Advertisement

International Journal of Economic Policy Studies

, Volume 13, Issue 1, pp 193–216 | Cite as

Budgetary institutions with or without coalition government: political economy of parliamentary democracies

  • Makoto NakanishiEmail author
Research article
  • 46 Downloads

Abstract

In comparative political economy, two elements likely to cause fiscal profligacy have received particular attention from researchers: weak budgetary institutions and governmental fragility, especially the fragility related to coalition government. A recent trend in political economy studies integrates both of these aspects, stating that strong budgetary institutions can overcome the tendency of fiscal mismanagement often seen in coalition governments. However, these recent studies have dismissed an important point. Namely, a more rigorous implementation of “interaction effect” analysis might have led to the conclusion that effects of budgetary institutions could weaken under a single party government. Such argument, if it is true, can be contradictory to the results of the pioneering studies in this area, which acknowledged the effectiveness of budgetary institutions in preserving fiscal sanity in the case of a single or nearly single-party government (Hallerberg and von Hagen, in Electoral institutions, cabinet negotiations, and budget deficits in the European union. NBER, pp 209–232, 1999). This paper refutes the hypothesis of ineffectiveness of budgetary institutions on fiscal discipline in the case of a single-party government, by using two additional empirical strategies. One is the adoption of Error Correction Model, used especially for the study of long-run (steady state) effects, taking into full account the particularity of “institutions”. The other strategy involves enlarging the sample. Traditional samples consist of 15 EU countries [e.g. the database of von Hagen and other researchers (Hallerberg et al. in J Polit Econ 23(2):338–359, 2007)]. In this paper the sample was enlarged using the database of Fabrizio and Mody (Econ Polit 22(3):362–391, 2010), which includes East European countries with further addition of 4 more countries (Japan, Australia, Canada and New Zealand). The resultant sample of 29 countries covers most of the parliamentary democracies among industrial nations. Using this new sample, we found that budgetary institutions are still effective in restraining fiscal mismanagement under a single-party government, but such effectiveness may be weak. However, dividing all the available indices of budgetary institutions into two categories, one being insensitive to the number of governmental parties, the other dependent on this number, enables us to find a clearer effect in the former group even under a single-party government, while the latter category is found to be effective under a coalition government. So far the only explanatory background of the fiscal effectiveness of budgetary institutions was provided by the theory of Common Pool Resources. To make a new classification, we introduced two additional aspects to the theoretical background: one being Time Consistency, and the other being a Multi-Principal Model. A new theoretical classification of budgetary institutions suggests that for single-party governments we should concentrate efforts of budgetary reform to assure Time Consistency. We must have a good numerical fiscal target, top-down budgeting in agenda setting, and only rare use of supplementary budgets. For coalition governments, incomplete contracts concerning the multiparty agreement makes fiscal planning less effective. To overcome this, parliamentary and administrative budgetary procedures should be carefully controlled according to the theory of Common Pool Resources and Multi-Principal Model.

Keywords

Coalition government Budgetary institutions Error correction model 

JEL Classification

C23 D72 E62 E63 H61 

Introduction

After the financial crisis of 2008 most industrialized countries had to resort to discretionary fiscal policy, leading to soaring government debt and the need for fiscal consolidation. However, some countries, such as Germany, where debt outstanding per GDP decreased during the last couple of years, regained relative fiscal sanity. Thus, we should be aware that there is a difference of fiscal performance among industrialized countries, which is still prominent now, and was already considerable before the recent crisis.

Two sources of such difference in fiscal performance have received attention in the field of comparative political economy: one is budgetary institutions [26, 27, 40], and the other is governmental fragmentation, especially the fragility inherent in coalition governments [3, 6, 21, 33, 36, 37, 39]. Weak budgetary institutions have been thought to be responsible for poor fiscal management, while governmental weakness seen in coalition governments is likely to inflate governmental expenditure to satisfy conflicting budgetary demands of stakeholders. Recent attempts to integrate these two kinds of political economy research produced the thesis of mitigating the pervasive fiscal effect of coalition governments by means of good budgetary institutions [16, 32, 43].

Yet, these studies do not address an important point. The interaction between the number of governmental parties and the quality of budgetary institutions might mean that a country with good budgetary institutions is not able to restrain fiscal mismanagement in the case of a single-party government. If true, this finding would contradict the previous assertion of comparative political economy concerning the effectiveness of budgetary institutions in “delegation states” [25], which are typically countries with single-party government, or nearly so.

Our study is the first to attempt to reaffirm the effectiveness of budget institutions for sound fiscal management even in the case of a single-party government, while integrating all the implications of interaction effect analysis. To achieve this, we integrate two additional research strategies. Firstly, we adopt a single equation Error Correction Model and use it to explore long-term effects of “institutions,” rather than to deal with cointegration. Such use of this Model is justified because it takes time for institutions to influence and change economic and social situations. This “long-term” effect is hypothetical, because the marginal effect of one variable is measured in steady-state equilibrium, which does not exist in the real world. However, the concept of steady-state effects helps to represent the effect of one institution as a “tendency”, subject to constant environmental changes.

Secondly, we changed the sample of countries. The traditional sample of von Hagen and Hallerberg [26, 27, 40] is composed of 15 (originally 12) EU countries. We enlarged the sample to include 29 Parliamentary Democracies in order to improve the quality of estimation results1and to gain universality, as we expanded the traditional West European sample of von Hagen and Hallerberg to integrate East European countries with the help of the database of Fabrizio and Mody [22], and also added Japan, famous for its huge government debt, and non-European Westminster countries (Australia, Canada, and New Zealand), that have been so far dismissed despite their popularly “mediatized” budgeting style. Improving the sample should enhance empirical results, so far relatively weak, and correct the bias in interpretation arising from the use of the traditional narrow “European” sample.

Our estimation shows that budgetary institutions are still effective for constraining fiscal profligacy under a single-party government, although this effect may be somewhat smaller than in the case of coalition governments.

After confirming a positive (in the sense of contributing to fiscal discipline) effect of budgetary institutions in the case of a single-party government, we tested another hypothesis. Namely that it is possible to divide all budgetary institutions into two categories, so that one category of budgetary institutions is effective regardless of the number of government parties, while the other category is sensitive to this number and its effectiveness improves as the number of government parties increases. If this hypothesis holds, at least one category of budgetary institutions will not lose effectiveness towards fiscal discipline when we decrease the number of government parties. Thus, we can confidently recommend improving or fully reforming this category of budgetary institutions in single-party governments, concentrating our efforts on a more limited reform agenda. This is our second research interest, and we tested our hypothesis using the same estimation technique as the one adopted for all budgetary institutions. We also introduced some theoretical considerations to test the effectiveness of reclassifying budgetary institutions. Using only Common Pool Resources theory, as is done by von Hagen and other researchers [25, 42] leads to an oversimplification, so we introduce two additional theoretical aspects to enable more policy recommendations to be considered compared to previous studies.

Firstly, this paper reviews previous studies and analyzes the background of the problem. Secondly, it shows the empirical strategy that is especially suitable for the study of institutions, and the data that were used. Then, it tests the hypothesis of the conditionality of the effectiveness of budgetary institutions on fiscal discipline and shows the basic results. Further, we consider a theoretical hypothesis, which can justify our new classification of budgetary institutions. After introducing this theoretical background, we show that one category of budgetary institutions can be effective under a single-party government, while the other category is only effective under coalition governments. We close our paper by making some policy recommendations.

