Number 3: Estimates of the costs of children in Australian families, 1993–94

This report was published by the former Department of Families, Community Services (FaCS).

1 Introduction

The purpose of this study  was to estimate the costs of children in Australian  two parent families. In the study, the costs of children were taken  to be money  expenditures—that is, the amounts  that parents actually spent  on their  children.

Estimating  the costs of children even  when  this is restricted to just parental expenditures is inherently difficult, as many items of family expenditure are often shared  among  all family members or incurred indirectly by parents. In practice, it is also likely  that there  are wide variations in the amounts  that families  spend  on children, both as family incomes vary and as the sense  of what  it is proper to spend  varies. Not surprisingly, discussions of the costs of children are often directed towards what  should be spent, as much  as to what  is spent.

As the focus of the study  was on what  families  in Australia  do spend  on their  children, an approach was required that would  make use of available information on family expenditure patterns and that would  allow  the costs of children to be calculated separately from other expenditures by their  parents.

After consideration of possible approaches for estimating the expenditure by parents on their children, a modified  version  of a methodology developed by Espenshade (1984) was used. Given detailed family expenditure  information, Espenshade’s methodology (in common  with similar  methodologies) allows  estimates of the average costs of children to be calculated for different  types  of families  and at different  levels  of aggregation (that  is, the total costs can be broken  down  into their  component expenditures).

While  every  effort was made  to develop a variation of the Espenshade methodology that was considered most valid for deriving estimates of average parental expenditures on children from survey  data on Australian  household expenditures, it should  be borne  in mind that other researchers may prefer  other  methodologies which may produce different  results.

Section  2 of the paper  outlines the methodology that was used in estimating the cost of children. Section  3 describes the data source used in the estimations and presents some analysis of the income distribution among  families. Section  4 presents the costs of children and, in Section 5, the study  is summarised and concluded.

[ Return to Top   Return to Section ]

2  Methodology

2.1   Estimating the cost of children

The methodology used in this study  to estimate the costs of children largely followed that developed by Espenshade to estimate the costs of parental expenditures on children in the United States (1984). This methodology has subsequently been  used by Lee (1988) to provide estimates of the costs of children in Australia  in 1984.

Broadly, Espenshade’s methodology estimates the cost of a child  as the difference in average expenditures between households where only a couple is present, and households where a couple and one or more children are present, given  that the households enjoy  an equivalent standard  of living.

As Espenshade points  out (1984, p. 19) the central problem in estimating the cost borne  by parents in raising  children is that it is difficult  to separate the costs of each  family member from the total costs of the household. In many instances these  costs cannot  be easily  assigned to particular individuals (for example, housing costs or electricity costs). As well, many
expenditure items that are directly for children are simply  recorded as expenditure items by their  parents (for example, food costs).

Espenshade’s solution  to this problem was to use an index  of the material standard  of living  of different  families. Such an index  allows  families  with  the same standard  of living  to be compared, even  though  their  family composition and incomes may differ. The problem, of course, is how  to determine when  families  have a similar  standard  of living—that is, how  to construct such an index. Simply  comparing incomes is not sufficient, as families  with  the same income may have widely differing  demands placed on that income and, hence, differences in their  standards of living.

Once it is possible to tell if differently constituted families  have the same standard  of living, it is reasonably straightforward to estimate the costs of their  children. For example, consider two families, the first a couple with  no children and a weekly expenditure of $500  and the second  a couple with  one child  and a weekly expenditure of $600. If both spend  15 per cent  of their total expenditure on food (or on an alternative estimator, see discussion in following section) then the difference in their  expenditures can be said to equal  the cost of the child  in the second  family—that is, $100  per week.

For the purposes of the study, this comparison was generalised in the form of two equations. The first (1) was used to estimate total household consumption expenditure, given  information on parental incomes and the number  (and ages)  of the children. The second  (2) estimated household living  standards, given  information on household consumption expenditure and the number  (and ages)  of the children.

Equation (1) varied  from that used by Espenshade (1984) in that family income was included as an aggregate term.

C = f (fY, fY2, Agei…Agen)                                               (1)

Where

C         = household consumption expenditure fY  = total household weekly income
fY2      = the square  of total household weekly income

Agei      = number  of persons of age i in household

For the second  equation (2) a functional form specified by Betson (1990) was used.

LNPF = g (LEFS, LEFS2, LNF, CKAi…CKAn)                           (2)

Where

LNPF = the logit of the proportion of household consumption expenditure spent  on food at home
LEFS  = the log of per capita  consumption

LEFS2 = LEFS squared

LNF   = the log of family size

CKAi    = number  of persons of age i in household divided  by family size

This form was preferred to that used by Espenshade (1984) as investigation indicated that this form both provided a better  fit for data and was better  able to differentiate the costs of children of different  ages. Betson’s  equation is derived from a class of equations that are referred in the literature as Working–Leser (WL) equations (Working 1943, Leser 1963). These have the general form of:

w = l (lnC, f, d)                                                                 (3)

Where

w        = the expenditure share  of a particular item (for example, food)

lnC      = the log of total expenditure f      = household size
d         = vector  of selected demographic variables (for example, number  and ages of children, number  of adults)

Equations of this form have been  widely used in estimating expenditure-based equivalence scales  (see, for example, Carlucci and Zelli 1998, Tran Nam and Whiteford 1990).

Having estimated both equations, the total  cost of a child  was calculated as follows:

  1. Assuming  values  for hY, sY and Agei…Agen, estimate the value  of C1  using  equation (1) for a couple family with  one child.
  2. By substituting this estimated value  of C1  for the couple with  one child  into equation (2), calculate the predicted value  of LNPF.
  3. Assuming  the same value  of LNPF (that  is, the same standard  of living), substitute this value into equation (2) for a couple without a child  and solve for C2.
  4. The difference between the two estimated costs (that  is, C1  – C2) is the estimated cost of a child  at income level  hY + sY.

An additional equation was required to calculate the disaggregated or component costs  of a child. This equation (4) estimated the expenditure by a family on a particular expenditure category, given  the family’s  income level  and number  and ages of their  children.

Ci= h (C, C2, Agei…Agen)                                                (4)

Where

Ci           = Household  consumption expenditure on item i
C         = Total household consumption expenditure

C2       = C squared

Agei      = number  of persons of age i in household

Component costs were calculated as:

  1. Assuming  values  for hY, sY and Agei…Agen, the consumption of a couple family with  one child  at income level  hY+sY was estimated as C1  using  equation (1).
  2. By substituting the value  of C1  into equation (2), calculate LNPF as the predicted standard  of living  of a couple family with  one child  at income level  hY+sY.
  3. Assuming  the same value  of LNPF, substitute in equation (2) and estimate C2  as the consumption of a couple family with  no children at this standard  of living.
  4. Estimate, using  equation (4), expenditure on category i at consumption levels  C1  and C2, for couples with  and without children (as Ei1 and Ei2 respectively).
  5. The difference between Ei1 and Ei2 is thus the estimated cost of a child  for expenditure category i at income level  hY + sY .

(Using this methodology it is possible that some families  will  exhibit ‘negative’ expenditure on particular items. This would  occur  where families  with  children spend, on average, less on a particular item than do comparable families  without children.)

2.2   Alternative estimators of comparable living standards

1. Engel Method

As noted, the Espenshade methodology used in this study  to calculate the cost of children relies on being  able to determine when  families  of different  sizes and with  different  incomes have similar  living  standards. In his earlier work  on costs of children, Espenshade examined several different  methods that had been  used to determine family living  standards (Espenshade 1972, pp. 63–74). These included per capita  income, level  of adult expenditure, proportion of income saved and proportion of income spent  on food.

After considering each, Espenshade concluded that the most appropriate measure to use was the proportion of family income spent  on food (the  so-called ‘Engel estimator’). Espenshade, following Ernst Engel, accepted the assumption that if the share  of expenditure devoted to the consumption of food at home was the same for two families, then they  had the same standard of living. The justification for this choice was held to be the substantial support that has been found for the observation first made  by Engel in 1857  that the poorer  a family, the greater the proportion of their  income that will  be spent  on food (‘Engel’s  Law’). Further, subsequent research found that the proportion of income spent  on food continues to decline with  income
when  the number  and ages of children in the family are held constant (Espenshade 1972, p. 71).

Figure 1 shows  explicitly how  this information is then used to calculate child  costs. At expenditure level  C1, the food share  of total expenditure of a couple family with  a single  child (Family  1) would  be PF1 (point  A1). A couple family without children (Family  2) would  have the same food share  at expenditure level  of C2 (point  A2). Thus, the methodology suggests that the cost of the child  in Family 1 is C1 – C2.

Clearly, if the assumptions underlying this method  are reasonable and can be accepted, it has several  advantages. These have been  summarised by van der Gaag (nd)  as, (1) the inherent plausibility of the measure given  that food needs  are likely  to have first call on the incomes of most families  and, (2), it is a measure that is easy to estimate (p. 10).

Figure 1: Using Engel estimators to calculate the cost of children

Figure 1: Using Engel estimators to calculate the cost of children

However, the assumptions underlying the use of an Engel estimator to calculate the cost of children have attracted a range  of criticisms. One is that the main argument used in its favour (that  is, basic  needs  are met first from a family’s  income) should  logically see it extended to also include other  basic  necessities, such as housing, clothing and transportation (van der Gaag (nd), p. 11). Arguably, doing  so should  better  establish the link between relative household consumption and relative well  being. However, as van der Gaag goes on to note, this still presents problems in advanced economic societies (such  as Australia)  where basic  necessities may arguably be seen  to extend to goods that fulfil social  rather  than physical needs.

Perhaps  the most trenchant criticism of the method  has come  from Deaton and Muellbauer (1986, p. 741)  who  assert  that the Engel method  is fundamentally incorrect in assuming that food share  indicates the welfare level  of households of different  sizes. They further  argue  that it can be theoretically shown  that the Engel method  will  produce estimates higher  than the ‘true’ cost of children.

Since the publication of the paper  by Deaton and Muellbauer, the Engel approach has often been described as establishing an upper  bound  to child  cost estimates. However, there  are grounds for viewing this widely held assertion with  some caution. In particular, as Bradbury rightly points  out, the argument that the Engel approach establishes an upper  bound  to child costs relies  on the assumption that food constitutes a relatively high proportion of total child expenditures. However, as this would  more typically be the case in developing countries (which are the focus of Deaton and Muellbauer’s analysis), in more developed countries such as Australia, such a conclusion is less certain (Bradbury 1994, p. 2).

What is perhaps more likely  to be correct is that an Engel estimator will  overestimate the costs of larger  families. This is because there  is likely  to be greater economies of scale  present in a household’s non-food consumption than there  is in their  food consumption.

Whether the Engel methodology establishes an upper-bound to child  costs or not, the problem remains that its acceptance hinges  on whether a sufficient relationship can be established between a family’s  food share  and their  standard  of living. Unfortunately, while a family’s  food share  can be measured, there  is no empirical measure of living  standards. Instead, proxies for living  standards (such  as food share)  are used and these  have to be justified on theoretical grounds prior to their  use. Confidence in them thus relies  on the extent to which their theoretical underpinning can be accepted and the extent to which the outcomes they  produce accord  with  what  is more generally known about family expenditure behaviour.

2. Rothbarth Method

One of the criticisms of the use of the Engel estimator in estimating the cost of children holds that the best measure should  be one that is related to changes in the consumption of parents. Essentially, the cost of a child  is the variation in the previous consumption of their  parents, all else having  remained the same. This suggests a measure that estimates the proportion of family income spent  on ‘adult’ goods (Williams, Price and Venohr 1993, p. 10). Such a method  has become known as the Rothbarth  method, after a 1943  paper  by Erwin Rothbarth
(Rothbarth 1943).

Practically, the Rothbarth  method  is very similar  to the Engel method. The difference is that the measure used to establish comparable household living  standards is the level  of expenditure on a particular good rather  than its share of total expenditure.

Figure 2 shows  how  expenditure on the adult good(s) is used to calculate child  costs.

In this instance, the expenditure curves  rise with  the assumption that, as family expenditure increases, so will  their  expenditure on adult goods. As well, it can be seen  that the expenditure curve  for the family with  no children lies above the curve  for the family with  children. This reflects the method’s assumption that the additional costs associated with  the presence of children will  result  in less expenditure by adults  on goods that are solely  for their  consumption. At expenditure level  C1, the expenditure of a couple family with  a single  child  (Family  1) on adult goods would  be E1 (point  A1). A couple family without children (Family  2) would  have the same expenditure on adult goods at a level  of C2 (point  A2). Again, the cost of the child  in Family 1 is estimated as C1 – C2.

In their  analysis, Deaton and Muellbauer (1986) provide  only indirect support for the Rothbarth method. That is, they  first suggest that the method  that should  produce estimates closest  to the ‘true’ costs of children is the Gorman-Barten  model1. However, they  go on to concede that this is not a method  that is easily  implemented—a conclusion supported by Betson after an attempt to do so (Betson  1990, p. 56). Accordingly, they  conclude that a method  that produces results closest  to the Gorman-Barten  model  should  be used. This, they  argue, is the Rothbarth  method.

Figure 2: Using Rothbarth estimators to calculate the cost of children

Figure 2: Using Rothbarth estimators to calculate the cost of children

However, despite such support for its use, the Rothbarth  method  has also attracted its share  of criticism. Even while advocating its use, Deaton and Muellbauer point out that it relies  on the assumption that the consumption preferences of couples are unaffected by the arrival  of children. If, instead, their  preferences shift either  away  from or towards adult goods, the Rothbarth  method  will  produce biased  estimates of the costs of children. Their conclusion? That the ‘true’ costs of children would  most likely  lie somewhere between the results  produced by the Engel and the Rothbarth  methods.

