Unemployment in New Zealand at the peak of the Great Depression
by Keith Rankin, Economics Department, University of Auckland
Paper first prepared for the
1994 Conference of the Economic History Association of Australia and New Zealand,
at the University of New England, Armidale, NSW, Australia; July 6-10, 1994
The Government of this country seems to have withdrawn into a kind of mental euthanasia. It sits there sublimely satisfied that all is well ... The official figures register eighty thousand unemployed. These figures do not include women. If we include women and youths we find that the number is practically double; ... add to those the number of people who are in employment but who are working only half time or part time, and earning no more than relief rates of pay, and we will find that they total practically as many as the total registered unemployed.
(Elizabeth McCombs MP, Parliamentary Debates, 28 September 1933)
While the depression of the 1930s - the Great Depression - is well known as a period of hardship arising from lost employment opportunities in New Zealand there is a dearth of published data by which we can assess the extent of unemployment and underemployment. In these circumstances it has been easy for inadequate data to become embedded into New Zealand's historiography. Accounts of the past which utilise misconstrued data have contributed to the creation of the myth that unemployment was much less severe in New Zealand than it was in the other countries we generally compare ourselves with: Australia, Great Britain, Canada and the USA.
Censuses were held in New Zealand in 1921, 1926 and 1936. The census scheduled for 1931 was cancelled as an economy measure. Australian censuses were taken in April 1921 and June 1933. Registered unemployment in New Zealand peaked in the third quarter of 1933, soon after the Australian census. By that time, Australia had experienced two consecutive years of economic growth.
In the inter-war years, migration flows across the Tasman Sea, although not large, were very sensitive to the relative health of the New Zealand and Australian economies (Rankin 1992, p.64). For New Zealand they confirmed the picture revealed by gross product data (Rankin 1992, Chapple 1994, Rankin 1994): a much greater recession in 1927‑28, a recovery in 1929‑30 which Australia did not experience, a later depression trough, and a stronger expansion in the late 1930s. There was a net migration flow to Australia in 1933 and 1934, suggesting that Australia's labour market was not as depressed as New Zealand's in 1933. Demand-deficient "cyclical" unemployment must have been at least as high in New Zealand as in Australia.
Structural unemployment may have been lower in New Zealand, given the greater ability of rural communities to absorb the urban jobless. Indeed the less industrialised Australian economies - Queensland and Tasmania - had lower unemployment than other parts of Australia in much of the Depression. There was almost certainly more seasonal unemployment in New Zealand in July 1933 than there was in Australia at the time of the census, given the dominance within manufacturing of primary product processing for export.
A peak unemployment estimate should reflect cyclical, structural and seasonal factors. New Zealand labour market statistics used for comparative purposes have generally been taken in March or April, thereby ignoring seasonal unemployment. The New Zealand censuses of the 1920s were held immediately following business cycle peaks (Rankin 1992, p.61); thus it may be assumed that they reveal the respective workforces at their "natural" rate of unemployment. In 1936, the census predated the expansion which followed the election of New Zealand's first Labour Government. While unemployment at that census remained well above any normal level, three years of economic recovery had ensured that unemployment levels were well below their depression peaks.
This paper uses Australian census data to derive an estimate of New Zealand's male unemployment rate in July 1933. It utilises census income data to estimate the size of the true 1933 female workforce. By deducting those who were employed from the workforce total, we get a more realistic measure of female joblessness in New Zealand's Great Depression.
At the depth of the Depression, New Zealand had approximately 80,000 registered unemployed, about 12% of the labour force. There is a great deal of uncertainty about unemployment among married women and Maoris, and in rural areas, but the broad relativity between this [12%] and figures of 44% in Germany, 25‑30% in the United States and Australia, and 20‑25% in the United Kingdom is not really in doubt. Industrialised countries had more redundancies, a quantity reaction, while agricultural countries had more wage reductions, a price reaction.
(Hawke 1985, p.124)
The most commonly cited set of mid-depression unemployment estimates for New Zealand are by Macrae and Sinclair (1975). They are summarised in Table 1. Unfortunately these estimates contain a number of basic errors, which mostly have been corrected in Table 1(b). Other errors include differing definitions of the commencement of working-age for the three years considered; the authors used both age 15 and age 16 despite noting that the correct age is really 13 (p.39, n11).
Macrae and Sinclair assumed constant participation rates for both males and females through the period. They used the Unemployment Board registration total for July 22 (NZOYB 1934, p.594) as an estimate of adult male unemployment. They stated that their unemployment estimate was "conservative" (p.38); ie really a lower bound. As a corollary, they subtracted unemployment from their male labour force estimate to give an employment total of 416,648 which amounts to 85% of the population aged 16‑64. Comparing this with a fulltime employment rate of 66% of 15-64 year‑old males at the time of the latest population census (March 1991), it is clear that their employment total is unrealistically high. Even when using the modern labour force approach which plays down the extent of unemployment by making male participation rates highly cyclical, male employment rates show big falls during any depression.
Youth unemployment was evidently much greater in 1933 than 1935. Despite the lower wages payable to "boys", it is likely that they had unemployment rates well in excess of the adult rates assumed by Macrae and Sinclair (p.39).
In the cities, boys have been absorbed in mercantile, industrial and professional pursuits by the efforts of committees in placing them in touch with prospective employers. Many such positions have doubtless been 'made', and but for the interest displayed by the organisations referred to the employers concerned would probably not have provided vacancies. ... the problem now seems to be confined generally to those who left school during the depth of the depression. Very little trouble seems to be experienced now in placing boys immediately they leave school.
("Report on Juvenile Unemployment"; NZAJHR H-35, 1935, p.23)
For females, the Macrae/Sinclair workforce estimate was an interpolation of the 1926 and 1936 census data. They made a downward adjustment of 6,000 on the interpolated employment estimate, giving, by subtraction, an unemployment total of 6,760. Yet they note that in the ten years from 1926 to 1936, female employment grew "much faster" than male employment (p.40) - somewhat contradicting an earlier claim (p.39) that female participation rates were "virtually the same" in the two census years - suggesting that this may have been a result of wives actively seeking work because of their "husbands' meagre allowances" (p.40).
