The IDEA book, has among the advantages of first past the post one that might answer your question (on majority-biased outcomes):
[b.] It gives rise to single-party governments. The ‘seat bonuses’ for the largest party
common under FPTP (e.g. where one party wins 45 per cent of the national vote but
55 per cent of the seats) mean that coalition governments are the exception rather than
the rule. This state of affairs is praised for providing cabinets which are not shackled by the restraints of having to bargain with a minority coalition partner.
Likewise John Curtice (2015) says:
According
to its advocates, the single-member plurality electoral system enables the electorate
to choose directly between alternative governments by ensuring that whichever
party comes first in votes secures an overall majority in seats, even though it may
have won much less than half the vote.
And he actually cites this argument to Powell's book. But Curtice goes on to criticize it as not being delivered in practice as often as its (unnamed) proponents claim.
And David Cameron has phrased roughly the same argument as:
Throughout history, [FPTP] has risen to the demands of the time, often with a brutal decisiveness.
That’s what happened when it brought in the Thatcher government in 1979.
The British people recognised it was time for change – and the electoral system didn’t let them down.
He doesn't seem wrong with that example, as the Conservative bonus was approximately 9.5% in that election (53.4% of seats vs 43.9% of votes). But the example Cameron gave is about average for the UK; according to Norris (1997)
In the postwar period, for example, British governments have received, on the average, 45 percent of the popular vote but 54 percent of seats.
And Norris also elaborates on the single-party government issue:
The classic argument for majoritarian systems is that they tend to produce stable and responsible
single-party governments, so that the electoral outcome is decisive. In contrast, unless one party wins a
majority of votes, PR is closely associated with coalition cabinets. A survey of twenty countries found that
single-party governments were formed after 60 percent of majoritarian elections, but only 10 percent of
PR elections (Blais and Carty 1987). If we compare the parliamentary democracies in this analysis
56.3 percent of elections under majoritarian systems produced single-party governments, compared with
36.4 percent of elections under mixed systems, and 34.8 percent of PR elections. In countries with PR
and fragmented party systems, like Italy, the Netherlands and Switzerland, all governments tend to be
coalitions. But majoritarian electoral systems can also result in coalition governments, such as in Britain
between the wars. Moreover PR systems may also have single-party governments, such as long periods
of dominance by the Austrian Socialists, the Norwegian Labour party, and the Swedish Social Democrats.
The pattern of government formation is therefore far more complex than any simple linear relationship
might lead us to expect (Laver and Shepsle 1995), although as expected there is a significant
relationship between the production of single party governments and majoritarian electoral systems.
And Blais appears to have been a major researcher of this question (although hardly the first one). A 1991 paper of his notes:
There can be no doubt that one-party government is
more likely to occur under plurality than under proportionality rule, as Rae’s (1969:
99) data indicate: "In 75 legislatures elected under P.R. formulae, the mean minimal
majority was 1.96 parties. Typically, the support of the two largest parties was
required for the formation of the majority. In 45 legislatures elected under majority
and plurality formulae, the mean minimal majority was only 1.15 parties, suggesting
that one-party majorities were more common."
Blais and Carty (1988) indicate that 72% of single-member
district plurality elections produce a one-party legislative majority, compared to 10%
of PR elections. Blais and Carty (1987) also show that, everything else being equal,
the probability of a one-party majority government is 40 percentage points higher in
a plurality than in a PR election.
These findings can be interpreted in two different ways. On the one hand, the
plurality rule (in single-member constituencies) generates majorities most of the time
and much more frequently than PR. On the other hand, it fails to achieve its basic
stated objective three times in ten and it is not even the most efficient procedure in
that regard: as Blais and Carty (1988) point out, multi-member district majority
elections have produced one-party majority governments nine times in ten. In short,
the plurality rule greatly increases the likelihood of a one-party government but is
not entirely successful on that score.
Although not as explicit, the following passage from a Fraser Institute paper advocating FPTP,
is probably arguing the same point (that a winner bonus is good because it makes changing an
existing government easier):
It is the ability to “throw the bums out,” more even than the ability to
choose a new government, that is the most striking practical virtue of
FPTP. Our governments are responsible, must answer to the voters, and
are regularly defeated. Joseph Schumpeter (1987: 272) and Karl Popper
(1963 and 1988, April 23) saw the ability to get rid of an unsatisfactory government
as the purpose and test of democracy and condemned proportional
representation for not seeing this. To “throw the bums out” is almost
impossible with proportional representation. In the 50 years after 1945 in
103 elections in Belgium, Germany, Italy, Japan, the Netherlands, Sweden,
and Switzerland, the major governing party was only thrown from office
six times (Pinto-Duschinsky, 1998, September 25). Major parties have remained
in government for decades under proportional representation despite
wide fluctuations in their votes. Minor parties often seem to share in
government in inverse proportion to their electoral success, turfed out when
their vote grows and they look threatening, and brought in when it sags.
As you noted below, such bonuses occur even in some semi-proportional systems, e.g. it's an intended feature of MBS. But it's also a less intended feature in SNTV. And
In his famous study of the impact of electoral systems, Douglas Rae observed that a high degree of proportionality is hardest to achieve in single-member electorates.
(of which FPTP is a prime example.) Also, an alternative name is (consequently) "winner's bonus".
