unemployment matters: voting patterns during the economic transition in poland, 1990–1995

This article was downloaded by: [Northeastern University]On: 13 November 2014, At: 12:05Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Europe-Asia StudiesPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/ceas20

Unemployment matters: Votingpatterns during the economictransition in Poland, 1990–1995Janice Bell aa Sussex European Institute, University of Sussex ,Published online: 06 Nov 2007.

To cite this article: Janice Bell (1997) Unemployment matters: Voting patterns duringthe economic transition in Poland, 1990–1995, Europe-Asia Studies, 49:7, 1263-1291, DOI:10.1080/09668139708412499

To link to this article: http://dx.doi.org/10.1080/09668139708412499

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoeveras to the accuracy, completeness, or suitability for any purpose of the Content. Anyopinions and views expressed in this publication are the opinions and views of theauthors, and are not the views of or endorsed by Taylor & Francis. The accuracyof the Content should not be relied upon and should be independently verifiedwith primary sources of information. Taylor and Francis shall not be liable for anylosses, actions, claims, proceedings, demands, costs, expenses, damages, and otherliabilities whatsoever or howsoever caused arising directly or indirectly in connectionwith, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms& Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/loi/ceas20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/09668139708412499

http://dx.doi.org/10.1080/09668139708412499

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

EUROPE-ASIA STUDIES, Vol. 49, No. 7, 1997, 1263-1291

Unemployment Matters: Voting Patternsduring the Economic Transition in

Poland, 1990-1995

JANICE BELL

AFTER THE VICTORY of the Democratic Left Alliance (SLD) in Poland's 1993 parlia-mentary elections, the general consensus was that the increase in support for thereformed communist party was linked to the course of economic performance duringthe transition. What was less certain, however, was the precise nature of the relationbetween the economy and the election.

One explanation was that the election resulted from mass disillusionment with the'social costs' of reform. In this view, the voters, disappointed with 'unduly cruelmarket reforms' and the 'rapid enrichment of the few and impoverishment of many','...led the post-communist parties to victory'.1 The SLD and Polish Peasant Party(PSL) benefited from protest votes which expressed a uniform disapproval of thecourse of market reform. A weakness of this type of analysis is that it relies onimprecise language. For instance, Kabaj & Kowalik cite the existence of massimpoverishment without reference to any of the exhaustive research on the scale anddistribution of poverty during transition.2

In contrast, this article argues that there are identifiable patterns of interest-basedvoting in Poland. These patterns are not only present in the more stable economy of1992-95 but are observable and consistent across each of Poland's four elections.Unlike other studies of voting using regional demographic and socioeconomic data,3

conclusive results can be obtained through the use of only two economic variables.Distributional issues play a central role in economic voting. Despite a highlyfragmented party system, the development of interest-based voting is discerniblethrough the ongoing consolidation of political parties. Of the conclusions reached inthis article, the most significant is the strong relation between unemployment ratesand voting. The rise in support for the SLD is closely related to the rise in joblessnessacross time and its variation across regions.

While this article concentrates on economic voting, it is acknowledged thatnon-economic issues also influence voters' decisions. For example, internal rifts andinfighting in Solidarity, revealed by the 1990 'war at the top' between Walesa andMazowiecki, damaged the organisation's public image. The much-discussed problemof the lack of parliamentary representation of the anti-communist centre-right ismostly a reflection of fragmentation at this end of the spectrum. The large number ofsmall rightist parties have been reluctant to trade their separate identities and

0966-8136/97/071263-29 © 1997 University of Glasgow

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

1264 JANICE BELL

autonomy for a greater probability of sharing seats gained through an electoralcoalition.4 In terms of institutions, the 1991 electoral law was too proportional; 29electoral lists were represented in the Sejm after that year's election. Amendmentspassed prior to the 1993 Sejm election raised thresholds for parties and electoralcoalitions, reducing the number of groupings elected to the Sejm in 1993 to six, plustwo parties representing the German minority.

Despite these institutional and political factors, as well as the ongoing debate overthe role of the Catholic Church, lustracja, abortion, foreign policy and so on, thecentral battleground has been over economic policy and performance. The area ofdebate can even be narrowed further: what is being debated is not whether Polandshould develop towards a market-oriented economy modelled on the advancedsystems of the West. The goal of a developed market economy is widely accepted.Rather, the core of the debate has centred on the distribution of costs and gains frommarket reforms, and increasingly on the role of the state in the economy.5

Economic voting

What is the link between economic performance and voting?

In simple terms, the theory of economic voting states that economic performanceinfluences an incumbent's chances of being re-elected. Higher growth, larger incomesand lower inflation in the months before an election boost popular support for theserving government, party and individual politician.

The most distinctive characteristic of this public choice approach to voting is therational voter hypothesis, which assumes that the individual voter behaves the sameway in the polling booth as in the marketplace—rationally and in pursuit of his ownself-interest.6 On the basis of each candidate's policy platforms, the voter calculateswhich candidate or party would deliver the highest 'streams of utility' once elected.7

The calculation of potential benefit is self-centred, as the voter takes his ownwelfare—past and/or future—into account. Unlike rational expectations analyses,rational choice allows for imperfect and costly information, whereby voters can beunsure about the post-reform distribution of incomes and welfare.

While rational choice theories of voting are predicated on the 'selfish voter',Smith's (1975) analysis of voting on tax equalisation showed that people can act inapparent contradiction to their own economic interests and welfare.8 Again and again,the question arises about how to account for the contradiction between what politicalscience terms 'self-interest' versus 'social interest'. Public choice allows this conflictto remain within the rational framework by adapting the voter's preference set toinclude wider social concerns. Conceivably, people make the connection betweentheir own personal welfare and the 'public interest' of the national or regionaleconomy. Mueller9 provides a utility function which incorporates these consider-ations, in that voter i seeks to maximise:

0,-=!/,•+02 £/,•

where 0 equals 0 for the selfish voter and 1 for the altruistic or socially oriented voter.If we disregard concerns of whether it is rational to vote and instead concentrate on

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

VOTING PATTERNS IN POLAND 1265

the choice of candidate, we are still left with the problem of not being able to tellfrom the wojewodztwo level data what weights these utility functions are given in theoverall set of preferences. In this article we shall proceed onto operational votingfunctions, which shed light on the relative weight of these factors. For the presentdiscussion, it will be assumed that 0 < 0.5, whereby the average voter weighspersonal experience more heavily than social welfare in his or her utility function.Voting behaviour is a combination of both self-interest and socially oriented voting,but it is assumed that the weight is greater on the former than the latter.

As implied, a refinement of this argument is that more value or utility may beattached to the welfare of people close to the individual, whether in their family, townor wojewodztwo, than to the welfare of the country as a whole. Therefore, we mightexpect that the regional level data will reveal some relationships which do not emergefrom the national data, just as cross-sectional data can reveal aspects which are notvisible in time-series data. This incorporation of self and society into individualpreference sets can be illustrated through the case of unemployment. In areas of highunemployment, even people who are still employed may perceive a higher probabilityof losing their own job in that environment, or they may consider high unemploymentto lower the entire community's welfare. Therefore we can expect a greater prefer-ence for redistributive or pro-employment policies in high unemployment regions.

Unemployment, income and voting

For more than 30 years, scholars have debated whether personal experience ofunemployment and stagnating real wages affects individual voting behaviour morethan wider social preferences. In his study of US congressional elections, Kramer10

argued along the lines of Downs' model11 that fluctuations in personal income, risinginflation and higher unemployment do affect voting behaviour. Kramer relegates theeffects of incumbency, campaign tactics, and 'coattail effects' of presidential incum-bency to error terms. If this theory is correct, coefficients for real and nominal incomeshould be positive, and price and unemployment coefficients negative. Kramer'sregressions showed significant connections between voting and price and incomevariables, but the coefficient for unemployment had the wrong sign, possibly becauseunemployed Americans are less likely to vote. The rationale and the argument arevery persuasive. The assumption that individual financial and employment conditionsinfluence voting behaviour has permeated assumptions about the political economy ofelections.

The connection between personal incomes, unemployment and voting has not goneunchallenged. Kramer's model has been criticised for not providing sufficientlyconsistent results. Kinder & Kiewiet12 also examined voting in US congressionalelections. They operated on the null hypothesis that people dissatisfied with their ownfinancial situation should vote against the incumbent, and likewise, that the unem-ployed should also prefer the opposition to the incumbent. In accordance with theirhypothesis, Kinder & Kiewiet found that general 'collective judgements' were moreimportant for election outcomes than 'personal economic grievances'. Kinder &Kiewiet's regressions showed that general trends such as the business cycle, economicissues of national importance, and the managerial competence of parties had a greater

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

1266 JANICE BELL

influence on voting than voters' personal financial situation. They found littleevidence of consistent and predictable voting patterns by employment status, satisfac-tion, or even amongst the working class and other groups who are vulnerable to policyshifts. If Kinder & Kiewiet's thesis holds, then coefficients for personal income andunemployment should not be significant.

Of all the possible economic variables which could be used to try to explain voting,these two classic models agree that unemployment and changes in real income are themost politically salient factors, because of the very real social costs they impose.

Polish data on priorities

The case of Poland is different from the established and (more or less) stable USeconomy, on which these models are based. Because of the transition from a plannedeconomy, economic liberalisation and the emergence of open unemployment compli-cate attempts to find a simple Phillips' curve trade-off between inflation andunemployment. Both inflation and unemployment continue to dominate public con-cerns about the economy and personal well-being in transition economies, but theseare not given equal weights. Over the course of three years, the polling agency CBOSasked Poles what were the most important problems facing Poland.13 Two-thirdsmentioned unemployment in 1992; this rose to 69% in 1993 and fell slightly to 65%in 1994. The second highest response was for low wages and high prices, thepercentage of people selecting this answer went from 59% in 1992 to 49% in 1993and 54% in 1994. Rose & Haerpfer14 found that, when asked to compare these twothreats, in 1995, 76% of Polish respondents thought inflation posed a greater threat totheir household than unemployment, whereas this figure two years earlier was 57%.The opinion data suggest that unemployment and inflation have loomed large asPoland's two most important problems, and this is not only limited to economicproblems. Secondly, it appears that the threat of unemployment peaked in 1993 andis now declining.

