John Beshears James J. Choi Christopher Harris

advertisement
John Beshears
James J. Choi
Christopher Harris
David Laibson
Brigitte C. Madrian
Jung Sakong
October 18, 2014
} 
If people have self-control problems
◦  People are tempted to spend money that they had
previously wanted to save
◦  E.g. quasi-hyperbolic intertemporal discounting
} 
If people have self-control problems
◦  People are tempted to spend money that they had
previously wanted to save
◦  E.g. quasi-hyperbolic intertemporal discounting
} 
And if people wish to curb their actions
◦  People anticipate their self-control problems
◦  Sophisticated agents
} 
If people have self-control problems
◦  People are tempted to spend money that they had
previously wanted to save
◦  E.g. quasi-hyperbolic intertemporal discounting
} 
And if people wish to curb their actions
◦  People anticipate their self-control problems
◦  Sophisticated agents
} 
Then they may take up commitment devices
that restrict their future choice set
} 
Subjects recruited for an online experiment
} 
Receive an endowment of money
} 
Divide money between two types of accounts
◦  Liquid account
◦  Illiquid account
} 
Make withdrawals over the next 6-12 months
Billions of Dollars (2006)
$80
Leakage from 401(k) Plans (2006)
$74
$70
Loans
$60
$50
Hardship
withdrawals
$40
$30
$20
$10
$0
Source: GAO-09-715, 2009
$1
$9
Cashouts at
job change
} 
Illiquidity creates costs – individuals cannot
access funds when spending needs arise
} 
} 
Illiquidity creates costs – individuals cannot
access funds when spending needs arise
But individuals may actually value greater
illiquidity if it combats self-control problems
Assigned to
one of seven
study arms
(three today)
Subjects
from ALP
participate
over web
Read
rules
Randomizing
device selects $50,
$100, or $500
Make allocations
for $50, $100,
$500 cases;
if appropriate
choose goal dates
After one year
study ends; all
money is
disbursed
Subjects can
start making
withdrawals 7
days after initial
allocation
Freedom Account
Goal Account
v  Subject picks a goal date
v  Liquid - can withdraw money
v  Illiquid before goal date
any time within the period of
v  Liquid after goal date, just like
experiment (1 year)
freedom account
v  22% interest per year
v  X % interest per year
Freedom Account
Goal Account
v  Subject picks a goal date
v  Liquid - can withdraw money
v  Illiquid before goal date
any time within the period of
v  Liquid after goal date, just like
experiment (1 year)
freedom account
v  22% interest per year
v  22% interest per year
Freedom Account
Goal Account
v  Subject picks a goal date
v  Liquid - can withdraw money
v  Illiquid before goal date
any time within the period of
v  Liquid after goal date, just like
experiment (1 year)
freedom account
v  22% interest per year
v  22% interest per year
What does illiquid mean? Three cases in our study:
•  10% withdrawal penalty: you get X but your account is debited (1.1)X.
•  20% withdrawal penalty: you get X but your account is debited (1.2)X.
•  No withdrawals
Average
Horizon
10% penalty
Fraction
0.39 (0.03)
81.8 (9.1)
20% penalty
0.45 (0.03)
100.5 (10.9)
No withdrawal
0.56 (0.04)
131.8 (13.9)
Subject allocates $100 between…
Freedom Account
Goal Account(s)
•  Subject picks a goal and
a goal date
•  Liquid – can withdraw
•  Illiquid before goal date;
money any time within the liquid after goal date, just
period of experiment
like Freedom Account
•  22% interest per year
•  22% interest per year
Goal Account characteristic
Arm 1
10% Penalty before goal date
Arm 2
No Withdrawal before goal date
Arm 3
•  10% Penalty
•  No Withdrawal
Average
Horizon
10% penalty
Fraction
0.46 (0.03)
64.4 (5.5)
No withdrawal
0.54 (0.02)
74.8 (4.4)
Both
0.50 (0.03)
71.3 (4.8)
10% penalty
0.16 (0.01)
No withdrawal
0.34 (0.02)
} 
Three period model with present-­‐biased agent ◦  Period 0: make alloca9ons ◦  Period 1: first consump9on opportunity ◦  Period 2: final consump9on opportunity } 
Three period model with present-­‐biased agent ◦  Period 0: make alloca9ons ◦  Period 1: first consump9on opportunity ◦  Period 2: final consump9on opportunity } 
U9lity func9on (note u is CRRA) βθu(c1) + βu(c2)
θu(c1) + βu(c2)
u(c2)
} 
Three period model with present-­‐biased agent ◦  Period 0: make alloca9ons ◦  Period 1: first consump9on opportunity ◦  Period 2: final consump9on opportunity } 
U9lity func9on (note u is CRRA) βθu(c1) + βu(c2)
θu(c1) + βu(c2)
u(c2)
}  Budget constraint: c1
+ c2 ≤ y (note c1 ≥ 0, c2 ≥ 0) } 
In period 0, the agent does not observe θ but can force the choice in period 1 to be from a subset of the budget set, with the constraint that the slope cannot be more extreme than –(1 + π)
} 
In period 0, the agent does not observe θ but can force the choice in period 1 to be from a subset of the budget set, with the constraint that the slope cannot be more extreme than –(1 + π)
} 
Assume (1 – β)θf(θ) + F(θ) increases then decreases }  The subset of the budget set chosen in period 0 is equivalent to dividing the endowment y between two accounts, one that imposes a penalty π on withdrawals in period 1, and another that imposes no penalty on withdrawals in period 1 }  The subset of the budget set chosen in period 0 is equivalent to dividing the endowment y between two accounts, one that imposes a penalty π on withdrawals in period 1, and another that imposes no penalty on withdrawals in period 1 } 
If the coefficient of rela9ve risk aversion is one, the frac9on of the endowment allocated to the penalty account increases in π
} 
Evidence that individuals recognize future
self-control problems and adopt commitment
devices
} 
} 
Evidence that individuals recognize future
self-control problems and adopt commitment
devices
Implications for firms and governments
} 
} 
Evidence that individuals recognize future
self-control problems and adopt commitment
devices
Implications for firms and governments
◦  Firms may not provide illiquid accounts because the
government already has (401(k)s in the U.S.)
} 
} 
Evidence that individuals recognize future
self-control problems and adopt commitment
devices
Implications for firms and governments
◦  Firms may not provide illiquid accounts because the
government already has (401(k)s in the U.S.)
◦  The availability of borrowing may undo illiquidity
Download