Economic growth as elections’ result: world experience

Yüklə 3.3 Mb.
ölçüsü3.3 Mb.
  1   2   3   4   5   6   7   8   9   ...   16
Economic growth as elections’ result: world experience


Alla Kobylianska

A thesis submitted in partial fulfillment of the requirements for the degree of

Master of Arts in Economics

National University “Kyiv-Mohyla Academy” Master’s Program in Economics


Approved by

Mr. Volodymyr Sidenko (Head of the State Examination Committee)

Program Authorized
to Offer Degree Master’s Program in Economics, NaUKMA


National University “Kyiv-Mohyla Academy”


economic growth as elections’ results: World experience

by Alla Kobylianska

Head of the State Examination Committee: Mr. Volodymyr Sidenko,

Senior Economist Institute of Economy and Forecasting, National Academy of Sciences of Ukraine

This paper provides estimation of influence of type of party in power on economic growth. It fits into discussion of different economic policies provided by different parties and different levels of economic growth attributed to them. To achieve the goal we attempt to extend the bunch of existing instruments used in the researches in the area. In doing so, measure of averaged GDP growths under different executives was suggested. Using the extensive set of econometric methods and tests it was shown that these instruments are robust and have good predictive power. Additionally, these instruments are in accordance with theory of economic voting and, thus, could be used in further researches. The construction of such instruments is particularly important in the situation of absence of common instruments for all countries.

Based upon the results of FE model with IV we found out that effect of left-wing parties on economic growth is higher than for right-wing ones, which is in accordance with results of previous studies. This relationship is supported by investigation of interregional and intergroup differences.

Table of Contents

List of Figures ii

Acknowledgement iii

Chapter 1. Introduction 1

Chapter 2. Literature Review 5

Chapter 3. Data description and methodology 10

A. Model specification 10

B. Methodology 14

C. Instruments’ construction 15

B. Data description 18

Chapter 4. Empirical tests and estimation results 22

A. Data discussion 22

B. Choice of correct specification 27

C. Lag for policy implication: How do we know who is in fact responsible for economic prosperity 34

D. Regional differences: Do they exist? 36

E. Do countries with different level of income differ from each other? 39

Chapter 5. Robustness check 42

A. 3 party-types vs 5 party-types 42

B. Balanced vs unbalanced data 45

Conclusions 50

Bibliography 51

Appendices A 56

Appendices B 58

Appendices C 61

Appendices D 64

Appendices E 66

Appendices F 70

Appendices G 76

Appendices H 79

List of figures


Figure 5.1 GDP growth under different executives in Europe 26

Figure 5.2 GDP growth in different geographical regions 26

Figure 5.3 GDP growth in High income OECD countries 39


To my brother devoted…

I would like to express the deepest appreciation to my thesis supervisor, Professor Tom Coupé for his support, giving me invaluable comments and reading my drafts. Especially I would like to thank him for responding on my e-mails during weekends, late in evening and after hard working days.

I also feel gratitude for Olexandr Shepotylo, Denis Nizalov and Mark Schaffer for their support during thesis preparation.

I thank Hanna Vakhitova, Olesia Verchenko, Pavlo Prokopovich, Sergey Maliar, Roy Gardner, Olena Nizalova and all other professors who reviewed my paper and gave me recommendations.

I thank to my father who does not stop believing that I am the best, cleverest and prettiest person in the World and to Artem Dishkant, I want to be the best one for whom. I thank my mother who always reminds me that the best is “an enemy of the better”. I also thank my classmate Victor Sannikov for being patient friend.


FEM. Fixed effect model

IV. Instrumental variables

LSDV. Least squares dummy variables model

OLS. Ordinary least squares

2SLS. Two stages least squares

REM. Random effect model

Chapter 1


The general theory concerning the subject of political economy and its influence on economic development is presented by the real business cycles theory (RBC) and its later extension in the form of political cycles’ theory (Lucas, 1977; Persson and Tabellini, 2000, 2002; Nordhaus, 1989). RBC theory investigates the main determinants of GDP fluctuations (among them are imported oil prices, labor force participation, innovations, environmental policy etc. –which are subject partly to governmental policy). Political cycle theory corresponds to politically oriented business cycles models, suggesting a decrease in inflation in the early periods of a party in office and an increase in inflation shortly before the new elections. As a consequence, there will be rises and falls in the economy. The empirical evidence so far indeed shows that the main economic indicators such as GDP, public debt, inflation, exchange rate, investment, change through the period before elections, in the year of elections and after (Verstyuk, 2004; Khemani, 2000; Alesina, 1987,1991; Rogoff and Sibert, 1987; Barreira and Baleiras, 2004).

