Inequality changes in the post World War II period : extent and possible causes

1
Inequality changes in the post World War II
period extent and possible causes

• Giovanni Andrea Cornia
• University of Florence
• (giovanniandrea.cornia_at_unifi.it)
• --------------------------------------------------
  -----------
• UNIDEA-University of Florence-EUDN Summer School
  on
• Inequality, Growth and Human Development
• Civita Castellana, 1-6 July 2007

2
Table of contents

• Past and recent perspectives on inequality
  trends
• Methodology of analysis of within-country income
  inequality and related results
• III. Groups most affected
• IV. Sources of recent inequality rise
• Causes of recent inequality rise
• A tentative econometric test
• Tentative conclusions

3
I. Until recently,inequality seen as stable

• The US income distribution is a facet of
  economic life which changes slowly when it
  changes at all (Solow, 1960)
• 'Decadal averages of inequality indexes are
  relatively stable through time but differ
  substantially across regions/countries(Deininger
  and Squire, 1996)
• ..there is no evidence of a time trend in 65
  of countries Growth is main solution for poverty
  alleviation (Li-Squire-Zou 1998) on 49 countries

4
Kuznets
Recent Literature
Solow
5

• More recent perspectives on inequality
• (taken mainly from
  Atkinson 2005)
• By the early 1980s, inequality in the US had
  reached 1948 levels .. the 1950s and 1960s ..
  were periods of unmatched equality (Gottschalk
  and Smeeding 2000) on USOECD
• the Golden Age witnessed declines in income
  inequality. This trend was reversed over the last
  two decades as country after country experienced
  an upsurge in inequality (Cornia and Court 2003,
  on 73 countries )
• in most countries inequality has tended to
  increase .. mainly because of an increase in
  the income share of the richest fifth
  (Jantti/Sandstrom 2005, on 115 countries)

6
II.Analysis of within-country inequality
methodology

• (i) Review-synthesis of case studiesof reviews
  (LIS, Szekely, Transmonee, LSMS-SSA, Milanovic,
  You, Ragayah) abt 150 studies collected
• For most countries included in trend analysis,
  collected 1-2 quanti qualitative studies placing
  distributive changes in broad context
• (ii) Trend analysis based on WIID2a database
  (28.06.05). It includes
• 4664 Gini coefficients up to 2003 for 154
  countries
• data on income decile/quintile for over 62
  countries
• Each Gini is fully documented. Contains info on
  territory, population, age group, income concept,
  unit of analysis of survey on which it was
  computed, equiv.scales eventually used, 1ary -
  2ary data sources
• Gini are rated 1, 2, 3, 4

7

Methodology From original 4664 Gini data
points, we dropped - data prior to 1950, data
for 67 countries with lt 6 data, data concentrated
only over short periods or prior to 80, with
changes in income concepts or national
boundaries, outliers. This left 1165 high
quality Gini for 85 countries (20 developed, 25
transitional, 40 developing) derived from fully
documented, comparable, represent. surveys of
entire economy For each country same income
concept, though this varies across countries
(table)

8
From earnings to household income inequality
9
Methodology

• In 20 of 85 countries with data on 2 income
  concepts, the less frequent Gini were
  standardized into the most frequent one through
  correction factor, i.e. mean ratio of 2
  concepts in years with both info
• In remaining 65, the series selected were the
  most complete. Among equally complete series (as
  in OECD) precedence was given to, net income,
  then gross income, net earnings, gross earnings,
  consumption.
• There is theoretical possibility of differences
  in trends among income concepts. Yet, this is
  very rare event (see Finland - France).
• except for SSA and MENA, 85 countries retained
  for trend analysis account for 89-99 of
  regional population and 95-99 GDP-PPP. But in
  SSA they represent only 35 55, and in MENA 50
  47

10
Methodology

• Data were interpolated with linear quadratic
  functions so as to capture trend reversals
• Choice between constant, rising or falling
  inequality made on basis of F, t (and R2)
  statistics
• Allowance made for flattening of rising trend
• In a few cases, third degree functions (which
  display a local maximum and minimum) were tested
  