Rothbarth’s method  has been  further  criticised on the grounds that it relies  on the assumption that all household goods are private (ie., all are either  adult or child  goods). Given that some goods are jointly  consumed, Bradbury  shows  that, theoretically, it will  most likely  overestimate the cost of children (Bradbury 1997, p. 76)

Finally, an ‘adult’ goods approach, while having  some intrinsic appeal, would  seem  to present several  practical disadvantages. First, definitions of ‘adult’ goods commonly include ‘sin’ goods, such as alcohol  and tobacco, and these  are known to be subject to significant reporting inaccuracies (see Wright  and Dolan, 1992  for a discussion on this issue  in an Australian context). Second, adult clothing would  be a relatively infrequent purchase for many people and expenditure information is typically only collected over a short period. In Australia, for example, the most comprehensive source of information on family expenditure is the Australian  Bureau of Statistics’ Household  Expenditure Survey  and it only records clothing expenditure over a two- week period. As a result, many families  are likely  to be shown  to have no expenditure for adult clothing.

Given this uncertainty, this study  has used the Espenshade approach. However, the estimator of comparable living  standards used was a variation on the Engel method, with  the food-at-home share  estimator being  expanded to include other  basic  expenditure items (such  an estimator is sometimes referred to as an ISO-PROP estimator, following Watts 1977) (Estimated  costs of children were also produced using  alternative estimators. These are presented in appendix B).

The basic  goods estimator was preferred as investigation found that it was able to provide  a better  fit for the data and its use answers, at least in part, one of the more telling  criticisms of the Engel estimator: that it imposes the relative lack of economies of scale  that exist  in food consumption on all other  items of non-food consumption. By using  the basic  goods estimators at least some account is taken  of household economies of scale.

Two basic  goods estimators were examined. These are listed  in Table 1. The expenditure items included in each ‘basket’ were selected to meet  two criteria:

  1. that each  item be capable of being  readily observed as a household necessity;
  2. that there  should  be a mix of items which are likely  to introduce either  an ‘upward’ or a ‘downward’ bias to the estimates (as Whiteford sets out, upward biases  are likely  to result  if an item is consumed more by a child  than an adult and downward biases  where the item [such  as housing] is subject to economies of scale  [Whiteford 1985: 52]).

 

Table 1: Basic goods ‘baskets’ used in ISO-PROP estimators
Basic goods—basket 1
  • Food at home
  • Fuel and power
  • Household non-durables for use inside the home
Basic goods—basket 2
As for basket 1, plus:
  • Postal, telephone and telegram charges
  • Personal care products and services

Not included in either  basket  were the costs of housing, clothing and health. While  these  have often been  included by other  researchers when  compiling a basket  of basic  goods (see, for example, Betson 1990; Merz and Faik 1992), their  use is problematic.

Housing  was not included, as it has characteristics that set it apart from other  essential expenditures. As a recent study  (which did not include the costs of housing in any of several baskets  of basic  goods ) noted, expenditure on housing is ‘peculiar’ (Carlucci and Zelli 1998). What makes  it peculiar is that it can differ markedly between households that are similar  in every  aspect other  than their  tenure type. As well, in contrast to other ‘basket’ items, the housing costs of many families, particularly those  with  children, tend to diminish across  the lifecycle. That is, families  with  older  children are much  more likely  to have lower relative housing costs, given  that expenditure by home purchasers will, over time, be reduced by the effects  of inflation  and, as well, will  increasingly go to paying off the capital component of their mortgage. For these  reasons, it would  seem  that the inclusion of housing in the measure of living standards is likely  to introduce significant distortions.

Clothing was excluded from the basic  basket  of goods, as analysis of the Household Expenditure Survey  indicated that, in contrast to other  basic  expenditures, its share  of total expenditure rose as household expenditure increased. This mirrors  the finding  of Carlucci and Zelli (1998) who  categorised clothing as a ‘comfort’ good, rather  than a ‘necessity’ good. (Some clothing is, of course, a necessity. However, it is also in many instances very much  a luxury item. Unfortunately, when  compared to food, it is more difficult  to separate out what  is basic expenditure on clothing from what  is luxury expenditure. There would  also appear to be more scope  for wealthier households to increase their  expenditure on clothing than there  is to increase their  expenditure on food at home.)

Finally, household expenditure on health was not included in the basic  basket  of goods, given the role played by the Medicare system  in Australia  to meet  all or most of the basic  components of these  expenses.

2.3   Comparison with other studies

The only closely comparable Australian  study  on the costs of children using  an expenditure- based  approach appears to be that of Lee (1988). These estimates have subsequently been updated and used by the Australian  Institute  of Family Studies  to present current estimates of cost of children in Australia  (see, for example, Australian  Institute  of Family Studies  1998, p.62). Lee describes his methodology as following that of Espenshade (1984) and, thus, the broad methodology used in this study.

From the available description of Lee’s analysis it appears that the main differences between his study  and this study  lie with:

  • the data on which each  is based. Specifically, Lee used the 1983–84 Household  Expenditure Survey  and limited  his analysis to single  family households where all children were under  25 years, and where the head of the household was under  60 years  and not unemployed or out of the labour  force. This study  is based  on a more recent expenditure survey  (1993–94) and uses a slightly differently defined  sub-population for the analysis (described below);
  • the specification of the standard  of living  equation. Lee used the proportion of food at home as his living  standards estimator, while this study  uses expenditure on a selection of basic goods.
  • the treatment of family income. Lee reported his results  for three  family types  defined  by the contribution to total family income made  by the wife’s  income. The main findings  in the study, however, are reported only for a single  family type. This was done following concerns about the ability  of the Lee/Espenshade  methodology to differentiate between family types defined  by the wife’s  labour  force status (see  Appendix E for further  analysis of this issue).

Apart from these  studies, there  are several  other  studies  which have been  undertaken in Australia and which consider the costs of individuals within households based  on reported expenditures. While  these  have mostly  been  concerned with  estimating equivalence scales  that, typically, can then be applied to distributional analysis, their  purpose and methodologies are closely similar  to those  of this study  (see  Bradbury  1997; Valenzuela 1996; Bradbury  1994; Tran Nam and Whiteford 1990).

In addition, Apps and Rees (1995) have developed a model  of household production to estimate the cost of raising  children using  data from income distribution and time use surveys.

Most recently, Valenzuela (1999) has used an extended linear  expenditure system  to estimate living  standards and costs of children in Australia  between 1984  and 1993–94 for a variety  of family types  and across  different  commodity groups.

[ Return to Top   Return to Section ]

3  Data source

The data source used in this study  was the publicly released 1993–94 Household  Expenditure Survey  unit record  file (ABS 1994). This ABS survey  collected information on household expenditures, incomes and a wide  range  of other  socio-economic characteristics. In total, some 8,389  confidentialised records were included in the publicly released file, representing some 6.6 million  households.

However, only included in the analysis for this study  were couple-only households where the spouse  was aged  between 25 and 54 years  and where there  was either  no other  persons present or where there  was one or more dependent children under  18 years  of age. This restriction was imposed to reduce any age-related effects  that might  be associated with  the two different  types  of families—those with  and those  without children—having different characteristics.

Also excluded from the analysis were households:

  • whose principal source of income was from self-employment or reported as being  from ‘private sources’ (for example, from interest or superannuation);
  • with  negative or zero consumption expenditure;
  • with  negative or zero income;
  • where the ratio of household total expenditure to household total income was greater than two.

The final data set thus defined  contained 2,658  records, representing some 1.976  million families.

3.1   Distribution of incomes

The analysis began  with  an examination of the data to determine the number  and distribution of records by income ranges.

Income was defined  as total current household income—that is, income from all sources before  deductions for income tax or other  compulsory payments are made. (Excluded by the ABS from the definition of current income are amounts  such as lump-sum  receipts, windfall gains and withdrawals from savings  (ABS 1995, p. 35).)

The distribution of total household income by the number  of survey  records is shown  in Tables 2 and 3.

Most apparent from the tables  is the decline in the number  of records as income increases, particularly above $75,000, and as the number  of children per household increases. Also apparent is the smaller  number  of households with  older  children present—most probably a result  of their  being  more likely  to live in households with  non-dependent children or other relatives present (and thus be excluded from the analysis).

Table 2: Number of records by parental income and number of children under 18 years of age
  Total family incomes
  0– 24 999 (records) 25 000– 49 999 (records) 50 000– 74 999 (records) 75 000– 99 999 (records) 100 000+ (records) All (records)
Children under  18 years
0 214 277 244 117 47 899
1 103 236 152 54 20 565
2 87 353 191 61 30 722
3 46 173 90 17 14 340
4 17 43 29 4 2 95
5 3 13 5 1 0 22
6+ 0 11 3 0 1 15
All 470 1106 714 254 114 2658

Source: ABS (1994)
Note: see text  for families  included in the analysis

Table 3: Number of records by parental income and children’s ages
  Total family incomes
  0–
24 999 (records)
25 000–
49 999 (records)
50 000–
74 999 (records)
75 000–
99 999 (records)
100 000+ (records) All
(records)
Couple  only 214 277 244 117 47 899
1 or more children 0 to 4 years 125 403 151 41 19 739
1 or more children 5 to 9 years 128 445 252 67 30 922
1 or more children 10 to 14 years 86 312 205 56 34 693
1 or more children 15 to 17 years 36 135 111 32 21 335

Source: ABS (1994)
Notes: includes multiple instances where families  have children in more than one age range; see text  for families included in the analysis.

Because of this decline in the number  of records, reporting of child  costs by the number  of children in a family was restricted to one, two or three  children. (However, families  with  larger numbers of children were included in the analysis which estimated the equations.)

3.2   Definition of expenditure

In this study, expenditure has been  defined  as current household expenditure. This is total household expenditure as recorded on the Household  Expenditure Survey  less the following:

  • repayment of mortgage principal for the family home;
  • other  capital housing payments (includes additions, extensions and renovations to family home, and purchase of other  dwellings;
  • expenditure on superannuation and life insurance.

These items are excluded because they  represent saving  rather  than consumption.

However, it needs  to be acknowledged that expenditure on at least some of the above items could  be expected to be influenced by the presence of children in the family—that is, they
could  be said to have a non-discretionary element. For example, in the case of life insurance, it is quite  possible that families  with  children are more likely  to purchase these  products.

Similarly, in the case of housing, once  the decision has been  made  to become a home owner, the subsequent housing costs are not discretionary, including those  that represent a form of saving.

What would  be the effect of including items of non-current expenditure in the cost estimates? Most likely  this would  see some increase in the costs of older  children. For example, typical housing repayment patterns see, over time, an increasing proportion go towards paying-off the mortgage principal. As a result, the housing costs of older  families  (who  would  be more likely  to have older  children) could  be biased  towards greater repayment of this mortgage component. (This may, however, be offset by the methodology not directly comparing families  at the same point in their  lifecycles. That is, couples with  older  children, who  would  be older  themselves, are not compared with  older  couples without children. Instead, they  are compared with
couples of all ages without children.) As well, it is also possible that families  with  older  children could  be more likely  to be paying for home extensions and renovations.

[ Return to Top   Return to Section ]

4     Costs of children

The first step in estimating the cost of children was to use Ordinary  Least Squares  (OLS) regression to model  each  of the equations described in Section  Two. (The full set of regression equations and their  coefficients are presented in Appendix F.) These equations were then used to estimate the costs of children of different  ages in families  with  incomes between $400  and $3,100 per week (equivalent to annual  total family incomes of approximately $20,800 and $161,600, respectively) (see  Appendix C for discussion of the relationship between family incomes and expenditures).

4.1   Costs of children by age

The costs of children were first estimated for a single  child  in each  of the defined  age and income ranges.

The results  are shown  in Table 4 and in Figure 3 (see  Appendix D for equivalent estimates for 1998).

Table 4: Estimated average costs of a single child, by age of child, Australia 1993–94
Weekly household income Age of child
0 to 4 5 to 9 10 to 14 15 to 17
  ($ pw) ($ pw) ($ pw) ($ pw)
400 51 61 88 126
700 76 89 120 164
1000 102 117 151 201
1300 126 143 181 236
1600 150 169 210 269
1900 172 194 237 301
2200 194 217 263 331
2500 214 239 287 359
2800 234 260 311 385
3100 252 280 332 410

Source: ABS (1994) and authors’ calculations

The rate of this rise was found to be consistent across  different  incomes and across  the different  ages. That is, wealthier families  were found to spend  a greater amount  on their children, whatever their  ages.

As Figure 3 shows, the cost of a single  child  was found to accord  with  what  would  be most people’s prior expectations. That is, it was found to be lowest for children in the youngest age group  (0 to 4 years) and highest for children in the oldest  age group  (15 to 17 years). Similarly, the estimated costs of children were also found to rise in accordance with  rising  family incomes.

Figure 3: Estimated average costs of children,  by age of child, Australia  1993–94

Figure 3: Estimated average costs of children, by age of child, Australia 1993–94

While  there  was a steady  rise in the cost of a child  as family income increased, when  total costs were considered as a proportion of family income, there  was a fall as incomes rose. However, the rate of decline was reduced as family income rose (Table  5 and Figure 4).

Table 5: Estimated average costs of a single child as a proportion of total family income, by age of child, Australia 1993–94
Weekly household income Age of child
0 to 4 5 to 9 10 to 14 15 to 17
  (%) (%) (%) (%)
400 12.8 15.3 22.0 31.5
700 10.9 12.7 17.1 23.4
1000 10.2 11.7 15.1 20.1
1300 9.7 11.0 13.9 18.2
1600 9.4 10.6 13.1 16.8
1900 9.1 10.2 12.5 15.8
2200 8.8 9.9 12.0 15.0
2500 8.6 9.6 11.5 14.4
2800 8.4 9.3 11.1 13.8
3100 8.1 9.0 10.7 13.2

Source: ABS (1994) and authors’ calculations

For families, the costs of a child  as a proportion of their  combined income range  between a little  over 8 per cent  (for a child  aged  0 to 4 years  in a family with  an income of $3,100 per week) to just under  32 per cent  (for a child  aged  15 to 17 years  in a family with  an income of $400  per week).

Figure 4: Estimated average costs of children as a proportion of family  income, by age of child, Australia  1993–94

Figure 4: Estimated average costs of children as a proportion of family  income, by age of child, Australia  1993–94

4.2   Costs of children by number of children in household

The next  stage  of the analysis was to consider the cost of children according to the number  of children in the family. For this purpose, the estimates presented represent the costs of children averaged across  all the age ranges. (An alternative would  be to estimate average total costs for hypothetical families—for example, for a family with  one child  aged  four years  and another child aged  15, and so on.) It should  be noted  that this is not effectively an estimate of the average costs of children up to the ages of 17 years, as this would  assume  that family incomes remain  constant across  a child’s different  ages. This is, of course, not the case, as incomes typically tend to increase across  a family’s  lifecycle. To calculate such lifecycle costs would require the use of estimates of the average variations in family incomes over their  child  rearing years. Such an approach has been  adopted by Espenshade in presenting estimates of the total parental expenditure on children (Espenshade 1984).

