MGI Poorer Than Their Parents Flat or Falling Incomes in Advanced Economies Full Report PDF
MGI Poorer Than Their Parents Flat or Falling Incomes in Advanced Economies Full Report PDF
MGI Poorer Than Their Parents Flat or Falling Incomes in Advanced Economies Full Report PDF
In the 25 years since its founding, the McKinsey Global Institute (MGI) has sought to develop
a deeper understanding of the evolving global economy. As the business and economics
research arm of McKinsey & Company, MGI aims to provide leaders in the commercial,
public, and social sectors with the facts and insights on which to base management and
policy decisions. The Lauder Institute at the University of Pennsylvania ranked MGI the
worlds number-one private-sector think tank in its 2015 Global Think Tank Index.
MGI research combines the disciplines of economics and management, employing the
analytical tools of economics with the insights of business leaders. Our micro-to-macro
methodology examines microeconomic industry trends to better understand the broad
macroeconomic forces affecting business strategy and public policy. MGIs in-depth reports
have covered more than 20 countries and 30 industries. Current research focuses on six
themes: productivity and growth, natural resources, labor markets, the evolution of global
financial markets, the economic impact of technology and innovation, and urbanization.
Recent reports have assessed the economic benefits of tackling gender inequality, the
global consumers to watch, a new era of global competition, Chinese innovation, and
digital globalization.
MGI is led by JacquesBughin, JamesManyika, and JonathanWoetzel, and chaired by
EricLabayeall four are McKinsey & Company senior partners. MichaelChui, SusanLund,
AnuMadgavkar, and JaanaRemes serve as MGI partners. Project teams are led by the
MGI partners and a group of senior fellows, and include consultants from McKinsey offices
around the world. These teams draw on McKinseys global network of partners and industry
and management experts. Input is provided by the members of the MGI Council, who
co-lead projects and provide guidance: AndresCadena, RichardDobbs, KatyGeorge,
RajatGupta, EricHazan, AchaLeke, ScottNyquist, GaryPinkus, ShirishSankhe,
OliverTonby, and EckartWindhagen. In addition, leading economists, including Nobel
laureates, act as research advisers.
The partners of McKinsey & Company fund MGIs research; it is not commissioned by
any business, government, or other institution. For further information about MGI and to
download reports, please visit www.mckinsey.com/mgi.
PREFACE
The rise of income inequality in advanced economies has generated serious
debate and academic research, with much of the recent attention focused
on the increasing concentration of wealth in the richest segments of the
population. In this report, the McKinsey Global Institute has approached the
issue of inequality from a different perspective by examining the share of
the population whose incomes have stopped advancing when compared
to people in the past with similar incomes or demographic profiles. This is
an aspect of inequality that has received relatively little attention, perhaps
because prior to the 2008 financial crisis less than 2percent of households
in advanced economies were worse off than similar households in previous
years. That has now changed: two-thirds of households in the United States
and Western Europe were in segments of the income distribution whose real
market incomes in 2014 were flat or had fallen compared with 2005.
In this research we set out to quantify the proportion of households in
advanced economies with flat or falling incomes. We try to understand how
much the recession and slow recovery since the financial crisis were the
primary causes, and how much is attributable to other long-run forces. Finally,
to help inform a debate, we catalog interventions that have been used around
the world to address the problem and that could become part of a societal
agenda to overcome the issue.
This research was led by Richard Dobbs, a McKinsey senior partner and
a member of the MGI Council based in London, and AnuMadgavkar, an
MGI partner based in Mumbai. MGI directors Jacques Bughin, James
Manyika, and JonathanWoetzel, and MGI chairman EricLabaye guided
and contributed to the research. We thank MGI partners MichaelChui and
JaanaRemes and MGI senior fellow JanMischke for their insights. We also
thank McKinsey senior partners KalleBengtsson, Heinz-PeterElstrodt
(emeritus), VivianHunt, ScottNyquist, GaryPinkus, SvenSmit, KevinSneader,
and LeonardoTotaro for their contributions. The research team was led by
PranavKashyap and LiesbethHuisman, and comprised OlgaBalusova,
AbhisekGhosh, CatherineHart, ChristyLauridsen, AlexanderMansilya-Kruz,
AditiRamdorai, and RavindranShanmugam. MGI senior editors PeterGumbel
and GeoffreyLewis worked on this report, as did MattCooke, MGI director
of external communications; JuliePhilpot, editorial production manager;
MarisaCarder, JasonLeder, and PatrickWhite, senior graphic designers, and
Margo Shimasaki, graphic designer; and RichardJohnson and MaryReddy,
senior editors, data visualization. MGI and McKinsey colleagues TimBeacom,
AlanFitzgerald, ShishirGupta, DeadraHenderson, JeanHocke, and
MekalaKrishnan also contributed.
We would like to thank our academic advisers for their invaluable insights
and guidance: MartinBaily, BernardL.Schwartz Chair in Economic Policy
Development and senior fellow, economic studies, Brookings Institution;
RichardN.Cooper, MauritsC.Boas Professor of International Economics at
Jacques Bughin
Director, McKinsey Global Institute
Senior partner, McKinsey & Company
Brussels
James Manyika
Director, McKinsey Global Institute
Senior partner, McKinsey & Company
San Francisco
Jonathan Woetzel
Director, McKinsey Global Institute
Senior partner, McKinsey & Company
Shanghai
July 2016
CONTENTS
In brief
HIGHLIGHTS
23
47
Changing demographics
72
Focusing on skills
Bibliography Page 95
IN BRIEF
FLAT OR FALLING
a new look at
INCOME INEQUALITY
The population with flat or falling incomes has surged in advanced economies
6570%
<2%
19932005
6570%
200514
MARKET INCOME
DISPOSABLE INCOME
<10 MILLION
540M580M
<10 MILLION
170M210M
Both the extent of flat or falling incomes and the forces driving the
phenomenon vary considerably among countries.
% of population in groups with flat or falling market income, 200514 1
AGGREGATE DEMAND
97
81
DEMOGRAPHIC CHANGES
70
70
6570
63
LABOR-MARKET SHIFTS
Lower share of GDP flowing to wages;
weak demand for low- and
medium-skill labor
CAPITAL INCOME
Lower investment returns and
business income
20
AVERAGE OF
25
ITALY
UNITED
STATES
UNITED
KINGDOM
NETHERLANDS
FRANCE
SWEDEN
ADVANCED
ECONOMIES2
+4
+3
+6
+4
+7
+4
+4
15
+10
+1
+1
+2
+5
+4
+10
+3
+5
+4
In a worst-case scenario, 7080% of income groups might not advance in the coming decade
WHAT CAN
BE DONE?
ENABLE BUSINESSES TO
GROW AND CREATE JOBS
INCREASE OPPORTUNITIES TO
IMPROVE EARNING POTENTIAL
2014 or latest available data for market income (wages and income from capital); population measured in income deciles.
Population-weighted average.
SOURCE: McKinsey Global Institute analysis
1
2
SECURE INCOMES
Adjust taxes, transfers, and labor
policies; encourage business
initiatives in profit-sharing and
employee benefits
EXECUTIVE SUMMARY
Most people growing up in advanced economies since World War II have been able to
assume that they and their children will be better off than their parents and grandparents
and for most of the time, that assumption has been correct. Over the past 70 years, except
for a brief hiatus in the 1970s, buoyant economic and employment growth has meant
that all households, especially those of the baby boomer generation, experienced rising
incomes, both before and after paying taxes and receiving government transfers such as
unemployment or social security benefits.
65-70%
Households in
income groups
with flat or falling
market incomes in
200514
That positive income trend came to an abrupt halt in the past decade. Our research shows
that in 2014, between 65 and 70percent of households in 25 advanced economies were
in income segments whose real market incomesfrom wages and capitalwere flat or
below where they had been in 2005.1 This does not mean that individual households
wages necessarily went down but that households earned the same as or less than similar
households had earned in 2005 on average. In the 12 preceding years, between 1993 and
2005, this flat or falling phenomenon was rare, with less than 2percent of households not
advancing. In absolute numbers, while fewer than tenmillion people were affected in the
19932005 period, that figure exploded to between 540milllion and 580million people
in 200514. Taxes and transfers helped soften the blow, but disposable incomes were
nonetheless flat or down in 20 to 25percent of income segments on average.
The severe recession that followed the 2008 financial crisis and the slow-growth recovery
since are a fundamental cause of this phenomenon, but we find that deep-rooted
demographic and labor-market factors also played a roleand will likely continue doing
so, even if economic growth accelerates. These factors include shrinking households, a
smaller share of GDP going to wages, and increased automation in the workplace. Even in
the 200514 period, market incomes in most of the countries we studied would have risen
slightly had it not been for such changes. In this report, we detail the extent of the flat or
falling phenomenon and the underlying factors, and outline some options for dealing with
what is potentially a corrosive social and economic development.
poor, those with insufficient income to provide for their basic needs, often calculated as a
percentage of the median income.
Our research looks at a third aspect, which has not been as widely studied or documented:
the very rapid growth in the proportion of income segments in advanced economies whose
earnings both before and after taxes and transfers have been flat or falling. This goes
beyond the degree of inequality measured in the standard Gini index by providing a detailed
view of the trajectory of all income segments, which can be lost in a consolidated index. We
focus on income rather than on wealth or consumption, and we also look at the evolution of
incomes over time, rather than at a fixed point.
In our research, we used three approaches to size this flat or falling phenomenon. The
first analyzed changes by income segments, or households divided into deciles (tenths),
quintiles (fifths), and evenpercentiles (one-hundredths) depending on where they rank in the
national income distribution.3 We examined income segments in six advanced economies
(France, Italy, the Netherlands, Sweden, the United Kingdom, and the United States) to
determine how they have fared over the past two decades.4 We then scaled up the findings
to include 19 other advanced economies with similar growth rates and income distribution
patterns, for a total of 25 countries with a combined population of about 800million that
account for just over 50percent of global GDP.5 Our second approach was an analysis of
a detailed data set for 350,000 people in the three countries with microdata available
France, Italy, and the United States. For these countries we examined income by age
bracket and educational attainment. Finally, we sought to understand perceptions through
conducting detailed surveys of more than 6,000 people in France, the United Kingdom, and
the United States that tested how people felt about the evolution of their income.
We did not conduct a longitudinal study to examine intergenerational changes in income
level or social mobility. The numbers of people or households that we report are thus based
on income or population segments rather than on individuals. Nonetheless, the overall trend
is striking, given the hundreds of millions of people in segments with flat or falling income.
Full details of our methodology are to be found in the technical appendix at the end of
this report.
Executive summary
Exhibit E1
The percentage of households in income segments with flat or falling incomes
exploded in the past decade
Share of households with flat or falling incomes1
%
By market income
By disposable income
6570
2025
<2
Millions of people
<2
19932005
200514
<10
540580
19932005
<10
200514
170210
1 Population-weighted average of 25 countries extrapolated from six country deep dives; for each country we use the
latest year the data are availableFrance (2012), Italy (2012 market incomes, 2014 disposable incomes), the
Netherlands (2014), Sweden (2013), United Kingdom (2014), and United States (2013). The base year for France is
1996 and for Sweden is 1995.
SOURCE: McKinsey Global Institute analysis
The impact was smaller when measured in disposable income. But even after accounting
for higher net transfers to households because of the recession, disposable incomes on
average were flat or down in 20 to 25percent of income segments.
The distribution of flat or falling incomes varies across the six economies we studied in
depth. At one extreme is Italy, which experienced a severe economic contraction in the
recession after the 2008 financial crisis and has had a very weak recovery since. There, real
market incomes were flat or falling for virtually the entire population. At the other extreme is
Sweden, where only 20percent of the population had flat or falling market incomes. In each
of the four other focus countriesFrance, the Netherlands, the United Kingdom, and the
United Statesthe proportion of segments whose market incomes did not advance was in
the 60 to 80percent range.
The variation was greater at the level of disposable income. The share of income segments
whose disposable income did not advance between 2005 and 2014 ranged from
100percent in Italy to 10percent in France and less than 2percent in Sweden and the
United States. These variations reflect differences in policy approaches; labor institutions
such as the strength of unions and their role, or services for the unemployed; and widely
varying national economic, fiscal, and monetary policy responses to the recession.
ExhibitE2 shows how income segments in each of our six focus countries fared during the
200514 period.
Income inequality
ES
Poorer than theirmc
parents?
Flat or falling incomes in advanced economies
0712
Exhibit E2
How income groups in our six focus countries fared before taxes and transfers
Real household market income
change, 2005141
%
France
63
Italy
40
40
30
30
20
20
10
10
-10
-10
-20
-20
-30
-30
10
Decile
70
40
30
30
20
20
10
10
-10
-10
-20
-20
2
-30
10
Decile
70
40
30
30
20
20
10
10
-10
-10
-20
-20
2
10
20
10
United States2
40
Decile
United Kingdom 2
-30
Sweden
40
97
Decile
Netherlands2
-30
10
-30
81
Quintile3
Decile
1 Growth numbers are standardized to make both periods comparable for all countries. For each country we use the latest year the data are availableFrance
(2012), Italy (2012), the Netherlands (2014), Sweden (2013), United Kingdom (2014), and United States (2013).
2 Data show that the increase in the bottom decile incomes in the Netherlands is driven by increase in self-employment income, while in the United Kingdom,
the bottom three deciles saw gains in both self-employment and wage income. In the United States, the decrease in the incomes of the top 5% is driven by a
decrease in capital income for the top 1% from 2005 to 2013 based on available CBO data.
3 US income is available only in quintiles except for the top quintile which is broken up into the 81st to 90th, 91st to 95th, 95th to 99th and top 1 percentiles.
SOURCE: Institut national de la statistique et des tudes conomiques (INSEE); Bank of Italy; Centraal Bureau voor de Statistiek (CBS); Statistics Sweden; UK
Office for National Statistics (ONS); US Congressional Budget Office (CBO); McKinsey Global Institute analysis
Executive summary
20X
as many single
mothers in the
United States were
in the lowestincome households as in the
highest-income
ones
This second set of data confirmed our sizing results from the first analysis by income
segments. We found that income from wages fell for all population segments between 2002
and 2012, regardless of age or level of education.
In all three countries, less-educated workers, and especially younger ones, have been most
affected. Moreover, the recession and weak recovery in some of the countries have led to
persistently high levels of youth unemployment, preventing young people across advanced
economies from launching careers. These are the people who are literally at risk of growing
up poorer than their parents.
Women are also overrepresented in lower income deciles. Single mothers were more likely
to be in segments that were not advancing, although there is a variance among countries.
In the United States, 20 times as many single mothers were in the lowest-income decile as
in the highest. In Italy, there were eight times as many single mothers in the lowest income
households as in the highest-income households. For France this number was 11 times.
Our microdata for the United States show that single-mother households not only earn
less than the average household, but their real household income also declined nearly one
percentage point faster than all other households in the decade from 2003 to 2013.
30-40%
of survey
respondents said
their incomes were
not advancing
The survey provided an indication of the potentially corrosive social and economic
consequences of flat or falling incomes. Along with questions about income trends, we
asked about peoples views on trade and immigration. The citizens who held the most
negative views on both were the same group who felt their incomes were not advancing and
did not expect the situation to improve for the next generation. More than half of this group
agreed with the statement, The influx of foreign goods and services is leading to domestic
job losses, compared with 29percent of those who were advancing or neutral. They were
also twice as likely to agree with the statement, Legal immigrants are ruining the culture
and cohesiveness in our society, compared to those advancing or neutral. Our survey also
found that those who were not advancing and not hopeful about the future were more likely
than those who were advancing to support nationalist political parties such as Frances
National Front or, in the United Kingdom, to support the move to leave the European Union.
Wealthier households have a lower marginal propensity to consume. For a discussion of this phenomenon
and its effect on growth, see A window of opportunity for Europe, McKinsey Global Institute, June 2015.
8
In our analyses of factors causing flat or falling incomes, we standardize the growth rate from 1993 to 2005
and 2005 to 2014 in order to make them comparable. For details, see the technical appendix.
7
Executive summary
Five factors underlie the changes in median incomes that we observe in our focus countries:
Aggregate demand factors. When aggregate demand (or GDP) grows, employment,
and labor-force participation also increase, enabling incomes to rise. Conversely, lower
labor-force participation rates, rising unemployment, and waning productivity (output
per worker) can all lead to stagnating or falling incomes. Unemployment in particular can
have a dampening effect on household income.
Demographic factors. These capture changes in the number of working-age people in
each household. This number has fallen in several of our focus countries because of the
shrinking size of households, the result of changing family structures and lower fertility
rates, and aging, which decreases the number of people available to work.
Labor-market factors. These include the evolving pattern in labor demand and supply.
This is manifested in the wage share of GDP and the median households share of
wages. Among the forces that can explain movements in these two factors are income
gains for high-skill workers and negligible income gains or declines for low- and mediumskill workers, and the share of part-time and temporary work, which is often less well
paid proportionately than permanent or full-time work. Labor-market factors can vary
depending on the role and influence of unions, different national labor regulations
and practices, trade and immigration, and the degree to which jobs are affected
by automation.
