You are on page 1of 28

AN INDEX OF DEMOCRATIC

POLITICAL STABILITY:
A Methodological Note

LEON HURWITZ
Cleveland State University

democracies of the world present varied picture of differing


The a

structural forms and attributes. One such attribute is the of concept


&dquo;political stability&dquo; and this paper is concerned with a discussion and
analysis of the &dquo;stability&dquo; performance of selected democratic polities. A
study of stability has relevance and importance for several reasons. It may
help in understanding how and why some democracies have been able to
remain viable and successful-able to adapt to a changing environment
without losing essential forms and patterns-and why other countries have
been less fortunate in coping with the exigencies of change. This adaption
to the environment, this change without alienation, is perhaps the most
relevant concept in politics today and it is readily apparent that several
countries are not currently achieving this end. The young especially are
estranged from the democratic process and the continuance of specific
systems may be open to question. A study of political stability, then, will
identify the location of this dissatisfaction-this failure of the democratic
system to adapt-how it is manifested, and, perhaps, an explanation of its
existence.

AUTHOR’S NOTE: An earlier version of this paper was delivered at the 1970 Annual
Meeting of the Ohio Association of Economists and Political Scientists, Akron, Ohio,
April .?-~,
~4pn/ 3-4, 7970.
1970. 7 would M~e
I wouM like ~o
to ~c/t~
thank 7~’o/~Mor
Professor Yo~M
John ~Vo~e
Nagle o/~rccMse
of Syracuse MfHferN<y
University
for his assistance throughout the research and writing of this paper.

[41]
[42]

The purpose of this paper is to bring some cohesion and meaning to the
concept of democratic political stability. Although quite hidden, obscured,
and often submerged in confusion, there are several identifiable strands of
agreement in most of the works devoted to political stability. A composite
view from the relevant literature concludes that when one speaks of the
&dquo;stability&dquo; of a certain democratic polity, one is referring to (a) the ability
of the political system to persist (among those authors who stress
&dquo;endurance&dquo; or &dquo;persistence&dquo; are Banks and Textor, 1963; Blondel, 1968;
Eckstein, 1966a and 1966b; Goldberg, 1968; Lijphart, 1968; Upset, 1959;
Russett et al., 1964); (b) the existence of a legitimate political system
(stressed by Almond and Verba, 1965; Bwy, 1968; Eckstein, 1966a and
1966b; Lijphart, 1968; Lipset, 1959; Needler, 1968); and (c) the presence
of effective decision-making by the political system (the following are
relied upon for the concept of &dquo;effectiveness&dquo;: Almond and Verba, 1965;
Eckstein, 1966a and 1966b; Goldberg, 1968; Lijphart, 1968, Lipset, 1959;
Russett et al., 1964) Most of the scholars cited do agree that democratic
political stability means persistence, legitimacy, and effectiveness, but
confusion abounds due to the lack of consensus as to the meaning of these
latter concepts. The concept of &dquo;legitimacy,&dquo; for example, frequently
means all things as various individuals attempt to measure the degree of

legitimacy present in their particular political system. Many of the


empirical and quasi-empirical studies cited were measuring disparate
phenomena when attention was directed toward the legitimacy of a
system. To state that stable democracies are persistent, legitimate, and
effective only begs the question-what needs to be examined are these
latter concepts. This paper attempts to transform these abstract concepts
into valid operational definitions and observable phenomena. A coherent
set of twenty democratic countries is reviewed from the end of World War
II to January 1, 1969, in order to construct an empirical measure of
democratic political stability.22
The proposed index of democratic political stability can be employed
for a variety of purposes. One could first offer statements about the
relative position of any one country by comparing its stability score with
the scores of other countries. The index could also be utilized to examine
the empirical relationship between democratic political stability and
several hypotheses which have traditionally been offered to &dquo;explain&dquo; the
observed variations in stability levels. Judicious use of correlation analyses
would be able to identify the degree of relationship between some of these
traditional explanations of stability (level and rate of economic develop-
twnt, the degree of Roman Catholicism, cultural homogeneity, multi-
j
[43]

partism, and degree of democratic performance) and the proposed measure


of stability. Although the proposed index is well-suited for further
research in cross-national comparison, such correlation analyses are not
included in this exploratory paper. The present analysis is limited to the
construction of the stability index and a discussion of the correlates of
stability must await further research.
The Index of Democratic Political Stability is composed of two basic
indicators: a Persistence of Pattern Index and an Index of Legitimacy//
System Support. The Persistence of Pattern Index is itself a composite
index, the component parts being an Index of Government Persistence and
an Index of Legislative Agreement/Election Frequency. Data are collected
for each country on each indicator and, by use of a well-adapted statistical
3
operation-T-scoring-the final index is constructed.3

THE PERSISTENCE OF PATTERN INDEX

The concept of &dquo;endurance&dquo; was a strand of common agreement in


most of the works cited above. &dquo;Durability&dquo; or &dquo;longevity&dquo; is perhaps the
basic component in any discussion of political stability. The first
component of the Persistence of Pattern Index is presented as an Index of
Government Persistence but not from the viewpoint of mere longevity or
survival of specific governments. This devotion to survival has been a major
characteristic of previous studies concerned with political stability. Bruce
Russett’s (Russett et al., 1964: Table 30, pp. 101-104) index is perhaps the
best-known measure of political stability, and his index gives the rate of
turnover in office of the legally designated chief executive. The index,
however, as Russett readily admits, is far from perfect. Each and every
change in the chief executive represents evidence of instability even
though the change may have been a result of persistence of pattern
elements (general parliamentary elections), neutral elements (accidental
death of the prime minister), or from more unstable elements (assassina-
tion of the chief executive). Russett’s index thus reduces the concept of
stability to a monomeasure of longevity. Jean Blondel (1968: 190-191)
approaches stability in a like manner. Stability is reduced to a specific
government’s duration and this measure, as is Russett’s, is presented as
years per government. These measures are rejected, for they reduce
stability to endurance. Persistence is seen to be the government’s ability to
hold on, longevity, perhaps even stubborness. All changes in the chief
executive or the cabinet were seen as evidence of instability and by so
[44]

