Dynamic Inconsistency in Real Effort Tasks

Working Over Time: Dynamic Inconsistency in Real Eﬀort Tasks ∗

Ned Augenblick † Muriel Niederle‡ UC Berkeley, Haas School of Business Stanford University and NBER Charles Sprenger§ Stanford University

First Draft: July 15, 2012 This Version: September 25, 2014

Abstract Experimental tests of dynamically inconsistent time preferences have largely relied on choices over time-dated monetary rewards. Several recent studies have failed to find the standard patterns of present bias. However, such monetary studies contain often- discussed confounds. In this paper, we sidestep these confounds and investigate choices over consumption (real effort) in a longitudinal experiment. We pair this effort study with a companion monetary discounting study. We confirm very limited time inconsistency in monetary choices. However, subjects show considerably more present bias in effort. Furthermore, present bias in the allocation of work has predictive power for demand of a meaningfully binding commitment device. Therefore our findings validate a key implication of models of dynamic inconsistency, with corresponding policy implications.

JEL classiﬁcation: C91, D12, D81 Keywords: Time Discounting, Demand for Commitment, Real Eﬀort, Convex Time Budget

∗We are grateful for many helpful discussions including those of Steﬀen Andersen, James Andreoni, Colin Camerer, Yoram Halevy, David Laibson, Matthew Rabin, and Georg Weizs¨acker. We thank Wei Wu for helpful research assistance and technological expertise. †University of California Berkeley, Haas School of Business, University of California, Berkeley, 545 Student Services Building, 1900, Berkeley, CA, 94720-1900. [email protected] ‡Stanford University, Department of Economics, Landau Economics Building, 579 Serra Mall, Stanford, CA 94305; [email protected], www.stanford.edu/∼niederle §Stanford University, Department of Economics, Landau Economics Building, 579 Serra Mall, Stanford, CA 94305; [email protected]. 1 Introduction

Models of dynamically inconsistent time preferences (Strotz, 1956; Laibson, 1997; O’Donoghue and Rabin, 1999, 2001) are a pillar of modern behavioral economics, having added generally to economists’ understanding of the tensions involved in consumption-savings choices, task performance, temptation, and self-control beyond the standard model of exponential discounting (Samuelson, 1937). Given the position of present-biased preferences in the behavioral literature, there is clear importance in testing the model’s central falsiﬁable hypothesis of diminishing impatience through time. Further, testing auxiliary predictions such as sophisticated individuals’ potential to restrict future activities through commitment devices can distinguish between competing accounts for behavior and deliver critical prescriptions to policy makers.1 In this paper we present a test of dynamic inconsistency in consumption and investigate the demand for a meaningfully binding commitment device. To date, a notably large body of laboratory research has focused on identifying the shape of time preferences (for a comprehensive review to the early 2000s, see Frederick, Loewenstein and O’Donoghue, 2002). The core of this experimental literature has identiﬁed preferences from time-dated monetary payments.2 Several confounds exist for identifying the shape of time preferences from such monetary choices. Issues of payment reliability and risk preference suggest that subject responses may be closely linked to their assessment of the experimenter’s reliability rather than solely their time preferences.3 Furthermore, monetary payments may not

1Sophistication is taken to mean the decision-maker’s recognition (perhaps partial recognition) of his predilec- tion to exhibit diminishing impatience through time. Appendix sectionA outlines the model which follows the framework of O’Donoghue and Rabin(2001). 2Recent efforts using time dated monetary payments to identify time preferences include Ashraf, Karlan and Yin(2006), Andersen, Harrison, Lau and Rutstrom(2008), Dohmen, Falk, Huffman and Sunde(2010), Tanaka, Camerer and Nguyen(2010), Benjamin, Choi and Strickland(2010) Voors, Nillesen, Verwimp, Bulte, Lensink and van Soest(2012), Bauer, Chytilova and Morduch(2012), Sutter, Kocher, Glatzle-Ruetzler and Trautmann (2013), and Dupas and Robinson(2013). 3This point was originally raised by Thaler(1981) who, when considering the possibility of using incentivized monetary payments in intertemporal choice experiments noted ‘Real money experiments would be interesting but seem to present enormous tactical problems. (Would subjects believe they would get paid in five years?)’. Recent work validates this suspicion. Andreoni and Sprenger(2012a), Gine, Goldberg, Silverman and Yang(2010), and Andersen, Harrison, Lau and Rutstrom(2012) all document that when closely controlling transactions costs and payment reliability, dynamic inconsistency in choices over monetary payments is virtually eliminated on aggregate. Further, when payment risk is added in an experimentally controlled way, non-expected utility

2 be suitable to identify parameters of models defined over time-dated consumption. Arbitrage arguments imply that choices over monetary payments should only reveal subjects’ borrowing and lending opportunities (Cubitt and Read, 2007).4 Chabris, Laibson and Schuldt(2008) describe the difficulty in mapping experimental choices over money to corresponding model parameters, casting skepticism over monetary experiments in general. In this paper we attempt to move out of the domain of monetary choice and into the domain of consumption. Our design delivers precise point estimates on dynamic inconsistency based upon intertemporal allocations of effort and provides an opportunity to link parameter measures with demand for commitment. Delivering such a connection and contrasting present bias measured over money and over consumption are key contributions of our study. There are few other experimental tests of dynamic inconsistency in consumption. Lead- ing examples document dynamic inconsistency in brief, generally a few minutes, intertemporal choices over irritating noises and squirts of juice and soda (Solnick, Kannenberg, Eckerman and Waller, 1980; McClure, Laibson, Loewenstein and Cohen, 2007; Brown, Chua and Camerer, 2009). On a larger time scale, perhaps closer to everyday decision-making, there are two key contributions. Read and van Leeuwen(1998) identify dynamic inconsistency between choices over snack foods made one week apart. Ariely and Wertenbroch(2002) document demand risk preferences deliver behavior observationally equivalent to present bias as described above (Andreoni and Sprenger, 2012b). 4In a monetary discounting experiment, subjects often make binary choices between a smaller sooner pay- Y ment, $X, and a larger later payment, $Y. The ratio, X , defines a lab-offered gross interest rate. An individual who can borrow at a lower rate than the lab-offered rate should take the larger later payment, finance any sooner consumption externally, and repay their debts with the later larger payment they chose. An individual who can save at a higher rate than the lab-offered rate should take the smaller sooner payment, pay for any sooner consumption and place the remainder in their savings vehicle. These two strategies deliver a budget constraint that dominates the lab-offered budget constraint. Hence, monetary discounting experiments should reveal only external borrowing and lending opportunities. And, unless such opportunities change over time, one should reveal no present bias. The logic extends to the convex decisions of Andreoni and Sprenger(2012a). Subjects should allocate only at corner solutions and such solutions should maximize net present value at external interest rates. This point has been thoughtfully taken into account in some studies. For example, Harrison, Lau and Williams(2002) explicitly account for potential arbitrage in their calculations of individual discount rates by measuring individual borrowing and saving rates and incorporating these values in estimation. Cubitt and Read(2007) provide excellent recent discussion of the arbitrage arguments and other issues for choices over monetary payments. One counterpoint is provided by Coller and Williams(1999), who present experimental subjects with a fully articulated arbitrage argument and external interest rate information and document only a small treatment effect.

3 for deadlines for classroom and work assignments, a potential sign of commitment demand for dynamically inconsistent individuals. Though suggestive, neither exercise allows for precise identification of discounting parameters, nor delivers the critical linkage between present bias and commitment demand. With the exception of Ashraf et al.(2006) and Kaur, Kremer and Mullainathan(2010) virtually no research attempts to make such links. Ashraf et al.(2006) employ monetary discounting measures and link them to take-up of a savings commitment product. Kaur et al.(2010) use disproportionate effort response on paydays to make inference on dynamic inconsistency and link this behavior to demand for a dominated daily wage contract. There are several major differences between our research and this prior work, which are discussed in detail in Section 3.4. Most important is the measurement of dynamic inconsistency. As opposed to monetary measures or measuring potential correlates of present bias, our effort allocations yield precise parametric measures linked directly to the theory of present bias. 102 UC Berkeley students participated in a seven week longitudinal experiment. Subjects allocated units of effort (i.e., negative leisure consumption) over two work dates. The tasks over which subjects made choices were transcription of meaningless Greek texts and completion of partial Tetris games. Allocations were made at two points in time: an initial allocation made in advance of the first work date and a subsequent allocation made on the first work date. We then randomly selected either an initial allocation or a subsequent allocation and required subjects to complete the allocated tasks. This incentivized all allocation decisions. Differences between initial and subsequent allocations allow for precise measurement of dynamic inconsistency. A first block of the experiment, three weeks in length, was dedicated to this measurement effort. In a second block of the experiment, also three weeks in length, the design was augmented to elicit demand for a commitment device. The commitment device of the second block allowed subjects to probabilistically favor their initial allocations over their subsequent allocations in the random selection process. Hence, commitment reveals a subject’s preference for implementing the allocations made in advance of the first work date. We investigate demand for our offered commitment device and correlate identified dynamic inconsistency with commitment demand.

