The Joneses Visit an Economics Lab

Technical Summary: The Joneses Visit an Economics Lab

Problem and Motivation
Existing literature establishes that individuals care about how their consumption compares to peers, yet there is a proliferation of theoretical models explaining this behavior. These models range from "Keeping up with the Joneses" (self-image concerns driven by relative position) to Veblenian conspicuous consumption (social-image concerns driven by observability). Theoretical frameworks vary significantly in their assumptions: some emphasize comparison to the mean, others to ordinal rank, and others to pairwise differences with asymmetric weights for envy (upward comparisons) and pride (downward comparisons).

Empirical discrimination among these models is difficult due to limitations in field data (observability, endogenous peer groups) and a lack of laboratory experiments capable of eliciting preferences for conspicuous consumption. This paper aims to address this gap by proposing a unified quasi-linear framework to compare these models and testing their predictive power against new laboratory data.

Methodology
The authors design a laboratory experiment involving 224 participants across 14 sessions at the University of Essex. The experiment consists of two blocks:

1. Block 1 (Rounds 1–10): Simultaneous move rounds.
2. Block 2 (Rounds 11–15): Last-mover rounds, where participants can unilaterally revise their choices after observing peers' decisions.

Experimental Design:

• Task: Participants earn tokens via a "slider" real-effort task.
• Consumption Good: Participants allocate tokens to purchase "decals," a purely positional good with no intrinsic monetary value but visible to peers.
• Treatments:
  • Jones Treatment: Peers are anonymous (usernames suppressed). This isolates self-image concerns ("Keeping up with the Joneses").
  • Veblen Treatment: Peers' identities are visible. This introduces social-image concerns (status signaling).
• Framework: The study assumes quasi-linear preferences $U(m_0, y_0, y_{-0}) = m_0 + \phi(y_0, y_{-0})$, where $m_0$ is remaining income and $y_0$ is conspicuous consumption (CC). This framework implies zero income effects for interior solutions.

Theoretical Models Compared
The paper evaluates seven models within this framework:

1. Simple Average Model (SAM): Utility depends on the difference between own CC and the peer mean.
2. Conformity Model: Utility is maximized by minimizing the squared distance from the peer mean.
3. Rank-Dependent Model: Utility depends solely on ordinal rank, not cardinal values.
4. Peerwise Envy-Pride (PEP): Linear utility based on pairwise differences, with asymmetric weights for envy and pride.
5. Mean Envy-Pride (MEP): Linear utility based on the difference between own CC and the peer mean.
6. Prospect Theory - Mean Envy-Pride (PT-MEP): Uses an S-shaped value function (loss aversion) applied to the difference between own CC and the peer mean.
7. Prospect Theory - Peerwise Envy-Pride (PT-PEP): Uses an S-shaped value function applied to pairwise differences between own CC and each peer's CC.

Key Results
The data reveal several robust patterns that allow for the rejection of several standard models:

1. Quasi-Linearity: Income effects are negligible. The coefficient of earnings on CC is statistically indistinguishable from zero, supporting the quasi-linear framework and ruling out models with strong multiplicative income effects.
2. Visibility Effects: CC is significantly higher and more persistent in the Veblen treatment compared to the Jones treatment. The "Last Mover" treatment (Block 2) causes an immediate and sustained increase in CC, primarily driven by the extensive margin (more participants choosing positive CC) rather than the intensive margin (amount spent by existing consumers).
3. Rejection of Rank and Conformity Models:
   • Rank-Dependent: Contrary to the prediction that players target specific ranks at minimal cost (choosing just above a peer's level), data shows very few "on-target" choices. Active adjustments often move players away from target ranks.
   • Conformity: The model predicts convergence toward the mean and shrinking dispersion over time. The data shows no such trend; dispersion remains stable or increases slightly.
4. Rejection of Linear Models:
   • PEP and MEP: These linear models predict that CC should either be a corner solution (0 or max income) or, in the case of MEP, track the mean exactly. They fail to capture the observed interior choices and the specific response to peer dispersion.
   • MEP/PT-MEP: These models predict that peer dispersion (measured by $L$, the normalized CC of higher-ranking peers) has no effect on CC once the mean is controlled for.
5. Support for PT-PEP: The Prospect Theory - Peerwise Envy-Pride (PT-PEP) model best explains the data.
   • 1:1 Response to Mean: CC increases nearly one-for-one with the peer mean.
   • Negative Response to Dispersion: Crucially, controlling for the mean, CC is negatively correlated with peer dispersion ($L$). This aligns with the PT-PEP prediction that an S-shaped value function creates diminishing sensitivity; a mean-preserving spread increases the "pain" of envy from distant leaders more than the "pleasure" of pride from distant followers, leading to lower optimal CC.
   • Interior Solutions: The S-shaped function stabilizes interior choices, preventing the extreme corner solutions predicted by linear models.

Significance and Claims
The paper claims to provide the first laboratory evidence distinguishing between various theoretical models of conspicuous consumption. Its primary contribution is demonstrating that conspicuous consumption arises from reference-dependent, loss-averse social comparisons applied at the interpersonal level.

Specifically, the authors argue that:

• The PT-PEP model is the superior explanatory framework because it accounts for both the strong sensitivity to the peer mean and the negative sensitivity to peer dispersion.
• The S-shaped value function (diminishing sensitivity) is essential for explaining why individuals choose interior levels of consumption rather than extreme corner solutions or strict rank targeting.
• Visibility (Veblen effects) significantly enhances the intensity and sustainability of conspicuous consumption, likely by adding social-image concerns to self-image concerns.

The authors remain modest regarding the generalizability of their findings, noting that the "decal" good is a bland form of conspicuous consumption and that the "Last Mover" effect might be a restart effect requiring further investigation. They suggest that future work should explore different tasks and more dramatic forms of signaling to confirm the robustness of the diminishing sensitivity and loss aversion mechanisms identified.