We shall look for a relation between unemployment and income and voting patternsin the 1990 and 1995 presidential elections. Wojewodztwo level figures for per capitaincome and unemployment will be compared with election results from the sameregional level. After looking at the raw data and making simple correlations, we canthen perform OLS regressions on unemployment, income and voting in Poland'sregions.

A model of economic voting

Voting functions can incorporate a wide range of economic and non-economic data,such as rate of GDP and/or personal income growth, recent trends and levels ofpopularity in public opinion polls, the exchange rate, incumbency and so on. In thisarticle a simple multivariate model will be used to analyse how well per capitaincome and unemployment rates can explain voting patterns in Poland. Research intovoting can use time series of aggregate macroeconomic indicators to explain thedependent variable of the vote distribution across several decades of elections (e.g.Fair (1996) on US presidential elections15). Since Poland has only had two presiden-

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014


tial and two freely contested parliamentary elections since 1990, any regressionsbased on four observations would be patently unreliable.

Instead, I will use a technique which is not only better suited to the available databut which will allow an investigation into the politics of the distribution of economiccosts and benefits. The equation is as follows:

V= bi + b2l + b3 U

where V is the share of the vote gained by the designated candidate in each of 49provinces (wojewodztwa), I is average per capita monthly income (on an annualaverage) in the wojewodztwo, and U is the provincial unemployment rate.16

This regression uses data for per capita income for the year as a whole, as theseare the data available from Rocznik Statystczny in terms of overall per capita income,not just average wages in industry or construction. As each of Poland's post-commu-nist elections was held at the end of the third or in the fourth quarter of the year, thisis an adequate proxy for incomes on the day of the election.

The 1990 presidential election

By November 1990 Poland was entering a recession worse than originally andoptimistically anticipated. GDP fell by approximately 12% in 1990, a large drop evenconsidering the upward bias of statistical distortions. Nearly 6% of the labour forcewas unemployed in November, up from 0.3% in January 1990. While inflation wasgradually easing, and both imports and exports were rising, unemployment and thedelayed threat of further labour shedding17 increased the level of personal economicinsecurity.

Table 1 contrasts the actual polling results from the first round of elections withpreferences one week before the election as surveyed by CBOS, and PGSS responsesfrom their 1992 survey.18 In the first round of Poland's first post-communistpresidential election, held on 25 November, no single candidate won an absolutemajority, thus requiring a second round on 9 December. While the CBOS and officialfigures are slightly different, the candidates finish in the same order in both lists andcomparison of the CBOS results with official figures places the opinion data close toa standard error of 3%.

Let us consider the campaign stance of each of the main candidates. BothMazowiecki and Walesa were in the post-Solidarity camp but, as the serving primeminister, Mazowiecki was the policy-making incumbent and therefore the mostclearly connected with current government policy. Walesa had one foot in theanti-communist, pro-reform, Solidarity camp and the other in a more populist,anti-Mazowiecki position. While Cimoszewicz and the SdRP were still too closelyassociated with the largely discredited PZPR regime, a hard-core of about 10% votedfor Cimoszewicz. The superior PSL grassroots organisation was still not sufficient toget Bartoszcze elected, although his support was closely linked to the well-definedlobby of small private farmers. Yet the rural vote was split, with Walesa also doingwell in the countryside. Tyminski represented the complete political outsider; hecampaigned on a populist platform and appeared to have no ties to any pre-existingpolitical organisation.

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

1268 JANICE BELL

TABLE 1VOTE IN FIRST ROUND OF PRESIDENTIAL ELECTION, 27 NOVEMBER

1990 (%)

WalesaTyminskiMazowieckiCimoszewiczBartoszczeMoczulski

Official data

39.9623.1018.089.217.152.50

CBOSpoll

362622681

PGSS

46.714.717.84.53.92.8

Source: Rzeczpospolita, 1-2 December 1990; CBOS, 1990,'Elektorat na tydzien przed wyborami', Report No. 683/90,November, p. 15 (n = 1017); Institute for Social Studies,University of Warsaw, Polish General Social Survey, 1992(n = 1647).

With the advantage of several years, the protest vote analysis might easily concludethat the 1990 election marked the end of the political 'honeymoon' and thatMazowiecki's third place showing was inevitable. However, the results did surpriseMazowiecki and his supporters,19 as just a few months previously their candidate hadhigh and wide-ranging popular support. While Tyminski capitalised on peoples' fearsand scepticism, in the end Walesa's platform of progress (he criticised Mazowieckifor not advancing political reforms quickly enough) tempered with populism won outover the simple protest vote.

For the 1990 presidential election, the above equation was used to estimate theimpact of personal income and regional unemployment levels for support for each ofthe top six candidates in the first round of voting. Results are shown in Table 2. Theincome/unemployment regression was performed for each of the top five first roundcandidates. While the entire election outcome was not completely determined by thecombination of income and unemployment (the highest R1 was for Tyminski, at0.402), significant effects do emerge which tell us about their relative influence.

Votes for Walesa can be typified by a negative relation with unemployment andincomes. While there is a high and statistically significant coefficient for unemploy-ment, the total impact of lower incomes is not as conclusive but is still within aconfidence interval of 99.1%. Walesa's electorate can be characterised as the workingpoor; typically low-skilled industrial workers whose wages have fallen in real as wellas relative terms during transition.

Tyminski finished in second place and received more support from areas withhigher unemployment and higher incomes, a somewhat contradictory picture but inkeeping with the electorates of populist candidates (see results for PPPP in the 1991Sejm elections, below). Tyminski's populist and redistributionist themes attracted aproportion of the newly emergent group of unemployed, and received a greater shareof the jobless vote than did Walesa.

The coefficients for income and unemployment as regressed against election resultsfor Mazowiecki reveal that his electorate was in favour of economic policy asdescribed by Kramer's criteria: the coefficient for nominal income is positive and thatfor unemployment is negative. Income was the more important determinant, being

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014


TABLE 2PERSONAL INCOME AND UNEMPLOYMENT LEVELS, FIRST ROUND OF 1990 PRESIDENTIAL ELECTION

Variable CoefficientStandard

error t valueSignificanceoft value

WalesaCONSTANTINCOMEUNEMPLOY

R2: 0.171. Adjusted/?2: 0.135.

TyminskiCONSTANTINCOMEUNEMPLOY

R2: 0.402. Adjusted R2: 0.376.

MazowieckiCONSTANTINCOMEUNEMPLOY

R2: 0.262. Adjusted R2: 0.229.

BartoszczeCONSTANTINCOMEUNEMPLOY

R2: 0.321. Adjusted R2: 0.040.

CimoszewiczCONSTANTINCOMEUNEMPLOY

n = 49. R2: 0.133. Adjusted R2

0.867-0.368- 1.582

SEE: 0.081. F-ratio

- 0.2370.3521.885

SEE: 0.051. F-ratio

-0.1700.384

- 0.754

SEE: 0.060. F-ratio

0.477- 0.386-0.104

SEE: 0.040. F-ratio

0.055-0.001

0.603

0.096. SEE: 0.033.

0.1840.1710.580

(sig): 4.749

0.1150.1070.361

4.699-2.139- 2.727

(0.013)

- 2.0603.2895.219

(sig): 15.444 (0.000)

0.1370.1280.431

(sig): 8.146

0.0910.0850.285

(sig): 10.86;

0.0750.0700.236

F-ratio (siel

- 1.2393.005

-1.751

(0.001)

5.252- 4.564- 0.363

> (0.000)

0.729-0.019

2.551

i: 3.543 (0.037)

0.0000.0380.009

0.0450.0020.000

0.2220.0040.087

0.0000.0000.718

0.4700.9850.014

All numbers rounded to third decimal place.Source: J. J. Parysek, Z. Adamczak & R. Grobelny, 'Regional Differences in the Results of the1990 Presidential Election in Poland as the First Approximation to a Political Map of the Country',Environment and Planning A, 23, 1991, pp. 1315-1329 own calculations from GUS data.

significant within 99.6%. This is entirely consistent with public opinion data whichshow that UD and subsequently UW voters typically come from the professional,managerial and intellectual classes. People with higher incomes and more secureemployment, and the education which enables this status to be attained, are morelikely to support market-oriented reform than more vulnerable groups, such asindustrial workers and farmers.

Bartoszcze was the candidate of the Peasant Party (PSL), and public opinion datashow that the bulk of his support comes from the rural and agricultural population.As such, it is not surprising that unemployment rates have no significant impact onvoting for Bartoszcze. However, lower per capita incomes show a strong relation.Several studies, including the author' s doctoral research, have definitively shown that

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

0.1720.1600.540

6.006- 0.839- 4.893

0.0000.4060.000

1270 JANICE BELL

TABLE 3REGRESSION RESULTS FOR WALESA IN SECOND ROUND, (1990)

Standard SignificanceVariable Coefficient error lvalue of lvalue

CONSTANT 1.031INCOME -0.134UNEMPLOY -2.641

n = 49. R2: 0.345. Adjusted R2: 0.317. SEE: 0.076. F-ratio (sig): 12.117 (0.000)

All numbers rounded to third decimal place.Source: as Table 2.

rural households were disproportionately affected by falling real incomes. Therefore,this finding on Bartoszcze's electorate is consistent with the empirical data.

Finally, the candidate of the ex-communist SdRP, Cimoszewicz, attracted just over9%, nearly 2% more than Bartoszcze. That a former communist managed to attractnearly one in ten votes surprised some observers. However, it should be rememberedthat in the 1989 semi-free parliamentary elections, candidates from the combinedgovernment list (PZPR, ZSL and SD) received 3.3 million votes,20 an electoratewhich in 1990 was mostly split into 1.2 million for Bartoszcze and 1.5 million forCimoszewicz. For the 1990 election, it is important to note that higher unemploymentrates were positively linked to support for Cimoszewicz, with a coefficient of 0.6,significant to 1.4%. There was little sign of any relation to income levels, but overallthis result is consistent with the eventual consolidation of the post-communists—andnot the eccentric Tyminski—as the legitimate recipient of the unemployed vote.