4 The repeated interaction of our seven-week study allows us to complement measures of effort discounting with measures of monetary discounting taken from Andreoni and Sprenger (2012a) Convex Time Budget (CTB) choices over cash payments received in the laboratory. In these choices, subjects allocated money over two dates. Variation in whether the first payment date is the present delivers identification of monetary present bias. Hence, we can compare dynamic inconsistency measured over work and money at both the aggregate and individual level within subjects. A second study, essentially a between-subjects replication exercise, was also conducted to provide corroboration of the within-subject conclusions. We document three primary findings. First, in the domain of money we find virtually no evidence of present bias. Monetary discount rates involving present dates are effectively indistinguishable from those involving only future dates. Further, subjects appear to treat money received at different times as perfect substitutes, suggesting they treat money as fungible. Sec- ond, in the domain of effort we find significant evidence of present bias. Subjects allocate roughly nine percent more work to the first work date when the allocation of tasks is made in advance compared to when it is made on the first work date itself. Corresponding parameter estimates corroborate these non-parametric results. Discount rates measured in advance of the first work date are around zero percent per week while discount rates measured on the first work date are around eleven percent per week. We reproduce these two primary study results in our between-subjects replication exercise with an additional 200 UC Berkeley students. Our third finding is that 59 percent of subjects demand commitment at price $0, preferring a higher likelihood of implementing their initial pre-work date allocations. We show that the choice of commitment is binding and meaningful in the sense that initial preferred allocations differ significantly from subsequent allocations for committing subjects. Importantly, we show that present bias measured in the first block of the experiment is predictive of this (later) commitment choice. A corresponding investigation on the extent of sophistication and commitment demand indicates that subjects potentially forecast their present bias. This link delivers key validation and support for our experimental measures and well-known theoretical models of

5 present bias. We draw two conclusions from our results. First, our results show evidence of present bias in the domain of consumption with a design that eliminates a variety of potential confounds and provides precise parameter estimation. Second, our subjects are at least partially aware of their dynamic inconsistency as they demand binding commitment. The paper proceeds as follows: Section 2 provides details for our longitudinal experimental design. Section 3 presents results and section 4 concludes.

2 Design

To examine dynamic inconsistency in real effort, we introduce a longitudinal experimental design conducted over seven weeks. Subjects are asked to initially allocate tasks, subsequently allocate tasks again, and complete those tasks over two work dates. Initial allocations made in advance of the first work date are contrasted with subsequent allocations made on the first work date to identify dynamic inconsistency. If all elements of the experiment are completed satisfactorily, subjects receive a completion bonus of $100 in the seventh week of the study. Otherwise they receive only $10 in the seventh week. The objective of the completion bonus is to fix the monetary dimension of subjects’ effort choices and to ensure a sizable penalty for attrition. Subjects are always paid the same amount for their work, the question of interest is when they prefer to complete it. We present the design in five subsections. First, we describe the Jobs to be completed. Second, we present a timeline of the experiment and the decision environment in which allocations were made. The third subsection describes the elicitation of commitment demand. The fourth subsection addresses design details including recruitment, selection, and attrition. The fifth subsection presents the complementary monetary discounting study. In addition to this primary within-subjects study, we also conducted a between-subjects replication exercise. The between-subjects design is discussed primarily in section 3.5 and note is made of any design differences.

6 2.1 Jobs

The experiment focuses on intertemporal allocations of effort for two types of job. In Job 1, subjects transcribe a meaningless Greek text through a computer interface. Panel A of Figure 1 demonstrates the paradigm. Random Greek letters appear, slightly blurry, in subjects’ transcription box. By pointing and clicking on the corresponding keyboard below the transcription box, subjects must reproduce the observed series of Greek letters. One task is the completion of one row of Greek text with 80 percent accuracy.5 In the first week, subjects completed a task from Job 1 in an average of 54 seconds. By the final week, the average was 46 seconds. In Job 2, subjects are asked to complete four rows of a modified Tetris game, see Panel B of Figure1. Blocks of random shapes appear at the top of the Tetris box and fall at a fixed relatively slow speed. Arranging the shapes to complete a horizontal line of the Tetris box is the game’s objective. Once a row is complete, it disappears and the shapes above fall into place. One task is the completion of four rows of Tetris. If the Tetris box fills to the top with shapes before the four rows are complete, the subject begins again with credit for the rows already completed. In the first week, subjects completed a task from Job 2 in an average of 55 seconds. By the final week, the average was 46 seconds. In contrast to a standard Tetris game, one cannot accelerate the speed of the falling shapes, and one does not pass through ‘levels’ of progressive difficulty. Hence, our implementation of Tetris should not be thought of as being as enjoyable as the real thing.

2.2 Experimental Timeline

The seven weeks of the experiment are divided into two blocks. Weeks 1, 2, and 3 serve as the ﬁrst block. Weeks 4, 5, and 6 serve as the second block. Week 7 occurs in the laboratory and is only used to pay subjects. Subjects always participate on the same day of the week throughout

5 Our measure of accuracy is the Levenshtein Distance. The Levenshtein Distance is commonly used in computer science to measure the distance between two strings and is deﬁned as the minimum number of edits needed to transform one string into the other. Allowable edits are insertion, deletion or change of a single character. As the strings of Greek characters used in the transcription task are 35 characters long our 80 percent accuracy measure is equivalent to 7 edits or less or a Levenshtein Distance ≤ 7.

7 Figure 1: Experimental Jobs

Panel A: Job 1- Greek Transcription

Panel B: Job 2- Partial Tetris Games

the experiment. That is, subjects entering the lab on a Monday allocate tasks to be completed on two future Monday work dates. Therefore, allocations are made over work dates that are always exactly seven days apart. Weeks 1 and 4 occur in the laboratory and subjects are reminded of their study time the night before. Weeks 2, 3, 5, and 6 are completed online. For Weeks 2, 3, 5, and 6, subjects are sent an email reminder at 8pm the night before with a (subject-unique) website address. Subjects are required to log in to this website between 8am and midnight of the day in question

8 and complete their work by 2am the following morning. At each point of contact, subjects are first given instructions about the decisions to be made and work to be completed that day, reminded of the timeline of the experiment, given demonstrations of any unfamiliar actions, and then asked to complete the necessary actions. The second block of the experiment, Weeks 4, 5, and 6, mimics the first block of Weeks 1, 2, and 3, with one exception. In Week 4, subjects are offered a probabilistic commitment device, which is described in detail in subsection 2.4. Hence, we primarily describe Weeks 1, 2 and 3 and note any design changes for Weeks 4, 5 and 6. To summarize our longitudinal effort experiment, Table1 contains the major events in each week which are described in detail below.

Table 1: Summary of Longitudinal Experiment

10 Eﬀort Minimum Allocation-That- Complete Commitment Receive Allocations Work Counts Chosen Work Choice Payment Week 1 (In Lab): x x Week 2 (Online): x x x x Week 3 (Online): x x Week 4 (In Lab): x x x Week 5 (Online): x x x x Week 6 (Online): x x Week 7 (In Lab): x

2.3 Eﬀort Allocations

In Week 1, subjects allocate tasks between Weeks 2 and 3. In Week 2, subjects also allocate tasks between Weeks 2 and 3. Subjects were not reminded of their initial Week 1 allocations in Week 2. Note that in Week 1 subjects are making decisions involving two future work dates, whereas in Week 2, subjects are making decisions involving a present and a future work date. Before making decisions in Week 1, subjects are told of the Week 2 decisions and are aware that exactly one of all Week 1 and Week 2 allocation decisions will be implemented.

