See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/225617595

Inequality and business cycles in the U.S. and European Union

countries

Article  in  International Advances in Economic Research · August 1999

DOI: 10.1007/BF02296415

CITATIONS READS

26 204

2 authors:

Sophia P. Dimelis Alexandra Livada

Athens University of Economics and Business Athens University of Economics and Business
48 PUBLICATIONS   1,541 CITATIONS    31 PUBLICATIONS   204 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Reconciliation between work and private / family life : a pilot statistical study for Greece View project

FDI and technical efficiency View project

All content following this page was uploaded by Sophia P. Dimelis on 10 January 2015.

The user has requested enhancement of the downloaded file.

 Inequality and Business Cycles
in the U.S. and European Union Countries

SOPHIA DIMELIS AND ALEXANDRA LIVADA*

This paper derives the business cycle properties of some aggregate and disaggregate
inequality indices for the U.S. and three European Union countries (United Kingdom, Italy,
and Greece). The findings suggest that inequality indices move countercyclically with output
in the U.S. and the United Kingdom, a procyclical behavior prevailed in Greece, and a mixed
cycle influenced Italy. A common countercyctical pattern of inequality indices with inflation and
unemployment characterizes the three large economies (U.S., United Kingdom, and Italy).
Also, in most countries, the top income group seems to lose at the benefit of the rest during
inflationary periods while, in all four countries, the poor will gain from inflation and suffer
from unemployment. (JEL E32, D63, 057)

Introduction

A challenging topic in income inequality research has always been the analysis of how
aggregate and disaggregate inequality indices fluctuate and how they are influenced by
macrovariables. The majority of studies in relevant literature use traditional econometric
techniques to examine the distributive impact of the most important macroeconomic
aggregates such as inflation, unemployment, and growth. These studies suggest that
unemployment increases inequality while growth and inflation ambiguously affect income
distribution. Some studies for the U.S. are Schultz [1969], Mirer [1973a, 1973b], Beach
[1977], Blinder and Esaki [1978], and Blank and Blinder [1986]. Studies for other
countries include, among others, Schultz [1969] for the Netherlands, Buse [1982] for
Canada, Nolan [1989] for the United Kingdom, Bjorklund [1991] for Sweden, and
Achdut [1996] for Israel. However, the cyclical influences in these studies were either
ignored or captured ad hoc by a trend factor among the regressors.
Income inequality has also been viewed extensively in literature in connection with
development economics. The question of whether or not a trade-off exists between
inequality and growth remains controversial. This literature can be traced back to
Kuznets' [1955] inverted U-curve hypothesis which postulates that, as economic
development occurs, income inequality widens in the early phase and narrows at later
stages of development. This hypothesis has been investigated both theoretically and
empirically in various studies [Chenery et al., 1974; Danziger and Gottschalk, 1982;

*Athens University--Greece.The authors appreciate the helpfulcomments from the participantsof the
InternationalAtlanticEconomicConference,Rome, Italy,March 14-21, 1998and fromthe participantsof the
1997 meetingsof the Societyfor EconomicDynamics,Oxford, UnitedKingdom,July 1997.

321
322 IAER: AUGUST 1999, VOL. 5, NO. 3

Lindert and Williamson, 1985]. Further evidence and recent critiques on the U-curve
hypothesis can be found in Canlpano and Salvatore [1988], Ram [1988], and Anand and
Kanbur [1993].
On the other hand, a series of studies in growth literature explores the opposite
linkage, that is, the causal relation from income distribution to growth. There is empirical
evidence [Clarke, 1995] that inequality, as captured by conventional measures, exerts a
significant and negative impact on the level of output as well as on its growth. These facts
have given rise to many theoretical and empirical works that explore the relationship
between income distribution and growth through various channels such as investment in
human capital, tax policies, capital market imperfections, or political processes (see, for
example, Galor and Zeira [1993], Perotti [1993], Persson and Tabellini [1994], and
Alesina and Rodrik [1994] ).
Within business cycle literature, however, the evidence is very scarce. In an earlier
classic study, Kuznets [1953] found a tendency for overall inequality to move
countercyclically in the U.S. during the interwar period (1919-38). He argued that income
shares of upper income groups decline during business expansion and increase during
business contraction. Some years later, Creamer [1956] examined U.S. personal income
and its components over four decades and his findings were in line with those of Kuznets.
However, these conclusions were not accepted by all due to the limitations of the official
statistics.
This paper takes up the issue of how income inequality behaves over the business
cycle. It examines the cycles of income inequality and determines the stylized facts of
their comovements with the cycles of the main macroeconomic aggregates of real per
capita gross domestic product (GDP), inflation rate, and unemployment rate. Analysis is
performed for the U.S. and three European Union (EU) countriesmthe Unite~ Kingdom,
Italy, and Greece--using the Kydland and Prescott [1990] methodology. The intention
here is to identify any empirical regularities displayed by the income inequality data
without actually imposing any theoretical structure. Also, some of the results are
contrasted to the evidence implied by a Kuznets type of curve obtained from the time
series data.
The type of analysis performed here is useful in comparing countries of different
socioeconomic and political backgrounds related to the pattern of their business cycles.
Any empirical regularities observed in these countries will have important implications
for income distribution and the transmission of the business cycles between the U.S. and
EU countries. This study may also provide useful information about the distributive
impact of any socioeconomic programs or policies undertaken in the EU or the U.S.
These facts can finally be used for theorization and calibration in the growing literature
of business cycles and income distribution.
The rest of the paper is organized as follows. The second section describes the
methodology and the data series used in the analysis. The third section presents the main
statistics that summarize the business cycle characteristics of the income inequality indices
and their comovements with the real per capita output and the rates of inflation and
 DIMELIS AND LIVADA: INEQUALITY AND BUSINESS CYCLES 323

unemployment for each country. The fourth section provides a summary and concluding
remarks.