Capital income factors. These include capital gains from asset sales, interest and
dividends from investments, rental income, income from business, or income received
from private pension plans.
Tax and transfer factors. Transfers include a range of cash payments to beneficiaries
such as social security payments, disability or workers compensation, and
unemployment benefits.9
The first three of these categoriesaggregate demand, demographic, and labor-market
factorscontribute to changes in labor income. Changes in market income are driven by
changes in this labor income, together with changes in capital income. Disposable income
is the amount households receive after taxes, and transfers are applied to market income.
ExhibitE3 shows how each of these factors played a role in the 200514 period, and the
difference with the previous 1993 to 2005 period, by country.
Let us now explore each of these in turn.
In-kind transfers such as the Supplemental Nutrition Assistance Program, Medicare, and Medicaid are
counted for the United States but not for the other five countries due to lack of data on in-kind transfers
by decile.
Exhibit E3
Five factors determine changes in disposable income
Change in disposable income for middle-income households,
19932005 and 2005141
%
DisposAggregate
able
demand
income,
factors2
start year
France
Italy
Netherlands
Sweden
United
Kingdom
United
States
130
120
110
100
90
80
130
120
110
100
90
80
130
120
110
100
90
80
130
120
110
100
90
80
130
120
110
100
90
80
130
120
110
100
90
80
Demographic
factors3
Market
income
change
+6
119
-1
118
100
100
-3
-5
-2
+6
+18
106
+3
-5
+7
+15
+1
+3
-1
+19
+4
-1
109
-1
+4
97
-12
111
-15
-6
Tax and
transfer
factors6
-5
Disposable
income,
end year
114
+3
103
-6
-5
Capital
income
factors5
+4
+29
200514
Labor
income
change
-4
-2
Labormarket
factors4
+16
+13
19932005
-1
108
-2
93
-2
105
+10
98
95
-5
107
86
+2
88
+3
121
121
-1
120
+10
112
113
+5
117
+5
121
+4
125
-1
124
-8
94
+1
95
+4
100
-9
109
+2
111
+4
115
+5
+1
-7
96
96
101
1 Middle-income, or median, households are households in the middle (3rd) quintile or the 5th and 6th decile or the 40th to the 59th percentile. For each
country we use the latest data availableFrance (2012), Italy (2012), the Netherlands (2014), Sweden (2013), United Kingdom (2014), and United States
(2013). The base year for France is 1996 and for Sweden is 1995. All growth numbers are standardized to make results comparable.
2 Change in aggregate output, measured by output per employed worker, multiplied by change in number of employed workers in the working-age population.
3 Change in number of working-age people per household.
4 Change in wage share of GDP, adjusted for difference between consumer price inflation and inflation of overall output, and median household share of
wages.
5 Includes profit from own business, income from capital, and other sources of market incomes that cannot be classified as income from labor.
6 Includes income from private and public pension transfers, other transfers such as social security benefits, and taxes on labor income and capital income.
NOTE: Numbers may not sum due to rounding.
SOURCE: INSEE; Bank of Italy; CBS; Statistics Sweden; ONS; CBO; McKinsey Global Institute analysis
Executive summary
The recession and subsequent slow recovery sharply reduced the effect of
aggregate demand on market incomes
After the global financial crisis in 2008, GDP contracted in each of the economies we
studied in depth, raising unemployment rates sharply and reducing median incomes.
Labor productivity growth, which was already slowing in the 200007 period, has slowed
even further since the crisis. In the 12-year period before the recession that is our baseline
(1993 to 2005), growth in aggregate demand contributed 19 percentage points to median
disposable income growth in the United States and 17 points, on average, in the five
European countries we focused on. In 2005 to 2014, which included the recession and
its aftermath, that figure plunged to just four percentage points in the United States and
in Europe.
On average,
18
percentage points
of growth in
median disposable
incomes in
19932005 came
from aggregate
demand
The recession in the United States was severe but relatively short-lived: GDP dropped
by 3.4percent from peak to trough from 2008 to 2009, and growth was negative for five
quarters, ending in late 2009. Unemployment doubled from less than 5percent to nearly
10percent between 2007 and 2010. Europe overall suffered a double-dip recession,
when growth stalled in 2012 during the Eurozones sovereign debt crisis. Italy suffered a
quadruple-dip recession with growth stalled or falling almost continuously from 2007
through 2015; over that period, GDP contracted by 12.2percent. Unemployment rates
in Europe rose at an accelerated pace after the second dip, doubling from less than
4percent in the Netherlands in 2008 to nearly 8percent in 2014. In France, unemployment
reached its highest level since the crisis10.8percentin the third quarter of 2015. Italys
unemployment rate peaked at 12.9percent in the third quarter of 2014.
The recovery has been slow and uneven across countries. At the end of 2015, seven years
after the recession began, per capita GDP had not returned to pre-recession levels in Italy
and the Netherlands, though it had recovered in the other four countries. The US economy
has recovered faster than the other five, with GDP per capita rising 1.3percent per year
between 2009 and 2015. This compares with 0.9percent across the European Union (EU).
However, even as US GDP per capita growth rebounded and the US unemployment rate
returned to the pre-crisis level in 2015, median market incomes remained flat between
2011 and 2014. The United Kingdom suffered a double-dip recession, but employment has
returned to the pre-recession level.
Slow productivity growth in turn has raised questions about the link between productivity
and inequality.10 While the largest change from the 19932005 period was the lower levels of
aggregate demand growth, that alone was not enough to depress incomes and determine
which income segments bore the pain to a greater or lesser degree.11 Indeed, aggregate
growth remained positive for all countries except Italy, and yet most income segments had
flat or falling incomes. That was because two other long-run factors also weighed heavily on
income advancement.
10
11
household incomes in the past two decades, especially in Europe, and will continue to
do so. The number of working-age adults per household fell in both the 19932005 and
200514 periods across the five European economies we analyzed, reducing income by
the equivalent of four percentage points in both periods. The drop in household size was
greatest in Italy, where there were 21 fewer working-age people per hundred households
in 2012 than in 2002.12 In the United States, by comparison, the number of working-age
people per hundred households dropped on average between 2002 and 2012 by just two.
However, in the United States, the bottom quintile of households has on average 50 fewer
working-age people in every hundred households than the richest quintile.13
Labor-market factors have depressed wage growth for middle- and lowskill workers
Two labor-market factors contributed to flat or falling incomes and have been particularly
pronounced in the United States, as well as the Netherlands and the United Kingdom.
5%
average decline in
wage share as a
percentage of GDP
in our 6 countries
First is the share of national income that is paid to workers, the so-called wage share.14
Specifically, we look at wages and salaries paid to workers, rather than all compensation
to employees, to remove the effects of non-wage labor costs such as employer pensions
and National Insurance contributions in the United Kingdom.15 From 1970 to 2014, with the
exception of a spike during the 197374 oil crisis, the average wage share fell by 5percent
on an indexed basis in the six countries we studied in depth, and in the most extreme case,
the United Kingdom, by 13percent. The decline in wage share has taken place despite
rising productivity, suggesting a disconnect between productivity and incomes. The wageshare decline is due in part to the growth of corporate profits as a share of national income,
as a result of rising capital returns to technology investments, lower returns to labor from
increased trade, rising rent incomes from home ownership, and increased depreciation on
capital.16 Indeed, profits for North American and Western European corporations in the past
three decades have been exceptional, with after-tax operating profits rising to 9.8percent
of global GDP in 2013 from 7.6percent in 1980, an increase of nearly 30percent. Between
2010 and 2014, after-tax profits of US firms measured as a share of national income even
exceeded the 10.1percent level last reached in 1929.17
The second factor is the uneven distribution of this wage share among different income
segments. Since 1993, households in the uppermost income segments in our six focus
countries have on average received a growing share of the total wages, even as the share for
low- and middle-income segments has either stagnated or fallen. This is not the case in all
Shrinking household size in a country may not affect per capita income but does lead to falling equivalized
household income, a measure of household income adjusted for the number of dependents. This attempts
to account for the economies of scale that come with living in larger households (additional household
members receive lower weighting to reflect economies of scale). Needs of households fall with size, but not
proportionately, since housing, utilities, and other necessities are not used on a per capita basis; one resident
uses the same amount of heat as two, for example. For a detailed discussion of this concept, see OECD
framework for statistics on the distribution of household income, consumption and wealth, OECD, June 2013.
13
This quintile also had falling market incomes in 200514.
14
Other factors that are included in the national income are rent, interest, and profits. In this report, we use GDP
as a proxy for gross domestic income due to the negligible statistical discrepancy between the two numbers.
15
In the United Kingdom, unfunded liabilities in defined-benefit pension schemes are creating downward
pressure on wages to workers. See Conor DArcy and Gavin Kelly, Securing a pay rise: The path back to
shared wage growth, Resolution Foundation, March 2015, and Brian Bell, Wage stagnation and the legacy
costs of employment, Centre for Economic Performance, London School of Economics, paper number
CEPCP 458, November 2015.
16
While overall spending on capital goods has been weak, there has been considerable investment in
information technology, whose prices have declined. See Loukas Karabarbounis and Brent Neiman, The
global decline of the labor share, NBER working paper number 19136, June 2013; Loukas Karabarbounis and
Brent Neiman, Declining labor shares and the global rise of corporate saving, NBER working paper number
18154, June 2012; and How CBO projects income, Congressional Budget Office (CBO), July 2013.
17
See Playing to win: The new global competition for corporate profits, McKinsey Global Institute, September
2015, and Diminishing returns: Why investors may need to lower their expectations, McKinsey Global
Institute, May 2016.
12
10
Executive summary
countries: in France, Italy, and Sweden, for example, the share of upper income households
actually declined somewhat in the 200514 period. However, in the Netherlands, the United
Kingdom, and the United States, upper income households experienced strong wage
growth while low-income and middle-income segments fell back sharply.
For a further description of the emergence of a global labor force, see The world at work: Jobs, pay and skills
for 3.5billion people, McKinsey Global Institute, June 2012.
19
Gianmarco Ottaviano, Offshoring and the migration of jobs, IZA World of Labor, volume 170, July 2015;
David Autor, David Gorn, Gordon H. Hanson, The China shock: Learning from labor market adjustment to
large changes in trade, NBER working paper number 21906, January 2016.
20
Ton Wilthagen and Frank Tros, The concept of flexicurity: A new approach to regulating employment and
labor markets, Transfer, volume 10, number 2, 2004.
21
OECD employment outlook 2015, OECD, July 2015.
18
11
Exhibit E4
Employment has been lower for low- and medium-skill workers, and they are more likely to be employed on
temporary contracts
United
Kingdom
Italy
France
High skill
Netherlands
Sweden
Great Recession
Axis midpoint
Medium skill
Low skill
Employment rate
%
90
90
90
80
80
80
70
70
70
60
60
60
50
50
50
40
1994
2004
2014
40
1994
2004
2014
40
1994
2004
2014
1994
2004
2014
Temporary employment
Thousand people
30
30
30
25
25
25
20
20
20
15
15
15
10
10
10
1994
2004
2014
1994
2004
2014
Differences in union rates and labor regulation influenced outcomes for some
income and demographic segments
National labor-market institutions and practices that shaped the outcomes in employment
and wages appear to have made a difference in some of our focus countries. For example,
the United States is known for its relatively light labor regulation and flexible labor markets
compared with most European economies. About 11percent of private-sector workers
See A labor market that works: Connecting talent with opportunity in the digital age, McKinsey Global Institute,
June 2015.
23
OECD labor database.
22
12
Executive summary
in the United States are represented by unions, compared with 30percent, on average,
in Europe. During the recession, US companies had greater freedom to cut jobs and
implemented permanent labor-cost savings, despite weak demand.24 In Europe, by
contrast, labor-market rigidity may contribute to youth unemployment. For example, in Italy,
employers hired fewer young workers in the recession following the 2008 financial crisis in
part because of income support schemes for permanent workers. This was compounded
by a 2012 pension reform that kept older workers in the workforce.25
The declining ability of labor to protect its share of national income, and of middle and lower
income segments to protect their share of the wage pool, reduced real median disposable
income growth by nine percentage points in the United States in the 19932005 period
and by seven points in the 200514 period, while only two European economies, Italy and
the Netherlands, experienced this negative effect in the 19932005 period. In the 200514
period, however, labor-market effects did not contribute to median disposable income
growth in France, and had a negative impact in the Netherlands and the United Kingdom,
where union membership has fallen the most steeply in our sample of countries. Italy, which
entered the recession in a weak state and has had the greatest prevalence of flat or falling
incomes, in 2015 introduced labor-market legislation aimed at simplifying rules and rigidities
that have held back employment.
68%
of workers in
Sweden are union
members
In Sweden, where 68percent of workers are union members, the median household
received a greater share of output that went to wagesand received more of the gains from
output growth than households in Swedens top and bottom income deciles in the 200514
period. This reflects Swedish labor policies such as contracts that protect wage rates and
hours worked. After the global financial crisis, the Swedish government worked with unions
to forge agreements for temporary reductions in work hours, which preserved jobs and
helped private-sector employers withstand the downturn.
Capital income factors had a relatively minor effect on median and lowincome households
Capital income is derived from a range of investment and business activities including
interest, dividends, and realized capital gains from financial-market investments, asset
sales, business income, and private pensions. For upper income households, capital
income is significantly more important than for other income segments, an issue that has
become a focus of other income inequality research, including that of French economist
Thomas Piketty.26 For example, in the six countries we study in depth, in 2014, capital
income amounted to 33percent of disposable income for households in the highest income
quintile. That compared with just 7percent of disposable income for the lowest income
quintile, and, for median income households, 14percent of disposable income.
For our analysis, capital income was not a major factor, as the shift between 2005 and 2014
was very small for median and low-income households. As a percentage of disposable
income, for example, the share of capital income in disposable income remained virtually
unchanged on average in our six focus countries for the low- and middle-income quintiles in
200514. In fact, the largest movement in capital income as a share of disposable income
was actually felt by high-income households in the top quintile. For them capital income fell
from 35percent of disposable income in 2005 to 33percent in 2014.
See An economy that works: Job creation and Americas future, McKinsey Global Institute, June 2011.
Antoine Bozio et al., European public finances and the Great Recession: France, Germany, Ireland, Italy,
Spain, and the United Kingdom compared, Fiscal Studies, volume 36, number 4, December 2015.
26
Thomas Piketty, Capital in the twenty-first century, Belknap Press, 2014.
24
25
13
Tax and transfer policies reduced or even reversed the impact of flat or falling
market incomes on disposable income
Taxes and transfers directly affected how declining market incomes translated into
disposable income, and in some countries made a significant difference in reducing or
even reversing the flat or falling phenomenon for some income groups, including middleincome households. Market incomes for median income segments were flat or falling in all
countries except Sweden between 2005 and 2014, but disposable incomes for the median
income segment fell in only two of our six focus countries, Italy and the Netherlands, by 2 to
10percent.
In the United States, net transfers raised median disposable income growth by the
equivalent of five percentage points between 2005 and 2014, turning a four-point decline
in median market income into a one percentage point gain in disposable income. As part
of its stimulus plan, the US government transferred more than $350billion to households
in the form of tax relief and assistance to workers affected by the downturn, including
raising unemployment benefits and extending their duration.27 In France, net transfers on an
indexed basis raised median disposable income by three percentage points above median
market income, while in the United Kingdom, transfers restored disposable income for
median income households to their 2005 level, offsetting the decline in market income.
Central
government debt
in Italy, the United
Kingdom, and the
United States is
close to
100%
of GDP
In the future, governments may find it difficult to sustain this level of spending without
substantial revenue increases; government debt as a share of GDP increased over the past
seven years in five of the six countries we studied. For example, central government debt is
close to 100percent of GDP or higher in Italy, the United Kingdom, and the United States.
In the United States, it rose from 56percent of GDP in 2005 to 97percent of GDP in 2016.
Sweden is the outlier with a relatively flat debt share of GDP between 2008 and 2015; while
it increased debt levels to fund the effects of recession, it started from a lower level, of less
than 40percent in 2008, which gave it greater freedom to spend during the crisis.28 In 2015,
Swedens central government debt remained steady at about 42percent.
Douglas W. Elmendorf, Estimated macroeconomic impacts of the American Recovery and Reinvestment Act
of 2009, CBO letter to Senator Charles Grassley, March 2, 2009.
28
Debt and (not much) deleveraging, McKinsey Global Institute, February 2015.
27
14
Executive summary
Exhibit E5
Wide variations in market and disposable incomes in the two periods were driven by differing tax and transfer
policies across countries
Total growth in income by quintile (%)
Quintiles (1 = bottom, 5 = top)
199320051
2005142
Italy
Netherlands
Sweden
United
Kingdom
United
States
60
50
40
30
20
10
0
-10
-20
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
Italy
Netherlands
Sweden
1 2 3 4 5
United
Kingdom
1 2 3 4 5
United
States
40
30
20
10
0
-10
-20
-30
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
1 All growth numbers are standardized to make results comparable across all countries and both time periods.
2 For each country we use the latest year the data are available: France, 2012; Italy, 2014 disposable incomes, 2012 market incomes; Netherlands, 2014;
Sweden, 2013; United Kingdom, 2014; and United States, 2013.