These regressions reveal a great deal about the impact of unemployment andincome voting in post-communist Poland's first presidential election. Furthermore,nearly all of the coefficients have the expected sign, as predicted by our model ofeconomic voting. The same regression can be used to analyse the results of the secondround of the 1990 presidential election, in which Walesa defeated Tyminski (Table 3).Again, the independent variables explain about one-third of the total deviation fromtrend. The coefficient for income is negative, but with a low coefficient andsignificance level. The most interesting finding is the negative relation betweenunemployment and support for Walesa. That is, higher unemployment is associatedwith a higher vote for Tyminski and, conversely, areas with lower unemploymenttended to support Walesa.

We can also conclude that voters who selected Mazowiecki in the first roundsupported Walesa in the second, as did farmers. The threat of Tyminski being electedmobilised support for Walesa across a wider range of social groups and interests thanin the first round. Mazowiecki voters were 'trapped' during the second stage of theelection. If we think of the candidates as being on a spectrum from full continuity ofreforms to populism, Mazowiecki was clearly pro-market and Tyminski populist, withWalesa at a vague location between the two. Although Walesa's policies were notcompletely desirable to Mazowiecki voters, the dilemma was that the overriding needto defeat Tyminski dominated the negative aspects of Walesa's platform. Therefore,

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014


TABLE 4RESULTS FROM THE FIRST ROUND OF THE 1995 PRESIDENTIAL

ELECTION (%)

KwasniewskiWalesaKuronOlszewskiPawlakZielinskiGronkiewicz-WaltzKorwin-Mikke

Official results

35.133.19.226.94.33.52.82.4

OBOP exit poll

33.733.110.07.03.73.03.02.8

Source: Official results: Maly Rocznik Statystyczny, 1996, pp.73-74.OBOP data: Gazeta Wyborcza, 6 November 1995, p. 1 (n = 1145).

the shift of support from Mazowiecki to Walesa is logical, and indeed was crucial inthe latter's victory.

The 1995 presidential election

The 1995 presidential campaign was characterised by inability of the anti-communistright to unite behind one nominee, and by bitter ad hominem attacks betweencandidates. In the first round, Olszewski and Gronkiewicz-Waltz arguably drew about10% of votes from Walesa's potential electorate. The poor showing of Pawlak (3.7%)could have reflected his personal unpopularity after a controversial tenure as primeminister.21 Despite being Poland's most popular and respected politician, concernsabout Kuron's health damaged his chances.22 As shown in Table 4, the figures revealthis to have been a two-man race.

While candidates assumed highly symbolic roles in the 1990 election, the personalattributes and personality of the candidates were more important in the 1995election—especially in terms of negative voting. In a survey conducted by OBOP forGazeta Wyborcza22' on 19 November 1995, 54% of Kwasniewski voters had chosentheir candidate in the first round because Kwasniewski had 'a responsible character[and] individuality', whereas 46% of Walesa supporters were motivated 'because itwas impossible to support any other candidate'! Kwasniewski's victory could be anindicator of the rise of rational, interest-based economic voting, or it could demon-strate the power of negative voting against Walesa.

The OBOP exit polls are very close to the official results, and an improvement onthe 1990 surveys. It is interesting to notice that the largest differences between surveydata and official figures were for Kwasniewski and Pawlak; one might speculate thatthere was some residual hesitancy to admit having voted for the post-communists orthe controversial Pawlak.

While unemployment in November 1995 was, at 14.7%, more than double that inNovember 1990, jobless rates were receding from the 1994 peak of 16.9%. Judgingfrom the economic data, the strong and sustained recovery, the falling rate ofunemployment and rising real wages should suggest a Walesa victory as presidential

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

1272 JANICE BELL

TABLE 5REGRESSION RESULTS FOR FIRST ROUND OF PRESIDENTIAL ELECTION, 1995

Gronkiewicz-Waltz (Ind.)

Korwin-Mikke (UPR)

Kuron (UW)

Kwasniewski (SLD)

Olszewski (ROP)

Pawlak (PSL)

Walesa (Ind.)

Zielinski (Ind.)

R2

0.1762

0.2782

0.2456

0.0684

0.1414

0.1866

0.2318

0.3032

Adj. R2

0.1404

0.2468

0.2128

0.0279

0.1040

0.1512

0.1984

0.2729

Unemployment(significance)

- 0.0172(0.3360)

- 0.0054(0.7987)0.1573

(0.0272)0.5273

(0.0773)- 0.2205

(0.0090)- 0.2278

(0.0884)- 0.9667

(0.0006)0.0437

Income(significance)

0.0284(0.0235)0.0540

(0.0005)0.1787

(0.0005)0.0833

(0.6791)- 0.0799

(0.1563)- 0.2882

(0.0024)-0.1866

(0.3042)0.0650

Source: Panstwowa Komisja Wyborcza, Wyniki Wyborow Prezydenta RzeczpospolitejPolskiej 5 listopada i 19 Hstopada 1995 (Warsaw, 1996); own calculations from GUS data.

incumbent. However, in the Polish political system the parliament and the cabinet aremore important loci of policy making, and both of these had been controlled by theSLD-PSL coalition since the September 1993 Sejm elections. Furthermore, SLDnominees had manned the key economic ministries and the premiership since 1994.Therefore, in this study, incumbency should be understood as being control of thegovernment rather than the presidency, as in models used for the USA. Because ofthe use of regional rather than time-series data, dummy variables for incumbency willnot be introduced into the regression.

In spite of Balcerowicz's often-cited argument that Poland's present growth wasmade possible only by the tough policies of 1990-91, actual voting behaviour appearsto link present governments with present economic performance. It is only since 1993that political and economic conditions have emerged from the volatility of 1988-91.In the mind of the average voter this greater stability probably owes more to thepresent than the preceding government.

For the first round of the 1995 election, the regressions for the fragmented votefor centre-right and right wing candidates, apart from Walesa, all demonstrateinconclusive results (Table 5). Several candidates, with varying economic and socialoutlooks, were fielded from the political right. Gronkiewicz-Waltz and Korwin-Mikke's electorates demonstrate the expected relation to lower unemployment andhigher income, but at insignificant levels. Olszewski has a better relation with lowerunemployment rates; a 1% drop in regional unemployment is related to a 0.22% risein his support.

The results for Walesa reveal the consistent relation of his electorate with lowerunemployment and lower personal incomes. Moreover, the coefficient for unemploy-ment is not only very high, but it has an extremely good level of significance.Consistant with our original hypothesis, support for Kwasniewski shows a stronglink to unemployment. While the coefficient is higher for Kwasniewski, the

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

0.1750.2200.321

1.1900.7174.392

0.2400.4770.000


TABLE 6REGRESSION RESULTS FOR KWASNIEWSKI IN SECOND ROUND, 1995

Standard SignificanceVariable Coefficient error t value of t value

CONSTANT 0.208INCOME 0.158UNEMPLOY 1.408

R2: 0.309. Adjusted/?2: 0.279. SEE: 0.101. F-ratio: 10.295 (0.000)

All numbers rounded to third decimal place.Source: as Table 2.

significance of the unemployment variable is stronger for the Unia Wolnosci candi-date, Jacek Kuron. Both of these candidates show a positive relation to higher income,although this is stronger for Kuron, for whom patterns of support have undoubtedlybeen strengthened by his popularity as a left-of-centre politician.

Finally, while Pawlak did not receive a sizeable share of the total vote, theregression run on regional levels of support reveals a comparatively strong relationbetween support for the PSL candidate and both lower unemployment and lowerincome, as is typical of the rural electorate. The similarity of this electorate toWalesa's suggests that the rural vote is split between agrarian interests of the PSL andthe conservative social values of centre-right parties in its voting preferences.

In the regression for Kwasniewski in the second round of the 1995 elections(Table 6), the clear conclusion is that unemployment was again the decisive factor.Higher unemployment rates are linked to support for Kwasniewski. Income was alsopositive, but much less important, suggesting that a wide range of income groups, andtherefore social groups, voted for the SLD candidate.

Public opinion data on the breakdown of first-round voting by occupational groupreinforce how much this was a race between Walesa and Kwasniewski from the start.More self-employed and private businesspeople supported Walesa than Kwasniewski,while the unemployed gave the largest share of their vote to Kwasniewski (Table 7).Kuron received his highest level of support from unemployed people, but this couldnot outweigh their much greater preference for the SLD candidate. Farmers were splitbetween Kwasniewski and Walesa, but notably farmers gave the least support to theUD candidate, Kuron. Workers were also divided between the leading candidates, but

TABLE 7ELECTION PREFERENCE IN THE FIRST ROUND OF 1995 PRESIDENTIAL ELECTION BY

SOCIO-ECONOMIC GROUP (%)

WorkersFarmersPensionersUnemployedEntrepreneurs

Kwasniewski

35.730.434.237.028.0

Walesa

35.230.639.628.539.5

Kuron

7.03.27.4

10.79.3

Olszewski

8.76.86.36.46.9

Source: OBOP survey cited in Gazeta Wyborcza, 6 November 1995, p. 4

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

1274 JANICE BELL

TABLE 8SHARES OF VOTES IN 1991 SEJM ELECTION

Unia DemokratycznaSojusz Lewicy DemokratycznejWyborcza Akcja KatolickaPorozumienie Obywatelskie CentrumPolskie Stronnictwo Ludowe 'SP'Konfederacja Polski NiepodleglosciKongres Liberalno-DemokratycznyRueh Ludowy 'PL'NSZZ 'Solidarnosc'Polska Partia Przyjaciol PiwaChrzescijanska DemokracjaUnia Polityki RealnejSolidarnosc 'Pracy'German minorityTotal others

000 votes

1382.11344.8980.3977.3973.0841.7840.0613.6566.6367.1265.2253.0231.0132.1630.8

% of votes

12.312.08.78.78.77.57.55.55.13.32.42.32.11.2

12.7

Number of seats

626049444846372827165347

24

% of seats

13.513.010.79.6

10.410.08.06.15.93.51.10.70.91.55.1

Source: Rocznik Statystyczny, 1993, p. 73.Others = parties with fewer than 2% of the vote, plus the German minority.

of these five groups, Olszewski received his highest level of approval from workers.Olszewski ran under the party Ruch dla Rzeczpospolitej, and was the leadingcandidate for the Catholic right and for those opposing unregulated marketisation.