11
(No Transcript)
12
OECD
or
UK
or
US over the long term
India
or
Russia
Brazil
or
13
In US and OECD there was a decline in wage
share and surge in skilled/unskilled wage
ratio. Causes ? See later
14
Similar trend in other Anglo-Saxons, e.g.NZ
15
UK inequality stabilizes in 1990, after sharp
rise over 1978-90
--------------------------------------------------
--------------- (ii) e Coef.
Std. Err. t Pgtt
-------------------------------------------------
--------------- year -1.080997 .0274643
-39.36 0.000 cons 2152.756
54.84639 39.25 0.000
-------------------------------------------------
----------------
16
Yet, France W.Germany (both with strong labor
institutions) show falling trend
France
ERF Gross income
ERF Gross income equivalized
ERF Net income equivalized
17
(No Transcript)
18
Inequality explodes till late 1990s in FSU
19
Yet in some transition economies, inequality
stabilizes since late 1990s (as in Romania
below), as GDP growth recovers
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
Nigeria shows a constant trend as none of
parameters of the time trend (see below) is
significantly different from zero
25
In India, inequality begins rising in 1991 with
the introduction of its first
internal and external liberalization program
26
A 3rd degree shape in S. Korea, with rising
inequality in the 1990s
27
MENA. Region underwent little liberalization-globa
lization. The few available Gini point to
broadly stable inequality
28
Summary of trends in within-country income
inequality, 1980-2003
Source Cornia and Rosignoli 2007 on WIID2a
(www.wider.unu.edu)
29
Causes of difference with LSZ results
30
Sources of difference with LSZ (in )

• 47 due to functional form chosesn (linear in LSZ
  linear, quadratic in our study )
• 20 due to different time span analyzed (till
  1992-3 in LSZ till 1998 in our study )
• 33 due to country coverage
  (49 in LSZ -77, including EE/FSU, in
  our study )

31
Years of inequality trend reversal

• Timing of reversal

- in 1 country over 1960-65, - in 4 over
1966-75, - in 6 over 1975-80, - in 8 over
1980-85, - in 10 over 1985-90, and - in 14
over 1991-97
Source Cornia and Kiiski 2001
32
Gini points rises ranged from moderate to
unprecedented

• The Gini points rise was of
  
  
• - lt5 pts in 6 countries,
• - 5-10 points in 20
• - 10-20 points in 23
• - gt20 points ..in 6.

Source Cornia and Kiiski 2001
33
III. Which group was most affected (rich,
middle class poor)?

• WIID2a includes data on income shares of top 20,
  middle 40 and bottom 40
• Out of 85 countries selected, 62 had time series
  on income shares of top 20, middle and bottom 40
• shifts in income shares offer insights on drivers
  of inequality , in case of hollowing out of
  middle class, on political sustainability of
  observed shifts

34
USA a hollowing out of middle poor class,
35
in UK poorest 40 affected most by
distributive changes
36
Italy only bottom 40 affected by rising
inequality, mid class gained
Gini coefficient
Share top 20
Share bottom 40
Share mid 40
37
A gradual but steady inequality rise affecting
mostly the bottom 40.
Income share top 20
Gini coefficient
Income share middle 40
Income share bottom 40
38
In China, inequality rise affects sharply bottom
40 (rural people, marginal regions, internal
migrants) - much less the middle 40
Gini index
Income share of top 20
Income share bottom 40
Income share middle 40
39
Distribution of gains/losses by social groups in
the 39 countries (out of 62 with quintile data)
experiencing rising inequality
40
IV. Sources of recent rise in inequality rely
on decomposition by income type

• What explains rises in inequality identified? To
  answer this question we decompose surge in total
  inequality by income type
• Gini of total income can be decomposed as
  weighted sum of concentration coefficients of
  wage income, capital income and transfer income,
  i.e.
• (1) G sw Cw Cci sci st Ct
  
