirely as a result of women doing evening classes, as it is unclear whether this represents hobbies rather than training these figures were omitted from subsequent totals. Primary and secondary pupils given as percentage of population aged 5-19. Tertiary education, further and higher, given initially as percentage of population 15-19 but later adjusted to include 20-24 year olds, so comparable with estimates from other countries when trying to ascertain the Achievable Millennium Standard.
Trade union membership. Membership of trade unions, GB and NI, male and female (M, p. 137). Given as a percentage of those working.
Expenditure on education. Expenditure of local authorities on education (M, p. 612 ff), calculated as a percentage of national income then adjusted for the proportion of children in the population for each year. Expenditure declines in recent decades due, for instance, to the opting-out of grant maintained schools, therefore total expenditure is assumed to remain at its previous highest level of 5.1 per cent after 1980.
Electorate. Numbers registered at general elections, UK (M, p. 793). Given as a percentage of the population 20 and over. Male and female electorate are based on the age distribution of the adult population in 1931 to account for differential patterns arising from men having the vote at age 21 and women age 30 from 1919-27.
Household size. Occupied dwellings at census, GB (M, p. 389; C). GB population divided by number of houses.
Domestic fuel used. Consumption of coal in the UK according to use, million tons, domestic (M, p. 258), sales of electricity by type of consumer GB, gigawatt hours, domestic and farms (M, pp. 263-4), gas sold to consumers, GB, million therms (M, p. 269). Converted into gigawatt hours at gas*29.31, coal*6096.3. Given as domestic fuel per capita.
Pollution. Consumption of coal in the UK according to use, million tons, domestic, industrial and transport (M, p. 258), multiplied by 100 to give approximately equal weight to car use. Vehicles in use, GB, private cars (M, p. 557).
Crimes. Crimes known to the police, thousands, GB, against property and persons (M, p. 776). Calculated as a percentage of the population.
Travel and communication. Train mileage, million passenger miles, GB (M, p. 545 ff). Vehicles in use, GB, private cars (M, p. 557), assume average annual mileage of 12,000. Airway mileage and traffic UK, passenger miles flown, millions (M, p. 561). Post-office mail traffic, UK, millions, letters, postcards and packets (M, p. 563), Telegraph and telephone services, UK, trunk calls made (M, p. 566), Broadcast receiving licences, UK (M, p.569). Given per capita.