45The Calculation andUse of Poverty Linesin Australia1.IntroductionPoverty lines are benchmark income levels for various types of family incomeunit. If the income of a family incomeunit is less than the benchmark applicable to it, then the family is consideredto be in poverty. A family income unitis the family group normally supportedby the income of the unit. The mostcommon types of family income unit aresingle person, couple, single parent withdependants and couple with dependants.Poverty lines may be divided into twocategories: absolute poverty lines andrelative poverty lines.An absolute poverty line is defined forsome base period as the cost of buyinga basket of goods and services to meeta minimum set of basic needs (food,shelter, clothing, etc.). The consumerprice index (CPI) is a commonly usedmeasure of the cost of goods and services. An absolute poverty line is updatedto any later period by multiplying itsvalue in the base period by the ratio ofthe cost of the basket of goods andservices in the later period to its costin the base period.A relative poverty line is defined forsome base period as a fixed proportionof some measure of average income,for example, average earnings or incomeper head. It is updated to a later periodby multiplying its base-period value bythe ratio of the value of the averageincome measure in the later period toits value in the base period.Changes in the standard of living canbe measured by changes in income perThe author is grateful to Brian Parmenter, DainaMcDonald, Nellie Lentini, Tiana Ellis and Bee Fungfor valuable comments and advice on earlier draftsof this article.The Australian Economic Review 4th Quarter 1987David Johnson*Institute of Applied Economicand Social ResearchUniversity of Melbournehead relative to changes in the CPI, thatis, by changes in the real purchasingpower of average income. On this measurement, since 1 973-74 Australia’sstandard of living has risen by about 17per cent. Consequently relative povertylines, which are updated by changes inaverage income, have risen faster thanabsolute poverty lines, which areupdated by changes in the CPI. Similarly, if living standards were to fall, thenrelative poverty lines would fall fasterthan absolute poverty lines.The welfare of poor families can beimproved in two ways. Firstly, improvements in the general standard of livingwill raise the income of everyone, including the poor. Secondly, with no overallchange in the general standard of living,changes in income distribution can raisethe income of the poor, but only at theexpense of the non-poor.Looked at from a different angle thismeans that in the absence of changesin income distribution the proportion offamilies below a relative poverty line willbe constant. However, when general living standards rise these families will bebetter off. Similarly when living standardsfall the material welfare of all people,including those below a relative povertyline, will fall, though the proportion ofthe population below a relative povertyline will not change.In Australia the best known and mostwidely used poverty lines are thoseestablished by the 1973 Commission ofInquiry into Poverty (Henderson 1975).Since 1975 the Institute of Applied Economic and Social Research (IAESR) hasupdated the Henderson poverty lines ona quarterly basis using a measure ofchanges in average income, rather thancosts. That is, they are treated as relative poverty lines.The remainder of this article is organised as follows. In Section 2 the development of poverty lines is traced brieflyand in Section 3 the method of calcu-lating the Henderson poverty lines isdescribed. The uses and a limitation ofHenderson poverty lines are discussedin Section 4. Section 5 summarises themain points made in the article.2.The Development of PovertyLines Overseas and in AustraliaEarly attempts to establish poverty linesin England and the United States werebased on the budgetary approach inwhich poverty is defined in terms of thecosts of meeting minimal physicalrequirements for living.In 1887 Charles Booth examined theincome and status of families of schoolage children from a survey in London,and found that 30 per cent of the population were in what he regarded as acutepoverty. In 1902 Seebohm Rowntreepublished the findings of a survey ofhousing, occupation and earnings ofevery wage-earning family in York, England. Rowntree established a povertyline by estimating how much it wouldcost to buy a basket of necessities underthe headings of food, clothing, fuel andhousehold sundries. Bowley in 1912,defined a poverty line in terms of theactual spending habits of the populationrather than the budgeted standarddefined by Rowntree. The most recentexample of a budget-based poverty lineis that provided by Mollie Orshansky inthe United States in 1965. Orshanskydeveloped a poverty line based on theestimated cost of achieving a minimumlevel of nutrition. She took account ofother necessary expenses by multiplying the cost of food by a factor of three.(She assumed that poor people spentone-third of their income on food.)It was not until 1964 that the firstdetailed study of poverty in Australia wasundertaken (Henderson, Harcourt andHarper 1970). This was the first povertysurvey by the IAESR under the directionof Professor Ronald Henderson. The
