Ex Extreme and and Persis istent In Ineq equality ty: : Ne - PowerPoint PPT Presentation

Ex Extreme and and Persis istent In Ineq equality ty: : Ne New Evidence for Br Brazil Co Combining National Acc Na Accounts, , Sur Survey eys and and Fis iscal al Data, a, 2001-2015 2015 Marc Morgan Paris School of Economics & EHESS World Inequality Lab & WID.world Fellow First WID.world Conference Paris School of Economics | 14-15 December, 2017

Mo Moti tivati tion Why Brazil? • A large developing country with a long and dramatic inequality history • High market income inequality (Gini = 0.50-0.60) • Improved data quality and access in recent years The 2000s • Return of growth (18% per adult) amid crises • First “PT” government since early 1960s • Strong decline in income inequality during 2000s according to surveys (‘success story’) : fall in Gini by 7-10 pp Barros et al. (2010) in López-Calva and Lustig (Eds.) Declining inequality in Latin America: A Decade of Progress? 2

Evolution of per adult national income in Brazil: 1985-2015 R$ 40,000 R$ 38,000 R$ 36,000 + 18.2% Constant (2015) Reais R$ 34,000 R$ 32,000 R$ 30,000 - 0.3% R$ 28,000 R$ 26,000 R$ 24,000 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 Notes: The figure shows the evolution of national income per adult (aged 20+). IBGE and WID.world.

Motivation Mo Distributional Na Di National Accounts (DI (DINA NA) = Adding inequality to SNA (using surveys and fiscal data) Applied to Brazil to offer a new analysis of inequality: - How unequal is Brazil? - How has inequality evolved over this new period of growth and government? - Is there evidence of a decline? Sources? - How was growth distributed? First attempt at combining three sources of data for Brazil 4

Da Data sour urces, Conc ncepts ts and nd Metho thod 5

Da Data Sources Sur Survey eys: PNAD (IBGE) E) § • Micro-files: 2001-2015 • Complemented with POF 2009 survey Pe Pers rsonal income tax declara rations (DIRPF) ) § • Tabulations of total gross income (taxable, withheld & exempt) before social contrib. • Coverage: 25-28 million declarations (20% of adult population) • Tax unit = individuals or couple Na National accounts (I (IBG BGE) § • Flow of funds (2001-2015) = SNA (2008) • Financial balance sheets (2010-2014) 6

Concepts Co Unit of observation: Eq Equal-sp split a t adults ts • Aged 20+, income of married couples is equally divided Income concept: Pr Pre-ta tax post-re replaceme ment t nat national nal i inc ncome me ( (DIN INA) • National income attributed to households • Including pensions + unemployment insurance & deducting social contributions • vs fiscal income = income assessed by tax administration • vs survey income = income appearing in surveys 7

Me Method: building fr from th the e bo bottom up up Na National income + Net government capital + government income Normalize distribution by income + net production taxes (SNA) national income Personal income + Imputed rent (7%) – social contributions (14%) + + non-fiscal income Assume a correlation structure insurance/pension fund between fiscal and missing income (SNA & surveys) income (1%) + household (Gumbel Copula) undistributed profits (6%) Fiscal income = Gross wages + pensions + + top incomes Correct survey incomes by ratio self-employed income + y fiscal / y survey by percentile where it (tax data) capital income ≈ 71% of NI is > 1 Survey income 8

Figure 1. Evolution of total real income in Brazil: 2001-2015 R$ 6,000,000 R$ 5,000,000 R$ 4,000,000 Million (2015) Reais R$ 3,000,000 R$ 2,000,000 National Accounts (national income) National Accounts (fiscal income) R$ 1,000,000 Survey + Tax data (fiscal income) Survey data Tax data R$ 0 2001 2003 2005 2007 2009 2011 2013 2015 Notes: The figure shows the evolution of total real income in billions of Reals across datasets and income concepts. The Survey + Tax data series uses survey incomes up to the percentile where average percentile income in the surveys is less than or equal