Some evidence of an interaction between different aspects of election results, such as type of party in power, and economic growth also has been found. Indeed, opportunistic parties act so as to maximize the probability of their reelection stimulating the economy before the election periods and making social transfers bigger. Ideological (left-wing and right-wing) parties, on the contrary, tend to provide stable economic policies targeted either at low inflation or at low unemployment. Thus, in the latter case economic policy changes not due to launching of elections, but due to change of party in power (Nordhaus, 1989; Hibbs, 1977, 1987).

As it was mentioned above differences in ideologies mean differences in economic policies and methods of their realization, e.g. environmental protection and its measures, social protection (Neumayer, 2003, 2004; Dunlap et al., 2001). These differences also determine targeted groups (business players or population) and, therefore, economic growth. Moreover, the results of elections create certain reaction of foreign and domestic players who make their decision based on their expectations about future economic policy: for example, foreign investors may face greater risks in case of left-wing party in power (Vaaler et al. 2004). At the same time, economic development can also influence the type of party chosen (Alesina et al., 1987), thus leading to an identification problem (simultaneity).

In this paper, I will use worldwide data on elections results trying to estimate their influence on economic performance. The reason is that the above mentioned works have one common feature. All of them consider the effect of elections in a specific regions (mostly: USA, countries in transition, Europe), as this division captures similarities in their development. We use worldwide data to detect interregional differences in estimations results and compare them. The same estimation procedure describing relationship between party in power and economic growth will be used for each geographical region. The use of the same data for each region will help us to reach comparable results and derive conclusions based on them. To make formal conclusions on existence of interregional differences we will use least squares dummy variable technique interacting the regional dummies and the executive variables we are interested in.

We could expect interregional differences to appear based on several assumptions. Some economies may be less responsive to election shocks. First of all, this could be due to more developed self-regulation of an economy or due to absence of some entrepreneurs’ interests’ lobby. Countries, which governments do not use populist strategies, may be also less responsive to election shocks. All these differences in responsiveness could be the consequence of different length of history of democracy or of governance in whole. Some countries have bigger experience in holding elections (choosing certain party), as well as, in living after elections (providing some policies). It is commonly known that Europe and North America have the richest political history. Latin America was mostly Spanish colony and that is why could posses some distinct features. Asia represents different culture and, thus, probably different attitudes towards election process itself. In most Asian countries monarchies still exist. African countries are the poorest ones and have very little experience of self-government. Thus, we would expect different dependence of economic growth on party in power in different regions both due to differences in elections’ process and differences in experience with governance.

It will be shown later on, that the level of income determines the type of party chosen. Thus, we will also present results of our estimation analysis for countries grouped by level of income. Generally such grouping of countries is considered as alternative to geographical one. However, if, while considering such grouping as the main, we do not pay attention to other relevant countries’ division, we should be cautious about interpreting results.

Our research also differs in the way we will tackle the simultaneity between elections’ results and economic performance measured by GDP growth. Electoral decision will be instrumented by election outcomes in neighboring countries. Evidence that a specific political power in a neighboring country is performing well might give an incentive to the population in the home country to vote for such a type of party. It is reasonable to assume that parties in power may differ by ideological content in neighboring countries. However, a voter of a specific country makes its choice based on the performance of all these parties. Thus, we would instrument our voters’ decision by averages of economic performance of countries under specific type of party in power. Contiguity, as the essence of this method, seems to capture not only similar mentality of voters, but also features of regional development, which are common to all countries in the specific area. Similarities in economic development may make people think that experience of its neighbors could be applied in native country and bring the same economic results. In other words, economic performance of certain party in power in neighboring countries could form certain expectations of native population. Thus, we could expect that this is neighbors’ experience, e.g. Russian and not USA, which influences Ukrainian voters’ decisions.