46Calculation and Use ofPoverty Lines in Australia1964 study set a poverty line for astandard family of two adults and twochildren with the head of the family working, at the level of the then basic wageplus child endowment. This amountedto 33 per week after paying personalincome tax or 56.5 per cent of seasonally adjusted average weekly earnings inAustralia. (Average weekly earnings areused as a measure of average income.)Henderson, Harcourt and Harper (p. 1)state in justifying this selection:This is a definition of poverty so austere as, we believe, to make it unchallengeable. No one can seriously arguethat those we define as being poorare not so.We chose this basic-wage contentof the poverty line because of itsrelevance to Australian concepts ofliving standards-the basic wage beingthe lowest wage which can be paidto an unskilled labourer on the basisof, in the famous phrase of Mr JusticeHiggins, ‘the norTal needs of an average employee regarded as a humanbeing living in a civilised community.’This poverty line also has internationalrelevance since, in its relationship toaverage earnings, to average incomesand to basic social service rates, it iscomparable to the poverty lines thathave been adopted in some surveyscarried out overseas, particularly inthe United Kingdom, the United Statesof America and Denmark.In this argument the authors appearto support their poverty line using bothabsolute and relative notions of poverty.At around the same time researchersconcerned with poverty in westerncountries were increasingly using therelative approach which recognised that,in a country such as Australia, even thepoorest people receive sufficient incometo exceed the minimum requirements offood, clothing and shelter for health. Yetsome people are still regarded as poorrelative to others. The starting point forthe measurement of poverty using thisapproach is the continually moving average standards of the community. Thepoor are those who fall far below theseaverage standards.In 1973 Professor Henderson wasappointed chairman of a Commission ofInquiry into Poverty, or the Hendersonpoverty inquiry. The inquiry updated thebenchmark income, from the 1964study, for the standard family to theSeptember quarter 1973 to account forincreases in average weekly earnings inthe intervening period. The new povertyline, now known as the Henderson poverty line, was set at 62.70 per weekafter tax, also equal to 56.5 per centof average weekly earnings. By usingaverage weekly earnings for updating,the Henderson inquiry confirmed theHenderson poverty lines as a relativemeasure of poverty.Poverty lines for family income unitsother than the standard family werederived from the benchmark, using a setof equivalence scales. EquivalenceTable 1 Equivalence Scales Using a Couple Plus Two Dependant ChildrenWhere the Head Works as the StandardType of family income unitIHead working184.108.40.206.220.127.116.11.20.21 .22.CoupleCouple plus 1Couple plus 2Couple plus 3Couple plus 4Couple plus 5Couple plus 6Couple plus 7Couple plus 8Couple plus 9Couple plus 1O Single personSingle parent plusSingle parent plusSingle parent plusSingle parent plusSingle parent plusSingle parent plusSingle parent plusSingle parent plusSingle parent plusSingle parent plus12345678910 Head not workingAll costs(11All costslesshousing(2)(3)All costslesshousing(4)0.71 1.97122.1 3670.52090.64750.77410.90071.0259.I 453,2734,3928.5209.6489.78850.61 150.75540.89931.04321.I 80.67340.80000.92521.04461.17271.29211.42011.5482I .6a780.53240.68350.82730.971 9211.62010.431 .70501.84170.25760.39140.51800.64460.771 20.89641.01581.14391.26331.39141.5194All costs18.104.22.168.22.214.171.124.9.0.1.I
47Calculation and Use ofPoverty Lines in Australiascales relate the income levels requiredby different types of family income unitto attain a given standard of living. Conventionally, the standard family has ascore of 1 on the equivalence scale.Hence, some other income unit with ascore of 1.2, say, requires, accordingto the scale, 20 per cent more incomethan the standard family in order to attainthe same standard of living.3.Calculation and Updating of theHenderson Poverty Lines3.1Updating the Benchmark IncomeFrom the September quarter 1973 untilthe December quarter 1980, the IAESRpublished updates of the poverty lines,maintaining the value for the standardfamily at 56.5 per cent of averageweekly earnings. The updates for a particular quarter were obtained by takingthe poverty line values in the Septemberquarter 1973, and multiplying them bythe ratio of average weekly