The Mixed Member Proportional representation (MMP / AMS)
tries to link two (partially opposed) principles:
- Elect the candidate for direct election nominated in the constituency.
- Proportional Party Representation.
For each candidate directly elected in the constituency his party list will lose a seat.
Hence, it is mostly worthless to win a constituency and the constituency vote is not useful
for parties.
But this only works, as long as the party wins more seats by the party vote
(secondary vote) than by the first vote (constituency vote).
If a Party is awarded less seats, as it needs to justify the constituency seats won, the
missing proportional seats are additionaly awarded as additional overhang seats
(see § 6 Abs. 5 and § 7 Abs. 3 sentence 2
BWahlG).
|
The two direct reasons, increasing the number of overhang seats
of a party's Land's List, are: |
| 1. |
Several won constituencies. |
| 2. |
Few Party votes. |
The week point in the legitimation of overhang seats
is Reason 2:
few or missing votes cause additional seats.
All other conceivable reasons cause overhang seats
by changing reason one or two:
-
Cast with the constituency vote a candidate of another party as
casted with the party vote "Splitting of votes".
Thus, the electors can target both reasons.
- Poor Polling A small election turnout in a country (Bundesland, federal state) reduces the number of
party votes (Reason 2) of the Land's List.
- Several (Three or more) relatively strong parties (week Constituency Winners) with
narrow majorities.
Especially in the new eastern countries, where there are three bigger competeting parties (CDU, PDS and SPD)
with chances to win constituencies.
The strongest party can win almost all of the constituencies (reason 1)
with close votes (reason 2).
- Distribution of constituencies on the lands
The number of constituencies in a land (federal state) is the maximum number of
constituency seats a party can achieve there (reason 1).
In the 1998 election four overhang seats were caused by
an unproportional distribution of constituencies (WP 96/98).
(3 in 1994).
- Gerrymandering (Zoning of constituencies)
It seems this is the only point of interest for the parties and
the Bundestag's reform commission, although it is rather irrelevant
in comparison to all other points.
If all the constituencies are too big, it would touch their distribution
into the countries (land's).
With gerrymandering, one could influence the winner in same constituencies (and with that, reason 1)
but practically, this seems to be really difficult.
Many won constituencies of one party (and thus overhang seats) are favoured by more
more homogenous
constituencies, because the conditions to win one constituency are the same as to win
all constituencies.
On the other hand,
overhang seats are more unlikely in a
heterogenous set of constituencies,
because the conditions to win a constituency change from constituency
to constituency. To win almost all of them, means to reach more than 50% of the votes,
by any means.
- Effect of subdivison by the Hamilton's method.
Because Hamilton's method (Hare/Niemeyer) is not consistent, the number of votes
for one land's list affect the allocation of seats between two others.
E.g. in 1994 additional votes for the SPD in North-Rhine/Westphalia would have cost
the SPD would have cost an overhang seat
In extreme cases this could mean that a land's list
which wins votes receives less seats.
(Example).
von Martin Fehndrich