doing, Russett and Blondel are unable to differentiate among the very real
and nontheoretical types of government change in democratic polities.
For the purposes of this paper, we shall employ &dquo;chief executive/
cabinet/government endurance&dquo; as one component of the Persistence of
Pattern Index but in a manner that differs greatly from the above studies.
This is done in order to avoid the problem of equating all incidences of
change with instability; this is accomplished by positing that some types of
governmental change are more stable (more reflective of the persistence of
the democratic pattern) than other changes. The Index of Government
Persistence is the result of the following expression:
number of total persistence
points
number of different governments = Index of Government Persistence

where :
Persistence Points = weighted score (range is 0 to 1) assigned to each type
of government change. Each different government in the 20 countries
reviewed (260 total governments) is assigned a weighted score which is
determined by the circumstances causing the new government’s formation.
This approach will enable one to differentiate among the various types of
government change and the actual duration of the government (expressed
in units of time) is not relevant.
Government = an administration which meets any one of the following
criteria: (a) post-election formation; (b) change in the prime minister; (c)
change in the party composition of the cabinet; and (d) an interelection
resignation/reformation with the same prime minister and cabinet com-
position. Government in nonparliamentary systems (United States) is
defined as the President, and the President is considered as being identical
to the prime minister in a parliamentary system.
Maximum Score = 1.00, denoting high persistence of pattern.
Minimum Score approaches 0, denoting low persistence of pattern.
The differentiated types of government formation and the assigned
persistence points are presented below in descending order on the arbitrary
interval scale 4

Type I. Postelection Formation,


Composition of Cabinet Not Relevant
~

Elections are considered an integral component of persistence of the


-

democratic pattern. What else can one mean by democracy if one does not
require some sort of citizen participation in the determination of elite
.composition? New governments resulting from the population’s mandate
[45]

reflect stability in the sense that the change is determined by institu-


tionalized, constitutional, and expected practices. The democratic pattern
of choosing the governors is maintained, and the resultant governments are
thus considered to be formed in a stable, democratic manner. The actual
composition of the newly formed government relative to the adminis-
tration immediately preceding it is not relevant. Labour’s 1964 electoral
victory in the United Kingdom may have represented a sense of
dissatisfaction by the population with the Conservatives, but Labour’s
accession to office was not the result of dysfunctional elements in British
society. The same example can be seen in the United States: Presidents are
elected, and it is this that one means, in part, by persistence of the
democratic pattern. Postelection formation is thus assigned the maximum
persistence score of 1.00. Only one country in the present analysis, New
Zealand, has had all government formations classified as Type I.

Type II. Interelection Change in Prime Minister Due to


Accidental Death but not from Assassination,
Composition of Cabinet Not Relevant
This type of change is not due to a resignation (voluntary or forced)
but to the death of the prime minister while in office. Such deaths (Holt in
Australia, Dupong in Luxembourg, Hedtoft and Hansen in Denmark) are
relatively neutral occurrences when applied to the persistence/nonper-
sistence scale. The deaths are not politically relevant for the accidental
nature of the death is unconnected with politics (except, of course, that
the death concerned a political person). Accidental deaths do not reflect
stability or instability, but since there have been governments formed from
this cause, a persistence score must of necessity be assigned. The
&dquo;nonpolitical&dquo; act of death receives the maximum value of 1.00 on the
proposed index. Deaths are neutral, but are posited to reflect maximum
stability if accidental in cause.s

Type III. Interelection Addition of Party to Cabinet,


Prime Minister Remains Constant

This type of formation also receives the maximum value of 1.00. Such
changes reflect persistence in the sense that the democratic process is
functioning: consensus has been broadened, the scope of representation in
the decision-making process has increased, and greater participation has
resulted. Parties that were previously in opposition or nonrepresented
[46]

support but then included in the government do not present evidence of


instability within the democratic system. In our view, such additions
strengthen the persistence of the democratic pattern. Very few govern-
ments are, however, classified as Type III, and it appears that this method
of consensus formation is not too prevalent in the twenty democracies
under review. There are several reasons for this: many countries operate
under the two-party system, and this type of formation would be almost
unheard of except in national emergencies (an all-party coalition to face an
external threat); other systems are so polarized that all acceptable
coalition partners are already in the government; and, perhaps, the
unwillingness of the governing coalition to admit extra but unnecessary
members to the coalition (the theory of the least or minimal winning
coalition-fifty percent plus one).

Type IV. Interelection Resignation/Reformation,


Prime Minister and Parties Remain Constant or
Additional Party Joins Government

This classification of government formation is assigned a persistence


score of 0.50. The fact that the government resigned leads one to assume
that some sort of political crisis transpired and, therefore, that there is
some degree of political instability in the system. Consensus among the

political participants is reduced and this type shows less persistence than
Types I-III. Since, however, the prime minister and parties in the reformed
administration remain identical to those of the previous administration,
one can posit that consensus among the political actors did not completely
break down. That there was a political crisis (indicating instability)
requiring the government’s resignation is not questioned, but it is evident
that such crises are not severe enough to warrant a change in the prime
minister or a party’s withdrawal from the cabinet. Type IV formations are
readily explained: governments reform with identical personnel because
there is a lack of acceptable coalition partners who are not already
represented in the governing coalition. There is no other choice but to
reform with the same personnel once the decision to resign is made. A
brief look at some specific instances of Type IV formations serve as an
illustration. Of the seventeen Type IV formations, ten are in countries that
had no acceptable coalition partners outside the actual governing
coalition: Israel (4), Italy (4), and Austria (2).
[47]

Type V. Interelection Change in Prime Minister


for any Reason Except Death, Parties in Cabinet
Remain Constant or Additional

This type of government formation is assigned 0.25 persistence points.