It also should  be noted  that the under-representation of older  children in the data included in the analysis (see  Section  3.1)  may, given  the higher  costs of older  children, mean  that averaged costs presented in this section may be under-estimated.

Figure 5 and Table 6 show  the cost of children by the number  of children. Again the cost of each  child  was found to rise with  family incomes.

Figure 5: Estimated average costs of children,  by number  of children,  Australia,  1993–94

Figure 5: Estimated average costs of children,  by number  of children,  Australia,  1993–94

Table 6: Estimated average costs of children, by number of children, Australia, 1993–94
Weekly household income 1 child 2 children 3 children
  ($ pw) ($ pw) ($ pw)
400 69 135 *
700 99 186 259
1000 129 235 322
1300 157 281 380
1600 183 324 435
1900 208 365 486
2200 230 402 533
2500 251 436 577
2800 270 467 616
3100 287 495 651

Source: ABS (1994) and authors’ calculations
* not calculated2

As Table 7 shows, the cost of a single  child  amounts  on average to between 9 and 17 per cent of family income, for two children 16 to 34 per cent, and, for three  children, about 21 to
37 per cent.

Table 7: Estimated average costs of children, as a proportion of total family income, by number of children, Australia, 1993–94
Weekly household income 1 child 2 children 3 children
  (%) (%) (%)
400 17.3 33.8 *
700 14.1 26.6 37.0
1000 12.9 23.5 32.2
1300 12.1 21.6 29.2
1600 11.4 20.3 27.2
1900 10.9 19.2 25.6
2200 10.5 18.3 24.2
2500 10.0 17.4 23.1
2800 9.6 16.7 22.0
3100 9.3 16.0 21.0

Source: ABS (1994) and authors’ calculations
* not calculated

The cost of each  additional child  in families  with  up to three  children is shown  in Table 8 and Figure 6. The cost of the first child  is the greatest across  all incomes (between $69 and $287). For a family on $1,000 per week the average cost of a second  child  was approximately 82 per cent  of cost of the first child, while the cost of the third child  was approximately 67 per cent  of the first. The reduction in the average cost of each  additional child  is a result  of both the expenditure constraints and the economies of scale  that families  experience as their  size increases.

It was also found that, in proportional terms, the additional costs of each  child  diverged as incomes rose. That is, the gap between what  families  spent  on a single  child  and the additional amounts  they  spent  on subsequent children contracted as incomes rose. At incomes of $700,
the cost of the second  child  was approximately 84 per cent  of that of the first, while the cost of the third child  was approximately 74 per cent  of the first. At incomes of $3,100, the cost of the second  child  was approximately 75 per cent  of that of the first, while the cost of the third child was approximately 54 per cent  of the first. While  the reasons  behind  this trend  are not clear, it may reflect  income constraints fall as incomes rise. It may also be that what  economies of scale there  are with  respect to children rise with  income. As well, the caution noted  at the start of this section with  respect to variations in lifecycle incomes needs  to be borne  in mind when considering these  estimates.

Table 8: Estimated average additional cost of each child, Australia, 1993–94
Weekly household income Number of children
1st child
($ pw)
2nd child
($ pw)
3rd child
($ pw)
400 69 66 *
700 99 87 73
1000 129 106 87
1300 157 124 99
1600 183 141 111
1900 208 157 121
2200 230 172 131
2500 251 185 141
2800 270 197 149
3100 287 208 156

Source: ABS (1994) and authors’ calculations
* not calculated

Figure 6: Estimated average additional costs of each child, Australia,  1993–94

Figure 6: Estimated average additional costs of each child, Australia,  1993–94

4.3   Costs of children by expenditure items

In the previous sections, the costs of children were estimated as the differences in total expenditure between couple families  with  no children and those  with  one or more children. In this part of the analysis the costs of children were re-estimated as the differences in expenditure across  broad expenditure categories. The expenditure categories were:

  • current housing costs;
  • fuel and power;
  • food (consumed both at home and away  from home);
  • alcohol;
  • tobacco;
  • clothing and footwear;
  • furnishings and equipment;
  • services and operation (includes child  care  expenditure);
  • medical care  and health;
  • transport;
  • recreation and entertainment;
  • personal care;
  • miscellaneous (includes expenditure on miscellaneous goods and services such as travel goods, clocks, allowances, legal  fees; interest payments on selected credit  services; education fees; and payments for other  property, such as rates and rent).

The estimated costs of a single  child  across  each  of the expenditure categories is shown  in Table 9 (see  also, Appendix A). (As noted  in Section  2, the estimates of negative costs in Table 9 occur  as a result  of the methodology and indicate that families  with  children spend, on average, less on a particular item than do comparable families  without children.)

Several  trends  are evident. First, with  the exception of expenditure on tobacco, expenditure on all categories was found to increase with  income. However, there  was considerable variation in the extent of the rise. Increased costs were least pronounced for fuel and power, alcohol, medical care  and health, and personal care. They were most pronounced for current housing costs, furnishings and equipment, transport, and recreation and entertainment.

In terms  of children of different  ages, expenditure on food, clothing and miscellaneous items tended to increase with  the age of the child  (the  miscellaneous category includes the income sensitive expenditure item of education). Other items such as current housing costs and fuel and power were found to vary considerably less across  children of different  ages. Expenditure on household services and operations was noticeably higher  for the youngest children, as a result  of the inclusion of child  care  services in this category. In contrast, recreation and entertainment expenditure on the youngest children (0 to 4 year  olds) was much  less than that spent  on all older  children. Transport  costs were found to be significantly higher  for families with  a child  under  five years  of age and those  with  a child  over 15 years  of age.

The analysis of component costs was further  extended to consider how  expenditure on children varied  for children across  income levels. As Figure 7 shows, the component costs of children in families  on $1,000 (the  group  whose incomes are closest  to median  family incomes) are compared across  each  of the four age ranges. As a child’s age increase, there  was a corresponding increase in expenditure on miscellaneous items, recreation and entertainment (although this fell away  for the oldest  children), clothing and food. By contrast, there  were only small changes in expenditure on fuel and power and medical care  while expenditure on housing fell as children aged. Trends in expenditure on furnishings and equipment and on household services and operation were found to be more complex. Expenditure on furnishings and equipment rose across  the three  youngest age groups  and then fell noticeably for children aged  15 to 17. Expenditure on household services and operation was highest for the children under  0 to 4 years  of age, then fell progressively for children aged  5 to 9 years  and those aged  10 to 14 years  (as would  be expected due to the high child  care  costs associated with young  children), before  rising  again  for children aged  15 to 17 years.

Figure 7: Estimated component costs of a single child for families with  income of $900 per week, Australia,  1993–94

Figure 7: Estimated component costs of a single child for families with  income of $900 per week, Australia,  1993–94

The analysis also considered how  the component costs of a single  child  might  change as family incomes change. For this part of the analysis, only children aged  5 to 9 years  were selected. As Figure 8 shows, expenditure on most expenditure components was found to rise with  incomes. The greatest rises were for the items of furnishings and equipment, recreation and entertainment, transport and housing costs. Lesser, but still significant, rises were found to occur in miscellaneous expenditure, food, clothing and household services and operation. Finally, only very small differences were found for children of this age when  expenditure on medical care and health, personal care  and fuel and power was considered.

Figure 8: Estimated component costs of a single child age 5 to 9 years, Australia,  1993–94

Figure 8: Estimated component costs of a single child age 5 to 9 years, Australia, 1993–94

Table 9: Estimated average costs of a single child, by individual expenditure categories and age of child, Australia, 1993–94
  Weekly family income
Expenditure category 400 700 1000 1300 1600 1900 2200 2500 2800 3100  
1 child,  0 to 4 years
Current  housing costs 15 18 21 25 28 31 34 37 40 43 **
Fuel and power 1 2 2 2 2 2 2 3 3 3 *
Food 12 15 17 19 21 23 25 26 28 29 **
Alcohol –1 –1 –0 0 1 1 1 2 2 2  
Tobacco 0 0 0 0 0 0 0 0 0 0  
Clothing  and footwear 2 4 6 8 10 12 13 15 16 17 **
Furnishings and equipment –4 –1 2 6 10 15 20 26 31 37 **
Services and operation 13 14 15 16 17 18 19 20 21 22 **
Medical  care  and health 2 3 3 4 4 5 5 6 6 6 *
Transport 16 21 26 30 33 37 39 42 44 46 **
Recreation & entertainment –5 –0 4 8 12 15 18 21 23 25 **
Personal  care 0 0 1 1 1 1 2 2 2 2 *
Miscellaneous –2 1 4 7 9 12 14 16 18 20 **
Total 51 77 102 126 150 172 194 214 234 252 **

 

  Weekly family income
Expenditure category 400 700 1000 1300 1600 1900 2200 2500 2800 3100  
1 child,  5 to 9 years
Current  housing costs 7 10 14 17 21 24 27 31 34 37 **
Fuel and power 1 2 2 2 2 2 3 3 3 3 *
Food 16 19 21 24 26 28 30 31 33 34 **
Alcohol –2 –1 –1 –0 0 1 1 1 2 2  
Tobacco 0 0 0 0 0 0 0 –0 –0 –0  
Clothing  and footwear 6 8 11 13 15 17 18 20 21 23 **
Furnishings and equipment 0 3 6 10 15 20 26 32 38 43 **
Services and operation 7 8 9 10 11 12 13 14 15 16 **
Medical  care  and health 0 1 2 2 3 4 4 4 5 5 *
Transport 8 13 18 22 26 30 33 36 38 40 **
Recreation & entertainment 10 15 20 24 28 32 35 38 40 43 **
Personal  care 1 1 1 2 2 2 3 3 3 3 *
Miscellaneous 7 10 13 16 19 22 24 26 29 30 **
Total 61 89 117 143 169 194 217 239 260 280 **

 

  Weekly family income
Expenditure category 400 700 1000 1300 1600 1900 2200 2500 2800 3100  
1 child,  10 to 14 years
Current  housing costs 4 8 12 16 20 24 27 31 35 38 **
Fuel and power 2 2 2 3 3 3 3 3 4 4 *
Food 24 28 30 33 36 38 40 41 43 44 **
Alcohol –0 0 1 1 2 2 3 3 3 4  
Tobacco 1 1 1 1 1 1 1 1 1 1  
Clothing  and footwear 10 13 16 18 20 22 24 26 27 29 **
Furnishings and equipment 0 4 8 13 19 25 31 38 44 51 **
Services and operation –1 0 2 3 4 5 6 7 8 9 **
Medical  care  and health 2 3 3 4 5 5 6 6 7 7 *
Transport 7 13 18 23 28 31 35 38 40 43 **
Recreation & entertainment 16 22 27 32 36 40 44 47 50 52 **
Personal  care 2 2 3 3 3 4 4 4 4 4 *
Miscellaneous 20 24 27 30 34 36 39 42 44 46 **
Total 88 120 151 181 209 237 263 287 310 332 **

 

  Weekly family income
Expenditure category 400 700 1000 1300 1600 1900 2200 2500 2800 3100  
1 child, 15 to 17 years
Current housing costs 3 7 12 16 21 25 30 34 38 42 **
Fuel and power 4 5 5 5 5 5 6 6 6 6 *
Food 37 41 44 47 50 53 55 57 59 60 **
Alcohol –1 –0 0 1 1 2 2 3 3 3  
Tobacco –1 –1 –1 –1 –1 –1 –1 –1 –1 –1  
Clothing  and footwear 19 22 25 28 30 32 35 37 38 40 **
Furnishings and equipment –13 –8 –3 3 10 17 24 32 40 48 **
Services and operation 4 5 7 8 10 11 12 13 15 16 **
Medical  care  and health 5 6 7 8 9 10 10 11 11 12 *
Transport 18 25 31 36 41 46 49 53 56 58 **
Recreation & entertainment 11 17 23 29 34 39 42 46 49 52 **
Personal  care 5 5 6 6 7 7 7 7 8 8 *
Miscellaneous 36 40 44 48 52 55 58 61 64 66 **
Total 126 164 201 236 269 301 331 359 385 410 **

Note: * indicates the regression equation used to estimate the value  had an R2  > .09 and < .20
** indicates an R2  > .19, (blank  indicates R2  < .10).
Source: ABS (1994) and authors’ calculations

 

[ Return to Top   Return to Section ]

5 Summary  and conclusions

This study  has estimated the costs of children in Australian  two-parent families, where the costs of children were defined  as parental expenditures on children up to 17 years  of age. The level of expenditure was determined by comparing the expenditures of couple families  with  and without children at the same material standard  of living.

There were several  important findings  made  in the study. Not unexpectedly, these  included that the average costs of children in Australian  families  were found to vary according to the age of the child, the income level  of the parents and the number  of children in the family. More specifically:

  • The cost of a child  was found to be lowest for children in the youngest age group  (between $51 and $252  week for 0 to 4 year  olds in families  with  incomes of $400  and $3,100 per week, respectively) and highest for children in the oldest  age group  (between $126  and $410  per week for 15 to 17 year  olds in families  with  incomes of $400  and $3,100 per week, respectively).
  • While  costs of children rose in line with  rising  family incomes, at the same time they  were found to fall as proportion of income. For example, expenditure on a child  aged  between 15 and 17 years  was found to require about 23 per cent  of a family’s  income when  that income was $700  per week, falling to about 14 per cent  for families  with  incomes of $2,500 per week.
  • The average expenditure per child  was found to be greatest for families  with  one child, and was found to fall for families  with  either  two or three  children. The average cost of a first child  was found to range  between $69 and $287, according to family income, while the cost of a second  child  was between $66 and $208  per week and that of a third child  between $73 and $156  per week.

