Quick glance: Developing Kampala’s Citizen Charter

Scope: Kampala, Uganda | 2,312 residents | 188 small-scale citizen meetings

Duration: Jan. 2019 – Dec. 2022 (pandemic interruptions)

Implementation: IPA Uganda

Funders: IGC, WZB, Columbia, an anonymous foundation

Goal: Provide citizen input into the construction of a Citizen Charter in Kampala, outlining rights and responsibilities between citizens and municipality

Results based on baseline survey, consultation behavior and decisions, and post-consultation survey.

The context

Primary goal: measurement of political inequality

How can we best measure citizens’ degree of political power (understood as influence)? How much inequality is there in this?

How do gaps in voice (Coffe and Bolzendahl 2011; Kasara and Suryanarayan 2015) relate to systemic responsiveness (Gaikwad and Nellis 2021)?

Probe complex linkages between:

1. input: inequality in intensity of participation
2. throughput: inequality in the system’s responsiveness to demands
3. output: inequality in how much decisions favor specific individuals / groups

Preview of findings

We uncover clear disparities in inputs, with more advantaged citizens participating more during meetings.

There is evidence of limited elite capture, but not of outright throughput inequality between groups.

Thankfully, no evidence of output inequalities, suggesting the possibility of effective compartmentalization.

Even in a “hard case” (trained facilitators, small group, grounded topics) we continue to observe input inequality.

Where does discrepancy come from?

Inequality in participation and preferences \(\nrightarrow\) inequality in outputs.

Our interpretation: evidence consistent with discussion leaders countervailing efforts by more powerful groups to skew outcomes.

Possible means:

• imposing their own views on discussions
• amplifying opinion of less powerful groups (some evidence for this)

Simple model of political action

We start from a simple model: a status quo policy, \(x\), which \(N\) players with ideal policy points \(x_i^*\) try to influence.

Each player takes action with intensity \(a_i^*\), and with \(\pi_i^*\) denoting how much their action would shape the outcome (causal quantity).

Utility is: \(u_{i} = -(x_{i}^{*} - x^{'})^2 - a_{i}^2\)

The new policy is the result of all individual actions: \(x^{'} = x + \sum_{i=1}^{N}\pi_ia_i\)

Players choose optimal actions, which can vary in direction and magnitude, and we seek a Nash equilibrium.

Similar to the Esteban and Ray (1999) model, though on a policy space.

Strategies depend on other actors

The best response of player \(i\) is:

\[ a_i = \frac{\pi_i(x^*_i - x)}{1+\pi_i^2} - \frac{\pi_i\sum_{-i}\pi_{j} a_{j}}{1+\pi_i^2} \]

Important to note that action taken by \(i\) depends on:

1. own effectiveness, \(\pi_i\)
2. the action of \(j\): \(a_j\)
3. the effectiveness of \(j\): \(\pi_j\)

Also tied to own dissatisfaction with reversion point, and extremity to others’ preferences.

Multiple forms of inequality

In equilibrium, welfare is:

\[ w_i = -(1+\pi^2_i)\left(\frac{(x^*_i - x) + \sum_j(x^*_i - x^*_j)\pi_j^2}{1+\sum_j\pi_j^2}\right)^2 \]

We can have inequality in:

1. inputs
2. throughput
3. outputs

These are distinct quantities, and inter-related in complex ways depending on where the status quo, x, is.

Illustration: complex dynamics

Ideals are fixed: \(x_1 = -0.5\) and \(x_2 = 0.5\)

Increases in input inequality sometimes result in corresponding increases in output inequality, but sometimes they don’t.

Capturing the dynamics

We try to assess these inequalities in the setting of our consultative meetings.

They allow us to measure:

1. preferences
2. the actions taken to promote these
3. outcomes (decisions)

We sacrifice some generalizability, but gain tight control over the process and the possibility to measure frequently.

Structure of meetings

• Small-scale consultation meetings, of around 1-1.5 hours
• Participants: 6-8 citizens recruited from the same village
• Facilitated by KCCA officials, or neutral facilitators (trained enumerators)
• Objective: collect input from citizens for construction of Charter
• Decisions: made by the group, and recorded by facilitator

Neutral facilitators underwent a special training focused on the importance of neutrality in such consultations.

Topics discussed in meetings

1. Report municipal expenditures: At what levels should budgets be reported to citizens?
2. Channels of communication: Through what communication channels should KCCA engage with citizens?
3. City allocations: growth vs equality: Should KCCA plan city investments focusing on maximizing growth or reducing inequality?
4. Raising fees and taxes: Should there be higher taxes for better services, or rather lower taxes even with fewer services?
5. Monitoring Charter: Should Charter be monitored externally or internally?
6. Extra: “Is KCCA going in the right direction?”

Features: (1) identified based on baseline survey; (2) vetted by KCCA; (3) plausibly contentious.

Mapping: questions-data

Dimensions of inequality we probe

Effect of leaders: KCCA going in right direction?

Facilitators in general, and IPA facilitators in particular, matter!

We see a similar visible difference for 2 more topics.

Do facilitators drive outcomes?

Adj. \(R^2\): bound on the degree influence . Facilitators drive 10–25% of variation in meeting outcomes.

Are they more influential in disadv. contexts?

The dynamic we expect is not seen for any of the topics.

Facilitator preferences do not drive the meeting outcome more in disadvantaged communities.

KCCA vs. IPA facilitators: is there outcome skew?

For one of five issues, we see outcomes skewed in a direction preferred by the institution (KCCA).

Do the disadvantaged do worse under some facilitators?

Example: \(Match_i = \beta_0 + \beta_1*gender + \beta_2*KCCA + {\color{purple}{\beta_3}}*gender*KCCA + \epsilon_i\)

No: we don’t see inequality in responsiveness.

Is there throughput inequality?

We find clear influence of facilitators in the process: anywhere between 10 and 25% (of variance in outcomes explained by facilitator identity).

For only one issue, clear differences in outcomes between KCCA and IPA facilitators.

No evidence that the preferences of some sub-groups are favored over those of other sub-groups.

Effects of meetings: satisfaction with outcomes?

Summing up

We find consistent patterns of input inequality during on consultations (by gender, age, wealth, and education), but not in attendance at consultations.

Discussion facilitators have a moderately-strong influence over the outcome of the consultation (some evidence of throughput inequality, but no disparities in responsiveness).

Encouragingly, we find no systematic evidence of output inequality.

Egalitarian process partly offset by inequalities in power.

Implications

Lijphart (1997): unequal participation produces unequal responsiveness (Hill and Leighley 1992). Might not always be the case.

A mistake to infer inequality in outcomes from inequality in inputs, or inequality in responsiveness from inequality in inputs.

In our setting, we believe facilitators play an offsetting role—what else could be at play?

Power in consultations

Output inequality

Top 4 explanatory factors (permutation-based)

Village preferences matter, but not as much as facilitator characteristics.

In the top 8 features, we frequently see political features at the village (political information, engagement, participation) as mattering in 3/6 cases.

Comparative model fit

Predictive accuracy does increase when adding substantive predictors, but only slightly.

Output inequality summary