The prime minister either resigned voluntarily on account of illness or
for other reasons or the resignation was not voluntary (a loss of a vote of
confidence). In both cases, the fact of the resignation leads one to assume
that there is some degree of instability in the system. There are valid
reasons to justify regarding all such changes as evidence of a political crisis.
If the resignation is a forced one, and is reported as such, the change is due
to a crisis by definition. If the change is reported as a voluntary resignation
or is ambiguous, there is still no other way to regard the resignation except
as &dquo;nonvoluntary.&dquo; How voluntary was the resignation? Did the prime
minister graciously step down or was perceived nonsupport present? How
&dquo;sick&dquo; was an ill prime minister? Was &dquo;illness&dquo; the real reason for the
resignation or was it &dquo;political illness&dquo; contracted from inability to
govern?6 Since this type of change excludes accidental death and
assassination, it is posited that all other interelection change of the prime
minister reflects some degree of nonpersistence.’ The change of the prime
minister, for whatever reason allowed, causes Type V to be less persistent
than Type IV. Although the latter type also experienced a resignation, the
political crisis is not considered so acute as to require a change in the
prime minister. A change in the prime minister is usually regarded as
indicating that a political crisis 8has taken place (executive longevity/
endurance) and is so regarded here .8

Type VI. Interelection Withdrawal of Party from the Government,


Prime Minister Remains Constant

This category of government formation reflects approximately as much


nonpersistence as Type III (party addition) reflects persistence and for
reasons converse to those of Type III.
Type VI changes are seen as evidence
of lesspersistence in thethat the
sense democratic process is not achieving
some posited major goals: consensus among the parties is weakened, the
area of representation in the decision-making process has decreased, and
reduced participation is the result. Type VI are assigned 0.25 persistence
points, an equal value assigned to Type V (change in prime minister). This
equality is based on the conception of what constitutes a new govern-
ment.99 It will be recalled that both the prime minister and party
[48]

representation helped to determine the criteria for a new administration. A


party withdrawal from the government without a concurrent change in the
prime minister is equated with a change in the prime minister without a
loss of party support.

Type VII. Interelection Change in Prime Minister


with a Simultaneous Withdrawal by a Party from the Government

This type of government formation is considered to be less persistent


than Types V and VI, for the present category has almost complete per-
sonnel turnover. The explanations offered above in Types V and VI as to
their tendencies toward instability are valid here but to a greater degree.
Type VII formations are assigned the minimum possible persistence score
of 0.00. This value is due to the severe elements of nonpersistence
present in this category of government formation when compared with
other types.

Type VIII. Change in the Prime Minister Due to Assassination,


Parties in the Cabinet Not Relevant

The assassination of a prime minister is not regarded as a change under


Type II (accidental death) even though, semantically, an assassination is an
accidental death. The formation of a new administration due to the
assassination of the prime minister is seen as evidence of dysfunctional
elements in the system and thus as evidence of extreme political instability
(nonpersistence of the democratic pattern). Type VIII formations are
assigned the minimum persistence score of 0.00. The assigned value of
0.00 is due to this writer’s conception of the severe dysfunctional elements
present in a society where the prime minister is assassinated. It is the
denial of the democratic process to determine policies and elites in this
fashion. The example of the United States in 1963 is instructive: the new
Johnson administration was not formed in terms of settled social and
political conditions, nor did it evidence, in Eckstein’s words, &dquo;a capacity
for adapting to changing conditions, for realizing political aspirations and
holding fast allegiances.&dquo; On the contrary, the new government was
formed from nonpersistence elements: violence and noninstitutionalized
practices.
The Index of Government Persistence, by considering those changes
from elections as evidence of persistence of pattern, is open to
resulting
[49]

question on both theoretical and empirical grounds. There is an instance


(New Zealand) where all government formations are due to general

elections, and thus New Zealand is ranked at the maximum possible


stability score. The criticism of the Government Persistence Index is that it
ranks country X at maximum score provided all government formations
are Type I (postelection). Since the frequency of such changes is far larger
than usual constitutional practices would warrant, this criticism has to be
met. Such a task leads to the second component of the Persistence of
Pattern Index. This indicator, the Legislative Agreement/Election Fre-
quency Index, serves a double function: (a) to compensate for any skewed
results which appear in the Government Persistence Index, and (b) as an
equal component with the Government Persistence Index in the calcula-
tion of the Persistence of Pattern Index.
It is posited that general parliamentary elections are necessary although
not sufficient evidence of the persistence of the democratic pattern.
Absence of such elections, therefore, must be assumed to denote
nonpresence of the democratic pattern. But the obverse is not true, that is,
that the greater frequency of elections is evidence of greater persistence of
pattern. In other words, it is not valid to state that the more elections a
country has, the more stable it will be. On the contrary-the more
elections a country experiences after a certain point will introduce
dysfunctional elements into the system and will be evidence of less
stability. This assumption is made after consideration of the nature of a
democratic society. General parliamentary elections are eufunctional in ~
the sense that they provide for institutionalized interest articulation and.
aggregation, elite replacement, support and legitimization of the regim -
and system, and, most basic to a democracy, consent of the governed. B~
its logical extreme-constant and recurring elections-moves a society aw~
from stable democracy and toward a more unsettled polity.
The above remarks are valid only in relation to each society’s perceiv
optimum number of elections per specific time span. Elections beyond this
societal perceived optimum are considered to be evidence of less stability.
Each country in the present study has a standard, a norm, in regard to its
election frequency. Only when this norm is overstepped are dysfunctional
elements introduced: constant politicalization of the society, rapid rather
than gradual elite replacement, and abrupt reversals rather than continuity
in output would be the most likely result once this norm is passed.
Constant democracy, however genuine or authentic it may be, is posited to
be less stable than a polity where the perceived norm is observed.
The basic approach to this indicator rests upon institutional prescrip-
tions and the level of agreement between actual practice and these formal
[50]

provisions. The index examines each country’s constitutional provisions


regarding the maximum time each parliament may sit, and compares this
maximum time to the actual duration of parliament. Since the number of
parliaments is determined by the number of elections (a new parliament is
formed after each election), the Legislative Agreement Index will reflect
the number of elections held. Such a procedure will allow each country to
individually determine its standard as to how many elections are &dquo;normal&dquo;
for a given time span. One election per constitutional time span is
represented as maximum stability and each country’s score is calculated in
relation to this maximum. Those countries with more elections than
required are not classified as &dquo;unstable,&dquo; but only as &dquo;less stable&dquo; than a
country in which actual practice agrees with formal provisions. This index
is the result of the following expression:

time in country study


number of legislatures x constitutional tenure =
Index of Legislative Agreement/Election Frequency
where:
Time in Country Study number of days between the first general
=

parliamentary election in the postwar period and the last such election
prior to January 1, 1969. Range is 5,231 days (Japan) to 8,410 days
(Canada).
Number of Legislatures number of general parliamentary elections minus
=

one (1). Range is 4 (Germany and Italy) to 9 (Denmark).

Constitutional Tenure formal constitutional provision, expressed in days,


=

relating to the maximum allowable time span between general elections.


Maximum Agreement 1.00, denoting high persistence of pattern.
=

Minimum Agreement approaches 0.00, denoting low persistence of


pattern.