Methodology and Data

The paper studies the business cycle features of income inequality measures by
applying the methodology of Kydland and Prescott [1990]. This procedure is based on
the Lucas [1977] definition of the business cycle as the component of a variable that is
derived by taking deviations from its smooth trend. That is, growth business cycles are
computed as opposed to the classical business cycles, which refer to the levels of the
variables.~ There are several ways to represent a smooth trend such as using deterministic
polynomial functions of time, differencing, or other stochastic procedures. This study
chooses the Hodrick and Prescott [1980] filter (hence, HP filter) for comparative
purposes since it has become a standard detrending technique in recent business cycle
literature. 2
The income inequality time series data used in this study are published estimates of
three aggregate inequality indices: the Gini coefficient, the coefficient of variation (CV),
and the first index of Theil. Quintile income shares (Qi, i - 1,..., 5) are also used as
disaggregate inequality measures. The main criteria for the choice of these indices are to
satisfy certain properties [Sen, 1973; Cowell, 1977] and to have the same measures for
a long period of time in all four of the countries examined. Analysis is not restricted to
only one index since there is almost universal consensus that no perfect inequality index
exists nor hypothetical statistical distribution to describe it. For example, the Gim
coefficient tends to attach more weight to income transfers that occur around the middle
income classes, the CV is very effective in reflecting inequality mainly among high
incomes, and Theil's index attaches equal weight to transfers at the lower and upper end.
In this sense, using different inequality measures can be viewed as a test of robustness on
the results.3 The inequality measures and their data source as well as the period covered
by country are as follows:
1) for the U.S., 1947-89: Gini, Theft, and Q~ [Slottje et al., 1989; Hayes et al., 1990];
2) for the United Kingdom, 1960-84: CV, Gini, and Qi [Atkinson and Micklewright
[1992];
3) for Italy, 1977-89: CV, Gini, Theft, and Qi [Brandolim and Cannari, 1994;
Brandolini, 1992]; and
4) for Greece, 1959-93: CV, Gini, Theil, and Q~ [Livada, 1991] (updated).
The above annual inequality measures are estimated using consumer income data from
the Current Population Report for the U.S., tax unit net income from the Blue Book for
the United Kingdom, family income from tax declaration forms for Greece, and family
income from surveys by the Bank of Italy for Italy. The macroeconomic variables
considered are the GDP per capita in constant prices, the inflation rate derived from the
Consumer Price Index (CPI), and the unemployment rate. The main data source for these
variables is the Economic Outlook of the OrganiTation of Economic Cooperation and
324 IAER: AUGUST 1999, VOL. 5, NO. 3

Development (OECD) [ 1996] for 1960 onward. For previous years, national sources were
used.
As an illustration, Figure 1 presents the evolution of the Gini coefficient along with its
trend computed by the HP filter. It becomes quite clear that both the U.S. and the United
Kingdom have experienced an increasing inequality trend since the early and mid-1970s,
respectively. After a period of stable inequality, Greece experienced an upward trend
since the mid-1980s, though much smaller, and, only in Italy is there a systematic
downward trend since the late 1960s. Similar observations have been made by Levy and
Murnane [1992], Gottschalk [1997], and others for the U.S.; by Johnson and Webb
[1993] and Jenkins [t995] for the United Kingdom; by Dimelis and Livada [1995] for
Greece; and by Brandolini and Sestito [1994] for Italy.

FIGURE 1
Actual and Fitted Values of the Gini Coefficient

0.47 0A4

0.46 USj
0.45 '/' 0.42

0.44~ _ .
0.40
0.43
0.42. 0.38
0.41
0.40 "5'0'' ?~' ' ' 6 b " ' 6 ' ~ " ~'b' ' ' ' 7 ~ ' ' ' 8 b 8'5
0.36
60 65 70 75 80 85 90

0.44
0.37 . UK
ITALY
0A2

0.40

038

0.36

V 0,34

0.31~ '7'0~ '7'5.... 810.... 8'5'' 0.32

60 65 70 75 80 85

The trend of high and growing inequality in the U.S. and the United Kingdom has been
an issue of great concern and has raised a major debate over the causes and consequences
of this phenomenon. It is widely accepted today that the rise in income inequality is due
to the widening of the dispersion in earnings or wages. Several explanations have been
put forth for this rise among which, the most often quoted for the U.S. economy are
 DIMELIS AND LIVADA: INEQUALITY AND BUSINESS CYCLES 325

shifts in labor demand away from low-skilled labor as a consequence of skill-biased