In the second round24 Kwasniewski was able to mobilise support from what havebeen termed the 'socialist professions': clerical workers and professionals from thebudget-funded, state administration sector. Interestingly, Walesa was apparentlyunable to capitalise on Solidarity's organising role in the 1993 public sector strikes.Farmers also gave the edge to Kwasniewski. Walesa attracted the majority of votesfrom various segments of the private sector: service sector employees and en-trepreneurs. More conservative groups, notably pensioners and housewives, opted forWalesa. However, it is worth noting that students, typically thought of as UW voters,split their vote nearly evenly between the two candidates.

Economic voting in Poland's parliamentary elections

So far in this article we have examined whether and to what extent economic factorscan explain voting in presidential elections. Because the cabinet and the parliament—and not the president—are the seats of policy making, it is important to apply thesame economic voting methodology to Sejm elections. With smaller shares for eachparty and a more fractionalised vote, the total variation which can be explained bythese regressions is less than for the presidential elections.25 One reason is the sheernumber of small centre-right parties all receiving small shares. The 1993 election wasa brutal wake-up call for the anti-communist parties; continued resistance to consol-idation would come at the considerable risk of becoming peripheralised. Yet movestowards consolidation can appear to be more motivated by the desire to oppose theSLD than to represent those interests currently without a parliamentary presence.

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014


October 1991 Sejm elections

Table 8 lists those parties which received more than 2%. The electoral system usedin 1991 resulted in a highly fractionalised parliament. The post-election fragmentationwas aggravated by the near-constant shifting of political alliances and groupings ofdeputies, particularly on the post-Solidarity right. Not only did numerous representa-tives not have any clear party identity, but there was apparently little continuity withthe interests of those who voted them into office. The Olszewski and Suchockagovernments were large coalitions, numbering seven partners at the time of theno-confidence vote in July 1993. While, as stated earlier, the income-unemploymentregressions are weaker than for the presidential elections, amongst the larger vote-gainers there are discernible patterns of economic voting.

The major features in the Polish economy in 1991 were the financial crises in thestate enterprise sector and the state budget. By the autumn of that year Poland wasnearing the depths of the transitional recession. Output and real wages were stillfalling, and the unemployment rate was escalating. Farmers, state sector employees,health sector workers, and of course pensioners were all voicing increasingly stridentopposition to fiscal austerity and structural adjustment.

In this environment the lower level of support for UD is not surprising. UD neverfully recovered from this shock. The only significant result is the link between higherincome and stronger support of UD, which is consistent with Mazowiecki's electoratein the 1990 presidential election. The coefficient for unemployment is negative, butsmall and with a weak significance level. Overall less than 20% of the UD vote canbe explained by these two variables. The pattern of coefficients is the same as forMazowiecki in 1990, positive for income and negative for unemployment, indicatingthat support for the pro-reform policies of the UD and KLD came from the winners,or at least from those who had lost less. Overall, the R2 for UD in 1991 is lower thanthat for Mazowiecki in 1990 (Table 9).

While the KLD membership and electoral support were always small, the party wasdisproportionately influential in deciding policy in the 1989-91 government. Theywere in a difficult position by autumn 1991, leading a minority coalition governmentcommitted to reducing the fiscal deficit. Implementing austerity measures on top ofa deepening recession ('transformational' or not) may have created the impressionthat the UD-KLD liberals were pursuing this policy on principle, and despite the costsinflicted on society. I am not going to delve into issues of the relative medium-termcostliness of fast versus slow reforms. Yet while the maintenance of liberal economicpreferences during the emergence of recession and unemployment may be better inthe long run, the effect of fiscal crisis plus controversial distributional issues such asprivatisation, pension reform and the budget may have convinced the public that theKLD and UD had become ideologically committed to liberal reform, even if at theexpense of pursuing the socially optimal pragmatic policy.

Such a close identification with radical reform suggests that the natural KLDconstituency will be even more strongly identifiable through higher income and lowerunemployment rates than even the UD, which retained a substantial social-democraticwing. As for the regression, the KLD yielded the best overall results for 1991. Morethan 40% of the variation in support is explainable by the two independent variables.

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

1276 JANICE BELL

TABLE 9REGRESSION RESULTS FOR 1991 SEJM ELECTIONS

Variable Coefficient

UDCONSTANT -0.164INCOME 1.643UNEMPLOY -0.018

n = 48. R2: 0.195. Adjusted R2: 0.159.

KLDCONSTANT -0.210INCOME 1.625UNEMPLOY - 0.034

n = 48. R2: 0.417. Adjusted R2: 0.391.

PLCONSTANT 0.594INCOME - 2.676UNEMPLOY - 0.443

n = 45. R2: 0.301. Adjusted R2: 0.268.

KPNCONSTANT 0.088INCOME 0.006UNEMPLOY -0.151

n = 48. R1: 0.047. Adjusted R2: 0.005.

PSL-PLCONSTANT 0.351INCOME -1.476UNEMPLOY 0.098

n = 46. R2: 0.117. Adjusted R2: 0.076.

SLDCONSTANT 0.031INCOME 0.303UNEMPLOY 0.311

n = 49. R2: 0.108. Adjusted R2: 0.069.

PPPPCONSTANT - 0.023INCOME 0.285UNEMPLOY 0.076

n = 41. R2: 0.284. Adjusted R2: 0.246.

Standarderror

0.0880.5040.138

SEE: 0.038. F-ratio

0.0510.2910.080

SEE: 0.022. F-ratio

0.1220.6920.195

SEE: 0.052. F-ratio

0.0650.3710.102

SEE: 0.028. F-ratio

0.1160.6590.180

SEE: 0.050. F-ratio

0.0860.4920.135

SEE: 0.037. F-ratio

0.0170.0960.026

SEE: 0.050. F-ratio

t value

- 1.8543.263

-0.134

(sig): 5.443 (0.0076)

-4.1245.579

- 0.425

(sig): 16.094 (0.000)

4.847- 3.864- 2.277

(sig): 9.036 (0.0005)

1.3490.015

-1.487

(sig): 1.120 (0.3351)

3.033- 2.240

0.546

(sig): 2.845 (0.0691)

0.3600.6172.310

(sig): 2.736 (0.0756)

-1.3812.9642.891

(sig): 7.543 (0.0017)

Significanceoft value

0.07020.00210.8940

0.00020.00000.6731

0.00000.00040.0279

0.18400.98790.1441

0.00410.03030.5880

0.72090.54060.0255

0.17550.00520.0063

All numbers rounded.Source: GUS data; Panstwowa Komisja Wyborcza, Wyniki Wyborow do Sejmu RzeczpospolitejPolskiej: 27pazdziernika 1991r., Part II (Warsaw, 1991).

The coefficient for income is close to that for UD but even more highly significant.Like the UD vote, the link with unemployment is negative but with a very lowreliability. The prediction for KLD was right in assuming that this party displays a

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014


strong positive relation to income and a negative weaker dependence on unemploy-ment rates.

Considering the split of Solidarity in 1990, what happened to the other post-Solidarity parties in the 1991 election? Wyborcza Akcja Katolicka (WAK, or CatholicElectoral Action), the leading partner of which was ZChN, came third in the pollswith 8.7%. Next in the election results was Porozumienie obywatelskie Centrum(Centre Alliance, abbreviated to POC, then PC).

By this analysis, none of these electoral groupings and parties have electorateswhich are explainable through the regional economic indicators. The results are notfully provided in the text because they reveal little. In this regression, voting patternsfor WAK revealed a negative coefficient for income, — 1.12 with a significance levelof 0.115. The coefficient for unemployment was also negative, as with Walesa's votein 1990, implying more support amongst areas which had encountered less unemploy-ment. POC also showed a negative coefficient for unemployment (— 0.246,significance of f-stat 0.0342), but unlike UD and KLD, there was no significantrelation with income. This coefficient is positive but very low at 0.015, and the R2

is below 6%.Of the groupings related to Solidarity, the only one with a basic relation to

economic performance is PL, which was rooted in Rural Solidarity. While there is afairly significant negative link to higher unemployment, there is a very convincingsignificance level for the high negative coefficient for per capita income. This isconsistent with economic conditions in rural areas, and is similar to the pattern forZChN and POC. PL supporters may have been more likely to live in areas wherehouseholds were doubly hit by lower incomes and redundancies in regional industries.Unlike the results for PL, KPN appears to base little of its electoral support oneconomic issues. Income has almost no effect at all, and support is negatively relatedto unemployment. Voting for KPN has little basis in economic dissatisfaction, andmay find more support from people in more secure positions.

Despite the different historical associations of the PL and PSL-PL, the regressionresults from the PSL election results (Table 10) are similar: greater support frompoorer regions, but with a weaker impact from unemployment. The R2 is very low,in line with the other 1991 results, suggesting that more detailed analyses are neededto explain voting.

In the 1990 presidential elections the former PZPR member Cimoszewicz garnerednearly 10% of the vote. In both elections there was a discernible positive relationbetween unemployment levels and approval of the SLD or its candidate. In 1991, foreach 1% increase in the wojewddztwo unemployment rate, support for the SLD roseby one-third. However, this relation was only enough to explain about one-tenth ofthe vote for the former communist party. The coefficient for income is also positivebut is not highly significant.

Finally for the 1991 elections, even peripheral, semi-serious parties can be analysedby economic voting techniques. PPPP started out as a stunt, but over the course of thesemi-free Sejm became identified with a pro-market ethos. This party shared alow-tax ethos with Tyminski, as well as an apparently split electorate. While only aquarter of the vote is explained by the regression, the results for this party showed the

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

1278 JANICE BELL

best significance of coefficients. The impact of changes in unemployment and incomeis quite low, but it is interesting to note that it varies positively with both higherincome and higher joblessness. This is an interesting comment on the protest party;it appears to have attracted not only a proportion of the unemployed but, one assumes,a proportion of maverick entrepreneurs as well.