This recognition of the added‑worker effect is very pertinent to any estimates of female participation rates during a recession or depression. In the interwar years, it was more pertinent with respect to single women than married women. In the 1920s, it was normal for young single women to live in their parents' homes as dependents. Furthermore, women in work in the early 1930s delayed marriage and pregnancy, the routes by which they traditionally took an early exit from the workforce. They had parents and siblings to support, and few potential marriage partners were in a position to support them. Realistically, most teenagers and women in families which were adversely affected by the depression would have taken up any socially acceptable opportunity to have added to the family budget. The true size of the workforce is the entire working age population, less those who make a positive choice in favour of activities that are outside the bounds of the labour market. During a depression, peoples choices become very fluid; opportunism becomes predominant.
Macrae and Sinclair, to their credit, attempted to estimate Maori unemployment. It should be noted that New Zealand census data on unemployment excludes the indigenous Maori population. Despite Maori being an essentially rural population at that time, there is no reason to doubt the authors' contention that Maori unemployment rates were higher than average. If a census had been taken in July 1933, by ignoring a significant minority it would have understated total unemployment. However, the 1933 census was taken in Australia, where the aboriginal population was proportionately smaller. Thus there will be only a small bias from the Australian data as a result of the exclusion of indigenous peoples. The New Zealand estimates presented here relate to a non‑Maori population base. The inclusion of Maoris would add to the workforce totals, but participation and unemployment rates would remain valid within the margin of error.
Recent Australian labour force data (Maddock and McLean 1987, p.354) show an increase of nearly 330,000 jobs (16%) in the three years after the depths of the depression. Similar economic recovery should have raised total employment in New Zealand by over 70,000, which translates to a fall in unemployment of the same magnitude if using Macrae and Sinclair's stable participation rate methodology. Adding 72,000 to the 38,000 fully unemployed in the 1936 census gives 110,000 unemployed in 1933 (171/2% of the workforce). Employment in New Zealand factories -13% of the employed workforce in 1936 - rose by 13,600 (24%) between 1932 and 1935 (Rankin 1990, p.169). Economic growth rates during the recovery were higher in New Zealand (Rankin 1992, pp.53,61; Maddock and McLean, pp.359-364), suggesting that employment growth in New Zealand from 1933 to 1936 should have been at least 16%. This estimate of 110,000 should be taken as a lower bound; an estimate which grossly understates female unemployment.
Any measure of unemployment must fall between two limits: a lower bound which includes those who habitually perform market work but are without any work on a specified day in July 1933, and an upper bound which includes everyone without a job who would accept one if offered. I have given 110,000 as such a lower bound. And I have suggested that there would have been comparatively few people who would have refused a reasonable job offer, given the high opportunity costs of parenthood, retirement and full-time education. This paper produces a range of estimates which fall between those limits. They are summarised in Table 2:
Measuring the labour force in this manner meant that certain marginal groups were not enumerated. Persons seeking jobs for the first time would not be included ... neither would part‑time workers who did not regard their work as customary.
(Mawer 1982, p.6)
There has been some discussion about whether the measured increases in participation rates over the first half of the twentieth century have been anything more than statistical artefacts (eg Jaffe 1956, Smuts 1960). One point of contention is over the difference between the "gainful worker" or the "labour force" approach. Early twentieth century censuses in most countries followed the former approach. In New Zealand the labour force approach was first adopted from 1945 (Mawer, p.7). A gainful occupation was a personal attribute which described a person's customary activity. Unemployment was a category within the wider category of a gainful worker. Thus an unemployed person was a gainful worker without present employment. A man of normal working age would only cease to be a gainful worker if he completely changed his lifestyle. Census questionnaires designed under the gainful worker approach revealed what they were designed to reveal; namely, that most men were gainful workers and most women were dependents. Participation rates were stable.
A key difficulty with the gainful worker census was the way it regarded workforce entrants. The problem is clearly revealed by the fact that in 1936, a year of very high unemployment by 1920s' standards, fewer women were classed as unemployed than in the full employment months of April 1921 and April 1926. The many women and teenagers who wanted employment during the depression but could not find jobs were no less unemployed than those who had lost their jobs. People who had never been employed could not be classed as "unemployed wage and salary earners". Not having been wage or salary earners prior to their becoming unemployed, they were classed as dependents. In addition, many teenagers who had stayed on at school or had left school were working on a casual basis. To be classified as both a (customary) dependent and a (casual) wage-earner would be in conflict with the requirement to give a single discrete answer to the census question.
The labour force approach, on the other hand, takes a snapshot of recent activity, such as whether a person is working or, if not, has been seeking work in the previous month (ABLMR 1985, p.3). It gives an objective upper bound to employment, and an equally objective lower bound to unemployment. Thus the labour force excludes discouraged workers and many seasonally unemployed workers. The labour force approach gives a misleading view of economic activity, producing a population subset that is very strongly correlated with the employed workforce. It makes the following value judgements: (i) that unemployment is determined by people's willingness to search for work, (ii) that people should be identified principally by their labour market activities - employment or actively seeking market work - regardless of how unimportant a part such activities may play relative to their overall activity. The labour force approach regards non‑labour‑market activities as residual.
An alternative "workforce" approach assumes that participation rates are stable, with the only element of periodicity being the added worker effect. Discouraged workers are classed as being in the workforce, as are people engaged in second choice activities while no job is available for them. Employment, defined restrictively as "formal employment", is the sum of people classified as entrepreneurs (employers or self-employed) or as employees. Unemployment is the residual difference. Macrae and Sinclair take such an approach, except that they regard employed males as a residual (all males excluding identified non-participants and identified unemployed).
Figure 1 summarises the key differences in the labour force and workforce approaches. Residual employment includes part-timers who do other activities, and who may or may not class their employment as their main economic activity. It also includes people working in family businesses for no wages. Residual unemployment includes all of those who are not employed, would like to be employed, but do not fit the contemporary official criterion of unemployment. The borders between residual employment, unemployment and non‑participation are blurred at the best of times, more so during an economic slump, when the opportunity cost to non‑breadwinners (people who are not households' main income earners) of not working becomes very high. At times of weak labour demand, any assessment of whether the large numbers of people with low or nil incomes are in the workforce or not is subjective; some would say meaningless (eg Smuts 1960, p.79).