technological change, trade with low-wage developing countries, and international
competition [Blank, 1995; Johnson, 1997]. In the United Kingdom, the causal forces are
a mixture of earnings inequality, employment structure, and changes in the tax and
benefit system as documented by Jenkins [1995]. In Greece, the rise in inequality seems
to have occurred right afXer the austerity program enforced in 1986 to control rising
deficits and inflation. In Italy, according to recent estimates, the trend of inequality has
also been rising after 1989 due to institutional reforms (see OECD [1996] ).
Following the methodology of Kydland and Prescott [1990], first compute the trend
component of each series using the HP filter. Then obtain the cyclical component by
taking deviations of the above trend from the actual series. All deviations are expressed
in percentage terms by taking natural logarithms of the level variables. For the share
variables, such as the inequality indices or rates, no other transformation was undertaken.
Next, for each detrended series, compute:
1) the standard deviation as a measure of the relative amplitude of the fluctuations;
2) the cross correlations of the reference variable with each one of the inequality
indices (contemporaneous correlations); and
3) the cross correlations of the reference variable with a number of lags and leads of
the same index to account for the degree of comovements.
If the contemporaneous correlation is positive (negative) and statistically significant, then
the inequality index is procyclical (countercyclical), meaning that the cycles of the index
move in the same (opposite) direction as that of real output or other pertinent variable.
A number close to zero indicates that no contemporaneous correlation exists. For the
statistical significance of the correlation coefficients, the two-sided t-test was used to test
the null hypothesis of a zero correlation at the 5 percent significance level. The cross
correlations of the reference variable with leads and lags provide information on the phase
shift of the inequality cycles relative to the cycles of the reference variable.

Business Cycle Facts

In this section, Table 1 will present the volatility of the inequality indices and
macrovariables by country. This is the standard deviation of the relevant detrended series
which measures the variability of the cyclical fluctuations. For comparison purposes, the
standard deviation is expressed in percentage terms for all series. Tables 2 through 5 will
summarize the comovements of the income inequality cycles relative to those of real
output, rates of inflation, and unemployment. Following the description in the previous
section, under column x(t), the contemporaneous cross correlations are reported to
determine procyclicality (positive correlation), countercyclicality (negative correlation),
or an acyclical behavior (zero correlation). Under the columns of x(t±j),
for j = 1, ..., 3, the cross correlations of the pertinent variable with three lags and leads
of the inequality index series x are reported. If this correlation has its largest absolute
value in entries x(t -j), x(t), or x(t ÷j), then the cycle of income inequality (x) is
leading by j periods, is synchronous with j periods, or is lagging by j periods the cycle
326 IAER: AUGUST 1999, VOL. 5, NO. 3

of the pertinent index, respectively. The facts are presented and discussed separately for
each case in the following sections.
Volatility
Looking first at the volatility of the cycles in Table 1, observe that the standard
deviation of the GDP fluctuations varies from a low of 1.8 percent in Italy to 2.4 percent
in Greece and the U.S. Inflation cycles are more volatile than GDP in all countries except
the U.S. while the unemployment cycles are highly volatile everywhere. Among the
aggregate inequality indices, the CV is the most volatile measure while Gini shows the
least volatility, except in the U.S. The cycles of income shares have, as expected, smaller
variability, with the top and bottom quintiles (Q5 and Q1) being the most variable.
Overall, the cycles of the main macrovariables and some of the most common inequality
measures show considerable uniformity among countries with respect to the relative
amplitude of the above cycles.

TABLE 1
Volatility of Cyclical Fluctuations

Variables U.S. United Kingdom Italy Greece

GDP 2.39 2.02 1.80 2.40

CPI Inflation Rate 1.25 2.52 2.78 3.82
Unemployment Rate 99.70 20.60 51.80 87.50
Gini .77 .55 1.46 .92
CV m 1.71 11.02 10.57
Theil .56 -- 3.08 1.59
Q1 .15 .32 .39 .45
Q2 .15 .23 .30 .29
Q3 .13 .24 .23 .21
Q4 .10 .22 .15 .29
Q5 .35 1.14 .91 .67

Notes: Figures are standard deviations.

Inequality and GDP

Table 2 presents the results with respect to real GDP per capita. The correlation
coefficients at time t show that in the U.S. and the United Kingdom, all aggregate
inequality indices move countercyclically with real output. In Italy, the CV moves
 DIMELIS AND LIVADA: INEQUALITY AND BUSINESS CYCLES 327

countercyclically but Gini and Theil move procyclically. In Greece, all indices are
procyclical. However, these coefficients are significant only in the case of Greece. 4 A
diagrammatic representation of the above comovements with regard to the Gini coefficient
is given in Figure 2. A more careful examination of these graphs suggests that the above
behavior for the U.S., the United Kingdom, and Italy has not been uniform in the entire
period examined. This may explain the low correlation coefficients computed for these
countries. On the other hand, Greece presents a dichotomy in the pattern of income
inequality over business cycles, resulting in countercyclical behavior from the late 1980s
onward: The pattern of the phase shift is variable among countries as indicated by the
cross correlations with leads and lags. The only exception is Greece where inequality is
consistently lagging.

TABLE 2
Real Per Capita GDP and Income Inequality Indices by Country

Cross Correlation of Real GDP with:

Index Country x(t - 3) x(t - 2) x(t - 1) x(t) x(t + 1) x ( t + 2) x ( t + 3)

CV U.S. -- . . . . . .
U.K. -. 10 0.22 -.31 -.24 -.30 -. 15 -.03
Italy .77 .47 -.08 -.28 -.32 -.36 -.31
Greece -. 15 -.24 -.26 .43 .47 .26 -.02

Gini U.S. .25 .26 .17 -. 15 -.24 -.22 -. 19

U.K. .03 -.01 -.17 -.24 -.31 -.33 .04
Italy .05 .02 .18 .12 -.11 -.54 -.16
Greece -. 16 -.21 -.05 .30 .32 .27 -. 12