SOURCE: INSEE; Bank of Italy; CBS; Statistics Sweden; ONS; CBO; McKinsey Global Institute analysis
15
First, government taxes and transfers can play a decisive role in limiting or reversing the
decline of market incomes at the level of disposable incomes. Of our six countries, this was
particularly striking in the United States, where a decline in market incomes for 81percent of
all income segments in the 200514 period translated into an increase in disposable income
for nearly all households. This type of large-scale intervention could be unsustainable,
however, given already high debt levels and the effect on budget deficits. Government
intervention can also accentuate income declines, as happened in Italy, where austerity
measures raised taxes and reduced some benefits, aggravating the drop in market incomes
for all quintiles.
Second, the lowest income groups were not always the segment to bear the brunt of flat
or falling incomes; in all of our focus countries except Sweden, middle-income segments
also felt the impact, as a result of declining income from labor. Higher income segments
also experienced a decline in market income in Italy, the United Kingdom, and the United
States as a result of lower income from capital, which was especially volatile during and after
the 2008 financial crisis. In the United States, higher capital income increased disposable
income growth for the top quintile by 24 percentage points from 1993 to 2005, but pushed it
down by six points from 2005 to 2014.
Looking at some countries individually, Sweden stands out as the only one where market
incomes rose for middle-income households. Sweden had gone through a previous steep
downturn in the 1990s. After 2008, the government focused on job preservation and
creation, adding temporary jobs to the public sector, reducing payroll taxes for businesses,
and providing tax incentives to hire young people and the long-term unemployed.29
In the United Kingdom, the pattern of disposable income from 1993 to 2005 highlights the
outcome of the redistributive policies of the government of Tony Blair at the time, with sharp
income increases for the lowest quintiles. The British economy is highly reliant on revenue
from the financial sector to balance its budgets, and after an initial period of increased
spending after the financial crisis, the government imposed a period of austerity when
financial revenue fell post-2008. More than four-fifths of the fiscal measures associated with
austerity were spending cuts that disproportionately affected working-age people (cuts to
benefits and public-sector jobs, for example), although state pensions were protected from
the cuts. In our data, this decrease in spending seems to have most affected the bottom
quintile from 2005 to 2014, with disposable income growth decreasing by six percentage
points because of taxes and transfers.30
In France, there was a notable difference in the impact of labor-market factors on
different quintiles. These labor-market factors reduced disposable income growth by four
percentage points for the lowest quintile, and increased it by two percentage points for the
top quintile. This could be a reflection of Frances two-tiered labor market, where lowerpaying jobs are often temporary and do not provide the same level of benefits or security.
Moreover, throughout the financial crisis, the unit cost of workers in France continued rising,
and some companies opted to stop hiring and end short-term contracts.31
Dominique Anxo and Thomas Ericson, Labor market measures in Sweden 2008-13: The crisis and beyond,
ILO, February 2016.
30
Antoine Bozio et al., European public finances and the Great Recession: France, Germany, Ireland, Italy,
Spain, and the United Kingdom compared, Fiscal Studies, volume 36, number 4, December 2015.
31
Mathias Andr et al., French public finances through the financial crisis: Its a long way to recovery, Fiscal
Studies, volume 36 number 4, December 2015.
29
16
Executive summary
We model these scenarios for three countriesFrance, Italy, and the United Stateswhere we had the
microdata to estimate employment and wage outcomes for different types of labor market participants,
based on education, age and gender. The consolidated results are based on a simple average of these three
economies, which we use as a proxy for outcomes across advanced economies.
33
Growth and renewal in the United States: Retooling Americas economic engine, McKinsey Global Institute,
February 2011.
34
Erik Brynjolfsson and Andrew McAfee, Race against the machine, Digital Frontier Press, 2011. For an
assessment of automation potential, see Michael Chui, James Manyika, and Mehdi Miremadi, Four
fundamentals of workplace automation, McKinsey Quarterly, November 2015.
35
Digital America: A tale of the haves and the have-mores, McKinsey Global Institute, December 2015.
36
Diminishing returns: Why investors may need to lower their expectations, McKinsey Global Institute,
May 2016.
37
Debt and (not much) deleveraging, McKinsey Global Institute, February 2015.
32
17
70-80%
of income
segments could
experience flat or
falling incomes in
the next decade if
growth remains low
Our high-growth case assumes higher annual GDP growth, of 1.3percent for Italy,
1.8percent for France, and 2.4percent for the United States, with productivity growth in
advanced economies reverting to the 30-year average preceding the financial crisis, about
2percent per year. Unemployment rates would fall as demand accelerates, and we assume
that factors such as demographic shifts and technology adoption would continue to affect
labor-market dynamics and incomes as they have in the past decade. With this sustained
economic upturn, the proportion of household income segments experiencing flat or falling
incomes would drop off sharply but not disappear.
Under these conditions, market incomes might be flat or falling in 10 to 20percent of income
segments across advanced economies. While that is considerably lower than the proportion
of households affected in 200514, it is five to ten times the pre-2005 level.
The relationship between productivity growth and income growth is uncertain, so using
similar productivity assumptions as the high-growth hypothesis, we also modeled a
variation on the high-growth case as a third hypothesis. For this, we incorporated a greater
disruptive impact of technology on employment. This reflects the potential for increasingly
powerful digital technologies to take on many activities now requiring workers, further
reducing demand for low- and medium-skill workers.39 To understand the potential range of
this sensitivity, we assumed, on the basis of prior MGI research, that advances in technology
might automate as much as 15percent of the work that medium-skill workers do.40
Unemployment and underemployment would rise, and the wage share would fall further.
Some 30 to 40percent of the population might be in income segments where real market
incomes in 2025 are flat or down compared with 2012. To sustain disposable incomes,
additional targeted transfers of as much as 5 to 10percent of 2012 net transfers might
be needed.
It should be noted that this labor disruption hypothesis does not fully model the normal
behavior of economies. In reality, the wealth and investment created by rising productivity
would create new types of demand, which would lead to jobs that do not exist today.
This has been the pattern when new technologies have disrupted labor markets in the
past: rising output leads to more profits, which enables new investment, leading to new
The secular stagnation hypothesis, which holds that an oversupply of savings and a lack of investment can
reduce growth, inflation, and the natural equilibrium interest rate, dates back to the 1930s and has gained
renewed attention recently. See Lawrence H. Summers, The age of secular stagnation: What it is and what to
do about it, Foreign Affairs, March/April 2016.
39
For a discussion of the potential effect of automation, see Michael Chui, James Manyika, and Mehdi Miremadi,
Four fundamentals of workplace automation, McKinsey Quarterly, November 2015.
40
Digital America: A tale of the haves and the have-mores, McKinsey Global Institute, December 2015.
38
18
Executive summary
employment and more demand. However, this sensitivity analysis serves to illustrate the
extent to which rapid technological changes could affect income inequality for a sustained
period if they outpace the rate at which workers and employers adapt to the new realities of
the labor market.
19
Reviving growth and enabling a thriving business environment that creates jobs
As we have seen, the economic downturn was a fundamental cause of the lack of income
advancement for a large majority of income segments since 2005. The corollary is that
revival of stronger economic growth will be a key to raising incomes, even in the face of
demographic shifts and labor-market changes that work against them. Conversely, if the
current low-growth world becomes the new normal, the phenomenon of flat or falling
incomes could become entrenched.
The paramount importance of boosting growth through improved productivity is a theme
we have covered extensively in 25 years of MGI research.41 About three-quarters of the
potential for productivity improvements comes from the adoption of existing best practices
and catch-up productivity improvements, while the remaining one-quarter comes from
technological, operational, and business innovations that push the frontier of the worlds
GDP potential. Governments have many opportunities to help boost productivity, including
through measures that would reduce waste and improve resource and energy efficiency,
increase competition and deregulation, or target infrastructure and other investment that
creates new jobs in the short run and shores up economic growth over the longer term.
20
Executive summary
and value-added taxes, payroll taxes, and property taxes can fall heavily on lowand middle-income households. These taxes could be adjusted to raise disposable
incomes for these households. Policy makers can also consider the impact of their
spending decisions on disposable incomes of segments whose incomes are not
advancing. A public transit system, for example, is likely to provide more value for a lower
income household than a new highway.
Where there is political consensus, direct payments such as a guaranteed basic income
scheme or expansion of programs such as the US earned income credit could be used
to maintain disposable incomes, although such measures can be highly controversial.43
Also, where appropriate, labor rules could help lift incomes for segments that have not
been advancing. This might include adjusting minimum wages or extending employment
protections and benefits to part-time and temporary workers, which some countries already
have done.
For example, voters in Switzerland in a June 2016 referendum overwhelmingly rejected the introduction of a
guaranteed basic income.
44
The world at work: Jobs, pay, and skills for 3.5billion people, McKinsey Global Institute, June 2012.
45
See, for example, Dominic Barton, Capitalism for the long term, Harvard Business Review, March 2011;
John Browne, with Robin Nuttall and Tommy Stadlen, Connect: How companies succeed by engaging
radically with society, PublicAffairs, 2016; Kathleen McLaughlin and Doug McMillon, Business and society in
the coming decades, McKinsey Quarterly, April 2015.
43
21
Executive summary
Exhibit 1
The share of households with flat or falling incomes in advanced economies now surpasses the share of
those not catching up or not getting by
Focus of our analysis
Not getting by
Definition of
indicator
Not catching up
Share of households in
income deciles below the
poverty line1
Based on
market income
% of households
Flat or falling
Share of households in
deciles where incomes are
flat or falling
6570
6065
5560
<2
Million people
19932005
2005142
19932005
2005142
480500
500540
<10
540580
6065
Based on
disposable
income
% of households
4550
2025
13
15
<2
Million people
2005
20142
19932005
2005142
19932005
2005142
110
120
370410
500540
<10
170210
1 Market income data is not shown for the not getting by segment as countries do not typically measure pre-transfer poverty rates.
2 For each country we use the latest year the data are availableFrance (2012), Italy (2014 disposable incomes, 2012 market incomes), the Netherlands
(2014), Sweden (2013), United Kingdom (2014), United States (2013).
SOURCE: McKinsey Global Institute analysis
24
Thomas Piketty, Capital in the twenty-first century, Belknap Press, 2014. Other works that have examined this
aspect include Philippe Aghion et al., Innovation and top income inequality, CEPR discussion paper number 10659,
June 2015; Anthony Atkinson, Inequality: What can be done? Harvard University Press, 2015; David Autor, Skills,
education, and the rise of earnings inequality among the other 99percent, Science, volume 344, issue 6186, May
2014; Franois Bourguignon, The globalization of inequality, Princeton University Press, 2015; Brian Keeley, Income
inequality: The gap between rich and poor, OECD, December 2015; Branko Milanovic, Global inequality: A new
approach for the age of globalization, Harvard University Press, 2016; Jos Gabriel Palma, Homogeneous middles
vs. heterogeneous tails, and the end of the Inverted-U: The share of the rich is what its all about, Development and
Change, volume 42, number 1, January 2011; Joseph E. Stiglitz, The price of inequality: How todays divided society
endangers our future, W. W. Norton, 2012.
2
Others who focus on social mobility include Philippe Aghion et al., Innovation and top income inequality, CEPR
discussion paper number 10659, June 2015.
3
Thomas Piketty, Capital in the twenty-first century, Belknap Press, 2014.
4
Franois Bourguignon, The globalization of inequality, Princeton University Press, 2015; Branko Milanovic, Global
inequality: A new approach for the age of globalization, Harvard University Press, 2016.
5
Thomas Piketty, Capital in the twenty-first century, Belknap Press, 2014; In it together: Why less inequality benefits all,
OECD, 2015.
1
25
The 25 advanced economies we scaled up to are Australia, Austria, Belgium, Canada, Czech Republic,
Denmark, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Luxembourg, the Netherlands,
New Zealand, Norway, Portugal, Slovenia, Spain, Sweden, Switzerland, the United Kingdom, and the United
States. Countries including Japan and South Korea are not included for lack of available comparative data.
47
For an overview of income growth by quintile in the United States between 1947 and 1989, see Paul
Krugman, The rich, the right, and the facts: Deconstructing the income distribution debate, American
Prospect, fall 1992.
48
World Banks World Development Indicators data, available starting in 1960.
49
See John Helliwell, Comparative macroeconomics of stagflation, Journal of Economic Literature, volume 26,
issue 1, 1988.
50
Our analysis shows that for middle-income households, net transfers are typically 5 to 15percent of
disposable income, while in the poorest households (the tenthpercentile), transfers can provide 50 to
70percent of disposable income. We cannot say that a specific family has less income than it had a decade
ago, because we did not track incomes for individual households. However, from studies of social mobility,
we know that most people remained in the same income decile over the period of a decade and are therefore
stuck or moving backwards in terms of household income. See sidebar, Social mobility has done little to raise
incomes in advanced economies, later in this chapter for a discussion of social mobility.
46
26
Exhibit 2
In the past decade 65 to 70 percent of income segments in advanced economies had flat or falling market incomes
and 20 to 25 percent had flat or falling disposable incomes
% of households in income segments with flat or falling income, 2005141
Market income
Disposable income
2025
6570
Weighted average2
Italy
97
81
United States
United Kingdom
70
Netherlands
70
100
<2
60
70
63
France
20
Sweden
10
<2
1 For each country we use the latest year the data are availableFrance (2012), Italy (2014 disposable incomes, 2012 market incomes), the Netherlands
(2014), Sweden (2013), United Kingdom (2014), United States (2013).
2 Population-weighted average of 25 countries extrapolated from six country deep dives.
SOURCE: INSEE; Bank of Italy; CBS; Statistics Sweden; ONS; CBO; McKinsey Global Institute analysis
In Italy, market incomes declined for virtually all income groups (97percent of the
population).51 In France, the Netherlands, the United Kingdom, and the United States, 60 to
80percent of income segments did not advance in market income. Sweden was the outlier:
in only two deciles were market incomes flat or declining.
Disposable income was also affected in every one of these six countries from 2005 to
2014, but the pattern was quite different from the pattern of market incomes, largely as a
result of different macroeconomic conditions and policy responses. Disposable income
in Italy was flat or fell across all income deciles. That was largely the result of tax increases
and reductions in benefits driven by austerity measures. Disposable incomes were flat
or falling in 60 to 70percent of the population groups in the Netherlands and the United
Kingdom. However, in the United States, only the top 1percent of the population had flat or
falling disposable income, while in Sweden disposable incomes rose for every decile. This
cushioning of disposable income was attributable to substantial government intervention
after the 2008 financial market crisis.
In Chapter 2 we discuss the factors underlying this phenomenonthe sharp economic
recession and weak recovery after the 2008 global financial crisis, long-run factors such
as demographic and labor-market changes, capital income movements, and government
Not all flat or falling groups are multiples of 10percent due to the availability of more granular data in some
countries. For Italy, we have access to household income microdata that we grouped intopercentiles to find
the percentage ofpercentile groups that had flat or falling market income. For disposable income in Italy,
we use data from the Bank of Italy, which groups households into deciles. In France, we use microdata from
72,000 households to calculate thepercentile groups that had flat or falling market and disposable income.
For the United States, the Congressional Budget Office provides quintile data for the bottom 80percent
of the population and also provides data on the 80th90thpercentile, the 90th95thpercentile, the 95th
99thpercentile and the top 1percent.
51
27
taxes and transfers. To ensure that the outcomes were not unduly influenced by our choice
of the years we compared, we also conducted an analysis from 2003 to include four years
of pre-recession growth. It found similar results in terms of the number of households not
advancing. On average, 40 to 50percent of deciles had flat or falling real market incomes,
and for 10 to 20percent, disposable incomes were not advancing either. This suggests that
other factors were affecting income growth even before the steep downturn.
This phenomenon has primarily affected advanced economies. In developing economies,
we find that even though income inequality is rising, the phenomenon of flat or falling
incomes has not been a widespread problem (see sidebar, With few exceptions, incomes
in developing economies have been rising for all groups).
