Comments on the New Zealand Time Use Survey

By Stuart Birks (9 December 1999)

A time use survey was conducted in New Zealand for the year July 1998 to July 1999, by Statistics New Zealand under contract to the Ministry of Women's Affairs which sponsored the survey at a cost of $2million. The first results are to be published on the Ministry of Women's Affairs web site (http://www.mwa.govt.nz/main.html) on 14^th December 1999.

General Issues

Fleming R and Spellerberg A (1999) discuss the motivation for and methodology of the New Zealand time use survey. The first section of the publication is written by Fleming. She discusses the history and use of time use surveys. Spellerberg concentrates on the methodology.

Fleming indicates the motivation for time use surveys:

"It is recognized that in both the developing and the industrial world women do the greater share of unpaid work, and men do the greater share of paid work. This imbalance has been identified as a major factor in women's lower status, lack of access to resources and increasing poverty." (p.17)

The situation is not a clear cut as Fleming claims, however. "Work" here is measured in terms of hours, not necessarily value of output. It is not clear that women have lower status, less access to resources, or increasing poverty in comparison to men. Status would depend on measures used and valuations placed on the results. For example, would a greater likelihood of award of custody of children indicate lower or higher status? Would an obligation to pay child support without any accountability in the use of the money or rights to contact with the children indicate lower or higher status? Would gender balance across all government boards and committees (and aim of the Ministry of Women's Affairs by the year 2000[1]) indicate lower status for women? Access to resources depends on income and intra-family transfers. Poverty depends on both income and wealth. The "gender wage gap", measured in terms of hourly ordinary time rates of pay, is either static or narrowing over time and women's workforce participation is increasing, whereas men's is decreasing. Prue Hyman reports that women's share of total wealth in New Zealand rose "from 30.5 per cent in 1980-81 to 38 per cent in 1987-88, partly as a result of more equal sharing of property after divorce" (Hyman, 1994, pp.219-220). This is a remarkable increase. Perhaps we should consider it more as an exercising of rights obtained through marriage, though, in which case the wealth distribution figures presumably significantly understate women's total claims on assets. For example, a married person's legal claim to half of their partner's superannuation should be taken into account by an reduction in the measured wealth of the one with superannuation and an increase in the wealth of the other.

A more recent study by Infometrics, using 1997 data, found that women’s average wealth was 86% of men’s[2]. This figure also does not make any adjustment to consider unrealised claims on partners’ assets through marriage, as with superannuation. Such adjustment would further reduce the gender difference[3].

Fleming proceeds to quote from the 1995 UN Human Development Report that, "a major index of neglect is that many of women's economic contributions are grossly undervalued or not valued at all" (p.17). Presumably this is a reference to inclusion in national accounts, because intra-family transfers, state benefits, public provision of goods and services, and matrimonial property legislation all indicate a high valuation of women and their contributions. These forms of recognition have far more substance than an economic measure which was designed for another purpose.

She contends that, "time use data provides empirical evidence of unequal sharing within households and the barriers to participation in paid work presented by caring responsibilities" (p.17). In fact this time use survey may distort and overstate women's contribution relative to men, it understates their financial position relative to men, and it could equally be argued that society's expectation that men provide financially for their families acts as a barrier to their participation in family life. Similarly, separation, more commonly the woman’s idea[4], and the associated non-custodial status of most of the fathers[5], is a barrier to fathers’ caregiving activity.

She also claims that time use data is of value because, inter alia, "consistency of data collection allows comparisons to be made between … different groups within the same population" (p.16). This is not the case if groups are poorly identified, and if groups are not treated identically in the survey. Both these problems apply with the current survey for certain groups, as illustrated below.

She claims further that, "Married women are increasingly involved in paid work, but the evidence is that men are taking up unpaid work only slowly or, in some countries, hardly at all, resulting in a double burden for working women" (p.16). One New Zealand study (Fletcher, 1978) found that full-time homemakers put in less time in total than their working husbands, and the increased hours (paid and unpaid) of women in paid work were essentially a catch-up to the hours put by men in full-time work. Given that women do 60-65 percent of the unpaid work (p.36), and men do the majority of paid work, it would be more accurate to say that both men and women have a "double burden".[6]

The Ministry of Women’s Affairs claims that its priority areas of work include: “measuring women's contribution to the economy and community through unpaid work” (my emphasis, from http://www.mwa.govt.nz/work/priority.html).

As Spellerberg points out in the development of the survey, "the needs of sponsors … were obviously paramount" (Fleming and Spellerberg, 1999, p.57).

It would appear that there are distortions in perception of the issues even before the data are analysed. These may have a significant impact on the questions asked and interpretation of results.

Specific Concerns

The results of a study can depend on the methodology used. This includes the definitions used and the groupings made.

