|
Field
Experimental Design (Public)
|
Before
Individuals will be randomly allocated to one of four treatment arms with 100 participants each:
(1) CBS no prompting / MLL no prompting / CBS prompting
(2) MLL no prompting / CBS no prompting / MLL prompting
(3) CBS prompting / MLL prompting
(4) MLL prompting / CBS prompting
We will test for differences between groups 1+2 and 3+4 for the first two questions. We will also test for differences between the first and third question for groups 1 and 2. We expect differences when these are measured with CBS, and not if they are measured with MLL.
|
After
Individuals will be randomly allocated to one of four treatment arms with 100 participants each:
(1) CBS no prompting / MLL no prompting / CBS prompting
(2) MLL no prompting / CBS no prompting / MLL prompting
(3) CBS prompting / MLL prompting
(4) MLL prompting / CBS prompting
We will test for differences between groups 1+2 and 3+4 for the first two questions. We will also test for differences between the first and third question for groups 1 and 2. We expect differences when these are measured with CBS, and not if they are measured with MLL.
The experimental design relies on our intervention to successfully alter expenditure expectations. See the full pre-analysis plan for details.
|
|
Field
Randomization Unit
|
Before
Sample size: 400
We will recruit participants using the on-line research platform Prolific.
Eligibility criteria:
● currently residing in New York State
● with a household income below the median household income (2017-21 in 2021 US dollars): $75,157 (source US Census bureau)
|
After
Sample size: 400
We will recruit participants using the on-line research platform Prolific.
Eligibility criteria:
● currently residing in New York State
● with a household income below 99,999 US dollars
|
|
Field
Intervention (Hidden)
|
Before
Individuals will choose between early and late outcomes. Some individuals will first be prompted to think about all their upcoming expenditures over the next two to three weeks category by category, including exceptional expenditures such as those for repairs or health care. Previous research has shown that this induces individuals to realize that the expenditures for their intended purchases is larger than anticipated, i.e., larger than reported without considering carefully all the expenditure categories that they are prompted about. Therefore, under prompting, individuals realize that their short-term costs are higher than expected, and if money is not perfectly transferable, this necessitates a cut in current background consumption as a consequence.
This change in perceived background consumption should induce individuals to desire more money early on. We therefore use prompting to induce a change in perceived background consumption, which should affect standard discount factor measurement.
This is investigated with the convex budget set method of Andreoni and Sprenger (2012) - CBS hereafter. We adopt both an across-subject and a within-subject design. In the across-subject treatment, one group chooses with prompting and one without, and we expect the first group to appear significantly more present-biased (even though assignment to groups is actually random). In the within-subject design, individuals who first answered without prompting are then prompted, and are again asked about time preferences. We expect them to appear significantly more present-biased after prompting.
On the other hand, we expect no such effect when using the method of Multiple Lottery Lists proposed by Belot et al (2021) – MLL hereafter. That method was designed to be unaffected by background consumption, and so changes in perceived expenditures and associated changes in background consumption should not affect discount factor measurement under MLL.
This would indicate that standard methods indeed suffer when individuals worry that they have less money in some periods than in others, while MLL does not display this feature.
We will also check that both methods correlate in terms of present-bias and long-run discount factor when individuals are not prompted. This indicates that in times without changes in background consumption both methods uncover similar elements.
|
After
Individuals will choose between early and late outcomes. Some individuals will first be prompted to think about all their upcoming expenditures over the next two to three weeks category by category, including exceptional expenditures such as those for repairs or health care. Previous research has shown that this induces individuals to realize that the expenditures for their intended purchases is larger than anticipated, i.e., larger than reported without considering carefully all the expenditure categories that they are prompted about. Therefore, under prompting, individuals realize that their short-term costs are higher than expected, and if money is not perfectly transferable, this necessitates a cut in current background consumption as a consequence.
This change in perceived background consumption should induce individuals to desire more money early on. We therefore use prompting to induce a change in perceived background consumption, which should affect standard discount factor measurement.
This is investigated with the convex budget set method of Andreoni and Sprenger (2012) - CBS hereafter. We adopt both an across-subject and a within-subject design. In the across-subject treatment, one group chooses with prompting and one without, and we expect the first group to appear significantly more present-biased (even though assignment to groups is actually random). In the within-subject design, individuals who first answered without prompting are then prompted, and are again asked about time preferences. We expect them to appear significantly more present-biased after prompting.
On the other hand, we expect no such effect when using the method of Multiple Lottery Lists proposed by Belot et al (2021) – MLL hereafter. That method was designed to be unaffected by background consumption, and so changes in perceived expenditures and associated changes in background consumption should not affect discount factor measurement under MLL.
This would indicate that standard methods indeed suffer when individuals worry that they have less money in some periods than in others, while MLL does not display this feature.
We will also check that both methods correlate in terms of present-bias and long-run discount factor when individuals are not prompted. This indicates that in times without changes in background consumption both methods uncover similar elements.
A prerequisite for testing the impact on discount factor measurement is that the treatment indeed shifts expected expenditures as envisioned. So analysis will only proceed if that is the case. See pre-analysis plan for details.
|