Chapter 21 Quiz: Self-Assessment

Instructions: Answer each question without looking back at the chapter. After completing all questions, check your answers against the key at the bottom. If you score below 70%, revisit the relevant sections before moving on to Part IV.

Multiple Choice

Q1. The term "cobra effect" refers to:

a) The danger of introducing invasive species into new environments b) A situation in which an incentive designed to solve a problem makes the problem worse c) The tendency of colonial governments to misunderstand local ecosystems d) A feedback loop in which predator and prey populations oscillate

Q2. The British cobra bounty in Delhi failed because:

a) The bounty was too small to motivate cobra hunting b) Citizens bred cobras to collect the bounty, increasing the cobra population c) Cobras learned to avoid hunters d) The bounty attracted foreign hunters who disrupted the local ecosystem

Q3. The French rat-tail bounty in Hanoi illustrates the cobra effect because:

a) Citizens killed rats but did not bring in the tails b) The rat population naturally increased during the bounty period due to seasonal breeding c) Citizens cut tails from live rats and released them to breed, and some bred rats for the tails d) The bounty was paid in a currency that was losing value

Q4. The HFC-23 emissions trading loophole is an example of the cobra effect because:

a) The carbon credit system created an incentive to produce more of the harmful substance whose byproduct generated valuable credits b) Carbon credits were traded on a speculative market that inflated their value c) Developing countries refused to participate in the emissions trading system d) The credits were denominated in a currency that depreciated

Q5. The welfare cliff creates a cobra effect because:

a) Welfare recipients are inherently less motivated to work b) The abrupt loss of benefits at income thresholds creates a financial incentive not to earn more, trapping recipients in poverty c) Government bureaucracies are too slow to process benefit adjustments d) Welfare benefits are too generous, removing any incentive to work

Q6. The Streisand effect is a cobra effect in the domain of:

a) Real estate law b) Celebrity culture c) Information suppression -- the attempt to suppress information amplifies it d) Environmental regulation

Q7. "Cash for Clunkers" exhibited cobra effects because:

a) The program had no environmental goals b) Consumers refused to participate c) It primarily shifted purchases in time rather than creating new demand, destroyed usable vehicles that raised used car prices for the poor, and produced questionable environmental benefits d) The auto industry used the program to raise prices on new vehicles

Q8. The chapter identifies incentives as a form of:

a) Market failure b) Government regulation c) Feedback loop d) Natural selection

Q9. The concept of "incentive ecology" refers to:

a) The environmental impact of economic incentives b) The full landscape of strategic responses -- intended and unintended -- that an incentive creates c) The use of incentives in ecological conservation d) The natural selection of effective incentive systems over time

Q10. Mechanism design is described as:

a) The engineering of physical mechanisms like levers and pulleys b) "Reverse game theory" -- starting with desired behavior and designing rules to produce it c) The study of how markets naturally evolve d) A branch of philosophy concerned with free will

Q11. The Vickrey auction achieves incentive compatibility by:

a) Requiring all bidders to bid simultaneously b) Having the highest bidder win but pay only the second-highest bid, making honest bidding the dominant strategy c) Allowing bidders to see each other's bids and adjust d) Setting a minimum price that prevents underbidding

Q12. Incentive compatibility means:

a) All agents have the same incentives b) Agents maximize their own welfare by behaving in a way that also maximizes the system's welfare c) The incentive is compatible with existing laws and regulations d) All agents agree to follow the rules

Q13. The Gibbard-Satterthwaite theorem shows that:

a) All incentive systems will eventually fail b) In many common settings, no mechanism can be simultaneously incentive-compatible, efficient, and budget-balanced c) Cobra effects are mathematically impossible to prevent d) Auctions are always the most efficient allocation mechanism

Q14. Munger's dictum -- "Show me the incentive and I'll show you the outcome" -- is interpreted in this chapter as:

a) A celebration of the power of incentives to control behavior b) A warning that the outcome of an incentive is determined by the ecology it creates, not by the designer's intention c) An argument against using incentives in public policy d) A recommendation to always use monetary incentives rather than non-monetary ones

Q15. The first law of perverse incentives states:

a) Incentives always produce the intended behavior b) Every incentive creates its own ecology of strategic responses c) Monetary incentives are always more effective than non-monetary ones d) Perverse incentives only affect large organizations