earnings inthe particular quarter to average weeklyearnings in the September quarter 1973.Table 2 Simplified Costs for FamilyIncome Unit Members Related toTheir Employment orDependency StatusFamily incomeunit memberNumber ofpointsHead not working13.0Spouse working18.5Spouse not working9.5DeDendant child7.5Source:Henderson (1975), AppendixF, p. 356.The Australian Economic Review 4th Quarter 1987There were two main problems withthe use of average weekly earnings forupdating. Firstly, whilst the poverty linesrefer to income after personal incometax has been paid, average weekly earnings include tax. This implies that thepoverty lines will be maintained as constant proportions of before-tax income.It does not follow that they will be maintained as constant proportions of aftertax income (which would be necessaryto maintain consistency with the aftertax definition of the Henderson povertylines). The second problem is that average weekly earnings, which measure theaverage income of wage and salaryearners only, are a narrow measure ofincome. A change in the fortunes ofwage and salary earners relative to theself-employed, or to those on socialsecurity benefits, will affect the (relative)poverty of the poor but will not affectpoverty lines updated by movements inaverage weekly earnings.Since 1976 the Australian Bureau ofStatistics (ABS) has published quarterly,seasonally adjusted estimates of household disposable income in its QuarterlyEstimates of National Income andExpenditure (Catalogue No. 5206.0).Household disposable income is the totalincome of households from all sources(that is, it includes wages, social security payments, rents, interest, dividends,and earnings from private business) lesspersonal income tax payments. The ABSalso publishes population estimates ona quarterly basis (in Australian Demographic Statistics, Catalogue No.3101 .O). Consequently, it is possible tomake quarterly estimates of per capitahousehold disposable income. Per capita household disposable income is amore appropriate series than averageweekly earnings for updating the povertylines because it is calculated after taxand it is a wide measure of income.From the March quarter 1981 the IAESRhas updated the Henderson poverty lineson the basis of movements in per capitahousehold disposable income.However, the use of per capitahousehold disposable income as themeans of updating the poverty lines stillhas an important problem. Published dataon both household income and popula-Table 3 Standard Costs Used in the Adjustment of Incomefor Families of Varying SizeFamily sizeI1123456789101112 Source:Number of points for:Housing 4.2IFuel, power, ancillary enderson (1975), Appendix F, p. 40.044.0
48Calculation and Use ofPoverty Lines in Australiation are estimates, subject to revision.Consequently the values for per capitadisposable income and therefore for thepoverty lines will also be estimates. Asmore information becomes available theABS may revise the estimates for, eitheror both, population and household disposable income for some or all pastquarters. Whenever the estimates ofpopulation or household disposableincome are revised it is necessary to reestimate the poverty lines.3.2The Equivalence ScalesEquivalence scales describe the relationships between the living costs of astandard family income unit and othertypes of family income unit. The equivalence scales calculated by the IAESR,are shown in Table 1. The table showsvalues on the equivalence scale fortwenty-two types of family income unit.Column 1 shows the equivalence scalefor family income units in which the headof the family is workjng. The index isset at 1 .OOOO for the standard family,that is, for a couple plus two dependantchildren, where the head of the familyis working. The value on the scale fora couple with no dependant children is0.7122 and the value for a single working parent with two dependant childrenis 0.8273.Column 2 shows the equivalence scale for all costs except housingcosts, again for family income unitswhere the head of the family is working.The separate scale for the situation inwhich housing costs are excluded enables the effects of housing costs to beidentified. Columns 3 and 4 show equivalence scales for situations in which thehead of the family is not working. Thevalues on these scales are lower thanthe values for comparable family incomeunits shown in columns 1 and 2 becausein these scales certain costs associatedwith working, such as travel and extracosts of meals and clothing, have beenexcluded.Table 1 has been calculated usingdata from a study undertaken in NewYork in 1954 which obtained verydetailed costs for families in differentcircumstances. It may seen incongruousthat data derived from foreign sourcesand from such an old study should beused to determine equivalence scales inAustralia for 1 987.Manning (1 982)hasdefended the continued use of theequivalence scales based on the NewYork data by pointing