THE INDEX OF LEGITIMACY/SYSTEM SUPPORT


The second component of the proposed view of democratic political
stability was stated to be &dquo;legitimacy.&dquo; That is to say, in order to be
classified as &dquo;stable,&dquo; a democracy must be legitimate as well as persistent.
For the purposes of this paper, &dquo;legitimacy&dquo; is regarded as the acceptance
by the population of the democratic political system. Legitimacy is thus
seen as &dquo;system support&dquo; or &dquo;system affect.&dquo;! 0 The present task is to
construct a measure that will identify the degree of such system support
[51]

or, conversely, the degree of antisystem support existing in the twenty


polities under review.
The immediate question is how this support or nonsupport is to be
identified. Here we are concerned with the articulation and quantification
of a certain interest (system support). System support can be articulated
by various methods in a democracy, and one such method is through
inputs to the system as evidenced by specific voting behavior. A fruitful
approach, although a limited one because it is not valid to make inferences
about what individuals have in their minds when the electorate votes, is to
study system support through voting behavior. Since interest articulation
is given a structured expression when directed at political parties, we can
regard political parties as representing the interests of the population
vis-a-vis the political system. In is thus posited that the degree of pro- or
antisystem support in a democracy (legitimacy) can be measured by the
degree of support (votes) given to the political parties, according to
whether the specific party voted for is classified as a pro- or antisystem
party. The Index of Legitimacy/System Support is the result of the
following expression:
average number of votes per election given to system parties -

average number of total valid votes cast per election


Index of Legitimacy/System Support
where:
Time of Study time between the first general parliamentary election in
=

the postwar period and the last such election prior to January 1, 1969.
Election = general parliamentary elections. This is defined as an election
for a national, permanent, viable government. By-elections and elections
for constituent assemblies and provisional governments are excluded.
Elections refer to the lower house of parliament if there is more than one
house.
Political Party = Leon Epstein’s (1967: 9) criterion is borrowed to
determine what is and what is not a political party: &dquo;any group, however
loosely organized, seeking to elect government office-holders under a given
label.&dquo;
Total Vote =
the number of actual valid votes cast. The nonvoting
electorate as well as spoiled and blank ballots are excluded.&dquo;I
System Parties = all political parties not included under any of the
following classifications: Communist, neo-Fascist, monarchist, separatist,
federalist, minor, or splinter.
Maximum Score = 1.00, denoting high legitimacy/system support.
Minimum Score = 0.00, denoting low legitimacy/system support.
[52]

CLASSIFICATION OF ANTI-SYSTEM PARTIES

Type I. Left-Right Extremes


Parties placed in this category are operationalized as the radical left and
the intransigent right or as Communist and (neo-)Fascist parties. This view
relating to the anti-systemic tendencies of the left-right extremes in a
democracy is shared by several writers.12All Communist and (neo-)Fascist
parties are thus considered antisystem parties, and the votes given to these
extremes are regarded as antisystem votes in the final calculations.
Examples of right extreme are the German National Democrats and the
Italian Social Movement.

Type II. Monarchist


Those parties which are in favor of maintaining the monarch in a
constitutional monarchy (Great Britain) are not classified as Type II
parties. Only those parties which support the reinstitution of an effective
monarch in a democracy or those parties which desire a greater role for the
monarch in a constitutional monarchy are so classified. Type II parties are
considered antisystem, for it is posited that such parties are unable and
unwilling to support the democratic republic in its entirety. The ideology
which supports the king in any one of the twenty democracies is
antisystem almost by definition: the concept of democracy excludes the
monarch from effective decision-making (unless the king is elected every
four years by universal suffrage), and the determination of elites is to be
based on universal and not ascriptive criteria. the PDIUM in Italy is an
example of Type II.

Type III. Separatist-Federalist


This category is operationalized as those parties which demand an
autonomous region within the national state (separatist) or which favor
transforming a unity state into a federal union (federalist). Such parties are
classified as antisystem because of the very nature of their ideology and
goals relative to the existing constitutional order. These parties seek the
reconstitution of the existing state to form entities congruent with the
particular views (separate areas for ethnic minorities, for example). These
parties do not accept the current formal institutional framework, and
votes given to Type III parties are seen as antisystem votes. Examples are
the Scottish Nationalists in Great Britain (separatist) and the Flemish
Nationalists in Belgium (federalist).
[53]

Type IV. Minor-Splinter


Minor parties are defined as those parties which do not receive
votes to win at least one seat in the national
enough
parliament. Splinter parties
are defined those parties which win less than two
as
percent of the seats
and are notrepresented in the governing coalition. Parties with less than
two percent of the seats but also
represented in the cabinet are not seen as
splinter parties but as small parties acceptable to the system
coalition partners). Type IV parties are (acceptable
regarded as antisystem by Hans
Daalder and Arend Lijphart, for the
inability or unwillingness of this
segment of the population to direct support toward a major
the logic of segmentation to extremes and
party carries
injects dysfunctional elements
into the society (Daalder, 1966: 225;
Lijphart, 1968: 167-177). Examples
of Type IV parties are the Australian
Democratic-Queensland Labor
(minor) and the Belgium Democratic Union (splinter).

THE INDEX OF EFFECTIVENESS


The third and final component of the
proposed index of democratic
political stabilityis identified as &dquo;effectiveness.&dquo; A stable
democracy must
not only persist and be legitimate, but also exhibit a minimum of the
tendencies of a roi faindant. This writer shall borrow the view
offered by
Almond and Verba as to the basic
meaning of the concept,13 The term is
regarded here as signifying acceptance of and satisfaction with
mental outputs by the population, and thus effective govern-
action or output is
posited to be acceptable or satisfying action as determined
by the
population. Conversely, ineffective action is unacceptable decisional
output. This view of the concept allows one to classify &dquo;no action&dquo;
(whether due to immobilism or voluntary design) as &dquo;effective action&dquo;
provided the population is satisfied with the output (or
for these reasons that some current measures of nonoutput). It is
effectiveness cannot be
utilized. 14 These measures relate &dquo;positive action&dquo; to &dquo;effective action&dquo;
but overlook the possibility that measured
output may not necessarily be
effective in the sense of satisfying action. The
point is that specific action,
even though very
impressive in amount and success, may be unsatisfactory
to the population and thus would be
regarded as &dquo;ineffective.&dquo;1$S
The present task is to operationalize the
concept into an observable
indicator in order to employ effectiveness as a
component in the final
stability equation. Since the emphasis has been placed upon
acceptable
[54]