• - A D over time in total Gini can be
  decomposed by differentiating over time (1) to
  get
• (2) DG S Dsi Ci S DCi si S
  Dsi DCi
• where swscist 1, (share of wages, capital
  transfer incomes
• and Ct lt Cw lt Cci (concentration coeff. capital
  income isgtthat of wages/transfers
• -

41
Decomposition by income

• Decomposition shows that recent income inequality
  rise can be due changes in 5 parameters
• (i) a rise in capital share (sci) in total income
  
• (ii) a fall in transfer share (st) and wage share
  (sw)
• (iii) rising concentration of wage, capital
  income and transfers incomes (Cw Cci Ct)
• Which factors of last 20 years generated these
  changes ?

42
(No Transcript)
43
(iii) evidence of rise of capital share

• Atkinson (2002) shows that
• in UK, income share of top 1(60 coming from
  capital incomes) rose from 21 to 34 over
  1979-01, entailing a 8 point rise in sci.
• Bertoli-Farina (2007) over 1975-00 sci rose in
  19/20 Europ countries
• IMF (2007) in advanced economies sw declined by
  7-8 points
• Cornia (2004) presents evidence on developing
  countries
• Thailand, share nonfarm profitsprop.income rose
  19to25, 88-96
• Turkey, interests on public debt/GDP rose from 0
  to 15, 1980-98
• Venezuela, CS rose from 58 to 78 over 1978-96,
• S.Africa, property income rose from 18 to 30
  over 1981-2000,
• Chile Argentina, LS fell 5-6 points over
  1980-88, 10 in Mexico.
• India,operating surplus/net income rose from
  10to16 over 88-96  
• Economies in transition (Russia, Uzbek, Azer)
  data point same way

44
Source Atkinson .
45

• V. Causes of recent inequality rise

46
5 possible sets of causes (in red those
emphasized in current debate)

• (i) Core model worsening of traditional causes
  of inequality ?
• Concentration of land, hum. capital,
  industrial/fin. assets, credit
• Urban bias (China)
• Resource curse (Kazak, Azerb, Russia,
  Eq.Guinea)
• (affect sw , sci ,Cci)
• (ii) Cyclical effects (due to economic collapse
  of 1990s in EE, recession in LA-SSA in 80s-part
  90s) (affect sw ,Cw) and inflation
• (iii) Technical progress ? It
• raises the demand for skilled labor
• replaces unskilled labor with capital
• (affects sw ,Cw , sci )

47
Causes of recent rise in inequality

• (iv) Liberalization-globalization policies
• Trade liberalization (affect sw ,Cw , sci)
• Migration and remittances (affects Cw)
• FDI, outsourcing, enterprise relocation (affect
  sw ,Cw , sci)
• Financial liberalization (Cci, sci)
• (v) Changes in domestic institutions
• labor protection (min wage, unionization,
  collective bargaining) (affect sw sci)
• Tax and transfers (affect st ,Ct)
• Changes in asset distribution privatization
  (affect Ct)

48
(i) Limited affect of changes due to core
variables
49
The vanishing land reforms
50
Core causes explain less ineq than before except
education

• the total variance of income inequality (INCINEQ)
  explained by the old causes of inequality
  (LANDINEQ, INEQEDUC and MINRENT) fall, as their
  coefficients get smaller and less significant
• - (1960-75) INCINEQ 34.84 (22.76) 1.36 (2.59)
  INEQEDUC 64.04 (2.53) MINRENT
• R2 0.16, nobs 71
• - (1975-98) INCINEQ 36.22 (37.42) 1.22 (3.21)
  INEQEDUC 21.07 (1.84) MINRENT
• R2 0.10, nobs 135

51
income ineq. due to educ inequality
52
(ii) Inequality changes during the business cycle

• Inequality rises in developing countries
• Wages are downward flexible
• Enterprises shed labor
• Lack of safety nets to offset loss of labor
  income
• Inequality falls in industrialized countries
• Wages are inflexible (thanks to unionisation)
• Enterprises hoard labor (to reduce future
  selection costs)
• Safety nets compensate for loss of labor income
• Enterprises experience large compression in
  profits