Verification of coherence of previous studies concerning the effectiveness of budgetary institutions for different types of government

The three papers cited above, namely Wehner [43], Martin and Vanberg [32], and De Haan et al. [16], have not really established the compatibility of the two arguments concerning the effectiveness of budgetary institutions and the risks related to coalition governments. Although these studies demonstrate the effectiveness of budgetary institutions in the case of coalition governments, at the same time they might imply the “non”-effectiveness of budgetary institutions in the case of single-party governments. If the latter turns out to be true, there is a risk of generating an argument contradictory to the literature surrounding budgetary institutions, because such literature insisted on the effectiveness of budgetary institutions in the case of both coalition governments and single party governments, though in a different manner [25].2 It is ironical that emphasizing the value of budgetary institutions for coalition governments simultaneously leads to downgrading their benefits for single-party governments, now that among the countries recently confronting fiscal crisis those with single-party or nearly single party government are particularly affected (Greece, Portugal, Spain, Ireland as well as Japan if judged by its huge outstanding debt).3

Whenever we can show that the effect of budgetary institutions is “positive” in the case of coalition governments (we say a “positive” effect provisionally, although its actual sign depends on the definition of fiscal performance—the effect being positive if we speak about budget balance and negative if we refer to government spending), and if the effect of budgetary institutions is considered to be linear, depending on the measure of government fragility, we have only three possibilities for the case of a single-party government: (1) the effect of budgetary institutions in the case of a single-party government is negative, (2) the effect of budgetary institutions falls into the area of uncertainty and appears to be insignificant, (3) the effect of budgetary institutions remains positive but its magnitude appears to diminish.

Using interaction effect analysis to integrate the budgetary institutions and the problem of coalition government, as is done in the previous studies, necessarily implies one of the three possibilities above. That is why any study which insists on the effectiveness of budgetary institutions in controlling public finance in the case of coalition government, should check whether the same is true in the case of a single-party government. Previous studies have failed to do so. To deal with this problem, a marginal effect plot of the effectiveness of budgetary institutions dependent on some indicator of governmental fragility should be explicitly used in the analysis. However, only the reverse version of the related diagram (impact of the number of governmental parties dependent on the score of budgetary institutions) appears in two of the three cited papers [32, 43]4. De Haan et al. [16] have included the diagrams necessary to check the effectiveness of budgetary institutions depending on several indicators of governmental fragility in their paper, but their diagrams show that budgetary institutions are effective only in the case of coalition governments.5 However, the authors do not mention this problem in their paper, confining themselves to asserting the effectiveness of budgetary institutions in the case of coalition governments.

Thus, we face an empirical challenge consisting of testing whether the budgetary institutions are effective in cases of both single-party governments and coalition governments. As we know that the analysis of De Haan et al. [16] was not successful in showing this, this task is not an easy one. This paper intends to overcome this empirical weakness by introducing a new empirical methodology and enlarging the sample.

Error correction model for institutional study

In this paper, we formulated the following empirical equations based on the Error Correction Model (ECM). Here, ECM is used mainly to deal with long-term effects of variables, rather than cointegration as is usually the case. ECM for long-term effects is typically formulated using two equations: an ordinary ECM equation for short-term effects (Bardsen type ECM [5]), and Bewley Transformation ECM [13] for long-term effects. The following equations have been used here:
$$\begin{aligned} \Delta Y_{it} & = \alpha_{0} + \alpha_{1} Y_{it - 1} + \beta_{0} \Delta B_{it} + \beta_{1} B_{it - 1} + \gamma_{0} \Delta F_{it} + \gamma_{1} F_{it - 1} \\ & \quad + \zeta_{0} \Delta B_{it} *\Delta F_{it} + \zeta_{1} B_{it - 1} *F_{it - 1} + \eta_{0} \Delta P_{it} + \eta_{1} P_{it - 1} \\ & \quad + \theta_{0} \Delta E_{it} + \theta_{1} E_{it - 1} + \mu_{i} + \lambda_{t} + \varepsilon_{it} \\ \end{aligned}$$
(1)
$$\begin{aligned} Y_{it} & = \alpha_{0}^{*} + \alpha_{1}^{*} \Delta Y_{it} + \beta_{0}^{*} \Delta B_{it} + \beta_{1}^{*} B_{it - 1} + \gamma_{0}^{*} \Delta F_{it} + \gamma_{1}^{*} F_{it - 1} \\ & \quad + \zeta_{0}^{*} \Delta B_{it} *\Delta F_{it} + \zeta_{1}^{*} B_{it - 1} *F_{it - 1} + \eta_{0}^{*} \Delta P_{it} \\ & \quad + \eta_{1}^{*} P_{it - 1} + \theta_{0}^{*} \Delta E_{it} + \theta_{1}^{*} E_{it - 1} + \mu_{i}^{*} + \lambda_{t}^{*} + \varepsilon_{it}^{*} \\ \end{aligned}$$
(2)

In the above equations, we use variables of fiscal performance (Y), budgetary institutions (B), governmental fragmentation (F), other political control variables (P) and economic control variables (E). We also introduce country and time fixed effects as ηi + λt to compose Two Way fixed effects estimation. We apply the usual F Test to reject null hypotheses of μi ≠ 0, λt ≠ 0, μi ≠ 0 AND λt ≠ 0 and confirm that we can safely use Two Way fixed effects estimation.6

Bewley transformation is considered convenient for assessment of the long-term effects of variables because it automatically reports standard errors without manipulation of short-term ECM. As this is an instrumental variable estimation, we set ΔYit as an endogenous variable, and Yit-1 as an instrumental variable [4, 15].7 When Bewley transformation is used, only long-term effects (marginal effects of level-independent variables) are considered, and the effects of first-difference variables are neglected for policy objectives. On the contrary, in the original ECM only short-term effects are considered and level-effects are neglected.

As far as we know, this paper is the first to use ECM for long-term effects in the political economy of budgetary institutions.8 We use this model for long-term effects because we are interested in the effects of institutions. In fact, political economy of government budgets is a study of institutions. Institutions do not change frequently and thus can be considered as predetermined, at least in the short-to-medium term [1]. However, a typical empirical work analyzing budgetary institutions adopts fixed effects estimation (without ECM), considering the effects of budgetary institutions on fiscal or other explained variables to be contemporary or lagged depending on the modeled lag structure. This conjecture seems unrealistic, and it is also difficult to say how long it might take for the real effects of such reform to become visible.

Institutions are formal and informal rules with related enforcement functions [35]. In this setting, institutional change is, usually and necessarily, incremental [35], so a certain time will pass until economic, fiscal or social performance is improved, especially because informal rules are included and the stakeholders also need time to adjust to the new environment.

Besides, ECM in institutional studies only suggests a “tendency”. The real effects are not necessarily readily visible, because real economy and society are constantly perturbed. It is possible that an effect of one institutional reform could be apparently erased by some environmental change. Since ECM only shows long-term (steady-state equilibrium) effects when all the deltas (changes of variables) are supposed to be null, adopting ECM for an institutional study is a reasonable choice. It may also overcome the weakness of empirical results often found in political economy of budgetary institutions when ordinary fixed effects estimation with or without lags is used.

Data used in this study

We gathered data for 29 parliamentary democracies for a relatively long time span from 1972 to 2010,9 although data availability is different for each country.10 The data for Central and East European countries are naturally limited in scope. With this in mind, the choice of an unbalanced panel is inevitable.