action rather than upon the action itself, it appears fruitful to approach
the measure from the viewpoint of the reaction to the relatively
unimportant output. There are two levels of analysis with this approach.
First is &dquo;systemic effectiveness&dquo;-to what extent does the population
regard the political system itself as satisfying? Second is &dquo;policy/personnel
effectiveness&dquo;-how acceptable is the concrete output (or nonoutput) and
the people who determine this output? It is evident, however, that such
questions have already been dealt with in the previous Indices of
Persistence of Pattern and Legitimacy/System Support.
Legitimacy was operationalized as the degree of system support as
measured by voting behavior in national elections. Since systemic
effectiveness is basically what is termed &dquo;system support&dquo; or &dquo;affect&dquo; in
the Legitimacy Index, this aspect of effectiveness needs no further
measure. The Persistence of Pattern Index measured changes in govern-
ment and election frequency which were primarily the result of output
ineffectiveness. That is to say, one imputes rationality and democracy to
the twenty countries reviewed: the people, through their elected represen-
tatives in parliament and through the electoral process, decide to change
the government. Changes in government, except Type II (accidental death),
are regarded to be evidence of the population’s nonacceptance of whatever
the government did or did not do. Cabinets fall, prime ministers lose votes
of confidence, parties leave the government-all are illustrations of where
the population does not accept the present policies or personnel. There is
no other way of conceiving of such changes: what else does one mean by

democracy if consent of the governed is not included? It is posited that


the government fell because the people withdrew their consent and,
therefore, the inescapable conclusion is that this consent was withdrawn
because whatever the government was doing or not doing was
unacceptable.
The above remarks are not intended to reduce the importance of
effectiveness in this analysis of democratic political stability. They only
demonstrate the reasons why a separate and distinct indicator for the
concept is not required. Effectiveness is a basic component in the stability
equation together with persistence and legitimacy and should be regarded
as such.

THE INDEX OF DEMOCRATIC POLITICAL STABILITY

The Index of Democratic Political Stability is a composite measure, the


component parts being the Persistence of Pattern Index and the Index of
[55]

Legitimacy/System Support. The two indicators are posited to be of equal


weight, and thus the Index is calculated by summing the T-scores of each
country on the two component indicators. A brief summary of the
attributes and findings of the Index follows:i

(1) The proposed Index of Democratic Political Stability improves


upon various other indices of stability found in the literature.
Represented as an interval scale permitting differentiation among
the countries in the study, the Index moves beyond the nominal
classifications employed by Upset (1959: Table I, p. 32) and
Banks and Textor (1963: 84-85, 98-100).
(2) The proposed index is a composite measure based upon several
attributes, rather than a monomeasure of any one attribute.
(3) The proposed index appears to have successfully synthesized
previous studies’ theoretical constructs and contributions, for there
is a high level of agreement (Spearman’s rank-order correlation)
between the proposed index and other measures of political
stability. The degree of replicability between the stability index
and the conflict/aggressive-behavior measures are less, however,
than with the longevity studies. This is a result of the decision not
to conceive of civil violence and unrest as a component of
democratic political stability unless such behavior had direct,
immediate, and observable consequences upon political phenomena
(assassination of the prime minister, the resignation of a govern-
ment resulting from labor strife, premature elections to allay civil
unrest). Table 9 presents the figures for the degree of replicability
between the proposed index and various other measures.
(4) There are small but distinct variations in stability performances
among the geographical-cultural areas represented in the study.
Anglo-America (Australia, Canada, Great Britain, New Zealand,
United States) scores higher than does Scandinavia (Denmark,
Finland, Iceland, Norway, Sweden) which, in turn, ranks above
Continental Europe (Austria, Belgium, France, Germany, Ireland,
Luxembourg, Italy, Netherlands) which, in its turn, ranks above
Japan-Israel. Table 10 presents the figures for these area groupings.
(5) The proposed Index of Democratic Political Stability has wide
application for further research. One could first offer statements
about the relative position of any one country by comparing its
stability score with the scores of other countries. The Index can
also be employed, through correlation analyses, to examine the
relationship between democratic political stability and several
hypotheses which have traditionally been offered to &dquo;explain&dquo; the
observed variations in stability levels (economic development and
party systems, for example). The proposed Index is well suited for
further research in cross-national comparison, but such analyses are
beyond the scope of this exploratory paper.
w
0
z
w
...
U)
M
cc
w
CL
t-
Z
w
2
z
-cc
uj uj
M>
mO
~ 0
U.
0
x
w
Q
Z
oc
0
LL
<
...
~
Q

[56]
[57]

Notes to Table 1
SOURCE: Keesing’s Contemporary Archives, Vol. V (1943-1945) through Vol. XVIIt
(1969-1970), London: Keesing’s Publications, Ltd.; New York Times, 1945-1969.
a. Canada: Trudeau’s one-month tenure as Prime Minister before the June 1968
general election is not regarded as a new government. Pearson resigned on April 20 in
anticipation of the forthcoming election and Trudeau dissolved the House of
Commons almost immediately upon succeeding Pearson. For our purposes, Trudeau’s
government begins in June-a result of the general election.
b. France: the transformation from the Fourth to the Fifth Republic presents some
difficulties for this measure but none so serious as to warrant excluding the Fifth
Republic. The position of the Prime Minister relative to the other branches of
government (the President, specifically) has been altered in the Fifth Republic and is
not truly comparable to the other countries in the study or, for that matter, even to
the Fourth Republic. But France’s score has been calculated mth the attendant
problems. France’s score and rank, therefore, should not be regarded as precise as
those for the other countries.
c. Great Britain: Eden’s 30-day tenure as Prime Mmister prior to the May 1955
general election is not considered as a new government. Churchill resigned on April 5
in anticipation of the forthcoming election and Eden dissolved the House of
Commons 30 days after his seccession to office. Eden’s government, for the present
analysis, commences after the general election and is a case of post-election
formation (Type 1).
d. Japan: the listed assassination was not in reality an assassination but only an
attempt. It did, however, have the same result as a successful assassination: Kishi was
forced to resign as Prime Mmister due to the wounds he received in the kmfe attack
(July 1960). It was a functional assassination and as such, is classified as Type VIIIl
with 0.00 persistence points.
e. Luxembourg: the last government for Luxembourg began on January 27, 1969
and included in the study even though the begin date is beyond the cut-off point of
is
January 1, 1969. This is due to the fact that the government’s cause-the December
15, 1968 parliamentary election-is within the time requirements of the study.
f. New Zealand: Holyoake’s brief tenure as Prime Minister before the November
1957 general election is not considered a new government. Holland resigned on
September 20 in anticipation of the forthcoming election and the House of
Representatives was dissolved shortly after Holyoake was named Prime Minister.
g. United States: the United States is ranked on this measure even though a true
comparison is not attamable. As stated above in the text, a resignation of the
President (prime minister) is defined as when the President was constitutionally
allowed to stand for a second term but did not so choose. The changes: 7
governments, 4 elections (1948, 1956, 1960, 1964), 2 resignations with change of
both the prime minister and parties (1952 and 1968), and one assassination (1963).
The two resignations received 0 points for the opposition was successful in
subsequent elections. The last government began on January 20, 1969 and is included
in the study although it falls beyond the cut-off date. This is due to the fact that the
cause of the government’s formation-the 1968 general election but, for our
purposes, Johnson’s &dquo;resignation&dquo; in March 1968-is within the time span.
[58]