When costs of children were broken  down  into their  constituent items, it was found that expenditure on food, clothing, furnishings and equipment, recreation and entertainment, and, most of all, on miscellaneous items, tended to increase with  the age of the child. This last item included expenditure on education and was also found to increase with  income—as did most expenditures, but particularly expenditures on furnishings and equipment, current housing costs and recreation and entertainment.

In assessing the results  of the study, it must again  be stressed that the costs of children that have been  presented are averaged estimates derived from reporting of what  parents spend  to meet all households costs. Being averages, actual  expenditures on children will  often be considerably higher  or lower for particular families  than the estimates we present—that is, nothing in the study  suggests a prescriptive level  of parental expenditure. As well, as the analysis was restricted to couple households it is not possible to say what  the cost of children would  be in other  types of households (for example, those  where there  is only one adult or those where there  are more than two adults,).

Finally, while much work was undertaken to develop a variation of the Espenshade methodology that was considered most valid for deriving estimates of spending on children from household expenditure data, it should be noted  that other  methodologies exist  for estimating the costs of children. Some of these  methodologies also attempt to estimate average parental expenditures on children, while others  are concerned with  other  issues  such as what parents “need” to spend  on children. Accordingly, it is recommended that the estimates in this report  should  not be treated as the “last word” on the costs of children, and that readers also give consideration to the substantial existing literature on costs of children issues  and estimates.

[ Return to Top   Return to Section ]

Appendix A: Expenditure by category

Table A1: Expenditure by category for couple only and couple with one child aged 0 to 4 years, Australia, 1993–94
Expenditure category Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child  
  $400 pw     $700 pw     $1000     $1300     $1600    
current housing costs 16 15 1 17 16 2 18 16 2 19 17 2 20 18 2 **
fuel and power 87 71 15 101 83 18 115 93 21 128 104 25 142 114 28 *
food 93 81 12 107 92 15 119 102 17 131 112 19 142 121 21 **
alcohol 13 14 –1 16 17 –1 19 19 0 21 21 0 23 23 1  
tobacco 9 9 0 9 9 0 9 9 0 9 9 0 9 9 0  
clothing and footwear 12 10 2 23 19 4 34 28 6 44 36 8 54 44 10 **
furnishings and equipment 24 28 –4 32 33 –1 42 40 2 54 48 6 66 56 10 **
services and operation 35 22 13 40 26 14 45 30 15 50 34 16 54 37 17 **
medical care and health 27 25 2 31 28 3 34 31 3 38 34 4 41 36 4 *
transport 66 50 16 92 71 21 116 90 26 138 108 30 158 125 33 **
recreation and entertainment 38 43 –5 63 63 0 86 82 4 108 100 8 128 116 12 **
personal care 7 7 0 9 9 0 11 10 1 12 12 1 14 13 1 *
miscellaneous 15 17 –2 30 29 1 45 41 4 59 52 7 72 62 9 **
Total 442 391 51 571 494 77 694 592 102 811 685 126 922 773 150 **

 

Table A1: Expenditure by category for couple only and couple with one child aged 0 to 4 years, Australia, 1993–94 cont
Expenditure category Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child  
  $1900 pw     $2200pw     $2500     $2800     $3100    
current housing costs 21 18 2 21 19 2 22 19 3 23  20 3 23  20 3 **
fuel and power 155 124 31 167 133 34 179 142 37 191 151 40 202 159 43 *
food 152 129 23 162 137 25 170 144 26 178 151 28 185 156 29 **
alcohol 25 24 1 27 26 1 29 27 2 30 28 2 32 30 2  
tobacco 9 9 0 9 9 0 9 9 0 9 9 0 9 9 0  
clothing and footwear 62 51 12 70 57 13 78 63 15 85 69 16 91 74 17 **
furnishings and equipment 80 65 15 94 74 20 109 83 26 124 93 31 139 102 37 **
services and operation 59 40 18 63 44 19 67 47 20 70 49 21 74 52 22 **
medical care and health 44 39 5 46 41 5 48 43 6 51 44 6 53 46 6 *
transport 177 140 37 194 154 39 209 167 42 223 179 44 236 190 46 **
recreation and entertainment 146 131 15 163 145 18 178 157 21 192 169 23 205 180 25 **
personal care 15 14 1 17 15 2 18 16 2 19 17 2 20 17 2 *
miscellaneous 84 72 12 95 81 14 105 89 16 115 97 18 124 104 20 **
Total 1028 856 172 1128 934 194 1222 1007 214 1310 1076 234 1392 1140 252 **

Note: * indicates the regression equation used to estimate the value  had an R2  > .09 and < .20
** indicates an R2  > .19, (blank  indicates R2  < .10).
Source: ABS (1994) and authors’ calculations

Table A2: Expenditure by category for couple only and couple with one child aged 5 to 7 years, Australia, 1993–94
Expenditure category Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child  
  $400 pw     $700 pw     $1000     $1300     $1600    
current housing costs 16 14 1 17 15 2 18 16 2 19 17 2 20 17 2 **
fuel and power 75 68 7 89 78 10 103 89 14 116 99 17 129 109 21 *
food 93 77 16 107 88 19 120 98 21 132 108 24 143 117 26 **
alcohol 12 13 –2 15 16 –1 17 18 –1 20 20 0 22 22 0  
tobacco 9 9 0 9 9 0 9 9 0 9 9 0 9 9 0  
clothing and footwear 13 7 6 24 16 8 35 24 11 45 32 13 55 40 15 **
furnishings and equipment 26 26 0 34 31 3 43 37 6 54 44 10 67 51 15 **
services and operation 28 21 7 33 25 8 38 28 9 42 32 10 47 35 11 **
medical care and health 24 24 0 28 27 1 31 30 2 35 32 2 38 35 3 *
transport 50 42 8 76 63 13 100 82 18 122 100 22 143 117 26 **
recreation and entertainment 46 36 10 71 56 15 94 75 20 116 92 24 136 108 28 **
personal care 7 7 1 9 8 1 11 10 1 13 11 2 14 12 2 *
miscellaneous 19 12 7 35 24 10 49 36 13 63 47 16 76 57 19 **
Total 417 356 61 546 457 89 669 552 117 786 643 143 897 728 169 **

 

Table A2: Expenditure by category for couple only and couple with one child aged 5 to 7 years, Australia, 1993–94 cont.
Expenditure category Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child  
  $1900 pw     $2200pw     $2500     $2800     $3100    
current housing costs 20 18 2 21 19 3 22 19 3 22 20 3 23 20 3 **
fuel and power 142 118 24 155 127 27 167 136 31 178 144 34 189 152 37 *
food 153 125 28 162 132 30 171 139 31 179 146 33 186 152 34 **
alcohol 24 23 1 26 25 1 28 26 1 29 28 2 31 29 2  
tobacco 9 9 0 9 9 0 9 9 0 9 9 0 9 9 0  
clothing and footwear 63 47 17 72 53 18 79 59 20 86 65 21 93 70 23 **
furnishings and equipment 80 60 20 94 68 26 108 77 32 123 86 38 138 94 43 **
services and operation 51 39 12 55 42 13 59 45 14 62 47 15 66 50 16 **
medical care and health 41 37 4 43 39 4 46 41 4 48 43 5 50 45 5 *
transport 161 132 30 179 146 33 194 158 36 208 170 38 221 181 40 **
recreation and entertainment 154 122 32 171 136 35 186 149 38 200 160 40 213 171 43 **
personal care 16 13 2 17 14 3 18 15 3 19 16 3 20 17 3 *
miscellaneous 88 66 22 99 75 24 110 84 26 120 91 29 129 98 30 **
Total 1003 809 194 1103 886 217 1196 957 239 1285 1025 260 1367 1087 280 **

Note: * indicates the regression equation used to estimate the value  had an R2  > .09 and < .20
** indicates an R2  > .19, (blank  indicates R2  < .10).
Source: ABS (1994) and authors’ calculations

Table A3: Expenditure by category for couple only and couple with one child aged 10 to 14 years, Australia, 1993–94
Expenditure category Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child  
  $400 pw     $700 pw     $1000     $1300     $1600    
current housing costs 16 14 2 17 15 2 18 16 2 19 17 3 20 17 3 **
fuel and power 72 68 4 86 78 8 100 88 12 114 98 16 127 107 20 *
food 101 77 24 115 87 28 128 97 30 140 106 33 151 115 36 **
alcohol 13 13 0 16 16 0 19 18 1 21 20 1 23 21 2  
tobacco 10 9 1 10 9 1 10 9 1 10 9 1 10 9 1  
clothing and footwear 17 7 10 28 15 13 39 24 16 49 31 18 59 38 20 **
furnishings and equipment 27 26 0 35 31 4 45 36 8 56 43 13 69 50 19 **
services and operation 20 21 –1 25 25 0 30 28 2 34 32 3 39 35 4 **
medical care and health 25 24 2 29 27 3 33 29 3 36 32 4 39 35 5 *
transport 49 42 7 75 62 13 99 81 18 121 98 23 141 114 28 **
recreation and entertainment 52 36 16 77 55 22 100 73 27 121 89 32 141 105 36 **
personal care 8 7 2 10 8 2 12 9 3 14 11 3 15 12 3 *
miscellaneous 32 12 20 47 24 24 62 35 27 76 45 30 89 55 34 **
Total 442 355 88 571 451 120 694 543 151 811 631 181 923 713 209 **

 

Table A3: Expenditure by category for couple only and couple with one child aged 10 to 14 years, Australia, 1993–94 cont.
Expenditure category Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child  
  $1900 pw     $2200pw     $2500     $2800     $3100    
current housing costs 21 18 3 22 18 3 22 19 3 23 19 4 24         20 4 **
fuel and power 140 116 24 152 125 27 164 133 31 176 141 35 187        149 38 *
food 161 123 38 170 130 40 179 137 41 186 143 43 193        149 44 **
alcohol 25 23 2 27 24 3 29 26 3 30 27 3 32         28 4  
tobacco 10 9 1 10 9 1 10 9 1 10 9 1 10           9 1  
clothing and footwear 67 45 22 75 51 24 83 57 26 90 63 27 96         68 29 **
furnishings and equipment 82 58 25 97 66 31 112 74 38 127 82 44 142         91 51 **
services and operation 43 38 5 47 41 6 51 44 7 55 46 8 58         49 9 **
medical care and health 42 37 5 45 39 6 47 41 6 49 42 7 51         44 7 *
transport 160 128 31 177 142 35 192 154 38 206 166 40 219        176 43 **
recreation and entertainment 159 119 40 176 132 44 192 145 47 206 156 50 218        166 52 **
personal care 17 13 4 18 14 4 19 15 4 20 16 4 21         17 4 *
miscellaneous 101 64 36 112 73 39 123 81 42 132 88 44 141         95 46 **
Total 1028 791 237 1128 865 263 1222 935 287 1310 1000 310 1392     1060 332 **

Note: * indicates the regression equation used to estimate the value  had an R2  > .09 and < .20
** indicates an R2  > .19, (blank  indicates R2  < .10).
Source: ABS (1994) and authors’ calculations

Table A4: Expenditure by category for couple only and couple with one child aged 15 to 17 years, Australia, 1993–94
Expenditure category Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child  
  $400 pw     $700 pw     $1000     $1300     $1600    
current housing costs 19 14 4 19 15 5 20 16 5 21 16 5 22 17 5 **
fuel and power 68 65 3 82 75 7 96 84 12 110 93 16 123 102 21 *
food 111 74 37 125 84 41 138 94 44 150 103 47 161 111 50 **
alcohol 12 13 –1 15 15 0 17 17 0 20 19 1 22 20 1  
tobacco 8 9 –1 8 9 –1 8 9 –1 8 9 –1 8 9 –1  
clothing and footwear 23 5 19 35 13 22 45 21 25 56 28 28 65 35 30 **
furnishings and equipment 13 25 –13 21 29 –8 31 34 –3 43 40 3 56 46 10 **
services and operation 24 20 4 29 23 5 34 27 7 38 30 8 43 33 10 **
medical care and health 28 23 5 32 26 6 36 28 7 39 31 8 42 33 9 *
transport 55 38 18 81 57 25 105 74 31 127 91 36 147 106 41 **
recreation and entertainment 42 31 11 67 50 17 90 67 23 111 82 29 131 97 34 **
personal care 11 6 5 13 8 5 15 9 6 16 10 6 18 11 7 *
miscellaneous 45 9 36 61 21 40 76 31 44 89 41 48 102 50 52 **
Total 460 334 126 589 425 164 712 511 201 829 593 236 940 671 269 **

 

Table A4: Expenditure by category for couple only and couple with one child aged 15 to 17 years, Australia, 1993–94 cont.
Expenditure category Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child Couple with child Couple only Cost of child  
  $1900 pw     $2200pw     $2500     $2800     $3100    
current housing costs 23 17 5 24 18 6 24 18 6 25 19 6 26 19 6 **
fuel and power 136 111 25 149 119 30 161 127 34 172 134 38 183 141 42 *
food 171 118 53 180 125 55 189 132 57 196 138 59 203 143 60 **
alcohol 24 22 2 26 23 2 27 25 3 29 26 3 30 27 3
tobacco 8 9 –1 8 9 –1 8 9 –1 8 9 –1 8 9 –1
clothing and footwear 74 41 32 82 47 35 89 53 37 96 58 38 103 63 40 **
furnishings and equipment 70 53 17 85 60 24 100 67 32 115 75 40 130 82 48 **
services and operation 47 36 11 51 39 12 55 41 13 59 44 15 62 46 16 **
medical care and health 45 35 10 48 37 10 50 39 11 52 41 11 54 42 12 *
transport 165 120 46 182 133 49 198 145 53 212 156 56 224 166 58 **
recreation and entertainment 149 111 39 166 123 42 181 135 46 195 146 49 208 156 52 **
personal care 19 12 7 20 13 7 22 14 7 23 5 8 24 16 8 *
miscellaneous 114 59 55 126 67 58 136 75 61 146 82 64 155 88 66 **
Total 1046 745 301 1146 815 331 1240 881 359 1328 942 385  1410 1000 410 **

 

[ Return to Top   Return to Section ]

Appendix B: Estimating child costs using alternative standard of living estimators

Alternative estimates of costs (A)

Table B1 and figures  B2a, B2b, B2c and B2d compare estimates of costs of children produced using  the following alternative estimators of living  standards:

  • Engel (Eng), where the estimator is the proportion of household expenditure that goes to food at home;
  • Basic goods  (BG1 and BG2), where the estimators are the proportion of household expenditure that goes to a basic  of ‘basic  goods’ (see  Table 1 in Section  2 for a description of the baskets’ contents);
  • Rothbarth (Rth), where the estimator is household expenditure on ‘adult  goods’ (tobacco, alcohol, coffee  and adult clothing)

To summarise the main points  of difference:

  • The Rothbarth  estimator (Rth) appears to produce reasonable estimates only for children in the two younger age groupings, unlikely results  for 15 to 17 year  old and questionable results for 10 to 14 year  olds. This appears to be a result  of households recording adult clothing when  the purchase was in fact for younger aged  children. For this reason, the Rothbarth  estimates should  be discounted. (It is, however, interesting to note that the estimates for the two youngest age groups  were the highest estimates of all the estimators.)
  • When compared to the basic  goods/ISO-PROP estimators, the Engel estimator was found to produce the highest costs estimates, except for children in the youngest age group. In all other  instances, the Engel estimates were highest and those  produced by the basic  goods estimator BG2 the lowest.
  • The basic  goods estimator BG1 produced higher  cost estimates (when compared to the Engel estimates) for children aged  0 to 4 years  and lower estimates for all other  ages. The basic  goods estimator BG2 produced lower cost estimates (when compared to the Engel estimates) for children of all ages. The greatest variation between the Engel and ISO-PROP estimators were for children aged  either  5 to 7 or 10 to 14.
  • The greatest variation in costs estimates (excluding the Rothbarth  estimates) was for children aged  5 to 9 years  and the least for children aged  15 to 17 years.
Table B1: Alternative estimates of child costs
Weekly household income Age of child
0 to 4 5 to 9 10 to 14 15 to 17
  ($ pw) ($ pw) ($ pw) ($ pw)
400 54 102 132 154
700 88 141 174 197
1000 119 178 213 239
1300 149 213 251 278
1600 178 247 286 315
1900 205 279 320 350
2200 230 309 352 384
2500 250 337 382 415
2800 278 364 411 444
3100 299 389 437 472
Basic goods 1 (BG1)
400 74 85 113 137
700 107 119 150 178
1000 138 153 187 216
1300 168 185 221 253
1600 196 215 254 288
1900 223 244 286 322
2200 249 272 315 353
2500 274 297 343 383
2800 297 322 369 411
3100 318 345 394 437
Basic goods 2 (BG2)
400 51 61 88 126
700 76 89 120 164
1000 102 117 151 201
1300 126 143 181 236
1600 150 169 210 269
1900 172 194 237 301
2200 194 217 263 331
2500 214 239 287 359
2800 234 260 311 385
3100 252 280 332 410
Rothbarth  (Rth)
400 76 95 71 6
700 119 139 115 46
1000 159 180 154 82
1300 195 219 191 114
1600 230 255 225 144
1900 262 289 257 173
2200 292 320 287 199
2500 321 350 315 224
2800 347 378 341 247
3100 371 403 366 268

Source: ABS (1994) and authors’ calculations

Figure B2a: Alternative estimates of child costs, for children aged 0 to 4 years

Figure B2a: Alternative estimates of child costs, for children aged 0 to 4 years

Figure B2b: Alternative estimates of child costs, for children aged 5 to 9 years

Figure B2b: Alternative estimates of child costs, for children aged 5 to 9 years

Figure B2c: Alternative estimates of child costs, for children aged 10 to 14 years

Figure B2c: Alternative estimates of child costs, for children aged 10 to 14 years

Figure B2d: Alternative estimates of child costs, for children aged 15 to 17 years

Figure B2d: Alternative estimates of child costs, for children aged 15 to 17 years

Alternative estimates of child costs (B)

An additional question that arises  with  respect to Espenshade’s methodology is: to what  extent is it reasonable to be comparing people who  may be at very different  points  of their  life cycles? To an extent, the analysis accounted for this by restricting the dataset  used in the analysis to couple-only families  and, for couples without children, to those  where the wife  was between the ages of 25 and 54 years. However, to determine how  sensitive the estimates were to further age restrictions, the costs of children were re-estimated using  two separate datasets. The first was used to estimate the costs of younger children (0 to 4 years  and 5 to 9 years) and the second  the cost of older  children (10 to 14 years  and 15 to 17 years).

The dataset  used for the younger age groups  was restricted to couples where both partners were aged  between 20 and 39 years  and the dataset  for older  children to couples between 34 and 54 years.

The resulting cost estimates by age of child  are shown  in Table B3. The differences between the original estimates (based  on the full dataset) and new  estimates (based  on the restricted datasets) varied  most for youngest and the oldest  children and least for children aged  5 to 9 and 10 to 14. In all instances the new  estimates were higher.

While  this appears to indicate that the cost estimates presented in this study  are conservative, it needs  to be noted  that about 40 per cent  of children fall outside  the definitions used to define the two datasets from which the estimates were derived. It may that these  families, where one or both parents are younger or older, have different  spending patterns and it is not clear  to which other  groups  they  should  most appropriately be compared.

Table B3: Alternative estimates of child costs
Weekly household income Age of child
0 to 4 5 to 9 10 to 14 15 to 17
  ($ pw) ($ pw) ($ pw) ($ pw)
Original  estimates
400 51 61 88 126
700 76 89 120 164
1000 102 117 151 201
1300 126 143 181 236
1600 150 169 210 269
1900 172 194 237 301
2200 194 217 263 331
2500 214 239 287 359
2800 234 260 311 385
3100 252 280 332 410
New estimates
400 85 82 105 148
700 115 113 137 191
1000 144 142 168 232
1300 171 170 198 271
1600 197 195 226 308
1900 220 219 253 343
2200 242 241 279 377
2500 262 262 303 408
2800 280 280 326 438
3100 297 296 347 465
Difference  ($pw)
400 34 21 17 22
700 39 24 17 27
1000 42 25 17 31
1300 45 27 17 35
1600 47 26 16 39
1900 48 25 16 42
2200 48 24 16 46
2500 48 23 16 49
2800 46 20 15 53
3100 45 16 15 55
Difference  (%)
400 67 34 19 17
700 51 27 14 16
1000 41 21 11 15
1300 36 19 9 15
1600 31 15 8 14
1900 28 13 7 14
2200 25 11 6 14
2500 22 10 6 14
2800 20 8 5 14
3100 18 6 5 13

Source: ABS (1994) and authors’ calculations

[ Return to Top   Return to Section ]

Appendix  C: Relating income to expenditure

Central  to Espenshade’s methodology is that child  costs are calculated as the difference in household expenditures (between couples with  and without children) at a particular household income level. This latter  facility  is particularly useful  as we are much  more used to thinking in terms  of income than in terms  of expenditure. However, it does create some problems as there are characteristics of the two concepts which means  their  relationship is not as straightforward as might  be assumed.

This partly  occurs  because of the way  the Household  Expenditure Survey  records incomes and expenditures. As the ABS (1995, p. 11) explains, total incomes and total expenditures can be expected to differ because:

  • expenditure does not cover  all current payments (this is because the ABS uses on acquisition-based approach to recording expenditure);
  • expenditure estimates for different  items refer to different  periods;
  • not all income is included (for example, receipts from drawing down  assets, capital gains and capital losses, income in kind from other  households);
  • income estimates for different  sources of income refer to different  periods.

As a consequence, the household incomes and total  expenditures will  not necessarily balance over the period  during  which the Household  Expenditure Survey  was recorded.
Table C1 shows  the relationship between household incomes and current expenditure, as predicted by equation (1).

Table C1: Predicted relationship between current expenditure and income, Australia, 1993–94
Household weekly income Household weekly current expenditure Current  expenditure as a proportion of income
400 457 1.14
700 586 0.84
1000 709 0.71
1300 826 0.64
1600 937 0.59
1900 1,043 0.55
2200 1,142 0.52
2500 1,236 0.49
2800 1,325 0.47
3100 1,407 0.45

Source: ABS (1994) and authors’ calculations
Note: see text  (Chapter 3) for households included in the analysis

Most noticeably, the model  predicts that the expenditures of families  on the lowest incomes levels  are actually higher  than their  incomes. This ratio (of expenditure to income) then decreases as income rises, so that families  with  an income level  of $2,500 or greater are predicted to use less than half of their  income in paying for items of current expenditure. It should  be noted  that the modelling estimates are a reasonably accurate reflection of what  was reported in the Household  Expenditure Survey  by respondents.

The implications of this characteristic relationship between income and expenditure means  that expenditure-based costs of children will  have a similar  relationship. That is, the costs of children in lower income family will  appear to consume a higher  proportion of their  income than if the costs were simply  related to their  expenditure. Again, this is not just true of their  expenditure
on children, but of all their  expenditure.

This effect can seen  in Table C2, which shows  the costs of children at current expenditure levels between $400  and $1,400 (as Table C1 indicates, these  expenditure levels  approximately cover  the family income range  of $400  to $3,100 per week). When the costs in Table C2 are compared to those  in Table 4, Section  4, it can be seen  that they  are very similar.

However, when  the costs in Table C2 are expressed as a proportion of family expenditure (Table C3) a different  picture emerges. In terms  of family income, the modelling suggests that families tend to spend  a decreasing proportion of their  incomes on children (see  Table 5, Section  4). However, when  considered in terms  of total household expenditure, expenditure on all
children, except for younger children as incomes initially increase, would  appear to be a more constant proportion.

Table C2: Estimated average costs of a single child, by age of child, Australia 1993–94
Weekly household current expenditure Age of child
0 to 4 5 to 9 10 to 14 15 to 17
  ($ pw) ($ pw) ($ pw) ($ pw)
400 43 57 77 108
600 82 101 127 167
800 124 147 178 227
1000 166 193 230 287
1200 210 240 282 347
1400 254 288 334 407

Source: ABS (1994) and authors’ calculations
Note: see text  (Chapter 3) for households included in the analysis

Table C3: Estimated share of household current expenditure of a single child, by age of child, Australia 1993–94
Weekly household current expenditure Age of child
0 to 4 5 to 9 10 to 14 15 to 17
  ($ pw) ($ pw) ($ pw) ($ pw)
400 10.8 14.3 19.3 27.0
600 13.7 16.8 21.2 27.8
800 15.5 18.4 22.3 28.4
1000 16.6 19.3 23.0 28.7
1200 17.5 20.0 23.5 28.9
1400 18.1 20.6 23.9 29.1

Source: ABS (1994) and authors’ calculations
Note: see text  (Chapter 3) for households included in the analysis

[ Return to Top   Return to Section ]

Appendix D: Estimates of expenditure on children in 1998

The Household  Expenditure Survey  from which the estimates used in this study  were derived was conducted over the period  1993–94. As with  all such large  and complex surveys, there  is inevitably a lengthy period  between the survey  being  conducted and the data becoming publicly available for analysis. As a result, the survey’s estimates are historic rather  than current.

To derive  more current estimates of the costs of children, the cost estimates presented in Section 4 are adjusted in this appendix to allow  for changes in the prices of items of household expenditure and in family income levels

There are two approaches that can be adopted to produce more current cost estimates. These are outlined in the following sections.

D1.  Inflating/deflating the original estimates

The first option  is to simply  inflate  or deflate  all expenditure items and incomes by a common factor to account for any change. Typically, the change in the Consumer Price Index  (CPI) between the period  of the survey  and a more recent point in time is used for this purpose.

For example, the All Groups CPI (weighted average of eight  capital cities) index  was 110.35 over the period  1993–94 and 121.90 in December 1998  (ABS 1998b). Accordingly, an adjustment factor for this period  would  be 121.90/110.35 or 1.105.

Tables D1, D2 and D3 show  the effect of applying this inflator  to the costs of children estimators. These are still 1993–94 estimates, but are now  expressed in 1998  values.

D2.  Ageing the input data

A more comprehensive method  for providing more current costs estimates is to attempt to simulate a more current world. This can be at least partially achieved by statically ageing the key components from which the cost estimates are derived. In this instance, these  are household incomes and expenditures.

Static ageing means  that these  items are adjusted (inflated/deflated) to match  the average changes that have occurred to them over the period  of interest (see  Percival  1994  for a fuller discussion of the techniques).

This adjustment can be undertaken with  various  degrees of complexity. In this instance, a reasonably simple  process was adopted. This consisted of the following:

  • Households were separated into those  whose principal source of income were from wages or salaries and ‘others’.
    The incomes of the former were adjusted by the change in median  incomes, as reported in Weekly Earnings of Employees  (Distribution) (ABS, 1993, 1994  and 1997) for the period 1993–94 to August 1997, and by Average Weekly Earnings (ABS 1998a), for the period August 1997  to November  1998. The incomes of the non-wage and salary  households (‘other’) were adjusted by the movement in the CPI between 1993–94 and December 1998.
  • Expenditures were adjusted by the movement in the CPI between 1993–94 and December 1998  (all groups, weighted average of eight  capital cities).

Tables D4, D5 and D6 show  the estimates of child  costs for December 1998.

