Tải bản đầy đủ - 0 (trang)
4 TSCS Analyses of the Developments in 21 Democracies Between 1995 and 2010 on a Government Term Basis

4 TSCS Analyses of the Developments in 21 Democracies Between 1995 and 2010 on a Government Term Basis

Tải bản đầy đủ - 0trang

258 



J.L. GARRITZMANN



To avoid this pitfall, I opt for a very straightforward solution and

instead use government terms (i.e., cabinets) as the unit of analysis in the

present section and control for cabinet duration, which additionally allows

testing of whether and how government duration matters. Thus, all variables used in the previous section are transformed into government term

format. This is done by calculating the average over the respective years of

a cabinet term for each variable.5 I exclude governments that have been

in office for less than 1 year, as such a short term might not even allow

these governments to run through the whole legislative process of the

household budget procedure, making it problematic to assume that the

preferences of such “transitory governments” have measurable effects on

financial outcomes. As in the country-year setting, governments cannot

be expected to affect the tuition-subsidy regime of the same year, but of

the subsequent year; therefore, the data are coded to include a 1-year lag

(vis-à-vis longer lags). In contrast to the country-year approach, however,

the assumption of a 1-year lag is less crucial here, because the government

terms are much longer, i.e., the effects are simply expected to materialize

at some point within the government’s term (plus 1 year). The averages

are also less sensitive to exogenous shocks than annual observations, facilitating analysis even further.6

Overall, this proceeding leaves 112 observations of governments in

the 21 countries between 1995 and 2010 who were in charge for more

than 1  year. The country-year approach, used above and customary

in the literature, however, used 274 cases instead (i.e., N× T – ­missing

­values = 21 × 15 – 41), thereby artificially adding 162 “observations” without adding any variance on the independent variable. Thus, we could reason that the results of the “government term setting” used in this section

might diverge from the results of the “country-year standard approach”

applied in the previous section and that it provides much more accurate

estimates.

Concerning model specifications, I am again mainly interested in

short-term effects, i.e., how changes in the government composition affect

changes in the tuition-subsidy system. Therefore, I regressed—as above—

the first-differenced spending variables on the first-differenced partisan

composition of government variable by government term.7



QUANTITATIVE LARGE-N EVIDENCE 



259



Findings from the government term-based analysis are presented

in Table  5.5 for tuition fees and in Table  5.6 for public spending on

­subsidies. Beginning with the effects of major theoretical interest, both

tables reveal—and echo the findings of the country-year analysis earTable 5.5  Effects of government composition on tuition fees in 21 democracies

between 1995 and 2010: TSCS regressions, government term data

DiD models

(1)

Left-wing parties’ cabinet seat

0.001

share (first difference)

(1.03)

“Welfare state” parties’ cabinet

seat share (first difference)

Non-Christian center and

conservative parties’ cabinet seat

share (first difference)

Social democratic parties’ cabinet

seat share (first difference)

Trade openness (first difference) 0.001

(0.47)

Public gross debt (first difference) 0.001

(0.38)

Age ratio (first difference)

−0.079

(0.19)

National income per capita

0.000

(first difference)

(1.51)

Total public disbursement

0.013

(first difference)

(2.39)**

Share of females in the working

−0.002

population (first difference)

(0.23)

Cabinet duration (year count)

0.000

(0.31)

Constant

−0.011

(0.32)

0.21

R2

N (cabinets)

64

Number of countries

21



(2)



(3)



(4)



0.000

(0.49)

−0.001

(1.38)



0.001

(0.44)

0.001

(0.39)

−0.068

(0.15)

0.000

(1.41)

0.012

(2.36)**

−0.002

(0.24)

0.000

(0.54)

−0.016

(0.42)

0.20

64

21



0.001

(0.43)

0.001

(0.49)

−0.089

(0.21)

0.000

(1.65)

0.013

(2.59)**

−0.001

(0.15)

0.000

(0.46)

−0.015

(0.40)