TABLE 2
SUMMARY OF GOVERNMENT CHANGES, 1945-1969

TABLE 3
INDEX OF GOVERNMENT PERSISTENCE,
EXPRESSED IN T-SCORES
’5.
c
0
cr0
0
c

4)
~
I-
z
w
2
w
w
cc

«CD
LLS Uj
SS
..J>
c(1-
I- < -i
M
C)
ui
-i
u.
0
x
w
a
2
lm
o
LL
<
5
0

[59]
[60]

Notes to Table 4

SOURCE: Keesing’s Contemporary Archives, Volume V (1943-1945) through Vol-


ume XVII (1969-1970), London: Keesing’s Publications, Ltd.; New York Times,
1945-1969; Parliaments and Electoral Systems: A World Handbook, London: Insti-
tute for Electoral Research, 1962.
a. France: the change from the Fourth to the Fifth Republic is not relevant for this
Index because the constitutional tenure of the French Chamber of Deputies remained
at five years. Thus France’s score is calculated for the entire period rather than
considering the Fifth Republic as a distinct data division as is the case with
Luxembourg (see Note d, below).
b. Germany: Article 68 of the Basic Law states that the Federal President may
dissolve Parliament on the Chancellor’s proposal-although he is not obliged to do
so-if the Chancellor is refused a vote of confidence by the Bundestag. The German
system thus permits early dissolution and elections but since this tactic has yet to be
employed, Germany’s score is reported as 1.0000. The discrepancy in time
(5876/5840) is due to a formal shift of election dates. There have been no premature
dissolutions of the Bundestag nor have there been any early elections.
c. Italy: Article 88 of the Italian Constitution reads: &dquo;The President of the Republic
may, having heard their respective Presidents, dissolve both Chambers or only one of
them.&dquo; Italy’s score is reported as 1.0000 for Article 88 has yet to be invoked against
the Chamber of Deputies. The discrepancy reported in Level of Agreement
(7321/7300) is due to a formal shift in election dates. Italy has had no early elections
or dissolution of Parliament.
d. Luxembourg: the constitutional tenure for Luxembourg’s Parliament was changed
in 1954 from 6 to 5 years and thus the country’s score has been calculated in a
composite manner.
Time Span (A): Oct. 21, 1945 - May 30, 1954 Time of (A) 3141
=

Time Span (B): May 30, 1954 - Dec. 15, 1958 Time of (B) 5309
=

Parliaments: (A) 3 Tenure: (A) 1095*


(B) 3 (B) 1825
*The constitutional tenure of the Chamber was actually six years but since one-half
the members were elected every third year, each election was considered to form a
new Parliament and thus the reported tenure of 1095.
Returning to the formula for the Agreement Index, Luxembourg’s scores for the two
periods can be calculated:i
Agreement Index =
time
parliaments x tenure
The Level of Agreement for each time division is:
A) 3141/3285 = raw score of .9562
B) 5309/5475 = raw score of .9697
The total time for Luxembourg is 8450 days of which
A) is 3141 or 37.17% and B) is 5309 or 62.83%.
Completing the final composite score then requires the raw score for each time span
to be multiplied by the percentage of the total time that each division include:
A) raw score .9562 x time span % (37.17) = .3554
B) raw score .9697 x time span % (62.83) = .6093
A) + B) = final raw score of .9647.
e. Norway: The Norwegian Constitution does not permit the Storting’s dissolution
before the constitutional tenure has elapsed and, therefore, no early elections are
possible in the present Norwegian system. Norway’s reported score of 1.0000 is thus
not truly comparable to the other scores and obviously skews the resultant T-scores.
It appears that Norway is being rewarded (1.0000 denotes maximum persistence
of pattern) not for any electoral stability but for constitutional observance. It is this
constitutionality, in part, which helps to classify Norway as a democracy-it does not
cause the country to be stable or unstable in itself.
(Table 4 Notes cant)j
[61]

To require perfect comparability in this index would require that Norway not be
ranked but the exclusion of the country from the Index would lead to more
problems than presently exist. The caveat is that the score of 1.0000 does not reflect
Norway’s actual position (the actual position, in fact, is nonexistent given the
assumptions and definitions of the Index), but the score is included in order to arrive
at a final persistence of pattern score. The score of 1.0000 is the only possible value
that can be assigned to Norway once the prior decision is made to include the
country in the study. The discrepancy in Level of Agreement (7294/7300) is due to
changing election dates.
f. United States: the same caveats must be mentioned for the United States as were
for Norway (see Note e, above). The Level of Agreement discrepancy (7303/7300) is
due to the changing election dates.