The data for this research is taken from the World Bank Development Indicators1 (WDI), Corruption Perception index (, Freedom House and Data on Political Institutions (DPI). WDI database covers history of economic development of around 200 countries over the period 1960-2006 years (GDP, BoP, social indicators, financial indicators, etc). DPI covers 199 countries over 1975-2006 and contains information on chief executive variables, party variables in legislature, electoral rules, stability and check and balances. Variables of special importance for us are those related to chief executive variables, because these are the executives who are responsible for economic policy in the country and, thus, influence their development.

The paper proceeds as follows: first we make a literature review. In chapter 3 we suggest empirical methodology for estimation. Afterwards, chapter 4 summarizes the results obtained from estimations and chapter 5 provides results of some robustness tests of our model. Afterwards we provide overall conclusions.

Chapter 2


Our research investigates the interrelationship between the economic situation of countries and the elections results. The section contains the description of interrelationships between type of party in power and economic growth and possibility for simultaneity bias.

Maybe one of the first attempts to depict economic outcomes of different electoral decisions, specifically, the type of party in power, was made in the late 80ies and the beginning of 90ies. Alberto Alesina et al. (1989) consider a natural experiment, which the history of American democracy gave to economists. Based on assumption of voters’ rational expectations they investigated the situations when the government was formed by one type of party (e.g. Democratic) while the president was the representative of another one (Republican and vice versa)2. They looked to the changes in rates of unemployment and inflation attached to changes in the position of the parties (whether party forms government or its representative is president), as they differ by their ideological content3. Within the model, which was an extension to Fischer (1977), GDP was described as the outcome of expected inflation rate, which is, in turn, determined by the party in office and output growth. In addition, they formally characterize parties’ goals. Elections' result (incumbent change) was also estimated by expected inflation as well as by GDP. In addition, it was found out that elections’ result depends on current growth rate and changes in ideologies of parties and voters.

Using MLE procedure authors found significant relationships between political performance and GDP. Based on the results of Hausman test they conclude that their estimation does not suffer from endogeneity bias. Nevertheless, they mention the necessity to extend the research to long-term horizon, when “the voters are risk-averse with regard to the growth rate”, and to improve the quality of data which is rather short. The last fact follows from the article’s result that only president elections are significant. As they are held once in four years, this limits the number of observations.

Another highly relevant article is that of Nordhaus (1989), which also looks at the American experience. One of the results of this empirical investigation claims that under a conservative incumbent the unemployment rate is higher than the inflation rate and under a liberal incumbent the situation is the opposite.

A somewhat more global view on elections outcomes is represented by the exploration of economic history of different societies- socialistic and capitalistic. This approach investigates in depth the reasons of dissimilarities in societies’ development. One of the explanations is different ideologies. Indeed, Haan and Storm (2000) found out that the governments of these countries differ in the prevalent goals of their policies. Socialistic governments pay more attention to social protection of population, capitalistic- to protection of property rights. The protection of property rights, according to Haan and Storm, made a sufficiently larger impact on economy, which grows faster than in the case of execution of socially oriented programs.

Furthermore, Barro (1991) found the evidence for the economic development being dependent on the level of government interventions. Taking into account, that left-wing parties provide more of these interventions, this claims for a relationship between the type of party in power and economic growth.

In this paper we would like to test whether this conclusion holds. To do this we would include dummies for left-wing, right-wing parties and defining centrist parties as control group to compare whether influence of right-wing and left-wing parties on economic growth is different (coefficients before these left/right wing dummies should be statistically different in this case).

Some attempts to investigate channels through which influence of party ideology on economic growth is transmitted, were done in previous researches. Bjorsnikov (2005), developing the idea of Barro, in his article concludes that political ideology influences economic growth through the government size and legal quality. Government size negatively affects economic growth, legal quality- positively. Nevertheless, as the author mentions, the endogeneity bias between economic growth and political ideology could exist. Therefore, the author used Hausman test to check for endogeneity of political ideology. The test shows that hypothesis of endogeneity could be rejected.

Extending the subject, Bjornskov came to another interesting conclusion in his later article (2006): ideologically right-wing countries strengthen legal system and system of property rights in response to economic crisis. Moreover, he made distinction between government’s and citizens’ ideology and found out that citizens’ ideology influences size of government, legal quality and regulation issue, while government’s ideology influences only size of the government.