In this paper's opening quotation, Elizabeth McCombs separated the unemployed into three categories; categories which match the three central categories of Figure 1. Following the workforce approach, I have classed all three categories as the "residual workforce", representing unemployment. Under the labour force approach, only the category "official unemployment" meets its unemployment criteria. In practice, however, the present official unemployment category includes many people who did not meet the official criteria in the days of the gainful worker approach. Thus the gap between a hypothetical unemployment total gathered via a modern household survey and the estimates presented here, is probably not as large as might be expected. Certainly, a modern survey would show the female labour force in the 1930s to be well in excess of the census measures.
There is no objective way of counting everyone who would accept a job at prevailing wages. The biggest source of uncertainty is the number of unemployed additional workers. The added worker effect has an impact on teenagers, single women and the retired, as well as married women. None of my estimates of unemployment can be a true upper bound. With respect to the census, unemployment is the sum of all employment categories other than employer, self-employed, employee, dependent and retired. "Not specified" is included as residual; hence as unemployed. Indeed, the incomes of those who did not specify an employment status were on average, well below those of the employed.
With respect to Table 2, my lowest unemployment estimate for 1933 is the Australian unemployment rate applied to the New Zealand population. A more realistic estimate is gained by applying male unemployment ratios to raise the size of the female workforce. Sufficient post‑war census evidence exists to indicate that females are at least as likely as males to be unemployed (Rankin 1993a). Indeed discrimination against females was almost certainly much greater in 1933 than in recent years of full employment and Equal Employment Opportunity programmes. My preferred estimate reallocates a number of additional female participants from dependency into the residual workforce. This is done via an analysis of incomes, disaggregated by sex and age. Table 2 shows this estimate to be comparable with an estimate of the number of people who had some income, but less than 30% of a basic wage.
All the unemployment figures available for the Depression are grossly deficient in that they do not give any real idea of the extent of female unemployment.
(Broomhill 1978, p.18)
It is necessary to check on the comparability between the workforce data of the two countries. Table 3 displays the data needed to give some perspective on the inter-war workforces of New Zealand and Australia. There is close comparability in the workforce structures of Australia in 1933 with those of New Zealand in 1936, with the exceptions of unemployment and of the participation of young people, especially young males. Australia's apparently lower participation rates for young people in 1933 reflect the way that school leavers of both sexes who were unable to secure employment were classed as dependants.
In April 1921 both countries had a census. There were no income data collected in either census. Other aspects of workforce data are comparable across the two countries but with several key reservations. Male unemployment was nearly twice as high in Australia, female unemployment was lower in Australia, female employment was lower in Australia, and more males in New Zealand were employers rather than employees (reflecting a greater proportion of farmers in the New Zealand population). This could suggest that Australia had a higher natural unemployment rate than New Zealand in the inter-war years, but for several reasons that is, at best, only part of the story:
• New Zealand lost a greater proportion of its male population during World War I. (Rankin 1990, Figure 2E, p.47; Neale 1929) This in itself is a reason for higher participation rates by females and employment rates by males in early 1921.
• New Zealand GNP estimates (Rankin 1992) indicate that New Zealand had higher economic growth rates than Australia in the two years prior to the 1921 census. If those statistics are valid, then the Australian macroeconomy was out of phase with New Zealand's in April 1921. Forster (1965, p.449) indicates that there was a sharp increase in unemployment in Victoria in the last quarter of 1920.
• A greater seasonal employment peak in New Zealand in April may indicate that New Zealand's overall employment rates are overstated.
• New Zealand male and female participation rates fell between 1921 and 1926. By contrast Keating's Australian data (1973, pp.340-341) show stable participation rates for males between ages 15 and 64, and rising participation rates for females during this period. Jones (p.74) has adjusted the Australian female workforce upwards for both the 1921 and 1933 censuses, having reconsidered the female rural labour force. His graph shows a rising female participation rate from 1921 to 1933, with no further increase until the 1950s. His adjustment procedures should also be relevant for the New Zealand censuses.
• The population of New Zealand became significantly more urban during the 1920s, so employment patterns should have been converging with those of Australia.
in Australia peaked in 1931/32, at least 12 months before New Zealand. Thus, if
New Zealand did have a lower natural unemployment rate, then labour force data
from the Australian census of mid-1933 may be a very good indicator of incomes,
employment and unemployment in New Zealand in 1933. If there was no significant
difference in the natural rate of unemployment between the two countries in the
1930s, then New Zealand workforce estimates based on the Australian census
would understate the degree of unemployment in New Zealand. Migration flows and
GNP data certainly imply that unemployment in New Zealand was higher than in
Australia in July 1933.
The New Zealand exchange rate was devalued by 15% in January 1933 (NZOYB 1934, p.652) bringing New Zealand's currency to parity with the value of the Australian pound set in 1931 (Schedvin, p.155); ie 1.25 New Zealand pounds equalled 1 pound sterling (UK). A highly valued exchange rate inhibits the growth of employment in the tradeable sector, and there tends to be a lag between, say, GDP growth and employment growth. Thus July 1933 was a date in which New Zealand's unemployment should have been higher than usual, relative to Australia's, and is exactly when employment levels could be expected to have been at a minimum.
Census workforce totals are shown in Table 4. The Australian census data have been applied to the New Zealand population on an age-group basis to produce a first estimate of the mid-depression New Zealand workforce.
Table 4B includes revised estimates for females for all three years. The revisions have been made by simply assuming that, for each age group, female employment rates were the same as those for males. Age-composition effects bring the aggregated female employment rate to below the male rate.
From the above summaries, it appears that nearly 30 percent of the mid-depression workforce were not formally employed. This is strong evidence for the added worker effect, because an adequate allowance for unemployment raises the female participation rate in 1933 from 271/2 percent to 321/4 percent. Female employment increased from 201/2 percent of total employment in 1926, to 231/4 percent in 1936. It was an estimated 231/2 percent of the total in 1933. Put another way, female employment barely fell from 1926 to 1933, whereas male employment fell by over 13 percent.
The census income statistics give an alternative perspective on labour supply. Income data were first collected in 1926 - despite some misgivings relating to concerns about personal privacy. They have always been collected in bracketed form; people would be asked to state whether their annual incomes were in certain ranges, such as nil-£52, £52-£103 or £103-£155. The upper bracket was £364 or more (£260 in Australia), so that work on income aggregation and distribution had to be done in conjunction with data derived from income tax returns. There was one difference in the definition of income between 1926 to 1936. In 1936, people classed as employees were asked to include the value of payment in kind such as free housing as part of their incomes. This appears to have had little effect, given the significantly higher income levels in 1926. It does not affect people classified as dependents. The Australian definition of income is essentially the same as that for New Zealand in 1936. The self‑employed were asked to indicate their net incomes.