9 2.3.1 Allocation Environment

Allocations are made in a convex environment. Using slider bars, subjects allocate tasks to two dates, one earlier and one later, under different interest rates.6 Figure2 provides a sample allocation screen. To motivate the intertemporal tradeoffs faced by subjects, decisions are described as having different ‘task rates.’ Every task allocated to the later date reduces the number of tasks allocated to the sooner date by a stated number. For example, a task rate of 1:0.5 implies that each task allocated to Week 3 reduces by 0.5 the number in Week 2.7 For each task and for each date where allocations were made, subjects faced five task rates. These task rates take the values, R ∈ {0.5, 0.75, 1, 1.25, 1.5}. The subjects’ decision can be formulated as allocating tasks e over times t and t + k, et and et+k, subject to the present-value budget constraint,

et + R · et+k = m. (1)

The number of tasks that subjects could allocate to the sooner date was capped at ﬁfty such that m = 50 in each decision in the experiment.8

2.3.2 Minimum Work

In each week, subjects are required to complete 10 tasks of each Job prior to making allocation decisions or completing allocated tasks. The objective of these required tasks, which we call “minimum work,” is three-fold. First, minimum work requires a few minutes of participation at each date, forcing subjects to incur the transaction costs of logging on to the experimental website at each time.9 Second, minimum work, especially in Week 1, provides experience for

6The slider was initially absent from each slider bar and appeared in the middle of the bar once a subject clicked on the allocation. Every slider bar was thus clicked on before submission, avoiding purely passive response. 7We thank an anonymous referee for noting a small error in our instructions which inverted the task rates when ﬁrst introducing them. Though this appears not to have aﬀected response as allocations move appropriately with task rates, we do correct this error in our replication exercise and document very similar behavior. See section 3.5 for detail. 8 We use R for present value budget constraints of the form et + R · et+k = m, and P for future value budget constraints of the form P · et + et+k = m. 9A similar technique is used in monetary discounting studies where minimum payments are employed to eliminate subjects loading allocations to certain dates to avoid transaction costs of receiving multiple payments

10 Figure 2: Convex Allocation Environment

subjects such that they have a sense of how eﬀortful the tasks are when making their allocation decisions. Third, we require minimum work in all weeks before all decisions, and subjects are informed that they will have to complete minimum work at all dates. This ensures that subjects have experienced and can forecast having experienced the same amount of minimum work when making their allocation decisions at all points in time.

2.3.3 The Allocation-That-Counts

Each subject makes 20 decisions allocating work to Weeks 2 and 3: ﬁve decisions are made for each Job in Week 1 and ﬁve for each Job in Week 2. After the Week 2 decisions, one of these 20 allocations is chosen at random as the ‘allocation-that-counts’ and subjects have to complete the allocated number of tasks on the two work dates to ensure successful completion of the experiment (and hence payment of $100 instead of only $10 in Week 7). The randomization device probabilistically favors the Week 2 allocations over the Week 1 allocations. In particular, subjects are told (from the beginning) that their Week 1 allocations will be chosen with probability 0.1, while their Week 2 allocations will be chosen with probability or cashing multiple checks (Andreoni and Sprenger, 2012a).

11 0.9. Within each week’s allocations, every choice is equally likely to be the allocation-that- counts.10 This randomization process ensures incentive compatibility for all decisions. This design choice was made for two reasons. First, it increases the chance that subjects experienced their own potentially present-biased behavior. Second, it provides symmetry to the decisions in Block 2 that elicit demand for commitment.

2.4 Commitment Demand

In the second block of the experiment, Weeks 4, 5, and 6, subjects are offered a probabilistic commitment device. In Week 4, subjects are given the opportunity to choose which allocations will be probabilistically favored. In particular, they can choose whether the allocation-that- counts comes from Week 4 with probability 0.1 (and Week 5 with probability 0.9), favoring flexibility, or from Week 4 with probability 0.9, favoring commitment. This form of commitment device was chosen because of its potential to be meaningfully binding. Subjects who choose to commit and who differ in their allocation choices through time can find themselves constrained by commitment with high probability. In order to operationalize our elicitation of commitment demand, subjects are asked to make 15 multiple price list decisions between two options. In the first option, the allocation- that-counts will come from Week 4 with probability 0.1. In the second option, the allocation- that-counts will come from Week 4 with probability 0.9. In order to determine the strength of preference, an additional payment of between $0 and $10 is added to one of the options for each decision.11 Figure3 provides the implemented price list. One of the 15 commitment decisions is chosen for implementation, ensuring incentive compatibility. Subjects are told that the implementation of the randomization for the commitment decisions will occur once they submit their Week 5 allocation decisions. Given this randomization procedure, an individual choosing commitment in all 15 decisions will complete a Week 4 allocation with probability

10For the description of the randomization process given to subjects please see instructions in AppendixF. 11We chose not to have the listed prices ever take negative values (as in a cost) to avoid subjects viewing paying for commitment as a loss.

12 0.9. Each row at which a subject chooses ﬂexibility reduces this probability by 5.3 percent.12 Hence a subject choosing to commit at price zero (the eighth row) and lower will complete an initial allocation with probability 0.53. Naturally, if subjects treat each commitment decision in isolation, the incentives are more stark as each decision moves the probability of facing an initial allocation from 0.1 to 0.9.13 This isolation is encouraged as subjects are told to treat each decision as if it was the one going to be implemented (See Appendix F.4 for detail).

Figure 3: Commitment Demand Elicitation

Our commitment demand decisions, and the second block of the experiment, serve three purposes. First, they allow us to assess the demand for commitment and ﬂexibility. Second, a key objective of our study is to explore the theoretical link, under the assumption of sophistication, between present bias and commitment demand. Are subjects who are present-biased more likely to demand commitment? Third, a correlation between time inconsistency and commitment validates the interpretation of present bias over other explanations for time inconsistent choices. For example, a subject who has a surprise exam in Week 2 may be observationally indistinguishable in her Week 2 eﬀort choices from a present-biased subject. However, a subject 12Each row changes the probability of implementing an initial allocation by (1/15 * (0.9 - 0.1)) = 0.053. 13In assessing the value of commitment we make this assumption, ignoring the second stage randomization inherent to the commitment demand elicitation.

13 prone to such surprises should favor ﬂexibility to accommodate her noisy schedule. In contrast, a sophisticated present-biased subject may demand commitment to restrict her future self.

2.5 Design Details

102 UC Berkeley student subjects were initially recruited into the experiment across 4 experimental sessions on February 8th, 9th and 10th, 2012 and were told in advance of the seven week longitudinal design and the $100 completion bonus.14 Subjects did not receive an independent show up fee. 90 subjects completed all aspects of the working over time experiment and received the $100 completion bonus. The 12 subjects who selected out of the experiment do not appear different on either initial allocations, comprehension or a small series of demographic data collected at the end of the first day of the experiment.15 One more subject completed initial allocations in Week 1, but due to computer error did not have their choices recorded. This leaves us with 89 subjects. One critical aspect of behavior limits our ability to make inference for time preferences based on experimental responses. In particular, if subjects have no variation in allocations in response to changes in R in some weeks, then attempting to point identify both discounting and cost function parameters is difficult, yielding imprecise and unstable estimates. In our sample, nine subjects have this issue for one or more weeks of the study.16 For the analysis, we focus on the primary sample of 80 subjects who completed all aspects of the experiment with positive variation in their responses in each week. In Appendix Table A10, we re-conduct the aggregate analysis including these nine subjects and obtain very similar findings.

14Student subjects were recruited from the subject pool of the UC Berkeley Experimental Laboratory, Xlab. Having subjects informed of the seven week design and payment is a potentially important avenue of selection. Our subjects were willing to put forth eﬀort and wait seven weeks to receive $100. Though we have no formal test, this suggests that our subjects may be a relatively patient selection. 153 of those 12 subjects dropped after the ﬁrst week while the remaining 9 dropped after the second week. Including data for these 9 subjects where available does qualitatively alter the analysis or conclusions. 16Appendix Tables A6 and A7 provide estimates for each individual based on their Block 1 data. The 9 individuals without variation in their responses in one or more weeks are noted. Extreme estimates are obtained for individuals without variation in experimental response in one of the weeks of Block 1.

14 2.6 Monetary Discounting

Subjects were present in the laboratory in the ﬁrst, fourth, and seventh week of the experiment. This repeated interaction facilitates a monetary discounting study that complements our main avenue of analysis. In Weeks 1 and 4 of our experimental design, once subjects complete their allocation of tasks, they are invited to respond to additional questions allocating monetary payments to Weeks 1, 4, and 7. In Week 1, we implement three Andreoni and Sprenger(2012a) Convex Time Budget (CTB) choice sets, allocating payments across: 1) Week 1 vs. Week 4; 2) Week 4 vs. Week 7 (Prospective); and 3) Week 1 vs. Week 7. Individuals allocate monetary payments across the two dates t and t + k, ct and ct+k, subject to the intertemporal constraint,

P · ct + ct+k = m. (2)

The experimental budget is fixed at m = $20 and five interest rates are implemented in each choice set, summarized by P ∈ {0.99, 1, 1.11, 1.25, 1.43}. These values were chosen for comparison with prior work (Andreoni and Sprenger, 2012a).17 In Week 4, we ask subjects to allocate in a CTB choice set over Week 4 and Week 7 under the same five values of P . We refer to these choices made in Week 4 as Week 4 vs. Week 7 and those made in Week 1 over these two dates as Week 4 vs. Week 7 (Prospective). Hence, subjects complete a total of four CTB choice sets. The CTBs implemented in Weeks 1 and 4 are paid separately and independently from the rest of the experiment with one choice from Week 1 and one choice from Week 4 chosen to be implemented. Subjects are paid according to their choices. Subjects are not told of the Week 4 choices in Week 1. As in Andreoni and Sprenger(2012a), we have minimum payments of $5 at each payment date to ensure equal transaction costs in each week, such as waiting to get paid. AppendixF provides the full experimental instructions. While the monetary discounting experiment replicates the design of Andreoni and Sprenger (2012a) to a large extent, there are two important differences. First, Andreoni and Sprenger (2012a) implement choices with payment by check. Our design implements payment by cash

17Additionally, P = 0.99 allows us to investigate the potential extent of negative discounting.