Me Method: est stimating th the l e labou our ea earnings d s distr tributi tion on Combine surveys + tax data as above but for labour earnings Note: Labour income is not purely observable in tax returns Estimation: Taxable income (wages, self-employed labour income, rent) Ta § Distribution of taxable income – property rent (= 2% of fiscal taxable § income in national accounts) Assumptions: § • 20% of rental income à Middle 40% taxable income share • 80% of rental income à Top 10% (incl. 40% à Top 1%) share 10

Re Results: : inc income ine inequality quality in in Br Brazil 2001 2001-2015 2015 Extreme and consolidated concentration at the top vs compression in bottom 90% Surveys seem to massively underestimate levels and overestimate decline Overall the distribution narrowed over the period Labour incomes seem to do most of the work . 11

Figure 2. Income inequality in Brazil: DINA estimates 60% 55% 50% Top 10% 45% Middle 40% Bottom 50% 40% 35% 30% 25% 20% 15% 10% 5% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Notes: Distribution of pretax national income (before taxes and transfers, except pensions and unemployment insurance) among adults. Corrected estimates (combining survey, fiscal and national accounts data). Equal-split-adults series (income of married couples divided by two).

Figure 2. Income inequality in Brazil: DINA estimates 60% +2.0% 55% 50% Top 10% 45% Middle 40% Bottom 50% 40% 35% 30% -7.6% 25% 20% +11.2% 15% 10% 5% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Notes: Distribution of pretax national income (before taxes and transfers, except pensions and unemployment insurance) among adults. Corrected estimates (combining survey, fiscal and national accounts data). Equal-split-adults series (income of married couples divided by two).

Figure 8. Top 1% in Brazil: survey vs fiscal vs DINA series 40% 38% National income series 36% 34% Fiscal income series 32% Survey income series 30% 28% 26% 24% 22% 20% 18% 16% 14% 12% 10% 8% 6% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Notes: Distribution of income (before taxes and transfers, except pensions and unemployment. insurance) among adults in our three series, raw estimates from surveys, a fiscal income series (combining surveys and fiscal data) and a national income series (combining national accounts, surveys and fiscal data). Equal-split-adults series (income of married couples divided by two).

Inequality of labour income more clearly falls Figure 9. Labour income shares in Brazil: corrected estimates 50% -7.0% 45% 40% +2.2% 35% Top 10% 30% Middle 40% 25% Bottom 50% 20% +19.5% 15% 10% 5% 0% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Notes: Distribution of pretax taxable labour income (before taxes and transfers, except pensions and unemployment insurance) among adults. Corrected estimates (combining survey, fiscal and national accounts data). Equal-split-adults series (income of married couples divided by two).

Capital income makes up the difference Figure 10. Top 1% in Brazil: labour income vs total income 40% 38% Top 1% (national income) 36% Top 1% (fiscal income) 34% 32% Top 1% (labour income) 30% 28% 26% 24% 22% 20% 18% 16% 14% 12% 10% 8% 6% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Notes: Distribution of pre-tax income (before taxes and transfers, except pensions and unemployment insurance) among adults compared to a fiscal total income series (combining surveys and fiscal data) and a national income series (combining national accounts, surveys and fiscal data). Equal-split-adults series (income of married couples divided by two).

Role of labour income inequality for Gini Figure A.4. Gini coefficients in Brazil: 2001-2015 0.700 0.680 0.660 0.640 0.620 0.600 0.580 0.560 0.540 Gini (national income) 0.520 Gini (survey income) 0.500 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Notes: Distribution of income (before taxes and transfers, except pensions and unemployment insurance) among adults. Raw estimates from surveys and a national income series (combining national accounts, surveys and fiscal data). Equal-split- adults series (income of married couples divided by two).

Explaining Ex plaining Recent Trends nds in in Braz azilian ilian ine inequalit quality Most of the action from the labour income side Minimum wage (+ 64% real growth since 2000) Mi Incidence: Ferreira et al. (2017); Brito et al. (2016) • Fall in education premium Fa Higher wage growth of lower skilled labour • Exceptional educational expansion, 1990-2010 (Barro and Lee, 2013) • “Horizontal wage inequalities” “H ” (Ferreira et al. 2017) Men vs women; blacks vs whites; rural vs urban; formal vs informal sectors • 18