0.22

64

21



0.001

(1.06)

0.001

(0.44)

0.001

(0.40)

−0.094

(0.22)

0.000

(1.53)

0.013

(2.41)**

−0.002

(0.25)

0.000

(0.33)

−0.012

(0.34)

0.21

64

21



Dependent variable: change in private household spending on higher education institutions between t and

t–1

DiD difference-in-differences, TSCS time-series–cross-section

* p < 0.1, ** p < 0.05, *** p < 0.01



260 



J.L. GARRITZMANN



lier—that there essentially is no effect of the partisan composition of

cabinets on countries’ private household spending on tertiary education

or on public subsidy spending between 1995 and 2010. This finding

holds ­irrespective of whether we test the influence of left-wing parties

Table 5.6  Effects of government composition on subsidies in 21 democracies

between 1995 and 2010: TSCS regressions, government term data

DiD models

(1)

Left-wing parties’ cabinet seat

0.000

share (first difference)

(0.80)

“Welfare state” parties’ cabinet

seat share (first difference)

Non-Christian center and

conservative parties’ cabinet seat

share (first difference)

Social democratic parties’

cabinet seat share (first

difference)

Trade openness (first difference) 0.000

(0.09)

Public gross debt (first

0.001

difference)

(1.25)

Age ratio (first difference)

0.140

(0.60)

National income per capita

0.000

(first difference)

(2.50)**

Total public disbursement

0.002

(first difference)

(0.25)

Share of females in the working −0.007

population (first difference)

(0.92)

Cabinet duration (year count)

0.000

(1.22)

Constant

−0.091

(1.47)

0.30

r2

N (cabinets)

64

Number of countries

21



(2)



(3)



0.000

(0.87)

0.000

(0.23)

0.000

(0.52)

0.000

(0.04)

0.001

(1.23)

0.162

(0.70)

0.000

(2.54)**

0.001

(0.20)

−0.008

(0.94)

0.000

(1.28)

−0.094

(1.51)

0.30

64

21



0.000

(0.19)

0.001

(1.17)

0.114

(0.47)

0.000

(2.54)**

0.001

(0.17)

−0.007

(0.88)

0.000

(1.28)

−0.093

(1.50)

0.30

64

21



Dependent variable: change in total public subsidy spending between t and t–1

DiD difference-in-differences, TSCS time-series–cross-section

* p < 0.1, ** p < 0.05, *** p < 0.01



(4)



0.000

(0.10)

0.001

(1.24)

0.127

(0.54)

0.000

(2.47)**

0.001

(0.21)

−0.007

(0.91)

0.000

(1.25)

−0.092

(1.49)

0.30

64

21



QUANTITATIVE LARGE-N EVIDENCE 



261



(i.e., communists, socialists, greens, social democrats) or “welfare state

parties” (i.e., left-wing parties and Christian democrats), of non-Christian center and conservative parties, or solely of social democrats. This

is entirely in line with my arguments about the increasing strength of

positive feedback-effects (Chaps. 1 and 6) leading partisan effects on the

tuition-subsidy regimes to become weaker and radical change increasingly unlikely over time.

One might object, however, that these non-findings are at least partly

due to the very low number of cases available to test the effects: the number of cases, which had already been small in the first place due to data

availability, was further reduced because of the first-difference operators.

Thus, very few degrees of freedom remain to estimate the effects. One has

to keep in mind, however, that exactly the same objection applies to TSCS

analyses on a country-year basis, but that these “standard approaches”

cover up this weakness by artificially inflating the number of cases. At the

end of the day, it remains a fundamental problem of data availability. Thus,

it would be highly valuable to re-estimate the models for longer time-­

series, if these data were available. However, my theoretical model predicts

that the results, i.e., the non-finding of partisan effects, should remain the

same for the more recent period.