“Household” or “family”?

This survey uses the household as a defining criterion, rather than the family. In particular, there is a distinction drawn between caring for children who live in the same household and children who do not live in the same household. Note that there are a large number of families where the parents do not live together in the same household, as indicated by the more than 200,000 parents who are paying child support[7]. According to the definitions, non-custodial parents do not live in the same household as their children. Their time spent caring for their children is therefore classified differently. As the criterion is simply one of household (“unpaid work for own household: looking after a child who lives in the same household as you”), the relationship of caregiver to child is not identified. It is therefore not possible to determine whether a caregiver in the same household as the child is a parent, a step parent, someone in the parenting role, or another adult (older sibling, cousin, aunt, or uncle, say) living in the household. Similarly a non-custodial parent’s time with a child would be classified as “informal unpaid work outside the home: looking after a child who does not live in the same household as you”.

What is “caregiving”?

The “looking after” category only applies for children under 14 years old. For children older than that, the only activities considered are, “coaching, training, teaching, or helping with schoolwork, etc.”, unless the person needs special care because of illness or disability. It is not clear that this covers the wide range of parenting activities described in Birks (1999), as distinct from the narrow “nurturing” and traditional-female-focused definitions of parenting used in such places as the Family Court.

Simultaneous activities

People can be engaged in several activities at the same time, as when cooking a meal while listening to the radio and keeping an eye on children. Spellerberg states that, “if only main or primary activities are recorded, information will be lost on childcare (and passive leisure activities)” (Fleming and Spellerberg, p.51). To measure time use, it is important to identify all the activities being undertaken at any time, but it does reveal a problem.

In time use surveys, time spent on an activity is used as a measure of input, which is then equated to output of that activity. Can you be as productive in each of two or more activities conducted together as you would if you spent the same amount of time on one alone? If you can, then time spent does not reflect the opportunity cost of the activity because it does not preclude other activities. If you cannot, then time spent on one activity does not reflect output in that activity unless allowance is made for any other activities undertaken at the same time.

While there may be problems with the concept of a main, or primary activity, it can be useful to identify what might be considered the most significant of several activities being undertaken at the same time. It might be imagined that such an activity would be the one that took up most time, attention, or effort, or was considered the most important by the person concerned. The survey uses a different definition of primary activity:

“Primary activity is not determined by the respondent in that the diary does not allow identification of primary and secondary activities. Rather, the order has been determined by the MWA [Ministry of Women’s Affairs] in conjunction with SNZ [Statistics New Zealand].” (Guide p.21)

This is quite amazing. It means that the MWA determined the order of priorities for the various possible activities, with the primary activity being defined as whichever of the activities undertaken is highest on the list. The list is at the “second level classification” of activities, giving twelve groups in the following ranking from most important downwards: personal care; labour force activity; education and training; caregiving for household members; household work; purchasing goods and services for own household; unpaid work outside of the home; religious, cultural and civic participation; social entertainment; sports and hobbies; mass media and free time activities; residual.

It is notable that the ranking fits closely with the activity classification which was designed in part to maintain consistency with international classification used overseas. The only difference is that the ranking reverses the order of “caregiving for household members” and “household work”, putting the former ahead of the latter.

The approach leads to some anomalous results. One subcategory of caregiving for household members is, “available for care of household members”. It appears that this would rank above actually undertaking household work. The primary activity of someone doing household work while a child sleeps is caregiving. However, consider someone working from home and actively involved in caring for children in the household at the same time. That would give a primary activity of labour force activity. Even more strange, a non-custodial parent actively caring for a child while undertaking household work would have a primary activity of household work. Caring for the child does not count as caring for a household member because the child does not live in the same household (even if the child is in that parent’s sole care for anything up to 145 nights in a year, given the New Zealand Child Support Act definition of non-custodial). Instead, the caring is classified as, “unpaid work outside the home (informal), caring for non-household members”. Unlike the case with household members, there is no sub-category to allow for a non-custodial parent who is “available for care”.

Grouping and variability within groups

Ideally for analysis, there should be homogeneity within groups (group members should be identical in relevant characteristics). If there is much variability within a group, then average group characteristics will not accurately reflect individual characteristics, so results will imprecise and could be misleading.

As has already been mentioned, people are grouped by household rather than by family, making it difficult to identify whether someone is a parent or other adult. Non-custodial parents also cannot be identified. There could be much variability within groups on this basis.

Households in the study can be classified by "family type". Family type is defined on page 34 of Statistics New Zealand (1999). Categories are: couple without children; couple with child(ren), some or all aged less than 18 years; couple with child(ren), all aged 18 years or more; one parent with child(ren), some or all aged less than 18 years; one parent with child(ren), all aged 18 years or more; non-family.

