Does executive autonomy reduce second-order election effects?

Figures & data

Table 1. Hypothesised impacts of executive autonomy and authority on second-order election effects.

Authority	Executive autonomy
	Low	High
Low	A: +	C: + +
High	B: −	D: − −

Table 2. Included countries, regions and regional and national elections.

Country	Regions	N	Regional elections			National elections
			First	Last	N	First	Last
Austria	Länder	9	1949	2019	137	1949	2019
Belgium	Deutsche Gemeinschaft	1	1990	2019	7	1987	2019
	Gewesten and Brussels	3	1989	2019	19	1987	2019
Denmark	Amter	14	1970	2001	113	1968	2001
	København	1	1946	2001	15	1945	2001
	Regioner	5	2005	2017	20	2005	2015
France	Régions	22	1986	2010	109	1986	2007
	Régions (after mergers)	13	2015	2017	14	2012	2017
Germany	Länder (West)	11	1949	2019	176	1949	2017
	Länder (East)	5	1994	2019	30	1990	2017
Greece	Nomoi	49	1994	2006	196	1993	2004
	Peripheria	13	2010	2019	39	2009	2015
Italy	Regioni a statuto ordinario	15	1970	2019	158	1968	2018
	Regioni a statuto speciale	5	1948	2019	71	1948	2018
	Trento, Süd-Tirol	2	2003	2018	8	2001	2018
Netherlands	Provincies	12	1946	2019	218	1946	2017
Norway	Fylkeskommuner	19	1975	2015	209	1973	2013
	Fylkeskommuner (after mergers)	11	2019	2019	11	2017	2017
Portugal	Açores, Madeira	2	1976	2019	23	1976	2015
Spain	Ceuta, Melilla	2	1995	2019	14	1993	2019
	Comunidades autónomas	17	1980	2019	175	1979	2019
Sweden	Län/Regioner	29	1946	2018	498	1944	2018
	Städer	5	1946	1994	44	1944	1994
Switzerland	Cantons	24^a	1945	2019	346	1943	2015
United Kingdom	Greater London Authority	1	2000	2016	5	1997	2015
	Scotland	1	1999	2016	5	1997	2015
	Wales	1	1999	2016	5	1997	2015
Total		282			2665

Note: The total number of regions does not add up to 292 but to 282 because six French and four Norwegian regions were not merged.

^aGraubünden and Appenzell Innerrhoden are not included because they hold elections during public meetings.

Table 3. Regional Executive Autonomy Index (REAI) indicators.

Indicator	Scores
Mode of appointment of the head of the executive	By largest party (=0), parliament (=0.33), members executive (=0.66), citizens or central government (=1)
Mode of appointment of the other members of the executive	By party (=0), parliament (=.33), citizens (=0.66), head executive (=1)
Guaranteed parliamentary support for the executive	No seat bonus (=0), seat bonus <50% (=0.5), seat bonus >50% (=1)
The extent to which citizens elect the members of the executive	None (=0), all members (=0.5), only the head (=1)
Mode of dissolution of parliament	By parliament (=0), both parliament and executive (=0.33), not possible (=0.66), executive (=1)
Mode of termination of the executive	By nominating party (=0), simple (=0.25) or qualified majority (=0.5), constructive vote of no confidence (=0.75), not possible (=1)
Overlap between membership executive and parliament	Not possible (=0), not clearly separated (=0.5), must overlap (=1)
Extent of involvement of executive head in parliamentary affairs	None (=0), chairs parliament (=0.5), chairs parliament and committees (=1)
Extent of power sharing within the executive	Head is primus inter pares (=0), head is elevated (=0.5), monocratic head (=1)
Regional Executive Autonomy Index	Scores varying between 0 (non-autonomous) and 9 (autonomous)

Figure 1. Marginal effects of executive autonomy and regional authority on party vote share swings.

Notes: Shown are predicted vote share swings between regional and previously held national elections and their 95% confidence intervals when executive autonomy and regional authority, self-rule and shared scores go from their minimum to their maximum (Online Appendix Table B1b). The estimates in Figure 1 are based on the model results presented in Online Appendix Table B2 which include 41,603 vote share swings for 2665 elections held in 282 regions in 14 countries (Table 1).

Table 4. Average vote share swings won by personal lists and regionalist parties.

Party type	Regional authority	Executive autonomy
		Low	High	Diff.
Personal lists	Low	−2.0%	7.6%	9.6%*
	High	−0.1%	2.3%	2.4%*
	Diff.	1.9%*	−5.3%*
Regionalist parties	Low	1.0%	0.9%	−0.1%
	High	1.4%	2.4%	1.0%*
	Diff.	0.4%	1.5%*

Note: ^*p < .01. (ANOVA, Games-Howell nonparametric post hoc analysis for non-equal variances and sample sizes, bootstrap; Field et al. 2012). Low = below the median score; high = medium score or above (Online Appendix Table B1b). Shown are average vote share swings between regional and previously held national elections won by 142 personal lists in 573 regional elections and by 371 regionalist parties in 1373 regional elections. Details are provided in Online Appendix D.

Field, Andy, Jeremy Miles, and Zoë Field (2012). Discovering Statistics Using R. London: Sage.

 Google Scholar