53
Businness cycle and inequality changes

• Country Period Phase Urban Sector Rural
  Sector
• 
  Gini 10/40 Gini 10/40
• --------------------------------------------------
  --------------------------------------------
• Brazil 79-87 Recess-Exp. 4 32 7
  9
• 87-90 Recession
  2 9 1 53
• Colombia 80-86 Recess-Exp. -3 -16
  -2 - 1
• 86-90 Expansion
  -5 0 -9 -2
• Costa Rica 81-88 Recession 7 22 14
  3 88-90
  Expansion -6 -13 -6 -15
• Chile 78-88 Recess-Exp. - 23 -
  -
  87-90 Expansion -2 - 3 -2
  -3
• Uruguay 81-86 Recession 7 20 - -
  86-89 Expansion -9 -19 -7 -12

54
Inflation and inequality

• inflation affects most the poor as they
• are less able to protect their assets by
  dollarizing /buying assets (minimize
  transaction/hgher cost)
• depend more than rich on non-indexable social
  transfers such as pensions and poor relief
• assign high share of their income to consumption
  of tradable wage goods likely to be affected by
  faster price rises.
• consider inflation as their top concern
• But reducing inflation is very costly and will
  cause a recession that also affects the poor most

55
Impact of these factors

• Cyclical effects -inflation
• plaid an initial (irreversible) role in European
  transition economies
• Perhaps in some L.A./SSA country not generally
• When recovery sets in, Gini stabilizes, but not
  drops suggesting the role of structural
  inequality changes

56
(iii) technological change

• Contributed to worsening Gini where new
  technology was adopted on large scale (E.Asia,
  OECD) not in countries with low-negative growth
  
• Also domestic impact is not large if supply of
  skilled labor adjusts to demand (no scarcity
  rents)
• Brazil vs Taiwan in the 1960s-70s
• US vs Canada in the 1980s-90s
• Increases in inequality in developing countries
  importing state of the art (capital skill
  intensive) technology

57
Changes in wage inequality due to shifts in
technical progress and skilled labor supply
Relative supply of Skilled/unskilled workers
Skilled/ unskilled relative wage
Technology shift
A
B
A
C
skilled/unskilled intensity choice Labour supply
58
(iv) Liberalisation- Globalisation policies (a)
Trade Liberalisation

• Trade Theory(HO-SS) free trade reduces ineq in
  LIC, raises it in OECD, reduces global inequality
• Observed trends A mixed increasingly different
  picture
• Improved distribution in SEA in 60/70s (Wood) and
  ceteris paribus in Coastal China in 1990s
• Worsening in LA,Philippines, EE in 1990s
• More and more evidence accumulates to show that
  the distributional changes went in the opposite
  direction suggested by conventional wisdom
  (Koujianou-Goldberg and Pavcnik 2007)

59
(a) continued

• Theoretical explanations that go beyond H-O/SS
• Factors immobility most studies find labour does
  not move from contracting to expanding sector -
  due to lack of roads, housing, human capital,
  over-unionisation, wage or profit compression,
  low social mobility due to sub-caste networks
  providing social assistance, etc. (Topalova
  2004, Burgess et al 2006 on India)
• Trade induced skill-bias (via import of
  investment goods made cheaper by trade
  liberalisation) raise capital intensity ( sci)
  complementary with skilled labour
• Differential rates of protection (unskilled one
  was most protected, and suffers most rather
  than gain from trade liberalisation)
• Trade reorientation following capital account
  liberalisation(Taylor)