Once again, survey data can be used to compare socioeconomic voting patternswith the conclusions from the regression analysis.26 Although the total percentages ofthe vote for each party are not provided, this source provides information on moreparties than a CBOS poll taken two weeks before the election.27

The SLD attracted proportionally more votes from white collar workers, pensionersand clerical workers than from workers. PSL received its highest share of votes fromfarmers, primarily, and also people with lower education levels. These trends appearin weaker form for PL. UD gained moderate to high levels of support relative to theaverage from all groups except farmers. UD voters are concentrated amongst whitecollar employees and professions, clerical workers (10% more than SLD) andentrepreneurs. KLD displays similar patterns but with much lower overall results.POC attracted support from across many groups, re-emphasising the widespreadsupport for Walesa in 1990. WAK, on the other hand, has its electorate concentratedamongst pensioners and people with only elementary education. This is one elementwhich did not appear even in the extended regressions, but is important nonethe-less. KPN mobilised workers for their support, again re-emphasising that the anti-communist right finds votes amongst workers with lower incomes.

September 1993 Sejm elections

Even considering various arguments about the crystallisation of political opposition tothe 'social costs' of transition, the surge in support for the SLD in the 1993 electionstook many observers by surprise. After the initial panic in some quarters that marketreforms were in danger of reversal or at least serious impediment, many analysts havesince concluded that the SLD won votes on the basis of unity (compared with thefractious cabinet politics of the Bielecki, Olszewski and Suchocka governments) anda promise to proceed with reforms but with greater responsiveness to the needs of the'average Pole'. This responsiveness amounted more or less to a promise of entitle-ments, a moderate slowing of reform, and a more equitable distribution of costs andbenefits. Whether or not the coalition government which has remained in power since1993 has actually fulfilled these promises is another issue.

Table 10 provides a breakdown of the shares of votes and seats won in the pivotal1993 parliamentary elections. The post-communist SLD and PSL were able to forma government with a substantial majority. The return to power of the formercommunists was mirrored by the contemporaneous loss of power by pro-reformparties in Lithuania, Hungary and Russia.

Blazyca & Rapacki28 identified the key economic factors in the election to beunfulfilled high expectations of improved living standards, especially for urbanworkers and farmers. In this view, the ability of the SLD-PSL government to maintainpopular support despite adhering to a general continuity in macroeconomic policy,and in particular reduction of the budget deficit to below Maastricht requirements, is

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014


TABLE 10SHARE OF VOTES IN SEPTEMBER 1993 SEJM ELECTION

SLDPSLUDUnia PracyKatolicki Komitet Wyborczy 'Ojczyzna'KPNBezpartyjny Blok Wspierania ReformNSZZ 'Solidarnosc'Porozumienie CentrumKLDUPRSamoobrona-LepperaPartia 'X'German minorityOthers

000 votes

2815.22124.41461.01005.0878.4795.5746.7676.3610.0550.6438.6384.0377.5

84.2848.8

% of votes

20.415.410.67.36.45.85.44.94.44.03.22.82.70.66.1

Number of seats

1711327441—2216

—

——

31

% of seats

37.228.716.18.9—4.83.5——

——0.60.2

Source: Rocznik Statystczny, 1994, p.77.

very interesting. As the senior coalition partner, the SLD has controlled the economicsministries, which have been a source of sharp contention between the two parties.

Along the lines of the partisan model of political economy, there are indicationsthat the governing coalition has pursued some of its goals of easing the costs oftransition. The rate at which inflation rates are falling has slowed. Year-on-yearinflation for November 1996 was high at 17.1%. Inflation may not be falling quicklyenough, not least because of the maintenance of the creeping exchange rate peg, butat least it has become more stable. Even within a deficit of less than 3% of GDP, thegovernment has maintained a large role in the economy with a total budget stillequalling 54% of GDP in 1994. Transfer payments have increased as a share in totalgovernment expenditure, from 32.6% in 1993 to 36.4% in 1994. Total governmentdebt has also increased, from 12% of GDP in 1991 to 24% in 1994.29

At the time of the election the SLD appeared to be largely indebted to urban,state-sector workers. Once in government, the party has resisted making openconcessions to workers, excepting a possibly more lax attitude towards tax arrears inthe enterprise sector. These tensions between the spending demands of key constitu-ency groups and the awareness of the need for economic stability have pervaded theSLD-PSL coalition. Cautious notice has also been taken of the SLD tendency towardscentralisation and strengthening the position of managers in state enterprises, epito-mised in the revised Law on Commercialisation and Privatisation of State Enter-prises,30 versus privatisation minister Kaczmarek's push to implement the massprivatisation plan and several important capital privatisations. This is indicative of thesomething-for-everybody nature of policy making under the PSL-SLD government:workers in threatened state-sector industries favour a continued state role in industry,other losers of transition favour higher social spending, while private entrepreneursand small-scale businesspeople prefer political stability and perhaps identify with theformer nomenklatura who are prominent in the business community. The PSL wantsto grant all sorts of subsidies to farmers,31 and the SLD has to struggle to balance

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

1280 JANICE BELL

TABLE 11REGRESSION RESULTS FOR 1993 SEJM ELECTIONS

Variable CoefficientStandard

error t valueSignificance

of t value

SLDCONSTANTINCOMEUNEMPLOY

n = 47. R2: 0.262.

PSLCONSTANTINCOMEUNEMPLOYn = 17. R2: 0.125.

UDCONSTANTINCOMEUNEMPLOYn = 47./?2: 0.220.

KLDCONSTANTINCOMEUNEMPLOYn = 47. R2: 0.182.

OjczyznaCONSTANTINCOMEUNEMPLOY

-0.0110.3550.515

Adjusted R2: 0.228.

0.524-0.916- 0.040

Adjusted R2: 0.086.

- 0.0360.415

-0.162Adjusted/?2: 0.185.

- 0.0460.2140.013

Adjusted/?2: 0.144.

0.195- 0.270-0.197

0.0800.1940.132

SEE: 0.045. F-ratio

0.15403760.256

SEE: 0.087. .F-ratio

0.0660.1600.108

SEE: 0.037. F-ratio

0.0290.0700.048

-0.1431.8223.894

(sig): 7.808 (0.0013)

3.397- 2.436-0.157

(sig): 3.151 (0.0526)

- 0.5552.593

- 1.493(sig): 6.220 (0.0042)

- 1.6033.0500.276

SEE: 0.016. F = ratio (sig): 4.880 (0.0122)

0.0500.1220.083

3.892- 2.208- 2.372

0.88680.07530.0003

0.00150.01900.8759

0.58180.01290.1495

0.11610.00390.7835

0.00030.03250.0221


BBWRCONSTANT 0.053 0.029 1.811 0.0767INCOME 0.040 0.071 -0.556 0.5818UNEMPLOY -0.102 0.049 -2.106 0.0404n = 47. R2: 0.1159. Adjusted i?2: 0.0770. SEE: 0.0166. F-ratio (sig): 3.0164 (0.0588)

All numbers rounded off.Source: GUS, Panstwowa Komisja Wyborcza, Wyniki Wyborow do Sejmu Rzeczpospolitej Polskiej:19 wnesnia 1993 r., Part II (Warsaw, 1993).

more redistributionist views (notably those expressed by Leszek Miller while socialaffairs minister) with the pragmatic desire to maintain financial and economiccredibility in world markets.

Taking the income and unemployment voting function, let us start with the resultsfor the SLD. The former communists campaigned on a pragmatic rather thanideological basis. In a survey of parties' positions,32 the SLD claimed to be willingto cooperate with the PSL, UP and UD, while the last two parties were somewhatmore reluctant to cooperate with the former communists. Yet with the actual electionresults, the SLD-PSL agreement was the 'minimum winning coalition', in that it hadthe maximum preponderance of seats over a majority and the lowest number ofpartners, thus minimising future need for concessions to supporters.

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014


The R2 for the SLD is higher than in 1991, and the level of significance is alsomuch better. However, the connection with unemployment is tighter than for income.There is a positive relation with income, but this is outweighed both in terms ofcoefficient and significance by unemployment. Once again, the SLD vote is higherin regions more seriously affected by job losses. Furthermore, the rise in thesignificance of unemployment occurred as unemployment itself increased. Despiteconstant media attacks, in the polling booth the SLD became the legitimate recipientof the votes of those most adversely affected by economic transition.

Looking at the results for the other coalition partner, the share of votes for the PSLgrows in poorer areas. However, the relation to unemployment is even more tenuousthan in the 1991 election. The R2 for 1993 is almost identical to the very low levelsfor 1991. Again, we can conclude that these two variables do not capture the mostimportant factors for the PSL vote.

Next let us consider the results for the parties with the most consistent identificationwith economic liberalism, UD and KLD. Of the post-Solidarity parties, UD wasjoined in the Sejm only by BBWR and UP, and was the only one of these to gain asubstantial number of seats. The R2 is slightly improved for UD, but the level ofsignificance of the positive relationship with income is fractionally less than in 1993.However, there is a more distinguishable pattern of lower support in high unemploy-ment regions.

The results for the 1991 regression for KLD displayed the greatest explanatorypower. For 1993 this is much reduced. This party only received a small proportion ofthe total vote, and this perhaps helps to explain this poor result. However, the positivecoefficient for income remains statistically significant, if less so than in 1991.

In the 1993 elections Unia Pracy won 7.3% of the total vote. This party, whichattracted the social-democratic wing of the 1989 Solidarity coalition, is revealed bythe income-unemployment regression to attract more votes in areas of higher percapita incomes and lower unemployment rates: Warsaw and the urban centres ofwestern Poland. Since this time, UP has not capitalised on this early surge in support;in December 1996 it received approval ratings of 5%.33

The final political tendency to consider is the anti-communist right, which has beenidentified with the personalities of Olszewski and Walesa, but also increasingly withMarian Krzaklewski, the popular leader of NSZZ 'Solidarnosc'. Over time, the PC,ZChN and Solidarity have become more clearly identified with a conservative socialstance (pro Church, anti-abortion), anti-communism (continuing to push for lustrationlaws) and a more interventionist economic stand, at least concerning employmentpolicy and pensions.

