Comparison of Alternative Models of Household Equivalence Scales: The Australian Evidence on Unit Record Data (original) (raw)

The eyuivtilence scule is un importtint parumeter in view of its use in welftire comparisons. This paper investigates sensitivity qf the scale to models. methotls tiritl commotlities. The estimtitioris rringe from sin& rqutitioti. OLS iti case of Engel und Rothburth models to mti.rirnum likrliiiooil iri cmr of drmogruphic JemunJ sytrrns. The rstimiitetl scrilrs jrotn clrtncind smtents genercillv lie between the E n p 4 und Rothhiirtlr scdes. The Equivtilence Sctile E.rtictnrss Ir,vpothesis. cori.sitlerucl tircessciNj7~r interpretutioti ofthe scule cis chill i'ost. sirflers only t i mild rejrctioti. The Airstrtilitiri rviilericr .sii~q,qrsts tlirit rirrik I~L Y I ilrmtrntl moclels. rejected by tlir cltitti. inti! e.rtigx,rr(ite c~liilil cm.vt.r. I Iiitrciclirction The Equivalence Scale seeks to answer questions such as: how much income does a household with two adults and one child need, in relation to a childless couple. to enjoy the same level of 'welfare' as the latter? Notwithstanding severe , conceptual and methodological problems in any interpersonal welfare comparisons (see Sen 1987). let alone those involving the household rather than the individual as the unit of decision making. such comparisons are inevitable in major policy exercises e.g., in the measurement of inequality and poverty, in studying the effects of a set of tax changes on the welfare levels of different house-' holds, in calculating the compensation that a * The research in this paper was supported by P grant from the Australian Research Council. The authors are grateful to Mr Elkana Ngwenya for considerable programming support. Helpful comments from two anonymous referees and from seminar participants at Sydney, Monash. Rome, Verona and Florence Universities on an earlier version are gratefully acknowledged. The disclaimer applies. household with a child requires for the additional cost of that child. etc. The calculation of household equivalence scales ha.. a long and chequered history originating with the pioneering study of Engel (1895) on Belgian working class expenditure data. There has. recently, been renewed and substantial interest in the subject-see Browning (1992) and Nelson (1993) for up-to-date reviews of the literature. Following Pollak and Wales (1979). the recent literature has generated considerable controversy on the interpretation and use of equivalence scales, as conventionally calculated, in welfare comparisons across households (see, also. Fisher 1987, Blackorby and Donaldson 1994). Pollak and Wales (1979) argue that conventional budget data does not provide sufficient information for rneaningful calculation of equivalence scales of the sort that one requires in welfare comparisons. Deaton and Muellbauer (1986). Binh and Whiteford (1990). Pashardes (I99 I), among others, disagree with the Pollak and Wales view and see virtue in conventional calculations on 'cost of children'. I 1998. The Economic Society of Australia.