Skip to main content

Advertisement

SpringerLink
Log in

Was Pro-Poor Economic Growth in Australia for the Income-Poor? And for the Multidimensionally-Poor?

  • Published:
Social Indicators Research Aims and scope Submit manuscript

Abstract

We investigate the pro-poorness of Australia’s strong economic growth in the first decade of the twenty-first century using anonymous and non-anonymous approaches to the measurement of pro-poor growth. The sensitivity of pro-poor growth evaluations to the definition of poverty is evaluated by comparing the results for the standard income-poverty measure with those based on a multidimensional definition of poverty. We find that Australian growth in this period can be only categorized as pro-poor according to the weakest concept of pro-poorness that does not require any bias of growth towards the poor. In addition, our results indicate that growth was clearly more pro-income poor than pro-multidimensionally poor. Counterfactual distribution analysis reveals that differences in the distribution of health between these two groups is the non-income factor that most contributes to explain this result.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2

Similar content being viewed by others

Notes

  1. Ranking derived using the series of GDP at purchasing power parity per capita elaborated by the OECD and available at http://stats.oecd.org/Index.aspx.

  2. Some of the results presented in this paper were already discussed in Azpitarte (2013). This is an improved and augmented version with new results that were not available by the time the first version was written.

  3. Because of the changes in the methodology used by the ABS, the estimates for 2007–2010 are not directly comparable with those for previous years. The comparison of the figures for 2000 and 2010 suggests an even larger increase than the one observed for the period 2000–2007.

  4. Groll and Lambert (2012) show using simulation analysis with parametric distributions that pro-poor growth generally leads to a decline in relative inequality. There exist, however, pro-poor growth patterns that exacerbate inequality.

  5. For a discussion on the development of social exclusion agenda in Australia and its relationship with the policy initiatives in Europe and the UK see Scutella et al. (2009a).

  6. As it is common in the pro-poor literature, we will assume that the poverty line remains constant in real terms over time. Deutsch and Silber (2011) analyse the pro-poorness of growth in Israel between 1990 and 2006 considering alternative ways of defining the poverty line and concepts of pro-poor growth. They find that although these choices affect the results, the overall characterization of the growth pattern is robust to these choices.

  7. For the FGT α family the individual poverty function is equal to \(\theta (y,z)=(\frac{z-y}{z})^{\alpha }\), where α is the parameter of inequality aversion. When α is set equal to 0,1, or 2, this expression leads to the headcount measure, the poverty gap ratio and the severity of poverty index, respectively. In the case of the Watts index the poverty function is given by \(\theta (y,z)=Ln(\frac{z}{y}).\)

  8. In particular, this Theorem covers any poverty measure P whose individual poverty function is decreasing and convex. The headcount index clearly fails to satisfy this property.

  9. These necessary conditions correspond to the case of positive income growth. This is precisely the type of growth observed in Australia for the period under analysis so we decided not to discuss the case of negative growth. For more on this see Essama-Nssah and Lambert (2009).

  10. This is defined as the area under the GIC up to the headcount index divided by the headcount measure, and it can be expressed as \(\frac{1}{H}\int_{0}^{H}g(p)dp\).

  11. When P is set equal to the Watts index of poverty, then the \(PEGR=\frac{1}{H}\int_{0}^{H}g(p)dp\), where the term on the right hand side is the pro-poor growth index proposed by Ravallion and Chen (2003).

  12. Grimm’s original formulation is in terms of the initial income of individuals. However, the framework is still valid when \(\Upomega _{t-1}\) refers to any other welfare indicator.

  13. Differently to the anonymous pro-poor growth measures, to the best of our knowledge no formal relationship between the anonymous measures and the variation of a particular poverty measure has been established in the literature.

  14. For a detailed description of the HILDA sample see Wooden and Watson (2007).

  15. The use of weights is particularly necessary for the longitudinal analysis due to the non-randomness of non-response patterns. A discussion on this issue is presented later in Section 5.2.