With few exceptions, incomes in developing economies have been rising for all groups
From the 1980s to the early 2000s, hundreds of millions of 36percent across income deciles between 2005 and
people were brought out of extreme poverty in developing 2012.3 In all three countries, the poorthose in the bottom
half of the income distributionbenefited from increases
economies. In 1981, twobillion people, or 54percent
in government transfers, often funded by resource
of the population in low- and middle-income countries,
exports, and from rising public-sector wages.
lived below the extreme poverty line; by 2011, that figure
was down to onebillion people, or only 15percent of
As in the advanced economies we studied, income trends
the population in low- and middle-income countries.1 In
in developing economies are heavily influenced by GDP
China, the share of the population in extreme poverty
growth. In 2015, with high inflation in Brazil and the sharp
dropped from 88percent in 1980 to approximately
drop in oil prices that affected Russia, the economies
5percent in 2015. Extreme poverty still averaged
of both countries changed course and fell into deep
43percent in sub-Saharan Africa in 2015.2
recession, accompanied by falling wages and increased
While inequality has risen along with wealth in many
poverty rates.4
developing economies, people at all levels of the income
distribution were advancing during most of the 2000s
3
Russia Longitudinal Monitoring Survey (RLMS-HSE) conducted
and early 2010s. In Brazil and Russia, for example, labor
by the National Research University Higher School of Economics,
income rose across all income deciles during the period
and ZAO Demoscope together with Carolina Population Center,
University of North Carolina at Chapel Hill and the Institute of
of robust output growth between 2003 and 2013. In India,
Sociology of the Russian Academy of Sciences; Instituto Brasileiro
total household consumption climbed between 29 and
1
2
28
Exhibit 3
Wide variations in market and disposable incomes in the two periods were driven by differing tax and transfer
policies across countries
Total growth in income by quintile (%)
Quintiles (1 = bottom, 5 = top)
199320051
2005142
Italy
Netherlands
Sweden
United
Kingdom
United
States
60
50
40
30
20
10
0
-10
-20
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
Italy
Netherlands
Sweden
1 2 3 4 5
United
Kingdom
1 2 3 4 5
United
States
40
30
20
10
0
-10
-20
-30
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
1 All growth numbers are standardized to make results comparable across all countries and both time periods.
2 For each country we use the latest year the data are available: France, 2012; Italy, 2014 disposable incomes, 2012 market incomes; Netherlands, 2014;
Sweden, 2013; United Kingdom, 2014; and United States, 2013.
SOURCE: INSEE; Bank of Italy; CBS; Statistics Sweden; ONS; CBO; McKinsey Global Institute analysis
29
In terms of magnitude, middle-income groups in the United States, the Netherlands, Italy,
and the United Kingdom experienced the largest fall in market incomes, of 4 to 10percent.
In France, the market incomes of middle-income groups were flat, while in Sweden
they increased.
The picture is more varied at the level of disposable income. Italy was the outlier, as
disposable income fell in all quintiles from 2005 to 2014. In the Netherlands, disposable
income fell for all but the richest quintile, while in the United Kingdom, disposable income
fell for all but the second quintile. In Sweden, real disposable income grew across all income
segments, while in the United States it rose for all quintiles except the top one.
Italians situation actually worsened after government taxes and transfers, with their
disposable incomes decreasing an additional 3 to 5percent. In Sweden, disposable income
growth during the second period was very similar to the first period. Transfers reduced the
extent of the decline in disposable income of middle-income deciles in the United Kingdom
and the Netherlands, although the trend was still down. But in France and the United States,
transfers reversed the decline in market incomes, with disposable income for middleincome deciles rising slightly.
We selected these three variables based on multivariable regressions that included other factors such
as location and family status. As in the case of income deciles, this analysis does not take into account
movement between segments over time (a worker completing a college degree, for example).
53
In France, lower educated is lower secondary school diploma or less; medium, secondary school diploma;
higher, bachelors degree or higher. In Italy, lower educated is middle school diploma or less; medium, high
school diploma; higher, bachelors degree or higher. In the United States, lower educated is less than a high
school diploma; medium, high school diploma or associates degree; higher, bachelors degree, graduate
degree, or higher. US data are calculated from 2003 to 2013.
54
Incomes of young and lower-educated Italians dropped by 30 percent or more, but the sample size was too
small to be reliable.
52
30
under 30 with high education experiencing half the decline of the least educated workers
under 30 in the United States and Italy, and a fifth of the decline in France.
Exhibit 4
Younger and less-educated workers were more likely to face income declines than other types of workers
France1
Italy2
United
States3
Age
Years
Share of
labor force
in age group
%
<30
20
3045
39
>45
41
-3
<30
16
NA4
3045
42
NA4
>45
43
<30
24
3045
31
>45
44
Lower educated
Medium educated
-10
Higher educated
-10
-7
-2
-10
-6
-9
-27
-3
-21
-15
-13
-14
-15
-14
-15
-14
-11
-8
-6
-7
-4
-6
-2
1 In France, lower educated is lower secondary school diploma or less; medium, secondary school diploma; higher, bachelors degree or higher. Incomes in
France include self-employment income.
2 In Italy, lower educated is middle school diploma or less; medium, high school diploma; higher, bachelors degree or higher.
3 In the United States, lower educated is less than a high school diploma; medium, high school diploma or associate's degree; higher, bachelors degree,
graduate degree, or higher; US data compare 2003 to 2013.
4 Sample size too small for reliable measurement.
NOTE: Numbers may not sum due to rounding.
SOURCE: ONS; Bank of Italy; INSEE; US Current Population Survey; McKinsey Global Institute analysis
In instances where incomes fell for highly educated workers, it was mostly due to a decline
in hourly wages. For less-educated workers, hours worked and hourly wages both fell. This
could imply that more highly educated workers have been taking jobs that did not require
their level of skills, while less-educated workers faced difficulties finding any type of work at
all. As we will see in Chapter 2, high youth unemployment in the economic downturn that
followed the 2008 financial crisis likely aggravated the income declines for young people,
especially in Europe.
31
In Europe, the share of single mothers in the population has stabilized, but in the United
States it continues to rise. A recent McKinsey study found that single motherhood is
prevalent across all 50 states and can often trap women in a cycle of low opportunitythey
drop out of school earlier and, with limited access to child care, have little chance to improve
their skills and raise their income.55
55
The power of parity: Advancing womens equality in the United States, McKinsey Global Institute, April 2016.
32
Exhibit 5
In our survey, 3040 percent of respondents said their incomes were not advancing
Share of respondents per category (self-identified)
Weighted % in category1
United Kingdom
N = 2,204
United States
N = 2,172
France
N = 2,006
31
Not advancing2
39
28
Neutral3
24
41
Advancing4
41
27
37
31
1 IN 3
respondents who
feel they are not
advancing expect
their children will
advance more
slowly in the future
than their
generation
The 30 to 40percent who felt they were not advancing held more pessimistic views about
their futures and the futures of their children than those who felt they were advancing
(Exhibit6). Nearly half of those not advancing expected not to advance in the future,
compared with just one-quarter of those who felt they were advancing. One in three of those
not advancing also expected their children to advance more slowly in the future than their
own generation. That is three times the proportion of those who felt they were advancing, or
those who were neutral.
In our survey, younger people tend to be more optimistic than older generations. More than
one-third of respondents younger than 35 agreed or strongly agreed with the statement, I
expect to advance significantly in the next five years. By contrast, only 11percent of people
35 and over agreed or strongly agreed with that assertion. In other words, 89percent of
members of these older generations are considerably more pessimistic about advancing
in the future. This may be due a broad expectation among young people entering the
workforce that their incomes will increase as their careers advance.
Surveys were conducted online in August 2015 in the United Kingdom and the United States and in
September 2015 in France, with representative samples of about 2,000 respondents in each country. For
more detail about the survey methodology, see the technical appendix.
56
33
Exhibit 6
Respondents who were not advancing were most pessimistic about their future and the future of their children
Views on own financial position
% who agree or strongly agree
36
31
22
11
I am advancing more
slowly than other
people with similar
experience and
qualifications
25
11
My economic
progress has
substantially
slowed in the last
5 years
I expect my financial
position to improve
in the next 5 years,
but more slowly than
that of my peers
10
I do not expect
to advance
significantly in
the next 5 years1
I expect my children/
the next generation
to advance more
slowly in the future
1 Percentage of people who strongly disagreed or disagreed with the positively framed statement, I expect to advance significantly in the next five years.
SOURCE: McKinsey Global Institute survey on income inequality (2015) results for United States, France, and United Kingdom, about 2,000 respondents per
country; McKinsey Global Institute analysis
57
58
34
income growth for the bottom 40percent of the population and its effects on the rest of the
economy.59
If market incomes continue to decline for large segments of the population, governments
could be under pressure to maintain disposable income growth by increasing transfers and
reducing taxes. However, countries that are still struggling with slow growth and high public
expenditure would have limited capacity to fund higher transfer payments or large tax cuts
particularly when aging places additional demands on public services. Stagnant or falling
wage incomes could reduce income tax receipts, exacerbating the considerable fiscal
challenges that many advanced economies already face, with government debt as a share
of GDP increasing steadily over the past seven years in five of the six countries we studied.
Even maintaining current levels of taxes and transfers could become more challenging.
For example, central government debt is close to 100percent of GDP or higher in Italy, the
United Kingdom, and the United States, where it rose from 56percent of GDP in 2005 to
97percent of GDP in 2016 (Exhibit7). Sweden is the outlier with a relatively flat debt share of
GDP between 2008 and 2015; while it increased debt levels to fund the effects of recession,
it started from a lower level, of less than 40percent in 2008, which gave it greater freedom
to spend during the crisis. In 2015, Swedens central government debt remained steady at
about 42percent of GDP.
Exhibit 7
Government debt rose sharply in all countries except Sweden during and after the 2008 financial crisis
Debt of central government
% of GDP
Great Recession
Axis midpoint
160
150
140
130
120
France
110
Italy
100
Netherlands
90
Sweden
80
70
United Kingdom
60
United States
50
40
30
1/1/1995
1/1/2000
1/1/2005
1/1/2010
1/1/2015
59
35
Stagnant incomes could also undermine the financial viability of existing government
programs. Many programs and public investments are predicated on the assumption of
rising incomes. Public pension systems, for instance, assume rising payments by current
workers to help support retirees.60 And economic assessments for major infrastructure
projects typically assume that future taxpayers will have higher incomes.61 If future
generations are, indeed, poorer than their parents, these long-standing assumptions
about funding public expenditures would need to be revised and assumptions about
intergenerational trade-offs would need to be updated.62
For example, the CBOs budget outlook relies on the assumption of steadily rising average real wages for
workers covered by social security in the United States. See Congressional Budget Office, 2015 long-term
budget outlook, June 2015.
61
For instance, a cost-benefit analysis for the ten-year, $5.6billion Dulles Metrorail project in the United
States, calculates the value of reduced travel times based on an assumption of rising hourly wages. See
Lauren Donnelly, Dulles Corridor Metrorail project. A cost-benefit analysis, Policy Perspectives, volume 16,
May 2009.
62
Anthony B. Atkinson, Inequality: What can be done? Harvard University Press, 2015.
63
See, for example, Alberto Alesina and Roberto Perotti, Income distribution, political instability, and
investment, European Economic Review, volume 40, number 6, 1996.
64
Rafael Di Tella, John Haisken-De New, and Robert MacCulloch, Happiness adaptation to income and to
status in an individual panel, Journal of Economic Behavior and Organization, volume 76, issue 3, 2010.
60
36
Exhibit 8
Respondents who were not advancing and not hopeful about the future (1015 percent of the sample) held strong
negative attitudes about immigration and free trade
Views on trade and immigration
% who agree or strongly agree
36
19
24
56
43
41
29
1 Respondents that agreed or strongly agreed with the statement, "My non-adult children/the next generation will advance more slowly in the future" were
counted as those who were not advancing and not hopeful about the future. All other respondents who were not advancing were counted in the group that
was not advancing but hopeful about the future.
SOURCE: McKinsey Global Institute survey on income inequality (2015) results for United States, France, and United Kingdom, about 2,000 respondents per
country; McKinsey Global Institute analysis
37
Exhibit 9
In most economies, the recession following the 2008 financial crisis was deeper and longer than previous
downturns
Employment and GDP recession and recovery periods over time
Downturn
start year
197374
197982
199596
200708
197677
198993
200103
2014
Employment
GDP
5
0
Italy
-10
-10
United
States
39
-5
-10
-10
5
28
-5
-5
-10
-10
5
24
-5
-5
-10
-10
53
-5
-5
-10
-10
27
-5
-5
-10
-10
40
-5
United
Kingdom
0
-5
Sweden
30
-5
Netherlands
34
28
19
21
13
65
66
41
Taxes and transfer factors. Taxes and government transfers determine the difference
between market income and disposable income. Transfers include a range of cash
payments to beneficiaries such as social security payments, disability or workers
compensation, and unemployment benefits.67 Taxes generally include central and local
government taxes.68
The first three of these five factors contribute to changes in labor income. Changes in
market income are driven by changes in this labor income, together with changes in capital
income. Once taxes and transfers are applied to market income, it becomes disposable
income. What role these factors played in the post-2005 flat or falling income phenomenon
that we have outlined above, and what role they could play in the future, will be considered
in the rest of this chapter. In Exhibit10, we show how each of the five factors contributed to
the development of median household income in the most recent 200514 period and in
19932005.
In-kind transfers such as the Supplemental Nutrition Assistance Program, Medicare, and Medicaid
are counted for the United States but not for the other five countries due to a lack of available data by
income segment.
68
In France, we include personal income taxes, housing taxes, and other social contributions. In Italy, we
include all income taxes, municipal taxes (such as waste and water tax), and corporate income taxes. In
the Netherlands, we include income taxes (and wealth tax until 2001) and dividend taxes. In Sweden, we
include federal, local, and municipal taxes such as property taxes. In the United Kingdom, we do not include
value-added taxes (VAT) in order to make our UK data comparable to the other European countries and the
United States. VAT data are not available in any of the other European countries. In the United States we use
CBO data that include only federal taxes until 2013. In our analysis that extends this time period to 2014 using
Current Employment Statistics data, we also include state and local taxes (which account for about 6percent
of market income for the middle quintile of households).
69
The end date of our period varies slightly from country to country in our analysis, depending on
data availability.
67
42
Exhibit 10
Five factors determine changes in disposable income
Change in disposable income for middle-income households,
19932005 and 2005141
%
Disposable
income,
start year
France
Italy
Netherlands
Sweden
United
Kingdom
United
States
130
120
110
100
90
80
130
120
110
100
90
80
130
120
110
100
90
80
130
120
110
100
90
80
130
120
110
100
90
80
130
120
110
100
90
80
Market
income
change
+6
119
-1
118
-3
100
100
-5
-2
-4
+4
+6
+18
Capital
income
factors5
+16
+29
200514
Labor
income
change
Demographic
factors3
-2
Labormarket
factors4
Aggregate
demand
factors2
+13
19932005
106
+3
+4
97
+15
+1
+3
-1
+19
+4
-1
114
+3
-12
111
-15
-1
108
-2
93
-2
105
+10
98
95
-5
107
86
+2
88
+3
121
121
-1
120
+10
112
113
+5
117
+5
121
+4
125
-1
124
-8
94
+1
95
+4
100
-9
109
+2
111
+4
115
-5
+7
109
-1
-6
-5
-5
Disposable
income,
end year
103
-6
Tax and
transfer
factors6
+5
+1
-7
96
96
101
1 Middle-income, or median, households are households in the middle (3rd) quintile or the 5th and 6th decile or the 40th to the 59th percentile. For each
country we use the latest data availableFrance (2012), Italy (2012), the Netherlands (2014), Sweden (2013), United Kingdom (2014), and United States
(2013). The base year for France is 1996 and for Sweden is 1995. All growth numbers are standardized to make results comparable.
2 Change in aggregate output, measured by output per employed worker, multiplied by change in number of employed workers in the working-age population.
3 Change in number of working-age people per household.
4 Change in wage share of GDP, adjusted for difference between consumer price inflation and inflation of overall output, and median household share of
wages.
5 Includes profit from own business, income from capital, and other sources of market incomes that cannot be classified as income from labor.
6 Includes income from private and public pension transfers, other transfers such as social security benefits, and taxes on labor income and capital income.
NOTE: Numbers may not sum due to rounding.
SOURCE: INSEE; Bank of Italy; CBS; Statistics Sweden; ONS; CBO; McKinsey Global Institute analysis
43
Exhibit 11
Flat or falling incomes spread during the recession and the weak recovery
Periods without flat or falling income
198396
France
19962003
1.7
200307
1.7
200710
-0.8
22
89
0.6
201012
198393
2.3
1.4
19932005
Italy
200508
200810
100
0.2
70
-2.4
201012
100
-1.5
198393
2.3
19932005
Netherlands
3.4
200507
200710
100
-0.7
100
-0.3
201013
198395
2.9
19952005
Sweden
3.4
200507
-0.8
200710
0.4
201013
2.3
198393
United
Kingdom
19932005
2.7
2.0
200507
200710
100
0.6
198393
2.4
19932005
2.3
1.3
200507
200710
201013
60
-1.1
1.2
SOURCE: INSEE; Bank of Italy; CBS; Statistics Sweden; ONS; CBO; OECD; McKinsey Global Institute analysis
44
30
60
-1.7
201013
United
States
10
61
See An economy that works: Job creation and Americas future, McKinsey Global Institute, June 2011.
GDP per capita increase calculated by comparing Q4 to Q4.
72
OECD labor database, harmonized unemployment measured quarterly.