The author also mentioned other directions of ideology’s influence on economic growth: tolerance of right-wing ideology towards income inequality (Mitchell., 1993; Scott et al., 2001; Michelbach., 2003), influence of party in power on happiness (Di Tella et al. 1994) and in turn on GDP, the perception of trade-off between efficiency and equity basing on a shared ideology (Mitchell et al. 1993). Although, he did not concentrate on a specific region, he disregarded the possibility of existence of interregional differences in the results. In this paper we would try to fill in the lack of research in the area by estimating influence of party in power on GDP growth in each specific region.

As it was told at the very beginning, there exists a list of articles, which concentrate their attention on another side of elections-growth relationship. They consider economic variables as determinants of electoral decision. Belanger et al. (2006) in their research follow the experience of French elections and account for such independent variables as: ideological identification of population (left-right), party identification of society (left/right), religion and perception of economic well-being, - to estimate the intention to vote for left/right-wing party.

To control for endogeneity bias they use lagged ideological variables as instruments for estimation of vote intentions. Applying two-stage ordered logit estimation (vote intentions were ordered from 0 to 7- from left to right) to a system of equations, the scientists found out that ideological identification has more influence on vote intention than party identification. Besides, perception that economy is in bad state positively affects probability of choice of left-wing parties.

The other approach to the problem of electoral choice is based on the concept of strategic voting. Using micro data from survey conducted by University of Pittsburg, Canache et al. (1994) investigate voters’ electoral behavior in Honduras. Strategic voting corresponds to the situation, when a voter acts as to insure the election of certain party. They use economic status of respondent as economic variable, ranging it form low to high. After that, taking into consideration ideology of voter and using multinomial logit procedure, they try to estimate probability of wins of several Honduras parties which ranges from left to right. According to this research economic status positively affects the probability of switching from right-wing to left-wing party choice.

These researches were based upon the experience of specific countries, which are very different from economic and political point of view. The strength of these works is quite robust results and their policy applicability in mentioned countries. It is possible that such specificity was motivated by lack of data needed to conduct broader research and, linked to this lack, estimation problems. We would try, indeed, to search for the influence of party in power on economic growth throughout the world. Consideration of countries division based on geographical characteristics is important because of following, additional to already mentioned, arguments. This division would allow us to account for history of colonization and, thus, inherited from colonizer way of doing business and political system. There was shown that such dissimilarities may influence further economic development. Geographical situation also reflects into list of main trade partners (when the trade with closest ones is more extensive due to smaller transportation costs) and participation in different international organizations, which are mostly formed by neighbors and help to boost economic growth (e.g. through cessation of exporting quotes etc.). All this influences GDP growth. Therefore, these geography related features of economic development, if left without attention could lead to a bias in estimation results.

To sum up, significant evidence on the existence of a relationship between the ideology of an elected party and economic growth has been found. In most of the cases, the experience of specific regions showed that economic growth is faster under right-wing parties in office. Previous researches used various procedures to avoid simultaneity bias problem. The most often used one among them is implementation of such instruments as self-positioning of citizens in political spectrum.

At the end of the current research we expect to enrich the literature on the subject of non-economic determinants of economic development. As the data on cultural characteristics in different countries of the world is not comparable, we would try to create new instrument of economic voting. The specific contribution is the use of averaged neighboring economic growth as determinant of certain party being in office as an instrument for election results within a specific country. A rather rich data set, which covers around 11 years of worldwide experience (116 countries), will help us to broaden topic even further by considering cross-countries differences and patterns in the relationship between elections results and speed of economic growth.

Chapter 3


For the sake of logical ordering of paper let us start with an explanation of the regression specification. The data description and its sources will be presented later

In this research we would try to use next growth model suggested by Jones and Olken (2005)

,where j-country index, t- time index

Where Y-annual GDP growth, where GDP computed in constant prices (Jones and Olken, 2005) , xj- economic and institutional variables , political dummy- vector of political dummies equal to winner in election in specified country (e.g. right-wing party), j-country index. In such a kind of specifications coefficients before country dummy show us country specific effects and interaction terms - differences in effect of independent variable from country to country. Least squares dummy variable technique (LSDV) is used for estimation of such models, however, it is equivalent to FE estimation.

