This paper analyzes poverty dynamics in 5 Latin American countries between 2003-2012 by examining rates of entry and exit from poverty. It finds that labor market events, such as changes in wages and job losses, were most strongly associated with transitions into and out of poverty. Public cash transfers played a limited role in exits from poverty. Households with children were more vulnerable, as negative labor market events were more likely to trigger entry into poverty for these households. The results suggest precarious labor markets in the region contributed significantly to poverty dynamics during this period.
1. Dynamics of Income Volatility in the US and in
Europe, 1971-2007: The Increasing Lower Middle
Class Instability
Louis Chauvel and Anne Hartung
University of Luxembourg
IARIW 2014
Catherine Van Rompaey for Sylvie Michaud
2. Purpose of paper
• Compare short term income mobility (volatility)
at different levels of the income distribution over
time and space.
– Are poor/rich particularly mobile ?
– Has this changed over time ?
– Are there cross country differences in mobility
patterns ?
• Use continuous measure (logit- income ranks)
instead of income per se
3. The logit-rank method
• Uses a continuum of income ranks as opposed to
income or centile groupings
– Gets at intra-group volatility
– Removes element of structural volatility due to
changes in the income distribution over time
– Avoids volatility due to changes in the top-code
• Better measure at the upper tail
– More stable results
– Precedents in other areas of social science
– Rank matters in terms of utility
4. Measuring volatility of income ranks
• Panel Survey of Income Dynamics (PSID)
– 1970-2007: Comparison of ‘70s and 2000s
– Equivalized post-governmental household income
– Households with heads between 25 and 59
– Black and white Americans, other ancestries such as
Asians and Hispanics excluded
• Short term income rank volatility: changes in the
logit of percentile ranks in two year intervals
• Compare US profiles of volatility to Europe using
EU-SILC (2008 to 2010)
8. Change in profile: Increased volatily in lower
income & decreased volatility in higher income
9.
10.
11.
12. When controls added
• Income
– Higher income less volatility; third quartile is turning point
• Age of head of household
– Younger and older head of households have more volatility
• Marital status / family structure
– Change in marital status or number of children impacts volatility
– Less volatility for families with children
• Hours worked
– Changes in hours impacts
– Head has full time contract less volatile (except if heads works more
than 3000 hours)
13. US vs Europe
• Levels of volatility lower in Nordic countries,
Portugal and Italy, higher in the UK, Austria and
Spain
• Portugal different: smaller at top and stronger at
bottom with very volatile poor and stable elite.
• Denmark and Italy present different model than
US: rich more unstable relative to poor
• U shape fairly similar everywhere, except
Luxemburg with less volatile middle class and
more shaky extremes
14. Discussion /questions
• Very interesting paper
• Agree with choice of household income
• Limitations of non parametric methods; rank may not
discriminate properly, especially at bottom of the
distribution
– Could be due to rounding errors
– Could be due to ties
– Include discussion on possible limitations of the non-parametric
method?
• Are there limitations in the data themselves
– Top coding, source of income
– Compare 2008/2010 in Europe to US in the 2000s and 1970s
• Investigate goodness/badness at tails?
15. Entry and exit rates in Latin America: the
role of Labor Market and Social Policies
Luis Beccaria, Roxana Maurizio, Gustavo Masquez
and Manuel Espro
Universidad Nacional de General Sarmiento,
Argentina
IARIW 2014
Catherine van Rompaey for Sylvie Michaud
16. Purpose
• Poverty dynamics in 5 Latin America countries
between 2003-2012
– Do different levels of incidence impact rates of
entry/exit ?
– Impact of events that affect poverty transition
• How do they affect different subgroups?
17. Context
• Latin America experienced period of expansion
between 2003-2012
– Rapid growth in GDP per capita (2.9% annually)
– Reduced inequality and poverty
– Despite progress, 28% still live in poverty, 12% in
extreme poverty
18. Measurement
• Poverty dynamics examined through household surveys in
5 countries
– Argentina, Brazil, Costa Rica, Ecuador and Peru
– Slightly faster growth, different incidence levels
– Surveys where households interviewed at least twice
– Periods slightly different, study limited to urban areas
• Absolute poverty measure; income to satisfy basic needs
– Specific consumption patterns, same set of basic needs
– Household income adjusted for size and composition
• Other measures
– ECLAC, lines by stat. agency, World Bank
19. Entry and exit rates in Latin America: the
role of Labor Market and Social Policies
Luis Beccaria, Roxana Maurizio, Gustavo Masquez
and Manuel Espro
Universidad Nacional de General Sarmiento,
Argentina
IARIW 2014
Catherine van Rompaey for Sylvie Michaud
20. Purpose
• Poverty dynamics in 5 Latin America countries
between 2003-2012
– Do different levels of incidence impact rates of
entry/exit ?
