Demanding more than what you want

Demanding more than what you want (DOI: https://doi.org/10.1017/psrm.2025.25) was written by Zuheir Desai and Scott A. Tyson in 2025. It was published in Political Science Research and Methods.

The authors consider platforms to be "pledges that pull policy imperfectly in the direction of their proposed platform and away from a status quo." No one expects platforms to become policy directly. There is a dimension of capability attached to candidates which explains how platforms become policy. Furthermore, platforms are relative to the status quo. A platform that is significantly different should be more difficult to realize; a platform to maintain the status quo is trivial.

Past research has considered non-ideological aspects of candidates to be an orthogonal valence issue space. Studies are divided on the effect of valence advantages between candidates; some have found it leads to ideological moderation, others to ideological extremism.

Downs argued that platforms cannot be compared directly; instead reasonable expectations of what policies will be realized should be compared. Platforms are effectively discounted.

The authors iterate on the median voter theorem. There are two candidates and there is a given representative voter whose incentives are examined.

There is a status quo: s in one dimensional real space R.

Each candidate has two inherent fixed attributes: an ideological position (y_i for each candidate i and also in R) and a capability (c_i in [0,1]). They also have a platform (x_i also in R) that is given the status quo and their capability. The realized policy outcome function π is defined as π(x_i; c_i, s), shorthand as π_i. The utility function they maximize is -|y_i - π_i|.

• 'Capability' extends not only to their ability to realize their platform but also to perform important duties that have common support, e.g. national defense.
• Candidates are singly motivated by realizing policy close to their ideological position.

• |x_i - π_i| monotonically decreases on c_i

• |s - π_i| monotonically increases on c_i

• For c_i in (0,1) the authors assume...

  • 

  • 

  • 

All individuals have an ideological position in R, but this mdoel cares only about the two candidates and a given voter. For simplicity, the real space is normalized such that the position of that given voter is 0.

Big assumption: candidates' positions are assumed to be symmetric about the given individual: y₁ = -y₂. The authors more generally argue that the case of a 'moderate' voter (i.e., y₁ < 0 < y₂) is the only interesting case.

The given voter has two fixed attributes:

1. The salience of capability is parameterized as α (in [0,1])

   • this is known to candidates
   • This 'salience' is the degree to which they care about capability.

   • The complement (1-α) then is the degree to which they care about ideology.

2. Residual valence is parameterized as ε_i.

   • this is not known to candidates, but the authors assume that candidates expect it is uniformly distributed between ±1/ψ

     • Necessarily, ψ > 0.

   • 'Residual valence' is the preference for a candidate independent of both x_i and c_i

The given voter's utility function, if candidate i wins, is αc_i - (1-α)|π_i| + ε_i.

The authors derive a unique equilibrium given a few more assumptions, that mostly are in the same direction as above (i.e., only interesting case is when both candidates have a non-zero chance of winning).

It becomes helpful to parameterize the capability gap as γ (defined as |c₁ - c₂|).

The important conclusions from the model though are:

• If the status quo is moderate, both candidates set platforms that pull the status quo away from the given voter's position.

• As α increases, the platform of the candidate with a capability advantage moves further from the voter's position. (They are able to extract greater utility from their comparative advantage.) Conversely, as α decreases, the platform of the candidate with a capability disadvantage moves further from the voter's position. (They are not credible, so the more extreme platform is discounted.)

• If the status quo is moderate, as c_i increases, that candidate moves their platform further from the voter's position. Conversely, the other candidate moves their platform towards the voter's position. (They need to compete against a more competitive opponent.)

• There is a critical level of the status quo in the opposite direction compared to a candidate where, as c_i increases, they move their platform towards the voter's position. (They must moderate their platform because it is more extreme than their opponent's realized policy position.)

• There is a critical level of the status quo in the same direction compared to a candidate where, as c_i decreases, that candidate moves their platform towards the voter's position, and possibly past it towards the opponent's position. (A moderate platform is not credible enough, so they set the platform to be extreme in the opposite direction.)

  • This can explain circumstances where both candidates set platforms on one side of the ideological space: s < 0 <= x₁ < x₂.

The authors connect this to ideological polarization.

Reading notes

The authors chose a shorthand notation whereby the leftist candidate has a negative platform (i.e. π_L < π_R), has a capability advantage (i.e. c_L > c_R), and therefore is more extreme. This is a strange choice given how easy it is to re-express everything in abstract terms, as seen above.

I also have trouble taking this model seriously when there are so many arbitrary assumptions about the residual valence.

CategoryRicottone