As we would like to look for interregional differences we should use regional, not country, dummies and their interactions with variables of interest, thus the regression transforms to

, where subscript j states for specific region.

The choice of independent variables is supported by empirical results and theoretical works. On of the variables of interest is corruption score. Dreher and Herzfeld (2005) found out that the corruption level negatively affects GDP: “an increase in corruption by one index reduces GDP growth by 1.13 percentage points”, as high level of corruption de-motivates possible investors and make allocation of resources ineffective. The article investigates the experience of around 70 countries over 30 years and uses corruption perception index as corruption variable.

The other variables to include are taken from fundamental works of Sala-i-Martin (1997). In the papers he tried to estimate the influence of different indicators on GDP growth and check for robustness of results looking for the whole distribution of coefficients got. He ended up with a list of indicators from which we will use:

  1. Sub-Sahara dummy

  2. Absolute latitude

  3. Religious variables: Muslims, Protestants, Catholics

  4. Former Spanish colonies dummy

  5. War dummy

  6. Rule of law, Civil liberties and Political rights.

It was found by Glaeser at al. (2004) that government effectiveness positively affects GDP growth. Their study was based on cross-country data, which includes observations on poor countries as well as on rich. Thus, results of their work could serve us as guidance for future conclusions.

Rule of law as another institutional determinant of economic growth was discussed in the article of Henisz (2000). He investigates institutional environment for economic performance of 157 countries over 1960-1994 years. Based on the results of OLS and GLS he found out that rule-of-law has significant positive influence on economic growth.

The influence of other two institutional variables: civil liberties and political rights, - was discussed by Gwartney et al. (1999) in the framework of the notion of political freedom. It was found out that improvement in political freedom positively affects per capita GDP growth. However, earlier Levin and Ranelt (1992) concluded that civil liberties insignificantly influence economic growth. Moreover, Barro (1989) noted that with political instability variables in regressions (coups, revolutions) such variables as political rights and civil liberties could become insignificant. Thus, the effect of these indices on economic growth is not straightforward.

We will also include civil war dummy as was suggested by Sala-i-Martin (1997). He argued that civil war negatively relates to economic growth as it lowers rate of return. Another work which investigated long- and short-run impact of civil war on economic growth is that of Murdoch and Sandler (2001). Based upon Solow growth model they considered cross-country experience and found empirical evidence that civil war has negative significant effect on country and its neighbors.

Religious variables, shares of people who belong to specific confession, were found to have impact on economic growth. Sala-i-Martin (1997) suggests considering them as country-specific variables, e.g. countries which poses huge oil reserves are mostly Muslim countries. Besides, currently religious variables are considered as indicators of specific ideology which could influence general attitudes to life, cultural values and, thus, influence economic development (Noland, 2005).

Among others geographically related variables, latitude influences economic growth in the sense that it shows the climate conditions and, thus, agricultural output and results in trade independence/dependence (Masters and McMillan (2004)).

Consideration of Sub-Sahara dummy and Spanish colony dummies as determinants of GDP growth in recent literature is provided both due to religion differences and differences in the level of growth in these groups of countries.

To sum up, we expect to get next sign of coefficient before independent variables in regression.

Table 3.1 Expected signs of independent variables

Group of variables

Variable name

Expected sign

Theoretical\ empirical justification

Religious affiliation

Share of Catholics


As found by Sala-i-Martin, supported by Noland (2005)

Share of Muslims


Share of Protestants


Governance indicators

Civil liberties


Gwartney at al. (1999)

Political rights


Corruption score


Dreher and Herzfeld (2005) found negative relationship between corruption and economic growth

Government effectiveness


Glaezer at al.(2004)

Rule of law


Barro (1996)

Political variables

Executive elections


Parties tend to boost economic growth spatially in the year of elections

Geographical variables

Absolute latitude

+ (far away from equator is good)

As found by Sala-i-Martin, supported by Grier (1997)

Spanish colony


Sub-Sahara dummy



Civil war

Murdoch and Sandler (2001)

  1   2   3   4   5   6   7   8   9   ...   16

Verilənlər bazası müəlliflik hüququ ilə müdafiə olunur © 2016
rəhbərliyinə müraciət

    Ana səhifə