New database of budgetary institutions

One innovation attempted in this paper was to radically enlarge the scope of countries included in the sample. We adopt the database of Fabrizio and Mody [22] as the basis of our sample as it includes the data on East European countries (which was originally developed by Gleich [24] and Yläoutinen [45]). We also adopt their database, because this is the result of a more up-to-date research compared to the database of von Hagen and Hallerberg [26]. Then, we go further and append 4 more countries to the database of Fabrizio and Mody [22]. These countries are, firstly, 3 Westminster countries, Australia, Canada and New Zealand, and secondly, Japan. Thus, our sample is composed of 29 countries, and also reintegrates the data of France and Ireland not treated in Fabrizio and Mody’s database, but included in the database of von Hagen and Hallerberg.

Following the table format of Fabrizio and Mody (for classification of these features and other details of budget institutions, including the data of other countries, the reader is referred to Fabrizio and Mody [22]), our supplementation of the budget scores for these countries could be represented as follows (Tables 1, 2).11
Table 1

Indices of Budgetary Institutions

Source: Fabrizio and Mody [22]

Stage of budget process

No.

Budgetary indicators

Negotiation stage

N1

General constraint (spending, debt and balance…etc.)

N2

Agenda setting (whether it is MF or cabinet who is the first to decide the budget norm, etc.)

N3

Negotiation (bilateral, multilateral or all cabinet)

Voting stage

V1

Amendment limit

V2

Initial/final vote on total aggregates

V3

Vote of confidence

Implementation stage

I1

Intra-annual changes in budget law

I2

Transfer limit

I3

Carryover limit

I4

MF power to block expenditure if necessary

Table 2

Budgetary indices of 4 added countries (Plus 2 reintegrated countries)

 

N1

N2

N3

V1

V2

V3

I1

I2

I3

I4

Australia (85-)

2

0~97

1

2

0~82

0

0

4

4

1.28

2.56~85

1.33

4~85

0

Canada

2

0~94

4

1~94

4

0~94

4

0

4

4

1.28

3.2~92

1.33

4~86

0

NZ (85-)

3

0~93

2

1~93

2

0~88

0

4~95

0

4

4

1.92

3.2~88

4

0

Japan

0

2

0

0

0

4

0

3.2

2.67

0

France

4

4

4

4

4

4

0

4~97

1.28

1.92~05

1.33

4

Ireland

4

2~92

4

1~92

2

2~92

4

0

4

4

1.28

3.2~92

1.33

4~92

4

0~92

Source: The author’s own qualitative research which supplemented the database of Fabrizio and Mody [22]

As for Japan, [41] was also referred although we readjusted it to reflect local information, while for France and Ireland, the scores from Hallerberg et al. [26] are used (French I2 score was modified)

To create the supplementary data for 3 Westminster countries, we relied on qualitative research including our own fieldwork12 based on document research. Detailed information about our database and related data sources appears in the working paper of the author [34].13

Other control variables

After defining the variable of budget institutions (B), we define our variables of fiscal performance (Y), governmental fragmentation (F), other political control variables (P) and economic control variables (E).

Expenditure of general government as a proportion of GDP is introduced as a dependent variable (Y) for fiscal performance, following the study by Martin and Vanberg [32] taken from OECD Economic Outlook No. 92. This choice of government expenditure rather than budget balance as a dependent variable can be fully justified, since for budget balance, different complexities could enter into budgetary decision making. Balance is an intersection of two decision-making objects: revenue and expenditure. Using budget balance as a dependent variable supposes a very complex causal linkage and blurs the policy effects of institutions. Variables are directly related to expenditure rather than balance. For example, a leftist government might prefer a big government, but not necessarily prefer a deficit if it is conscious of the sustainability of the welfare state. All theoretical reflections, as well as the general weakness of the empirical results for budget balance, as remarked by Perotti and Kontopoulos [36], justify the use of government expenditure as a dependent variable.14

For governmental fragmentation (F) it is necessary to count the number of government parties. We used the database of Andersson et al. [2]. However, after carefully checking the correctness of this database, we had to manually correct it to create our own handmade database. The same is true for some other political control variables (P).

The list of our variables of political controls (P) is as follows; (P1) Ratio of the number of seats of government parties to total number of parliamentary seats (lower house), (P2) Ideology of the prime minister, (P3) Elections. The ratio of governmental seats to the total number of parliamentary seats is taken from our handmade database based on Andersson et al. [2]. The ideology data of prime ministers is derived from Manifesto Project Information (Volkens et al. [38]).15

We also added one dummy variable representing the occurrence of legislative or executive elections, including the year prior to the year of such occurrence (P3). This indicator is a kind of political budget cycle variable, which has been very important in the history of political economy research. We have created this variable using the database of the World Bank on political institutions originally developed by Beck et al. [10]. We have not only counted the year of legislative elections, but also that of executive elections since typically in “semi-presidential” regimes executive elections could have political effects similar to those of legislative elections. We also counted the year prior to the year of elections so as to take into account the political ambiance, which drives politicians to spend more before the elections.

Economic control variables (E) adopted in this study are as follows; (E1) Real GDP growth rate, (E2) Unemployment rate: first difference, (E3) Trade openness measured by all export and import divided by GDP: first difference. We believe that it is necessary to use the real GDP growth rate because we must control for the influence of economic performance on fiscal performance. We adopted the first difference of the unemployment rate as a “social” economic control variable. As an “international” economic control variable, we chose the first difference of trade openness. The first differences of these variables are chosen to assure weak exogeneity (with GDP). We used the World Bank Development Indicators as data source for real GDP growth and trade openness ratio. We relied on the OECD database of labor force statistics for unemployment rates. For several East European countries, which are not present in OECD databases, we used the data from the AMECO database.

Descriptive statistics

Table 3 shows the descriptive statistics of the dependent and independent variables adopted in our paper (in total 769 observations were used in our empirical analysis).
Table 3

Descriptive Statistics

Variables

Mean

SD

Min

Max

Expenditure

45.39

7.74

23.88

70.54

No. of gov. parties

2.37

1.36

0.00

8.95

Budget indices

0.51

0.16

0.22

0.88

Ideology

− 1.37

16.92

− 58.00

64.53

Seat share

0.54

0.95

0.00

0.86

Elections

0.32

0.47

0.00

1.00

Real GDP growth

2.73

3.05

− 17.95

12.23

Trade openness (D.)

1.27

6.86

− 38.63

43.84

Unemployment (D.)

0.12

1.35

− 5.40

10.20

SD standard deviations. D. stands for the first difference of related variable

Basic results

Using ECM with marginal effect plot and redefinition of variables in the regression table, we test our first hypothesis formulated as follows:

H1

Hypothesis of ineffectiveness of budgetary institutions for fiscal discipline under a single-party government: The effectiveness of budgetary institutions in enhancing fiscal performance is lost under a single-party government (in which case the effect of budgetary institutions might become insignificant or even harmful to fiscal discipline).

The focus of our estimation is the interaction between the variable of budgetary institutions and that of the number of government parties. More precisely, it is the conditional marginal effect rather than the interaction effect that is at stake in most “interaction” studies. Thus, focusing only on the coefficient of the interaction term in regression tables does not make sense. Studies of interaction effects typically use marginal effect plots for hypothesis testing [14]. Such a conditional marginal effect can only be confirmed when the whole range of constitutive variables is used.