70
71
45
Exhibit 12
Income growth followed GDP down, but did not necessarily recover when GDP recovered
GDP per capita, median market income, median disposable income (Index: 100 = 1993)
Compound annual growth rate, 19932005, 200514 (%)
France1
Italy1
160
160
140
140
120
120
100
100
80
1993
2005
20143
80
1993
2005
20143
1.9
0.3
1.6
-1.0
3.2
0.0
1.3
-0.8
1.3
0.5
0.3
-1.8
Netherlands
Sweden2
160
160
140
140
120
120
100
100
80
1993
20143
2005
80
1993
2005
20143
2.6
0.6
3.5
0.7
0.6
-1.1
2.2
2.4
0.7
-0.1
1.9
2.8
United Kingdom
United States
160
160
140
140
120
120
100
100
80
1993
2005
20143
80
1993
20143
2005
3.1
0.0
2.6
0.2
4.0
-1.4
1.4
-0.5
3.8
-0.3
2.0
0.1
1 Market incomes are calculated only for the beginning and end of periods; graphs show a smoothed curve across these three data points.
2 1993 is estimated by taking average of 1991 and 1995 data; all years are available after 1995.
3 Latest data available ranges from 2012 to 2014.
SOURCE: INSEE; Bank of Italy; CBS; Statistics Sweden; ONS; CBO; OECD; US Bureau of Labor Statistics; McKinsey Global Institute analysis
46
40%
Youth
unemployment
in Italy in 2015
Unemployment has been a significant factor for young people, whose incomes have been
particularly affected, as we saw in the previous chapter. This is especially the case in
Europe, where the unemployment rate among workers aged 15 to 24 rose to 20percent in
2015, more than twice the overall unemployment rate (9.4percent), and sharply above the
16percent rate in 2007. Youth unemployment soared to 40percent in Italy in 2015, up from
21percent in 2008, and stood at nearly 25percent in France and 20percent in Sweden,
according to OECD data.
Differences in labor markets and education systems may explain much of the variation in
youth unemployment among countries. Higher employment protection can lead to low
turnover of older workers and higher youth unemployment. For example, in Italy, income
support programs for permanent workers meant that employers hired fewer young
workers in the recession following the 2008 financial crisis, an effect compounded by
a 2012 pension reform that kept older workers in the workforce.73 However, a review of
OECD countries found no systemic effect of increase in overall unemployment from higher
employment protection.74 The impact of employment protection is likely compounded
by other institutional factors that contribute to labor-market rigidity. For instance, youth
unemployment is also higher where the education system is disconnected from the labor
market and produces too many university graduates qualified for the public sector and too
few skilled tradespeople.75 Where educators work with employers, as in Germany, youth
unemployment is considerably lower than European Union averages. We discuss some of
these issues later in this chapter and in Chapter 3.
Antoine Bozio et al., European public finances and the Great Recession: France, Germany, Ireland, Italy,
Spain, and the United Kingdom compared, Fiscal Studies, volume 36, number 4, December 2015.
74
Prabirjit Sarkar, Does Employment protection lead to unemployment? MRPA, 2011.
75
Juan J Dolado, ed., No country for young people? Youth labour market problems in Europe, CEPR
Press, 2015.
76
Families are changing, in Doing better for families, OECD, 2011.
77
For a detailed discussion of this concept, see OECD framework for statistics on the distribution of household
income, consumption and wealth, OECD, June 2013.
73
47
Exhibit 13
The working-age population per household decreased in middle-income households between 2002 and 2012
in France, Italy, and the United States
Number of individuals per household and age
for each decile
2002
2012
France
Italy
United States1
2.5
2.5
2.5
2.0
2.0
2.0
1.5
1.5
1.5
1.0
1.0
1.0
0.5
0.5
0.5
1 2 3 4 5
6 7 8 9 10
1 2 3 4 5 6 7 8 9 10
1 2 3 4 5
6 7 8 9 10
Decile
1 United States data between 2003 and 2013.
SOURCE: National government microdata on 8,000 households in Italy, 72,000 households in France, and 75,000 households in the United States; McKinsey
Global Institute analysis
78
79
48
Global growth: Can productivity save the day in an aging world? McKinsey Global Institute, January 2015.
Working-age population consists of people between the ages of 16 and 64.
Factors such as financialization of the economy and the deteriorating power of unions have been cited as
leading factors in wage stagnation as well. See, for example, Global wage report 2012/13, ILO, December
2012; OECD economic outlook 2012 volume 1, OECD, June 2012; Andreas Hornstein, Per Krusell, and
Giovanni L. Violante, The effects of technical change on labor market inequalities, Center for Economic Policy
Studies working paper number 113, July 2005.
81
In the United Kingdom, unfunded liabilities in defined-benefit pension schemes are creating downward
pressure on wages to workers. See Conor DArcy and Gavin Kelly, Securing a pay rise: The path back to
shared wage growth, Resolution Foundation, March 2015, and Brian Bell, Wage stagnation and the legacy
costs of employment, Centre for Economic Performance, London School of Economics, paper number
CEPCP 458, November 2015.
82
Loukas Karabarbounis and Brent Neiman, The global decline of the labor share, NBER working paper number
19136, June 2013, and Loukas Karabarbounis and Brent Neiman, Declining labor shares and the global rise
of corporate saving, NBER working paper number 18154, June 2012.
83
This finding is in line with a central thesis of Thomas Pikettys research, which attributes rising income
inequality to rising returns on capital. See Thomas Piketty, Capital in the twenty-first century, Belknap
Press, 2014.
80
49
decreased labors bargaining power. Finally, rising capital incomes from owner-occupied
housing also contributed to wage share decline.84
Exhibit 14
The wage share of national incomes has declined over the past 44 years in every country
except France and Sweden
Wages as a share of total GDP1
Index: 100 = 1970
Weighted average2
France
Italy
120
120
110
+7
100
-5
90
80
110
100
90
-5
-8
80
70
1970
2014
70
1970
Netherlands
Sweden
120
120
110
110
100
-5
90
-9
80
100
90
2014
+1
-5
80
70
1970
2014
70
1970
United Kingdom
United States
120
120
110
110
100
-5
90
-13
100
90
2014
-5
-8
80
80
70
1970
2014
70
1970
2014
1 Measured as full compensation to employees. (This is slightly different from our model, which uses only wages and salaries for data availability reasons). Full
compensation includes wages, salaries, and social insurance contributions payable by employers (e.g., social security) and does not include other employerpayable labor costs such as training, transportation, and employment or payroll tax. We use GDP as a proxy for GDI.
2 The absolute change in population-weighted average employee compensation share of GDP during 19702014 was from 53% to 50%; France, 50% to 53%;
Italy, 43% to 40%; Netherlands, 54% to 50%; Sweden 48% to 47%; United Kingdom, 56% to 49%; and United States, 58% to 53%.
SOURCE: OECD; McKinsey Global Institute analysis
84
50
How CBO projects income, Congressional Budget Office, July 2013. For more detail, see the
technical appendix.
The wage share decline began in manufacturing and has since spread to other business
sectors. In the United States, a study by the Congressional Budget Office found that
wage share was relatively constant from 1950 to 2000, averaging 62.4percent of GDP. In
those years, the rise of the non-profit sector (such as hospitals) compensated for the fall
in wage share in manufacturing; in education, health care, and other parts of the public
sector, output is usually measured as being equivalent to wages, and thus raises the
average wage share of GDP. However, the situation changed sharply after 2000, when
wage share began to fall across the entire incorporated business sector that includes both
manufacturing and non-manufacturing industries. In incorporated businesses, the wage
share fell from 66.1percent in 2001 to 59.3percent in 2011. This decline was not offset by
either household, public, and non-profit sectors, or by non-corporate businesses such as
proprietorships and partnerships, which tend to have a higher wage share.85
A shift in demand toward high-skill labor has affected the income growth of
low- and middle-income groups
While the recession and recovery touched most households up and down the income
distribution, labor-market effects have been very highly varied. In general the top quintiles
have fared better than lower income groups; they have held ground or gained a higher share
of the total wage pool, except in Italy and Sweden (Exhibit15).
This development is a reflection of shifting market demand for workers depending on
their skill levels, which in turn is feeding through to the income of the groups we analyze.
According to OECD research, wage growth has been more limited in low-skill occupations
construction, non-finance services, and low-tech manufacturing, for examplethan in highskill industries such as finance and high- and medium-tech manufacturing.86
97%
Wage premium in
the United States
of a college
graduate over a
high school
graduate
In the past two decades, there has been a clear pattern of consistent job growth for highskill workers and little or no growth for low- and medium-skill workers. In 1981, collegeeducated workers in the United States earned a 48percent wage premium over high school
graduates. By 2005, that premium had risen to 97percentin other words, an American
college graduate earns almost twice as much as a high school graduate.87 In Sweden
and the United Kingdom, where overall employment now surpasses pre-recession levels,
employment for low- and medium-skill workers remains below 2007 levels. In France, Italy,
and the Netherlands, employment rates for middle- and low-skill workers are lower than
employment for high-skill workers, and they declined rapidly from 2007 to 2014.
The level of educational attainment is reflected in income groups. In Italy, in the top quintile,
30percent of workers have college degrees or advanced degrees. In the third (middle)
quintile, only 6percent of workers have four-year degrees; 28percent have high school
diplomas, some college, or two-year degrees. These relatively low skill levels help explain
the flat or falling incomes of low- and middle-income households.
Ibid.
The labour share in G20 economies, ILO and OECD, February 2015.
87
David Autor, Skills, education, and the rise of earnings inequality among the other 99percent, Science,
volume 344, issue 6186, May 2014.
85
86
51
Exhibit 15
Since 1993, upper-income households have received a growing share of total wages in most countries
Change in share of total wage pool to each quintile
%
19932005
Quintile
Average1
200514
Bottom
2nd
3rd
4th
Top
Bottom
2nd
3rd
4th
-1
1
0
-1
France
0
Italy
1
0
0
0
Top
-1
-1
Netherlands
-1
0
-2
0
-1
-2
Sweden
United
Kingdom
1
0
-1
-1
2
1
-1
-1
-2
-1
-1
-1
1
0
-1
-2
0
-1
United
States
0
0
0
-1
-1
-1
-1
-1
1 Population-weighted average for advanced economies. For each country we use the latest data available France (2012), Italy (2012), the Netherlands
(2014), Sweden (2013), United Kingdom (2014), and United States (2013).
SOURCE: INSEE; Bank of Italy; CBS; Statistics Sweden; ONS; CBO; McKinsey Global Institute analysis
52
40%
Contribution of
immigration to
labor-force growth
in advanced
economies
The rise of digital technologies has also made it possible for companies to grow with fewer
workers overall and with a smaller number of low- and medium-skill workers, in particular
(see sidebar, How technology affects labor demand). Technology not only automates
tasks of low- and medium-skill workers (in both manufacturing and service industries),
but it also tends to raise demand for high-skill talent. In the United States, 4.8million new
jobs were created in the 2000s in occupations requiring high levels of interaction with
other people and independent problem solving (such as doctors, lawyers, and corrections
officers), compared to a decrease of 3.4million jobs that were transaction-oriented (such
as cashiers and accountants) and production-oriented (such as food preparation and
manufacturing).90
For a further description of the emergence of a global labor force, see The world at work: Jobs, pay and skills
for 3.5billion people, McKinsey Global Institute, June 2012.
89
This was the case in Miami in 1980, for example, when the city experienced a significant if short-lived
unemployment increase and wage decrease following the migration of 100,000 Cubans, the equivalent of
8percent of Miamis workforce. However, the effect proved to be only a short-term one. See David Card, The
impact of the Mariel boatlift on the Miami labor market, Industrial and Labor Relations Review, volume 43,
number 2, 1990; George J. Borjas, The wage impact of the Mariellos: A reappraisal, NBER working paper
number 21588, September 2015; and Christian Dustmann, Tommaso Frattini, and Ian Preston, The effect of
immigration along the distribution of wages, Review of Economic Studies, volume 80, issue 1, January 2013.
90
For more details, see An economy that works: Job creation and Americas future, McKinsey Global Institute,
June 2011.
91
OECD labor database. Statistics use a non-weighted average and common definitions of part-time work, and
they exclude self-employment.
92
OECD employment outlook 2015, OECD, July 2015.
93
Ibid.
88
53
Michael Chui, James Manyika, and Mehdi Miremadi, Four fundamentals of workplace
automation, McKinsey Quarterly, November 2015.
2
Growth and renewal in the United States: Retooling Americas economic engine, McKinsey
Global Institute, February 2011.
3
Disruptive technologies: Advances that will transform life, business, and the global economy,
McKinsey Global Institute, May 2013.
4
Annual revenue per employee, comparing global revenue and employees in 2015 from
Hoovers. The top five legacy US retailers based on Kantar Retail IQ 2015 are Costco, Home
Depot, Kroger, Target, and Walmart.
5
US Bureau of Labor Statistics; Digital America: A tale of the haves and have-mores,
McKinsey Global Institute, December 2015.
6
Playing to win: The new global competition for corporate profits, McKinsey Global Institute,
September 2015.
1
54
Exhibit 16
Employment has been lower for low- and medium-skill workers, and they are more likely to be employed on
temporary contracts
United
Kingdom
Italy
France
High skill
Netherlands
Sweden
Great Recession
Axis midpoint
Medium skill
Low skill
Employment rate
%
90
90
90
80
80
80
70
70
70
60
60
60
50
50
50
40
1994
2004
2014
40
1994
2004
2014
40
1994
2004
2014
1994
2004
2014
Temporary employment
Thousand people
30
30
30
25
25
25
20
20
20
15
15
15
10
10
10
1994
2004
2014
1994
2004
2014
Flexible work arrangements have benefits for many workers. For example, they have
helped raise labor participation among individuals whose domestic responsibilities make
full-time employment impractical. In the Netherlands, for example, 61percent of women
worked part-time in 2014, compared with 20percent of men.94 A small but rapidly growing
number of workers is also actively seeking contingent work on online labor platforms,
including TaskRabbit and Upwork. Even if these platforms touch only a fraction of the global
workforce, they can generate significant benefits for both individuals and economies.95
However, temporary work and part-time arrangements can also leave workers with shorter
hours and lower incomes. For some workers, part-time employment is a stopgap measure;
the average share of workers in our six sample countries who are working part-time
involuntarily (that is, they sought full-time employment but accepted part-time work) doubled
94
95
55
from 3percent of the labor force in 1993 to 6.4percent in 2014.96 The share of temporary
work increased between one and three percentage points in each country except Italy.
There, the share of workers on temporary contracts jumped by nine points, from 1percent
of the labor force in 1993 to 10percent in 2014. And while even highly paid professionals are
engaged on a part-time or temporary basis, low- and medium-skill workers make up the
largest share of contingent employees.
Differences in union roles and labor regulation influenced outcomes for some
income segments
The differences in how labor-market shifts affected our focus economies are likely the result
of national labor-market institutions and practices as well as prevailing economic models.
The share of workers in OECD nations represented by a union fell from 35percent in 1974 to
17percent in 2014 (Exhibit17). Research has found that countries where more workers are
represented by unions tend to have lower wage inequality. However, union density by itself
is not a reliable indicator; in some countries, such as France, unions negotiate at a national
level on behalf of their members and all other workers in a given industry, so even where
membership has fallen, unions can retain strong influence on wages and work rules.97
Exhibit 17
Union membership has declined in all countries but Italy, and remains exceptionally high in Sweden
Union members, share of total employed
%
Recession
and slow
recovery
90
80
70
France1
60
Italy1
50
Netherlands1
40
Sweden
30
United Kingdom
United States
20
OECD countries
10
0
1964
1974
1984
1994
2004
2014
Sweden and the United States represent contrasts in how market incomes of middleincome segments were affected in the past decade. In Sweden, a country with a long
tradition of social democracy where 68percent of workers are represented by unions, the
median household received a greater share of output that went to wagesand received
OECD labor database.
Florence Jaumotte and Carolina Osorio Buitron, Inequality and labor market institutions, IMF staff
discussion note number 15/14, July 2015; Global wage report 2014/15: Wages and income inequality, ILO,
December 2014.
96
97
56
more of the gains from output growth than households in Swedens top and bottom income
deciles. This reflects Swedish labor policies such as contracts that protect wage rates and
hours worked. After the global financial crisis, the government worked with unions to forge
agreements for temporary reductions in work hours, which preserved long-term jobs and
helped private-sector employers withstand the downturn.
In the United States, labor institutions are very different and generally place fewer
restrictions on employers. There are fewer rules than in Europe limiting dismissals, and just
11percent of private-sector workers are unionized, compared with 30percent, on average,
in Europe. Rather than accepting lower productivity as a consequence of falling demand
(or using temporary layoffs to reduce costs), as had been the practice during most postwar
recessions, US employers took steps to maintain productivity when the 2008 financial
crisis hit, cutting jobs that were slow to return. (US productivity growth nonetheless fell
back below the pre-recession rate.) This helped US companies restore profits rapidly in
the recovery, despite weak demand growth.98 Union membership is higher in the United
Kingdom and the Netherlands than in the United States, though it has fallen steeply since
the early 1990s. In these countries, labor-market shifts had even greater negative effects on
median incomes than they had in the United States.
14%
Contribution of
capital income to
disposable income
of median income
households in
2014
For our analysis, however, capital income is not a major factor, as the shift between 2005
and 2014 was very small. As a percentage of disposable income, for example, the share
of capital income remained unchanged on average in our six focus countries for the lowand middle-income quintiles in 200514. Looking back further in time, capital income also
shifted only modestly for most income groups between 1995 and 2005, dropping slightly for
the bottom four quintiles.