60
(a) continued

• Theoretical explanations that go beyond H-O/SS
• Changing relative endowments of countries
  participating in multi-country, multi-factor
  multi-goods trade(US-vsMexico-vs China)
• Asymmetric liberalisation protectionism
  (Slaugther)
• Export of land- (not unskilled labour-) intensive
  goods under unequal land distribution
• Liberalisation in countries specialising in the
  export of primary commodities with falling prices
  (Birdsall-Hamoudi)
• G.E.effects Porto (2006, on Argentina) analyzes
  distribution of hh welfare via changes in
  employment and incomes but also in relative
  prices in T/NT sectors

61
(b) Liberalisation of FDI

• FDI Theory (same as n.classical trade theory)
  greenfield FDI reduces ineq as it raise labour
  demand wages of unskilled workers
• Observed trends A mixed picture
• Alternative theoretical explanations
• advantages of FDI are greatest in
  labour-intensive manufacturing, not in
  capital-or-resource int. sectors
• MA in utilities sector. The equity effect of
  this operation depends on the sale price of
  assets, prices of services supplied and
  industrial restructuring.

62
(b) continued

• Business stealing from Small and Medium local
  Enterprises
• N-S plant relocation skill-biased tech. change
  (Behrman et al 2000).
• Regional distribution of FDI and spatial
  inequality (Demurger,et al. 2002) as observed in
  China, (but spatial disequalization of FDI may be
  endogenous)
• Systemic effects in a world of mobile capital and
  immobile labour (Rodrik).

63
(c) Liberal.of portfolio flows

• Theory portfolio flows reduce inequality due to
  greater accumulation employment creation.
• Observed trends
• Modest worsening on inflows, large ones during
  crises over medium term
• Alternative theoretical explanations
• inflow of portfolio funds trigger
• Appreciation RER discourages labor
  absorptionformal empl
• Trigger credit booms with high i.r.strong e.r
  raise CS (Taylor)
• Intersectoral alloc funds go to rent and capital
  intensive FIRE

64
(c) continued

• mass outflows on occasion of financial crisis
  trigger
• Panic, heard behaviour, contagion, generate
• Recession, unemployment and formal-informal
  labour flows
• price changes (T prices rise following dev, NT
  people most affected)
• Fiscal retrenchments
• People in FIRE affected first, so inequality
  falls over short term (but rises over long term),
  but poor affected most by poverty
• Evidence from Indonesia, Mexico (Baldacci et al,
  2002) falling Gini
• Evidence from S.Korea (rising Gini in short and
  long term)
• Halac-Smuckler effect recapitalisation of
  bankrupt banks with tax-revenue collected through
  regressive taxes

65
(d) International migration

• Theory by increasing unskilled labor supply, it
  worsens wage inequality in rich countries (fall
  in unskilled wages), it reduces it in source
  countries as it diminishes unskilled labor
  supply
• Facts 191m.migrants in 2005 (3world pop, 2.2
  in 1980). Large is illegal, migrants work in
  informal sector
• Evidence above theory confirmed during first
  globalization. Not true now as ineq rises also
  in source countries
• Large supply of unskilled labour (migration,
  outsourcing via FDI and imports from South)
• IMF (2007) estimates that effective global labor
  supply increased fourfold during last two
  decades via the 3 channels identified above

66
(d) Continued

• Alternative explanation
• Hump theoryof migration migrants are well
  educated (better able to bear high costs of
  illegal migrations, function in new environment,
  have greater chance of finding work)
• Unskilled workers migrate when networks reduce
  migration costs, or organized migration
• Wages in source countries rise for skilled
  workers (many migrate), not for unskilled ones
  (e.g. farm workers)
• Remittances (250 bn a year, gt 10 GDP in gt20
  countries) accrue to middle income families
• Illegal nature of much present migration has
  disequalizing effect
• Due to bottlenecks and Dutch disease, remittances
  often give rise to import-intensive and job-less
  growth that raises inequality
• Brain drain effects and income inequality

67
Main global financial flows
68
(e) Domestic financial liberalisation

• Standard Theoryfinancial de-repression (Mc
  Kinnon) leads to financial deepening and growth
• Observed trends
• favourable effect in OECD some developing
  country
• In most of latter rises in financial instability
  spreads
• Alternative explanations (Diaz Alejandro,
  Stiglitz)
• Sequencing problem liberalisation with large
  deficits
• Fail raise competition? rise in real i. rates
  spreads (sci)
• Weak regulation instability and banking crises
  (- sw)
• exclusion of poor reduction in directed
  credit (India)