As additional robustness tests, I tested other model specifications

(e.g., EC models, “levels-on-levels,” FE specifications), other combinations of control variables, and other operationalizations of the

dependent and independent variables (e.g., spending on grants/loans,

household expenditure only to private or only to public HEIs). The

results remain unaltered, i.e., during the period 1995–2010 there is no

evidence for effects of governing parties on the tuition-subsidy regimes

anymore.



5.5   Conclusion

This book argues that in order to understand the development and

sustainability of the Four Worlds of Student Finance, we need to

investigate the partisan composition of government and pay particular attention to the sequence and duration of parties in office. While

Chap. 3 provided qualitative evidence for this Time-Sensitive Partisan

Theory by tracing the developments in four diverse country cases over

the seven post-war decades, and while Chap. 4 tested the assumptions



262 



J.L. GARRITZMANN



of this argument by investigating party positions on higher education

finance empirically, the present chapter probed the argument in a largen setting by investigating determinants of tuition-subsidy systems in 21

countries.

Graphical and cross-sectional analyses of lagged data and TSCS analyses of the recent two decades reveal strong support for my arguments, as

­current public spending levels on subsidies and private household education spending were found to be strongly associated with the partisan

composition of government. In countries where left-wing parties predominated during the post-war decades, low-tuition–high-subsidy regimes

emerged. Vice versa, right-wing predominance led to high tuition fees

and low subsidy spending.

Moreover, the analyses show that governments’ leeway in re-designing the tuition-subsidy systems decreased over time. In line with the

Time-­Sensitive Partisan Theory (Chap. 1, Hypothesis  6), the government composition during the roughly four immediate post-war decades

is a much better predictor of the current tuition-subsidy systems than the

government composition during the more recent decades. This finding

is supported by TSCS regressions investigating the last 15 years (1995–

2010) in depth: across different model specifications and two different

designs (country-years and cabinets as the unit of analysis), the results

do not reveal any effects of governments on the tuition-subsidy systems

during the last 15 years—rather, the systems have followed further along

their respective paths. Unfortunately, data availability only allowed partial and indirect testing of the arguments; thus, future research could try

to make longer time-series data available in order to test the arguments

more fully.

The next and final empirical chapter provides the reason for these path

dependencies by demonstrating that strong positive feedback-effects have

emerged from the existing tuition-subsidy systems, which in turn make

departure from the respective regime paths increasingly costly for political

parties.



16.07

16.07

16.07

16.07

16.07

16.07

16.07

0.00



0.00



0.00



282

282

282

282

282

282

282

336



336



336



Private household expenditure on HEIs as

a share of GDP

Private household expenditure on public

HEIs as a share of GDP

Private household expenditure on private

HEIs as a share of GDP

All types of financial transfers from private

households to HEIs as a share of GDP

Public spending (all levels) on financial aid to

higher education students as a share of GDP

Public spending (all levels) on grants to higher

education students as a share of GDP

Public spending (all levels) on loans to higher

education students as a share of GDP

Cabinet seat share of left-wing parties

(communists, socialists, social democrats,

greens)

Cabinet seat share of “welfare parties”

(communists, socialists, social democrats,

greens, Christian democrats)

Cabinet seat share of conservative and

non-Christian center parties



Missing (%)



Observations



Variable



Table 5.A  Descriptive statistics for country-year macro data (Chap. 5)



Appendix



36.937



46.811



37.167



0.113



0.174



0.289



0.307



0.102



0.113



0.220



Mean



41.242



40.212



38.735



0.165



0.140



0.234



0.306



0.178



0.152



0.267



SD



0



0



0



0



0



0



0



0



0



0



Minimum



(continued)



100.2



101



100



0.714



0.843



0.973



1.277



0.735



0.619



1.277



Maximum



QUANTITATIVE LARGE-N EVIDENCE 



263



0.89

0.89

0.00



333

333

336



GDP gross domestic product, HEIs higher education institutions, SD standard deviation



Source: Author’s compilation, based on OECD-Stat and Armingeon et al. (2011)



0.00

089

1.19

0.00



336

333

332

336



Cabinet seat share of social democratic parties

Trade openness (exports + imports/GDP)

