The Shape of Temptation: Implications for the Economic Lives of the Poor

Abhijit Banerjee, Sendhil Mullainathan

TL;DR

This paper proposes that the poor appear "myopic" not because they are inherently less patient, but because they face a higher "temptation tax." By modeling how the urge to spend on "temptation goods" (like snacks or cigarettes) declines as wealth increases, the authors explain why the poor struggle to save, avoid high-return investments, and fall into behavioral poverty traps.

What Problem Does This Paper Address?

Economists have long been puzzled by the behavior of the global poor. Why would a fruit vendor in India borrow money at 5% interest per day for nine years straight? Why do farmers fail to invest in high-return assets like fertilizer, even when they aren't strictly credit-constrained?

Standard economic models often attribute this to high "discount rates"—the idea that the poor simply value today much more than tomorrow. However, this doesn't explain why the same individuals might be far-sighted in other areas, like buying burial insurance. Banerjee and Mullainathan argue that the missing link is the composition of consumption. They tackle the gap between standard "time-inconsistent" models (like hyperbolic discounting) and the specific reality of living in poverty, where the struggle for self-control has much higher stakes.

Methods

The authors develop a theoretical microeconomic model centered on the distinction between two types of goods:

  1. Non-temptation goods ($x$): Goods that provide utility to both your current self and your future self (e.g., education, a sturdy house).
  2. Temptation goods ($z$): Goods that provide utility only to the "self" that consumes them in the moment, but are viewed as a waste by any previous "self" (e.g., cigarettes, sweets, or expensive but non-nutritious food).

The core of their method is a Modified Euler Equation. In standard economics, you save until the marginal utility of consuming today equals the discounted marginal utility of consuming tomorrow. In this model, saving for tomorrow is subject to a "temptation tax." Because you know your future self will "waste" a portion of every dollar on temptation goods you don't currently value, the incentive to save diminishes.

The authors specifically analyze the Declining Temptation Case (DTC), where the fraction of the marginal dollar spent on temptation decreases as total consumption rises.

Datasets

As a theoretical paper, it does not introduce a new dataset. Instead, it synthesizes and provides a framework for interpreting existing empirical findings from:

  • Surveys of moneylenders in Pakistan (Aleem, 1990).
  • Studies on fertilizer use in Kenya (Duflo, Kremer, and Robinson, 2009).
  • Financial diaries of the poor (Collins et al., 2009).
  • Consumption data from rural India regarding spending on fats and sugars (Subramanian and Deaton, 1996).

Key Insights and Ideas

  • The Temptation Tax: If you are poor, a larger share of your next dollar is likely to go toward satisfying immediate, visceral urges. This acts like a high tax on savings. If you are rich, that "tax" is negligible, making it easier to be "patient."
  • Behavioral Poverty Traps: The model shows that poverty traps can exist even without external barriers like bad credit markets. If you are just below a certain wealth threshold, the temptation tax is so high that you choose to dissave, keeping you poor. If you were just slightly richer, you would find it worthwhile to save, eventually becoming much wealthier.
  • The "Chaining" Effect: In a multi-period life, the problem compounds. You don't just worry about tomorrow's self wasting money; you worry about the self the day after that, and so on. This "chain" of future temptations makes long-term saving feel futile for the very poor.
  • Investment Scale Matters: The poor may reject a high-return investment if the payout is too small. If the profit isn't large enough to "lift" them into a lower-temptation wealth bracket, they anticipate the profit will just be eaten up by temptation goods, so they don't bother investing at all.

Key Results

  • Attributions of Impatience: The paper proves that an observer using standard models would mistakenly conclude the poor are more impatient ($\delta$ is lower), when in reality, they might have the same internal patience as the rich but are simply reacting to the "temptation tax" inherent in their consumption level.
  • Precautionary Savings Paradox: Unlike standard theory, which suggests people save more when the future is uncertain, this model shows that for the poor, increased uncertainty can actually reduce savings. The "tax" makes it too expensive to protect against low-income states.
  • Credit Demand: The model explains why the poor might demand "commitment" credit (like microfinance with rigid repayment) but fear "temptation" credit (like credit cards), which provides easy access to small-scale temptation spending.

Limitations and Future Work

  • Defining Temptation: The authors acknowledge that what constitutes a "temptation good" is subjective and context-dependent. They suggest that future empirical work should use "Engel curves" (how spending changes with income) to identify these goods.
  • The Starvation Margin: The model assumes the "Declining Temptation" starts above the level of absolute starvation. At the very bottom, the "temptation" to eat nutritious food might actually be a far-sighted necessity, suggesting the "temptation curve" might be S-shaped rather than strictly concave.
  • Policy Implications: While the paper discusses "sin taxes," it notes they can be regressive and might actually increase the temptation tax if the demand for those goods is inelastic, potentially harming savings further. More work is needed to design "temptation-aware" policies.

Summary updated

Abstract

This paper argues that the relation between temptations and the level of consumption plays a key role in explaining the observed behaviors of the poor. Temptation goods are defined to be the set of goods that generate positive utility for the self that consumes them, but not for any previous self that anticipates that they will be consumed in the future. We show that the assumption of declining temptations, which says that the fraction of the marginal dollar that is spent on temptation goods decreases with overall consumption, has a number of striking implications for the investment, savings, borrowing and risk-taking behavior of the poor, which would not arise if temptations were either non-declining or entirely absent. Moreover the predicted behaviors under the declining temptation assumption can help us explain some of the puzzling facts about the poor that have been emphasized in the recent literature.
BibTeX
@misc{banerjee2010shape,
  author = {Abhijit Banerjee and Sendhil Mullainathan},
  title = {{The Shape of Temptation: Implications for the Economic Lives of the Poor}},
  year = {2010},
  doi = {10.2139/ssrn.1598547},
  howpublished = {SSRN},
}

Loading related content...

Figure