15 with potentially lower transaction costs. Second, Andreoni and Sprenger(2012a) implement choices with present payment received only by 5:00 p.m. in a subject’s residence mailbox. If these payments are not construed as the present, one would expect no present bias. Here, we provide payment immediately in the lab. In both Weeks 1 and 4, the monetary allocations are implemented after the more central eﬀort choices. The monetary choices were not announced in advance and subjects could choose not to participate; ﬁve did so in either Weeks 1 or 4. In our analysis of monetary discounting, we focus on the 75 subjects from the primary sample with complete monetary choice data.

3 Results

The results are presented in five subsections. First, we present aggregate results from the monetary discounting study and compare our observed level of limited present bias with other recent findings. Second, we move to effort related discounting and provide both non-parametric and parametric aggregate evidence of present bias. Third, we analyze individual heterogeneity in discounting for both work and money. Fourth we present results related to commitment demand, documenting correlations with previously measured present bias and analyzing the value of commitment. Lastly, a fifth subsection is dedicated to a between-subjects replication exercise of the results concerning differences in discounting when comparing choices over monetary rewards to effort chocies.

3.1 Monetary Discounting

Figure4 presents the data from our monetary discounting experiment. The mean allocation to the sooner payment date at each value of P from P · ct + ct+k = 20 is reported for the 75 subjects from the primary sample for whom we have all monetary data. The left panel shows three data series for payments sets with three-week delay lengths while the right panel shows the data series for the payment sets with a six-week delay length. Standard error bars are

16 clustered at the individual level.

Figure 4: Monetary Discounting Behavior

3 Week Delay 6 Week Delay 20 15 10 5 Dollars Dollars to Date Early Allocated 0 1 1.2 1.4 1 1.2 1.4 P

(from Pct+ct+k=20) Week 1 vs. Week 4 Week 4 vs. Week 7 (Prospective) Week 4 vs. Week 7 Week 1 vs. Week 7 SE

Gr aphsbymonek

We highlight two features of Figure4. First, note that as P from P · ct + ct+k = 20 increases, the average allocation to the sooner payment decreases, following the law of demand. Indeed, at the individual level 98% of choices are monotonically decreasing in P , and only 1 subject exhibits more than 5 non-monotonicities in demand in their monetary choices.18 This suggests that subjects as a whole understand the implied intertemporal tradeoﬀs and the decision environment. Second, Figure4 allows for non-parametric investigation of present bias in two contexts. 19

18Subjects have 16 opportunities to violate monotonicity comparing two adjacent values of P in their 20 total CTB choices. 63 of 75 subjects have no identiﬁed non-monotonocities. Andreoni and Sprenger(2012a) provide a detailed discussion of the extent of potential errors in CTB choices. In particular they note that prevalence of non-monotonicities in demand are somewhat less than the similar behavior of multiple switching in standard Multiple Price List experiments. 19Though the six-week delay data are used in estimation, our non-parametric tests only identify present bias from choices over three-week delays. Without parametric assumptions for utility our data do not lend themselves naturally to the method of identifying present bias where short horizon choices are compared to long horizon

17 First, one can consider the static behavior, often attributed to present bias, of subjects being more patient in the future than in the present by comparing the series Week 1 vs. Week 4 and Week 4 vs. Week 7 (Prospective). In this comparison, controlling for P , subjects allocate on average $0.54 (s.e = 0.31) more to the sooner payment when it is in the present, F (1, 74) = 2.93, (p = 0.09). A second measure of present bias is to compare Week 4 vs. Week 7 (Prospective) made in Week 1 to the Week 4 vs. Week 7 choices made in Week 4. This measure is similar to the recent work of Halevy(2012). Ignoring income effects associated with having potentially received prior payments, this comparison provides a secondary measure of present bias. In this comparison, controlling for P , subjects allocate on average $0.47 (s.e = 0.32) more to the sooner payment when it is in the present, F (1, 74) = 2.08, (p = 0.15).20 Appendix Table A3 provides a corresponding tabulation of behavior, presenting the budget share allocated to the sooner payment date and the proportion of choices that can be classified as present-biased.21 Across all values of P subjects allocate around 38% (s.e. = 1.73) of their experimental budget to the sooner payment date when the sooner date is in the future (t 6= 0) and around 41% (1.34) to the sooner payment date when the sooner date is in the present (t = 0), F (1, 74) = 3.50, (p = 0.07). We find limited non-parametric support for the existence of a present bias over monetary payments. To provide corresponding estimates of present bias we follow the parametric assumptions of Andreoni and Sprenger(2012a) and assume quasi-hyperbolic (Laibson, 1997; O’Donoghue and Rabin, 2001) power utility with Stone-Geary background parameters. Hence, the quasi-hyperbolic discounted utility from experimental payments at two dates, ct, received choices to examine whether discount factors nest exponentially (see, for example Kirby, Petry and Bickel, 1999; Giordano, Bickel, Loewenstein, Jacobs, Marsch and Badger, 2002). 20Additionally, this measure is close in spirit to our effort experiment where initial allocations are compared to subsequent allocations. To get a sense of the size of potential income effects, we can also compare the Week 1 vs. Week 4 choices made in Week 1 to the Week 4 vs. Week 7 choices made in Week 4. Controlling for P , subjects allocate on average $0.07 (s.e = 0.31) more to the sooner payment in Week 1, F (1, 74) = 0.05, (p = 0.82), suggesting negligible income effects. 21 Budget shares for the sooner payment are calculated as (P · ct)/m for each allocation.

18 at time t, and ct+k, received at time t + k, is

α 1t=0 k α U(ct, ct+k) = (ct + ω) + β δ (ct+k + ω) . (3)

The variable 1t=0 is an indicator for whether or not the sooner payment date, t, is the present. The parameter β captures the degree of present bias, while the parameter δ captures long run discounting. β = 1 nests the standard model of exponential discounting. The utility function is assumed to be concave, α < 1, such that ﬁrst order conditions provide meaningful optima. Here, ω is a Stone-Geary background parameter that we take to be the $5 minimum payment of the monetary experiment.22 Maximizing (3) subject to the intertemporal budget constraint (2) yields an intertemporal Euler equation, which can be rearranged to obtain

ct + ω log(β) log(δ) 1 log( ) = · (1t=0) + · k + ( ) · log(P ). (4) ct+k + ω α − 1 α − 1 α − 1

Assuming an additive error, this functional form can be estimated at the aggregate or individual level.23 One important issue to consider in estimation is the potential presence of corner solutions. We provide estimates from two-limit Tobit regressions designed to account for the possibility that the tangency condition implied by (4) does not hold with equality (for discussion, see Wooldridge, 2002; Andreoni and Sprenger, 2012a). Discounting and utility function

22Andreoni and Sprenger(2012a) provide detailed discussion of the use of such background parameters and provide robustness tests with differing values of ω and differing assumptions for the functional form of utility in CTB estimates. The findings suggest that though utility function curvature estimates may be sensitive to different background parameter assumptions, discounting parameters, particularly present bias, are virtually unaffected by such choices. 23An additive error yields the regression equation

ct + ω log(β) log(δ) 1 log( ) = · (1t=0) + · k + ( ) · log(P ) + . ct+k + ω α − 1 α − 1 α − 1 The stochastic error term, , is necessary to rationalize any discrepancies between our theoretical development and our experimental data. One simple foundation for such an error structure would be to assume that individuals exhibit random perturbations to their log allocation ratios, log( ct+ω ). A more complete formulation ct+k+ω might follow macroeconomic exercises such as Shapiro(1984), Zeldes(1989), and Lawrance(1991). With a time series of consumption, one assumes rational expectations such that Euler equations are satisﬁed up to a mean zero random error, uncorrelated with any information available to the decisionmaker. Assuming constant relative risk aversion, as we do, this forecast error provides the structure for estimating utility function curvature and recovering discounting parameters in a way very similar to our exercise.

19 parameters can be recovered via non-linear combinations of regression coefficients with standard errors estimated via the delta method. AppendixA provides a detailed discussion of identification and estimation of discounting parameters for both monetary and effort choices.