From 1926, people in each workforce category have been cross-tabulated against income in the census tables. For New Zealand the 1936 tabulations are not strictly consistent because the income estimates cover the previous calendar year, 1935. Moreover, in both New Zealand censuses there were a number of employees without income. The most likely explanation is that the income given applies to the previous calendar year, rather than the previous 12 months. The income data for the 1926 census, which are for the year to March 1926, are taken as applicable to the 1925 calendar year. For Australia the income census is for the June year 1932/33.
Taking 1925 as a benchmark, retail prices in Australia in 1933 were at about the same level as in New Zealand in 1935. Retail prices were 20 percent lower in Australia than they were in 1925. New Zealand retail prices were 17 percent lower in 1935 than in 1925 (OYBCA, NZOYB). Incomes in Australian and New Zealand censuses were comparable.
The income statistics define the "income-force" rather than the work-force; the sum of all people with incomes. People without incomes were dependent on other family members, or were subsistence producers. Those with incomes included beneficiaries, superannuitants and people with investment income or pensions. The income-force included many people whose employment status changed during the year. The nearest income-force equivalent to "residual employment" can be called "marginal income". It applies to anyone receiving some income, but less than £1 per week.
Income tables were published in the 1926 census for each sex by age and, in a separate table, by workforce category. There was no tabulation of income and age for each workforce category, as there was in 1936. For 1936, income data could be used to reallocate female "dependents" into the residual workforce on an age-group basis. For 1926, any reallocation had to be done on an aggregate basis.
The Australian census of 1933 has very useful income information, collected in essentially the same way as that for New Zealand. Unfortunately, however, there is no simple tabulation of incomes by age. Rather, dependents were separated from breadwinners and then breadwinner incomes were tabulated by age. Thus the Australian definition of "breadwinner" was narrower than the New Zealand definition (which included all income recipients). Although breadwinners did include retired people living on pensions or on investment income, it was impossible to compare the workforce with an independently derived income-force on a complete age/sex basis. However, the published Australian data give some information, about the age distribution of dependents. By classing retired breadwinners separately from dependents, it is reasonably safe to assume that most female dependents with incomes were of working age.
The income statistics have been further grouped to facilitate the comparison of real incomes in 1925 with real incomes eight‑ten years later. Table 5 delineates the different income categories.
In 1933 the minimum adult male wage - eg for a textile worker or a builders' labourer - was 74s per week, over £190 per annum. A basic annual wage in 1925 was about £220. Thus, the medium income group includes more people in 1933 and 1935 who earned less than a basic annual wage than in 1925. Broomhill (1974, p.31) gives the basic wage in South Australia in 1932/33 as being £164. Because of the way the census data was tabulated, it is convenient to think of £155 (£3 per week) as being the basic male wage in both countries in the early 1930s, and £207 (£4 per week) as the equivalent in 1925.
Most people with marginal incomes could be regarded as underemployed or as part-time additional workers. Some had been fully employed for part of the year. On the other hand, many people classed as unemployed had incomes well above £52. The 1926 census - taken at a time of near‑full employment - did not reveal any social preference for part-time work, as there is amongst many caregivers today (NZHLFS 1993). Part-time employment was rarely a first choice activity; part‑timers were very much residual workforce participants.
One way of expanding the residual workforce concept would be to take the union of those classed as dependents with marginal incomes and those people classed by the census as being neither entrepreneurs nor employees. There would still be errors, however. Such a measure of the labour reserve would include housewives with investment income and it would exclude many unemployed additional workers.
It is well known that, for many, real wages and living standards rose during the depression. As long as one could remain in employment on a relatively fixed income, then a general decline in living costs ensured this result."
(Macrae and Sinclair 1975, p.40)
Figure 2 shows changes in the income distribution of young people from 1925 to 1935. For males over 16 there was no change in the numbers with no incomes. Rather, the main effect of the depression was the casualisation of work, with a big rise in males receiving incomes below £1 per week, well below the minimum wage level for adult labour. The graph shows that very few young males were able to find reasonably well-paid employment in 1935.
The graph shows an increase in the number of 14-year-old children of either sex with incomes. This reflects the increasing numbers of children who were withdrawn from school prematurely during the Depression. However, at 15, the most common school-leaving age in 1925, there was an increase in people without incomes. This reflects the greater difficulty young people faced in gaining employment, rather than an increasing preference for schooling (Rankin 1990, ch.3).
Many more females aged 16-24 had marginal incomes in 1935 than in 1925. Most of this increase represented women slipping from formal employment into residual employment. But, with increasing age, the marginally employed also represented additional workers. There were fewer young women with no income during the Depression.
Figure 3 focuses more closely on the 25-29 age group. This is one group strongly affected by delayed marriage and child raising. There was a 25% increase in the number of women in this age group with incomes, despite the fall in labour demand. It was not an age group for which there was likely to be an increased alternative source of income such as a pension, widow's benefit, or investment income. Virtually all of those women on marginal incomes were additional workers. For males on the other hand, there was an increase in those with no incomes - despite a halving of the non‑participation rate (Rankin 1990, p.83) - and a big increase in those receiving marginal sustenance. There was more long-term unemployment (males without incomes), and many more males in their late twenties - the normal age of marriage for men - whose ability to support a family had been destroyed.
For adult men (Figure 4), male income distributions were quite stable once the marginal income category was excluded. For adult men, the employment structure was preserved, with proportionate losses from each income group boosting the marginal income group. This is a pattern of changing income distribution which suggests widespread loss of employment and not a fall in real wage rates. While women were moving into casual employment from outside the workforce, men were moving into it from formal employment.
In both 1925 and 1935, adult women showed an increasing likelihood to have an income as their ages increased. The trend is much more marked in 1935, as more women became eligible for age benefits. Yet the main reason for increased numbers of women under 60 having an income is an increased need in 1935 over 1925 for women to work for an income. It is also interesting to note (see Table 6) the marital status of all women over 16 with nil incomes, and the proportion of women of each status with nil incomes. Additions to the income-force are apparent for all groups of women, but least for married women. Thus increasing numbers of women with incomes cannot be explained by the family benefit which was introduced in 1926.