out that divergences with alternative scales are notlarge and that similar scales have beenderived in different ways from differentbase data. Whiteford (1 985)conductedan appraisal of sixty sets of equivalencescales against a number of criteria (consistency, comprehensiveness, empiricalplausibility and relevance, and publicacceptability). He concluded that no individual scale performed well in allrespects. Overall the Henderson equivalence scales performed no worse thanthe other scales.The equivalence scales shown inTable 1 were derived using the information in Tables 2 to 4.Table 2 showscosts, derived from the New York data,Number of pointsType of family income unitHead workingIHead not workingother thanhousingother thanhousing1. Couple2. Couple plus 13. Couple plus 24. Couple plus 35. Couple plus 46. Couple plus 57. Couple plus 68. Couple plus 79. Couple plus 810. Couple plus 91 1 . Couple plus 1 O 8.046.855.664.372.681.589.898.7107.6117.312. Single person13. Single parent plus 114. Single parent plus 215. Single parent plus 316. Single parent plus 417. Single parent plus 518. Single parent plus 619. Single parent plus 720. Single parent plus 821. Single parent plus 922. Sinale- .Darent DIIJS I O .550.560.570.580.590.099.5109.01 1 5.6
49Calculation and Use ofPoverty Lines in Australiameasured in points for various membersof a family income unit. The table showsthat a working head incurs 20 pointsand a dependant child incurs 7.5 points.Table 3, also derived from the NewYork data, shows housing, fuel, powerand ancillary costs (again in points) forall family income units. These costs risewith increases in the number of members of the family. For instance a familycomposed of a single person incurs 12.1points for housing and 4.9 points forfuel, power and ancillary costs whereasa family consisting of a couple plus onedependant incurs 14.5 points for housing and 8.0 points for fuel, power andancillary costs.The information from Tables 2 and 3is used to calculate a table of equivalence costs. Table 4 shows a table ofequivalence costs for twenty-two incomeunits for two situations of employmentstatus (head working and head not working), where all costs are considered andwhere costs other than housing costsare considered. The entry for a couplewith the head working (49.5 points, seeline 1, column 1) is the sum of 20 pointsfor a working head and 9.5 points for anon-working spouse (from Table 2) plusa total of 20 points for housing, fuel,power and ancillary costs (from Table3). Similarly the total number of pointsfor a family unit consisting of a nonworking single parent with three dependant children for costs other than housing (line 15, column 4) is 44.8, madeup of 9.3 points for fuel, power andancillary costs plus 13.0 points for thenon-working single parent and 22.5 (thatis, 3 X 7.5) points for three dependantchildren.Table 1 is obtained from Table 4 simply by setting the value for the standardfamily (a couple plus two dependant children), on the scale where the head isworking and all costs are included, at1.OOOO and dividing all of the other cellsin Table 4 by 69.5 (the number of costThe Australian Economic Review 4th Quarter 1987points for the standard family).3.3Calculation of Poverty LinesThe equivalence scales are used to calculate poverty lines for all of the incomeunits described in Table 1 starting withthe benchmark income of 62.70 forthe standard family in the Septemberquarter of 1973.For example from Table 1 the equivalence value for a family income unitconsisting of a couple where the headworks and where all costs are includedis 0.71 22. Accordingly the poverty linefor this family income unit in the September quarter 1973 was 44.65 perweek (0.7122 X 62.70). Similarly fora single non-working parent with threedependant children the poverty line notincluding housing costs was 40.42(0.6446 X 62.70) in the SeptemberYear. 87-88quarter 1973.The poverty lines may be updatedfrom the September quarter 1973 toany quarter up to the current quarterusing data on per capita household disposable income. Table 5 shows estimates of per capita householddisposable income derived from the latest ABS estimates of population andhousehold disposable income. Per capita household disposable income was 48.50 in the September quarter of1973-74 and 220.60 in the September quarter of 1987-88. The povertyline for any family income unit for anyparticular quarter may be obtained bymultiplying the ratio of the per capitahousehold disposable income for thequarter to per capita household disposable income in the September quarter1973, by the product of the benchmarkincome and the family income unit’s 167.6183.3196.821 1.7 Sources: ABS, Quarterly Estimates of National Income and Expenditure, SeptemberQuarter 1987, Catalogue No. 5206.0; Australian Demographic Statistics,December Quarter 1986, March and June Quarters 1987, Catalogue No.3101 .O; and projections by the IAESR.