However, to make any correspondence between marginal effect plots and regression tables clearly visible, we include one adjustment of constitutive variables. In regression, the coefficient of one constitutive variable of the interaction model usually represents its marginal effect only when the other constitutive variable is zero. In our case, the number of government parties can be zero only in the case of a caretaker government. So we change our definition of the variable of government parties subtracting one from the original variable [Models 2(-1) and 3(-1) in Tables 4 and 5]. We also subtract three from our original variable of number of government parties [Models 2(-3) and 3(-3)]. Three parties are typical for coalition governments, and as a matter of fact, this is the third quartile of the number of government parties, while one party is the first quartile of the number of government parties in the whole sample. So we can read our regression tables in terms of typical cases of single-party government and coalition government (three governmental parties). We can also read the results for variables other than constitutive and interaction terms. Although interaction (conditional marginal) effects should be evaluated using marginal effect plot, readers can also refer to our regression tables to check the whole range of results of our estimation.
Table 4

Influence of long term effects on expenditure (Bewley transformation ECMs with IV)

 

1

2 (− 1)

2 (− 3)

3 (− 1)

3 (− 3)

Ideology

− 0.06** (0.03)

− 0.06** (0.03)

− 0.06** (0.03)

− 0.06** (0.03)

− 0.06** (0.03)

Elections

2.75** (1.38)

2.74** (1.37)

2.74** (1.37)

2.69** (1.34)

2.69** (1.34)

Seat share

− 5.52 (5.32)

− 5.46 (5.39)

− 5.46 (5.39)

− 6.14 (5.43)

− 6.14 (5.43)

Growth

− 1.27*** (0.32)

− 1.26*** (0.31)

− 1.26*** (0.31)

− 1.21*** (0.31)

− 1.21*** (0.31)

∆ Trade

− 0.06 (0.12)

− 0.06 (0.12)

− 0.06 (0.12)

− 0.06 (0.11)

− 0.06 (0.11)

∆ unemployment

0.99* (0.55)

0.97* (0.55)

0.97* (0.55)

0.99* (0.54)

0.99* (0.54)

Parties(-1/-3)

1.27*** (0.46)

2.71** (1.29)

2.71** (1.29)

3.11** (1.35)

3.11** (1.35)

Budget

− 18.99*** (5.49)

− 14.06** (6.71)

− 20.08*** (5.40)

  

Ps.(-1/-3)*budget

 

− 3.01 (2.47)

− 3.01 (2.47)

  

Category 1

   

− 7.77** (3.35)

− 8.42** (3.82)

Category 2

   

− 3.40 (6.77)

− 10.39** (4.61)

Ps.(-1/-3)*category 1

   

− 0.33 (1.76)

− 0.33 (1.76)

Ps.(-1/-3)*category 2

   

− 3.49* (1.99)

− 3.49* (1.99)

AIC

5802

5798

5798

5765

5765

***p < 0.01, **p < 0.05, *p < 0.1, panel corrected standard errors in parentheses. Some abbreviations are used to represent interaction terms, such as Ps.(-1/-3)*budget (interaction between the number of government parties minus one or three and the variable of budget institutions), or Ps.(-1/-3)*Category 1 (2) (interaction between the number of government parties minus one or three and the variable of “category 1 (2)” budgetary institutions)

Table 5

Influence of short-term effects on expenditure (Bardsen ECMs without IV)

 

1

2 (− 1)

2 (− 3)

3 (− 1)

3 (− 3)

Ideology

− 0.00 (0.01)

− 0.00 (0.01)

− 0.00 (0.01)

− 0.00 (0.01)

− 0.00 (0.01)

Elections

0.25** (0.12)

0.24** (0.12)

0.24** (0.12)

0.25** (0.12)

0.25** (0.12)

Seat share

− 2.31** (1.03)

− 2.51** (1.05)

− 2.51** (1.05)

− 2.61** (1.09)

− 2.61** (1.09)

Growth

− 0.23*** (0.03)

− 0.23*** (0.03)

− 0.23*** (0.03)

− 0.23*** (0.03)

− 0.23*** (0.03)

ΔTrade

− 0.02 (0.01)

− 0.02 (0.01)

− 0.02 (0.01)

− 0.02 (0.01)

− 0.02 (0.01)

ΔUnemployment

0.18** (0.07)

0.18** (0.07)

0.18** (0.07)

0.19** (0.07)

0.19** (0.07)

Parties(-1/-3)

0.34*** (0.11)

0.92*** (0.29)

0.92*** (0.29)

0.95*** (0.29)

0.95*** (0.29)

Budget

− 3.35* (1.90)

− 2.14 (1.96)

− 4.66** (1.95)

  

Ps.(-1/-3)*budget

 

− 1.26** (0.56)

− 1.26** (0.56)

  

Category 1

   

− 0.89 (1.32)

− 1.62 (1.33)

Category 2

   

− 1.09 (2.13)

− 2.97 (1.81)

Ps.(-1/-3)*category 1

   

− 0.37 (0.43)

− 0.37 (0.43)

Ps.(-1/-3)*category 2

   

− 0.94* (0.53)

− 0.94* (0.53)

Adj. R2

0.22

0.22

0.22

0.22

0.22

Residual ADF

− 19.52

− 19.52

− 19.52

− 19.47

− 19.47

Breusch–Godfrey χ2 (1)

0.82

0.85

0.85

0.69

0.69

***p < 0.01, **p < 0.05, *p < 0.1, Panel Corrected Standard Errors in Parentheses. Some abbreviations are used to represent interaction terms, such as Ps.(-1/-3)*Budget (interaction between the number of government parties minus one or three and the variable of budgetary institutions), or Ps.(-1/-3)*Category 1 (2) (interaction between the number of government parties minus one or three and the variable of “category 1 (2)” budgetary institutions)

Model without interaction

Before checking marginal effects of budgetary institutions conditional on the number of government parties, we must take a look at the basic fitting of our ECMs without considering the interaction effect. Properties of most of the control variables are also mentioned in this basic estimation since the general tendency is the same for models with or without interaction.16

First of all, we have to check the effects of budgetary institutions (as a simple mean of all the budget institutions) and of the number of government parties, without considering the interaction effect between them (Model 1 in each table). Our study is mainly concerned with the long-term effects of political and budgetary institutions, which we estimate using the instrumental variable (Table 4). Then we turn to ordinary ECM estimation focusing on short-term effects (Table 5), which are likely to add nuances to the time path of institutional effects.

Model 1 of Table 4 shows that the number of government parties has a positive effect at 1% significance, meaning that when the number of government parties increases, so does government expenditure. Similarly, the mean variable of budgetary institutions has a negative effect with 1% significance, implying a clear fiscal control effect.

As for other control variables, the government ideology clearly causes reduction of fiscal outlays, having 5% significance (left wing parties are measured as negative, so further leaning to the left increases government expenditure). Elections have a positive sign, with 5% significance, implying that frequent elections increase government expenditure. The negative sign of the seat share variable implies that a strong government party is advantageous for fiscal discipline, but this effect is not statistically significant. As for economic variables, positive real growth rate is advantageous to contain government expenditure, as expected, and this effect is clear with 1% statistical significance. Other economic variables, like unemployment rate and trade openness, have a blurred effect.

In the short-term effect model, that is, the usual ECM shown in Table 5, we can see an almost similar tendency, but with some nuances. The number of government parties has the effect of increasing government expenditure with statistically strong significance (1%). Budgetary institutions also have a short-term fiscal discipline effect, although its statistical significance is limited to 10%. The sign of government ideology remains negative, but not statistically significant. On the contrary, the sign of elections is as expected, and has 5% significance for all models. Thus, we can think that both timing (short-term effect) and frequency of elections are strongly related to government expenditure. The seat share, which was not significant in the long-term estimation, becomes significant at 5%. We can interpret this short-term and often short-lasting effect as an immediate effect after an absolute victory of the government party in elections, or after the establishment of a minority government.