Curtice notes that
The [FPTP] system will only provide the winner with a substantial bonus if a relatively large number of seats are highly competitive (or ‘marginal’) between the two parties. In those circumstances, seats readily change hands from one party to another, thereby making it likely that even a party with quite a small lead in votes will enjoy a substantial lead in seats. If, however, there are relatively few such seats, then a party might need a big lead in votes before it secures a majority of seats.
You may find of interest the paper of Shugart (2001) on "mixed-member system" (MMP). His theoretical preference is quite well laid out and contains what he considers to be the strong point of plurality systems (delivering identifiable governance):
Part of the problem is in figuring out what “society” wants. Of course, all the
standard problems of social choice are inherent in any expectation that elections
provide clues about collective preferences (Riker, 1982). Minimally, elections are
simply a devise for determining who should govern, and not an instrument for
determining what policies politicians should pursue once in office (Schumpeter,
1950). Indeed, empowering a government is a key component of my understanding
of efficiency, as much as it was for Bagehot. In order to empower a government,
elections must offer voters a choice from two parties or blocs of parties, one of
which will be likely to attain full control of the government. I define this aspect of
efficiency below as the “identifiability” of competing governmental options.
Systems based on the majoritarian pattern of democracy (Lijphart, 1999) by definition
offer high identifiability, but they do not assure that the government is supported
by a majority. On the other hand, systems based on proportional representation
usually assure that governments are based on coalitions representing a majority of
the electorate, but the government that forms is usually not identifiable in the election
campaign that precedes its formation. Systems that offer very low identifiability may
be termed hyper-representative systems. Pre-reform Italy is a prime example.
While necessary, identifiability is not sufficient for efficiency, because of the high
disproportionality that typifies majoritarian electoral systems. Given disproportionality,
the government that emerges from the electoral process might represent
only a plurality and thus leave the majority utterly unrepresented in the government.
Again, elections are at best “noisy” indicators of voters’ actual policy preferences,
due to the problems of social choice. Nonetheless, the risk is that governments that
are based on the electoral support of well under a majority of the voters will tend
to pursue policies that are not favored by a majority. To put it another way, such a
government is not constrained to follow more broadly supported policies because it
governs alone.
I shall call systems that generate governments representing well under a majority
of the electorate pluralitarian systems, thereby signaling that they are indeed not
representative of a majority, due the presence of a multiparty system in the electorate.
In such systems a less disproportional translation of votes into seats would almost
certainly lead to a coalition government, which would be more likely to represent
the preferences of a majority of the electorate, and would constrain any one party
from pursuing policies that were primarily of interest to its own constituency rather
than to the broader electorate. Pre-reform New Zealand is a paradigmatic case of a
pluralitarian system.
Thus, within the inherent limits of elections as instruments of collective choice,
the most efficient way for elections to connect government to the electorate is for
there to be both high identifiability and high proportionality. However, these two
key components of efficiency are likely to be in conflict. Identifiability is associated
with majoritarian electoral systems and proportionality is associated (obviously) with
proportional representation. It is because of these countervailing pressures that
mixed-member systems are likely to be more efficient. Theoretically we can expect
the tier of single-seat districts to encourage parties to aggregate into two principal
blocs — generating high identifiability — and the proportional tier to moderate or
eliminate (depending on specific details of how the tiers are combined) the disproportionality
of the outcome. The resulting governments can be expected to be
efficient in the sense that they are both empowered from the election outcome yet
constrained by the need for coalitions to take in a broader swath of the electorate’s
preferences.
His empirical (interparty efficiency) index he derives is not as impressive as one might hope from that. Essentially, he linearly combines an partly impressionistic measure of "electoral linkage" [L] (a measure of identifiablity) with a purely statistical plurality enhancement measure [P] (which actually works as a penalty, because the efficiency formula is L+P-1). For Westminster-style systems, the electoral linkage is basically 1, so only the deviation (from zero) by plurality enhancement counts as inefficiency.
A new paper of this kind is Raabe and Linhart (2017), which sadly uses almost completely different terminology for the roughly same notions.
Proportionality [... well, you know what it means]. The advantage of concentrated party systems, on the other hand, is that government
formation is connected more strongly to the voters’ choice. In the clearest cases, one
single party wins a majority of seats and forms a government – and thus can be held
responsible for its performance in the upcoming elections. The more fragmented a party
system is, the less clear it becomes who is an election winner and the more government
formation depends on coalition bargaining between parties instead of election results.
At the same time, more fragmented party systems generally lead to more parties in
government so that single parties in government can be held accountable by the voters
only partially (Powell, 2000).
As polar design options, pure PR electoral systems are associated with highly
representative parliaments that allow for a more nuanced representation of the
electorate, while plurality electoral rules are associated with the creation of
accountable single-party governments (Duverger, 1954; Rae, 1967; Farrell, 2011).
However, PR systems typically fail to concentrate the party system in order to
enable swift government formation and plurality systems fail to provide accurate
representation and to account for minority interests (Shugart, 2001).
However this 2017 is far better (compared to Shugart) at exploring the multi-dimensional design space between
(pure) plurality and (pure) propotionality:
We expect that the share of single-member
districts, the district magnitude, the legal threshold,
and the level of compensation each exert
individual effects on the propensity of an
electoral system to successfully provide both
proportionality and concentration.
Alas, they don't seem to consider an explicit bonus system as a means to bridge the two... anywhere in the paper. (An interesting pair of factoids is that Anglo-Saxon political scientists hold MMP is very high regard, while MBS has been held in outright derision by some political scientists, and generally gets little academic attention.)
