Time Horizons in Interdependent Security International Conference on Social Dilemmas Kyoto, Japan August 24, 2009 David Hardisty, Howard Kunreuther, David.

Time Horizons in Interdependent Security

International Conference on Social DilemmasKyoto, JapanAugust 24, 2009

David Hardisty, Howard Kunreuther, David Krantz, & Poonam Arora

Columbia University

Co-Authors

NSF Grants SES-0345840 & SES-0820496

Howard Kunreuther Dave Krantz Poonam Arora

Thank You ICSD 2009 Organizing Committee

• Satoshi Fujii

• Toshio Yamagishi

• Tsuyoshi Hatori

• Akira Kikuchi

• Haruna Suzuki

IDS Background

• Interdependent Security (IDS) is a social dilemma with stochastic losses

examples:

• border security

• pest/disease control

• risky investment

Previous Findings on IDS• Individuals cooperate less in IDS than in a

typical repeated PD (Kunreuther et al, in press)• However groups cooperate more in IDS (Gong,

Baron & Kunreuther, in preparation)

Group and Uncertai nty I nteract i on

0%

25%

50%

75%

100%

1 2

Pl ayer Type

Prop

orti

on o

fIn

vest

ing

Determi ni sti cGame

Stochasti cGame

Group I ndi vi dual

Research Motivation (1)

• Previous studies used probabilities of 20% to 80%

• Real life risks are often much lower


• In real life, players often precommit their strategy (whether to invest in protection) for several years in advance at a time

• example: CO2 reductions


• Normally, greater delay is associated with increased uncertainty

• example: $10 promised today or in 20 years

• However, with repeated low probability events, increasing time horizon may increase subjective probability

• example: chance of fire today or in the next 20 years

Study 1Question 1: Do previous findings of low

(~30%) cooperation under uncertainty hold with low probabilities? Or will it be even lower?

Question 2: Does precommitment raise investment rates?

IDS instructions (pg 1)

Scenario: Imagine you are an investor in Indonesia and you have a risky joint venture that earns 8,500 Rp per year. However, there is a small chance that you and/or your counterpart will suffer a loss of 40,000 Rp in a given year. You have the option to pay 1,400 Rp for a safety measure each year to protect against the possible loss. However, you will only be fully protected if both you and your counterpart invest in protection. The loss has an equal chance of happening each year, regardless of whether it occurred in the previous year.

IDS payoff matrixYour Counterpart

INVEST NOT INVEST

You INVEST - You definitely lose 1,400 Rp, and have a 0% chance of the large loss occurring.

- Your counterpart definitely loses 1,400 Rp, and has a 0% chance of the large loss occurring.

- You definitely lose 1,400 Rp and have a 1% chance of losing an additional 40,000 Rp.

- Your counterpart has a 3% chance of losing 40,000 Rp and a 97% chance of losing 0 Rp.

NOT INVEST

- You have a 3% chance of losing 40,000 Rp and a 97% chance of losing 0 Rp.

- Your counterpart definitely loses 1,400 Rp and has a 1% chance of losing an additional 40,000 Rp.

- You have a 4% chance of losing 40,000 Rp and a 96% chance of losing 0 Rp.

- Your counterpart has a 4% chance of losing 40,000 Rp and a 96% chance of losing 0 Rp.

PD payoff matrixYour Counterpart

INVEST NOT INVEST

You INVEST - You lose 1,400 Rp.- Your counterpart loses 1,400 Rp.

- You lose 1,800 Rp. - Your counterpart loses 1,200 Rp.

NOT INVEST

- You lose 1,200 Rp.- Your counterpart loses 1,800 Rp.


IDS: Choices

Will you invest in protection this year? INVEST | NOT INVEST

Do you think your counterpart will invest in protection this year?