Table 2: Parameter Estimates

Monetary Discounting Eﬀort Discounting (1) (2) (3) (4) (5) All Delay Three Week Delay Job 1 Job 2 Combined Lengths Lengths Greek Tetris Present Bias Parameter: β 0.974 0.988 0.900 0.877 0.888 (0.009) (0.009) (0.037) (0.036) (0.033) Weekly Discount Factor: (δ)7 0.988 0.980 0.993 1.007 0.999 (0.003) (0.003) (0.027) (0.029) (0.025) Monetary Curvature Parameter: α 0.975 0.976 (0.006) (0.005) Cost of Eﬀort Parameter: γ 1.624 1.557 1.589 (0.114) (0.099) (0.104)

# Observations 1500 1125 800 800 1600 # Clusters 75 75 80 80 80 Job Eﬀects Yes

2 2 2 2 2 H0 : β = 1 χ (1) = 8.77 χ (1) = 1.96 χ (1) = 7.36 χ (1) = 11.43 χ (1) = 11.42 (p < 0.01) (p = 0.16) (p < 0.01) (p < 0.01) (p < 0.01) 2 H0 : β(Col. 1) = β(Col. 5) χ (1) = 6.37 (p = 0.01) 2 H0 : β(Col. 2) = β(Col. 5) χ (1) = 8.27 (p < 0.01)

Notes: Parameters identified from two-limit Tobit regressions of equations (4) and (6) for monetary discounting and effort discounting, respectively. Parameters recovered via non-linear combinations of regression coefficients. Standard errors clustered at individual level reported in parentheses, recovered via the delta method. Effort regressions control for Job Effects (Task 1 vs. Task 2). Chi-squared tests inlast three rows.

In Table2, columns (1) and (2) we implement two-limit Tobit regressions with standard errors clustered at the individual level. In column (1) we use all 4 CTB choice sets. In column (2) we use only the choice sets which have three-week delays for continuity with our non-parametric evidence. Across speciﬁcations we identify weekly discount factors of around 0.99. The 95%

24The notation of AppendixA is slightly altered to discuss allocation timing and make links to partial sophistication and the value of commitment for eﬀort choices.

20 confidence interval in column (1) for the weekly discount factor implies annual discount rates between 40% and 140%.25 In column (1) of Table2 we estimate β = 0.974 (s.e. = 0.009), economically close to, though statistically different from dynamic consistency, H0 : β = 1: χ2(1) = 8.77, (p < 0.01). In column (2), focusing only on three week delays, we find β =

0.988 (0.009) and are unable to reject the null hypothesis of dynamic consistency, H0 : β = 1: χ2(1) = 1.96, (p = 0.16). These estimates demonstrate limited present bias for money and hence confirm the non-parametric results. In both specifications, we estimate α of around 0.975 indicating limited utility function curvature over monetary payments. Finding limited curvature over money is important in its own right, as linear preferences over monetary payments are indicative of fungibility. There is no desire to smooth monetary payments as there might be for consumption, with subjects treating money received at different points in time effectively as perfect substitutes. Supporting these estimates, note that 86% of monetary allocations are corner solutions and 61% of subjects have zero interior allocations in twenty decisions.26 Our non-parametric and parametric results closely mirror the aggregate findings of Andreoni and Sprenger(2012a) and Gine et al.(2010). 27 A potential concern of these earlier studies that carefully control transaction costs and payment reliability, is that a payment in the present was implemented by a payment in the afternoon of the same day, e.g. by 5:00 pm in the subjects’ residence mailboxes in Andreoni and Sprenger(2012a). In this paper, because subjects repeatedly have to come to the lab, a payment in the present is implemented by an immediate cash payment. The fact that we replicate the earlier studies that carefully control for transaction

25In AppendixA, we discuss identification of all parameters and note that discount factors are identified from variation in delay length, k. Our ability to precisely identify aggregate discounting was not a focus of the experimental design and is compromised by limited variation in delay length. In monetary discounting experiments it is not unusual to find implied annual discount rates in excess of 100%. 26A consequence of limited utility function curvature is that even a small degree of present bias can lead potentially to sizable changes in allocation behavior through time as individuals may switch from one corner solution to another. Hallmarks of this are seen in Appendix Table A3, which tabulates behavior across interest rates. Though a wide majority of observations are dynamically consistent, some significant changes in budget shares are seen at specific interest rates. 27In both of these prior exercises substantial heterogeneity in behavior is uncovered. In subsection 3.3 we conduct individual analyses, revealing similar findings.

21 costs and payment reliability alleviates the concerns that payments in the afternoon are not treated as present payments. To summarize, we confirm the finding of limited present bias in the domain of money. This could be either because the good in question, money, is fungible, a hypothesis for which we find some evidence (recall that we estimate α to be around 0.975). Alternatively, it could be because present bias in the form provided by models of dynamic inconsistency does not exist or exists in only very limited form. This motivates our exploration of choices over effort, which we believe is closer to consumption than money is.

3.2 Eﬀort Discounting

Subjects make a total of 40 allocation decisions over effort in our seven week experiment. Twenty of these decisions are made in the first block of the experiment, and twenty in the second block. One focus of our design is testing whether participants identified as being present-biased in Block 1 demand commitment in Block 2. Hence, we opt to present here allocation data from only the first block of the experiment. This allows the prediction of commitment demand to be conducted truly as an out-of-sample exercise. In Appendix E.5 we present results of present bias from both blocks of the experiment and document very similar findings.

In Figure5, we show for each value of R from et + R · et+k = 50, the amount of tasks allocated to the sooner work date, Week 2, which could range from 0 to 50.28 We contrast initial allocations of effort made in Week 1 with subsequent allocations made in Week 2 for the 80 subjects of the primary sample. Standard error bars are clustered at the individual level. As with monetary discounting, subjects appear to have understood the central intertemporal tradeoffs of the experiment as both initial and subsequent allocations decrease as R is increased. At the individual level, 95% of choices are monotonically decreasing in R, and only 5 subjects exhibit more than 5 non-monotonicities in their effort choices.29 This suggests that subjects as

28 The data are presented as a function of R from et + R · et+k = 50, as opposed to relative price, to provide a standard downward sloping demand curve. Recall that R ∈ {0.5, 0.75, 1, 1.25, 1.5}. When R is low, sooner tasks are relatively cheap to complete, and when R is high, sooner tasks are relatively expensive to complete. 29Subjects have 32 opportunities to violate monotonicity comparing two adjacent values of R in their 40 total

22 Figure 5: Real Eﬀort Discounting Behavior

Greek Transcription Tetris 40 30 20 Tasks to Date Early Allocated 10 .5 1 1.5 .5 1 1.5 R

(from et+Ret+k=50) Initial Allocation Subsequent Allocation Mean Mean SE

Gr aphsbytk

a whole understand the implied intertemporal tradeoﬀs and the decision environment. Apparent from the observed choices is that at all values of R average subsequent allocations lie below average initial allocations. Controlling for all R and task interactions, subjects allocate 2.47 fewer tasks to the sooner work date when the sooner work date is the present F (1, 79) = 14.78, (p < 0.01). Subjects initially allocate 9.3% more tasks to the sooner work date than they subsequently allocate (26.59 initial vs. 24.12 subsequent).30 Appendix Table A3 provides

CTB choices. 41 of 80 subjects are fully consistent with monotonicity and only 5 subjects have more than 5 non-monotonicities. Deviations are in general small with a median required allocation change of 3 tasks to bring the data in line with monotonicity. Three subjects have more than 10 non-monotonicities indicating upward sloping sooner effort curves. Such subjects may find the tasks enjoyable such that they prefer to do more tasks sooner to fewer tasks later. We believe the increased volume of non-downward sloping behavior in effort relative to money has several sources. Subjects may actually enjoy the tasks, they make more choices for effort than for money, and half of their allocations are completed outside of the controlled lab environment. Importantly, non-monotonicities decrease with experience such that in the second block of the experiment 97 percent of choices satisfy monotonicity while in the first block, only 93 percent do so, F (1, 79) = 8.34 (p < 0.01). 30The behavior is more pronounced for the first block of the experiment. For both blocks combined subjects allocate 25.95 tasks to the sooner date, 1.59 more tasks than they subsequently allocate (24.38 tasks), representing a difference of around 6%, F (1, 79) = 15.16, (p < 0.01). See Appendix E.5 for detail.

23 a corresponding tabulation of behavior, presenting the budget share allocated to the sooner work date and the proportion of choices that can be classiﬁed as present-biased.31 Across all values of R, subjects initially allocate around 53% (s.e. = 0.97) of their experimental budget to the sooner work date and subsequently allocate around 48% (1.02) to the sooner work date, when that sooner work date is in the present, F (1, 79) = 14.87, (p < 0.01). Motivated by our non-parametric analysis we proceed to estimate intertemporal parameters.