Figure 5 shows the increased spending power of women relative to men. For those with incomes, real average incomes appear to have fallen by about 10 percent across all ages except the very young and the old. However, when looking at all females aged 25-59, real average incomes have been maintained. This is the effect of additional workers on purchasing power. At high age groups, 1935 showed real increases in average female incomes. Moreover, the aggregate demand implications of this are more significant than the graph suggests, because, especially in 1935, a larger proportion of elderly women had incomes than did women in other age groups, and there were more of them than ever before. However, they may have had a low propensity to consume during the depression. For both sexes, the increasing purchasing power of the elderly indicates the increased real value of fixed investment incomes after a period of falling prices.
Was there a rise in the numbers of people on high real incomes, as a result of falling prices? Figure 5(b) suggests that there was for many retired people, who lived off investment income. Nevertheless, figure 4 shows that the numbers of men receiving high incomes fell from 1925 to 1935, despite the relaxation in the definition of "high income" (Table 5). The income data confirm that falling living standards had been the dominant experience during the Depression. Real wage rates may not have fallen, but real incomes certainly did. This was a quantity reaction, not a price reaction (ref. Hawke 1985, p.124).
The New Zealand and Australian populations are tabulated in Table 7. Workforce and income‑force data for each census year are summarised in Table 8, but with Australian percentage data transposed onto the New Zealand population for 1933.
During the depression, many people took in one or two girl boarders to help eke out their small income. Conditions have now improved and those women have had to find fresh quarters.
Because a [single] woman is forced by circumstances into the labour market, it should not follow ... she has no right to have a job of her own ... [for which] she has as much need as her sister with home duties.
(Woman Today, November 1937, p.180; & letter, December 1937, p.219)
During the Great Depression when more women were seeking work, they were told to stay home and leave the 'jobs to the boys'.
(Bacchi 1974, pp.421-422)
On account of the clear evidence for additional female workforce participants, a best estimate of the workforce requires further allocation of "dependent" women into the workforce category.
Women's Unemployment Committees maintained an unemployment register for women in the four main New Zealand cities in the latter years of the depression. The totals of females on this register were given in the reports of the Unemployment Board (NZAJHR 1933-36; H-35). The coverage was very low - covering only women with an occupation who lived in the main centres and who thought it might be to their advantage to avail themselves of the Committees' job‑finding services - but the data do give some idea on changes to female unemployment over time, and changes in the age composition of the female unemployed. The numbers of females registered as unemployed are tabulated below in Table 9.
These figures suggest that the female unemployment problem eased considerably after 1933. The fall in teenage job-seekers from 1933 to 1935 reflects a big rise in factory employment for young females (Rankin 1990, p.169). It is safe to conclude that female unemployment was much higher in 1933 than at the time of the 1936 census. The data given in Table 9 for 1936 shows that, as the recovery progressed, unemployed females were more likely to be in older age groups. This apparent rise in the labour supply of adult women probably represents a reduction of discouraged workers. By July 1936, it was becoming more worthwhile for women with a lower priority for assistance to seek training or help in finding work.
There was discrimination against employing women during and after the depression (Rankin 1990, ch.3). Furthermore, women had fewer formal skills. Therefore, unemployment rates ought to have been higher for females than for males. According to the Australian census, a greater proportion of female than male breadwinners in their thirties and forties had no incomes. It is in these age groups where the distinction between employment and dependence is most clearly differentiated. These age groups should have the least hidden unemployment. There are fewer alternative categories such as "at school", "at home waiting to get married", or "retired" for people in their forties. (Of course the alternative "married woman" remains.) With women in their forties having higher measured rates of unemployment than men in their forties, the problem of higher unemployment must have affected women of all ages.
There was much concern about male youths becoming unemployable (eg Report on Juvenile Unemployment; NZAJHR H-35B, 1933), a concern not applied to females. So, with providing employment for young males having a higher social priority than employing teenage females, and with there being only a smaller pay differential between the sexes for minors than for adults, the probability of finding formal employment must have been significantly lower for females than males. This may have been exacerbated by the fact that female employees remained protected by compulsory arbitration procedures which had been abandoned for males between May 1932 and May 1936 (Unwin, pp.96‑97). This provision in the Industrial, Conciliation and Arbitration (ICA) Act reduced the opportunity cost of employing males vis-à-vis females.
Raw census data for young people show a much smaller proportion of females than males in the labour force who were without incomes. Thus it is in the younger age groups that most of the workforce allocations have been made. For people under 21, the wage differences between males and females were much less than for adults, as figure 5(b) shows. Neither youths nor females were eligible for a "family wage", although wages for young males were graded towards the adult wage on an age basis. The corrected teenage workforce for 1933 and 1935 should yield an average female income as being of the same ratio to the average male income as in 1925, for each age group.
Not all young women had a greater need to work in the 1930s than the 1920s. Where fathers remained in fulltime employment and experienced a rise in real earnings, it could be that the added‑worker effect was working in reverse (Humphrey 1940, p.419). However, there is no evidence of a voluntary withdrawal from the workforce by middle-class females in the depression. Young middle-class women had been taking advantage of rising opportunities for office work in the 1920s. In the 1930s, with many men and women having to downgrade their job expectations, there was less social stigma from accepting low-status employment. Indeed, keeping a daughter at home during the depression might have led people to believe that she could not get a job. Thus, it is unlikely that many comparatively privileged "girls" or women actually withdrew from the workforce during the depression, although those who did lose their white-collar jobs may have been more choosy than many in accepting their next job. Such young women tried to get the best jobs through their superior family connections (Findlay 1989, p.32; In Those Days 1982, p.35).
Table 10 summarises the reallocations from dependent to workforce status which were made for females on an age-group basis. This is an alternative exercise to that summarised in Table 4B, except that income data, the added-worker effect and the stated intentions of school leavers (Rankin 1990, ch.3) were taken into account. The final allocations have been made to achieve consistency between males and females, and to achieve consistency between Australia and New Zealand (as is appropriate because the two countries' 1930s' income data were comparable in the more stable income and middle‑age groups). Thus a best-estimate of the New Zealand workforce in 1936 will be structurally consistent - for both sexes, for all age groups and for all income groups - with a best-estimate of the Australian workforce in 1933, allowing the revised Australian participation rates to be applied to New Zealand.