  16. Estimation results for alternative values of θ not presented here are available upon request.

  17. Interestingly, this shift did not lead to a significant change in social spending. This does not necessarily mean there was no welfare state retrenchment. Indeed, as Korpi and Palme (2003) show, replacement rates in the sickness and unemployment insurance programs in Australia substantially declined for the period 1975–1995. Unfortunately, no similar evidence is available for more recent periods.

  18. A key feature of the Australian Social Security System is the categorization of welfare payments into two groups: pensions and allowances. Pensions are meant for long-term support for those who are not expected to sustain themselves through paid work including mature-aged individuals and people with long-term health conditions and disability. Allowances are designed to be a transitional payment for those with capacity to work but are temporarily out of the labour market. Relative to pensions, allowances are paid at lower rates, face tighter means-tests and have more participation requirements. For more details on the structure of cash-transfers and its recent evolution see Herscovitch and Stanton (2008) and Australian Senate (2012).

  19. These and all the other estimates of pro-poor growth measures presented in this section were computed using the Distributive Analysis Stata Package developed by Araar and Duclos (2007).

  20. From Kakwani and Son (2008) we know that the growth rate in the mean, γ, is always less than the threshold \(\bar{\gamma}\) defined by these authors to characterize absolute pro-poor growth. Therefore, PEGR < γ implies that growth was not absolute pro-poor either.

  21. The choice of the cent cut-off points is completely arbitrary. Alternative thresholds for the bottom, middle, and top parts were considered and the main conclusions from the analysis remained unaltered.

  22. The extent to which this increase was due to the changes in the Australian social policy described above is an interesting issue that has not been analyzed yet.

  23. The comparison of our results with those from the literature on top income shares must be taken cautiously. The unit of analysis in this literature is usually the individual as results are based on records of personal income tax. Furthermore, the income variable used in these studies is gross income before tax. The figures presented here, however, refer to the distribution of disposable income and were derived by assigning each individual the equivalent income of her household.

  24. These authors analyze long-run trends using income tax data for the period from 1921 up to 2003.

  25. For more on these issues see the “Appendix”.

  26. Importantly, the larger growth of the income-poor could just be a consequence of the greater income mobility among those at the bottom of the distribution. To the best of our knowledge no methodological framework capable of distinguishing the effects of growth and income mobility on the pro-poorness of growth has been proposed yet. We propose a procedure that allows us to control for the income-mobility due to normal life-cycle income growth and the initial income conditions. As shown in the “Appendix”, we find that the main conclusions from the pro-poor analysis do not change when we control for these sources of mobility.

  27. All the results presented in this section correspond to the 15 % cut-off. Robustness checks carried out using the 5, 10, 20, 25, and 30th percentiles as thresholds yield similar results available upon request.

  28. For both the Oaxaca–Blinder and the DFL regression decompositions, e(S) is obtained setting all the other coefficients but those of the covariates in S equal to zero.

  29. Notice the aim of this analysis is to evaluate the contribution of the differences in the distribution of observed characteristics between the two poor groups to explain the growth gap. The econometric specifications are simply thought to identify the statistical association between individuals’ characteristics and benefits from growth. Issues of endogeneity and selection bias were not addressed which implies that no causal relationship can be assessed from our results.

  30. These are Sydney, other regions of New South Wales, Melbourne, other areas of Victoria, Brisbane, rest of Queensland, Adelaide, other regions of South Australia, Perth, rest of Western Australia, Tasmania, Northern Territory, and Australian Capital Territory.

  31. The general, physical, and mental health indices take values between 0 and 100 and are based on the SF-36 Health Survey included in HILDA.

  32. The results of the multiple regressions run to evaluate the contribution of each group of characteristics are not presented in the “Appendix”, but are available upon request.

  33. Note this conterfactual exercise provides an estimate of the income gains of the multidimensionally-poor assuming the characteristics of the income-poor. This implies that differences in returns between these two groups are weighthed by the characteristics of the income-poor. To check the robustness of the results we also estimated the alternative decomposition which weights differences in returns by the characteristics of the multidimensionally-poor. The results of this exercise, available upon request, are consistent with the ones presented here.

  34. The incidence of people with indigenous background is slightly higher among the multidimensionally-poor (3.1 vs. 2.3 %).