Subjects allocate eﬀort to an earlier date, et, and a later date, et+k. We again assume quasi- hyperbolic discounting and a stationary power cost function with Stone-Geary background parameters to write the discounted costs of eﬀort as

γ 1t=0 k γ (et + ω) + β δ (et+k + ω) . (5)

Here γ > 1 represents the stationary parameter on the convex instantaneous cost of eﬀort function. The Stone-Geary term, ω, could be interpreted as some background level of required work. For simplicity, we interpret ω as the required minimum work of the experiment and set

ω = 10 for our eﬀort analysis. The variable 1t=0 is an indicator for whether or not the sooner work date, t, is the present. As before, the parameter β captures the degree of present bias and the parameter δ captures long run discounting.

Maximizing (5) subject to (1) (et,t + R · et+k,t = 50) yields an intertemporal Euler equation, which can be rearranged to obtain

et + ω log(β) log(δ) 1 log( ) = · (1t=0) + · k − ( ) · log(R). (6) et+k + ω γ − 1 γ − 1 γ − 1

As before, we assume an additive error structure and estimate the linear regression implied by (6) using two-limit Tobit regression. The parameters of interest are again recovered from non-linear combinations of regression coeﬃcients with standard errors calculated via the delta method. AppendixA provides detailed discussion of identiﬁcation for such choices. 32

31 Budget shares for the sooner work date are calculated as et/m for each allocation. 32The notation of AppendixA is slightly altered to discuss allocation timing and make links to partial

24 Table2 columns (3) through (5) present two-limit Tobit regressions with standard errors clustered on the individual level. In column (3) the analyzed data are the allocations for Job 1, Greek Transcription. We find an estimated cost parameter γ = 1.624 (0.114). Abstracting from discounting, a subject with this parameter would be indifferent between completing all 50 tasks on one work date and completing 32 tasks on both work dates.33 This suggests non-fungibility in the allocation of tasks as individuals do desire to smooth intertemporally. A further indication of non-fungibility is that in contrast to the monetary choices, only 31% of allocations are at budget corners and only 1 subject has zero interior allocations. The weekly discount factor of δ = 0.993 is similar to our findings for monetary discounting. In column (3) of Table2 we estimate an aggregate β = 0.900 (0.037), and reject the null hypothesis of dynamic consistency, χ2(1) = 7.36, (p < 0.01). In column (4), we obtain broadly similar conclusions for Job 2, the modified Tetris games. We aggregate over the two jobs in column (5), controlling for the job, and again document that subjects are significantly present- biased over effort.34 The results of column (5) indicate that discount rates measured in advance of the Week 2 work date are around zero percent per week while discount rates measured on the Week 2 work date are around eleven percent per week. We therefore confirm our non-parametric findings on effort choices. Finally, our implemented analysis allows us to compare present bias across effort and money with χ2 tests based on seemingly unrelated estimation techniques. We reject the null hypothesis that the β identified in column (5) over effort is equal to that identified for monetary discounting in column (1), χ2(1) = 6.37, (p = 0.01), or column (2), χ2(1) = 8.27, (p < 0.01). Subjects are significantly more present-biased over effort than over money.35 sophistication and the value of commitment for effort choices. 33In many applications in economics and experiments, quadratic cost functions are assumed for tractability and our analysis suggests that at least in our domain this assumption would not be too inaccurate. 34For robustness, we run regressions similar to column (5) separately for each week and note that though the cost function does change somewhat from week to week, present bias is still significantly identified as individuals are significantly less patient in their subsequent allocation decisions compared to their initial allocation decisions. Appendix Table A11 provides estimates. 35In Appendix E.5 we conduct identical analysis using both Blocks 1 and 2 and arrive at the same conclusions. See Appendix Table A12 for estimates.

25 3.3 Individual Analysis

On aggregate, we find that subjects are significantly more present-biased over work than over money. In this sub-section we investigate behavior at the individual level to understand the extent to which present bias over effort and money is correlated within individual. In order to investigate individual level discounting parameters we run fixed effect versions of the regressions provided in columns (2) and (5) of Table2. 36 These regressions assume no heterogeneity in cost or utility function curvature and recover individual parameter estimates of βe, present bias for effort, and βm, present bias for money, as non-linear combinations of regression coefficients. The methods for identifying individual discounting parameters are

37 discussed in AppendixA. Appendix Tables A6 and A7 provide individual estimates of βe and

38 βm along with a summary of allocation behavior for both effort and money for each subject. Figure6 presents individual estimates and their correlation. First, note that nearly 60% of subjects have an estimated βm close to 1, indicating dynamic consistency for monetary discounting choices. This is in contrast to only around 25% of subjects with βe close to 1. The mean value for βm is 0.99 (s.d. = 0.06), while the mean value for βe is 0.91 (s.d. = 0.20). The difference between these measures is significant, t = 3.09, (p < 0.01). Second, note that for the majority of subjects when they deviate from dynamic consistency in effort, they deviate in the direction of present bias. Since correlational studies (e.g., Ashraf et al., 2006; Meier and Sprenger, 2010) often use

36We choose to use the measures of present bias based on three week delay choices for the monetary discounting for continuity with our non-parametric tests of present bias. Further, when validating our individual measures, we focus on allocations over three week delay decisions as in the presentation for the aggregate data. Very similar results are obtained if we use the fixed effects versions of Table2, column (1). 37One technical constraint prevents us from estimating individual discounting parameters with two-limit Tobit as in the aggregate analysis. In order for parameters to be estimable at the individual level with two-limit Tobit, some interior allocations are required. As noted above, 86% of monetary allocations are at budget corners and 61% of the sample has zero interior allocations. For effort discounting, 31% of allocations are at budget corners and 1 subject has zero interior allocations. To estimate individual-level discounting, we therefore use ordinary least squares for both money and effort. Nearly identical aggregate discounting estimates are generated when conducting ordinary least squares versions of Table2. Curvature estimates, however, are sensitive to estimation techniques that do and do not recognize that the tangency conditions implied by (6) and (4) may be met with inequality at budget corners. Se Andreoni and Sprenger(2012a) for further discussion. 38Appendix Tables A6 and A7 include data from the 9 subjects excluded from the primary study sample for having no variation in experimental response in one or more weeks of the study. These subjects are noted along with an explanation of which weeks they provided no variation in response.

26 binary measures of present bias, we define the variables ‘Present-Biased’e and ‘Present-Biased’m which take the value 1 if the corresponding estimate of β lies strictly below 0.99 and zero otherwise. We find that 56% of subjects have a ‘Present-Biased’e of 1 while only 33% of subjects have a ‘Present-Biased’m of 1. The difference in proportions of individuals classified as present-biased over work and money is significant, z = 2.31, (p = 0.02).39

Figure 6: Individual Estimates of Present Bias .6 .4 Fraction .2 0 .5 .75 1 1.25 1.5 Work Present Bias .6 .4 Fraction .2 0 .5 .75 1 1.25 1.5 Monetary Present Bias 1.2 1.1 1 .9 .8

Monetary Present Bias Present Monetary .5 .75 1 1.25 1.5 Work Present Bias

Two important questions with respect to our individual measures arise. First, how much do these measures correlate within individual? The answer to this question is important for understanding both the validity of studies relying on monetary measures and the potential consistency of preferences across domains. Signiﬁcant correlations would suggest that there may be some important preference-related behavior uncovered in monetary discounting studies.40

39Further, one can define future bias in a similar way. 17% of subjects are future biased in money while 29% of subjects are future biased over effort. Similar differing proportions between present and future bias have been previously documented (see, e.g., Ashraf et al., 2006; Meier and Sprenger, 2010). Two important counter- examples are Gine et al.(2010) who find almost equal proportions of present and future biased choices and Dohmen, Falk, Huffman and Sunde(2006) who find a greater proportion of future-biased than present-biased subjects. 40Indeed psychology provides some grounds for such views as money generates broadly similar rewards-related neural patterns as more primary incentives (Knutson, Adams, Fong and Hommer, 2001), and in the domain of discounting evidence suggests that discounting over primary rewards, such as juice, produces similar neural images to discounting over monetary rewards (McClure, Laibson, Loewenstein and Cohen, 2004; McClure et

27 Figure6 presents a scatterplot of βm and βe. In our sample of 75 subjects with both complete monetary and effort discounting choices, we find that βe and βm have almost zero correlation, ρ = −0.05, (p = 0.66). Additionally, we find that the binary measures for present bias,