DEFINITELY | PROBABLY | PROBABLY NOT | DEFINITELY NOT

Precommitted Condition

Will you invest in protection in year 1? INVEST | NOT INVEST

Do you think your counterpart will invest in protection in year 1?DEFINITELY | PROBABLY | PROBABLY NOT | DEFINITELY NOT

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~



~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ [...]



Feedback

Year 1 Results

Your choice: INVEST

Your counterpart's choice: NOT INVEST

The random number was: 88

This Means

For you, the large loss: did not occur

For your counterpart, the large loss: did not occur

Result: You lost 1,400 Rp, and your counterpart lost 0 Rp

Design Details

• participants played blocks of 20 rounds (“years”) with an anonymous partner

• 4 blocks total

• random pairing before each block

• 1 block paid out for real money

• all manipulations between subject, 30 subjects per group

PD vs IDS

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Block 1 Block 2 Block 3 Block 4

Inve

stm

ent

Pro

po

rtio

n

IDS rep

PD rep

PD vs IDS

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Inve

stm

ent

Pro

po

rtio

n

IDS rep

PD rep

PD vs IDS

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Inve

stm

ent

Pro

po

rtio

n

IDS rep

PD rep

Conclusion 1

• Uncertainty lowers cooperation between individuals playing loss framed dilemma

• Why? - Perhaps uncertainty makes players more greedy (Johansson & Svedsater, yesterday)- Perhaps uncertainty transforms the game from a social dilemma to a game of chance

IDS: repeated vs precommitted

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Inve

stm

ent

Pro

po

rtio

n

IDS rep

IDS pre

PD rep

IDS: repeated vs precommitted

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Inve

stm

ent

Pro

po

rtio

n

IDS rep

IDS pre

PD rep

Conclusion 2

• Under uncertainty, precommitment raises cooperation

• Why? Perhaps precommitment raises subjective probability of the loss

Precommitted Participants Estimated Higher Cumulative Probability

Interesting results, but a major confound:

Feedback

Study 2Question: Do individuals playing a

(non-dilemma) solo game invest more often when precommitting?

Solo payoff matrix

INVEST - You definitely lose 1,400 Rp, and have a 0% chance of the large loss occurring.

NOT INVEST - You have a 4% chance of losing 40,000 Rp and a 96% chance of losing 0 Rp.

IDS repeated vs Solo repeated

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Inve

stm

ent

Pro

po

rtio

n

IDS rep

IDS pre

Solo rep

Solo pre

IDS repeated vs Solo repeated

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Inve

stm

ent

Pro

po

rtio

n

IDS rep

IDS pre

Solo rep

Solo pre

Conclusion 3

• IDS players are mostly playing a game of chance, showing risk-seeking for losses

Support 3

Solo: repeated vs precommited

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Inve

stm

ent

Pro

po

rtio

n

IDS rep

IDS pre

Solo rep

Solo pre

Solo: repeated vs precommited

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Inve

stm

ent

Pro

po

rtio

n

IDS rep

IDS pre

Solo rep

Solo pre

Conclusion 4

• Precommitment raises investment rates by individuals

• Why? Perhaps subjective probability is increased

How else can we improve investment under uncertainty?

• Perhaps environmental framing can highlight social goals and raise investment rates

• However, earlier results are mixed

Study 3Question: Will environmental

framing will increase investment rates?

IDS environmental instructions

Scenario: Imagine you are a farmer in Indonesia. You get an annual yield of 8,500 Rupiah (Rp) from your potato crops. Both you and a neighboring farmer use the pesticide Aldicarb on your potato crops. However, there is a small risk of groundwater contamination each year from this pesticide, which is toxic. If contamination occurs, you and/or your neighboring farmer will suffer a loss of 40,000 Rp, to pay for groundwater cleanup. You have the option to switch to a more expensive, though safer, pesticide, at the cost of 1,400 Rp annually, to avoid groundwater contamination. However, you will only be fully protected if both you and your counterpart invest in the safer pesticide. The groundwater contamination has an equal chance of happening each year, regardless of whether it occurred in the previous year.