It is not clear why 18 is used, when 14 appears to be a crucial age in the survey. There is no identification of relationship between the couple and the children, so one could be a step parent or even short-term live-in partner. With "some or all" children aged under 18, there may be several young children, and there may or may not be older children able to care for them. An individual's unpaid work in the home may be highly dependent on the ages of all members of the household. Even in "couple" households, one or both might have repartnered, with other family members or non-custodial parents in other households. One parent households include never-married, separated, divorced and widowed parents. There may be large differences in the circumstances of these.

Data are given as average time in an activity either for the whole population, or for participants in the activity, where the population is people aged 12 and over. It may be that those aged 12-14, say, may be unrepresentative of the population as a whole.

Unfortunately the time use survey published tables give average times, with no measures of variability of time use among members of the group.

Measurement of income

The questions on income follow a similar pattern to the census, with the same problems, as discussed in appendices 4 and 5 of Birks (1998). Namely, they do not allow for child support paid, or tax on child support, there is no mention that child support is tax free to the recipient. The study is likely to understate custodial parents’ incomes and overstate the incomes of non-custodial parents, while failing to pick up the parenting input of non-custodial parents and associated reduction in parenting obligations of custodial parents. Moreover, the income measured is that of the surveyed individual, not of the household, and wealth and lump sum transfers are not identified. The latter can be important in that, for example, a separated spouse may receive a lump sum payment in place of an entitlement in an ex-partner's superannuation, thus receiving the income in advance without it being counted as income. For the other partner, when the superannuation is received, will be recorded as having the full amount as income with no account being taken of the payout made.

Valuing unpaid work

Fleming gives results from a 1992 Department of Statistics publication in which the value of unpaid work was estimated as a percentage of GDP (Fleming and Spellerberg, 1999, p.36). The lowest estimate is based on valuing unpaid work at the legal minimum wage, with the highest estimate using the average ordinary time wage. These figures are used as estimates of the opportunity cost of the time spent on unpaid work. Using legal minimum wage, unpaid work was valued at 29% of GDP, and with average ordinary time wage the figure was 66% of GDP.

It may be inappropriate to assume that individuals had the alternative option of paid work for the time spent on unpaid work, and there is the general problem of valuing all units at the value of the marginal unit. Nevertheless, these are fairly standard assumptions. There is another problem, though. Given that unpaid work primarily benefits the individuals and their households or families, it might be more appropriate to consider the apparent opportunity cost to them, namely after-tax income foregone. If we adjust for taxes at average rates of say 20% for the minimum wage and 25% for the average wage, then the percentages fall from 29 and 66 to 23.2 and 49.5.

Fleming states that women's share of unpaid work was 65% on the minimum wage valuation, and 60% on the average wage. Statistics New Zealand provides data on GDP and on compensation of employees. The latter would understate the value of paid work because it omits earnings of the self-employed. From these data, compensation of employees in 1997/8 is about 45% of GDP. In Birks (1994) I estimated that women contributed about half as much as men in terms of before tax earnings.

Table 1 gives estimates of men's and women's paid and unpaid work contributions as a percentage of GDP on this basis.

Note that labour costs are only part of a country’s GDP. Other components include a return to capital, for example, recognizing that inputs other than labour are needed for production. As it would be inappropriate to measure GDP purely in terms of payments to labour, similarly it is inappropriate to measure the value of goods and services provided through unpaid work simply on the basis of an estimated cost of the labour used. A more consistent approach would recognise the value of the capital input into production. Perhaps it is not only unpaid labour, but also unpaid capital, that goes to produce unpaid outputs in the home and elsewhere.

Table 1: Value of work contributions by gender (% of GDP)

In other words, not only are there problems with the measurement of unpaid work, but also there are difficulties with the further analysis which attempts to value the output from this work.

What we cannot say with the survey 

In June 1997 the Ministry of Women's Affairs claimed that[8]:

the information revealed by a time use survey would assist our understanding of the following:

• the proportion of time people spend in paid work, compared with the time in unpaid work such as caring for the sick, elderly or disabled and childcare
• the amount of time involved in voluntary work, for both groups and individuals, to identify the volume of unpaid voluntary work in the health sector, the labour market and other areas of the community
• the amount of time voluntary workers put into providing support services for people involved in early release programmes, as well as time volunteers, families and friends put into supporting people who are in prison or otherwise detained. This information would assist improved design and delivery of services in the justice sector
• the time people spend recovering from illness which cannot be classified as accidents, for example the effects of stress and overwork
• the range of activities of people in particular population groups, of interest to a wide range of organisations
• the geographical location of people at particular times of the day, of interest to civil defence planners
• information on people's participation in activities such as shopping, use of leisure facilities and cultural events
• identifying whether people usually combine education, training and skill development activities or whether the typical pattern is to pursue each separately

The classifications used in the survey are such that it would be difficult to get any precise information on any of these issues. "Proportions of time" or "amounts of time" cannot be identified accurately because of simultaneous activities and an imposed ranking of activities. There are few categories for voluntary work in the community. Information is not gathered on early release programmes, nor on recovery from illness such as stress and overwork. Geographical locations of a person's activities are not identified. Only selected leisure facilities are considered, with parks being a notable omission. Activities are widely grouped, hence sporting activity would be counted under "organized sport", "exercise" or "other sports and hobbies". Overall the ministry's list overstates what can actually be achieved by the current survey.