Theil U.S. .22 .19 .22 -.20 -.23 -.22 -. 13

U.K. -- . . . . . .
Italy .20 .52 .70 .27 -.40 -.57 -.41
Greece -.18 -.21 -.11 .37 .43 .31 -.11

Notes: Deviations from trend are from annual data. Sample size for the U.S. is from 1947-89; for the United
Kingdom (U.K.), 1960-84; for Italy, 1977-89; and for Greece, 1959-93.
328 IAER: AUGUST 1999, VOL. 5, NO. 3

FIGURE 2
Business Cycles: Log GDP Per Capita (LGDPH) and Gini Coefficient
0.04 0,08.
0.06. Greece
0.02 /'~ LGDPH
0.04
0.00t ~---11 ') 1, A LGDPH
0.02 .. )! Ok. ^

-0.02 0.00.
-0.02.
-0.04
4).04.
4).06 -0.06.
}.... 6's .... fo.... Is .... 8'o.... s'8.... 9'o

o.os ] UK
0.04

0.02 ALGDPH
0.04 A A I
I
I
i

o.ook
A 0.00i l t

-0,021
V "
v,..--l / .o.= vv .,..;,,,
"°'°4640.... 6'5.... fo.... ¢5.... 8'0.... s's.... 9'0' -0.04~0,, '6's' ' ' '¢0" ''"'7's. . . . s'o . . . . 8's . . . . 9'0'

The previous evidence is documented by looking at the movements of the disaggregate

indices in Table 3. The cyclical behavior of the quintile income shares in the U.S.
suggests that over the output expansions, the middle and lower income classes benefit the
most, while the upper-middle and top income classes seem to lose. This becomes evident
from the positive and significant correlation coefficients for Q1 and Q2 and the
significantly negative correlation coefficient for Q5. In the United Kingdom, only the
lower income class appears to benefit from output expansion while the rest of the classes
either lose or are unaffected. An opposite picture characterizes the case of Greece where
the top income class (Q5) benefits the most at the expense of the rest of the population
during expansions. In Italy, the lower income classes (Q1 and Q2) lose, the middle
income class gains, and the rest seem to be unaffected during the GDP upturns.
Therefore, the results might be interpreted as implying that recessions are bad for overall
equality or that inequality results in recessions, or both.
The previous evidence is also related to the well-known puzzle in growth literature of
whether inequality is harmful to growth or helpful. In this literature, the political regime
seems to play a significant role in determining the relationship between inequality and
 DIMELIS AND LIVADA: INEQUALITY AND BUSINESS CYCLES 329

growth. For example, van Wijck [1992] provides evidence for the U.S. that inequality
trends were negatively influenced by democratic administrations for the period 1959-89.
From a time series and cross section sample of developed and less-developed countries
in the period 1960-85, Persson and Tabellini [1994] found that over the periods of
democratic regimes, output growth and inequality were negatively correlated. However,
Clarke [1995] showed that this negative relationship holds for both democracies and
nondemocracies. A useful extension to the present analysis would therefore be to examine
whether the political regime differentiates the previous conclusions with respect to the
cyclical behavior of inequality.
TABLE 3
Cross Correlations with Quintile Indices by Country

Cross Correlations of Variable x(t) with:

Quintile Country Real Per Capita CPI Inflation Unemployment
GDP Rate Rate

First U.S. .45 .22 -.61

United Kingdom .31 .22 -.24
Italy -.16 .52 -.22
Greece .00 .04 -.12
Second U.S. .40 .28 -.57
United Kingdom -.10 .07 .21
Italy -.10 .51 -.37
Greece -.21 -.28 .00
Third U.S. .27 .28 -.35
United Kingdom -.04 .21 -.13
Italy .13 .60 -.44
Greece -.56 -.31 .44
Fourth U.S. -.22 .16 .29
United Kingdom .05 .32 -.28
Italy -.07 .17 -.35
Greece -.27 -.53 .38
Fifth U.S. -.38 -.32 .53
United Kingdom -.02 -.19 .06
Italy .08 -.57 .39
Greece .43 .27 -.22
330 IAER: AUGUST 1999, VOL. 5, NO. 3

This paper also attempts to combine its findings with evidence from a Kuznets type of
curve. Most of the empirical studies on the Kuznets hypothesis are based on cross-
sectional data. 6 Using the time series data, Figure 3 presents the association of inequality
as measured by the top 20 percent income share (Q5) with the logarithm of per capita
GDP at each point in time for the four countries examined. What characterizes these
figures is a sequence of inverse U-shaped curves rather than a single Kuznets type of
curve.

FIGURE 3
Kuznets Curves: Income Share of Top 20 Percent (Q5)
and the Log Real Per Capita GDP (LGDPH)

0.45
us y 0.49
0.48 73
Greece

0.44
0.47
0.43
0.46 59 9 ~ 883
Q5 61 Q5
0.42 80 0,45
0.41 0.44
0.40 0.43
8.0 8~2 8:4 8:6 8;8 9.0 5.0 5:5 6:0 6:5 7.0
LGDPH LGDPH
0.44 44
UK Italy
84 77
0.42 43
87
0.40 6 42
Q5 Q5
0.38 41