Q16. The third law of perverse incentives -- "Proxies are always vulnerable" -- means:

a) No measurement system can be accurate b) Every incentive system uses a measurable substitute for the thing it actually cares about, and agents will optimize the substitute at the expense of the underlying goal c) Government proxies are less reliable than private sector ones d) Proxy voting in elections is vulnerable to manipulation

Q17. The fifth law -- "Removing an incentive can be worse than never having created it" -- is illustrated by:

a) The success of the rat-tail bounty program b) The cancellation of the cobra bounty, which led to the release of bred cobras, making the problem worse than the starting point c) The continuation of the emissions trading system despite its flaws d) The gradual phase-out of welfare benefits

Q18. The Part III Retrospective argues that all eight failure modes are manifestations of:

a) Human greed and selfishness b) The gap between the model and the system -- interveners act on models that are simpler than the systems they are trying to manage c) Insufficient funding for government programs d) The inherent unpredictability of complex systems, which makes planning impossible

Q19. According to the diagnostic toolkit, a system is in "deep structural trouble" when:

a) One failure mode is operating b) Two failure modes are operating c) Three or more failure modes are operating simultaneously and reinforcing each other d) All eight failure modes are operating simultaneously

Q20. The chapter's threshold concept -- "Incentives Create Their Own Ecology" -- means:

a) Incentives should be designed by ecologists b) When you create an incentive, you create an entire landscape of strategic responses, many of which you never anticipated, and some of which directly undermine your goal c) Incentives are a form of natural selection that produces optimal outcomes d) Only ecological systems are vulnerable to perverse incentives

Short Answer

Q21. In two to three sentences, explain the structural similarity between the cobra bounty in Delhi and the welfare cliff in the United States. What is the "cobra" in each case, and what is the "bounty"?

Q22. The chapter argues that "every cobra effect involves a Goodhart failure, but not every Goodhart failure escalates to a full cobra effect." In your own words, explain the difference. What additional element must be present for a Goodhart failure to become a cobra effect?

Q23. Explain why mechanism design is so difficult in complex, real-world environments, even though it succeeds in controlled settings like the Vickrey auction. What features of complex environments make perfect incentive compatibility unachievable?

Q24. The chapter connects the cobra effect to feedback loops (Chapter 2). In your own words, explain how the intended feedback loop of the cobra bounty differed from the actual feedback loop that emerged. Why did the unintended loop dominate?

Q25. Choose one failure mode from Part III other than the cobra effect and explain how it interacts with the cobra effect in a real or hypothetical system. How do the two failure modes reinforce each other?

Answer Key

Multiple Choice:

Q1: b -- The cobra effect is a situation in which an incentive designed to solve a problem makes the problem worse, named after the Delhi cobra bounty. (Section 21.1)

Q2: b -- Citizens bred cobras specifically to collect the bounty, increasing the total cobra population rather than reducing it. (Section 21.1)

Q3: c -- Citizens discovered that cutting tails from live rats and releasing them was more profitable than killing rats, and some bred rats specifically for the bounty. (Section 21.2)

Q4: a -- The carbon credit system made it profitable to produce more HCFC-22 (generating HFC-23 as a byproduct) so that the byproduct could be destroyed for valuable carbon credits. Production increased because the byproduct was worth more than the product. (Section 21.3)

Q5: b -- The abrupt loss of benefits at income thresholds creates effective marginal tax rates that can exceed 100%, making it financially rational to stay below the threshold. The system designed to help people escape poverty creates a financial incentive to remain in poverty. (Section 21.4)

Q6: c -- The Streisand effect is a cobra effect in the information domain: the attempt to suppress information (the "incentive" to protect privacy through legal action) amplifies the information instead of suppressing it. (Section 21.5)

Q7: c -- Multiple cobra effects operated: temporal displacement of purchases, destruction of usable vehicles that raised costs for low-income consumers, and questionable net environmental benefits when manufacturing emissions were included. (Section 21.5)

Q8: c -- The chapter argues that an incentive is fundamentally a feedback loop -- an attempt to create a loop that guides behavior toward a desired outcome. (Section 21.6)

Q9: b -- Incentive ecology is the full landscape of strategic responses that an incentive creates, including intended responses, unintended responses, and responses that the designer never imagined. (Section 21.6)

Q10: b -- Mechanism design starts with the behavior you want and tries to design rules (mechanisms) that will produce it, the reverse of game theory which starts with rules and predicts behavior. (Section 21.7)