Economic growth has a significant influence in short-term estimations, while the sign of trade openness is as expected, but without statistical significance in short-term estimations. The unemployment rate is statistically significant at 5% in the short-term model and its sign is as expected. This represents the effect of a sharp change in the unemployment rate as we take the first difference of the unemployment rate. Considering the fact that people and the government are likely to react more to sudden than to gradual deterioration, this interpretation is reasonable from the point of view of the realities of political life. Finally, the coefficient of government expenditure (lag), which is also the coefficient of the error correction term, is negative and has the value of 0.15 with 1% significance for all models. This result is relevant from the theoretical perspective. The adaptation period, that is, the period from the moment, when the model diverges from equilibrium, until its return back to the equilibrium, is thought to be 6.67 years (speed of adjustment). This also means that on average budget reform needs 6.67 years to develop its full potential, as the coefficient of the related variable implies.

In the models from (2(-1)) to (3(-3)), that we describe in this and later sections, we can see the same tendency for control variables. Besides, with respect to our policy concerns, short-term effects are less important than long-term institutional effects and thus will not be mentioned in the descriptions that follow. Henceforth, we will mainly focus on marginal effect plots of long-term effect regression, and ask readers to refer to the regression tables to check the whole picture of estimation results.

Model with interaction

Next we are going to consider the model dealing with interaction between all budgetary institutions and government parties and test the aforementioned hypothesis of ineffectiveness of budgetary institutions for fiscal discipline under a single-party government.

Figure 1 shows the marginal effect of budgetary institutions on government expenditure, depending on the number of government parties.17 Including the error bands of 95%, we can see that the effect of budgetary institutions is always negative, at least from one party. The negative sign of the effect means here that appropriate budgetary institutions can decrease government expenditure.
Fig. 1

Marginal Effect Plot of The Model with Interaction: All Budget Institutions × Government Parties

Thus, in this section, we can conclude that budgetary institutions are effective in constraining government expenditure for any number of governmental parties, that is, both in single-party governments and in coalition governments. Thus we must refute the hypothesis H1 formulated above, which could be done with the help of Fig. 1. Readers can also refer to the models (2(-1)) and (2(-3)) in Table 4 and check the correspondence with Fig. 1. The coefficients of other variables are generally similar to those of the non-interaction effect model. The same is true of Table 5.

However, the marginal effect plot of Fig. 118 shows that the marginal effect of budgetary institutions apparently becomes smaller as the number of governmental parties decreases, even though the tangent is uncertain.19 Although we can still speak about the effectiveness of budgetary institutions for fiscal discipline, we cannot be sure as to the level of such effectiveness in a single-party government.

Does this mean that even though the effectiveness of budgetary institutions in the case of a single-party government cannot be denied, it is limited and less important? This may be so for all budgetary institutions considered together, yet we might be able to find subsets of budgetary institutions that are effective even under a single-party government, for which reformative efforts could be at least partly successful.

Next, we are going to reclassify all the features of budgetary institutions into two categories, one category having no relation to the number of government parties, and the other category being dependent on this number. To find appropriate classification criteria, we must further develop our theoretical background.

Interaction between the number of government parties and theoretically reclassified budgetary institutions

Next we test the following hypothesis:

H2

Hypothesis of theoretical reclassification of budgetary institutions: We can regroup all the budgetary institutions into two categories: one which is effective at least under a single-party government but may be or may not be effective regardless of the number of government parties, and the other one, which is effective at least under coalition government but may not be effective under a single-party government.

To test this hypothesis, we developed the following theoretical concepts and reclassified budgetary indicators according to these concepts.

Theoretical background

Using the database of von Hagen and Hallerberg [40], Hallerberg et al. [27], researchers have established the existence of an interactive relationship between the number of governmental parties (or the ideological gaps between these parties) and budgetary institutions. von Hagen [42] and Hallerberg and von Hagen [25] have used only one theory to justify the function of budgetary institutions, namely, the theory of Common Pool Resources (CPR).

A typical argument of the CPR problem in public finance was originally launched by Weingast et al. [44] in the context of pork barrel politics based on electoral districts. The logic is that n players could receive the benefits of the full amount of a grant in grant competition, while internalizing only 1/n cost for a grant financed from the national tax common pool resources. This risk is inherent in the public sector, whose basic characteristic is the discrepancy between benefits and costs, and between resources and expenditures. von Hagen [42] and Hallerberg and von Hagen [25] introduced this logic of the common pool problem in their political economy of budgetary institutions. In their model, instead of electoral districts each spending minister considers his or her private gain both from the achievement of a policy target and from the size of the budgetary allocation in his or her related field, the latter reflecting the political support of his or her constituency. At the same time the burden of taxation that they take into consideration is restricted to his or her constituency rather than society as a whole. In other literature on coalition governments where the CPR logic is adopted, the n players are more likely to be government parties, rather than spending ministers [36].

However, if the effect of constraining the mechanism of CPR depends on governmental fragility, and CPR is the only theory which can explain the effectiveness of budgetary institutions, we are not likely to find budgetary institutions that are effective irrespective of the number of governmental parties.

As an alternative, we introduce two additional theoretical concepts to explain the effects of budgetary institutions: Time Consistency and Multi-Principal Model.

Time Consistency suggests that we must establish a binding rule to prevent the government from changing an originally adopted promise by finding an opportunity to make use of the changed environment ex post [20, 30]. Often adopted to justify the independence of central banks from government, Time Consistency also justifies the establishment of fiscal rules (N1). Time Consistency means that original rules should be kept until budget execution. Thus, it also implies that, first, cabinet decisions should respect an already established fiscal plan by top-down budgeting (N2), and second, budgetary decisions should not change during the year via adopting supplementary budgets (I1).20 Time Consistency is supposed to be important to all governments irrespective of the number of governmental parties.

However, it is difficult to explain governmental budgetary behavior dependent on the number of governmental parties only by CPR, because it generally focuses on governmental decision-making, and not on its actual implementation, which mostly occurs through administrative processes, which is not the proper “decision process”. Administration is the agent of the government, and the Principal-Agent model applies well in this situation. However, if the number of governmental parties increases, we have more than one “Principal”, so the risk of the agent’s (Administration’s) moral hazard increases. In such “Multi-Principal” situations [11, 17, 31], it is necessary to cautiously monitor and finitely regulate the behavior of Administration, in order to avoid governmental inefficiency due to the “Third Best” nature of “Multi-Principal” situations.

Since efficiency slack increases in “Multi-Principal” situations in the case of substitutable tasks,21 we have good reason to finely divide administrative tasks and budgetary envelopes prohibiting transfers between them (I2). Limiting inter-annual carryover also helps such fine control (I3). Ultimately, the ministry of finance needs to have a position to block ministerial budget execution if necessary (I4), which also leads to stricter budgetary control. On the contrary, easing these finite control mechanisms in “Single-Principal” (single party) situations might be neutral, since such efficiency slack would be weaker, and there is some rewarding effect of simplifying administrative procedures and reducing related costs.

Thus, we can reclassify all the budgetary indices from three theoretical backgrounds. CPR still explains the background of all voting stage indices (V1–V3) and cabinet negotiation structures (N3). Multi-Principal model explains all indices in the implementation stage except the index of supplementary budgets. These two theoretical concepts assume that the effects of budgetary institutions depend on governmental fragility (Category 2). Time Consistency justifies good fiscal targets, top-down budgeting in agenda setting, and limitation of supplementary budgets. Time Consistency is considered to be important irrespective of the number of governmental parties (Category 1) (Table 6).
Table 6

Theoretical reclassification of budgetary institutions

No.