References

  • Araar, A., & Duclos, J. Y. (2007). DASP: Stata modules for distributive analysis. Statistical Software Components S456872, Boston College Department of Economics.

  • Atkinson, A. B. (1987). On the Measurement of poverty. Econometrica, 55, 749–764.

    Article  Google Scholar 

  • Atkinson, A. B. (2005). Top incomes in the UK over the twentieth century. Journal of the Royal Statistical Society, Series A, 168(February), 325–343.

    Google Scholar 

  • Atkinson, A. B., & Leigh, A. (2005). The distribution of top incomes in New Zealand. Australian National University, CEPR Discussion Paper 503.

  • Atkinson, A. B., & Leigh, A. (2007). The distribution of top incomes in Australia. Economic Record, 83, 247–261.

    Article  Google Scholar 

  • Atkinson, A. B., Leigh, A. (2013). The distribution of top incomes in five anglo-saxon countries over the long run. Economic Record. doi:10.1111/1475-4932.12004.

  • Australian Bureau of Statistics. (2006). Socio-economic indexes for areas (SEIFA)—Technical paper, 2006, cat. 2039.0.55.001, ABS Canberra.

  • Australian Bureau of Statistics. (2011). Household income and income distribution, cat. no. 6523.0, ABS, Canberra.

  • Australian Senate. (2012). Submission to the senate inquiry on the adequacy of the allowance payment system for job seekers and others. Prepared by Department of Education, Employment and Workplace Relations, the Department of Families, Housing, Community Services and Indigenous Affairs, the Department of Human Services and the Department of Industry, Innovation, Science, Research and Tertiary Education.

  • Azpitarte, F. (2013). Has economic growth in Australia been pro-poor? Forthcoming In P. Smyth (Ed.), Inclusive growth: An Australian approach. NSW: Allen & Unwin.

  • Blinder, A. S. (1973). Wage discrimination: Reduced form and structural estimates. Journal of Human Resources, 8(4), 436–455.

    Article  Google Scholar 

  • Bourguignon, F. (2010). Non-anonymous growth incidence curves, income mobility and social welfare dominance. Journal of Economic Inequality, 9(4), 605–627.

    Article  Google Scholar 

  • Buhmann, B., Rainwater, L., Schmaus G., & Smeeding, T. J. (1988). Equivalence scales, well-being, inequality, and poverty: Sensitivity estimates across ten countries using the luxembourg income study (LIS) database. Review of Income and Wealth, 34, 115–142.

    Article  Google Scholar 

  • Datt, G., & Ravallion, M. (1992). Growth and redistribution components of changes in poverty measures: A decomposition with applications to Brazil and India in the 1980s. Journal of Development Economics, 38, 275–295.

    Article  Google Scholar 

  • Deutsch, J., & Silber, J. (2011). On various ways of measuring pro-poor growth. Economics: The Open-Access, Open-Assessment E-Journal, 5, 2011–2013. http://dx.doi.org/10.5018/economics-ejournal.ja.2011-13.

  • DiNardo, J., Fortin, N., & Lemieux, T. (1996). Labor market institutions and the distribution of wages, 1973–1992: A semiparametric approach. Econometrica, 64, 1001–1044.

    Article  Google Scholar 

  • Duclos, J.-Y (2009). What is pro-poor?. Social Choice and Welfare, 32(1), 37–58.

    Article  Google Scholar 

  • Essama-Nssah, B., & Lambert, P. J. (2009). Measuring pro-poorness: A unifying approach with new results. Review of Income and Wealth, 55, 752–778.

    Article  Google Scholar 

  • Foster, J., & Shorrocks, A. F. (1988). Poverty orderings. Econometrica, 56, 173–177.

    Article  Google Scholar 

  • Foster, J., Greer, J., & Thorbecke, E. (1984). A class of decomposable poverty measures. Econometrica, 52, 761–765.

    Article  Google Scholar 

  • Goodger, K., & Larose, P. (1999). Changing expectations: Sole parents and employment in New Zealand. Social Policy Journal of New Zealand, 12(July), 53–70.