Unlike the standard centre-right pattern, the voting results for BBWR in 1993reveal a positive, although low and insignificant, relation to income. However, whilethe generally pro-reform official stance of BBWR attracted a higher income bracketthan the Christian right, the significant link with lower unemployment rates is foundhere.

The Ojczyzna coalition continued in the same vein as WAK. As shown in Table11, their share of the vote is proportionately higher in regions with lower income buthigher employment rates, implying support from the working poor and employees instate enterprises. These are not the best significance levels, but still pass the 95%

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

1282 JANICE BELL

confidence test for one tail. They also show clearer economic links and a better, if stilllow, R2 than those for WAK in the 1991 election. While a sizeable proportion forChristian parties spans socio-economic groups, we can still discern the impact ofthese selected economic factors.

Each of the unrepresented anti-communist parties, including those not stemmingfrom Solidarity, such as KPN, received a very small share of the total vote. Onceagain anticipating in hindsight the current development of Solidarity Electoral Action,the shares of the votes gained by each party currently involved in AWS was summedand regressed against the economic data. The R2 is just below that for the ChristianDemocratic Ojczyzna electoral coalition, but is higher than that for the KPN, PC andSolidarity alone. Furthermore, the negative relation with unemployment is better thanthat for Ojczyzna, and the relation with lower incomes is also stronger.

In the run-up to the September 1993 election, CBOS conducted a survey34 of thevoting preferences of different social and occupational groups. In this poll the shareof the vote given to the SLD and PSL is lower than the actual result, and theproportion going to UD and BBWR is overestimated. This could be because of asample bias, a reluctance to admit voting for the former communists, or because therewas a swing towards the SLD and PSL in the last weeks of the campaign. As in the1995 presidential race, this could be evidence that the median vote is up for grabs.

Nearly every socio-occupational group divided their support between two adversar-ial blocs. Managerial and white-collar employees take the middle ground, giving thegreatest share of votes to UD, then UP and SLD. These groups all have a secular,social-democratic orientation, with UP being to the left of UD. More unskilled thanskilled workers are likely to prefer KPN and PSL. Lower wages and poor jobprospects may cause workers to prefer a more gradual pace of reform. Skilledworkers, who have better chances of finding private sector jobs at better wages, preferBBWR and then SLD. Pensioners displayed divided support for redistributionistparties (SLD, PSL) and anti-communist parties (UD, BBWR) The PSL received itsstrongest support from farmers, but farmers gave the second largest share of their voteto the centre-right, represented here by KPN and BBWR. Here we can sense theconflict between economic interests (PSL) and the nationalist, anti-communist views.

Interpreting the regression results

Unemployment matters for elections

So far in this article we have thought about how regional data can help to shed lighton the connection between unemployment, income and voting behaviour. Even fromthe first year of transition, there has been an identifiable variation in politicalpreferences according to regional levels of income and unemployment, with the latterbeing undoubtedly the most significant economic factor.

Areas which have sustained greater job losses during transition gave a falling shareof their votes to those candidates most closely identified with radical, market-orientedreform. As unemployment rose, so did levels of support for candidates running on amore redistributive platform. Likewise, people in wojewodztwa less affected byunemployment or falls in real income are more likely to support liberal parties.

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014


Furthermore, we can identify interest-based voting amongst peasant farmers. We candecisively conclude that, in contradiction to Kabaj & Kowalik (1995), interest-basedvoting models along the lines of those used in advanced market economies can beapplied fruitfully to Poland.

It is a long-standing tenet of political economy that interest groups will bedifferentially sensitive to inflation and unemployment, as well as to economic reformprogrammes such as trade liberalisation or privatisation. Phillips curve-type analysesargue that trade unions, the poor and lower-skilled groups will tend to be moreconcerned about unemployment than inflation, and therefore will tend to supportpolitical parties which emphasise employment policies.

The two questions which prefaced this article were whether and to what extentvoting patterns in Poland can be explained by economic variables, and whetherinterest-based voting patterns are emerging. The answer to both of these questions ispositive. There are clear and persistent patterns in political preferences according tounemployment and income levels. More detailed regressions and public opinion datareinforce the intuitive conclusions about the socio-economic constituencies of politi-cal parties.

Further evidence that political parties have economic constituencies is provided inTable 12, where the most important political parties have been categorised into fourtendencies. The first tendency includes the PSL, which is historically based in thecommunist-era United Peasant Party. The second tendency includes the non-agrarian,post-communist parties. The main players in this group are the reformed communistand now social democratic SdRP and the political coalition SLD, which includesSdRP and the communist-era trade union OPZZ.

The third and fourth tendencies can be categorised as having their origins in theopposition to the communist system. The third group is associated with the 'left wing'of Solidarity. After the 1990 split in Solidarity this branch was organised first intoMazowiecki's electoral coalition ROAD (Citizen's Movement - Democratic Action)and then was transformed into the political party UD in 1991. After the 1994 mergerwith KLD this tendency is represented by Unia Wolnosci. The fourth tendency is thatwhich is currently called the 'centre-right'. This group is dominated by the offspringof the trade union Solidarity and Walesa supporters. Yet it also includes Olszewskiand Moczulski's KPN. In this group are the parties which are either members of orcooperating with Akcja Wyborcza 'Solidarnosc', as discussed below.

Cleavages can be defined along several axes, and the four groups presented here donot reflect any agreed definitive schema. Lewis (1994), for instance, categorisedparties into three groups: the post-communist parties (SdRP/SLD, PSL), the post-Soli-darity parties (UD, ZChN, PC, KLD, PL, UP), and KPN. However, since KPN iscurrently cooperating with AWS, and since it shares some policy preferences with thecentre-right, the historically separate identity of KPN has been somewhat supersededby opposition to the SLD-PSL government and general anti-communist views, not tomention similar economic constituencies. In terms of constituency, PL is includedunder the heading of peasant parties. However, PL also has a strong anti-communistorientation, as demonstrated by its origins in Rural Solidarity and current participationin AWS. While PL is included on economic grounds in the first category, it may alsobe analysed in relation to AWS.

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

1284

Peasant

SLD

UD, KLD, UW

Anti-communist,Catholic

Others

JANICE BELL

TABLE 12RELATION BETWEEN REGIONAL VARIATIONS IN UNEMPLOYMENT AND VOTES

1990

Bartoszcze

Cimoszewicz

Mazowiecki

Walesa 1Walesa 2

Tyminski

__

—

_ **_ ***

+ ***

1991

PSL-PL +PL - * *SLD + **

KLDUDPOC - **WAK

KPNPPPP + *

1993

PSLPSL-PLSLD + ***

KLDUDPC +"S" - *Ojczyzna - **BBWR - *KPN - * *UP - * *

1995

Pawlak

Kwasniewski 1Kwasniewski 2Kuron

Walesa 1Walesa 2GronkiewiczOlszewskiKorwin-MikkeZielinski

+ *+ ***4. **

__ ***_ ***-_ **-

Note: Significance levels (one-tail test): * = 0.10, ** = 0.05, *** = 0.005.

For each of the four elections, the candidate identified with the party or tendencyis listed along with a sign indicating whether the income-unemployment regressionprovided a positive or negative coefficient. In other words, the sign indicates whethervotes for the specified candidate or party were positively or negatively relatedto regional variations. The asterisks designate the level of significance of thecoefficient.

Each of the political tendencies in Table 11 has a consistent—and in half of thesea statistically significant—relation to unemployment levels. This is especially true forthe two post-Solidarity tendencies. As for the positive PSL coefficient for unemploy-ment in 1991, this can be attributed to the party's success in gaining urban supportfor its redistributive and egalitarian policy statements. Since that time the level ofnon-agricultural support for the PSL has fallen.

The results are even stronger if we compare the two political forces which canrealistically compete for the median voter, SLD and the post-Solidarity parties of theright. Taking the election results for these parties (including only those parties withparliamentary representation), only WAK in 1991 did not demonstrate a statisticallysignificant relation to unemployment, although the negative sign is consistent withthose for the other centre-right parties. Table 11 demonstrates that the SLD hasbecome the legitimate recipient of the votes of the unemployed. The post-Solidarityparties and those of the right, including the nascent AWS electoral coalition, aredrawing in votes from regions with higher employment levels but lower per capitaincomes. Surprisingly, the rapid emergence of unemployment in Poland has not givenrise to greater support for radical populist parties. The most populist tendency of thesefour is the centre-right. Among politicians of this tendency, Olszewski is the mostpopular proponent of both anti-communism and anti-neoliberalism. Judging from hissix-month tenure as prime minister in 1992, electoral victory of this bloc may lead topressures for higher levels of protectionism and state intervention.

Party consolidation

As has been emphasised by Lewis,35 political parties do not develop in an institutional

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014


TABLE 13PROJECTED RESULTS FOR 1991 SEJM ELECTION.- 'UNIA WOLNOSCI'

Standard SignificanceVariable Coefficient error t value of lvalue

CONSTANT -0.359 0.116 -3.107 0.0033INCOME 3.184 0.660 4.823 0.0000UNEMPLOY -0.049 0.181 -0.271 0.7873


All numbers rounded to third decimal place.Source: GUS Panstwowa Komisja Wyborcza (1991).

vacuum, and are influenced by electoral laws and social cleavages as much as by thehistorical origins of political movements. The legislative basis for formation ofpolitical parties was not passed until August 1990. Yet by the 1991 election Poland'sparty system was highly fragmented, with more than 100 parties contesting thenational elections. Despite the proliferation of small political groups and movements,only a handful qualify as significant players.