IDS environmental payoff matrixYour Counterpart

INVEST NOT INVEST

You INVEST - You definitely lose 1,400 Rp, and have a 0% chance of groundwater contamination.

- Your counterpart definitely loses 1,400 Rp, and has a 0% chance of groundwater contamination.

- You definitely lose 1,400 Rp and have a 1% chance of groundwater contamination occuring and losing an additional 40,000 Rp.

- Your counterpart has a 3% chance of losing 40,000 Rp due to groundwater contamination and a 97% chance of losing 0 Rp.

NOT INVEST

- You have a 3% chance of losing 40,000 Rp due to groundwater contamination and a 97% chance of losing 0 Rp.

- Your counterpart definitely loses 1,400 Rp and has a 1% chance of groundwater contamination occuring and losing an additional 40,000 Rp.

- You have a 4% chance of groundwater contamination occurring and losing 40,000 Rp and a 96% chance of losing 0 Rp.

- Your counterpart has a 4% chance of groundwater contamination occurring and losing 40,000 Rp and a 96% chance of losing 0 Rp.

IDS Environmental: Choices

Will you invest in the safer pesticide this year? INVEST | NOT INVEST

Do you think your counterpart will invest in the safer pesticide this year?

DEFINITELY | PROBABLY | PROBABLY NOT | DEFINITELY NOT

Feedback

Year 1 Results

Your choice: INVEST

Your counterpart's choice: NOT INVEST

The random number was: 88

This Means

For you, groundwater contamination: did not occur

For your counterpart, groundwater contamination : did not occur

Result: You lost 1,400 Rp, and your counterpart lost 0 Rp

Environmental Frame Results

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Inve

stm

ent

Pro

po

rtio

n

IDS rep

IDS pre

IDS rep E

IDS pre E


0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Inve

stm

ent

Pro

po

rtio

n

IDS rep

IDS pre

IDS rep E

IDS pre E


0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Inve

stm

ent

Pro

po

rtio

n

IDS rep

IDS pre

IDS rep E

IDS pre E

Conclusion 5

• Environmental framing may not have a significant effect on investment rates

Study 4Question: How does precommitment

affect investment rates in a deterministic prisoners dilemma

IDD payoff matrixYour Counterpart

INVEST NOT INVEST

You INVEST - You lose 1,400 Rp.- Your counterpart loses 1,400 Rp.

- You lose 1,800 Rp. - Your counterpart loses 1,200 Rp.

NOT INVEST



PD: Repeated vs Precommitted

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Inve

stm

ent

Pro

po

rtio

n

PD rep

PD pre

PD: Repeated vs Precommitted

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Inve

stm

ent

Pro

po

rtio

n

PD rep

PD pre

Conclusion 6

• Precommitment reduces investment rates in deterministic social dilemmas

• Why? Perhaps individuals realize there is no opportunity for reciprocity and are worried about being a sucker

Summary

• Precommitment lowers cooperation in regular prisoner’s dilemma, but raises it in interdependent security situations

• Why? In IDS, precommitment raises subjective probability of loss, but in the deterministic case it removes the possibility of reciprocity

Thank You!

References

Gong, M. J. Baron and H. Kunreuther (2008). Group Cooperation under Uncertainty. Wharton Risk Center Working Paper # 2008-11-24

Kunreuther, H., G. Silvasi, E. Bradlow, and D. Small (in press). Deterministic and Stochastic Prisoner's Dilemma Games: Experiments in Interdependent Security. Judgment and Decision Making.

Xiao, Erte & Kunreuther, H. (in preparation). Punishment and Cooperation in Stochastic Social Dilemmas.

Questions

• Do you believe the risk perception story?

• What new conditions would be most interesting? All-or-nothing precommitment? Group play? Change the payoffs (so non-investment dominates)?

• What journals do you recommend publishing in?

• What are the biggest holes in this story?

What about individual differences?