0.36. 40
77 ~
0.34 39
8.2 813 814 815 816 817 8.8 9.3 9.'4 9:5 9~6 9,7
LGDPH LGDPH

More specifically, in the U.S., the top income share is steadily rising over a period of
sluggish growth of per capita GDP, mostly from the mid 1970s onward, despite the fact
that in the entire period examined, inequality was found to be decreasing over business
cycle upturns. A similar analysis holds for the United Kingdom. However, in Greece, this
phenomenon is more severe since over the years of recession and contractionary policy
measures (1986-93), an almost vertical increase in the top income share has occurred,
leading to a more unequal income distribution. On the other hand, in Italy, with
 DIMELIS AND LIVADA: INEQUALITY AND BUSINESS CYCLES 331

reservation of the limited sample, a downward movement of the top income share has
prevailed since 1977. The above findings therefore support Atkinson' s [1993] observation
that for the United Kingdom and the U.S., "the trend in recorded income inequality
departs from Kuznets' hypothesis .... The inverse-U shape is replaced by a curly AV ." Of
course, in order to determine the exact form of relationship, a rigorous econometric
analysis is required which is beyond the scope of this paper.
Inequality and Inflation
The distributional impact of inflation has been a controversial issue in literature. Some
argue that inequality is unaffected by inflation [Buse, 1982; van Wijck, 1992]. Others
claim that only some of the poor would gain [Blinder and Esaki, 1978; Nolan, 1989]. On
the other hand, the incomes of some groups, such as wage earners and pensioners, are
slower to adapt in inflationary periods, whereas other groups' incomes (for example, the
self-employed) respond relatively quickly to inflation. Attention should also be paid to
the definition of income or the type of inflation employed when comparing the
redistributive effects of inflation among various studies. 7 Schultz [1969] argues that in the
case of demand-pull inflation, inequality rises because prices rise faster than costs,
resulting in higher profit shares and thus higher shares of the upper income groups. In
the case of cost-push inflation, however, inequality falls because the profit share decreases
relative to wages.
The contemporaneous cross correlations reported in Table 4 reveal that most aggregate
inequality indices move countercyclically with inflation in all countries except Greece.
The correlation coefficients are significant only in the U.S. and in the following cases:
for the United Kingdom, the Gini; for Italy, the Theil; and for Greece, the CV and Theil.
Regarding the timing in cycle phases, a variable pattern emerges. In Table 3, the behavior
of disaggregate indices regarding inflation shows that low, middle, and upper-middle
income groups move procyclically with inflation in the U.S., the United Kingdom, and
Italy, but countercyclically in Greece. The top income group seems to lose at the benefit
of the rest during inflation periods in all countries but Greece.

TABLE 4
CPI Inflation Rate and Income Inequality Indices by Country

Cross Correlation of CPI Inflation Rate with:

Index Country x ( t - 3) x ( t - 2) x ( t - 1 ) x(t) x(t + l) x(t + 2) x(t + 3)

CV U°S~ . . . . . .

U.K. -.06 -.17 -.24 -.21 -.27 -.14 -.29

Italy .40 -.30 -.59 -.18 -.34 -.28 -.11
Greece -.21 -.06 .65 .50 -.10 -.30 -.26
332 IAER: AUGUST 1999, VOL. 5, NO. 3

TABLE 4 (CONT.)

Cross Correlation of CPI Inflation Rate with:

Index Country x ( t - 3) x ( t - 2) x ( t - 1) x(t) x ( t + 1) x ( t + 2) x ( t ÷ 3)

Gini U.S. .14 .11 -.18 -.14 -.06 .10 .14

U.K. .06 -.26 -.32 -.37 -.21 -.08 -.03
Italy .27 .23 .24 .09 -.26 -.37 -. 13
Greece .03 -.01 .15 .21 .06 -.20 -.25
Theil U.S. .17 -.05 -.01 -. 19 -.05 .07 .23
U.K. -- . . . . . .
Italy .39 .18 .17 -.50 -.71 -. 13 -.06
Greece -.04 -.05 .32 .35 -.05 -.30 -.34

Notes: Deviationsfrom trend are from annualdata. Samplesize for the U.S. is from 1947-89; for the United
Kingdom(U.K.), 1960-84; for Italy, 1977-89; and for Greece, 1959-93.

The current evidence should be interpreted as in the previous discussion. In the three
large economies (U.S., United Kingdom, and Italy), overall inequality exhibits a common
countercyclical behavior regarding inflation. These results can further be interpreted as
having progressive implications. On the other hand, a rather regressive behavior is
dominant in Greece. This different cyclical behavior exhibited by the inequality indices
relative to inflation can be attributed to the different market structure of the individual
economies. It is also a matter of dominance between demand-pull and cost-push inflation
or the type of income concept used. Thus, the countercyclical behavior in the U.S., the
United Kingdom, and Italy implies a possible dominance of cost-push inflation as opposed
to Greece where the demand-pull inflation may have prevailed in the periods examined.
Inequality and Unemployment
Unemployment has distributional effects as well as dead-weight losses since a
substantial amount of unemployment compatible with zero inflation is involuntary and not
optimal. The social costs of involuntary unemployment are obvious and immediate,
compared with those of inflation [Tobin, 1972]. Studies on unemployment and income
inequality assert that unemployment hurts the poor and that the impact of unemployment
is stronger than that of inflation [Blinder and Esaki, 1978; Nolan, 1989; Blank and
Blinder, 1986; Danziger and Gottschalk, 1989].
Table 5 presents the cross correlations of the unemployment rate with the inequality
indices. The results show aggregate inequality indices move procyclically with
 DIMELIS AND LIVADA: INEQUALITY AND BUSINESS CYCLES 333