Q11: b -- In a Vickrey auction, the highest bidder wins but pays the second-highest bid. This makes honest bidding the dominant strategy because there is no advantage to bidding above or below your true valuation. (Section 21.7)

Q12: b -- Incentive compatibility means that agents acting in their own self-interest will also produce the outcome that is best for the system as a whole. Individual interest and collective interest are aligned. (Section 21.7)

Q13: b -- The theorem establishes fundamental limits on mechanism design: in many settings, you cannot simultaneously achieve incentive compatibility, efficiency, and budget balance. You must sacrifice at least one. (Section 21.7)

Q14: b -- The chapter interprets Munger's dictum as a warning: the outcome is determined by the incentive ecology, not by the designer's intention. The designer controls the signal, but the ecology controls the response. (Section 21.8)

Q15: b -- The first law states that an incentive is not a simple push toward a desired behavior but a signal that creates an entire ecology of strategic responses, many of which the designer did not anticipate. (Section 21.8)

Q16: b -- Every incentive system uses a proxy (a measurable substitute) for the underlying goal. Agents will optimize the proxy, and when manipulating the proxy is cheaper than performing the intended behavior, the proxy decouples from the goal. (Section 21.8)

Q17: b -- When the British cancelled the cobra bounty, the cobra farmers released their stock, leaving Delhi with more cobras than before the program began. The system did not return to its starting state. (Section 21.8)

Q18: b -- All eight failure modes occur because someone acts on a model that is simpler than the system. The gap between model and system is where each failure mode lives. (Section 21.9)

Q19: c -- Three or more simultaneously operating failure modes indicate deep structural trouble because they reinforce each other. Addressing one without addressing the others is likely to be iatrogenic. (Section 21.9)

Q20: b -- The threshold concept holds that creating an incentive creates an entire ecology of strategic responses, extending far beyond the intended behavior, many of which undermine the designer's goal. (Section 21.6)

Short Answer Rubric:

Q21: In both cases, the system rewards the condition it is designed to alleviate. The cobra bounty rewards the presence of dead cobras (creating incentives to produce more cobras). The welfare system rewards having low income (creating incentives to keep income low). The "cobra" in the welfare case is poverty itself; the "bounty" is the benefits that disappear when income rises. Both systems create perverse incentives because the reward structure makes the target condition (cobras existing, income being low) financially valuable.

Q22: A Goodhart failure occurs when a metric being used as a target decouples from the underlying goal under optimization pressure. A cobra effect requires an additional element: the gaming of the metric must make the original problem worse, not merely fail to solve it. A cobra effect is a Goodhart failure where the divergence between metric and goal is not just a gap but a reversal -- the optimized behavior actively undermines the objective. Example of Goodhart without cobra: teaching to the test reduces education quality but does not make students less educated than they were before testing existed. Example of cobra: the cobra bounty increased the cobra population beyond pre-bounty levels.

Q23: Mechanism design succeeds in controlled settings (like the Vickrey auction) because those settings have limited agents, simple strategy spaces, independent valuations, and enforceable rules. Real-world environments are complex because they have many heterogeneous agents with diverse strategies, correlated valuations, information asymmetries, incomplete enforcement, and the ability to collude, innovate new strategies, or exit the mechanism entirely. The Gibbard-Satterthwaite theorem formally demonstrates that perfect incentive compatibility is unachievable in many realistic settings.

Q24: The intended feedback loop was balancing: bounty motivates cobra killing, cobra killing reduces cobra population, reduced population reduces need for bounty. The actual feedback loop was reinforcing: bounty makes cobras valuable, valuable cobras are bred, more cobras generate more bounty payments, more payments justify more breeding. The unintended loop dominated because breeding cobras was easier and more profitable than hunting them -- the reinforcing loop had a stronger gain than the balancing loop.

Q25: Any coherent connection between the cobra effect and another Part III failure mode, with a specific example showing mutual reinforcement. For instance: Goodhart's Law (Ch. 15) and the cobra effect reinforce each other when a metric used as a management target creates a cobra-effect incentive to game the metric. The Goodhart failure (metric decoupling from goal) enables the cobra effect (gaming makes the problem worse), and the cobra effect validates continued reliance on the metric (gaming improves the metric while degrading reality, so the metric says the system is succeeding while it is failing).