Comprehensive Health Insurance

Gender

Gender

First-Round Comparison

Overall Comparison

20-Round Profile

80 Round Profile

What did people say?

IDS - Repeated• The probability of loosing was too low so I didn't decide to invest.• For the most part, it seemed better NOT to invest than to invest. I

found that the initial 1-5 years influenced how I invested in the remaining 15 years. So if I mainly did NOT invest the first 5 years, then I didn't invest for the remaining 15 years. I also found that my partner followed how I invested if I had no losses.

• I chose not to invest dut to low probality of loss• It was a little intimidating, but after awhile an understanding

occurred between myself and my partner and we flowed fairly well over the years.

• I chose to invest in the beginning so I wouldn't end up with negative numbers. Afterwards, I chose not to invest because I figured the chances of losing 40,000 were small, and even if I lost the money once, I would probably not lose it again in the 20 year span.

IDS - Precommitted• I chose to invest almost all 20 times except for one or 2 years to

make an extra bonus• An interesting game where I can observe that some people do take

risks. I chose to invest in every year and every scenario since statistically it makes more sense to invest in protection. The loss is big compared to the amount spent on protection and there is an average of more than one occurrence of loss in a 20 year period even if only one person invests.

• I invested too many times the first round, and when I saw that my partner rarely invested and suffered little losses, I invested less as well by the next round. I felt rather smug when he/she lost 40k while I invested-- but I thought it was interesting that I invested a lot more often than most of my counterparts.

• Initially I was just playing it safe, and then for the 2 later rounds I pretty much went with the Nash Equilibrium choice of Not Investing.

Solo - Repeated

• At first, it seemed as though investing would be a good idea. After the second round, I realized that not investing would probably give a better payout because the chances of actually suffering a loss was so slim.

• I mainly choose to not invest because there was only a 4% chance of losing 40000 Rp. But then every 4 or 5 turns, I would randomly decide to invest in protection. Now that I think about it, even if I hadn't gone with investing at all, the outcome may have been the same or maybe even better.

• I would invest at points where I thought that I was happy enough with my cumulative gains that I didn't want to risk losing what I had already gained.

Solo - Precommitted

• It's obvious safer to invest, and the investment totally worth that. Even if you invest for 20 years, the cost is lower than one large loss which might happen every year. However, as there was no large loss happened in the first session, I therefore chose to take some risks in following sessions.

• The potential loss from investing every time seemed less than from not investing more often where even one bad year could produce less income.

• 40000 Rp possible loss * 4% chance = 1600 Rp loss expected. So in general it's worth it to invest. However, given the low likelihood, it might be worth it to take a risk & skip some years.

Why Groups Cooperate More than Individuals in SPD

- Safety Oriented to Avoid Guilt and Blame

• People take less risk when their decisions affect others (Charness and Jackson 2008)

• In DPD– Defection is always better off– No ex post guilt or blame

• In SPD – Defection has a higher expected payoff, but also a

higher probability of suffering a loss.– If a large loss follows defection ex post guilt and ex

post blame for the one who suggested defection– Group members favor safety-oriented strategy

(cooperation) to avoid ex post guilt and blame Reversed discontinuity effect

Why Groups Cooperate More than Individuals in SPD -Social Pressure to Conform to Certain Norms

• Three norms are most relevant in PD games: Being pro-group, Being smart, and Being nice

• Similar to the group morality and individual morality argument (Cohen 2006; Pinter 2007)

• In DPD – Both pro-group (group morality) and smart norm

clearly indicate defection – Niceness (individual morality) is clouded

• In SPD– Unclear what strategy is pro-group and smart– Being nice is socially desirable and more salient than

in DPD.

Time Horizons in Interdependent Security International Conference on Social Dilemmas Kyoto, Japan August 24, 2009 David Hardisty, Howard Kunreuther, David.

Documents

rp slide

lower slide

preparation slide

group slide

counterpart investnot

previous year

pd vs ids

ids individuals