Table 11 summarises the different added-worker elasticities (Rankin 1990, p.29), which vary in accordance with the workforce estimates. We would expect a slightly stronger added worker effect in 1933 than in 1936. The elasticities with respect to workforce measures (b) and (c) in Table 2 show this appropriate relativity between the years 1933 and 1936. Measure (c) reflects estimates that fully accounted for the effect in both years. The income-force elasticities are lower, especially in 1933, in that they do not make any allowance for the many unemployed added workers who had no incomes. We can say that for a ten percent fall in real average male incomes, 21/2 to 71/2 percent of females aged from 14 to 59 were added to the New Zealand workforce. For the estimated 18% fall in average male incomes from 1925 to 1933, up to 13 percent of working age women who would not otherwise have done so were supplying labour to the market in the middle of the Depression.
This paper utilises the Australian Census of 1933 and the New Zealand Censuses of 1926 and 1936 to estimate New Zealand's employment, workforce participation and unemployment rates during the mid‑1933 trough of the Great Depression. Income data for females has been used to reclassify many women - especially young women - from "dependent" status into an unemployment or "residual workforce" category, which includes the fully and partially unemployed, those working without pay in family businesses, and those who did not specify an employment status.
The alternative workforce and unemployment estimates, summarised in Table 2, are not as precise as they are presented. They are, nevertheless, meaningful estimates which make full sense in the light of the research undertaken which shows adequate comparability between Australia and New Zealand and which reveals a greater level of female activity relative to male labour market activity.
Census income data give a more complete picture of the extent and diversity of labour market activity than do data on gainful employment, which show sharp divisions where there are none, and tend to reflect the expectations built into the census design. In trying to understand the Depression, it can be useful to focus on the income-force in its own right, since income is more easily defined than unemployment. The income approach emphasises economic activity over a whole year rather than at a single point in time, revealing many of the shades of grey in the labour market. It reminds us that income constitutes the motivation for labour supply, and that employment is an economic sacrifice. After all, households are income-maximisers and work-minimisers. Households are ambivalent towards employment, which is a means to an end - income - and not an end in itself. We need incomes to make the non‑labour‑market activities that really matter to us possible to undertake. Such activities are not residual, as the labour force approach implies.
The major conclusions of this paper are that, in the Great Depression (i) the working age population cannot be easily divided into the employed, the unemployed and non-participants, (ii) a large proportion of the mid-depression workforce was not formally employed in the sense of receiving a regular wage or a viable income, and (iii) that there was a significant addition of females to the New Zealand workforce in the 1930s.
The introduction to this chapter quoted Elizabeth McCombs, New Zealand's first woman MP, as claiming that the unemployed and underemployed totalled about 240,000. Was she correct? In estimating a total of 224,000 members in the non-Maori workforce without employment or with grossly inadequate employment, this paper supports her claim that there were 240,000 fully or partially unemployed people in the New Zealand workforce in 1933. Furthermore, there were 240,000 people with incomes above zero, but less than £1 per week, at a time when an adequate living wage was in excess of £4 per week.
________ Census of New Zealand: 1921, 1926, 1936, 1945, 1991.
________ Census of the Commonwealth of Australia: 1921, 1933.
________(1982) In Those Days, Wellington Branch Society for Research on Women
ABLMR (1985) Who's in the Labour Force? A Study of Labour Force Participation; Research Report No. 7, Australian Bureau of Labour Market Research.
Archer, T.M. (1934) Youth Unemployment in New Zealand , Canterbury M.Com thesis.
Bacchi, C. (1986) "The 'Woman Question' in South Australia"; in Richards (ed.)
Boston, S. (1987) Women Workers and the Trade Unions (1st edn.1980) London, Davis-Poynter.
Broomhill, R (1974) "Underemployment in Adelaide during the Depression"; Labour History 27:31-40.
Broomhill, R (1978) Unemployed Workers in Adelaide during the Depression; St. Lucia, Queensland University Press.
Brosnan, P. & Wilson, M. (1989) "The Historical Structuring of the Labour Market", Working Paper 4/89, Industrial Relations Centre, Victoria University of Wellington
Campbell, R.M (1927) "Family Allowances in New Zealand", Economic Journal 37:369-383.
Chapple, S. (1994) "How Great was the Depression in New Zealand? Neglected Estimates of Inter-War Aggregate Income", NZ Economic Papers 28(2):195-203.
Condliffe, J.B. (1959) The Welfare State in New Zealand; London, Allen & Unwin.
Costar, B.J. (1974) "The Great Depression: Was Queensland Different?" Labour Hist. 26:32-48.
Findlay, M. (1989) Tooth and Nail; the Story of a Daughter of the Depression, Auckland, Penguin (originally published: Wellington, Reed 1974).
Fisher, A.G.B. (1929) "The Drift to the Towns"; Economic Record 5:234-252.
Fisher, A.G.B. (1930) "Distribution of Income in New Zealand", Economic Record 6:221-234.
Forster, C. (1965) "Australian Unemployment 1900-40", Economic Record 41:426-450.
Giles, G.R. (1935) "Unemployment among Young People in Australia", International Labour Review 31:811-836.
Hare, A.E.C. (1946) Report on Industrial Relations in New Zealand; Wellington, Whitcombe and Tombs.
Hawke, G.R. (1985) The Making of New Zealand; Cambridge, Cambridge University Press.
Humphrey, D. (1940) "Alleged 'Additional Workers' in the Measurement of Unemployment"; Journal of Political Economy 48:412-419.
Jaffe, A.J. (1956) "Trends in the Participation of Women in the Working Force"; Monthly Labor Review 79: 559-567.
Jones, F.L. (1987) "Occupational Statistics Revisited: The Female Labour Force in Early British and Australian Censuses"; Australian Economic History Rev. 27:56-76.
Keating, M. (1973) The Australian Workforce 1910/11 - 1960/61, Australian National University
Lundberg, S. (1985) "The Added Worker Effect", Journal of Labour Economics 3(1):11-37.
Macrae, J. & Sinclair, K. (1975) "Unemployment in New Zealand during the Depression of the late 1920s and early 1930s", Australian Economic History Review 15: 35-45.