50Calculation and Use ofPoverty Lines in Australiaon the equivalence scale. (The productof the benchmark income and the valueon the equivalence scale is the povertyline for the income unit in the Septemberquarter 1973.) The poverty line for thestandard family in September 1987 is 285.20 (62.70 X 1.000 X (220.6/48.5)), and for a couple with no dependants where the head works is 203.1 0(62.70 X 0.7122 X (220.6/48.5)).The general formula for calculating thepoverty lines can be expressed as:Poverty line forincome unit x 62.70 Xin quarter yper capita householddisposable incomevalue onin quarter yequivalence scale xpercapitahouseholdfor incomeunitdisposable income inSept. quarter 1973The poverty lines are published bythe IAESR in the quarterly newsletter,'Poverty Lines : Australia'. Table 6 pre-Table 6 Poverty Lines: Australia, September Quarter 1987aType of family income unitAll costsincludinghousingcostsother thanhousing( 1( 1Head in workforceCoupleCouple with 1 childCouple with 2 childrenCouple with 3 childrenCouple with 4 07.2248.3289.373.5111.6147.7183.8219.9personparent withparent withparent withparent withHead not inCoupleCouple withCouple withCouple withCouple ildrenchildrenchildrensents poverty lines for the Septemberquarter 1987 for the most common family income units.3.4Re-estimating the Poverty LinesPast estimates of the poverty lines aresuperseded with the release of a newissue of the quarterly poverty linesnewsletter. As mentioned earlier thepoverty lines are based on ABS estimates of population and household disposable income which are subject torevision. Therefore it is not appropriateto use poverty lines from different issuesof the poverty lines newsletter. To enable valid comparison of the poverty linesat different points in time, the IAESRprovides, with each issue of the newsletter, information like that given in Tables5 and 6. This is sufficient for the calculation of poverty lines for all quartersfrom the September quarter 1973onwards. For instance, if you wish toknow the poverty line for the Junequarter 1981 for any household type,multiply the current value of its povertyline, by the ratio of per capita householddisposable income in the June quarter1981 to that in the current quarter; forexample, the poverty line for a standardfamily in June 1981 would be 285.20X (122.7/220.6), or 1 rsonparent withparent withparent withparent with1234childchildrenchildrenchildren(a) Based on the preliminary estimate of seasonally adjusted householddisposable income from all sources after taxes per head per week forSeptember quarter 1987 of 220.60 published in ABS, QuarterfyEstimates of National Income and Expenditure, Catalogue No. 5206.0.4.The Uses and a Limitation ofthe Henderson Poverty LinesManning (1982) has stated that thereare three main uses to which the (Henderson) poverty lines are put. These are(i) to assess the relative incidence ofpoverty among differentgroups,social(ii) to measure changes in incidence ofpoverty over time, and(iii) to act as a standard of adequacyfor social security payments.