Table D1: Estimated average costs of a single child, by age of child, Australia, December 1998
Weekly household income Age of child
0 to 4 5 to 9 10 to 14 15 to 17
  ($ pw) ($ pw) ($ pw) ($ pw)
442 56 67 97 139
773 84 98 133 181
1105 113 129 167 222
1436 139 158 200 261
1767 166 187 232 297
2099 190 214 262 333
2430 214 240 291 366
2762 236 264 317 397
3093 258 287 344 425
3424 278 309 367 453

Source: ABS (1994) and authors’ calculations

Table D2: Estimated average costs of a single child, by number of children, Australia, December 1998
Weekly household income Number of children
1 child 2 children 3 children
($ pw) ($ pw) ($ pw)
442 76 149 *
773 109 205 286
1105 143 260 356
1436 173 310 420
1767 202 358 481
2099 230 403 537
2430 254 444 589
2762 277 482 637
3093 298 516 680
3424 317 547 719

Source: ABS (1994) and authors’ calculations
* not calculated

Table D3: Estimated average costs of a single child, by individual expenditure categories and age of child, Australia, December 1998
  Weekly family income
Expenditure category 442 773 1105 1436 1767 2099 2430 2762 3093 3424
1 child,  0 to 4 years
Current  housing costs 17 20 24 27 31 34 37 41 44 47
Fuel and power 2 2 2 2 2 3 3 3 3 3
Food 13 16 19 21 23 25 27 29 30 32
Alcohol –1 –1 0 0 1 1 2 2 2 2
Tobacco 0 0 0 0 0 0 0 0 0 0
Clothing  and footwear 2 5 7 9 11 13 15 16 18 19
Furnishings and equipment –4 –1 2 7 12 17 23 28 34 41
Services and operation 15 16 17 18 19 20 21 22 23 24
Medical  care  and health 2 3 4 4 5 6 6 6 7 7
Transport 18 23 28 33 37 40 44 46 49 51
Recreation & entertainment –5 0 5 9 13 17 20 23 25 28
Personal  care 0 0 1 1 1 2 2 2 2 2
Miscellaneous –2 1 4 7 10 13 15 18 20 22
Total 56 85 112 139 165 190 214 237 258 278
1 child,  5 to 9 years
Current  housing costs 8 12 15 19 23 26 30 34 37 41
Fuel and power 2 2 2 2 3 3 3 3 3 3
Food 18 21 24 26 29 31 33 35 36 38
Alcohol –2 –1 –1 –0 0 1 1 2 2 2
Tobacco 0 0 0 0 0 0 0 0 0 0
Clothing  and footwear 7 9 12 14 16 18 20 22 24 25
Furnishings and equipment 0 3 7 12 17 23 29 35 41 48
Services and operation 8 9 10 11 12 14 15 16 17 18
Medical  care  and health 0 1 2 3 3 4 4 5 5 6
Transport 8 14 20 25 29 33 36 39 42 44
Recreation & entertainment 11 16 22 26 31 35 38 42 44 47
Personal  care 1 1 2 2 2 3 3 3 3 3
Miscellaneous 7 11 15 18 21 24 27 29 31 34
Total 67 98 129 158 187 214 240 264 287 309
1 child,  10 to 14 years
Current  housing costs 5 9 13 18 22 26 30 34 38 42
Fuel and power 2 2 3 3 3 3 4 4 4 4
Food 27 30 34 37 39 42 44 46 48 49
Alcohol 0 0 1 2 2 3 3 3 4 4
Tobacco 1 1 1 1 1 1 1 1 1 1
Clothing  and footwear 11 14 17 20 22 24 27 28 30 32
Furnishings and equipment 0 4 9 14 21 27 34 42 49 56
Services and operation –1 0 2 3 4 6 7 8 9 10
Medical  care  and health 2 3 4 5 5 6 6 7 7 8
Transport 8 14 20 26 30 35 38 42 45 47
Recreation & entertainment 18 24 30 35 40 44 48 52 55 58
Personal  care 2 2 3 3 4 4 4 4 5 5
Miscellaneous 22 26 30 34 37 40 43 46 49 51
Total 97 132 167 200 231 262 290 317 343 367
1 child,  15 to 17 years
Current  housing costs 3 8 13 18 23 28 33 37 42 46
Fuel and power 5 5 5 6 6 6 6 7 7 7
Food 41 45 49 52 55 58 61 63 65 66
Alcohol –1 0 0 1 2 2 3 3 3 4
Tobacco –1 –1 –1 –1 –1 –1 –1 –1 –1 –1
Clothing  and footwear 21 24 27 30 33 36 38 40 42 44
Furnishings and equipment –14 –9 –3 3 11 19 27 35 44 53
Services and operation 4 6 7 9 11 12 13 15 16 17
Medical  care  and health 6 7 8 9 10 11 11 12 12 13
Transport 19 27 34 40 46 50 55 58 62 65
Recreation & entertainment 12 19 26 32 38 43 47 51 54 57
Personal  care 5 6 6 7 7 8 8 8 8 9
Miscellaneous 40 45 49 53 57 61 64 68 71 73
Total 139 181 222 260 297 332 365 396 426 453

Source: ABS (1994) and authors’ calculations

Table D4: Estimated average costs of a single child, by age of child, Australia, December 1998
Weekly household income Age of child
0 to 4 5 to 9 10 to 14 15 to 17
  ($ pw) ($ pw) ($ pw) ($ pw)
400 51 61 90 127
700 77 90 122 165
1000 102 117 153 202
1300 126 144 183 237
1600 150 170 212 271
1900 173 194 240 303
2200 195 218 266 334
2500 216 241 291 363
2800 235 262 315 390
3100 254 283 338 416

Source: ABS (1994) and authors’ calculations

Table D5: Estimated average costs of a single child, by number of children, Australia, December 1998
Weekly household income Number of children
1st child
($ pw)
2nd child
($ pw)
3rd child
($ pw)
400 69 137 *
700 100 189 264
1000 130 238 328
1300 158 285 387
1600 185 330 444
1900 210 371 496
2200 234 410 545
2500 255 446 591
2800 276 478 632
3100 294 508 670

Source: ABS (1994) and authors’ calculations
* not calculated

Table D6: Estimated average costs of a single child, by individual expenditure categories and age of child, Australia, December 1998
  Weekly family income
Expenditure category 400 700 1000 1300 1600 1900 2200 2500 2800 3100
1 child,  0 to 4 years
Current  housing costs 15 18 22 26 30 33 37 41 44 47
Fuel and power 1 2 2 2 2 2 3 3 3 3
Food 12 16 19 22 24 27 29 31 33 34
Alcohol –2 –1 0 0 1 1 2 2 3 3
Tobacco 0 0 0 0 0 0 0 0 0 0
Clothing  and footwear 1 4 7 9 12 14 16 18 20 21
Furnishings and equipment –5 –3 1 5 9 14 20 26 31 37
Services and operation 14 16 17 18 19 21 22 23 24 25
Medical  care  and health 1 2 3 4 5 6 6 7 7 8
Transport 15 22 28 33 38 42 46 50 53 55
Recreation & entertainment –8 –1 5 10 15 20 24 27 30 33
Personal  care 0 0 1 1 2 2 2 3 3 3
Miscellaneous –4 0 4 8 11 14 17 19 22 24
Total 41 76 108 139 169 197 223 248 272 294
1 child,  5 to 9 years
Current  housing costs 10 15 20 24 28 33 37 41 45 49
Fuel and power 2 2 2 3 3 3 3 3 4 4
Food 20 24 28 32 35 37 40 42 44 46
Alcohol –1 0 0 1 2 2 3 3 3 4
Tobacco 0 0 0 0 0 0 0 0 0 0
Clothing  and footwear 9 12 16 19 22 24 26 29 31 32
Furnishings and equipment 0 3 8 13 18 24 31 38 45 52
Services and operation 8 10 11 13 14 15 17 18 19 20
Medical  care  and health 1 2 3 4 5 6 7 7 8 8
Transport 13 21 28 35 40 45 50 54 57 60
Recreation & entertainment 15 23 30 37 42 47 52 56 60 63
Personal  care 1 2 2 3 3 4 4 4 4 5
Miscellaneous 10 15 20 24 28 31 35 38 40 43
Total 89 130 168 205 239 272 303 333 361 387
1 child,  10 to 14 years
Current  housing costs 7 11 16 21 26 31 35 40 44 48
Fuel and power 2 3 3 3 3 4 4 4 4 4
Food 30 34 38 42 45 48 51 53 55 57
Alcohol 0 1 2 3 3 4 4 5 5 5
Tobacco 1 1 1 1 1 1 1 1 1 1
Clothing  and footwear 14 18 21 24 27 30 32 35 37 39
Furnishings and equipment –1 3 8 14 20 27 34 42 49 57
Services and operation 0 1 3 5 6 7 9 10 11 13
Medical  care  and health 3 4 5 6 7 8 8 9 9 10
Transport 13 21 29 35 41 46 51 55 59 62
Recreation & entertainment 25 33 40 47 52 58 63 67 71 74
Personal  care 3 4 4 5 5 6 6 6 7 7
Miscellaneous 25 30 35 39 43 47 50 54 57 59
Total 121 164 205 244 280 315 349 380 409 437
1 child,  15 to 17 years
Current  housing costs 2 7 12 18 23 28 32 37 42 46
Fuel and power 4 5 5 5 6 6 6 6 7 7
Food 42 46 50 54 57 60 63 66 68 70
Alcohol –1 0 1 1 2 2 3 3 4 4
Tobacco –1 –1 –1 –1 –1 –1 –1 –1 –1 –1
Clothing  and footwear 21 25 29 32 35 38 40 43 45 47
Furnishings and equipment –16 –12 –7 –1 6 13 20 28 36 45
Services and operation 4 6 7 9 10 12 13 15 16 17
Medical  care  and health 6 7 9 10 10 11 12 13 13 14
Transport 21 29 37 43 49 55 60 64 68 71
Recreation & entertainment 14 22 30 36 43 48 53 58 62 65
Personal  care 5 6 7 7 8 8 9 9 9 9
Miscellaneous 40 45 50 55 59 63 66 70 73 76
Total 141 185 228 268 306 343 377 410 441 469

Source: ABS (1994) and authors’ calculations

[ Return to Top   Return to Section ]

Appendix  E: Child costs and working mothers

One of the more discussed trends  over the past couple of decades has been  the increase in labour  force participation by women and the associated rise in two-earner families. In the context of this study, these  trends  raise the question of whether there  are differences in child costs according to whether or not a child’s mother  is in or out of the paid labour  force. Intuitively, it would  seem  that there  should  be a difference and that this most likely  would  see families  with  working mothers  having  to spend  more on their  children than families  with  non- working mothers.

Both the earlier studies  by Espenshade and Lee attempted to estimate separate child  costs for families  according the employment status of the wife.4

In Espenshade’s 1984  estimates for the United States this consisted of including separate terms for husband’s income and wife’s  income in the equation he used to estimate family consumption. His findings  (which are presented as total child  costs across  a synthetic lifecycle as children age from birth through to 18 years) are that families  spend  more on children where the wife  is employed (about  8 per cent  more if the wife  works  part-time  and about 23 per cent more if she works  full-time).

In his Australian  study, Lee used a similar  method  to examine the impact  of the employment status of the wife  on child  costs. However, the estimates of child  costs he presented were where the wife  had no income or where her income was a proportion of her husband’s (either a third, a half, or equal  to his). In his discussion, Lee concludes that families  where the wife  works  are more likely  to spend  more on their  children than those  where the wife  does not work. However, examination of the results  that are presented do not appear to support this conclusion. For example, in a two child  family (both  aged  between 5 and 14 years) and where the family income is $400  per week, if all the income comes  from the father  the child  costs are given  as $170  per week, while if the mother  earns  $200  and the father  earns  $200, the costs are given  as $162  per week.

The apparent reason  for this outcome is that the data used in the study  (the  1983–84 Household  Expenditure Survey)  indicate that the overall  expenditure of households where the wife  works  (defined as contributing income) is slightly less than that where the wife  does not work. Espenshade’s methodology then feeds this difference through to the final cost of children. The results  that this produces appear to be counter-intuitive, given, for example, that child  care  costs are likely  to be a significant cost of children and one that would (mostly) be borne  by families  where both parents work. This, however, does not necessarily mean  that the total costs of children will  be less in families  where the wife  does not work. For example, it may be that expenditure on other  items may be proportionally higher  in such families.

Because of uncertainty about the methodology used by Lee and Espenshade in capturing anything more than the differences in overall  expenditure levels  between households where wives  do and do not work, the previous analysis in this study  did not treat these  two groups separately. However, the original question remains: is there  a difference between their  child costs?

Accordingly, to better  understand the impact  of a mother’s labour  force status on child  costs we undertook additional analysis whereby the original data sample  was split into two groups  — families  where the wife  worked and families  where the wife  did not work. The child  costs were then separately estimated for each  group. To avoid any confounding effect that might  arise from estimating family expenditure from a given  level  of family income, child  costs were only estimated using  family expenditures in the range  between $400  and $1,400 per week (as noted in Appendix C, this corresponds to a family income range  of $400  to $3,100 per week). Similarly, to limit the increased variability that would  occur  as the sample  size was reduced, only the two labour  force types  were defined  (that  is, working included wives  working either  full or part time).

The results  of the analysis are presented in figures  E1a, E1b, E1c and E1d. Overall, they  appear to confirm  the supposition that families  tend to spend  more on children when  the wife  works. This is true across  all incomes and, with  the exception of children aged  10 to 14 years, across  all ages. The differences, in percentage terms, were greatest for families  on low incomes and where children were aged  between 15 and 17 years.

However, in considering these  results, it is important to note that they  should  at best be considered only a preliminary exploration of the question. In particular, they  may be sensitive to the reduced samples on which they  are based. As well, the categorisation of labour  force status in the Household  Expenditure Survey, whereby a part-time  worker is defined  as one who  works more than one hour but less than 35, tends to hamper the analysis of female  labour participation.