The survey is really very limited in its usefulness. Among some of the things it cannot tell us are:

• Differences in workload according to:
  • the number of adults and children in a household;
  • relationship of the adults to the children;
  • the nature of the relationship between the adults (recently married, de facto, reconstituted family…);
  • the working situation of other adults in the household (unless both were surveyed and the household identified).

·        Relative parenting contributions of custodial and non-custodial parents.

• Total time spent on caregiving.

• The amount of paid childcare undertaken.

• Household incomes.

• Whether the children in the household spend much time living in another household.

• Whether the time spent on unpaid work is productively and efficiently used.

• What the self-perceived priorities are during multiple activities, and which activities took most effort or concentration.

• The extent to which people chose their particular allocation of time, as in the amount of paid work undertaken, for example, identifying the relationship between paid and unpaid work obligations.

The use of “household” as a classifying unit and the absence of clear identifiers of family relationships seriously limits the usefulness of the information provided by the survey.

However, the most significant problem is the method of determination of “primary activity”. This arises because of people’s simultaneous engagement in several activities. The survey does not determine people’s assessment of the relative importance of these, or of their allocation of effort over them. Instead, it imposes the Ministry of Women’s Affairs’ own ranking of activities irrespective of individual preferences or effort. For some issues, the ranking of an activity depends on the family situation (intact or living apart) of the people concerned. In a survey specifically intended to give clearer information on unpaid work, it is hard to see how this approach can be justified.

References

Birks S (1994) Women, Families and Unpaid Work, Discussion Paper No.94.9, School of Applied and International Economics, Massey University

Birks S (1998) Gender Analysis and the Women’s Access to Justice Project, Issues Paper No.2, Centre for Public Policy Evaluation, Massey University

Birks S (1999) "Parenting and the Family Court: An Economist's Perspective", Chapter 8 in Birks S and Callister P (1999) Perspectives on Fathering II, Issues Paper No.6, Centre for Public Policy Evaluation, Massey University

Child Support Review Working Party (1994) Child Support Review 1994: a consultative document, Inland Revenue Department

Fleming R and Spellerberg A (1999) Using Time Use Data: A history of time use surveys and uses of time use data, Statistics New Zealand

Fletcher G J O (1978), "Division of Labour in the New Zealand Nuclear Family", New Zealand Psychologist, 7(2), pp.33-40.

Hyman P (1994) Women and Economics: a New Zealand Feminist Perspective,  Wellington: Bridget Williams Books

Maxwell G M and Robertson J P (1993) Moving Apart: A Study of the role of Family Court Counselling Services, Department of Justice

Statistics New Zealand (1999) Time Use Survey Users' Guide, Statistics New Zealand

Wheeler Y C H (1999) Women's Equal Status: "Equal Worth", Final Report: Output 4, prepared for The Australian Commonwealth/ State and New Zealand Standing Committee of Advisers for the Status of Women

http://www.dpmc.gov.au/osw/content/publications/ec_worth/index.html

[1] The Ministry of Women's Affairs has a Nominations Service, for which women can register. Names are then put forward to Ministers and departments in response to requests from them. See: http://www.mwa.govt.nz/work/nom.html

[2] In Wheeler (1999), section 5.2.

[3] Noting that there are about 6% more women than men aged 20 and over (Statistics New Zealand PCInfos database), this translates into women directly holding 47.5% of total wealth, and having actual but unrealised claims on portions of the other 52.5%. Superannuation is 18.1% of males' total wealth, and 8.1% of females' total wealth (from Wheeler, 1999, figure 5.2.3).

[4] Maxwell and Robertson (1993), table 1.4.3 on p.37

[5] In 1994, 84% of lone parents were women (The Child Support Review Working Party, 1994, p.24).

[6] See also “Valuing unpaid work” below.

[7] Inland Revenue figures for 31 August 1999 indicate 200,738 persons paying child support, in relation to about 300,000 children.

[8] http://www.mwa.govt.nz/work/whttus.html

Prepared for the web, 10 December 1999

Revised 14 December 1999