Public gross financial debt as a share of GDP

Age ratio (share of population ≥65 years/

share of population 5–29 years)

National income per capita

Total public disbursement

Share of females in the working population

(15–65 years old)



Missing (%)



Observations



Variable



29,610.25

45.060

60.519



35.081

73.334

54.508

0.497



Mean



8114.389

7.190

9.466



37.690

34.142

32.432

0.118



SD



13,487.82

31.278

32.524



0

16.750

4.922

0.274



Minimum



61,048.94

64.949

75.444



100

183.624

183.53

0.935



Maximum



264 

J.L. GARRITZMANN



QUANTITATIVE LARGE-N EVIDENCE 



265



Notes

1.The countries included in this chapter are Australia, Austria, Belgium,



Canada, Denmark, Finland, France, Germany, Ireland, Italy, Japan,

Netherlands, New Zealand, Norway, Sweden, Switzerland, UK, USA, and

Portugal and Spain after democratization. Data for Greece were missing.

2. The results do not change when different intervals are coded; see below.

3. The same countries as in the previous sections are used, plus Greece.

4. All estimations were conducted in STATA® 13.1 (StataCorp. LP, College

Station, TX, USA).

5.For example, if a government was in office from January 1, 1995 until

December 31, 1997, the dataset includes one entry, containing for each variable the average of the values for 1995, 1996, and 1997. If a government

changed during a calendar year, only those years are used to calculate the

averages during which the government held at least half a year in office

(June 30 is used as the cut-off point). As cabinets governing for less than

1 year are disregarded, no coding problems occur.

6. As an additional benefit, slightly weaker assumptions about (the imputation

of) missing values have to be made, because it is sufficient if one value is

available per government term.

7. I also regressed “levels-on-levels” and the results are unaltered. The same

holds for EC models. Results are available on request.



References

Armingeon, K., Knöpfel, L., Weisstanner, D., Engler, S., Potolidis, P., & Gerber,

M. (2011). Comparative political data set I 1960–2011. Bern: University of

Bern.

Beck, N. (2001). Time-series-cross-section data: What have we learned in the past

few years? Annual Review of Political Science, 4(1), 271–293.

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.

Busemeyer, M. R. (2006). Der Kampf um knappe Mittel: Die Bestimmungsfaktoren

der öffentlichen, privaten und sektoralen Bildungsausgaben im OECD-Länder-­

Vergleich. Politische Vierteljahresschrift, 47(3), 393–418.

Busemeyer, M. R. (2009). Social democrats and the new partisan politics of public

investment in education. Journal of European Public Policy, 16(1), 107–126.

Busemeyer, M. R. (2015). Skills and inequality: Partisan politics and the political

economy of education reforms in western welfare states. Cambridge: Cambridge

University Press.



266 



J.L. GARRITZMANN



Busemeyer, M. R., & Garritzmann, J. L. (2014). Globalization, compensation, and

social investment: A growing mismatch between public demand for and supply of

welfare policies?. Paper presented at the ESPAnet conference, Odense.

Garritzmann, J. L., & Seng, K. (2016) Party politics and education spending.

Challenging some common wisdom. Journal of European Public Policy. 23(4):

510–530.

Huber, E., & Stephens, J. D. (2001). Development and crisis of the welfare state:

Parties and policies in global markets. Chicago: University of Chicago Press.

Iversen, T. (2005). Capitalism, democracy, and welfare. Cambridge: Cambridge

University Press.

Iversen, T., & Cusack, T.  R. (2000). The causes of welfare state expansion:

Deindustrialization or globalization? World Politics, 52(3), 313–349.

Wooldridge, J. M. (2002). Introductory econometrics: A modern approach. Mason:

South Western.