Figure E1a: Estimates of child costs by mother’s labour  force status, children aged 0 to 4 years

Figure E1a: Estimates of child costs by mother’s labour  force status, children aged 0 to 4 years

Figure E1b: Estimates of child costs by mother’s labour  force status, children aged 5 to 9 years

Figure E1b: Estimates of child costs by mother’s labour  force status, children aged 5 to 9 years

Figure E1c: Estimates of child costs by mother’s labour  force status, children aged 10 to 14 years

Figure E1c: Estimates of child costs by mother’s labour  force status, children aged 10 to 14 years

Figure E1d: Estimates of child costs by mother’s labour  force status, children aged 15 to 17 years

Figure E1d: Estimates of child costs by mother’s labour  force status, children aged 15 to 17 years

[ Return to Top   Return to Section ]

Appendix  F: Regression equations

Description of variables

C = household consumption expenditure (X1000)

fY = weekly total household income (X1000)

fY2 = the square  of total household income

Age1 = number  of persons 0 to 4 years

Age2 = number  of persons 5 to 9 years

 Age3 = number  of persons 10 to 14 years  

Age4  = number of persons 15 to 17 years  

Age5 = number  of persons 18 to 24 years  

Age6 = number  of persons 25 years  and over

CKA1 = number  of persons of aged  0 to 4 years  divided  by family size

CKA2 = number  of persons of aged  5 to 9 years  in divided  by family size

CKA3 = number of persons of aged  10 to 14 years  divided  by family size

CKA4  = number  of persons of aged  15 to 17 years  divided  by family size

CKA5 = number  of persons of aged  18 to 24 years  divided  by family size

CKA6= number  of persons of aged  25 years  and older  divided  by family size

Kids  = number  of persons aged  0 to 17 years

CKA = number  of persons aged  0 to 17 years  divided  by family size

LNPF= the logit of the proportion of household consumption expenditure spent  on food at home

LNPF1        = the logit of the proportion of household consumption expenditure spent  on basic  goods—‘basket 1’

LNPF2        = the logit of the proportion of household consumption expenditure spent  on basic  goods—‘basket 2’

A_GOODS = expenditure on adult clothing, alcohol  and tobacco

LEFS = the log of per capita  consumption

LEFS2  = LEFS squared

LnF = the log of household size (persons)

Tables A1.1–A1.18—by  ages of children

Table A1.1: Equation to predict  total  family  consumption expenditure

Dependent variable: C

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 81150.84083 10143.85510 343.363 0.0001
Error 2555 75481.58564 29.54269    
C Total 2563 156632.42647      

Root MSE 5.43532 R-square 0.5181
Dep mean 0.65625 Adj R-sq 0.5166
C.V. 828.23607    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –0.145663 0.04433861 –3.285 0.0010
FY 1 0.464668 0.01586040 29.297 0.0001
FY2 1 –0.032250 0.00471892 –6.834 0.0001
AGE1 1 0.039297 0.00675616 5.817 0.0001
AGE2 1 0.014373 0.00513274 2.800 0.0051
AGE3 1 0.039865 0.00606161 6.577 0.0001
AGE4 1 0.057255 0.01000196 5.724 0.0001
AGE5 1 0.178105 0.01795534 9.919 0.0001
AGE6 1 0.183870 0.02181733 8.428 0.0001

Table A1.2: Equation to predict  percentage  of family  consumption spent on food  at home  (Engel estimator)

Dependent variable: LNPF

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 9 307544.34833 34171.59426 170.356 0.0001
Error 2616 524742.01976 200.58946    
C Total 2625 832286.36809      

Root MSE 14.16296 R-square 0.3695
Dep mean –1.81777 Adj R-sq 0.3673
C.V. –779.14040

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –4.488753 0.71573788 –6.272 0.0001
LEFS 1 –1.301335 0.07250869 –17.947 0.0001
LEFS2 1 –0.191550 0.02214398 –8.650 0.0001
LNF 1 0.276000 0.10700925 2.579 0.0100
CKA1 1 0.157978 0.69904092 0.226 0.8212
CKA2 1 0.452723 0.69731292 0.649 0.5162
CKA3 1 0.579868 0.70244297 0.826 0.4092
CKA4 1 0.673841 0.70878327 0.951 0.3418
CKA5 1 0.744035 0.69955259 1.064 0.2876
CKA6 1 1.003336 0.70141008 1.430 0.1527

Table A1.3: Equation to predict  percentage  of family  consumption spent on ‘basic goods’—
basket 1

Dependent variable: LNPF1

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 9 330461.99727 36717.99970 270.177 0.0001
Error 2602 353620.83198 135.90347    
C Total 2611 684082.82924      

Root MSE 11.65776 R-square 0.4831
Dep mean –1.46622 Adj R-sq 0.4813
C.V. –795.09122    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –3.991654 0.58949055 –6.771 0.0001
LEFS 1 –1.271795 0.05991008 –21.228 0.0001
LEFS2 1 –0.160435 0.01829383 –8.770 0.0001
LNF 1 0.129528 0.08849429 1.464 0.1434
CKA1 1 0.337373 0.57547279 0.586 0.5578
CKA2 1 0.439215 0.57409906 0.765 0.4443
CKA3 1 0.570149 0.57834985 0.986 0.3243
CKA4 1 0.690715 0.58354628 1.184 0.2367
CKA5 1 0.717521 0.57588804 1.246 0.2129
CKA6 1 0.954216 0.57752206 1.652 0.0986

Table A1.4: Equation to predict  percentage  of family  consumption spent on ‘basic goods’—
basket 2

Dependent variable: LNPF2

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 9 310849.88228 34538.87581 305.967 0.0001
Error 2586 291918.67044 112.88425    
C Total 2595 602768.55272      

Root MSE 10.62470 R-square 0.5157
Dep mean –1.21189 Adj R-sq 0.5140
C.V. –876.70218    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –3.967329 0.53746302 –7.382 0.0001
LEFS 1 –1.169718 0.05474230 –21.368 0.0001
LEFS2 1 –0.124146 0.01672355 –7.423 0.0001
LNF 1 0.105267 0.08088029 1.302 0.1932
CKA1 1 0.560944 0.52449886 1.069 0.2850
CKA2 1 0.653529 0.52324806 1.249 0.2118
CKA3 1 0.784424 0.52715738 1.488 0.1369
CKA4 1 0.998200 0.53184674 1.877 0.0607
CKA5 1 1.034676 0.52495039 1.971 0.0488
CKA6 1 1.309346 0.52645584 2.487 0.0129

Table A1.5: Equation to predict  expenditure on adult  clothing, coffee, tobacco and alcohol  (Rothbarth estimator)

Dependent variable: A_GOODS

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 9 1715572665.4 190619185.04 72.551 0.0001
Error 2652 6967841090.2 2627391.0597    
C Total 2661 8683413755.6      

Root MSE 1620.92290 R-square 0.1976
Dep mean 49.76974 Adj R-sq 0.1948
C.V. 3256.84445    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 146.641149 81.84073467 1.792 0.0733
LEFS 1 122.206358 8.03913801 15.201 0.0001
LEFS2 1 21.998165 2.44603839 8.993 0.0001
LNF 1 53.910702 12.15451275 4.435 0.0001
CKA1 1 –52.938402 79.96334385 –0.662 0.5080
CKA2 1 –63.873061 79.76483125 –0.801 0.4233
CKA3 1 –51.467277 80.31440349 –0.641 0.5217
CKA4 1 –25.641568 81.05062742 –0.316 0.7518
CKA5 1 –0.215117 80.02748541 –0.003 0.9979
CKA6 1 –15.004015 80.23543013 –0.187 0.8517

Table A1.6: Equation to predict  family  expenditure on housing Dependent variable: EXP01 Household exp for current housing costs

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 4747074999.7 593384374.96 124.474 0.0001
Error 2588 12337310221 4767121.4145    
C Total 2596 17084385220      

Root MSE 2183.37386 R-square 0.2779
Dep mean 107.44962 Adj R-sq 0.2756
C.V. 2031.99775    
Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 130.459189 17.70620043 7.368 0.0001
C 1 96.094986 6.74119190 14.255 0.0001
C2 1 13.511103 1.91109778 7.070 0.0001
AGE1 1 9.902509 2.72503228 3.634 0.0003
AGE2 1 0.631409 2.06115071 0.306 0.7594
AGE3 1 –4.940286 2.40390481 –2.055 0.0400
AGE4 1 –10.936264 3.94222391 –2.774 0.0056
AGE5 1 –36.487996 7.20310750 –5.066 0.0001
AGE6 1 –49.372737 8.82410980 –5.595 0.0001

Table A1.7: Equation to predict  family  expenditure on fuel and power

Dependent variable: EXP02 Household expenditure for fuel and power

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 26513918.433 3314239.8042 42.436 0.0001
Error 2588 202120016.1 78098.924306    
C Total 2596 228633934.54      

Root MSE 279.46185 R-square 0.1160
Dep mean 18.99276 Adj R-sq 0.1132
C.V. 1471.41251    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 4.091954 2.26631251 1.806 0.0711
C 1 7.936032 0.86284167 9.198 0.0001
C2 1 –0.250806 0.24461176 –1.025 0.3053
AGE1 1 1.000133 0.34879164 2.867 0.0042
AGE2 1 1.004694 0.26381784 3.808 0.0001
AGE3 1 1.310817 0.30768880 4.260 0.0001
AGE4 1 3.239666 0.50458659 6.420 0.0001
AGE5 1 2.728937 0.92196475 2.960 0.0031
AGE6 1 3.790136 1.12944561 3.356 0.0008

Table A1.8: Equation to predict  family  expenditure on food  and non-alcoholic beverages

Dependent variable: EXP03 Household exp food & non alcoholic bevs

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 3327821530.1 415977691.26 280.253 0.0001
Error 2588 3841352825.4 1484293.982    
C Total 2596 7169174355.5      

Root MSE 1218.31604 R-square 0.4642
Dep mean 129.29593 Adj R-sq 0.4625
C.V. 942.26944    
Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –43.881348 9.88000655 –4.441 0.0001
C 1 120.309006 3.76156479 31.984 0.0001
C2 1 –12.534915 1.06638681 –11.755 0.0001
AGE1 1 6.417791 1.52055981 4.221 0.0001
AGE2 1 9.147227 1.15011589 7.953 0.0001
AGE3 1 14.625808 1.34137165 10.904 0.0001
AGE4 1 23.163873 2.19974908 10.530 0.0001
AGE5 1 32.651979 4.01931231 8.124 0.0001
AGE6 1 39.752108 4.92382671 8.073 0.0001

Table A1.9: Equation to predict  family  expenditure on alcohol

Dependent variable: EXP04 Household expenditure for alcohol

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 109495267.79 13686908.474 34.956 0.0001
Error 2588 1013316851.1 391544.37833    
C Total 2596 1122812118.9      

Root MSE 625.73507 R-square 0.0975
Dep mean 16.94836 Adj R-sq 0.0947
C.V. 3692.00870    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 4.088194 5.07443584 0.806 0.4205
C 1 25.180446 1.93196422 13.034 0.0001
C2 1 –2.990256 0.54770323 –5.460 0.0001
AGE1 1 –2.308712 0.78096944 –2.956 0.0031
AGE2 1 –3.110282 0.59070703 –5.265 0.0001
AGE3 1 –2.353950 0.68893723 –3.417 0.0006
AGE4 1 –4.050497 1.12980548 –3.585 0.0003
AGE5 1 0.349278 2.06434503 0.169 0.8657
AGE6 1 0.380534 2.52890953 0.150 0.8804

Table A1.10: Equation to predict  family  expenditure on tobacco

Dependent variable: EXP05 Household expenditure for tobacco

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 6286895.213 785861.90162 3.545 0.0004
Error 2588 573722352.15 221685.60748    
C Total 2596 580009247.37      

Root MSE 470.83501 R-square 0.0108
Dep mean 9.61044 Adj R-sq 0.0078
C.V. 4899.20557    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 23.999353 3.81826458 6.285 0.0001
C 1 0.187701 1.45370851 0.129 0.8973
C2 1 –0.280823 0.41211987 –0.681 0.4957
AGE1 1 0.289737 0.58764128 0.493 0.6220
AGE2 1 0.098174 0.44447812 0.221 0.8252
AGE3 1 1.225603 0.51839154 2.364 0.0181
AGE4 1 –0.625851 0.85012332 –0.736 0.4617
AGE5 1 –3.284734 1.55331859 –2.115 0.0346
AGE6 1 –7.495966 1.90288063 –3.939 0.0001

Table A1.11: Equation to predict  family  expenditure on clothing and footwear

Dependent variable: EXP06 Household exp clothing and footwear

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 1842267255.8 230283406.97 90.612 0.0001
Error 2588 6577237665.2 2541436.5012    
C Total 2596 8419504920.9      

Root MSE 1594.18835 R-square 0.2188
Dep mean 39.55257 Adj R-sq 0.2164
C.V. 4030.55597    
Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –51.021927 12.92816544 –3.947 0.0001
C 1 95.661834 4.92207487 19.435 0.0001
C2 1 –6.519466 1.39538623 –4.672 0.0001
AGE1 1 –2.677928 1.98967973 –1.346 0.1784
AGE2 1 0.504584 1.50494723 0.335 0.7374
AGE3 1 2.300023 1.75520882 1.310 0.1902
AGE4 1 7.195003 2.87841105 2.500 0.0125
AGE5 1 17.556759 5.25934211 3.338 0.0009
AGE6 1 12.265962 6.44291540 1.904 0.0570

Table A1.12: Equation to predict  family  expenditure on furnishings  and equipment

Dependent variable: EXP07 Household exp furnishings and equipment

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 15038365497 1879795687.1 325.175 0.0001
Error 2588 14960908741 5780876.6387    
C Total 2596 29999274238      

Root MSE 2404.34537 R-square 0.5013
Dep mean 50.74763 Adj R-sq 0.4997
C.V. 4737.84771    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 74.061569 19.49818202 3.798 0.0001
C 1 –9.262028 7.42344396 –1.248 0.2123
C2 1 71.022839 2.10451319 33.748 0.0001
AGE1 1 –6.202127 3.00082311 –2.067 0.0389
AGE2 1 –2.741378 2.26975244 –1.208 0.2272
AGE3 1 –3.774053 2.64719547 –1.426 0.1541
AGE4 1 –18.595537 4.34120238 –4.283 0.0001
AGE5 1 –22.263281 7.93210840 –2.807 0.0050
AGE6 1 –26.759325 9.71716658 –2.754 0.0059

Table A1.13: Equation to predict  family  expenditure on services and operation

Dependent variable: EXP08 Household exp services and operation

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 578148768.31 72268596.039 97.303 0.0001
Error 2588 1922143316.9 742713.80095    
C Total 2596 2500292085.2      

Root MSE 861.80845 R-square 0.2312
Dep mean 38.32746 Adj R-sq 0.2289
C.V. 2248.54026    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 8.668115 6.98888696 1.240 0.2150
C 1 37.735827 2.66084349 14.182 0.0001
C2 1 1.323920 0.75433724 1.755 0.0794
AGE1 1 11.264333 1.07560866 10.473 0.0001
AGE2 1 4.563049 0.81356525 5.609 0.0001
AGE3 1 –4.312808 0.94885512 –4.545 0.0001
AGE4 1 –1.161190 1.55605136 –0.746 0.4556
AGE5 1 –5.196246 2.84316809 –1.828 0.0677
AGE6 1 –0.717015 3.48300056 –0.206 0.8369