The one tendency which has suffered from true fragmentation of its support hasbeen the anti-communist centre-right. In the 1993 elections the centre-right vote wasdivided between at least seven parties at national level. The price of this fragmen-tation was the failure of one-third of the electorate to attain representation in the Sejm.The one centre-right coalition running in 1993, the Christian Democratic Ojczyznacoalition, did not surpass the 8% threshold. Perhaps the centre-right needed toexperience loss of office to overcome its divisiveness, but the course of consolidationhas been slow.

Much of the explanation why consolidation of the fragmented right took severalyears and several electoral failures can be found in the changes in the rules of theelectoral game. For the 1991 parliamentary election, seats in the Sejm were dis-tributed by a proportional representation list system. The lack of vote thresholdsdiscouraged small parties from merging. Before the 1993 election the system waschanged to territorial representation for most seats, with a minority distributed by alist system. Furthermore, thresholds of 5% for parties and 8% for coalitions wereinstituted. There was hesitation in the centre-right because the total level of popularsupport for these parties was quite strong, but it was not clear how these votes wouldbe distributed. The election results were conclusive enough to convince even thepolitical entrepreneurs of the right that the switch from low-threshold proportionalrepresentation to high-threshold majoritarian seats now requires cooperation withinthe tendency. The parliamentary-cabinet structure of government, in which victoryallows the governing coalition considerable discretion in policy making, strengthensthe need for the opposition to fight on a unified front.

In 1994, Unia Demokratyczna and Kongres Liberalno-Demokratyczny merged toform Unia Wolnosci (UW). In 1995, former prime minister Balcerowicz was electedto succeed Mazowiecki as party leader. If consolidation of these two parties into onefront representing the most pro-reform forces is to work, then the combination of theirelectorates should show a stronger relation to lower unemployment and higher incomelevels.

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

1286 JANICE BELL

TABLE 14PROJECTED RESULTS FOR 1993 SEJM ELECTION: 'UNIA WOLNOSCI'

Standard SignificanceVariable Coefficient error t value oft value

CONSTANT -0.090 0.084 -1.063 0.2934INCOME 0.648 0.205 -3.155 0.0029UNEMPLOY -0.151 0.140 -1.084 0.2845


All numbers rounded to third or fourth digit.Source: GUS; Panstwowa Komisja Wyborcza (1993).

Table 13 adds together the votes for these two parties in 1991 and regresses themagainst unemployment and income levels, using the same method and data as in thepreceding section. This method approximates the potential vote for the current UWin the 1991 election. The coefficient for income doubles, and retains the very highlevel of significance of the KLD estimates. The coefficient for unemployment remainsnegative, even if the coefficient is relatively small. The R2 is closer to that for theKLD than UD, and thus the overall explanatory power of the regression is higherthan the average of the individual R2 values. This indicates that these parties hadsimilar constituencies, with a positive coefficient for income and a negative relationto unemployment. We can conclude that the political merger was logical and apositive step, in the presence of higher thresholds for parliamentary representation andthe electoral strength of the SLD.

The same exercise was performed for the 1993 election. The coefficient for incomeis lower, but retains its sign and significance. Income shows a higher negativecoefficient than the 1991 'UW figures, with improved significance of the t-stat. Whilethese results may appear to be weaker than those for 1991, the more importantobservation is that the consolidated vote share provides better results in coefficients,significance and R2 than the individual regressions for UD and KLD.

While our regressions show that there are grounds for concluding that UD andKLD electorates shared interests, and thus that their merger was conducted on soundprinciples, consolidation still has not led to a resurgence of popularity for UW.Between 1993 and 1996, support for UW was close to 10%. In January 1997 supporthovered around only 7%,36 and that same month sank below the parliamentarythreshold for the first time, to 4%.37 As unemployment rose, the UW and itspredecessor parties lost votes which they have been unable to win back. Survey datasuggest that UW is losing its electorate to the new formation of Akcja Wyborcza'Solidarnosc' (Table 14).

In the summer of 1996 the Solidarity trade union agreed to establish an electoralcoalition of the centre-right to mobilise its unrepresented votes. Under its popularleader, Marian Krzaklewski, Solidarity has aimed to join together ZChN, KPN, PCand numerous other small parties and groups.38 Krzaklewski has expressed hisintention eventually to transform AWS into a political party. It is still unclearwhether, like the Solidarity coalition of 1990, AWS represents more than a loosecoalition formed in common opposition to the SLD-PSL government. Given the

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014


TABLE 15PROJECTED RESULTS FOR 1991 SEIM ELECTION: 'AKCJA WYBORCZA WOLNOSCI "SOLIDARNOSC" '


CONSTANT 0.918 0.188 4.888 0.0000INCOME -2.481 1.073 -2.312 0.0254UNEMPLOY -0.596 0.293 -2.027 0.0486


All numbers rounded to third decimal place.Source: GUS; Panstwowa Komisja Wyborcza (1991).

diverse emphases on religion, nationalism and continued economic reform, there aredoubts whether AWS can formulate and stick to a coherent policy programme.However, most of these parties do share a conservative social outlook and a pro-tradeunion orientation, such as favouring job creation policies and slower privatisation tofurther workers' interests.

Within the present framework, we can ask whether there are identifiable economicinterests amongst supporters of the parties on the right. In the regressions for the 1991and 1993 parliaments, the votes in each wojewodztwo for the parties in AWS whichhave a national presence39 were totalled. Taking this total vote as an AWS proxy, itwas regressed against unemployment and per capita income. If consolidation of thecentre-right into AWS makes sense in terms of common economic interests, theresults should show improvement on the simple average of the individual parties'results.

The improvement in the results suggests that the very low shares of votes disguisedthe presence of voting trends. The higher level of support in regions with lowerincomes but relatively lower unemployment rates aligns with Walesa's electorate in1990 (Table 15).

The aggregated vote provides much more precise information on the economicelectorate of AWS. Approval of these parties is characterised by a significant negativerelation with income and unemployment, although income is the more decisive factor.In Poland, working-class organisations and trade unions tend to be socially conserva-tive. This helps to explain why the AWS parties have been able to mobilise the votesof blue-collar workers (Table 16).

TABLE 16PROJECTED RESULTS FOR 1993 SEJM ELECTION: 'AKCJA WYBORCZA "SOLIDARNOSC"'


CONSTANT 0.400 0.116 3.424 0.0013INCOME -0.488 0.285 -0.157 0.8757UNEMPLOY -0.572 0.194 -2.955 0.0050


All numbers rounded.Source: GUS data; Panstwowa Komisja Wyborcza (1991, 1993).

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

1288 JANICE BELL

If AWS continues to attract the support of about 20% of voters, as in December1996, then it stands a good chance of remaining the primary competitor for the SLD.Given the pragmatic, deal-making nature of the PSL, we can expect the peasant partyto continue making serious overtures to AWS. If AWS is able to make itself into morethan an opposition movement, it would probably become a labour party appealing toemployed trade unionists, rather than being a radical party of the unemployed.

One interesting conclusion from this survey is that the party system in post-communist Poland is definable along cleavages dating from the outset of politicalmobilisation.40 The two largest parties under the socialist system, the Polish UnitedWorkers' Party and the Polish Peasant Party, have reorganised and prospered underdemocracy. The two other tendencies as defined in this article have their roots inthe opposition, with the majority—but by no means all of them—linked to Solidarity.The first cleavage is between post-communist and anti-communist parties. A cross-cleavage can be defined between secular and Catholic-traditionalist tendencies.Among the post-communist parties, we can place the SLD in the former and the PSLin the latter category. As regards the anti-communist parties, UW is more closelyidentified with secular values of civil liberty and tolerance, while parties of thecentre-right tend towards greater emphasis on Catholic and traditional values. Excep-tions to this are the KPN, which is economically protectionist but secular (and whichpredates and is separate from Solidarity), and the small Union of Real Politics (UPR),which is closer to the classic libertarian model. Furthermore, these historical cleav-ages will continue to define the consolidating political system for the foreseeablefuture. As part of democratic development, these four tendencies are becoming moreclearly identified with particular economic interests and constituencies.

Prognoses for the 1997 Sejm elections

Our proposal that voting behaviour in the 1993 Sejm election was more complex thansimple pro-or anti-reform reactions receives further support from other transitioncountries. Taking the 1993 Russian elections as their case study, Whitefield & Evans41

asked whether the loss of electoral support for the most pro-reform parties should beunderstood as a protest vote or as part of an iterative progression towards interest-ledvoting. A protest vote would indicate that the electorate has accepted democratic andmarket principles but nevertheless votes to express a general dissatisfaction with themanner in which reforms have been enacted. On the other hand, if democratisationinvolves a learning process on the part of both voters and politicians, a greaterdiversification of the vote may indicate that voters who were initially ignorant of theimplications of radical economic reform have gained experience and gradually cometo identify their interests and learn how to vote accordingly. Whitefield & Evanspredict that, as part of an iterative process, Russian voters will continue to supporteconomic reform but of a less radical variant.

The interpretation of politics as a learning process leads to some realistic conclu-sions for the case of Poland. Parties have to learn how to represent their constituencyin as much as voters only understand what their interests are as information isrevealed over time. As the economy recovers, voters whose situation improves maymove to more pro-reform positions. Evans & Whitefield propose that more competent

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014


governments may increase support for reform. For Poland, Gibson & Cielecka predictthat economic growth will reduce the share of the vote going to the SLD.42

However, economic growth in itself has not been sufficient to reduce the SLD vote.At the time of the 1993 elections Poland was in its second year of solid growth. Ourmodel predicts that the shift in voting will occur when GDP growth leads to growthin employment. Therefore, electorates will shift as the economy undergoes structuraladjustment. In the first three years of the Polish recovery employment growth laggedconsiderably behind the pace of growth.43 By December 1996 the unemployment ratehad fallen to just over 13%. As the rate of job creation picks up, and if economicgrowth absorbs more unemployed labour, then the current success of AWS couldpotentially be translated into a stable electorate.

There are a few specific reasons why decreasing unemployment through jobcreation will benefit the anti-communist right rather than UW. Considering thesocio-economic composition of the unemployed (lower educational levels, morewomen than men), those who cease to be jobless will probably enter lower-paidemployment in light industry and services. If AWS is able to turn itself into a socialdemocratic labour party then it would be a much more natural association than UWfor these voters. UW voters tend to have higher education and higher incomes, andto be in employment already. Solidarity voters, on the other hand, work but tend tohave lower per capita incomes.