Budgetary indicators

Theoretical group (category)

N1

N2

I1

General constraint

Agenda setting

Intra-annual change

A. Time consistency (1)

N3

V1

V2

V3

Negotiation

Amendment limit

Initial vote on total

Vote of confidence

B. Common pool resources (2)

I2

I3

I4

Transfer limit

Carryover limit

MF block

C. Multi-principal (2)

Budget institutions suitable for single-party government and those suitable for coalition government

Based on the above hypothesis, we will see that there is a difference between the two categories of budgetary institutions, depending on the number of government parties.

Figure 2 shows the marginal effect of category 1 budgetary institutions conditional on the number of government parties. This figure shows that budgetary institutions of this category are effective under one party, two parties and three parties with 95% significance band, but not significantly effective for four and more parties in terms of controlling fiscal expenditure. This means that the effect on fiscal discipline of budgetary institutions of category 1 is relatively universal over almost the whole range of governments, although uncertainty remains for coalitions of four or more parties, i.e. for very wide coalition governments.
Fig. 2

Marginal Effect of Category 1 Budget Institutions (N1,N2 and I1) Conditional on the Number of Government Parties

Figure 3 shows the marginal effect of category 2 budgetary institutions conditional on the number of government parties. We can observe that budgetary institutions of category 2 are effective for coalitions of three or more parties, while their effectiveness is uncertain for two or fewer parties because the range of outcomes with significance under 95% covers the horizontal zero line. This means that the effect for fiscal discipline of budgetary institutions of category 2 is limited to typical coalition governments, i.e. governments with three or more government parties. This might mean that the effectiveness of budgetary institutions of this category is strongly related to and is conditional on the number of parties. The effectiveness of this category for single-party governments is uncertain.
Fig. 3

Marginal Effect of Category 2 Budget Institutions (N3, V1-3 and I2-4) Conditional on the Number of Government Parties

Readers can refer to the general regression results of this model listed in Tables 4 and 5 for the models (3(-1)) and (3(-3)).

The above considerations lead us to conclude that countries which expect a single-party government to continue should focus on the reform of budgetary institutions imposing time consistency constraint (N1, N2, I1), rather than aiming to reform all budgetary institutions. To impose a time consistency constraint, we need to set a clear numerical target (N1) to enhance the function of agenda setting for pre-budget expenditure control (N2), and hinder supplementary budgets in order to keep a pre-established fiscal target through the fiscal year (I1).

On the contrary, countries that expect a multiparty coalition government to last should centralize the parliamentary budget process (V1–V3), and avoid liberalizing administrative budgetary management (I2–I4), as well as sufficiently centralize cabinet negotiations (N3).

Strategies for reforming governmental budgetary institutions

This study has investigated the hypothesis of non-effectiveness of budgetary institutions in the case of single-party governments, and our empirical test refuted it. Thus, we still need budgetary reform for the countries which expect that their single-party or nearly single-party governments would last. However, under our new classification, budgetary institutions have differing effectiveness depending on the number of government parties. Our results suggest that we can focus on budgetary reforms of a more limited scope both for single-party governments and coalition governments, although the reform strategies for these different types of governments will be different.

For countries that are likely to continue to have a single-party government (or a nearly single-party government, such as the Japanese conservative coalition of Liberal Democratic Party and a small party called “New Komeito (Clean Government Party)”), it would be safe to postpone the budget reform of institutions based on the CPR logic or the common agency problem. The same is true of other countries, where one can expect to have a lasting single or nearly single-party government (especially those with deteriorated public finances such as Greece, Portugal, Spain… etc.). One needs to consider budget reform only in connection with Time Consistency, and can skip parliamentary reform of budgetary institutions. For single-party governments, government and parliament can be considered to be acting as a single unit. Parliamentary disputes are harmless as long as the government has strong party discipline. Though one had better avoid giving more discretion to public managers under a coalition government, we have a more ambiguous effect from a “letting managers manage” type of reform under a single-party government. However, when there are other reasons to proceed with such managerial deregulation (for the sake of procedural simplification, for example), then it is acceptable to make this type of reform under a single-party government.

For coalition governments with more than three parties, budget institutions based on the CPR logic and the common agency problem may be effective. Italy is a typical case in this category, but some of the East European countries need relevant budgetary reform adapted to coalition governments. One of the typical parliamentary budget reforms, double voting (V2), is often adopted in coalition governments, but is rarely performed under single-party governments. Coalitions are often fragile and easily broken in the process of parliamentary disputes. Coalition partners can exercise their veto power so that fiscal management can be destabilized. A sensitive problem can make some parliamentary dissidents leave the government to join an opposition group. Thus, coalition governments need a particularly rigid parliamentary budget process. Although the power structure of cabinet negotiations should be sufficiently centralized, we cannot be sure about the effectiveness of budgetary institutions fulfilling the Time Consistency constraint or “governmental” budget institutions, except in coalition governments with less than 3 parties. Because of confusion triggered by lack of consensus among (too many) stakeholders, such budgetary institutions might be ineffective for coalitions with too many partners, however strong those institutions might be. Multiparty coalition is an incomplete contract, where detailed procedural rules are more important than general fiscal planning function.22

Conclusion

In this paper, we enlarged the traditional framework of European studies of budgetary institutions by considering data concerning 29 parliamentary democracies, and checked the long-term relationship between budgetary institutions and coalition governments, using Error Correction Model focusing on steady-state equilibrium effects. We refuted the hypothesis of ineffectiveness of budgetary institutions on fiscal discipline in the case of a single-party government, without neglecting all the important implications of interaction effect analysis.

Our results also show the effectiveness of an approach that distinguishes between reforms of budgetary institutions according to the government type, that is, an approach adopting a fiscal target strategy for single-party governments, and a parliamentary and administrative implementation strategy for coalition governments. The former strategy implies elaborated fiscal planning, top-down budgeting in cabinet decision making, and rare use of supplementary budgets. The latter strategy recommends a strict parliamentary budget procedure and detailed control of ministerial budget execution. We can therefore concentrate on one of the two strategies of budgetary reform depending on the perspective of governmental fragility (or non-fragility).

Footnotes

  1. 1.

    To have relatively clear results, it is important to enlarge the sample in institutional studies that aim at discovering inherently subtle institutional effects. For example, Hallerberg et al. [27] divided the sample into delegation states and commitment states and obtained relatively imprecise results. We avoided such a division of the samples.

  2. 2.

    Hallerberg and von Hagen [25] classified their sample of 15 EU countries into delegation states and commitment states. The former group is mainly composed of single-party governments (often with a majority electoral system), and the latter is composed of coalition governments (often with proportional representation). They considered partially different mixes of budgetary institutions for each group [26, 27].

  3. 3.

    The tendency towards fiscal profligacy in coalition governments has gathered a lot of attention and related research effort from (especially European) researchers, although in reality, the outstanding debt of countries of typically multiparty coalition governments (Nordic countries and Benelux) was relatively limited, while most of the countries with single or nearly single party government (Greece, Portugal, Spain and Ireland as well as Japan) accumulated substantial outstanding debt. The only exception is Italy. There is some discrepancy between “mediatized” public finance of coalition governments and the very serious fiscal condition of single-party governments.

  4. 4.

    The use of both the original figure and its “inverse” figure is recommended in Berry et al. [12] to check for possible inconsistency in the logic (policy implication).

  5. 5.

    In the case of a single-party government, the number of governmental parties is obviously "one", and the ideological gap indicator is zero, because there is no ideological gap between “different” governmental parties. De Haan et al. [16] have indeed shown that budgetary institutions are effective in keeping fiscal discipline when the number of governmental parties or the ideological range (“fragmentation” indicator or maximal ideological gap) is superior to one (zero), implying that the budgetary institutions are effective only for coalition governments (thus, not for single-party governments).

  6. 6.