In fact, the largest movement in capital income as a share of disposable income was actually
felt by high-income households in the top quintile. For them, capital income rose sharply in
weight from 1995 to 2005, rising from 28percent of disposable income to 35percent, and
then falling back slightly from 2005, to 33percent in 2014.
An economy that works: Job creation and Americas future, McKinsey Global Institute, June 2011.
See Thomas Piketty, Capital in the twenty-first century, Belknap Press, 2014.
98
99
57
The data do not capture the value of in-kind payments, such as medical care or food stamps, except in the
United States.
100
58
Exhibit 18
Inequality as measured by market Ginis has increased in most advanced economies, but transfer payments have
limited the impact on net Gini (based on disposable income)
Market Gini and net Gini coefficients (index 0100), and difference between market and net Gini, 19932012
Market Gini coefficient
Italy
United Kingdom
55
53
50
50
50
49
48
53
France
Netherlands
Sweden
53
50
49
47
47
48
48
47
46
46
47
46
45
40
37
35
37
35
34
34
33
34
35
35
31
30
29
28
26
27
26
25
24
24
22
20
25
25
23
20
19
16
15
10
24
13
13
18
18
18
20
21
19
19
20
16
13
12
1993 2005 2012 1993 2005 2012 1993 2005 2012 1993 2005 2012 1993 2005 2012 1993 2005 2012
NOTE: We use data from the Standardized World Income Inequality database, which provides market and net Gini data for 174 countries for the periods we
examine in this report.
SOURCE: Frederick Solt, The standardized world income inequality database, working paper, SWIID version 5.0, October 2014; McKinsey Global Institute
analysis
59
Exhibit 19
Transfer payments had a greater impact on disposable incomes for the middle-income households
in our focus countries1
Effect of changes in taxes and transfers on disposable incomes from 2005 to 2014
%
Impact of transfers
Quintile
Bottom
France
2nd
Impact of taxation
3rd
4th
Top
Bottom
2nd
3rd
4th
-3
-2
-2
Top
10
7
5
6
4
-1
Netherlands
-5
9
5
3
0
0
-6
-7
Sweden
7
5
5
3
0
-2
United
Kingdom
-7
United
States
8
4
3
1
0
-1
1 Data for Italy do not separate the effects of taxes from transfers and hence we exclude Italy in this analysis.
SOURCE: INSEE; CBS; Statistics Sweden; ONS; CBO; McKinsey Global Institute analysis
60
Public pension payments formed the largest share of increased transfers in our focus
economies. In France, public pension payments accounted for about 85percent of the
increase in transfers between 2005 and 2012, while in the United Kingdom, they accounted
for about 45percent of the rise. Even in Sweden, where a drop in unemployment payments
reduced overall transfers, pension payments rose.
Patrick Aubert and Marion Bachelet, Disparits de montant de pension et redistribution dans le systme de
retraite franais, Lconomie franaise, INSEE, 2012.
102
Peter Haan, Daniel Kemptner, and Victoria Prowse, Inequality and defined benefit pensions when life
expectancy is heterogeneous, DIW Berlin, February 2015.
103
George Kudrna, Chung Tran, and Alan Woodland, The dynamic fiscal effects of demographic shift: The case
of Australia, Economic Modelling, volume 50, November 2015.
101
61
Swedish and US policy responses varied, but both countries boosted disposable
income for median households
The United States and Sweden represent contrasting policy approaches and responses to the
recession that followed the financial crisis.
As we have seen, market incomes in Sweden in 200514 rose for middle-income households,
whereas in the United States they fell for the same median income segments. In both countries,
however, the outcome for median households was a boost in disposable income.
In Sweden, which went through a deep four-year recession and banking crisis in the 1990s,
GDP shrank by 5percent in 2009, but it quickly bounced back, growing by 6percent in 2010, or
twice the US rate that year. The Swedish government focused on job preservation and creation,
adding temporary jobs to the public sector, reducing payroll taxes for businesses, and providing
tax incentives to hire young people and the long-term unemployed.1 The US approach was more
geared toward stabilizing sectors such as banking and autos through bailouts and stimulating
demand in the economy.
In both countries, tax cuts and stimulus spending including transfers to households led to
higher disposable income. Sweden expanded eligibility for unemployment benefits and raised
payments, while reducing income taxes. In the United States, tax cuts and unemployment
benefits more broadly reversed the decline in market incomes for median households,
transforming that drop into gains in disposable incomes.
In Sweden, only 20percent of the population was in deciles with flat or falling market incomes,
and median household incomes grew from 2005 to 2014. The median disposable income in
Sweden rose by 17percent from 2005 to 2014.
In the United States, market incomes were flat or falling for 81 percent of income segments in the
past decade, but the effect of more generous tax and transfer policies maintained disposable
income across deciles. And in fact, as a result of tax and transfer policiesparticularly since
2000median US disposable income has risen sharply, even as market income has dropped.
In Swedens case, the response was influenced by the previous crisis in the 1990s. In its
aftermath, the government recapitalized banks, cut taxes on capital gains to encourage
entrepreneurship, and scaled back some of its generous welfare programsreducing sick pay
and increasing co-pays for prescription drugs, for example.2
What lessons from the Swedish experience could apply to other economies? One of the most
important is to use recoveries and expansions to repair national balance sheets to enable the
spending to get out of the next crisis. After its financial crisis, Sweden set a cap on government
spending and set an official goal of generating a 1percent surplus every yearcompared
with a deficit of 7percent of GDP in 1995. By the time the 200809 recession arrived, central
government debt was only 36percent of GDP (compared with 40 to 100percent in the other five
countries we analyze in depth). This gave Sweden the wherewithal to fund tax cuts and other
stimulus efforts in the recovery.
Swedish reforms from the 1990s also helped raise labor participation among older workers and
women, which increased productivity and contributed to GDP. After the 1990s crisis, Sweden
switched its national pension program to a defined-contribution model. This has helped raise
labor participation among 55- to 64-year-old workers aged to 79percent, compared with
64percent in the United Kingdom and the United States. Sweden, which declared itself the
worlds first feminist government in 2014, has adopted a series of policies to encourage female
labor participation, including 90 days of parental leave per parent. The result is that 79percent
of Swedish women are in the labor force, compared with 72percent in the United Kingdom and
67percent in the United States.
Dominique Anxo and Thomas Ericson, Labor market measures in Sweden 200813: The crisis and beyond, ILO,
February 2016.
2
Growth and renewal in the Swedish economy, McKinsey Global Institute, May 2012.
1
62
Second, the lowest income groups were not always the segment to bear the brunt of flat
or falling incomes; in all of our focus countries except Sweden, middle-income segments
also felt the impact, as a result of declining income from labor. Higher income segments
also experienced a decline in market income in Italy, the United Kingdom, and the United
States as a result of lower income from capital, which was especially volatile during and after
the 2008 financial crisis. In the United States, higher capital income increased disposable
income growth for the top quartile by 24percent in 19932005 but pushed it down by
6percent in 200514.
Looking at some countries individually, Sweden stands out as the only one where market
incomes rose for middle-income households, Sweden had gone through a previous steep
downturn in the 1990s, and applied some lessons learned during that crisis to the post2008 recession. Among other measures, it focused on job preservation and creation.
In the United Kingdom, the pattern of disposable income in from 1993 to 2005 highlights
the outcome of the redistributive policies of the government of Tony Blair, with sharp income
increases for the lowest quintiles. The British economy is highly reliant on revenue from
the financial sector to balance its budgets, and after the financial crisis, the government
imposed a period of austerity when the financial revenue fell. More than four-fifths of the
fiscal measures associated with austerity were spending cuts that disproportionately
affected working-age people (cuts to benefits and public-sector jobs, for example), although
pensions were protected from the cuts. In our data, this decrease in spending seems to
have affected the bottom quintile the most from 2005 to 2014. Disposable income growth
declined by six percentage points.104
63
for market incomes. Italy has an above-average tax rate, and it raised taxes as part of its
economic and financial policies during the Eurozone debt crisis in 201014. Unlike in some
other countries, including the United States, taxes and transfers thus had a negative impact
on the bottom three quintiles of the income distribution; disposable incomes for households
in these low- and middle-income segments were below market incomes.
Exhibit 20
Depending on the pace of GDP growth and automation adoption, as little as 10 percent of income groups and as
many as 80 percent might not advance through 2025
201225
sensitivity
Low growth
High growth
Labor-market
dynamics
1520
05
510
The pace of GDP growth in advanced economies is one of the most variable of our five
factors and while it will not be the sole determinant of future income growth, it will be a major
one. As we have seen, the post-2008 global recession and slow recovery that followed
had a significant impact on incomes, by substantially reducing the aggregate demand
component of income growth compared with buoyant growth in the 19932005 era.
According to our analysis, aggregate demand factors nonetheless had a positive effect on
incomes even in the 200514 period in five of our six focus countries, with Italy being the
sole exception.
We model these scenarios for three countriesFrance, Italy, and the United Stateswhere we had the
microdata to estimate employment and wage outcomes for each income decile. The consolidated results
are based on a simple average of these three economies, which we use as a proxy for outcomes across
advanced economies.
107
64
The demographic factorsthat is to say, the decline in household size and a drop in the
number of working-age adults per householdare long-run trends resulting from lower
fertility rates, increased longevity, and changes in family structures, with more single-parent
families. The labor-market factorsthe wage share and its uneven distribution among
different income segmentswill likely continue to be affected by a range of developments.
Growing automation in the workplace could further reduce the need for low- and mediumskill workers, even as it increases the demand for high-skill ones. At the same time, digital
technology platforms such as LinkedIn and Monster could help overcome mismatches
between workers and jobs, while firms such as Uber or TaskRabbit could provide new
opportunities for freelance employment.108
Changes in capital income have not been a significant factor affecting middle-income
households over the past two decades, but significant shifts in global capital markets could
play a role in the future. In particular, after a period of exceptional increases in 19852014,
US and Western European stock and bond returns could fall back, and this would affect
both public and private pensions.109 Taxes and transfers will continue to influence disposable
income, at a time when many governments sovereign debt has risen to historic levels and
they have not yet begun the process of deleveraging.
Our sensitivity models are based on different assumptions about economic growth, wage
share, and labor demand. We use growth projections and labor-market data for France,
Italy, and the United States, which we then extrapolate to our universe of 25 advanced
economies based on the outcomes for the three economies.
The sensitivity analyses we conducted are not fully fledged scenarios, nor are they
general equilibrium models. Rather, they represent a hypothetical closed system where
we isolate the effect of changes in productivity and employment on market incomes of
populationpercentiles but do not adjust other factors that, in reality, would also change.
For example, if productivity were to rise to the extent we describe in our high-growth
and labor disruption scenarios, other economic factors including consumption could
also grow, leading to more employment.110 Our scenarios do not include any potentially
positive links between increased innovation and social mobility.111 To adjust for inflation, we
apply a constant deflator linked to the consumer price index for all income segments. In
reality, the consumption basket is different for each income segment and could warrant a
different deflator.
A continuation of current low GDP growth could further increase the proportion
of flat or falling households
For our first, low-growth sensitivity analysis, we assume that the slow average growth in
productivity, employment growth, and the markedly higher rate of decline in wage share
that marked the 200212 decade continue throughout the next decade, and that the
demographic effects continue as before. This hypothesis is broadly consistent with the
view of economists such as former US treasury secretary Lawrence Summers who are
predicting many years of secular stagnation due to weak demand growth and low levels of
investment.112 Other factors such as the effects of Chinas decelerating growth could also
For more details, see A labor market that works: Connecting talent with opportunity in the digital age,
McKinsey Global Institute, June 2015.
109
See Diminishing returns: Why investors may need to lower their expectations, McKinsey Global Institute,
May 2016.
110
See the technical appendix for details of how we constructed our three scenarios.
111
Philippe Aghion et al., Innovation and top income inequality, CEPR discussion paper number 10659,
June 2015.
112
The secular stagnation hypothesis, which holds that an oversupply of savings and a lack of investment can
reduce growth, inflation, and the natural equilibrium interest rate, dates back to the 1930s and has gained
renewed attention recently. See Lawrence H. Summers, The age of secular stagnation: What it is and what to
do about it, Foreign Affairs, March/April 2016.
108
65
result in continuing slow growth. The slow growth of advanced economies in recent years
about 0.5percent annual growth in GDP per capita in Europe and 1.5percent in the United
States since 2009would become a new normal.
In this low-growth environment, an even larger proportion of income groups in advanced
economiesfrom 70 to 80percentcould experience flat or falling real market incomes in
the next decade to 2025 than did during the 200512 period. Net transfers would need to
be 15 to 20percent higher, on average, to avoid losses in disposable incomea burden that
would be difficult for many governments to contemplate.
This finding underscores the continuing impact of demographic and labor-market factors.
Based on previous MGI research, we can expect that aging will continue to be the most
important demographic force affecting advanced economies and that labor-market effects
including the rising use of contingent labor and declining need for low-skill labor will continue
to influence household income.
If governments choose to preserve or lift disposable incomes, transfers may need to rise
sharply (as apercent of current transfers). In the United States, for instance, they would
need to rise by 20 to 30percent of the 2012 level. Such increases in transfers would be
challenging today, due to the fiscal constraints imposed by high levels of government debt;
another decade of slow growth would exacerbate those fiscal challenges.
10-20%
Proportion of
income groups that
could have flat or
falling incomes
even in the event of
high GDP growth
113
66
We also looked at a third hypothesis in which GDP growth is similarly high but is
accompanied by labor disruption, as increasingly powerful digital technologies take on
many activities now requiring workers, further reducing demand for low- and medium-skill
workers. As noted, previous MGI research has found that technological innovation in the
past has created more jobs than it destroyed.114 However, the spread of digitization, which
increases the automation potential of many sectors of the economy, has also prompted
forecasts that this historic link between productivity growth and employment growth could
change.115 MGI has estimated that automation could accelerate displacement of mediumskill jobs to nearly twice the rate of recent decades, with as much as 15percent of such jobs
being affected.116 For 60percent of existing US jobs, MGI has estimated that 30percent
or more of current work activities could potentially be automated by adapting currently
available technologies, representing $2trillion annually in wages in the United States
alone.117
If advances in technology were to have such a significant impact on the workforce,
unemployment for the middle- and low-skill segments would rise and the number of
employed workers per household would fall faster than in the past decade. The share of
national income going to wages would decline further, as even more output would come
from machines and information technology. Rising productivity would translate into rising
wages across the economy in absolute terms, but low- and middle-income households may
not benefit unless they are nimble in adjusting to the new realities of the labor market.
According to this analysis, 30 to 40percent of the population might be in income groups
whose real market incomes in 2025 are flat or down compared with 2015. To sustain
disposable incomes, net transfers would need to rise by 5 to 10percent. Even this would not
result in growth in disposable incomes, but merely arrest the decline.
It should be noted that this hypothesis regarding labor disruption does not fully model
the normal behavior of economies. In reality, the wealth and investment created by rising
productivity would create new types of demand, which would lead to jobs that do not exist
today. This has been the pattern when new technologies have disrupted labor markets in
the past: rising output leads to more profits, which enables new investment, leading to new
employment and more demand. However, this sensitivity analysis serves to illustrate the
extent to which rapid adoption of technological advances can affect income inequality if the
advances outpace the rate at which workers acquire new skills.
The recession and sluggish recovery after the 2008 financial crisis were primary causes
of the flat or falling income trend we have detailed from 2005, but demographic and
labor-market factors also played a role and will continue to do so even when the global
economys scars from the financial crisis eventually heal. Even in the face of a very steep
downturn, policy can make a substantial difference. In both Sweden and the United
States, the disposable income of households in median income groups was cushioned by
taxes, transfers, and, in Swedens case, muscular intervention in the labor market. How
could policy more generally reduce income inequality? In our final chapter, we look at a
range of optionsfor governments and also for businessthat could limit or reduce the
phenomenon of non-advancement we have described in this report.
Growth and renewal in the United States: Retooling Americas economic engine, McKinsey Global Institute,
February 2011.
115
Erik Brynjolfsson and Andrew McAfee, Race against the machine, Digital Frontier Press, 2011. For an
assessment of automation potential, see Michael Chui, James Manyika, and Mehdi Miremadi, Four
fundamentals of workplace automation, McKinsey Quarterly, November 2015.
116
Digital America: A tale of the haves and the have-mores, McKinsey Global Institute, December 2015.
117
Michael Chui, James Manyika, and Mehdi Miremadi, Four fundamentals of workplace automation, McKinsey
Quarterly, November 2015.
114
67
Exhibit 21
Possible interventions to promote rising household incomes for all income groups
Impacted factors
Aggregate
demand
Demographics
Labor
market
Measure the
phenomenon
Revive growth
through
productivity
Increase
opportunities
for individuals
to improve
their earning
potential
Sustain
disposable
incomes
despite weak
growth
Step up
engagement
from business
leaders
70
Capital
Taxes
and
transfers
71
Secular stagnation and low investment: Breaking the vicious circle, McKinsey Global Institute discussion
paper, April 2016.