unemployment for all countries except Greece. Both the U.S. and the United Kingdom
show significant contemporaneous correlations. In Italy, only the CV is significant while,
in Greece, all coefficients are insignificant. The pattern of the phase shift varies with the
cycles being synchronous in the U.S., variable in the United Kingdom, and lagging in
Italy and Greece. The disaggregate indices in Table 3 suggest that in most countries,
unemployment harms the poor and middle income classes while the top income class (and
the upper-middle class for the U.S.) gains. The exceptional behavior in Greece, where
the very poor and very rich seem to be harmed by unemployment, may be attributed
either to the nature of unemployment data or to the country's labor market structure. In
fact, the Greek unemployment series is partly estimated, whereas the Greek labor markets
are largely regulated. Finally, comparing the results of unemployment with those of
inflation, notice that the cross correlations of aggregate inequality indices and quintiles
with unemployment were consistently stronger than those of inflation in the case of the
U.S.

TABLE 5
Unemployment Rate and Income Inequality Indices by Country

Cross Correlation of Unemployment Rate with:

Index Country x(t - 3) x ( t - 2) x(t - 1 ) x(t) x(t + l) x(t + 2) x(t ÷ 3)

CV U.S. -- . . . . . .
U.K. .26 .44 .40 .33 .22 .04 .01
Italy -.33 .08 .19 .39 .40 .28 .18
Greece .02 .12 .01 -. 13 -.25 -. 14 .06
Gini U.S. -.32 -.23 -.04 .33 .23 .10 .06
U.K. .07 .18 .32 .33 .29 .14 -.09
Italy -.07 -.04 -.23 .06 .34 .24 -.07
Greece .17 .18 .07 -.06 -. 19 -.21 -.09
Theil U.S. -.24 -. 17 -.07 .32 .22 .11 .02
U.K. -- . . . . . .
Italy -.03 .01 -.14 .18 .17 .21 .22
Greece .11 .16 .04 -.11 -.23 -.16 .04

Notes: Deviations from trend are from annual data. Sample size for the U.S. is from 1947-89; for the United
Kingdom (U.K.), 1960-84; for Italy, 1977-89; and for Greece, 1959-93.
334 IAER: AUGUST 1999, VOL. 5, NO. 3

In further analyzing these results, it would be useful to refer to recent discussions in

literature about the empirical evidence of both high and persistent unemployment and
rising (wage) inequality in OECD countries. In particular, there is the view that wage
inequality has been the crucial factor that explains the better employment performance in
the U.S. [European Commission, 1996]. This assertion of a trade-off between persistent
unemployment and inequality was based on specific regulatory institutions in labor
markets across nations. Thus, the U.S., one of the most unregulated labor markets, kept
unemployment low at the expense of higher wage inequality as a result of the downward
wage flexibility. In contrast, the more rigid labor market institutions in the EU have
prevented wages of low-skilled workers from falling but have caused unemployment to
persist. However, more recent evidence casts doubts on this argument as in the case of
the United Kingdom which has experienced the largest rise in inequality and
unemployment despite the high degree of unregulation. Blank [1995] extensively reviews
existing literature on these issues and discusses policy issues as well. The results in this
paper for the U.S., the United Kingdom, and Italy are also against this assertion since the
cyclical relation between unemployment and income inequality was found to be positive
and significant. The fact that Greece has been one of the most highly regulated labor
markets may partly explain the lack of correlation found in the data.

Conclusions

This paper studies the business cycle properties of some aggregate and disaggregate
income inequality indices for the U.S. and three EU countries (United Kingdom, Italy and
Greece). Following the Kydland-Prescott methodology, the main stylized facts regarding
business cycles were obtained and compared among countries. These findings suggest that
some regularities can be observed in the cyclical pattern of inequality indices in these
countries.
At the aggregate level, inequality indices move countercyclically with real per capita
GDP in the U.S. and the United Kingdom. The evidence is mixed for Italy while a
procyclical behavior prevailed in Greece. A better picture emerged from the disaggregate
indices which support the above findings. That is, during expansions, the lower income
classes benefit in the U.S. and the United Kingdom while, in Greece and Italy, they either
lose or are unaffected. On the other hand, the graphical examination of the cyclical
patterns and the Kuznets curves suggest that breaks exist in the above patterns which are
possibly associated with economic, political, or institutional reforms, specific for each
country.
Regarding the comovements with the other macroeconomic variables, a common
pattern characterizes the three large economies (U.S., United Kingdom, and Italy). Thus,
inequality indices showed a countercyclical behavior with both inflation and
unemployment in these countries. On the other hand, the small economy of Greece, with
highly regulated labor markets, exhibited a procyclical behavior with respect to inflation
and no correlation with unemployment. Also, in most countries, the top income group
seems to lose at the benefit of the rest during inflationary periods while, in all four
 DIMELIS AND LIVADA: INEQUALITY AND BUSINESS CYCLES 335

countries, the poor gain from inflation and suffer from unemployment. It is interesting
that in this study, the observed correlations between cyclical inequality and inflation were
quite strong in some cases as opposed to previous studies that showed weak or no such
correlation. In addition, in the case of the U.S., the cross correlations of aggregate
inequality indices and quintiles with unemployment were consistently stronger than those
of inflation.
These results are helpful for formulating and calibrating business cycle models.
Interesting extensions of this study would be, first, to combine the time series evidence
with a cross-sectional panel of countries and, second, to examine the above business cycle
features in association with the socioeconomic, historical, and political structure of each
economy. Hence, further theoretical and empirical work concerning the issues of income
distribution, macroeconomic behavior, and politics should be undertaken toward this
direction.

