You are on page 1of 28

ECA5101 Presentation:

Do people drive more because they paid more for


their cars??
Literature Review on Sunk Cost Fallacy in Driving
the Worlds Costliest Cars
Presented by:
Fan YiJun
Cao WanZhong
Zhou BaiQi
Jonathan Chan
Introduction to:
Economy Theory
& Singapores Car Policy
YiJun
Main Findings
& Model Specification
WanZhong
Empirical Design,
Econometrics Interpretation,
Alternative explanation
BaiQi
Robustness testing
Policy Implications
& Opinion on paper
Jonathan
Introduction to underlying economic theory
Key Microeconomic Concept:
Sunk Cost Fallacy: To continue doing something if substantial amounts of money, time or
effort have already been invested, even when continuing is not the best thing to do.
Appropriate setting to test theory?
Buying a car is a big ticket item in Singapore, where there is substantial variation in new car
purchase prices over time due to COE policies
2 Possible effects of higher sunk cost:
More effort invested to correct irrational bias
Irrational bias still exist (Proven in this paper)
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Introduction to Singapores Car Policies
= 1 +

+ + -
OMV, Open Market Value: Import price of a car, may fluctuate according to
manufacturers pricing strategy
ARF, Additional Registration Fee: Fee for new car registration, currently set at
100% of OMW
Therefore ARF=f(OMV)
COE, Certificate of Entitlement AKA Quota Premium:
Right to own a car for 10 years from day of purchase
cannot be detached from original car of registration
Floating price is determined by a bi-monthly auction
Supply is fixed by a monthly quota based on:
i.Specific growth rate of car population
ii.Number of de-registration in preceding time period
Main cause of variations of car retail prices
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Introduction to Singapores Car Policies
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Note the large swings in COE prices, highly volatile
Rapid over the years due to quotas released coupled with Income (QE)
Introduction to Singapores Car Policies
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Note the immediate 20% in COE value the moment the new car is handled over to the
owner
Linear of 8% yearly for the next 10 years
0 1 2 3 4 5 6 7 8 9 10
% COE Remaining 80% 72% 64% 56% 48% 40% 32% 24% 16% 8% 0%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Years
% COE Value remaining after x no of years
Introduction to Singapores Car Policies
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Note the immediate 25% in ARF value the moment the new car is handled over to the
owner
No further decline for the 1
st
5 years, after which there is a step-down 15% yearly
1 2 3 4 5 6 7 8 9 10
% ARF Remaining 75% 75% 75% 75% 75% 60% 45% 30% 15% 0%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Years
% ARF Value remaining after x no of years
Main Findings
Proposition 1: Sunk CostUsage:
Notice the correlation between Policy-related Sunk Cost & Average monthly usage
Robust evidence of sunk cost fallacy (Irrational Behaviour)
1
st
4 years: S$1,000 sunk cost 142.666km annual usage (3.4%)
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Main Findings
Proposition 2: Attenuation
Age of carsunk cost
> Sunk Cost> (Rate of Sunk Costover time)
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Model Specification
2 Proposed Behaviour Models:
Conventionally Rational Behaviour
Mental Accounting for Sunk Cost
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Model Specification
Conventionally Rational Behaviour (CRB):

, = (

, ) (

, ) ()

, = f (Novelty, breakdown frequency)


, = ( / , / )
= (
+++
)
Find Optimal usage by maximising utility
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
*
Model Specification
Mental accounting for sunk cost (MASC):

, |

, Sunk Cost Adjustment

, | >

,
Where Sunk Cost(SC) Adjustment
= 0 SC sensitivity factor 1


Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Model Specification
Mental accounting for sunk cost (MASC):

, |

, Sunk Cost Adjustment


Find Optimal usage by maximising utility
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
in optimal usage due Sunk Cost Adjustment

*
Model Specification
= +
= [0.2

+ 0.25

] + [ 1 +

+ ]
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Policy-related Sunk Cost Parameters from Jun 2001-May 2009:
Mean: S$16,364
Standard Deviation: S$3400
Empirical Design
Data Sources:
Car Retail price & Usage: Car servicing records from a single undisclosed
authorised dealer for middle class brand of cars in Singapore
Monthly OMW, ARF, Retail Price & COE: Land Transport Authority (LTA),
Singapore
Fuel Cost: Consumer Price Index (CPI) of 98 Octane Petrol
Key features of sample set:
Panel data
Exclusion of resold cars
Exclusion of outliers (> 2 Standard Deviations of average monthly usage)
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Econometrics Interpretation
Proposition 1: Sunk CostUsage:
OLS Regression with Robust Standard Errors:

= 2.395 +0.013 +1.674 0.002


2
(2.362) (0.009) (0.068***) (0.001***)

2
= 0.541

= Cumulative usage
= Price (S$000)
= Age in months
*** Level of significant effort: p<0.01
Price coefficient is not significantly different from 0 for this regression
specification, hence cannot prove +ve correlation with

.
Need to re-specify model
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Econometrics Interpretation
Proposition 1: Sunk CostUsage:
Re-specified OLS Regression with Robust Standard Errors:

= 8.549 + 1.409

0.001

+1.775 0.003
2
(2.153***) (0.169***) (0.121) (0.071***) (0.001***)

2
= 0.545

= Cumulative usage

= Sunk Cost Proportion of COE Premium

= Sunk Cost Proportion of ARF


= Age in months
*** Level of significant effort: p<0.01
Most coefficients are significantly different from 0, there is a significant +ve
correlation between sunk cost proportion of COE & cumulative usage.
Proposition 1: Sunk CostUsage: Correlation proven
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Econometrics Interpretation
Proposition 2: Attenuation
Age of carsunk cost
> Sunk Cost> (Rate of Sunk Costover time)
Mental amortizing sunk cost relative to some target cumulative usage
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Econometrics Interpretation
Proposition 2: Attenuation
i) Maximum Simulated Likelihood(MSL) of Mental Accounting of Sunk Cost
(MASC) model:
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Notice how all the Sunk Costs, over the different time
horizons are +ve & significantly different from 0
Consistent with Attenuation
*4 Year Horizon preferred as it has the highest Log
likelihood value of -20736.6, indicating best fit among
the 3 time horizons
Econometrics Interpretation
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
i) Maximum Likelihood Estimate(MLE) of Conventionally Rational Behaviour
(CRB) model:
*
Can reject the Conventionally Rational Behaviour (CRB) model in
favour of the 4 year horizon specific model as the latter has a > Log
Likelihood with more variable coefficients significantly different
from 0, hence yielding a better fit.
Alternative Explanation
Bias Selection:
Higher prices (Higher sunk cost) would be more likely to attract people who
are planning to drive more.
Shown that there is correlation between prices and usage
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Pay more Sunk Cost Drive more
Mental Accounting of Sunk Cost (MASC)
Bias Selection
Establishing Causation
Alternative Explanation
Why Bias Selection cannot hold:
If the people who were planning to drive more were the ones paying the
higher sunk cost, then there should not be attenuation over time as their
usage should remain constant in accordance with their initial usage plan
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Pay more Sunk Cost Drive more
Mental Accounting of Sunk Cost (MASC)
Bias Selection
Establishing Causation
Robustness Testing
Check the sensitivity of initial model by running multiple MSL model variations controlling for
other factors (Car Size, Sunk Cost as % of retail price, COE Salience, Heterogeneous target usage)
Benchmark model is the 4 Year Time Horizon MSL equation as specified on slide 19
Notice how the Sunk Cost Elasticity of Usage stays relatively constant


=
%

%
Where = Sunk Cost sensitivity factor
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Policy Implication
COE & ARF were introduced to control the Singapore Car population:
& Cost
(-1.96 Million Km/ Month)
Contradictory effect as highlighted by paper:
& Cost
(+0.14 Million Km/Month)
Effectiveness of Policy:
Net -1.82 Million km/ Month of Total Road Usage Highly effective still (~93%)
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Policy Implication-Beyond Cars
Pricing strategy of durable goods:
Common practice: cheap durables, expensive complementary consumables
usage of durable good hence purchase complementary consumables
Implication of sunk cost fallacy: expensive durables, cheaper complementary
consumables
usage of durable good hence purchase complementary consumables
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Policy Implication-Beyond Cars
Escalation of commitment to rationalise the decision-makers earlier investments/ losses
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
Gambling losses

Further gambling to recoup initial losses


War Casualties/ resources spent

Commitment even
with no successful outcome foreseen
Opinions on paper?
Extensively comprehensive with ample Robustness checks done
given the data constraints:
Exclusion of 2nd hand carsIncome Selection bias with poor excluded
Study can be further expanded on in the future if the relevant datasets are
available:
Different brands of cars
Separation of discretionary & non-discretionary usage
Income of owner:
Sunk cost sensitivity may vary across income levels
Income Effect: Sunk cost Discretionary Incomefuel spending Usage
Introduction to:
Economy Theory
& Singapores Car Policy
Main Findings
& Model Specification
Empirical Design,
Econometrics Interpretation
& Alternative explanation
Robustness testing
Policy Implications
& Opinion on paper
The End
Thank you for your attention.
Questions?

You might also like