69
(f) Labour market liberalisation

• Theory Deregulation of labor market ?equalizing
  employment-creation effect and disequalizing
  wage-inequality effect as a whole it is
  indeterminate
• Observed trends
• Widespread surge in wage inequality in most
  countries
• Alternative theoretical explanations (Atkinson
  2002) erosion of reference norm (FIRE)(P90/P10)
  
• theories of CEOs remuneration in hierarchical
  structure (their mean wage rose fast over last 20
  years, also because of stock options)
• superstar theory(the winner takes all).
  Improvements in ITC enlarge markets/rents of
  those with highest abilities and earnings. Income
  gradient is much steeper.

70
Labour market liberalisation

• new social norms affect the level of bottom wages
  
• minimum/average wage ratio fell in most L.A.,
  Africa, EE/FSU and OECD. Fall correlates with
  earnings ineq.
• in USA 85-95 44 of ineq rise due to fall in
  minwage (Gottshalk-Smeeding)
• de-unionization decline collective bargaining
• informalization (L.A.-FSU) due to international
  trade/finance

71
Erosion of minimum wages

• 1982 1984 1986 198
  8 1989 1990 1991
• Brazil GDP/capita 92 89 99 99 100 94 93
  Average wage 121 105 121 103 102 88 84 Minimum
  wage 107 87 89 69 72 53 62
• Chile GDP/capita 90 89 92 101 109 109 113
  Average wage 109 97 95 101 103 105 110 Minimum
  wage 117 80 74 74 79 87 95
• Colombia GDP/capita 99 100 107 113 114 116 116
  Average wage 104 118 120 118 119 118 117
  Minimum wage 103 113 114 110 111 108 108
• C.Rica GDP/capita 86 90 90 92 94 95 94 Average
  wage 71 85 98 85 86 87 -- Minimum
  wage 86 104 119 115 119 120 --
• Mexico GDP/capita 102 97 92 90 91 92 94 Average
  wage 102 75 71 72 75 78 77 Minimum
  wage 93 72 65 54 51 45 42
• Peru GDP/capita 102 89 92 86 75 70 70 Average
  wage 110 87 98 76 41 43 39 Minimum
  wage 80 62 56 52 25 23 17
• Uruguay GDP/capita 88 80 86 93 93 93 94 Average
  wage 106 71 72 76 76 70 73 Minimum
  wage 104 89 88 84 78 69 62

72
(g) Reform of taxes/transfers ineq

• size of st Ct depend on societys preferences
  (demand for equality)
• Theories of justice (libertarians, liberal,
  marxist)
• Fiscal capacity of the state (chart on Chile)
• Theory of inequality impact of tax-transfer
  reform
• Tax reform not inspired by OTT/equity but by
  administrative simplification
• Lower progressivity offset by broadening tax base
  VAT. Neutral effect yields (Laffer)
• targeting of subsidies
• Observed trends Tax incidence survey in 36
  poor/transit. countries 75-95 (Chu 2004)
• Average tax/GDP ratio fell by 1 pt over 80s-90s
  (against 1.6 pts. rise over 70s-80s),
• declining weight of direct taxes in total,
• fall in overall tax progressivity,
• negative effect of the latter on Gini of final
  income
• as a consequence ? subsidies reduction

73
Redistributive effect of transfers in Chile

• 
  
  income quintiles
• 
  1
  2 3 4 5
• A.Share of private cons before transfers
  3.3 8.3 13.0 19.5 55.8
• B. Share of private cons after cash transfers
  5.2 9.3 13.3 19.2 53.1
• C. Share of private cons after inflation control
  5.4 9.4 13.3 19.2 52.8
• D. Share of priv cons after transfers in
  cash-kind 6.4 9.8 13.4 19.0 51.4