Footnotes

1) Classical business cycles refer to the much quoted definition of business cycles first given by
Wesley Mitchell in 1927 (and revised by the same author [Bums and Mitchell, 1946]) as
sequences of expansions and contractions of the level variables with an emphasis on turning
points and phases. For an extensive survey on various phases of business cycle theories, see
Zarnowitz [1991].
2) For a short explanation and application of the HP filter, see the EuropeanEconomy [1995].
3) See the special issue of the Journal of Econometrics [1990] for a representative sample of new
methods for analyzing and measuring economic inequality with emphasis on estimation and
modeling.
4) In the case of U.S. where the sample of the examined series is the largest (43 observations),
a value of the correlation coefficient higher than 0.29 is statistically significant at the 5 percent
level according to the two-sided t-statistic.
5) Figure 2 refers to the Gini coefficient, but a similar picture emerged from the other indices as
well.
6) Although Kuznets hypothesis refers to the time relationship between economic development
(as measured by the level of income) and income inequality, most of the existing literature is
dominated by cross-sectional studies, a large number of which confirms it while others refute
it [EuropeanEconomy, 1995].
7) It has also been argued [Shorrocks and Marlin, 1982] that the distributional effects of inflation
depend not only on the change in money income due to inflation, but also on changes in the
relative prices during inflation. That is, since there are different expenditure patterns for
different income groups (or different households), there are different price indexes for these
groups as well.

References

Achdut, Lea. "Income Inequality, Income Composition and Macroeconomic Trends: Israel,
1979-93,"Economica, 63, 1996, pp. S1-$27.
336 IAER: AUGUST 1999, VOL. 5, NO. 3

Alesina, Alberto; Rodrik, Dani. "Distributive Politics and Economic Growth," Quarterly Journal
of Economics, 109, 1994, pp. 465-90.
Anand, Sudhir; Kanbur, S. M. R. "Inequality and Development," Journal of Development
Economics, 41, 1993, pp. 19-43.
Atkinson, Anthony B. "What is Happening to the Distribution of Income in the UK?,"Welfare State
Program Discussion Paper 87, Suntory-Toyota International Centre for Economics and Related
Disciplines, London School of Economics, 1993.
Atkinson, Anthony B.; Micklewright, John. Economic Transformation in Eastern Europe and the
Distribution of Income, Cambridge, MA: Cambridge University Press, 1992.
Beach, Charles M. "Cyclical Sensitivity of Aggregate Income Inequality, "Review of Economics and
Statistics, 59, 1, 1977, pp. 56-66.
Bjorklund, Anders. "Unemploymentand Income Distribution: Time-Series Evidence from Sweden,"
The Scandinavian Journal of Economics, 93, 3, 1991, pp. 457-65.
Blank, Rebecca M. "Changes in Inequality and Unemployment over the 1980s: Comparative
Cross-National Responses," Journal of Population Economics, 1995, pp. 1-21.
Blank, Rebecca M.; Blinder, Alan S. "Macroeconomics, Income Distribution and Poverty," in S.
Danziger; D. H. Weinberg, eds., Fighting Poverty, Cambridge, MA: Harvard University Press,
1986.
Blinder, Alan S.; Esaki, H. Y. "Macroeconomic Activity and Income Distribution in the Postwar
United States," Review of Economics and Statistics, 60, 4, 1978, pp. 604-9.
Brandolini, Andrea. "A Description and an Assessment of the Sample Surveys on the Personal
Distribution of Incomes in Italy," Servizio Studi, Rome, Italy: Banca d'Italia, 1992.
Brandolini, Andrea; Cannari, L. "Methodological Appendix: The Bank of Italy's Survey of
Household Income and Wealth," in A. Ando; L. Guiso; I. Visco, eds., Saving and the
Accumulation of Wealth. Essays on Italian Households and Government Saving Behavior,
Cambridge, MA: Cambridge University Press, 1994, pp. 369-86.
Brandolini, Andrea; Sestito, Paolo. "Cyclical and Trend Changes in Inequality in Italy, 1977-1991,"
Servizio Studi, Rome, Italy: Banca d'Italia, 1994.
Burns, Arthur F.; Mitchell, Wesley C. Measuring Business Cycles, Cambridge, MA: National
Bureau of Economic Research, 1946.
Buse, Adolf. "The Cyclical Behavior of Size Distribution in Canada: 1947-78," Canadian Journal
Economics, 15, 2, 1982, pp. 189-204.
Campano, Fred; Salvatore, Dominick. "Economic Development, Income Inequality and Kuznets
U-Shaped Hypothesis," Journal of Policy Modeling, 10, 2, 1988, pp. 265-80.
Chenery, M.; Ahluwalia, M.; Bell, C. L. G.; Duly, J.; Jolly, R. Redistribution with Growth,
Oxford, United Kingdom: Oxford University Press, 1974.
Clarke, George R. G. "More Evidence on Income Distribution and Growth," Journal of
Development Economics, 47, t995, pp. 403-27.
Cowell, Frank A. Measuring Inequality, Oxford, United Kingdom: Phillip Allan, 1977.
Creamer, D. Personal Income During Business Cycles, Princeton, NJ: Princeton University Press,
1956.
Danziger, Sheldon; Gottschalk, Peter. "Increasing Inequality in the United States: What We Know
and What We Don't," Journal of Post Keynesian Economics, 11, 2, 1989, pp. 174-95.
Dimelis, Sophia; Livada, Alexandra. "Business Cycles and Income Inequality in Greece," Cyprus
Journal of Economics, 8, 1995, pp. 23-40.
European Commission. "Labor Costs, Wages, and Employment in the EU," II/482-EN, 1996.
Galor, Oded; Zeira, Joseph. "Income Distribution and Macroeconomics," Review of Economic
Studies, 60, 1, 1993, pp. 35-52.
 DIMELIS AND LIVADA: INEQUALITY AND BUSINESS CYCLES 337