    Google Scholar 

  • Gottschalk, P., & Danziger, S. (2001). Income mobility and exits from poverty of American children. In B. Bradbury, S. P. Jenkins, & J. Micklewright (Eds.), The dynamics of child poverty in industrialised countries, (pp. 135–153). Cambridge: Cambridge University Press.

    Chapter  Google Scholar 

  • Gottschalk, P., & Moffitt, R. (1994). The growth of earnings instability in the US labor market. Brookings Papers on Economic Activity, 2, 217–272.

    Article  Google Scholar 

  • Greenville, J., Pobke, C., & Rogers, N. (2013). Trends in the distribution of income in Australia. Productivity commission staff working paper, Canberra.

  • Gregory, B. (2013). The henderson question? The Melbourne Institute and fifty years of welfare policy. Australian National University, CEPR Discussion Paper Series, No. 682.

  • Grimm, M. (2007). Removing the anonymity axiom in assessing pro-poor growth. Journal of Economic Inequality, 5(2), 179–197.

    Article  Google Scholar 

  • Groll, T., & Lambert, P. J. (2012). The pro-poorness, growth and inequality nexus: Some findings from a simulation study. Review of Income and Wealth. doi:10.1111/j.1475-4991.2012.00522.x.

  • Harding, A., Vu, Q. N., Percival, R., & Beer, G. (2005). Welfare-to-work reforms: Impact on sole parents. Agenda, 12(3), 195–210.

    Google Scholar 

  • Herscovitch, A., & Stanton, D. (2008). History of social security in Australia. Family Matters, 80, 51–60.

    Google Scholar 

  • HILDA. (2011). HILDA survey annual report 2011. Melbourne Institute of Applied Economic and Social Research, The University of Melbourne.

  • Kakwani, N. (1993). Poverty and economic growth with application to Cote D’Ivoire. Review of Income and Wealth, 39(2), 121–139.

    Article  Google Scholar 

  • Kakwani, N., & Pernia, N. (2000). What is pro-poor growth. Asian Development Review, 16(1), 1–22.

    Google Scholar 

  • Kakwani, N., & Son, H. (2008). Poverty equivalent growth rate. Review of Income and Wealth, 54(4), 643–655.

    Article  Google Scholar 

  • Klasen, S. (2008). Economic growth and poverty reduction: Measurement issues using income and non-income indicators. World Development, 36(3), 420–445.

    Article  Google Scholar 

  • Korpi, W., & Palme, J. (2003). New politics and class politics in the context of austerity and globalization: Welfare state regress in 18 countries 1975–1995. American Political Science Review, 97, 426–446.

    Article  Google Scholar 

  • Mendes, P. (2009). Retrenching or renovating the Australian welfare state: The paradox of the Howard government’s neo-liberalism. International Journal of Social Welfare, 18, 102–110.

    Article  Google Scholar 

  • Osmani S. (2005) Defining pro-poor growth. One pager, 9. Brazilia: UNDP International Poverty Centre.

    Google Scholar 

  • Oaxaca, R. L. (1973). Male–female wage differentials in urban labor markets. International Economic Review, 14(3), 693–709.

    Article  Google Scholar 

  • Piketty, T., & Saez, E. (2003). Income inequality in the United States 1913–1998. Quarterly Journal of Economics, 118, 1–39.

    Article  Google Scholar 

  • Ravallion, M., & Chen, S. (2003). Measuring pro-poor growth. Economics Letters, 78, 93–99.

    Article  Google Scholar 

  • Saez, E., & Veall, M. (2005). The evolution of high incomes in Northern America: Lessons from Canadian evidence. American Economic Review, 95, 831–849.

    Article  Google Scholar 

  • Sastre, M., & Trannoy, A. (2002). Shapley inequality decomposition by factor components: Some methodological issues. Journal of Economics, 9, 51–89.

    Google Scholar 

  • Saunders, P., & Bradbury, B. (2006). Monitoring trends in poverty and income distribution: Data, methodology, and measurement. The Economic Record, 82(258), 341–364.