    ECM has often been implemented with the cointegration relationship between nonstationary variables, but this is not the focus of this paper. In this paper’s settings it might be natural to limit the analysis to stationary data. We tested the unit root and used the panel ADF test, and applied both to dependent and all the independent variables of stochastic nature, as well as to the residuals of OLS estimation, which might detect unit root in variables, in cases without cointegration relationship. The results show that we should not worry about the non-stationarity in our study's setting (see Table 4 and 5).

  7. 7.

    Besides, ECM used in this study is a single equation version of ECM. However, using the single equation ECM approach, weak exogeneity test is required and implemented in this paper, without which exogeneity of independent variables cannot be assured.

  8. 8.

    Franzese [23] is an example of ECM study in the field of political economy of political (not budgetary) institutions, but it does not use ECM for long-term effects. Martin and Vanberg [32] use Autoregressive Distributed Lag Model for the study of budgetary institutions, but they do not deal with long-term effects either.

  9. 9.

    Since the data structure adopted in this paper is supposed to be “Time Series Cross Section” (long panel data), rather than usual “Panel Data” with a shorter time-horizon, we adopted Panel Corrected Standard Errors [7, 8, 9] for our estimation.

  10. 10.

    The data availability of the sample countries adopted in this paper is generally from 1972 to 2010, except Australia (90-), Bulgaria (99-), Czech (97-), Estonia (97-), France (79-), Hungary (96-), Ireland (91-09), Latvia (95-), Lithuania (98-), Luxembourg (91-), New Zealand (87-), Poland (96-), Portugal (78-), Romania (96-), Slovakia (95-), Slovenia (96-), Spain (79-).

  11. 11.

    NB We modified French I2 indicator to 1.28 from 2006 reflecting French LOLF budgetary reform.

  12. 12.

    July–August 2012 for Australia, September 2012 for Canada, and March 2013 for New Zealand.

  13. 13.

    Although the scores of budgetary institutions vary from 0 to 4 points for each case, we use these scores in our estimation by normalizing all the scores to 0-1 scores. Thus, any composite indicator gathering the whole or some part of these indicators is estimated as the average of these 0-1 scores.

  14. 14.

    Another reason to avoid budget balance is that it is likely to be linked to other variables, and violate the weak exogeneity condition of single equation ECM.

  15. 15.

    Recent studies [6, 27, 32] tend to integrate variables created by (a) weighting the original variable for each government in a calendar year by the portion of the year the government was in power and then (b) summing the weighted measures across all governments in that year. We follow suit and generalize this procedure to all political variables except the elections variable.

  16. 16.

    Multicollinearity was also checked for this model without interaction, and no problems were detected. However, this does not exclude multicollinearity, which is often inevitable in models with interaction [14].

  17. 17.

    In all the marginal effects plots, the number of government parties is used without subtraction of one or three parties. Estimation results are basically the same for regression tables with variable adjustment.

  18. 18.

    As is claimed in Berry et al. [12], it is recommended to include the “inverse” version of marginal effects plots to confirm all the implications of interaction analysis, in this case, those of the impact of the number of governmental parties on government expenditure on the whole range of budgetary index. Such “inverse” diagram is omitted in this paper due to the page limit. The inverse version clearly shows that the bottom limit of the error band around the downslope line is above zero until around 0.6 of budgetary index. The author can provide the corresponding diagram on request.

  19. 19.

    Here, the interaction effect itself is not established because a straight horizontal line can be included inside the error band. Although the effect of budgetary institutions for single-party governments “may” appear weaker, we cannot clearly say, with statistical significance, that the effect of budgetary institutions “changes” depending on the number of government parties. Some discriminatory policy recommendations are only obtained in the following section by using theoretical reinterpretation and a new classification of budgetary institutions.

  20. 20.

    Originally von Hagen [40] mentioned “Time Consistency” to explain the importance of long term fiscal planning indices. Since then this argument vanished from the argument of von Hagen. We cannot use such long term fiscal planning indices as they are absent in the database of Fabrizio and Mody [22]. However, we could find elements interpretable in terms of “Time Consistency” (N1, N2, I1), among other dimensions of their database (and ours).

  21. 21.

    Dixit [18, 19] formalized this idea, based on the tradition founded by Holmstrom and Milgrom [28, 29], as follows. Given a linear reward system, which adds a bonus payment m per unit of xi (outcome of agency activity ai with normally distributed error term εi (its variance is ν, i is the number of tasks and principals, i = 1, … ,n)) to the base salary, and agent’s quadratic cost function, the variable part of reward m, summing for all the activities, is formalized as follows when there are n principals (supposed to be risk-neutral for simplicity) and k (−1 < k < 1) is designed to show substitutability (positive) or complementarity (negative) among tasks: m = 1/(1 + nrcν(1 + (n-1)k)), where c is a coefficient of cost function and r is the constant absolute risk aversion for the agent. When k is positive, the incentive m decreases dramatically with increasing n.

  22. 22.

    Hallerberg and von Hagen [25] have classified sample countries into delegation states and commitment states. For the former group detailed procedural rules are recommended. For the latter group, a fiscal planning approach based on multiparty agreement is recommended. As their dichotomy of delegation/commitment is loosely tied with single or nearly single-party government/coalition government, our conclusion and policy recommendations for budgetary reform are contrary to their argument. The main reason for Hallerberg et al. [27] to be skeptical about the effectiveness of procedural rules in coalition governments (commitment states) is that, using CPR logic, they regard such rules as being mainly related to the concentration of budgetary power in the minister of finance, who belongs to one of the coalition parties, so that other coalition partners do not desire such concentration of power. However, this schema, which is too abstract, should be avoided and procedural rules should be understood in their concrete meaning. A more precise examination of, say, parliamentary budget rules would lead to the conclusion that such rules are especially necessary for coalition governments.

Notes

Acknowledgements

The author is profoundly grateful to the two anonymous referees who have made very constructive and encouraging remarks, which helped very much to enhance the quality of the paper. The author is also grateful for the editorial efforts of the editor in chief, which were very helpful in completing this work.