Gottschalk, Peter. "Inequality, Income Growth and Mobility: The Basic Facts,"Journal of Economic
Perspectives, 11, 2, 1997, pp. 21-40.
Hayes, Kathy; Slottje, D. J.; Porter-Hudak, Susan; Scully, Gerard. "Is the Size Distribution of
Income a Random Walk?," Journal of Econometrics, 43, 1990, pp. 213-26.
Hodrick, Robert J.; Prescott, Edward C. "Post-War U.S. Business Cycles: An Empirical
Investigation," manuscript, Carnegie-Mellon University, 1980.
Jenkins, Stephen P. "Accounting for Inequality Trends: Decomposition Analyses for the U.K.,
1971-86," Economica, 62, 1995, pp. 29-63.
Johnson, George E. "Changes in Earnings Inequality: The Role of Demand Shifts," Journal of
Economic Perspectives, 11, 2, 1997, pp. 41-54.
Johnson, Paul; Webb, Steven. "Explaining the Growth in U.K. Income Inequality: t 979-1988, "The
Economic Journal, 103, 1993, pp. 429-35.
Journal of Econometrics. Special issue, 42-43, 1990.
Kuznets, Simon. "Economic Growth and Income Inequality, "American Economic Review, 45, 1,
1955, pp. 1-28.
. "Shares of Upper Income Groups in Income and Savings," Cambridge, MA: National Bureau
of Economic Research, 1953.
Kydland, Finn E.; Prescott, Edward C. "Business Cycles: Real Facts and a Monetary Myth,"
Federal Reserve Bank of Minneapolis Quarterly Review, 14, 2, 1990, pp. 3-18.
Levy, Frank; Murnane, Richard J. "U.S. Earnings Levels and Earnings Inequality: A Review of
Recent Trends and Proposed Explanations," Journal of Economic Literature, 30, 1992, pp.
1333-81.
Lindert, Peter H.; Williamson, Jeffrey G. "Growth, Equality, and History," Explorations in
Economic History, 22, 1985, pp. 341-77.
Livada, Alexandra. "Income Inequality in Greece: A Statistical and Econometric Analysis," Oxford
Bulletin of Economics and Statistics, 53, 1991, pp. 69-82.
Lucas, Robert E., Jr. "Understanding Business Cycles," in K. Brunner; A. H. Meltzer, eds.,
Stabilization of the Domestic and International Economy, Carnegie-Rocbester Conference Series
on Public Policy 5, Amsterdam, Netherlands: North-Holland, 1977.
Mirer, Thad W. "The Effects of Macroeconomic Fluctuations on the Distribution of Income,"
Review of Income and Wealth, 19, t973a, pp. 385-406.
. "The Distributional Impact of the 1970 Recession," Review of Economics and Statistics, 55,
1973b, pp. 214-24.
Nolan, Brian. "Macroeconomic Conditions and the Distribution of Income: Evidence from the United
Kingdom," Journal of Post Keynesian Economics, 11, 2, 1989, pp. 196-221.
Organization for Economic Cooperation and Development. "Earnings Inequality, Low-Paid
Employment and Earnings Mobility," Employment Outlook, 1996, pp. 59-108.
Perotti, Roberto. "Political Equilibrium, Income Distribution, and Growth, "Review of Economic
Studies, 60, 1993, pp. 755-76.
Persson, Torsten; Tabellini, Guido. "Is Inequality Harmful for Growth?," American Economic
Review, 84, 1994, pp. 600-21.
Ram, Raft. "Economic Development and Income Inequality: Further Evidence on the U-Curve
Hypothesis," World Development, 16, 1988, pp. 1371-6.
Schultz, T. Paul. "Secular Trends and Cyclical Behavior of Income Distribution in the United States:
1944-1965," in Lee Soltow, ed., Six Papers on the Size Distribution of Wealth and Income,
Cambridge, MA: National Bureau of Economic Research, 1969, pp. 75-100.
Sen, Amartya K. On Economic Inequality, London, United Kingdom: Clarendon Press, 1973.
 338 IAER: AUGUST 1999, VOL. 5, NO. 3

Shorrocks, Anthony F.; Marlin, E. "Inflation and Low Incomes," Canadian Public Policy, 1982,
pp. 334-6.
Slottje, Daniel J.; Basmann, R. L.; Nieswiadomy, Michael L. "On the Empirical Relationship
Between Several Well-Known Inequality Measures," Journal of Econometrics, 42, 1, 1989.
Tobin, James. "Inflation and Unemp 1oyment, "American
" Economic Review, 62, 1, 1972, pp. 1-18.
van Wijck, P. "Cyclical and Political Influences on the Size Distribution of Income in the U.S.,
1959-1989," Applied Economics, 24, 1992, pp. 169-73.
Zarnowitz, Victor. "What is a Business Cycle?," Working Paper No. 3863, Cambridge, MA:
National Bureau of Economic Research, 1991.

View publication stats