    Article  Google Scholar 

  • Saunders, P., & Hill, T. (2008). A consistent poverty approach to assessing the sensitivity of income poverty measures and trends. The Australian Economic Review, 41(4), 371–388.

    Article  Google Scholar 

  • Scutella, R., Horn, M., & Wilkins, R. (2009a). Measuring poverty and social exclusion in Australia: A proposed multidimensional framework for identifying socio-economic disadvantage. Melbourne Institute Working Paper No. 4/09.

  • Scutella, R., Kostenko, W., & Wilkins, R., (2009b). Estimates of poverty and social exclusion in Australia: A multidimensional approach. Melbourne Institute Working Paper No. 26/09.

  • Sen, A. (2000). Social exclusion: Concept, application and scrutiny. Social Development Papers No. 1, Office of Environment and Social Development, Asian Development Bank, Manila, Philippines.

  • Shorrocks, A. F. (1999). Decomposition procedures for distributional analysis: A unified framework based on the shapley value. University of Essex: Mimeo.

  • Son, H. (2004). A note on pro-poor growth. Economic Letters, 82, 307–314.

    Google Scholar 

  • Ware, J. E., Snow, K. K., Kosinski, M., & Gandek, B. (2000). SF-36 health survey: Manual and interpretation guide. Lincoln, RI: QualityMetric Inc.

    Google Scholar 

  • Watson, N. (2004). HILDA wave 2 weighting. HILDA Project Technical Paper Series No. 4/04.

  • Watts, H. (1968). An economic definition of poverty. In D. P. Moynihan (Ed.), On understanding poverty. New York: Basic Books.

    Google Scholar 

  • Whiteford, P. (2010). The Australian tax-transfer system: Architecture and outcomes. Economic Record, 86(275), 528–544.

    Article  Google Scholar 

  • Wilkins, R. (2007). The changing socio-demographic composition of poverty in Australia: 1982–2004. Australian Journal of Social Issues, 42(4), 481–501.

    Google Scholar 

  • Wooden, M., & Watson, N. (2007). The HILDA survey and its contribution to economic and social research (So Far). Economic Record, 83(261), 208–231.

    Article  Google Scholar 

Download references

Acknowledgments

Financial support from the Ministerio de Ciencia e Innovación (grant ECO2008-03484-C02-01/ECON and ECO2010-21668-C03-03) Xunta de Galicia (10SEC300023PR) is gratefully acknowledged. This paper uses unit record data from the Household, Income and Labour Dynamics in Australia (HILDA) Survey. The HILDA Project was initiated and is funded by the Australian Government Department of Families, Housing, Community Services and Indigenous Affairs (FaHCSIA) and is managed by the Melbourne Institute of Applied Economic and Social Research (Melbourne Institute). The findings and views reported in this paper, however, are those of the author and should not be attributed to either FaHCSIA or the Melbourne Institute.

Author information

Authors and Affiliations

  1. Melbourne Institute of Applied Economic and Social Research, University of Melbourne and Brotherhood of St Laurence, Melbourne, VIC, 3010, Australia

    Francisco Azpitarte

Authors
  1. Francisco Azpitarte
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Francisco Azpitarte.

Appendices

Appendix

1.1 The Index of Multidimensional Poverty

The poverty index proposed in Scutella et al. (2009a, b) combines information on twenty-one indicators from seven different domains: material resources; employment; education and skills; health and disability; social; community; and personal safety. Table 8 presents a description of the indicators included in each domain. For any individual i the measure of social exclusion, x S i , is defined as seven minus the weighted sum of the level of social exclusion experienced within each domain, x id , where every domain is assigned equal weight:

$$ x_{i}^{S}=7-\sum_{d=1}^{7}x_{id}. $$

The level of exclusion in any domain is given by the actual proportion of indicators within the domain in which the individual is deprived, which can expressed as follows

$$ x_{id}=\frac{\sum_{k=1}^{K_{d}}x_{id}^{k}}{K_{d}}, $$

where x k id is a binary indicator taking value 1 when the individual is deprived in the indicator k of social exclusion included in the domain d, and K d refers to the total number of indicators for domain d.