The inverse of this equation is that lower unemployment rates and higher wagesmay decrease support for SLD, even though this would be a fulfilment of theircampaign promises. More employed voters may decide to support AWS on thegrounds that the SLD have too much power in controlling the Sejm and presidency(although Kwasniewski is presently a member of neither SdRP nor SLD).

In February 1997 AWS received 29% support in an opinion poll, with SLD receiving22%, PSL and UW 12% each, ROP 10% and UP 6%.44 There were still several monthsbefore the autumn 1997 elections, and while AWS was then in the lead, victory is notguaranteed. Akcja Wyborcza 'Solidarnosc' consists of a wide array of small, indepen-dent groupings, not all of which are unanimously in favour of joining the coalition.KPN, for instance, recently split over exactly this issue, but with only 3% popularsupport this will not threaten Poland's political stability. AWS has been discussingcooperation with both UW and Olszewski's Ruch Odbudowy Polski at the local level,but any group which tries to unite Balcerowicz and Olszewski will inevitably faceserious tensions. Solidarity chief Krzaklewski is popular but relatively untested as apolitical leader. There are also questions as to whether AWS will be able to movebeyond opposition to SLD and agree upon a coherent set of policies. Although anythingcan happen between now and the general election, the indications are that Poland maybe heading towards a party system where SLD and AWS are competing for power, butwhere the PSL is likely to continue in its role as coalition maker.

Sussex European Institute, University of Sussex

The author would like to thank Tomasz Mickiewicz for his comments during the writing of thisarticle.

1 Mieczyslaw Kabaj & Tadeusz Kowalik, 'Who is Responsible for Postcommunist Successesin Eastern Europe', Transition (World Bank), 6, 7-8, 1995, pp. 7-8.

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

1290 JANICE BELL

2 Examples of this research, which often reach very different conclusions, include: BrankoMilanovic, 'Social Costs of the Transition to Capitalism', Working Paper 1165, World Bank. 1993;A. Szulc, 'Poverty in Poland during the Transition Period: 1990-1992 Evidence', Discussion PaperNo. 29, Polish Policy Research Group. University of Warsaw, 1994; World Bank, UnderstandingPoverty in Poland (Washington DC, World Bank, 1995); G.A. Cornia, 'Poverty, Food Consumption,and Nutrition During the Transition to the Market Economy in Eastern Europe', American EconomicAssociation Papers and Proceedings, 84, 2, 1994, pp. 297-302.

3 Geographical analysis of 1990 elections: J.J. Parysek, Z. Adamczak & R. Grobelny, 'RegionalDifferences in the Results of the 1990 Presidential Election in Poland as the First Approximation toa Political Map of the Country', Environment and Planning A, 23, 1991, pp. 1315-1329. See also:J. Gibson & A. Cielecka 'Economic Influences on the Political Support for Market Reform inPost-Communist Transitions: Some Evidence from the 1993 Polish Parliamentary Elections', Europe-Asia Studies, 47, 5, 1995, pp. 765-785.

4 The right wing of Solidarity was in opposition to the communist regime before politicalchange and democracy seemed even remotely possible. A logical conclusion is that former oppositionleaders are not only ideologically committed but also risk takers. The hesitation of the centre-rightand right-wing parties to give up a degree of autonomy in return for a greater probability of gainingseats may be outweighed by the possibility that a fragmented vote may leave one of these parties inthe position of kingmaker, thus gaining political power disproportionate to their membership andelectorate. This second strategy is riskier, but the rewards of the less likely outcome would be muchgreater than being one of several partners in an electoral coalition. For an application of this publicchoice concept to Poland see: T. Mickiewicz, 'Dobra publiczne, gra tchórza', Zycie Gospodarcze, 31October 1993.

5 This conflict between liberalisation and redistribution may be illustrated by the conflictbetween external and internal pressures on economic policy. While the present SLD-PSL governmenthas been rapidly passing legislation to liberalise Poland's banking and tax regulations in preparationfor joining the OECD (in August 1996) and eventually the European Union, recent changes to theprivatisation law have generated serious concern about the possibility of an increasingly intervention-ist role of the state in the economy.

6 While traditional, liberal economics examines economic behaviour in terms of utility, modernpublic choice texts tend to model behaviour in terms of preferences. Thus the utility function whichthe rational actor pursues can be defined in terms of preferences.

7 D. C. Mueller, Public Choice II (Cambridge and New York, 1989), p. 349.8 J. W. Smith, 'A Clear Test of Rational Voting', Public Choice, 23, 1975, pp. 55-67.9 D. C. Mueller. "The Voting Paradox', in C. K. Rowley (ed.), Democracy and Public Choice

(Oxford, 1987), p. 89.10 G. H. Kramer 'Short Run Fluctuations in US Voting Behaviour, 1896-1964', American

Political Science Review, 65, 1971, pp. 131-143.11 A. Downs, An Economic Theory of Voting (New York, 1957).12 D. R. Kinder & D. R. Kiewiet, 'Economic Discontent and Political Behavior: The Role of

Personal Grievances and Collective Economic Judgements in Congressional Voting', AmericanJournal of Political Science, 23, 1979, pp. 495-517.

13 CBOS, 'Poland's Most Pressing Problems', Polish Public Opinion, February 1994, p. 1(n = 1226).

14 R. Rose & C. Haerpfer, 'Fears and Hopes: New Democracies Barometer Surveys', Transition(World Bank), 7, 5-6, pp. 13-14.

15 R. C. Fair, 'Econometrics and Presidential Elections', Journal of Economic Perspectives, 3,1996.

16 A similar but less extensive analysis is in A. C. Pacek. 'Macroeconomic Conditions andElectoral Politics in East Central Europe', American Journal of Political Science, 38, 3, 1994, pp.723-744. Independent variables in this article are unemployment, education and lagged 1989Solidarity vote. Unemployment is found to be significant for the 1990 and 1991 elections, but theimplications for the political party system are relatively unexplored.

17 J. Bell & T. Mickiewicz, 'Unemployment and State Sector Insiders during the EconomicTransition in Poland', MOCT-MOST Economic Policy in Transforming Economies, 20, 1997.

18 Polls taken close to the date of the election are preferable to ones taken months or even yearslater. Shifts in popular approval of politicians and parties appear to cause voters to remember theirchoices differently as time passes.

19 See L. Balcerowicz, Socialism, Capitalism, Transformation (Central European Press, 1995),p. 298.

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014


20 A. Kublik & P. Pacewicz, 'Nasza wyborcza szesciolatka ' , Gazeta Wyborcza, 1 December1995, p . 2.

21 Fo r instance, the controversies over P S L connections with B G Z in both the recapitalisationprogramme and PSL renting office space from B G Z at absurdly low rates; constant PSL blockadingof any progress in privatisation, its constant lobbying for protection and tax breaks for the sole benefitof peasant farmers, etc.

22 No t to mention the recurring theme of anti-Semitism.23 'Jak Kwasniewski glosy zebral', Gazeta Wyborcza, 21 November 1995, p. 2 (n not given in

text).24 OBOP poll cited by Gazeta Wyborcza, 20 November 1995, p. 4 (n not given in the text).25 The number of cases also varies with the number of regions in which the parties ran

candidates.26 INFAS data quoted in 'Wyborcza Statystyka', Gazeta Wyborcza, 29 November 1991, pp. 2-3

(n not given).27 CBOS, 'Preference wyborcze spoleczenstwa na 2 tygodnie przed wyborami', Report No.

801/91, October 1991 (n = 850).28 G. Blazyca & R. Rapacki, 'Continuity and Change in Polish Economic Policy: The Impact

of the 1993 Election', Europe-Asia Studies, 48, 1, 1996, pp. 85-100.29 Ibid., pp . 9 3 - 9 4 .30 Which was finally passed in the late summer of 1996. This law provides for greater cabinet

and Sejm control over privatisations in sensitive and strategic sectors and places greater emphasis onchanging corporate governance structures rather than proceeding with sales and ownership transfers.

31 One example of the PSL influence is that in the middle of 1994 a price equalisation schemewas implemented. While these programmes are found in most industrial economies , it protectedPolish farmers by raising the price of food imports to the domestic level, thus eliminating undercut-ting by more efficient, mostly European producers.

32 Polityka, 17 September 1993.33 O B O P poll cited in Donosy, 17 December 1996 (n not given in text).34 Centrum Badania Opinii Spolecznej, 1993, 'Preferencje wyborcze spoleczenstwa pod koniec

czerwcu '93' , Report No. 1028/93, July 1993, pp. 12-13 (n = 617).35 P. G. Lewis, 'Political Institutionalisation and Party Development in Post-Communist

Poland', Europe-Asia Studies, 46, 5, 1994, pp. 779-799.36 OMRI Daily Digest, No. 16, Part II, 23 January 1997.37 OMRI Daily Digest, No. 19, Part II, 28 January 1997.38 J. Paradowska. 'Akcja ostatnej szansy' Polityka, 7 December 1996, pp. 2 0 - 2 4 .39 One omission is the lack of 1993 figures for B B W R . However , other analysis suggests that

the inclusion of the data would not counteract the results as shown.40 S. M . Lipset & S. Rokkan 'Cleavage Structures, Party Systems and Voter Alignments:

An Introduction' , in S. M . Lipset & S. Rokkan (eds), Party Systems and Voter Alignments:Cross-National Perspectives (New York, 1967), pp . 1-63.

41 S. Whitefield & G. Evans, "The Social Context of the December Elections in Russia: PublicAttitudes and the Transit ion Exper ience ' , mimeo, 1994.

42 Gibson & Cielecka, pp. 781-782.43 Bell & Mickiewicz.44 C B O S poll cited in Donosy, 17 February 1997.

Dow

nloa

ded

by [

Nor

thea

ster

n U

nive

rsity

] at

12:

05 1

3 N

ovem

ber

2014

unemployment matters: voting patterns during the economic transition in poland, 1990–1995

Documents