Maddock, R. & McLean, I.W. eds. (1987) The Australian Economy in the Long Run; Cambridge University Press.
Mawer, J.M. (1982) "Female Labour Force Involvement", Waikato University M.Soc.Sci. Thesis
Neale, E.P. (1929) "The Post-War Sex- and Age-Constitution of the New Zealand Population and some of its Economic Consequences", Economic Record 5:75-82.
NSWOYB New South Wales Official Year Book
NZAJHR Appendices to the Journal of the House of Representatives, New Zealand Official Publication
NZHLFS The New Zealand Labour Force; Dept. of Statistics.
NZMS (1938) "Employment and Incomes"; Miscellaneous Statistics, NZ Govt. Statistician
NZOYB New Zealand Official Year Book , New Zealand Government Statistician
NZPD New Zealand Parliamentary Debates, New Zealand Official Publication
NZSPB New Zealand Statistics of Population and Buildings Government Statistician
OYBCA Official Year Book of the Commonwealth of Australia
OECD (1987) "Employment and Unemployment"; Economic Outlook , June 1987
Rankin, K. (1990) Labour Supply in New Zealand and Australia: 1919-1939, MA Thesis, Victoria University of Wellington.
Rankin, K. (1991) "New Zealand's Labour Supply in a Long Term Perspective", in Long Run Perspectives on the New Zealand Economy, Proceedings of the Sesquicentennial Conference of the New Zealand Association of Economists; Auckland, 20‑22 August 1990, v.2:338-359.
Rankin, K. (1992) "New Zealand's Gross National Product: 1859-1939", Review of Income and Wealth 38(1): 49-69.
Rankin, K. (1993a) "The New Zealand Workforce 1950-2000"; New Zealand Journal of Industrial Relations, 18(2): 214-235.
Rankin, K. (1993b) "Lies, Damn Lies and the 'Spirit of Recovery'", New Zealand Political Review 2(4): 9-13.
Rankin, K. (1994) "Comment: How Great was the Depression in New Zealand? Neglected Estimates of Inter-War Aggregate Income", New Zealand Economic Papers 28(2):204-209.
Revell, P. & Brosnan, P. (1986) "New Zealand labour force participation: the ninety years to 1981"; New Zealand Journal of Industrial Relations 11:77-89.
Richards, E. (1974) "Women in the British Economy since about 1700: an Interpretation", History 59: 337-357.
Richards, E. ed.(1986) Flinders History of South Australia - Social History; Netley, South Australia, Wakefield Press.
Schedvin, C.B. (1970) Australia and the Great Depression; Sydney, Sydney University Press.
Smuts, R.W. (1960) "The Female Labor Force: A Case Study in the Interpretation of Historical Statistics"; Journal of the American Statistical Assosiation.
Sutch, W.B. (1966) The Quest for Security in New Zealand 1840-1966; Wellington, Oxford University Press.
Tsokhas, K. (1988) "Work Practices and Sheet Metal Workers"; ANU Working Papers in Economic History 116.
Unwin, D. (1944) Women in New Zealand Industry; MA Thesis, University of Otago.
Westrate, C. (1956) "Unemployment in New Zealand during the Great Depression", Economic Record 32:139-141.
Woytinsky, W (1940) "Additional Workers on the Labour Market in Depressions: A Reply to Mr Humphrey", Journal of Political Economy 48:735-739.
. This paper is drawn from my MA Thesis Labour Supply in New Zealand and Australia, 1919-1939.
. Introducing a 1938 survey of incomes and employment, the NZ Government Statistician commented that "the absence of comprehensive, detailed, and up-to-date statistics of employment and incomes has long been the cause of important lacunae in the basic data required in the fields of administration and social statistics"; Employment and Incomes; NZMS1938
. Many of those living in the cities had parents on farms (Fisher, 1929). New Zealand census data shows, for young males, a big movement away from agricultural occupations in the 1920s, and an equally big shift back to the farm in the early 1930s, and then a bigger shift away from farm employment from 1936-45 (Rankin 1990, p.160). Macrae and Sinclair (1975, p.38) note a rise of 22% in employment in primary production from 1926 to 1936 of 30,000 males In Australia, however, primary sector employment fell between 1921 and 1933 (Rankin 1990, p.165)
. Costar (1974, p.36) has estimated unemployment in Australia (1931) to be well above recorded rates, with Tasmania (38.6%, 27.4% recorded) above the Australian average of 37.8% (27.4% recorded). Queensland's figures were 30.1% estimated, 16.2% recorded. Queensland was a bastion of protected agriculture. Australian trade union statistics (OYBCA 1935, pp.394-395) - the source of Costar's recorded data - show Tasmania's unemployment to be very close to Victoria's. 1933 census data confirms the accuracy of this Victoria/Tasmania relativity.
. Macrae and Sinclair note earlier claims: by Harry Holland, Leader of the Opposition in Parliament (NZPD 1933, v.235, p.1073), Sutch (1966, pp.439-441), Condliffe (1959, p.54), Westrate (1956).
. In their Table 1, the potential labour force (from which the participation rate is computed) covers 15-64 year-olds in 1926, and 16-64 year-olds in 1933. In 1926 and 1936 the population doesn't include an allocation for people whose ages were not specified. In their Table 3, male minors and females are interposed and the labour force (and by subtraction, the total employed) double‑counts male minors. (The total number of non-Maori males in the 1933 labour force is given in as 600,000, whereas the total number of such males aged 16-64 is given as 489,907!)
. In a personal communication, H.C. Berry informed me that single unemployed men (such as himself) who did not go to distant relief camps were excluded from the unemployment register. Such men preferred to take their chances within their communities. He says that there were 14 unemployed young men in a single street in working class Wellington (Holloway Road). None of them were on the unemployment register.
. By implication, this means fulltime employment, because the unemployed total includes men on part-time relief work, and is compared with 1936 census data in which "partly unemployed" is included in the unemployment total. In reality, it was the difficult-to-pin-down residual or marginal categories, rather than full unemployment, that made up the essence of the depression experience. For example, Bill Hall recalled coming out of his engineering apprenticeship in 1933: "I was lucky in that I was able to stay at home. I found the odd job, like carrying a theodolite for a surveyor, at £3 per week." (The Dominion; Wellington, 17 September 1990.)