Table 8 Index of multidimensional poverty
Full size table

1.2 Controlling for Selective Non-response

Every wave of the HILDA survey provides longitudinal weights designed to control for selective non-response including attrition. These weights are constructed adjusting the initial person weights for the probability of non-response in subsequent waves. This probability is derived using a logit specification estimated using data internal to the HILDA survey. Concretely, this model assumes that the probability of response is a function of the characteristics of the individuals and the household they belong to and other factors related to the conditions of the personal interviews. A detailed discussion of the model and how its estimates are used to construct the weights can be found in Watson (2004).

To assess the robustness of the pro-poor growth estimates to the way selective non-response is accounted for we compute our own longitudinal weights following the method outlined in Watson (2004). Concretely we estimate the probability of responding using the model employed in the conterfactual analysis of Section 5.3 Compared to the model used for the original weights, our model does not include any information on the conditions of the initial interview and incorporates more detailed information on the statistical regions, disabilities and the health conditions (general, physical, and mental) of individuals. Estimation results are available upon request. Table 9 compares the original estimates based on HILDA weights with those computed using the new weights and those derived without any weights. Further, the bottom panel shows the estimates obtained using 1 year incomes instead of two-period averages. The figures from this table suggest that pro-poor estimates are robust to the way selective non-response is accounted for and that those estimates are not driven by the averaging of incomes.

Table 9 Robustness of pro-poor growth measures to weighting and averaging, 2001–2008
Full size table

1.3 Pro-poorness and Income Mobility

We aim to partially eliminate the influence of mobility on our pro-poor growth measures by controlling for two sources of income-mobility: that due to the natural growth of life-cycle earnings and the one explained by the initial income conditions. Following the literature on income dynamics (see, for instance, Gosttchalk and Moffitt 1994), we assume that income gains depend, among other things, on some function of individuals’ initial age and income. We propose the following model

$$ \dot{y}_{i}=Age_{0}\lambda _{1}+(Age_{0})^{2}\lambda _{2}+(Age_{0})^{3}\lambda _{3}+(Age_{0})^{4}\lambda _{4}+y_{0}\lambda _{5}+u_{i}, $$

where the income gains of the individuals, \(\dot{y}_{i}\), depend on the life-cycle of earnings captured by a quartic function of age, the initial income, y 0, and other factors included in the residual term u i . Under this specification, u i measures the income variation that is not explained by life-cycle factors and the initial income status. We estimate the parameters of the model by ordinary least squares and we use these estimates to compute the residual for every individual in the panel. Table 10 compares the actual pro-poor growth measures with those derived using the residuals of the model for the period 2001–2008. To differentiate the effect of the two sources of mobility, we compute the pro-poor measures using residuals from two versions of the model: one that controls only for life-cycle earnings (MGRIP*) and a second one that controls also for initial income (MGRIP**). The comparison of MGRIPs suggests that part of the income gains of the poor and, therefore, the pro-poorness of growth can be partially explained by income mobility. Thus, for any combination of poverty measure and threshold, we find that MGRIP* and MGRIP** are always lower than the actual MGRIP. The effect is particularly important in the case of the multidimensional measure, as the MGRIP becomes negative when we eliminate the part of the income growth explained by the initial age and income. Interestingly, even after controlling for these factors, we still find that growth was particularly beneficial for those initially in low-income and that growth was clearly more pro-income poor than pro-multidimensionally poor.

Table 10 Income mobility and pro-poor growth measures, 2001–2008
Full size table

Counterfactual Analysis: Regressions

See Tables 11 and 12.

Table 11 Oaxaca–Blinder OLS regressions
Full size table
Table 12 DiNardo–Fortin–Lemieux probit regressions for the probability of being income and multidimensionally-poor
Full size table

Rights and permissions

Reprints and permissions

About this article

Cite this article

Azpitarte, F. Was Pro-Poor Economic Growth in Australia for the Income-Poor? And for the Multidimensionally-Poor?. Soc Indic Res 117, 871–905 (2014). https://doi.org/10.1007/s11205-013-0378-8

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11205-013-0378-8

Keywords

Search

Navigation