Table A1.14: Equation to predict  family  expenditure on medical care and health

Dependent variable: EXP09 Household exp for medical care & health

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 214829232.43 26853654.053 48.427 0.0001
Error 2588 1435092253.9 554517.87243    
C Total 2596 1649921486.3      

Root MSE 744.65957 R-square 0.1302
Dep mean 31.90616 Adj R-sq 0.1275
C.V. 2333.90556    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –11.783322 6.03886114 –1.951 0.0511
C 1 35.089060 2.29914498 15.262 0.0001
C2 1 –4.241063 0.65179733 –6.507 0.0001
AGE1 1 0.314461 0.92939711 0.338 0.7351
AGE2 1 –1.722220 0.70297425 –2.450 0.0144
AGE3 1 –1.164768 0.81987366 –1.421 0.1555
AGE4 1 1.444603 1.34453142 1.074 0.2827
AGE5 1 0.035523 2.45668551 0.014 0.9885
AGE6 1 11.708384 3.00954313 3.890 0.0001

Table A1.15: Equation to predict  family  expenditure on transport

Dependent variable: EXP10 Household expenditure for transport

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 7327393814.9 915924226.86 98.202 0.0001
Error 2588 24138073597 9326921.7918    
C Total 2596 31465467412      

Root MSE 3054.00095 R-square 0.2329
Dep mean 101.94226 Adj R-sq 0.2305
C.V. 2995.81444    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –37.585936 24.76660268 –1.518 0.1292
C 1 228.607424 9.42926304 24.244 0.0001
C2 1 –27.112740 2.67315394 –10.143 0.0001
AGE1 1 5.926858 3.81164734 1.555 0.1201
AGE2 1 –5.050791 2.88304093 –1.752 0.0799
AGE3 1 –11.118444 3.36246930 –3.307 0.0010
AGE4 1 –8.499608 5.51419791 –1.541 0.1233
AGE5 1 –5.902054 10.07536892 –0.586 0.5581
AGE6 1 1.028285 12.34275091 0.083 0.9336

Table A1.16: Equation to predict  family  expenditure on recreation  and entertainment

Dependent variable: EXP11 Household exp recreation & entertainment

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 7197151254.3 899643906.78 165.396 0.0001
Error 2588 14076958666 5439319.4226    
C Total 2596 21274109920      

Root MSE 2332.23486 R-square 0.3383
Dep mean 92.16138 Adj R-sq 0.3363
C.V. 2530.59879    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –53.373127 18.91339754 –2.822 0.0048
C 1 217.812127 7.20080193 30.248 0.0001
C2 1 –23.072309 2.04139517 –11.302 0.0001
AGE1 1 –15.033315 2.91082319 –5.165 0.0001
AGE2 1 –2.548596 2.20167861 –1.158 0.2471
AGE3 1 –1.495611 2.56780146 –0.582 0.5603
AGE4 1 –14.547403 4.21100215 –3.455 0.0006
AGE5 1 10.942086 7.69421064 1.422 0.1551
AGE6 1 7.337570 9.42573181 0.778 0.4364

Table A1.17: Equation to predict  family  expenditure on personal care

Dependent variable: EXP12 Household expenditure for personal care

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 52804972.856 6600621.607 43.902 0.0001
Error 2588 389102637.79 150348.77813    
C Total 2596 441907610.65      

Root MSE 387.74834 R-square 0.1195
Dep mean 11.98250 Adj R-sq 0.1168
C.V. 3235.95641    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –9.639186 3.14446830 –3.065 0.0022
C 1 17.236915 1.19717747 14.398 0.0001
C2 1 –2.282059 0.33939446 –6.724 0.0001
AGE1 1 –0.777127 0.48394220 –1.606 0.1084
AGE2 1 –0.211488 0.36604257 –0.578 0.5635
AGE3 1 0.448940 0.42691274 1.052 0.2931
AGE4 1 2.952275 0.70010493 4.217 0.0001
AGE5 1 4.625901 1.27920969 3.616 0.0003
AGE6 1 5.217890 1.56708570 3.330 0.0009

Table A1.18: Equation to predict  family  expenditure on miscellaneous goods and services

Dependent variable: EXP13 Household exp miscel goods and services

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 8 3847235012.7 480904376.58 115.471 0.0001
Error 2588 10778328484 4164732.7993    
C Total 2596 14625563497      

Root MSE 2040.76770 R-square 0.2630
Dep mean 57.48666 Adj R-sq 0.2608
C.V. 3549.98502    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –38.096054 16.54972726 –2.302 0.0214
C 1 127.424825 6.30089373 20.223 0.0001
C2 1 –6.572336 1.78627522 –3.679 0.0002
AGE1 1 –8.044127 2.54704792 –3.158 0.0016
AGE2 1 –0.679598 1.92652750 –0.353 0.7243
AGE3 1 9.067251 2.24689476 4.035 0.0001
AGE4 1 20.516796 3.68473919 5.568 0.0001
AGE5 1 4.232591 6.73263951 0.629 0.5296
AGE6 1 2.876494 8.24776671 0.349 0.7273

Tables A2.1–A2.4—by number of children

Table A2.1: Equation to predict  total  family  consumption expenditure

Dependent variable: C

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 5 79260.28285 15852.05657 536.743 0.0001
Error 2556 75488.36381 29.53379    
C Total 2561 154748.64666      

Root MSE 5.43450 R-square 0.5122
Dep mean 0.65535 Adj R-sq 0.5112
C.V. 829.25026    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –0.171751 0.04364612 –3.935 0.0001
FY 1 0.497506 0.02000509 24.869 0.0001
FY2 1 –0.044994 0.00682863 –6.589 0.0001
KIDS 1 0.029819 0.00286539 10.406 0.0001
AGE5 1 0.188587 0.01769582 10.657 0.0001
AGE6 1 0.190076 0.02135712 8.900 0.0001

Table A2.2: Equation to predict  percentage  of family  consumption spent on food  at home (Engel estimator)

Dependent variable: LNPF

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 6 301157.22582 50192.87097 248.676 0.0001
Error 2616 528014.02216 201.84022    
C Total 2622 829171.24799      
 
Root MSE 14.20705 R-square 0.3632
Dep mean –1.81863 Adj R-sq 0.3617
C.V. –781.19471
Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –4.526564 0.71586715 –6.323 0.0001
LEFS 1 –1.304345 0.07266124 –17.951 0.0001
LEFS2 1 –0.195827 0.02215379 –8.839 0.0001
LNF 1 0.409725 0.10493586 3.905 0.0001
CKA 1 0.202719 0.69759254 0.291 0.7714
CKA5 1 0.620840 0.70010283 0.887 0.3753
CKA6 1 0.953416 0.70175869 1.359 0.1744

Table A2.3: Equation to predict  percentage  of family  consumption spent on ‘basic goods’—
basket 1

Dependent variable: LNPF1

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 6 327274.57209 54545.76202 400.011 0.0001
Error 2603 354947.09895 136.36078    
C Total 2609 682221.67104      

Root MSE 11.67736 R-square 0.4797
Dep mean –1.46739 Adj R-sq 0.4785
C.V. –795.79257

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –3.937285 0.58876105 –6.687 0.0001
LEFS 1 –1.270227 0.05993580 –21.193 0.0001
LEFS2 1 –0.162228 0.01827781 –8.876 0.0001
LNF 1 0.194447 0.08682561 2.240 0.0252
CKA 1 0.294111 0.57354411 0.513 0.6081
CKA5 1 0.600990 0.57551472 1.044 0.2965
CKA6 1 0.861361 0.57699128 1.493 0.1356

Table A2.4: Equation to predict  percentage  of family  consumption spent on ‘basic goods’— basket 2

Dependent variable: LNPF2

Analysis of variance
Source DF Sum of squares Mean square F value Prob>F
Model 6 307952.51429 51325.41905 451.287 0.0001
Error 2587 294222.41873 113.73112    
C Total 2593 602174.93302      

Root MSE 10.66448 R-square 0.5114
Dep mean –1.21237 Adj R-sq 0.5103
C.V. –879.64040    

Parameter estimates
Variable DF Parameter estimate Standard error T for  H0: Parameter=0 Prob  >|T|
INTERCEP 1 –3.773414 0.53725423 –7.024 0.0001
LEFS 1 –1.147634 0.05166367 –22.214 0.0001
LEFS2 1 –0.127190 0.01672153 –7.606 0.0001
LNF 1 0.176897 0.07952521 2.224 0.0262
CKA 1 0.477563 0.52379532 0.912 0.3620
CKA5 1 0.886487 0.52568019 1.686 0.0918
CKA6 1 1.185316 0.52704705 2.249 0.0246

[ Return to Top   Return to Section ]

Endnotes

  1. This is a model  that allows  the required change in consumption of a household with  an additional child  to vary across  different  goods. By contrast, the Engel and Rothbarth  methods assume  that the change in consumption captured by the proxies food and adult goods, respectively, are indicative of a corresponding change in the consumption of all other  household goods (see  Betson 1990  for a fuller description and an implementation of the method).
  2. Estimates were not calculated for households with  three  children at incomes of $400  per week as it was felt that this level  was unrealistically low, given  levels  of government income support assistance.
  3. Estimated at a household income level  of $700  per week.
  4. Betson (1990) included several  terms  relating to a wife’s  labour  force status in his equations but presents no results  for the resulting different  family types.

[ Return to Top   Return to Section ]

References

Apps, P. F. & Rees, R. 1995,‘Household production, the costs of children and equivalence scales’, Working  Paper No. 285, Faculty  of Economics  and Commerce and Economics  Program, Research School of Social Sciences, Australian  National University, Canberra.

Australian  Bureau  of Statistics (ABS) 1994, Household Expenditure Survey 1993–94, Unit Record  Tape, Canberra.

Australian  Bureau  of Statistics (ABS) 1995, Household Expenditure Survey 1993–94, Australia, User Guide, ABS Catalogue no. 6527.0, Australian  Bureau  of Statistics, Canberra.

Australian  Bureau  of Statistics (ABS) 1998a, Average Weekly Earnings of Employees, AUSSTATS, Table 6302.2.

Australian  Bureau  of Statistics (ABS) 1998b, Consumer Price Index, AUSSTATS, Table 6401.3B.

Australian  Institute  of Family Studies  1998,‘Costs of children update’ in Family Matters, Winter 1998, Australian  Institute  of Family Studies, Melbourne.

Betson, M. 1990, Alternative Estimates of the Cost of Children From the 1980–86 Consumer Expenditure Survey, Institute  for Research on Poverty, University of Wisconsin-Madison, Special Report  Series  No. 51.

Bradbury, B. 1994,‘Measuring the cost of children’, Australian Economic Papers, June 1994, pp. 120–138.

Bradbury, B. 1997, Family Size and Relative Need, unpublished Ph.D. thesis, School of Economics, University of New South Wales.

Carlucci, M. & Roberto  Zelli, R. 1998,‘Expenditure patterns and equivalence scales’, paper presented to the 25th General  Conference of the International Association for Research in Income  and Wealth, Cambridge, 23–29  August.

Deaton, S. & Muellbauer, J. 1986,‘On measuring child  costs: with  applications to poor countries’, Journalof PoliticalEconomy, 94(4), pp. 720–44.

Espenshade, T. 1972, The Cost of Children In Urban United States, Princeton University, Ph.D., University Microfilms, Ann Arbor, Michigan.

Espenshade, T. 1984, Investing in Children: New Estimates of Parental Expenditures, The Urban Institutes  Press, Washington DC.

Lee, D. 1988,‘Estimates of direct  expenditures on children in Australian: results  from the Household  Expenditure Survey  1984’, paper  presented at the Conference of the Australian Population Association, Brisbane, August 31–September 2, 1988.

Leser, C. 963),‘Forms of Engel functions’, Econometrica, 31(4), pp. 694–703.

Merz, J. & Faik, J. 1992,‘Equivalence scales  based  on revealed preference consumption expenditure microdata—the case of West Germany’, Discussion  Paper No. 3., Forschungsinstitut Freie Berufe, University of Lunenburg, Lunenburg.

Percival, R. 1994,‘Building STINMOD’s base population’, NATSEM STINMOD Technical Paper No. 4, November  1994.

Rothbarth, E. 1943,‘A  note on a method  of determining equivalent income for families  of different  composition’, Appendix 4 in C. Madge (ed.)  War-time Patterns of Saving and Spending, Cambridge, Cambridge University Press.

Tran Nam, Binh & Whiteford, P. 1990,‘Household equivalence scales: new  Australian  estimates from the 1984  Household  Expenditure Survey’ The Economic Record, September, pp. 221–234.

Valenzuela, M. R. 1999,‘Costs of children and living  standards in Australian  households’, Melbourne Institute Working  Paper no. 8/99, Melbourne Institute  of Applied  Economic and Social Research, the University of Melbourne, Melbourne.

Valenzuela, M. R. 1996,‘Engel Scales  for Australia, the Philippines and Thailand: a comparative analysis’, The Australian Economic Review, 2nd Quarter  1996, pp. 189–198.

van der Gaag, J. (nd),‘On measuring the cost of children’, Institute  for Research on Poverty, University of Wisconsin, Madison, mimeograph.

Watts, H. 1977,‘The iso-prop index: an approach to the determination of deferential poverty income thresholds’, in M. Moon and E. Smolensky (eds.) Improving Measures of Economic Well- being, New York, Academic Press.

Whiteford, P. 1985, A family’s needs: Equivalence scales, poverty and social security, Research Paper No. 27, Development Division, Department of Social Security, Canberra.

Williams, R., Price, D. & Venohr, J. 1993, Economic Basis for Updated Child Support Schedule— State of North Carolina, Report  prepared for the Office of Courts, North Carolina  Supreme Court, State of North Carolina.

Working, H. 1943,‘Statistical laws  of family expenditures’, Journal of the American Statistical Association, (38), pp. 43–56.

Wright, J. & A. Dolan, A., 1992,‘The use of HES data in distributional analysis’, paper  presented at the Conference of Economists, 8–10 July.

[ Return to Top   Return to Section ]

Content Updated: 5 June 2013