– Impact of events that affect poverty transition
• How do they affect different subgroups?
21. Context
• Latin America experienced period of expansion
between 2003-2012
– Rapid growth in GDP per capita (2.9% annually)
– Reduced inequality and poverty
– Despite progress, 28% still live in poverty, 12% in
extreme poverty
23. Events impacting poverty transitions
Mutually exclusive events classified:
– Non-demographic
• Labour market
• Non-labour income events
• Labour / non-labour income
– Demographic
• Demographic only
• Demographic leading to income
• Demographic and income
24. Impact of events
• Decomposed probability of transition in/out of
poverty into two elements:
1. Probability that at-risk population experiences events
2. Probability that the events triggers a transition
P (Sij)=Ʃ P(SijǀEr) P(Er)
• r=1
– Where Sij : transition from state i to j between t, t+1
– Where Er : events that could be associated from exit to
poverty
25. Events associated with poverty exits
• High proportion of poor experienced an event
– No more than 50% actually exited poverty
• Events related to labour market most relevant
– Increase in self-employed non-registered wage earning jobs
(labour precariousness)
• Non-labour events
– Rise in income from pensions (Argentina, Brazil, Costa Rica)
– Remittances from migrants in foreign countries (Ecuador)
• Exclusively demographic events such as reduction in
household members had low impact
– Yearly observation window too short, events less frequent
• Public cash transfers appear relatively unimportant
– Scarcity, data limitations
26. Events associated with poverty exits
• Households with and without children
– Labour market events still most important for
both, more so for households with children
– Non-labour events more important for
households without children
• Elderly household members (pensions)
27. Events associated with poverty entries
• High share of non-poor experienced negative events
reducing their income (by between 36% and 67%)
– 20-30% moved into poverty, non-negligeable group when
overall poverty incidence declining
• Unlike exits, frequency of events more important than
their impact
• Reduction in hourly wages most frequent event, but
job losses had larger impact: high frequency of non-
registered and self employed job losses
• Exclusively demographic events relatively unimportant
• Income from cash transfer policies played no role
28. Events associated with poverty entries
• Households with and without children
– No substantial difference in frequency of events
– Impacts of events differed substantially
• Probability much higher (10 p.p.) that negative event
triggers entry for households with children
– Households without children more influenced by
non-labour income and demographic events
• More elderly and young people outside labour force
29. Comments / questions
• Interesting paper
• Questions:
– Can chosen categories for classification of events
impact results? Alternatives considered?
– Why does Peru have higher rate of poverty with
panel?
– Transitions :
• Households defined pre/post event (first or second
year of panel)?
Editor's Notes
Levels of volatility are lower in Nordic countries, Portugal and Italy (also Netherlands and Luxembourg ?)
Portugal is different from other countries on volatility between high and lwoer income (smaller at top and larger at bottom).
Denmark and Italy present different model from US less volatile at bottom, more volatile at the top than US)
Argentina, Brazil, Peru, Ecuador, Costa Rica have surveys where househodls are itnerviewed at least twice
Argentina, Brazil, Peru, Ecuador, Costa Rica have surveys where househodls are itnerviewed at least twice
Calculated rates both for households and individuals; present only households here
Rates vary by country
Ecuador being relatively high
Poverty rate usually decrease through the years
Most decrease is in Argentina
Poverty rates are usually higher for the xsect. Data than for panel data
Often the case due to attrition
Peru is different : any reason for that ?
Labour market events
Change in number of people employed (registered or non-registered wage earners, non-wage earners)
Change in hourly wages (for people in
Change in horus worked
Labour market events
Change in number of people employed (registered or non-registered wage earners, non-wage earners)
Change in hourly wages (for people in
Change in horus worked
Labour market events
Change in number of people employed (registered or non-registered wage earners, non-wage earners)
Change in hourly wages (for people in
Change in horus worked
Labour market events
Change in number of people employed (registered or non-registered wage earners, non-wage earners)
Change in hourly wages (for people in
Change in horus worked
Labour market events
Change in number of people employed (registered or non-registered wage earners, non-wage earners)
Change in hourly wages (for people in
Change in horus worked
Labour market events
Change in number of people employed (registered or non-registered wage earners, non-wage earners)
Change in hourly wages (for people in
Change in horus worked