122
A window of opportunity for Europe, McKinsey Global Institute, June 2015.
123
Game changers: Five opportunities for US growth and renewal, McKinsey Global Institute, July 2013.
124
Infrastructure productivity: How to save $1trillion a year, McKinsey Global Institute, January 2013.
125
Abdul Abiad, Davide Furceri, and Petia Topalova, The macroeconomic effects of public investment: Evidence
from advanced economies, IMF working paper number 15/95, May 2015.
126
Miltiades N. Georgiou, Does entrepreneurship create enough jobs in Europe? A note, Max Planck Institute
of Economics, Group Entrepreneurship, Growth and Public Policy discussion paper number 2006-08,
March 2006.
127
Philippe Aghion et al., Innovation and top income inequality, CEPR discussion paper number 10659,
June 2015.
128
The United States, for example, offers EB-5 visas to business owners, and in the United Kingdom
entrepreneurs with 50,000 in investment funds can apply for a Tier 1 visa.
129
Financing SMEs and entrepreneurs 2016: An OECD scoreboard, OECD, April 2016.
121
72
20%
Projected share
of STEM jobs in
the US labor force
by 2022
Most advanced economies provide primary and secondary education to all legal residents,
but quality can vary widely, and children from low-income families often perform less well
at school than children from higher-income families; some drop out altogether.131 While
there is considerable debate about how to improve the quality of K-12 education, McKinsey
research finds that recognized leaders in secondary educationCanada, Finland,
Singapore, and South Koreafocus on the quality of teaching.132 Schools in these nations
are selective about hiring teachers, pay relatively high salaries, and invest in professional
development. Putting the highest priority on teaching quality (over class size, facilities, and
curricula) has enabled these countries to score highly in international assessments without
outspending rivals.133
73
been shown to push risk-averse students into the humanities and other fields.136 Educators
and policy makers could nudge students into more promising tracks by communicating the
value of different subjects. For example, a recent US study found that university graduates
in the least-paid fields made $3.4million less over the course of their lifetime than those in
the best-paid fields, a fact that might persuade some early-tenured students to opt for more
skill-based courses.137 As we discuss later in this chapter, businesses could also become
involved in these efforts.
Vocational education and apprenticeships can raise skill levels and reduce
youth unemployment
Increased accessibility of college and vocational education and the job relevance of tertiary
education all can affect skill levels and, in turn, income levels. As noted in Chapter 2, the
wage premium paid for US workers with college degrees over workers with high school
diplomas has doubled since the 1980s.138 That premium is a result of both supply and
demand; employers require workers with greater skills, even as the growth in the supply
of college graduates declined from 1980 to 2005.139 Given the rising needs of employers
and the poor wage growth for workers without college degrees, making college financially
accessible to students from all income groups is increasingly important. This is especially
the case in the United States, where annual tuition now averages 22percent of median
disposable income at public universities and 56percent at private colleges. In Europe,
tuition at public institutions makes up only 4percent of median disposable income on
average, with many countries providing entirely free tuition.140
56%
Cost of annual
tuition at private
UScolleges as
proportion of
median disposable
income
A college education is no panacea, however. Many graduates lack relevant skills, according
to employers in some countries, who say that they have left positions unfilled because of a
dearth of qualified applicants.141 Universities can work with regional and national employers
to adapt curricula to the needs of employers to produce graduates with job-ready skills.
Government can facilitate and encourage connections among businesses, universities, and
students to stimulate dialogue between employers and educational providers.
Policy makers and employers could also look to develop more non-college options, such as
vocational training and community college certificate programs. In Germany and Norway,
for example, more young adults attend vocational schools, and in both countries youth
unemployment is lower than in other European economies.142Jobs such as electricians,
medical equipment operators, and advanced industry maintenance workers pay well, are
in high demand, and do not require a college degree.143 One approach is to offer a standard
core high school curriculum complemented by courses tailored to local employers.
Maria De Paola and Francesca Gioia, Risk aversion and major choice: Evidence from Italian students,
University of Calabria working paper number 7, July 2011.
137
Anthony P. Carnevale, Ban Cheah, and Andrew R. Hanson, The economic value of college majors,
Georgetown University Center on Education and the Workforce, 2015.
138
David Autor, Skills, education, and the rise of earnings inequality among the other 99percent, Science,
volume 344, issue 6186, May 2014.
139
Claudia Goldin and Lawrence F. Katz, The race between education and technology: The evolution of U.S.
educational wage differentials, 1890 to 2005, NBER working paper number 12984, March 2007.
140
Tuition data from Education at a glance 2015: OECD indicators, OECD, November 2015, available for Austria,
Belgium, Denmark, Estonia, France, Finland, Italy, the Netherlands, Norway, the Slovak Republic, Slovenia,
Sweden, and the United Kingdom. Median disposable income from OECD, only available for 2012.
141
An economy that works: Job creation and Americas future, McKinsey Global Institute, June 2011.
142
OECD labor statistics, 2016.
143
Patricia Cohen, Its a tough job market for the young without college degrees, The New York Times, May
10, 2016.
136
74
Apprenticeships are another option that could improve matching between graduate skills
and employer needs, for both university and vocational school graduates. Apprenticeships
combine in-school education with part-time employment and on-the-job training, and not
only provide young people with skills that are clearly job-relevant but also help smooth the
school-to-work transition. The key to effective apprenticeship programs is collaboration
between educators and employers to develop curricula and credentialing criteria, giving
young people needed skills and employers qualified workers. The state can help with fiscal
incentives (such as tax breaks), certification, and job placement services.
Across Europe, strong apprenticeship programs have been associated with lower youth
unemployment rates.144 In Germany, where more than 55percent of all 16- to 24-year-olds
go through apprenticeships, youth unemployment is half the European average, and it is
even lower for graduates of vocational programs. Apprenticeships and certification are
required in Germany for occupations ranging from car mechanics to bank clerks. In 2017,
the United Kingdom is introducing an apprenticeship levy of 0.5percent of the annual pay
bill to all UK businesses whose annual payroll exceeds 3million. Companies that pay the
levy will be able to access funds to run apprenticeship programs.145
55%
of German 16- to
24-year-olds are in
apprenticeships
The German system has been transplanted successfully to other countries. In California,
German-based Bayer AG and other biotech companies formed Biotech Partners with
the city of Berkeley in 1993 to train young workers for jobs in their Bay Area facilities. The
partnership now has 35 business and government partners and serves 100 to 125 students
per year from Berkeley High School, Oakland Technical High School, and the Peralta
Community College District through the Biotech Academy, established in 1996. In Oakland,
a district where only 60percent of students graduate from high school (and even fewer
members of minority populations graduate), the academy has had a 100percent graduation
rate for a decade. All academy graduates85percent of whom are from low-income
familiesare accepted to college.
Improving job matching and increasing labor mobility can create opportunities
for income advancement
Inefficient job markets can be a significant barrier to employment. Companies have difficulty
finding qualified workers, and workers have trouble signaling that they have needed skills
and qualifications.
Job matching systems can remove this barrier, helping employers fill vacancies and giving
students and workers information about what skills are in demand so they can choose
courses and training accordingly. The rise of online job platforms has made job matching
more efficient and transparent and, according to MGI research, could help fill the equivalent
of 72million full-time jobs per year by 2025.146 In addition, as many as 230million workers
could find new jobs more quickly, reducing the duration of unemployment, while 200million
who are inactive or employed part time involuntarily could gain additional hours through
freelance platforms.147 There may be an opportunity for government to migrate its own
temporary labor needs onto these platforms. Government could also harness the increasing
amount of data these platforms can provide to inform regulatory initiatives and public
investment in training and retraining.
Federica Origo and Monica Patrizio, The effectiveness and costs-benefits of apprenticeships: Results of the
quantitative analysis, European Commission, September 2013.
145
Apprenticeship levy policy paper, HM Revenue and Customs, February 2016; Tax information and impact
notes, HM Revenue and Customs, 2016.
146
A labor market that works: Connecting talent with opportunity in the digital age, McKinsey Global Institute,
June 2015.
147
Ibid.
144
75
Increasing geographic mobility through affordable housing and transportation can also
improve mobility. Housing can be an obstacle for workers who need to relocate to another
city to find work. For example, in the London metropolitan area, the ratio of house price
to income is about 12 (the reference home costs 12 times the average annual earnings),
compared with four to eight across other major UK cities.148 Unlocking urban land for
housingprimarily through better regulation and transit-oriented developmentcan
create more affordable housing units.149 Geographic mobility also matters for low-income
urban residents who can afford neither a car nor a high-demand neighborhood: they
need inexpensive transport links to jobs. Investment in metropolitan public transportation
can drive such mobility within cities, increasing employment.150 In addition, government
can enable urban redevelopment and expansion, allowing people to live closer to where
they work.
76
pressure on the unemployed to search for work, which substantially accelerated the rehiring
process.155 In the Netherlands, after young workers received unemployment benefits for six
months, they were placed in three-month internships at a special training wage; employers
got tax breaks for participating.156
Enabling older workers to remain employed. Governments can encourage older
workers to remain employed through tax incentives and pension reforms, as well as by
enforcing employment protections, such as antidiscrimination laws. Such measures will
be increasingly important in economies with rapidly aging populations. Until recently, the
trend was going in the opposite direction, especially in some European countries. In Europe
overall, life expectancy has increased by more than nine years since 1970, but the average
effective male retirement age has fallen by six years over the same period.157
Sweden has one of the highest labor-force participation rates of workers who are 65 years of
age and older in Europe, a rate that increased to 17percent in 2014 from 9percent in 2004.
Pension reforms were one measure that encouraged older Swedes to continue working.
Between 1998 and 2003, Sweden moved from a traditional defined-benefit pension system
to a defined-contribution system that takes into account both additional years worked and
longer expected life spans. The new system also decoupled the decision to stop work
from the decision to start drawing benefits and offered increased benefits to workers who
continue working, even if only part time. The result has been people postponing retirement
for several years. In 2007 the government lowered income taxes and employer social
security contributions for employees over 65. The share of workers in the target group who
stayed on after turning 65 subsequently rose by 1.5 percentage points.
In Italy, by contrast, the labor participation rate of workers aged 65 and older is just
4percent, but the government has taken steps to curb early retirement and raise average
retirement ages. Until 2012, women could retire from private-sector jobs with full pensions at
60 and men at 65. The retirement age for all workers is planned to rise to 66 in 2018, and to
67 in 2021.
77
30 to 40percent for the second quintile and 5 to 15percent for the middle quintile. A recent
European MGI study found that a redistribution equivalent to 1percent of GDP could create
additional spending of approximately 200billion ($225billion) by targeting lower-income
households that have a higher propensity to spend their income than wealthier ones.158
Targeting transfers. Transfers include both direct payments, such as unemployment
benefits, and a wide range of in-kind transfers, such as health insurance and subsidies
for housing and food. Researchers have found that direct transfers are usually more
effective than indirect methods because they can more precisely target households.159
If governments seek to address stagnant or falling incomes, they could create targeted
transfers (cash or in-kind) based on metrics of people whose incomes are flat or falling,
rather than solely on income. Trade-offs may be needed, however, as raising taxes to pay
for larger transfers can affect growth. Deficit spending to support higher transfers, which is
common during recessions and recoveries, is problematic for many advanced economies
that have a high ratio of public debt to GDP. Even so, at a time of concern about the
potential for increasingly powerful digital technologies to reduce labor demand, there has
been a revival of the idea of providing a universal basic income for all citizens (see sidebar,
Guaranteed basic incomean idea whose time has come [again]?).
Adjusting taxes. While most advanced economies have progressive income tax laws, the
tax burden for low- and middle-income households can still be substantial. In France, for
example, social security contributions amount to 13percent of annual wages for middleincome workers, about twice the US employee contribution. In both countries, employees
stop making contributions after reaching a certain income threshold (currently $118,500
in the United States and a multiple of 38,000 depending on the type of contribution
in France), which reduces the share of total income that high earners contribute in
payroll taxes. Consumption taxes (sales and value-added taxes, for example) also fall
disproportionately on lower- and middle-income households because they spend a
larger proportion of their incomes on consumption to meet basic needs than wealthier
households, which tend to save proportionately more. Low- and middle-income household
also get limited benefit from tax preferences for income on capital, such as reduced rates
on capital gains. Low tax rates on capital income, in effect, reduce the progressivity of the
tax system.
Reducing taxes on low- and middle-income households raises disposable incomes, but
there are trade-offs. Reducing consumption taxes for less wealthy income groups and
raising taxes on capital income for wealthier groups could shift some of the tax burden
from the poor to the rich. But shifting too much from consumption taxes to income taxes
can have a negative effect on economic output.160 Increased taxation on capital income
of individuals or companies might also have the unintended consequence of shifting
investment out of the country or encouraging high-skill workers to emigrate to lower-tax
countries. Governments seeking to adjust capital taxation policies may need to coordinate
such measures on an international level to avoid capital and talent flight.
78
Christian Hilber and Tracy Turner, The mortgage interest deduction and its impact on homeownership
decisions, Review of Economics and Statistics, volume 96, number 4, October 2014.
161
79
Arindrajit Dube, T. William Lester, and Michael Reich, Minimum wage effects across state borders: Estimates
using contiguous counties, Institute for Research on Labor and Employment working paper number 157-07,
November 2010.
163
Rachel Mack, A measure of help for the working poor: A study of the effects of the minimum wage on welfare
enrollment in the United States, 19902011, Georgetown University, 2014.
162
80
Different countries have different types of minimum wages, but best practices were
recommended by the OECD in its 2015 economic outlook.164 First, countries must account
for regional differences in the cost of living in the minimum wage calculation. Second,
governments should coordinate minimum wage policy with tax-benefit adjustments (such
as lower social security payments) to ease the burden on the employer. Third, adjustments
should be set on a regular schedule. In the United States, where an act of Congress is
required to change the federal minimum wage, the minimum wage in real terms has fallen for
the past 50 years (Exhibit22).
Exhibit 22
Over 50 years, France has increased its minimum wage and narrowed the wage gap between minimum and median
wage earners; in the Netherlands and the United States, low- and middle-wage income inequality has grown
France
Minimum
wage share of
median wage
%
Netherlands
69
59 61
42
1974
United
Kingdom
62
53
51 52
1964
65
United States
49 48
52
44
34
1984
41
36
32
43
37
48
1994
2004
2014
Real minimum
wage (hourly),
2014
$
12
10.8
9.5
10
9.5
7.3
7.5 7.9
7.2
6
4
2 2.8
0
SOURCE: OECD; McKinsey Global Institute analysis
164
165
81
Some countries, such as Denmark, Germany, and the Netherlands, have reformed
employment regulations to put temporary workers on a more equal footing with permanent
workers. In 1997, the Netherlands became one of the first countries to extend access to
training, pension plans, and job security to employees of temporary employment agencies,
and abolished permits for temporary contracts.166 Other economies can consider ways
to extend social safety net programs to contingent workers and to workers employed
as independent contractors via digital platforms such as Lyft. At the same time, policy
makers can review job protection policies for full-time workers that may discourage hiring,
encourage more investment in automation, and result in greater use of contingent labor.167
Studies in the United States have shown that greater flexibility in labor markets can lead to
lower unemployment rates, especially for young and less-educated workers.168
Encouraging work-sharing. Some countries have used work-sharing programs to bolster
employment and maintain incomes in times of reduced demand. These programs either cap
work hours (through mandates and incentives) so employers will add workers, rather than
relying on overtime when demand rises, or they reduce hours for all workers to avoid layoffs
when demand is slack.
When demand is
slack in Germany,
work-sharing can
reduce hours for
up to
18
months
Ton Wilthagen and Frank Tros, The concept of flexicurity: A new approach to regulating employment and
labor markets, Transfer, volume 10, number 2, 2004.
167
Hulya Ulku and Silvia Muzi, Labor market regulations and outcomes in Sweden: A comparative analysis of
recent trends, World Bank policy research working paper number 7229, April 2015.
168
Steven Davis and John Haltiwanger, Labor market fluidity and economic performance, NBER working paper
number 20479, September 2014.
169
Andreas Crimmann, Frank Wiessner, and Lutz Bellman, The German work-sharing scheme: An instrument for
the crisis, ILO Conditions of Work and Employment Series, number 25, 2010.
170
Anders Hayden, Frances 35-hour work week: attack on business? Win-win reform? Or betrayal of
disadvantaged workers? Politics & Society, volume 34, number 4, December 2006.
166
82
The world at work: Jobs, pay, and skills for 3.5billion people, McKinsey Global Institute, June 2012.
See, for example, Dominic Barton, Capitalism for the long term, Harvard Business Review, March 2011;
John Browne, with Robin Nuttall and Tommy Stadlen, Connect: How companies succeed by engaging
radically with society, PublicAffairs, 2016; and Kathleen McLaughlin and Doug McMillon, Business and
society in the coming decades, McKinsey Quarterly, April 2015.
173
John Browne, Heres a better way for companies to tackle big social problems, Harvard Business Review,
March 30, 2016.