TABLE 5
INDEX OF LEGISL·ATlVE AGREEMENT (election frequency),
EXPRESSED IN T-SCORES
[62]

TABLE 6
PERSISTENCE OF PATTERN INDEX
[63]

TABLE 7
THE INDEX OF LEGITIMACY/SYSTEM SUPPORT,
1945-1969

SOURCE: The following three entries were employed for all countries: Keesing’s
Contemporary Archives (London: Keesing’s Publications, Ltd.), Vol. V (1943-1945)
through Volume XVII (1969-1970); New York Times, 1945-1969; Review of
Elections (London: Institute for Electoral Research, 1960 through 1964). The
following entries were employed for the country cited: Parliamentary Handbook of
the Commonwealth of Australia, 1962-196S (Canberra: Commonwealth Parlia-
mentary Library, 1965); Canadian News Facts (Toronto: Marpep Publications, Ltd.,
1968); &dquo;The Constitution,&dquo; Reprint from Denmark: An Official Handbook (Copen-
hagen: Ministry of Foreign Affairs, Press and Information Department, 1964); Le
Monde (1946-1958); F. W. S. Craig, ed., British Parliamentary Election Statistics,
1918-1968 (Glasgow: Political Reference Publications, 1968); Marver Bernstein, The
Politics of Israel (Princeton: Princeton University Press, 1967); Robert A. Scalapino
and Junnosuke Masumi, Parties and Politics in Contemporary Japan (Berkeley:
University of California Press, 1962); Bulletin dlinformation (Luxembourg City:
Minist~re d’Etat, Service &dquo;information et Presse,&dquo; June 30, 1951, May-July 1954,
February-March 1959, July 23, 1964, December 20, 1968).
TABLE 8
THE INDEX OF DEMOCRATIC POLITICAL STABILITY

TABLE 9
MATRIX OF CORRELATIONS (Spearman’s Rs) OF MEASURES OF
DEMOCRATIC POLITICAL STABILITY WITH LEVELS OF EXECUTIVE
STABILITY, DURATION OF GOVERNMENTS, VIOLENT DEATHS,
AND AGGRESSIVE BEHAVIORS

a. Bruce Russett et al., World Handbook of Political and Social Indicators. Table 30.
&dquo;Executive Stability: Number of Years Independent/Number of Chief Executives,
1945-1961,&dquo; pp. 101-104.
b. Jean Blondel, &dquo;Party Systems and Patterns of Government in Western Democ-
loc. cit. Table I V. &dquo;Average Duration of Governments in Seventeen
racies,&dquo;
Democracies During the Post-War Period (years per government),&dquo; p. 191.
c. Bruce Russett et al., World Handbook of Political and Social Indicators. Table 29.
&dquo;Deaths from Domestic Group Violence per 1,000,000 Population, 1950-1962,&dquo; pp.
97-100.
d. Ivo K. and Rosalind L. Feierabend, &dquo;Aggressive Behaviors Within Polities,
1948-1962: A Cross-National Study,&dquo; Journal of Conflict Resolution, X, 3
(September 1966), Table 1, p. 253.
e. The numbers in parentheses represent the sample n.

[64]
[65]

TABLE 10
STABILITY PERFORMANCES BY
GEOGRAPHICAL-CUI.TURAL AREA

NOTES

1. An example of this confusion can be seen by comparing Lipset (1959) to


Needler (1968). Lipset regards legitimacy as a vital component of any serious study
of political stability and he operationalizes the concept by positing that the existence
of a legitimate polity is demonstrated by the "absence over the past 25 years of a
major political movement opposed to the ’democratic rules of the game’" (Lipset,
1959: 30). The latter is defined as any totalitarian movement, either communist or
fascist, receiving twenty percent of the vote during the time span. Thus, for Lipset,
legitimacy is the absence of antisystem political parties.
This view and measure can be contrasted to that of Needler, who regards (1968:
891) the degree of legitimacy as "the number of years during a given period in which
a country has been ruled constitutionally." A constitutional year is defined by
Needler (1968: 891; italics added) as "one in at least six months of which the
country was ruled by a government chosen by more or less free elections and in
which that government on the whole respected constitutional procedures and
individual civil liberties and in which no extra-constitutional changes of government
took place."
Lipset and Needler both agree that stability is related to legitimacy but each has
measured disparate phenomena. To this writer, Lipset is much closer to legitimacy
than is Needler. The latter appears to be measuring "degree of democratic
attainment" rather than "degree of legitimacy."
2. The twenty countries examined are: Australia, Austria, Belgium, Canada,
Denmark, Finland, France, Germany (West), Great Britain, Iceland, Ireland, Israel,
Italy, Japan, Luxembourg, the Netherlands, New Zealand, Norway, Sweden, and the
United States. Switzerland is not included-not because it is undemocratic (which it
is not)-but because the collegial nature of the Swiss executive is not amenable to the
proposed type of cross-national comparison.
3. The expression employed to calculate T-scores is:
[66]

where T represents the T-score, X 1 the successive values of the variable, X the
arithmetic mean of the distribution, and s the standard deviation of the variable.
Expressing the scores as T-scores yields a new distribution which enables one to
present different variables in standardized form. A T-score index has a mean of 50
and a standard deviation of 10, and since the T-scores are standardized, different
measures can be added to form composite scales (see Edwards, 1959: 111-113; 1957:

158-159). The Persistence of Pattern Index is calculated by summing the T-scores of


the Persistence of Pattern Index and the Index of Legitimacy/System Support.
There are numerous other attempts in the literature to construct a composite
index from a number of subindicators (see, for example, Cutright, 1963, for a
measure of development, and Neubauer, 1967, for a measure of democracy).
4. The Government Persistence Index attempts to differentiate among several
types of government formation by positing that some changes are more (or less)
stable than other changes-a valid ordinal scale. But the mechanics of the Index
require that a value be assigned to each formation-type in order to compute the final
persistence scores. This assignment of values takes one beyond what was originally
intended. That is, postelection formations are judged to be more persistent (stable)
than, say, inter-election resignations/reformations. But to assign the scores of 1.00
and 0.50, respectively, states in effect that post-election formations are twice as
stable as reformations. This latter development is not intended and this caveat should
be kept in mind.
5. There is, of course, the possibility that an accidental death is in reality
suicide. In such cases, the assigned persistence score would be 0.25 (voluntary
resignation) on the grounds that perceived non-support led to the suicide. Since there
is only one possible instance in the present analysis where this question may be
asked-Holt’s drowning in Australia&mdash;the problem is not too serious. I assume Holt’s
death to be accidental until proven otherwise and thus Australia receives a score of
1.00 for the change from Holt to McEwen.
6. The resignation of Anthony Eden is an excellent example of this problem of
"voluntary" resignations. This writer does not deny that Eden was ill and pleaded
such but to what extent did political considerations or political illness enter the
situation? Eden might very well have been forced out of office in the aftermath of the
Suez campaign.
7. This is saying, in effect, that if Eden were ill but remained in office until he
died, Great Britain would have received 1.00 points (accidental death) in place of the
assigned 0.25 points. But the point is that one does not know how "sick" is sick.
Sickness unto death is observable and is classified as death. Sickness short of death is
not illness and is classified as a resignation.
8. A voluntary resignation of the United States President is operationalized as
any instance where the President was constitutionally permitted to stand for a second
term but did not so choose (Harry Truman in 1952 and Lyndon Johnson in 1968).
9. In cases where political parties leave their members free to join or not join an
administration, the resignation of a cabinet minister belonging to such a party is not
regarded as a party withdrawal. The cabinet representation is done on an individual
basis-the parent party being neutral-and when the individual resigns, the resignation
does not lead to a different government. This is illustrated with the first government
in the French Fifth Republic: the Radicals left their members free to join Debre’s
administration and the subsequent resignation of the Radical member is not regarded
[67]