References

  1. 1.
    Alesina, A., & Perotti, R. (1999). In J. von Hagen (Ed.), Fiscal institutions and Fiscal performance (Vol. NBER, pp. 13–36).Google Scholar
  2. 2.
    Andersson, S., Bergman, T., & Ersson, S. (2012). The European representative democracy data archive. Riksbankens Jubileumsfond. (In2007-0149:1-E). http://www.erdda.se.
  3. 3.
    Balassone, F., & Giordano, R. (2001). Budget deficits and coalition governments. Public Choice, 106(3–4), 327–349.CrossRefGoogle Scholar
  4. 4.
    Banerjee, A., Dorado, J. J., Galbraith, J. W., & Hendry, D. F. (1993). Cointegration, error correction and the econometric analysis of non-stationary data. Oxford: Oxford University Press.CrossRefGoogle Scholar
  5. 5.
    Bardsen, G. (1989). The estimation of long-run coefficients from error correction models. Oxford Bulletin of Economics and Statistics, 51(3), 345–350.CrossRefGoogle Scholar
  6. 6.
    Bawn, K., & Rosenbluth, F. (2006). Short versus long coalitions: electoral accountability and the size of the public sector. American Journal of Political Science, 50(2), 251–265.CrossRefGoogle Scholar
  7. 7.
    Beck, N. (2001). Time-series cross-section data: What have we learned in the past few years? Annual Review of Political Science, 4, 271–293.CrossRefGoogle Scholar
  8. 8.
    Beck, N., & Katz, J. N. (1995). What to do (and not to do) with time series cross- section data. American Political Science Review, 89(3), 634–647.CrossRefGoogle Scholar
  9. 9.
    Beck, N., & Katz, J. N. (2011). Modeling dynamics in time-series-cross-section political economy data. Annual Review of Political Science, 14, 331–352.CrossRefGoogle Scholar
  10. 10.
    Beck, T., Clarke, G., Groff, A., Keefer, P., & Walsh, P. (2001). New tools in comparative political economy: The database of political institutions. World Bank Economic Review, 15(1), 165–176.CrossRefGoogle Scholar
  11. 11.
    Bernheim, D., & Whinston, M. (1986). Common agency. Econometrica, 54(4), 923–942.CrossRefGoogle Scholar
  12. 12.
    Berry, W., Golder, M., & Milton, D. (2012). Improving tests of theories positing interaction. Journal of Politics, 74(3), 653–671.CrossRefGoogle Scholar
  13. 13.
    Bewley, R. A. (1979). The direct estimation of the equilibrium response in a linear model. Economic Letters, 3(4), 357–361.CrossRefGoogle Scholar
  14. 14.
    Brambor, T., Clark, W. R., & Golder, M. (2006). Understanding interaction models: Improving empirical analyses. Political Analysis, 14(1), 63–82.CrossRefGoogle Scholar
  15. 15.
    De Boef, S., & Keele, L. (2008). Taking time seriously. American Journal of Political Science, 52(1), 184–200.CrossRefGoogle Scholar
  16. 16.
    De Haan, J., Jong-A-Pin, R., & Mierau, J. O. (2013). Do budgetary institutions mitigate the common pool problem? New empirical evidence for the EU. Public Choice, 156(3–4), 423–441.CrossRefGoogle Scholar
  17. 17.
    Dixit, A. K. (1996). The making of economic policy: A transaction-cost politics perspective. Cambridge: MIT Press.CrossRefGoogle Scholar
  18. 18.
    Dixit, A. K. (1997). Power of incentives in private versus public organizations. American Economic Review, 87(2), 378–382.Google Scholar
  19. 19.
    Dixit, A. K. (2002). Incentives and organizations in the public sector: An interpretative review. The Journal of Human Resources, 37(4), 696–727.CrossRefGoogle Scholar
  20. 20.
    Drazen, A. (2000). Political economy in macroeconomics. Princeton: Princeton University Press.Google Scholar
  21. 21.
    Elgie, R., & McMenamin, I. (2008). Political fragmentation, fiscal deficits and political institutionalization. Public Choice, 136(3–4), 255–267.CrossRefGoogle Scholar
  22. 22.
    Fabrizio, S., & Mody, A. (2010). Breaking the impediments to budgetary reforms: Evidence from Europe. Economics and Politics, 22(3), 362–391.CrossRefGoogle Scholar
  23. 23.
    Franzese, R. J., Jr. (2002). Macroeconomic policies of developed democracies. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
  24. 24.
    Gleich, H. (2003). Budget institutions and fiscal performance in central and eastern European countries. Working Paper No. 215 (Frankfurt, Germany: European Central Bank).Google Scholar
  25. 25.
    Hallerberg, M., & von Hagen, J. (1999). Electoral institutions, cabinet negotiations, and budget deficits in the European Union. In J. Poterba, J. von Hagen (Ed.), Fiscal Institutions and Fiscal Performance, NBER, pp. 209–232.Google Scholar
  26. 26.
    Hallerberg, M., Strauch, R. R., & von Hagen, J. (2007). The design of fiscal rules and forms of governance in European Union countries. European Journal of Political Economy, 23(2), 338–359.CrossRefGoogle Scholar
  27. 27.
    Hallerberg, M., Strauch, R. R., & von Hagen, J. (2009). Fiscal governance in Europe. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
  28. 28.
    Holmstrom, B., & Milgrom, P. (1987). Aggregation and Linearity in the provision of intertemporal incentives. Econometrica, 55(2), 303–328.CrossRefGoogle Scholar
  29. 29.
    Holmstrom, B., & Milgrom, P. (1991). Multitask principal agent analysis: incentive contracts, asset ownership, and job design. Journal of Law, Economics and Organization, 7(Special Issue), 24–52.CrossRefGoogle Scholar
  30. 30.
    Kidland, F. E., & Prescott, E. C. (1977). Rules rather than discretion: The inconsistency of optimal plans. Journal of Political Economy, 85(3), 473–491.CrossRefGoogle Scholar
  31. 31.
    Martimort, D. (1996). The multiprincipal nature of government. European Economic Review, 40(3–5), 673–685.CrossRefGoogle Scholar
  32. 32.
    Martin, L. W., & Vanberg, G. (2013). Multiparty government, fiscal institutions, and public spending. Journal of Politics, 75(4), 953–967.CrossRefGoogle Scholar
  33. 33.
    Mierau, J. O., Jong-A-Pin, R., & De Haan, J. (2007). Do political variables affect fiscal policy adjustment decisions? New empirical evidence. Public Choice, 133(3–4), 297–319.CrossRefGoogle Scholar
  34. 34.
    Nakanishi, M. (2017). Newly developed database of budget institutions in 29 parliamentary democracies. CRES Working Paper FY2017-08. Center for Regional Economic Studies. Saga University.Google Scholar
  35. 35.
    North, D. C. (1992). Transaction Costs, Institutions and Economic Performance. Occasional Paper. No.30. Center for International Economic Growth.Google Scholar
  36. 36.
    Perotti, R., & Kontopoulos, Y. (2002). Fragmented fiscal policy. Journal of Public Economics, 86(2), 191–222.CrossRefGoogle Scholar
  37. 37.
    Rubini, N., Sachs, J., Honkapohja, S., & Cohen, D. (1989). Government spending and budget deficits in the industrial countries. Economic Policy, 4(8), 100–132.Google Scholar
  38. 38.
    Volkens, A., et al. (2012). The manifesto data collection. Manifesto project (MRG/CMP/MARPOR). Berlin: Wissenschaftszentrum Berlin für Sozial- forschung(WZB)).Google Scholar
  39. 39.
    Volkerink, B., & De Haan, J. (2001). Fragmented government effects on fiscal policy: New evidence. Public Choice, 109(3–4), 221–242.CrossRefGoogle Scholar
  40. 40.
    von Hagen, J. (1992). Budgeting procedures and fiscal performance in the European Communities. Economic Papers. 96. Commission of the EC directorate general for economic and financial affairs.Google Scholar
  41. 41.
    von Hagen, J. (2005). Fiscal rules and fiscal performance in the EU and Japan. IMES discussion paper no. 2005-e-5Google Scholar
  42. 42.
    von Hagen, J., & Haden, I. (1996). Budget processes and commitment to fiscal discipline. IMF working paper. WP/96/78. IMF. Washington, D.C.Google Scholar
  43. 43.
    Wehner, J. (2010). Institutional constraints on profligate politicians: The conditional effect of partisan fragmentation on budget deficits. Comparative Political Studies, 43(2), 208–229.CrossRefGoogle Scholar
  44. 44.
    Weingast, B. R., Shepsle, K. A., & Johnsen, C. (1981). The political economy of benefits and costs: A neoclassical approach to distributive politics. Journal of Political Economy, 89(4), 642–664.CrossRefGoogle Scholar
  45. 45.
    Yläoutinen, S. (2004). Fiscal frameworks in the central and eastern European countries. Finnish Ministry of Finance Discussion Paper. No. 72. Helsinki: Ministry of Finance of Finland.Google Scholar

Copyright information

© Japan Economic Policy Association (JEPA) 2018

Authors and Affiliations

  1. 1.Department of EconomicsSaga UniversitySagaJapan

Personalised recommendations