41 ‘Present-Biased’e and ‘Present-Biased’m are also uncorrelated, ρ = 0.11, (p = 0.33). The second question concerning our estimated parameters is whether they can be validated in sample. That is, given that βe and βm are recovered as non-linear combinations of regression coefficients, to what extent do these measures predict present-biased allocations of tasks and money? In order to examine this internal validity question, we generate difference measures for allocations. For effort choices we calculate the budget share of each allocation for Week 2 effort. The difference in budget shares between subsequent allocation and initial allocation is what we term a ‘budget share difference.’42 As budget shares are valued between [0, 1], our difference measure takes values on the interval [−1, 1]. Negative numbers indicate present-biased behavior and values of zero indicate dynamic consistency. Each subject has 10 such effort budget share difference measures in Block 1. The average budget share difference for effort is -0.049 (s.d. = 0.115) indicating that subjects allocate around 5% less of their work budget to the sooner work date when allocating in the present.43 At the individual level, 49 of 80 subjects have an average budget share difference of less than zero, 13 have an average difference of exactly zero, and 18 have an average difference greater than zero, demonstrating a modal pattern of present bias. A similar measure is constructed for monetary discounting choices. Taking only the three week delay data, at each value of P we take the difference between the future allocation (Week 4 vs. Week 7 (Prospective)) budget share and the present allocation (Week 1 vs. Week 4 or Week al., 2007). 41Interestingly, when using both Blocks 1 and 2 of the data, we come to a slightly different conclusion. Though βm and βe remain virtually uncorrelated, with the additional data we uncover a substantial and significant correlation between Present-Biased’e and ‘Present-Biased’m ρ = 0.24, (p = 0.03). Further, ‘Present-Biased’m is also significantly correlated with the continuous measure βe, ρ = −0.27, (p = 0.02). More work is needed to understand the relationship between monetary and effort present bias parameters. 42 Specifically, given an initial Week 1 allocation of e2 of work to be done in Week 2 and a subsequent allocation 0 0 e2−e2 of e2 in Week 2 of work to be done in week 2, the budget share difference is 50 . 43As noted previously, this average value deviates significantly from the dynamically consistent benchmark of 0, F (1, 79) = 14.87, (p < 0.01).

28 4 vs. Week 7) budget share. This measure takes values on the interval [−1, 1], with negative numbers indicating present-biased behavior. Each subject has 10 such monetary budget share difference measures. The average budget share difference for money is -0.029 (s.d. = 0.134).44 At the individual level, 28 of 75 subjects have an average budget share difference of less than zero, 32 have an average difference of exactly zero, and 15 have an average difference greater than zero, demonstrating a modal pattern of dynamic consistency. The non-parametric budget share difference measures are closely correlated with our parametric estimates at the individual level. The correlation between βe and each individual’s average budget share difference for effort is ρ = 0.948, (p < 0.01). Of the 49 individuals with negative average budget share differences for effort, 47 have estimates of βe < 1. Of the 18 individuals with positive average budget share differences for effort, all 18 have estimates of βe > 1.

Of the 13 individuals with zero average budget share diﬀerences for eﬀort, 11 have βe = 1 and

2 have βe = 1.003 . The correlation between βm and each individual’s average Budget Share Difference for money is ρ = 0.997, (p < 0.01). Of the 28 individuals with negative average budget share differences for money, all 28 have estimates of βm < 1. Of the 15 individuals with positive average budget share differences for money, all 15 have estimates of βm > 1. Of the 32

45 individuals with zero average budget share differences for money, all 32 have βm = 1. This apparent internal validity gives us confidence that our parameter estimates for present bias are indeed tightly linked with present-biased data patterns, appropriately capturing the behavior. In the next section we move out-of-sample to investigate commitment demand. The investigation of commitment demand is critical to ruling out potential alternative explanations for time inconsistency in effort allocations. Our preferred explanation is the existence of a present bias in individual decision-making. However, many alternative explanations exist for rational- izing these data patterns. Chief among these alternatives are the existence of unanticipated shocks to the cost of performing tasks (either in general or specific to tasks in Week 2), resolv-

44 As noted previously, this average value diﬀers marginally signiﬁcantly from the dynamically consistent benchmark of 0, F (1, 74) = 3.50, (p = 0.07). 45Appendix Tables A6 and A7 provide all the corresponding estimates and average budget share data.

29 ing uncertainty between allocation times, and subject exhaustion or error. These alternative explanations are considered in detail in AppendixC. Importantly, we show in AppendixC that under none of these alternatives would we expect a clear link between the behavioral pattern of reallocating fewer tasks to the present and commitment demand. This is in contrast to a model of present bias under the assumption of sophistication. Sophisticated present-biased individuals may have demand for commitment. In the next section we document commitment demand on the aggregate level and link commitment to measured present bias.

3.4 Commitment

In Week 4 of our experiment, subjects are offered a probabilistic commitment device. Subjects are asked whether they prefer the allocation-that-counts to come from their Week 4 allocations with probability 0.1 (plus an amount $X) or with probability 0.9 (plus an amount $Y), with either $X=0 or $Y=0. The second of these choices represents commitment and $X - $Y is the price of commitment.46 We begin by analyzing the simple choice between commitment and flexibility at price zero ($X=0 and $Y=0) and in subsection 3.4.1 we explore the value of commitment and choices when X or Y are not zero. In the simple choice where neither commitment nor flexibility were costly, 59% (47/80) of subjects choose to commit. We define the binary variable ‘Commit (=1)’ which takes the value 1 if a subject chooses to commit in this decision. Figure7 presents Block 1 task allocation behavior separated by commitment choice in Block 2. Immediately apparent from Figure7 is that experimental behavior separates along commitment choice. Subjects who choose commitment in Week 4 made substantially present- biased task allocations in Week 2 given their initial Week 1 allocations. Controlling for all task rate and task interactions, subjects who choose commitment allocate 3.58 fewer tasks to the sooner work date when it is the present, F (1, 46) = 12.18, (p < 0.01). Subjects who do not

46To avoid cutting the sample further, here we consider all 80 subjects in the primary sample. 4 of 80 subjects switched multiple times in the commitment device price list elicitation. Identical results are obtained excluding such individuals.

30 Figure 7: Commitment Choice and Allocation Behavior

Panel A: Commit (=0)

Greek Transcription Tetris 40 30 20 Tasks to Date Early Allocated 10 .5 1 1.5 .5 1 1.5 R

(from et+Ret+k=50) Initial Allocation Subsequent Allocation Mean Mean SE

Gr aphsbytk

Panel B: Commit (=1)

Greek Transcription Tetris 40 30 20 Tasks to Date Early Allocated 10

.5 1 1.5 .5 1 1.5 R

(from et+Ret+k=50) Initial Allocation Subsequent Allocation Mean Mean SE

Gr aphsbytk

demand commitment make more similar initial allocations and subsequent allocations of eﬀort. Controlling for all task rate and task interactions, they only allocate 0.89 fewer tasks to the sooner work date when it is the present, F (1, 32) = 4.01, (p = 0.05). Furthermore, subjects who demand commitment in Week 4 altered their allocations by signiﬁcantly more tasks than subjects who did not demand commitment, F (1, 79) = 5.84, (p = 0.02).

31 Table3 generates a similar conclusion with parametric estimates. In columns (3) and (4), we find that subjects who choose commitment in Block 2 are significantly present-biased over effort in Block 1, χ2(1) = 9.00, (p < 0.01). For subjects who do not choose commitment, we cannot reject the null hypothesis of β = 1 at conventional levels, χ2(1) = 2.64, (p = 0.10). Further, we reject the null hypothesis of equal present bias across committers and non-committers, χ2(1) = 4.85, (p = 0.03).47 In columns (1) and (2) of Table3 we repeat this exercise, predicting commitment choice for effort using present bias parameters from monetary decisions. While subjects who demand commitment also seem directionally more present-biased for monetary decisions than subjects who do not demand commitment, the difference is not significant, (p = 0.26). These findings indicate that present bias in effort is significantly related to future commitment choice. Individuals who are present-biased over effort are substantially more likely to choose commitment at price zero. An important caveat for this exercise is that correlation is far from perfect. For example, the raw correlation between βe and commitment choice is ρ = 0.225, (p = 0.04), implying an R-squared value of around 5%. Substantial variance in the choice of commitment remains unexplained. There are several potential reasons for this lack of explanatory power. A natural first possibility is substantial naivete. Though our results suggest at least partial sophistication, on average, many subjects may be naive with respect to their dynamic inconsistency. Further, among partially sophisticated individuals, there may be limited correlation between behavior and beliefs such that individuals with both high and low values of βe may share similar beliefs as to their future behavior. Third, there may be uncertainty 47These results are stronger for the first block of the experiment prior to the offering of the commitment device, though the general patterns holds when we use both blocks of data. Appendix Table A13 provides analysis including the data from both blocks. It is worth noting that the estimates of weekly discount factors, δ also differ across committing and non-committing subjects. This difference is identified from differences in initial allocations. Non-committing subjects have an average initial budget share for sooner tasks of 50.7% (clustered s.e. = 1.6) and an average subsequent budget share of 49.0% (1.7), while committing subjects have an average initial budget share of 54.9% (1.3) and an average subsequent share of 47.7% (1.4). Committing subjects’ behavior is consistent with δ > 1. However, we hesitate to draw any firm conclusions from this observation as our experiment provides no variation in delay lengths to help identify δ. As discussed in AppendixA, δ is identified from the constant one week delay between work dates. Hence, any level differences across subjects are revealed as differences in estimated δ parameters.

