Back to All Research
Hanoi mobility transition | discrete choice evidence

Motorbike Phase-Out Policy Acceptance

Using a discrete choice experiment for Hanoi residents, this study estimates how acceptance changes across policy bundles that combine restriction scope, travel cost changes, public transport improvements, fare relief, green reallocation, park-and-ride, transition support, and revenue governance. The analysis estimates mixed logit models, derives willingness-to-pay values, and identifies heterogeneity by dependency, income, trust, and environmental concern. Results show that a pull-first and support-first pathway is more acceptable than a push-first pathway.

Study period context: 2025-2026 Study area: Hanoi, Vietnam Method: DCE + mixed logit Realized sample: n = 320
8 attributes 27 profiles 3 blocks 9 tasks per respondent Status quo included
Push coefficients are negative Pull and support coefficients are positive Trust weakens status-quo lock-in WTP confirms compensation need Deploy alternatives before restriction scale-up Use gate-based monitoring for rollout control Push coefficients are negative Pull and support coefficients are positive Trust weakens status-quo lock-in WTP confirms compensation need Deploy alternatives before restriction scale-up Use gate-based monitoring for rollout control
Section 1

Background

Hanoi faces a policy dilemma. Air quality and congestion conditions justify intervention, but motorcycles remain central to household mobility and income generation. The implementation gap appears when restrictions move faster than alternatives.

This study tests the conditions under which residents accept a fossil-fuel motorcycle phase-out package. It quantifies trade-offs among push, pull, and support attributes, then estimates preference heterogeneity across socioeconomic and attitudinal profiles.

Primary mode

Motorcycle = 78.3%

Motorcycle ownership

At least one = 85.8%

Model baseline

ASC = +1.621

Model fit

ρ2 = 0.28

Core policy signal

Pull + support outperform push-first

Metro and BRT expansion are policy pull anchors Restriction design must account for dependence by necessity Equity support is a central acceptance condition Distributional and procedural fairness shape support
Street traffic with high motorcycle density in Hanoi
Street traffic in Hanoi with high motorcycle density (illustrative context image).
Hanoi urban rail train in operation
Hanoi urban rail corridor used as transition-alternative context (illustrative image).
Section 2

Method

This study uses a stated-preference discrete choice experiment (DCE) under random utility theory. Respondents in Hanoi completed repeated choice tasks. In each task, they chose between policy bundles and a status-quo option.

Each bundle was built from eight attributes with three levels each: restriction scope, travel-cost change, alternative transport quality, fare relief, transition support, green-space reallocation, park-and-ride, and revenue governance. The design generated 27 orthogonal profiles, split into 3 blocks; each respondent completed 9 tasks.

We estimated mixed logit models for main effects and interaction effects. Interaction terms tested heterogeneity by motorcycle dependency, income, trust, and environmental concern. Willingness-to-pay values were calculated as coefficient ratios to the cost term, WTPk = βk / (-βcost). The design target was n = 400 and the realized sample was n = 320.

The scenario simulation in Section 3A is directly linked to these estimated coefficients. It is not a market-share forecast. It is a transparent comparison tool that helps non-technical readers understand direction and relative magnitude across policy bundles.

DCE attributes

8 attributes, 3 levels each

Design profiles

27 orthogonal profiles

Blocking

3 survey blocks

Tasks per respondent

9 choice tasks + status quo

Sample

Target n = 400, realized n = 320

Utility decomposition

Uni = Vni + εni

Systematic utility

Vni = β'Xni

Monetary trade-off

WTPk = βk / (-βcost)

Method sequence

Study area definition, attribute design, orthogonal profile generation, blocked survey fieldwork, mixed logit estimation, interaction estimation, and WTP derivation.

Section 3

Results

The model identifies a strong status quo pull, large disutility for broad restrictions, and strong positive utility for pull and support attributes. Heterogeneity is consistent with dependence, income, trust, and environmental concern channels.

Model fit

ρ2 = 0.28

Status-quo preference

ASC = +1.621

Strongest negative β

Ban all motorcycles = -1.752

Strongest positive β

Financial support = +0.956

Heterogeneity signal

σ significant on key random terms

Figure 1. Mixed Logit Main Effects

Reported coefficients and 95% confidence intervals from Table 3, grouped as push, pull, support-process, and status-quo terms.

Mixed logit main effects from Table 3
Mixed logit main effects: push coefficients are negative, while pull and support coefficients are positive.

Figure 2. WTP and Heterogeneity Effects

Reported WTP ordering from Table 5 with selected interaction terms from Table 4. Positive values support acceptance; negative values indicate compensation required.