51Calculation and Use ofPoverty Lines in AustraliaThe most reliable estimates of therelative extent of poverty between socialgroups at a point in time have beenderived from the Household Incomeand Expenditure surveys carried out bythe ABS. Results from the most recentsurvey in 1986 have not yet beenpublished. However, the 1981-82Household Income survey has beenwidely analysed. For instance Gallagherand Foster (1 986) estimated the percentage of adult income units (sincefamilies headed by juveniles areexcluded from the analysis the resultsrefer only to adult income units) of various compositions living below the Henderson poverty line. Some of theirfindings are shown in Table 7. The finalcolumn of Table 7 shows the estimatedpercentage of each adult income unitbelow the poverty line. The table showsthat whilst 12.4 per cent of the totalpopulation were living below the povertyline, the extent of poverty was verymuch higher among some adult incomeunits. For instance, over 50 per cent ofall single parent families with two or morechildren were below the poverty line.Readers interested in a detailed analysisof the extent of poverty among varioussocial groups should consult the articleby Gallagher and Foster.Manning's second suggested use hasalso been explored by researchers. Gallagher (1 985) compared the incidenceof poverty among different income unitsover time. Poverty was measured by theproportion of adult income units estimated to be below the Henderson poverty lines in 1972-73, 1973-74, 197879 and 1981-82. These four dates referto the years in which surveys gatheringdata on the income of Australian adultincome units were conducted. Resultsfrom the study are reported in Table 8.The 1985 study used a methodologywhich is slightly different to that whichformed the basis for Table 7 and,according to Gallagher and Foster, infeThe Australian Economic Review 4th Quarter 1987rior. The results are therefore not strictlycomparable. The cells in Table 8 showthe percentage of each adult income unitestimated to be below the poverty linein each of the four years. According toTable 8, the overall incidence of povertychanged little between 1972-73 and1981-82. There were 10.2 per cent ofTable 7 Income Unitsa with Incomes BelowEquivalent Disposable Income of 6 881 in 1981-82(11(2)Composition ofincome unit(3)Totalnumber of unitsI(4)(5)Number of unitsUnits belowbelow detailedb poverty line asHenderson povertyline'000Iper cent1111101234 2404.9142.7126.96.36.19961 .O54.344.523.87.915.038.056.873.393.22222201234 76.59.830.5Total single parent units261.9130.549.8Total couple parent units1490.1127.18.5Total population5290.7671.112.4Notes:(a) Excludes income units headed by a person who is self-employed andincome units not headed by an adult.(b) The Henderson poverty inquiry presented two sets of equivalencescales: detailed scales in which the age and sex of members of thefamily income unit affected the value of the unit on the scale, andsimplified scales in which the age and sex of members of the familyincome unit did not affect the value of the unit on the scale. Since theage and sex of members of a family income unit did not make muchdifference to the overall pattern, the IAESR has used the simplifiedscales in the calculation of the poverty lines. Only the simplified scalesare described in this article. However in this table poverty has beenmeasured by calculating the proportion of family income units belowHenderson poverty lines calculated using the detailed scales.Source: Adapted from Gallagher and Foster (1986, Tables 1 and 2).
52Calculation and Use ofPoverty Lines in Australiaall adult income units, excluding incomeunits headed by self-employed persons,below the Henderson poverty line in1972-73,and 10.3 per cent in 198182. However, the characteristics of thepoor did change. In the earlier years ahigher proportion of single person adultincome units were in poverty (1 8.3 percent in 1972-73declining to 10.6percent in 1981 -82)whereas in later yearsthere was a higher proportion of adultincome units with dependant children inType of income unitpoverty. The percentage of single parentincome units in poverty rose from 33.8per cent in 1972-73to 46.2 per centin 1981-82and the percentage of couple income units in poverty rose from3.4 per cent in 1972-73 to 6.7 percent in 1981 -82.The comparison of the incidence ofpoverty over time using the Hendersonpoverty lines should be treated with caution. In the introduction it was pointedout that the welfare of the poor can bePercentage of income units belowdetailed Henderson poverty lines in: Single parent with1 child2 or more childrenTotalCouple with0 children1 child2 children3 or more childrenTotalAll adult income units(excluding the self-employed)All persons1972-731973-741978-791981 710.211.19.310.38.2n.c.188.8.131.52'n.c.1 1 .2dNotes:n.s. - not stated; n.c. - not calculated.(a) Excludes income units headed by a person who is self-employed andincome units not headed by an adult.(b) See note (b) in Table 7.(c) Percentage for a couple with one or two children.(d) As stated in Section 4,Tables 7 and 8 are not calculated on the samebasis and are not strictly comparable.Some:Adapted from Gallagher (1 985,Table 1 ).raised by increases in general standardsof living, or by income redistributions infavour of the poor at an unchanged general standard of living. A limitation of theHenderson poverty lines is that they donot indicate changes in the real purchasing power of poverty line incomes.That is, they do not indicate changes inthe general standard of living. Estimatesof the change in the extent of povert
cable to it, then the family is considered to be in poverty. A family income unit . A relative poverty line is defined for some base period as a fixed proportion of some measure of average income, . McDonald, Nellie Lentini, Tiana Ellis and Bee Fung for valuab