. Australian mid-depression (1932) employment estimates (MW Butlin, in Maddock & McLean 1987, p.354) equate to an employment total of 470,000 in New Zealand, in contrast to the corrected Macrae/Sinclair total of 544,000. That points to a male employment low of no more than 350,000; 70% of the working-age population.
. Giles (1935, p.830) estimated that 60% of 16-20 year-olds in South Australia were unemployed in May 1931.
. Lundberg (1985) defines the added-worker effect as the impact on married women of their husbands' unemployment. Clearly, it is broader than that. It is the labour supply response of a household to a reduced real disposable income - whatever the reason - of the main breadwinner. It is the orthodox income effect acting in the labour market. As such, it affects all those who could work but whose normal activity would be outside the labour market. Particularly affected were women, teenagers, and men in the 60-75 age group.
. For example, Cez Blazey, former president of the New Zealand Rugby Union, met his wife in 1929 but delayed marriage until 1935 "for financial reasons" (Maggie Barry, Radio New Zealand National Program, 14 February, 1990).
. Combining Macrae & Sinclair's employment estimates with published real wage rates suggests that the majority of families were not adversely affected by the Depression and those on sticky nominal wages should have become significantly better off. It was in fact reduced hours of work available to family breadwinners that drove the added-worker effect.
. In the interwar years, there was a separate Maori census, which included information about occupations, but not about unemployment.
. As well as having higher unemployment rates generally, Maori have always had disproportionate numbers of seasonally unemployed, thanks to their disproportionately high representation in primary product processing industries. In the March 1991 census, 40% of Maori males aged 20-60 were not employed, full-time or part-time.
. University bursaries were suspended (NZOYB 1940, p.198) and only two Teachers Training Colleges were open, with a total of 439 students, compared to 1161 in 1931 (NZOYB 1935, p.146). In 1934 all Colleges were closed, further adding to labour supply.
. Remnants of the gainful worker approach were retained up to the 1986 census, the last census to invite respondants to select a single main activity, of which "unemployed" was one option. The number who said unemployed was 170,000, a figure that was cut back in accordance with another question whiched asked the unemployed if they were actively seeking work.
. Indeed, the general public - including radio talk‑back hosts - tend to confirm in their minds that the economics profession is mentally challenged when we have to explain why unemployment does not necessarily fall when employment rises.
. Note that many of those classing themselves as self-employed were in reality unemployed or underemployed casual workers, not able to earn a living wage (Broomhill, 1974).
. The "workforce" approach would, if practically possible, class all people in full-time education as outside the workforce, unless education was for them, a second choice activity.
. If married women working without wages for their self-employed husbands - esp. farmers' wives - had been included as employed, recorded female participation rates would have been much higher.
. Forster (1965), finds Australian unemployment in the 1920s to be mainly a structural problem of too many workers lacking skills. While this implies that female unemployment rates should be high because females are more likely to be unskilled, it does not necessarily apply to the depression years. Tsokhas (1988, p.38) pointed out that when work was slack, it would be cheaper to employ unskilled boys than skilled men because time was no longer at a premium.
. eg 6 months after GDP, which itself takes time to respond to a devaluation. Rankin (1993b, Figure 4, p.12) shows a 6-month lag between employment and GDP in the years 1976 to 1993.
. Such as Fisher(1930), "Distribution of Income in New Zealand" Economic Record, and Hare (1946) "Technical Appendix: The Distribution of Personal Income", in Industrial Relations in New Zealand.
. Re purchasing power parities. Campbell (1927, p.373) stated in the Economic Journal that New Zealand "prices are on a high level". In July 1926, a 2lb loaf of bread cost the same (61/2d) in Wellington and Sydney (NZOYB 1928, p.769; NSWOYB 1928, p.751), while rents on a 4 bedroom house were the same in August 1925. For larger houses rents in Sydney were cheaper than Wellington, but were very close to the New Zealand average (NZOYB 1927, p.807; NSWOYB 1928, p.750).
. New Zealand Statistics of Wage-Rates and Hours of Labour 1935, p.16
. Broomhill 1974, p.31. Although no research has been done into purchasing power parities, a lower basic wage in Australia could indicate a lower cost of living during the depression. But it could also mean that the basic industrial award wage in New Zealand was not a basic wage in the same sense that the Australian basic wage was.
. In the middle of the depression, average female factory wages were in the order of £80-£90 per annum. Many teenagers, however, were working long hours in factories and elsewhere for less than £1 per week (Archer 1934) While they may be described as having had marginal incomes, they were certainly not underemployed.
. Brosnan and Wilson (1989); Revell and Brosnan (1986); Campbell (1927). The family benefit was introduced late in 1926, a year of sharply deteriorating economic conditions and, as Brosnan and Wilson indicate, as an explicit recognition of the added worker effect. Family allowances were not included as wives' personal incomes in 1936. The introductory notes to the 1936 income census indicate that the census question stated that "allowances which are not is respect of paid services, such as housekeeping allowances received by wives from their husbands, are not to be counted by the recipient as income".
. "The tip of the iceberg", according to Macrae and Sinclair (1975, p.41). A rather small iceberg, in their case.
. Higher unemployment rates for females were also apparent in other countries despite their also having raised employment rates during the depression. In the British Parliament in 1931 it was claimed that Ministry of Labour data showed higher unemployment rates for women than men in 17 out of 25 trades. (Boston 1987, p.172.)
. A reverse added-worker effect may have been taking place in New Zealand - probably concentrated amongst "lower‑middle‑class" households - between 1921 and 1926 (see Table 3), as indeed it was in Great Britain for much of the nineteenth century (see Richards 1974).
cross-elasticity of male incomes on female participation:
Ef:m = (L'f* - Lf) / Pf / (I'm - Im) / Im
where Lf is the initial female workforce
L'f* is the new female workforce, population adjusted
Pf is the initial female working age population
Im is the initial average real male income
I'm is the new average real male income
. Perhaps even more than 13 percent. By being more liberal in interpreting 1925 income data as earnings, and by reclassifying into the workforce more "dependent" males in their teens and in their 60s, even more females would have been transferred from dependent to workforce status in 1933 and 1936. This reflects the very blurred boundary between residual unemployment and non-participation.
. Income data are not entirely objective, however, in that there are two areas where the concept of income is open to varying interpretations: the handling of income in kind, and the distinction between gross and net incomes by self-employed people (Fisher 1930, p.223).