WTP ranking and heterogeneity interactions from Tables 4 and 5
WTP ranking is led by e-motorcycle grants and stronger public transport quality; interactions show dependency, income, trust, and environmental-concern effects.

Restriction sensitivity

Ban intensity produces the strongest negative effect, with substantial heterogeneity in the random term (σ for broad ban is large).

Dependency and income effects

Motorcycle dependency strengthens opposition to broad bans (β = -0.324). Higher income reduces cost sensitivity (β = +0.005) and lowers grant valuation (β = -0.018).

Trust mechanism

Trust lowers status-quo pull through the ASC interaction (β = -0.281), supporting a governance-first communication strategy.

Environmental preference channel

Environmental concern raises support for green-space reallocation (β = +0.217), indicating stronger acceptance of visible co-benefits.

Policy Scenario Simulation

Coefficient-Based Scenario Simulation

This module converts estimated coefficients into scenario comparisons. It is a communication tool, not a demand forecast. In this index, push has a negative sign, while pull, support, and trust have positive signs.

Simulation calculation (concise)

1) Utility index

ΔU = (-1.967 × Push) + (2.093 × Pull) + (1.271 × Support) + (0.281 × Trust)

Inputs are entered on 0-100 sliders and normalized to [0,1].

2) Acceptance score

Score = clamp(50 + 18 × ΔU, 0, 100)

3) High-acceptance chance

Chance = 100 / (1 + exp(-2.7 × (ΔU - 0.25)))

Shown as 1-99 to avoid false certainty.

Definitions: Push = restriction scope and direct burden; Pull = alternative transport quality; Support = financial and implementation support; Trust = confidence in delivery and revenue governance.

Diagnostics: Alternative readiness (pull + support), Burden pressure (push with partial support offset), Transition trust (trust with support).

For scenario comparison only; not a direct adoption forecast.

39%

Higher values represent stricter scope and higher burden from push-side policy terms.

74%

Represents public transport frequency, fare relief, greenery, and park-and-ride quality.

74%

Represents transition grants and revenue governance transparency.

68%

Higher trust lowers status-quo attraction and improves acceptance feasibility.

Acceptance score

0/100

Rollout status

Pending

Pull Support Calibrate Restrict
Coefficient-based scenario display for comparing policy bundles under different slider settings.

Alternative readiness

0%

Burden pressure

0%

Transition trust

0%

Diagnostic gaps

Readiness gap

0 pts

Compensation gap

0 pts

Trust buffer

0 pts

Utility shift

ΔU = 0.00

High-acceptance chance

0%

Primary lever

-

Phase readiness

Pull

0%

Support

0%

Calibrate

0%

Restrict

0%

Policy guidance

Adjust policy mix to reduce burden before scale-up.

  • Waiting for scenario input.
Section 4

Managerial Implications

The managerial implication is sequencing discipline. Restriction scale-up should follow measurable readiness, support reach, and trust conditions.

Hanoi BRT bus in operation as pull-side transport capacity
Hanoi BRT operation as pull-side transport capacity context.
VinFast charging station infrastructure in Hoa Binh
Public charging infrastructure as transition-support context for cleaner vehicle adoption.

Pull-first deployment

Prioritize frequent and reliable public transport, fare relief, and usable interchange capacity before tightening restriction scope. This aligns with positive pull coefficients and high WTP for quality improvements.

Targeted equity support

Protect dependence-heavy and lower-income groups with e-motorcycle grants and transition subsidies. Interaction terms indicate that dependence strengthens ban opposition and income changes cost and grant valuation.

Governance and trust architecture

Use transparent revenue recycling and public oversight communication. Trust reduces status-quo pull, so implementation credibility is not supplementary; it is part of acceptance design.

Readiness gate

Transit frequency, reliability, and corridor coverage meet threshold.

Support reach gate

Grant and subsidy access reaches exposure-heavy households.

Trust and compliance gate

Complaint intensity and enforcement disputes remain stable.

Outcome gate

Air quality and social burden indicators improve together.

Source and image attribution

Model figures are directly based on manuscript Table 3, Table 4, and Table 5 values. Illustrative visuals in this section: hanoi-brt-bus.jpg (photo by Minseong Kim, CC BY-SA 4.0; source: Wikimedia Commons) and vinfast-charging-station-hoa-binh-01.jpg (photo by Nguyenvuxe, CC BY-SA 4.0; source: Wikimedia Commons).

Back to All Research