as a change in government. Conversely, in the case of a party being represented in the

government and then withdrawing en masse except for individuals remaining in the
cabinet without the parent party’s approval, a new and different government is
posited to exist. This was the situation in the French Fourth Republic when the MRP
withdrew from the government in June 1954 but individual MRP ministers continued
until Felxuary 1955. A different government is posited as of June 1954.
10. Almond and Verba (1965: 186) are among those who regard legitimacy as
system support. They comment, "Political systems, if they are to survive must...
be... relatively legitimate, that is ... the system, if it is to have a long run potential
of survival, must be generally accepted by citizens as the proper form of
government." Lipset (1959: 64) also regards legitimacy as system support.
11. It appears necessary to note that some people who do not vote may harbor
anti-systemic feelings but due to the lack of precise data separating that percentage of
the non-voting electorate not voting because of dissatisfaction from that percentage
not voting for other reasons (structural and physical impediments, apathy, system
support to such an extent that the individual regards all possible alternatives as
equally acceptable), it has been decided to exclude all such (non) voters from the
calculations. This may raise each country’s raw score but the relative positions and
T-scores, assuming an equal margin of error for each country, should be unchanged.
12. Lipset (1959: 30) regards political movements of the left and the right
(communist and fascist, to be exact) as "opposed to the democratic ’rules of the
game’ " and, being totalitarian movements, classifies them as antisystem parties.
Kogan (1962: 40-48) agrees with this conception of the left-right extremes in a
democracy. He sees antisystem parties "as those which would destroy the present
Republic (Italy) if they ever achieved the necessary power." Parties falling into this
category in Italy are mentioned by Kogan to be the extremes (Communist and Social
Movement). Eckstein (1966: 15) also believes that the left-right extremes inject
antisystemic elements and he employs the absence of such groups in Norway as
partial proof of that country’s legitimacy and, hence, stability.
13. Almond and Verba (1965: 186) comment, "Political systems, if they are to
survive, must... be relatively effective ...that is, what the government accom-
plishes must be at least satisfying enough to the citizens so that they do not turn
against the government." Eckstein (1966: 228-229), Goldberg (1968: 4) and Lipset
(1959: ch. 3) agree that the concept of effectiveness means satisfying or acceptable
action.
14. Lipset (1959: ch. 3: 70) appears to have overlooked his own prescriptions
and understanding of the concept when he equates effectiveness with economic
development. The data in Russett et al. (1964: esp. 56-81) relating to levels of
government expenditures may be valid for specific outputs of a polity, but the data
do not measure effectiveness in the sense of acceptable action.
15. A debatable example is found in the United States. The United States space
program is undoubtedly effective in actual output and success but it may very well be
unsatisfactory (misplaced priorities) to many citizens and thus be ineffective as the
term is used here. A second example might be the current Republican adminis-
tration’s policies toward desegregation. Here we may have very little output but if
this is satisfying and acceptable to the citizens-at least to a majority of the
citizens-then this minimum output is "effective" action.
[68]

REFERENCES

ALMOND, G. and S. VERBA (1965) The Civic Culture. Boston: Little, Brown.
BANKS, A. S. and R. B. TEXTOR (1963) A Cross-Polity Survey. Cambridge, Mass.:
MIT Press.
BLONDEL, J. (1968) "Party systems and patterns of government in Western
democracies." Canadian J. of Pol. Sci. 1 (June): 180-203.
BWY, D. P. (1968) "Political instability in Latin America: the cross-cultural test of a
causal model." Latin American Research Rev. 3 (Spring): 17-87.
CUTRIGHT, P. (1963) "National political development: measurement and analysis."
Amer. Soc. Rev. 28 (April): 253-264.
DAALDER, H. (1966) "Netherlands: opposition in a segmented society," pp.
167-177 in R. Dahl (ed.) Political Oppositions in Western Democracies. New
Haven: Yale Univ. Press.
ECKSTEIN, H. (1966a) Division and Cohesion in Democracy: A Study of Norway.
Princeton: Princeton Univ. Press.
---

(1966b) "A theory of stable democracy," pp. 225-288 in Division and


Cohesion in Democracy: A Study of Norway. Princeton: Princeton Univ. Press.
EDWARDS, A. (1959) Statistical Methods for the Behavioral Sciences. New York:
Rinehart.
(1957) Techniques
&mdash;&mdash;&mdash;
of Attitude Scale Construction. New York: Appleton-
Century-Crofts.
EPSTEIN, L. D. (1967) Political Parties in Western Democracies. New York:
Frederick A. Praeger.
GOLDBERG, A. S. (1968) "A theoretical approach to political stability." Presented
at the Annual Meetings of the American Political Science Association, Washing-
ton, D.C., September 2-7.
KOGAN, N. (1962) The Government of Italy. New York: Thomas Y. Crowell.
LIJPHART, A. (1968) The Politics of Accommodation: Pluralism and Democracy in
the Netherlands. Berkeley: Univ. of California Press.
LIPSET, S. M. (1959) Politican Man. Garden City, N.Y.: Doubleday.
NEEDLER, M. C. (1968) "Political development and socioeconomic development:
the case of Latin America." Amer. Pol. Sci. Rev. 62 (September): 889-897.
NEUBAUER, D. (1967) "Some conditions of democracy." Amer. Pol. Sci. Rev. 61
(Decernber): 1002-1009.
RUSSETT, B. et al. (1964) World Handbook of Political and Social Indicators. New
Haven: Yale Univ. Press.

You might also like