74
(g) Reform of tax transfers

• In OECD countries, over 1980-95(Atkinson 2004)
• tax-transfers worsened 1-ary distribution in 3
  cases out of 6
• in all, the generosity redistributivness of
  unempl.fell
• P.I.T.became less progressive, but tax base
  broadened
• Alternative theoretical explanations
• Lower progressivity/simplification prevailed on
  tax broadening
• Gradual dominance of (non-graduated) indirect
  taxes
• Race to bottom to attract FDI erode CIT yields
  
• Redistribution not enough to compensate rising
  ineq 1ary income (chart on Finland)

75
But in egalitarian Finland role of
redistribution rose
76
(No Transcript)
77
VI. Econometric test of equity impact of
liberalisation-globalisation reforms

• Global testing of relation L-G policies-income
  inequality is difficult, as measures of L-G are
  very rough and arbitrary
• L-G proxied by Overall Reform Index (ORI)
  developed by Lora for L.Amerca and E.Europe over
  1980-99 1989-2001.
• ORI was lagged 5 years, as reforms take time to
  alter income distribution. Introduced (see
  Behrman et al 1999) reform increments for
  subsequent 5 years
• ORI was interacted with initial Gini, while GDP/c
  its square were introduced to test if reforms
  impact was same at different income levels.
• several controls (M2/GDP, external debt, real
  interest rate, external terms of trade and
  inflation) were introduced
• Hausmann test suggests carry out econometric
  analysis with FE

78
Results of regression analysis dependent
variable is Gini coefficient
79
Preliminary econometric results

• lagged ORI raised significantly inequality in
  EE-LA
• so did reform increments during subsequent five
  years, with declining effect in EE.
• In LA, interaction betw. ORI initial Gini is
  negative/significant, suggesting that recent
  reform raised Gini most where initial level was
  low. Test has no sense in EE
• parameters of GDP GDP2 suggest policy reforms
  affect most low income nations.
• Control var. only RER (EE), public debt (LA)
  significant, explain small Gini rise

80
Preliminary econometric results

• The F of overall test is satisfactory, R2 is very
  low but is less important in panel data
• Precaution needed in interpreting results, as
  they may be affected by specification problems,
  heterogeneity, omitted variables, ORI measurement
  error, endogeneity
• Cannot attribute conclusively recent inequality
  rise to premature reform implement.
• Yet, ORI-Gini association is robust support
  conclusions of theoretical analyses and country
  studies above

81
VII.Overall tentative conclusions

• Various approaches used in this study
• They confirm that Ineq rose in 2/3 countries
  analyzed
• Several factors caused ineq rise business
  cycle, technology, resource curse. Core factors
  not central
• Ineq rise seems related in different ways to L-G
  reforms, with large variations across countries -
  instruments
• No uniform impact of trade liberalisation FDI.
  But rising evidence they are disequalizing
• Negative impact of domestic-international
  financial liberalisation
• Institutional change affecting redistribution)
  was weakened
• Reform of tax and transfers have been
  disequalizing
• Labour market liberalisation generates mostly
  negative effects (except in strongly dualistic
  labour markets)
• Impact of privatisation vary with approach and
  sector (land reform is equalizing, insider
  privatisation is not)

82
Conclusions (continued)

• Models on which the LG policies are promoted are
  rather sketchy and rely on restrictive
  assumptions ignoring institutional conditions
• Ex-ante analysis of effects of reforms requires
  more complex models capturing the reality of
  country analyzed
• Some policies (e.g.portfolio liberalisation) do
  not generate favourable outcomes
• Others, (as trade liberalisation) are potentially
  positive, but generate adverse distributional
  (and growth) outcomes if applied under
  asymmetric, poorly sequenced and incomplete
  market conditions characterized by considerable
  rigidities
• countries need policy space to determine which
  policy measures they can choose and this will
  vary with local conditions to reach fairly
  universal targets (e.g. greater trade
  integration)