171
172
83
Brad Stone. Costco CEO Craig Jelinek leads the cheapest, happiest company in the world. Bloomberg,
June 7, 2013; Aaron Taube, Why the Container Store pays its retail employees $50,000 a year, Business
Insider, October 16, 2014.
175
Talent investments pay off: Cigna realizes return on investment from tuition benefits, Lumina Foundation,
April 2016.
176
Tony Fang and Richard J. Long, Do employees profit from profit sharing? Evidence from Canadian panel data,
IZA discussion paper number 6749, July 2012.
177
Giovanni Andrea Cornia, ed., Falling inequality in Latin America: Policy changes and lessons, Oxford University
Press, 2014; 2013 investment climate statementEcuador, US Department of State, February 2013.
174
84
For government policy makers and business leaders alike, introducing changes that rekindle
income advancement is not straightforward and may require some difficult trade-offs.
Policies to raise productivity may not help reduce income inequality, for example, while
efforts to achieve a more equal income distribution may at times inhibit moves to increase
productivity growth. A number of the most effective policies and business practices we
describe in these pages, especially raising educational attainment standards of young
people and the overall skills of the labor force, could take years to develop. Increasing
transfers to boost disposable income and jumpstart stalled demand is especially complex
and controversial at a time when governments in many advanced economies are struggling
with historically high debt levels. Yet consideration of these and other measures is
essential, for the fundamental issue that we identify in this report, the massive increase in
the proportion of income groups in advanced economies whose income is flat or falling, is
in itself a threat to the well-being of economies, businesses and, more broadly, societies
themselves. All stakeholders, not just low- and middle-income workers and their families,
have a vested interest in ensuring that income advancement restarts and picks up, as it has
for almost the entire modern era.
The world at work: Jobs, pay, and skills for 3.5billion people, McKinsey Global Institute, June 2012.
178
85
TECHNICAL APPENDIX
The appendix has the following sections:
1. Estimating market incomes and disposable incomes for different population deciles
2. Scaling up our analysis to 25 countries
3. Estimating the factors influencing disposable income for median income households
4. Methodology for estimating sensitivities
5. Survey details
For a detailed discussion of this concept, see OECD framework for statistics on the distribution of household
income, consumption and wealth, OECD, June 2013.
179
provides two versions of some components of household income data for 2000, which
we use to adjust income data in years prior to 2000. We adjust our numbers for 1993 by
scaling up the wage and capital income figures based on changes in the gross income,
and we scale disposable income using the changes in disposable income between the
two versions in 2000.180 In the United States, Congressional Budget Office (CBO) data are
available until 2013. We use household incomes that are ranked by the CBO on the basis of
market incomes.
For our calculation of the earnings of the median household in each country, we use the
average values of the fifth and sixth income deciles in the Netherlands, Sweden, and the
United Kingdom, of the third quintile in the United States, and the average of the 40th to the
59thpercentile in France and Italy for market income. For the bottom quintile of households,
we use the average of the first and second income deciles in the United Kingdom, the
Netherlands, and Sweden, of the first 20percent in France and Italy, and of the first quintile
in the United States. We use similar approaches to estimate the averages for the 30th, 70th,
and 90th deciles.
For France, disposable income data is available at 10 percentile intervals. For market
incomes in France and all income data in Italy, data about sources of income by income
decile are not available. We estimate decile-level data by using household-level microdata,
which include income information about 72,000 households in France and 8,000
households in Italy.181 These data include total disposable incomes per household, split
by labor income, capital income, and transfer income. In these data, labor and capital
incomes are reported on an after-tax basis, so we use local tax rates to estimate pretax
labor incomes and to calculate the net value of transfers. Based on the incomes of individual
households, we construct the income distribution and with that we estimate the incomes
for the median households. To be consistent with the median income household data
for the other countries, we use the average incomes of households in the 40th to the
59thpercentile to calculate the median income. For France, this microdata is available only
for 1996 onward, so we use data from 1996, 2002, and 2012 to represent our three points in
time. For Italy, we use the data for 1993, 2002, and 2012.
For all economies, we use the national consumer price indexes from the OECD to calculate
all disposable incomes in 2010 real local currencies.
The CBS in the Netherlands also stopped reporting equivalized incomes by decile after 2000. To make the
incomes comparable we apply the same ratio of equivalized disposable income to unequivalized disposable
income across income deciles. In 1993, we make an additional adjustment for the two slightly different ratios
of equivalized to unequivalized disposable income for 2000.
181
Sample sizes in 2012; in earlier years sample sizes were slightly different.
180
88
Technical Appendix
of the population that is not advancing to calculate outcomes on the aggregate level for
advanced economies.
To ensure that our clustering methodology does not dictate the outcomes, we compare
outcomes of our main scaling methodology with two other approaches. First, we use an
alternative grouping of countries into six groups based on per capita GDP and net Gini
coefficients. Second, we use a simple population-weighted average of just the six countries.
The outcomes from all three methods of estimation were similar (ExhibitA1).
Exhibit A1
Our three scaling methodologies give similar outcomes
Average share of population affected between the two periods for three scaling methodologies, 200514
% of population
Our
methodologies
I. Similar GDP growth rates and change in the net Gini coefficients
II. Similar GDP per capita and net Gini coefficients
III. Population-weighted average for the six countries
Not getting by
Not catching up
Disposable
income
Market
income1
64
15
15
15
II
III
62
Flat or falling
57
23
II
III
63
62
63
68
II
III
28
23
II
III
74
78
II
III
1 Market income data is not shown for the not getting by segment as countries do not typically measure pre-transfer poverty rates.
SOURCE: Frederick Solt, The standardized world income inequality database, working paper, SWIID version 5.0, October 2014; OECD; McKinsey Global
Institute analysis
89
90
Technical Appendix
For taxes, we use different approaches for each country. In the Netherlands, we include
income taxes (and wealth tax until 2001) and dividend taxes. In Sweden, we include federal,
local, and municipal taxes such as property taxes. In the United Kingdom, we include all
direct taxes but exclude indirect and intermediate taxes. In the United States, we use CBO
data that include only federal taxes and are available until 2013. In our analysis extending
this time period to 2014 using Current Employment Statistics data, we also include state and
local taxes (which account for about 6percent of market income for the middle quintile of
households). In France, we include personal income taxes, housing taxes, and other social
contributions. In Italy, we include all income taxes, municipal taxes (such as waste and water
tax), and corporate income taxes.182
We calculate the changes in the five effects driving disposable income in two periods, 1993
2005 and 200514.183 To account for the different lengths of the periods and make figures
for the two periods comparable, we express all growth rates in our analysis and results in
terms of seven-year growth.
91
and gender. Within each micro-segment, we also divide labor-force participants into those
employed in interactive jobs (exchanges involving complex problem solving, experience, and
context), transactional jobs (exchanges that can be scripted, routinized, or automated), or
production jobs (the process of converting physical materials into finished goods), plus the
unemployed.184
We start with the supply of labor-force participants at the level of micro-segments. We
calculate the future supply of workers using national population forecasts for the three
countries for age and gender categories. We extrapolate educational achievement trends
of the past decade from our microdata to understand the total supply of working-age
population in 2025 per educational achievement micro-segment. We extrapolate trends in
labor-force participation rates of the past decade for micro-segments to create the size of
2025 micro-segments of labor-force participants. We use the same labor supply model for
all scenarios.
The demand for labor depends on total GDP and the wage share of income in each
economy. In the low-growth hypothesis, we assume that the trends of the past decade in
productivity per labor-force participant will continue to define total GDP. In the two variants
of the high-growth hypothesis, we assume that the trends in productivity growth of the three
decades prior to 2005 will resume. To link demand in the economy to demand for labor, we
use wage share of GDP at the economy-wide levelover the past decade in the low-growth
and labor disruption analyses and over the three decades before 2005 in the high-growth
analysis (2002 to 2012 for Italy and France and 2003 to 2013 for the United States). We
split labor demand between growth in jobs and growth in wages, also based on the trends
of the past decade. We split the total demand for jobs into three categories: interactive,
transactional, and production jobs, based on trends in the past decade.
We then distribute the total demand for each type of job across the micro-segments of
workers based on the trends observed in the past decade in jobs that involve interaction,
transaction, or production.185 To reflect variations across scenarios, we adjust the
employment rates as well as labor incomes of those employed in each micro-segment
according to the estimated changes in productivity and labor demand, split between job
types, in our three scenarios.
Based on the results for each micro-segment, we convert the labor incomes for each
individual in 2012 microdata to an estimated labor income in 2025. We also adjust the
weights of the individuals in 2012 to reflect the changes in composition of the labor force
by occupation and employment in 2025 (for example, if we assume unemployment will be
higher in 2025, for instance in the labor disruption scenario, the weights of those who are
unemployed in 2012 would be increased).
Help wanted: The future of work in advanced economies, McKinsey Global Institute, March 2012.
Ibid.
184
185
92
Technical Appendix
5. SURVEY DETAILS
We commissioned Research Now to carry out online consumer surveys in the United States
and the United Kingdom in August 2015 and in France in September 2015. In each country,
the surveys were answered by approximately 2,000 respondents, providing a representative
sample of the population by income, age, and gender (as well as by region in France).
The survey contained three sets of questions: sociodemographic questions to classify
respondents; questions regarding the respondents past, current, and expected future
economic situations; and questions regarding their perspectives on trade and immigration.
93
Technical Appendix
BIBLIOGRAPHY
A
C
Card, David, The impact of the Mariel boatlift on the
Miami labor market, Industrial and Labor Relations
Review, volume 43, number 2, January 1990.
Carnevale, Anthony P., Ban Cheah, and Andrew R.
Hanson, The economic value of college majors,
Georgetown University Center on Education and the
Workforce, 2015.
E
The Economist, The world economy: Out of ammo?
February 20, 2016.
Elmendorf, Douglas W., Estimated macroeconomic
impacts of the American Recovery and Reinvestment Act
of 2009, CBO letter to Senator Charles Grassley, March
2, 2009.
Emmerson, Carl, and Gemma Tetlow, UK public
finances: From crisis to recovery, Fiscal Studies, volume
36, number 4, December 2015.
European Commission, Europe 2020 in the
Netherlands, 2016.
D
DAddio, Anna C., Intergenerational transmission of
disadvantage: Mobility or immobility across generations?
A review of the evidence for OECD countries, OECD
Social, Employment and Migration working paper number
52, OECD, 2007.
DArcy, Conor, and Gavin Kelly, Securing a pay rise: The
path back to shared wage growth, Resolution Foundation,
March 2015.
Davis, Steven J., and John Haltiwanger, Labor market
fluidity and economic performance, NBER working paper
number 20479, September 2014.
De Paola, Maria, and Francesca Gioia, Risk aversion and
major choice: Evidence from Italian students, University of
Calabria working paper number 7, July 2011.
96
Bibliography
G
Georgiou, Miltiades N., Does entrepreneurship create
enough jobs in Europe? A note, Max Planck Institute of
Economics, Group Entrepreneurship, Growth and Public
Policy discussion paper number 0806, March 2006.
Goldin, Claudia, and Lawrence F. Katz, The race
between education and technology: The evolution of
U.S. educational wage differentials, 1890 to 2005, NBER
working paper number 12984, March 2007.
H
Haan, Peter, Daniel Kemptner, and Victoria Prowse,
Inequality and defined benefit pensions when life
expectancy is heterogenous, DIW Berlin, February 2015.
Hayden, Anders, Frances 35-hour work week: attack on
business? Win-win reform? Or betrayal of disadvantaged
workers? Politics & Society, volume 34, number 4,
December 2006.
Helliwell, John, Comparative macroeconomics of
stagflation, Journal of Economic Literature, volume. 26,
issue 1, 1988.
Hilber, Christian A. L., and Tracy M. Turner, The mortgage
interest deduction and its impact on homeownership
decisions, Review of Economics and Statistics, volume
96, number 4, October 2014.
HM Revenue and Customs, Apprenticeship levy policy
paper, February 2016.
HM Revenue and Customs, Tax information and impact
notes, 2016.
Hornstein, Andreas, Per Krussel, and Giovanni L.
Violante, The effects of technical change on labor market
inequalities, Center for Economic Policy Studies working
paper number 113, July 2005.
I
ILO, Global wage report 2012/13: Wages and equitable
growth, December 2012.
ILO, Global wage report 2014/15: Wages and income
inequality, December 2014.
ILO and OECD, The labour share in G20 economies,
February 2015.
IMF, Fiscal policy and income inequality, staff report,
January 2014.
K
Karaborbounis, Loukas, and Brent Neiman, Declining
labor shares and the global rise of corporate saving, NBER
working paper number 18154, June 2012.
Karabarbounis, Loukas, and Brent Neiman, The global
decline of the labor share, NBER working paper number
19136, June 2013.
Karsten, Jack, and Darrell West, Rising minimum wages
make automation more cost-effective, Brookings
Institution, September 2015.
Keeley, Brian, Income inequality: The gap between rich
and poor, OECD, December 2015.
L
Lawrence, Robert Z., Does productivity still determine
worker compensation? Domestic and international
evidence, in The US labor market: Questions and
challenges for public policy, Michael R. Strain, ed.,
American Enterprise Institute, 2016.
Lumina Foundation, Talent investments pay off: Cigna
realizes return on investment from tuition benefits,
April 2016.
M
Mack, Rachel, A measure of help for the working poor:
A study of the effects of the minimum wage on welfare
enrollment in the United States, 19902011, Georgetown
University, 2014.
Marino, M. Rosaria and Zizza, Roberta, The personal
income tax evasion in Italy: an estimate by taxpayers
type, Bank of Italy, 2014.
McKinsey & Company, Beyond corporate social
responsibility: Integrated external engagement,
March 2013.
McKinsey & Company, How the worlds most improved
school systems keep getting better, November 2010.
McKinsey & Company and LeanIn.Org, Women in the
workplace, September 2015.
McKinsey Center for Government, Education
to employment: Designing a system that works,
January 2013.
McKinsey Global Institute, Beyond austerity: A path to
economic growth and renewal in Europe, October 2010.
McKinsey Global Institute, A blueprint for addressing the
global affordable housing challenge, October 2014.
McKinsey Global Institute, Debt and (not much)
deleveraging, February 2015.
McKinsey Global Institute, Digital America: A tale of the
haves and have-mores, December 2015.
McKinsey Global Institute: Diminishing returns: Why
investors may need to lower their expectations, May 2016.
97
N
Neumark, David, The effects of minimum wages on
employment, FRBSF Economic Letter, volume 34,
December 21, 2015.
Nichols, Austin, and Jesse Rothstein, The earned income
tax credit, NBER working paper number 21211, May 2015.
O
OECD, Education at a glance 2015: OECD indicators,
November 2015.
OECD, Families are changing, in Doing better for
families, 2011.
OECD, Financing SMEs and entrepreneurs 2016: An
OECD scoreboard, April 2016.
98
Bibliography
P
Palma, Jos Gabriel, Homogeneous middles vs.
heterogeneous tails, and the end of the Inverted-U: The
share of the rich is what its all about, Development and
Change, volume 42, number 1, January 2011.
Piketty, Thomas, Capital in the twenty-first century,
Belknap Press, 2014.
Politi, James and Ghiglione, Davide. Matteo Renzis
early pension proposals draw mixed reactions, Financial
Times, May 29, 2016
R
Rosholm, Michael, Experimental evidence on the nature
of the Danish employment miracle, IZA discussion paper
number 3620, July 2008.
Rothwell, Jonathan, The hidden STEM economy,
Metropolitan Policy Program at Brookings, June 2013.
Z
Zipperer, Ben, How raising the minimum wage ripples
through the workforce, Washington Center for Equitable
Growth, April 2015.
U
Ulku, Hulya, and Silvia Muzi, Labor market regulations
and outcomes in Sweden: A comparative analysis of
recent trends, World Bank policy research working paper
number 7229, April 2015.
United Nations University, The rise and fall in income
inequality in Ecuador, UNU-WIDER, October 2014.
US Department of State, 2013 investment climate
statementEcuador, February 2013.
US Department of the Treasury, Income mobility in the US
from 1996 to 2005, November 2007.
W
Willetts, David, The pinch: How the baby boomers took
their childrens futureand why they should give it back,
Atlantic Books, 2011.
Wilthagen, Ton, and Frank Tros, The concept of
flexicurity: A new approach to regulating employment
and labour markets, Transfer, volume 10, number
2, 2004.
99
Bibliography
RELATED MGI
AND MCKINSEY RESEARCH
Diminishing returns: Why investors
may need to lower their expectations
(April 2016)
The forces that have driven exceptional
investment returns over the past 30 years are
weakening, and even reversing. It may be time
for investors to lower their expectations.
www.mckinsey.com/mgi
E-book versions of selected MGI reports are available at MGIs website,
Amazons Kindle bookstore, and Apples iBooks Store.
Download and listen to MGI podcasts on iTunes or at
www.mckinsey.com/mgi/publications/multimedia/
Cover image Sam Edwards/Getty Images.
Contents page images (top to bottom): Family on sofa Jannike Stratton/
Getty Images; man with walker and aide Johnny Greig/Alamy Stock Photo;
man reading Bill Cheyrou/Alamy Stock Photo.