CHAPTER 6



Individual-Level Attitudes Towards

Subsidies: How Positive Feedback-Effects

Prevent (Radical) Change in the Four

Worlds of Student Finance

6.1



INTRODUCTION1



The previous chapters focused on the origins of the Four Worlds of

Student Finance. In them, I probed my Time-Sensitive Partisan Theory,

both in qualitative in-depth analyses of four diverse country cases over

seven decades (Chap. 3) and quantitative analyses of party positions on

higher education finance (Chap. 4) and the effects of parties on the origin of the Four Worlds of Student Finance in a large-n setting (Chap. 5).

Overall, these analyses revealed broad support for the argument that the

Four Worlds of Student Finance can (only) be explained by investigating

the partisan composition of government, particularly the sequence and

duration of parties in office.

What is still puzzling, however, is the sustainability of the Four Worlds

during the more recent decades. Although parties had exercised enormous

influence on the development of the tuition-subsidy systems during the

1940s–1980s, shaping the systems according to their beliefs and interests,

recent decades have seen much less radical change (demonstrated in Chap.

2). Rather, countries followed further along their respective paths laid

out during the immediate post-war decades: high-tuition countries kept

increasing tuition, while low-tuition countries kept tuition low. Likewise,

countries with generous student support systems maintained or extended

these even further, while low-subsidy countries did not extend the subsidy

systems (cf. Fig. 2.5). In fact, it seems that no country witnessed radical

change in its tuition-subsidy regime after important decisions were made

© The Author(s) 2016

J.L. Garritzmann, The Political Economy of Higher Education

Finance, DOI 10.1007/978-3-319-29913-6_6



267



268



J.L. GARRITZMANN



in the 1970s and 1980s (with the exception of England; see Sects. 3.10

and 4.4). Thus, an important remaining question is why don’t the Four

Worlds of Student Finance change anymore?

The aim of this final empirical chapter is to offer an explanation regarding why the Four Worlds of Student Finance have become so sustainable

in the recent decades, i.e., so resistant to (radical) change. Following key

contributions in the recent historical institutionalist literature (for many,

see Pierson 1993, 2000a, b; Skocpol 1992), I argued in Chap. 1 that

the reasons for this is because the existing tuition-subsidy regimes create

strong positive feedback-effects on individual-level attitudes towards higher

education finance, thereby generating support for the countries’ respective paths and making (radical) policy change increasingly costly for governing parties. As Chap. 1 showed, this conclusion follows from both a

rational choice and a sociological perspective.

To underpin these claims empirically, this chapter investigates individuallevel policy preferences towards higher education finance. The chapter

has two main aims: on one hand, it studies individual-level determinants

of people’s preferences, thereby probing micro-level implications of my

Time-Sensitive Partisan Theory, such as that people’s left–right ideological standpoint affects their preferences towards higher education finance

(Chap. 1, Hypothesis 1); on the other hand, it tests whether the notion

of positive feedback-effects finds empirical support in the data, which

would help explain why the Four Worlds of Student Finance have become

increasingly stable over time (Chap. 1, Hypotheses 5 and 6).

The chapter thus poses and answers three questions: Who favors/

opposes subsidies? What are the determinants of the respective preferences? Does the existing subsidy regime have positive feedback-effects on

citizens’ attitudes on subsidy policies, generating support for the status

quo? Empirically, I concentrate on attitudes towards higher education

subsidies. While it would be equally interesting to investigate preferences

regarding tuition fees, I focus on subsidies, firstly, due to space limitations, secondly, because considerably better comparative data are available

for subsidies, and, thirdly, because the focus on subsidies provides the

“stronger test” (Van Evera 1997: 30f) for the rational choice arguments

of positive feedback—as argued in Chap. 1. However, preferences towards

tuition fees have been studied elsewhere (Garritzmann and Busemeyer

2015).

Empirically, I use questions posed in three International Social Survey

Program (ISSP) Role of Government survey waves, providing data for



Tài liệu bạn tìm kiếm đã sẵn sàng tải về

4 TSCS Analyses of the Developments in 21 Democracies Between 1995 and 2010 on a Government Term Basis

Tải bản đầy đủ ngay(0 tr)

×