T02 Cole et al. (2015): Incentivizing Calculated Risk-Taking

Question 1

What is the setup of the ‘framed field experiment’ (sample, task, treatment)?

Cole et al. (2015): Incentivizing Calculated Risk-Taking

Answer 1

• Sample: Loan officers
  
  • median 35y old + 10y experience
  • from commercial banks
  • representative
• Task: Evaluation of credit applications
  
  • outcomes are observed
  • no soft information
  • mass market for small business loans
• Treatment: Exogenously assigned monetary incentives with different
  
  • incentive power (high, low, origination)
  • time horizon of payment (deferred yes/no)
  • degree of limited liabilty yes/no
  • cost of information yes/no

Question 2

What do the incentives look like?

Cole et al. (2015): Incentivizing Calculated Risk-Taking

Answer 2

Question 3

What are the 4 outcome variables examined?

Cole et al. (2015): Incentivizing Calculated Risk-Taking

Answer 3

1. Screening effort (information reviewed)
2. Risk assessment (internal rating=credit score)
3. Lending decisions (approval rates)
4. Loan profitability

Question 4

What are the 5 predictions?

Cole et al. (2015): Incentivizing Calculated Risk-Taking

Answer 4

1. Origination incentives lead to lower screening effort, higher approval rates, and more defaults
2. High-powered incentives lead to higher screening effort, higher ratings of approved loans, lower approval rates, less defaults
3. Deferred compensation reduces power of performance-based incentives
4. Relaxing limited liability increases screening effort
5. Reputational concerns or intrinsic motivation increase effort

Question 5

What is the method used? Explain 5 general attributes of it.

Cole et al. (2015): Incentivizing Calculated Risk-Taking

Answer 5

Fixed-effects regression:

• Only for panel data
• Fixed effects control for time invariant unobservables
• Time invariant unobservables can be correlated with regressors
• Cannot estimate the effect of time invariant regressors
• Similar to first-differencing

Question 6

Results: Effect of Incentives on the 4 outcome variables

Cole et al. (2015): Incentivizing Calculated Risk-Taking

Answer 6

1. On effort:
   - High-powered incentives increase search effort, even stronger under costly information.
   - Origination bonus does not affect reduce effort
2. On Risk Assessment:

• Origination slightly increases internal ratings
• High-powered incentives reduces approval of loans with higher ex ane uncertainty

3. On Lending Decision:
   - High-powered incentives reduce approval rate of nonpeforming loans
   - Origination incentives increase approval rate of performing loans
4. On Profitability:
   - High-powered incentives improve profits by 3%

Question 7

Results: Role of deferred compensation in the incentive effect on the 4 outcome variables.

Cole et al. (2015): Incentivizing Calculated Risk-Taking

Answer 7

1. On Effort:
   - mutes effect of high-powered incentives on effort
2. On Risk Assessment:

• leads to more conservative lending under low-powered incentives
• leads to less conservative lending under high-powered-incentives

3. On Lending Decision:
   - mutes effect of high-powered incentives on loan approval rate
4. On Profitability:
   - decrease profits under high-powered incentives
   - increase profits under origination incentives

Question 8

What is the problem and the solution of limited liability? How is it relaxed in the paper and what are the results?

Cole et al. (2015): Incentivizing Calculated Risk-Taking

Answer 8

• Problem: Incentive to take excessive risk as loan officers have no downside risk
• Solution: Pay for each defaulted loan
• Modeling: Receive initial endowment from which negatie payments are subtracted
• Results:
  
  • Neither effort, lending decisions nor profits are different from high-powered treatment
  • Approved loans seem to have lower quality(!), result can be questioned

Question 9

What is the role of loan officer characteristics?

Cole et al. (2015): Incentivizing Calculated Risk-Taking

Answer 9

• Risk-aversion and altruism increase effort
• Overconfidence decreases effort
• Optimism results in higher risk ratings
• Risk-aversion leads to less approved loans
• Impatience increases approval rate of nonperforming loans
• Higher discount factor increases approval rates of nonperforming loans
• Effect of incentives does not vary with personality traits

Question 10

7 Conclusions

Cole et al. (2015): Incentivizing Calculated Risk-Taking

Answer 10

1. Origination incentives lead to lower effort, higher approval rates and more defaults
2. High-powered incentives lead to higher effort and higher profitability
3. Deferred pay reduces the effect of incentives
4. Deferred pay reduces the origination of nonperforming loans
5. Loan officer characteristics are important but do not alter the effect of incentives
6. Incentives may be bounded by fixed personaliyt characteristics
7. Incentives distort perception of risk