32 Table 3: Monetary and Real Eﬀort Discounting by Commitment

Monetary Discounting Eﬀort Discounting Commit (=0) Commit (=1) Commit (=0) Commit (=1) (1) (2) (3) (4) Tobit Tobit Tobit Tobit Present Bias Parameter: β 0.999 0.981 0.965 0.835 (0.010) (0.013) (0.022) (0.055) Weekly Discount Factor: (δ)7 0.978 0.981 0.917 1.065 (0.003) (0.005) (0.032) (0.039) Monetary Curvature Parameter: α 0.981 0.973 (0.009) (0.007) Cost of Eﬀort Parameter: γ 1.553 1.616 (0.165) (0.134)

# Observations 420 705 660 940 # Clusters 28 47 33 47 Job Eﬀects - - Yes Yes

H0 : β = 1 χ2(1) = 0.01 χ2(1) = 2.15 χ2(1) = 2.64 χ2(1) = 9.00 (p = 0.94) (p = 0.14) (p = 0.10) (p < 0.01)

H0 : β(Col. 1) = β(Col. 2) χ2(1) = 1.29 (p = 0.26)

H0 : β(Col. 3) = β(Col. 4) χ2(1) = 4.85 (p = 0.03)

Notes: Parameters identified from two-limit Tobit regressions of equations (4) and (6) for monetary discounting and real effort discounting. Parameters recovered via non-linear combinations of regression coefficients. Stan- dard errors clustered at individual level reported in parentheses, recovered via the delta method. Commit (=1) or Commit (=0) separates individuals into those who did (1) or those who did not (0) choose to commit at a commitment price of zero dollars. Effort regressions control for Job Effects (Job 1 vs. Job 2). Tested null hypotheses are zero present bias, H0 : β = 1, and equality of present bias across commitment and no commitment,

H0 : β(Col. 1) = β(Col. 2) and H0 : β(Col. 3) = β(Col. 4). in the work environment uncontrolled by the researcher. Even sophisticated present-biased individuals may wish to remain flexible. In subsection 3.4.1 and AppendixD we discuss uncertainty and the benefits of flexibility in detail, noting that the value of commitment is likely influenced by the unmodeled benefits of flexibility. Fourth, the allocation decisions may be subject to substantial noise, leading at least partially to a misestimation of preferences and a misclassification of subjects. Each of these forces may be at play to certain degree, reducing our

33 ability to tightly measure present bias and the extent of sophistication. However, our finding of a significant present bias and a correlation between present bias and commitment demand points to at least partial sophistication for some subjects. It is comforting for a theory of sophisticated present bias to find that present bias predicts commitment demand. However, the result is only meaningful if we can show that commitment places a binding constraint on subjects’ behavior. Do individuals who demand commitment actually restrict their own activities, forcing themselves to complete more work than they instantaneously desire?48 Given the nature of our commitment device, commitment will bind whenever initial allocations differ from subsequent allocations. Two such comparisons are considered. First, we consider the first block of the experiment when no commitment contract is available. How many more tasks would subjects have been required to complete in Week 2 had commitment been in place? To answer this question we examine budget share differences for Block 1. Non-committers have a mean budget share difference of −0.018 (clustered s.e. = 0.009) allocating about 2 percentage points less of each budget to Week 2 when deciding in the present. In contrast, committers have a mean budget share difference of −0.072 (0.020), allocating 7 percentage points less to Week 2 when deciding in the present. While both values are significantly different from zero (F (1, 79) = 4.14, (p = 0.05), F (1, 79) = 12.39, (p < 0.01), respectively), the difference between the two is also statistically significant, F (1, 79) = 5.88, (p = 0.02). Hence, had commitment been in place in Week 2 and had subjects made the same choices, committers would have been required to complete significantly more work than they instantaneously desired and would have been more restricted than non-committers. The same analysis can be done for Block 2 focusing on required work in Week 5. Non-committers have a mean budget share difference of 0.011 (0.017) while committers have a mean difference of −0.030 (0.013). The difference for committers remains significantly different from zero,

48Though our oﬀered commitment contract allows individuals only to meaningfully restrict themselves, this need not be the case. One example would be to have individuals commit to completing at least 1 task at the sooner work date. As virtually all initial allocations and subsequent allocations satisfy this condition anyways, such commitment would not be meaningful and as such, should not serve as evidence for the theoretically predicted link between sophisticated present bias and commitment demand.

34 F (1, 79) = 5.57, (p = 0.02), and the difference between the two remains significant at the 10% level F (1, 79) = 3.68, (p = 0.06).49 Hence, in the presence of commitment in Week 5, committed subjects are required to complete significantly more work than they instantaneously desire and are more restricted than non-committed subjects. We are aware of two prior exercises exploring the potential extent of present bias and its correlation with commitment demand. Kaur et al.(2010) link the apparently present-biased behavior of working harder on paydays with demand for a dominated wage contract wherein individuals choose a work target. If the work target is not met, an individual receives a low piece- rate wage, while if it is met or exceeded the individual receives a higher piece rate wage. As the dominated wage contract can be viewed as a commitment to complete a certain amount of work, this represents a potential link between commitment and present bias. Commitment levels are chosen by individuals themselves and are set to around one-sixth of daily production on average. Calculations indicate that committing subjects would have missed their target with probability around 0.091 in the absence of commitment, and do miss their target with commitment in place with probability 0.026. Hence, commitment can viewed as binding in about 7.5 percent of cases, effectively forcing an individual to do more work than they instantaneously desire. Ashraf et al.(2006) consider hypothetical intertemporal choices over money, rice and ice cream and link those to take-up of a savings commitment device. The authors show that present bias in the hypothetical monetary decisions is significantly correlated at the 10% level with take-up for women. We contrast two dimensions of our study with these prior findings. The first concerns the techniques used to measure dynamic inconsistency, and the second is the extent to which subjects are bound by commitment. As opposed to monetary discounting measures or dynamic inconsistency inferred from payday effects, we attempt to measure discounting directly with intertemporal allocations of effort delivering identification. As opposed to commitments with somewhat limited binding probabilities, our committing subjects are clearly bound by commit-

49The difference for non-committers is no longer significantly different from zero F (1, 79) = 0.39, (p = 0.53).

35 ment.

3.4.1 The Value of Commitment

A natural question is how much should subjects be willing to pay for commitment. In Appendix A we present the value of commitment, V , as the utility difference between the discounted costs of commitment and flexibility. Given our experimental structure we can only assess the monetary value of commitment. Virtually nobody is willing to pay more than $0.25 for commitment, with 91 percent of subjects preferring flexibility when the price of commitment is $0.25. Likewise, nobody is willing to pay more than $0.25 for flexibility, with 90 percent of subjects preferring commitment when the price of commitment is -$0.25. Taking the midpoint of each person’s price list switching interval, the data thus imply a median valuation of $0.125.50 For committers and non-committers, the median valuation is $0.125 and $-0.125, respectively. What do these monetary valuations imply for the extent of V and correspondingly for the extent of sophistication? In AppendixA, we theoretically investigate the valuation of commitment through the lens of the partially sophisticated quasi-hyperbolic model of O’Donoghue and Rabin(2001). We recover the valuation of commitment, V , for stationary cost functions. This analysis shows that the value of commitment is linked to the extent of sophistication, which is governed by sophistication parameter βb, reflecting an individual’s assessment of their future present bias. If βb = 1, an individual is perfectly naive, and if βb = β, an individual is perfectly sophisticated. Values of βb ∈ {β, 1} correspond to partial sophistication. That present bias is predictive of commitment demand at price zero indicates at least partial sophistication on average, βb < 1. The level of V can be calculated directly for the fully sophisticated benchmark of βb = β, which implies a perfect forecast for present-biased behavior. Using the parameters estimates of Table3, columns (3) and (4) and the actual allocations at R = 1, we can calculate the fully- sophisticated value of commitment for committing and non-comitting subjects. For comitting

50For this measure we exclude the four individuals with multiple switching.

36 subjects, we calculate VC=1 = 1.23, which can be expressed in equivalent number of tasks as

−1 c (1.23) = 1.14 tasks. For non-comitting subjects, we calculate VC=0 = −2.06, which can be expressed in equivalent number of tasks as -1.59 tasks. To relate the value of roughly two tasks to mooney, note that on average, using minimum work completion rate, subjects complete approximately 60 tasks per hour. Assuming earnings of around $12 per hour and a constant task value, a subject would be willing to complete 1 task for around $0.20.51 Hence the monetary value of commitment should be around $0.23 for committing subjects and the value of flexibility should be around $0.32 for non-committing subjects. These values compare favorably to the monetary valuations reported above. Hence, assuming complete sophistication and no additional benefits to flexibility, we predict monetary commitment valuations reasonably close to the valuations expressed by subjects.52 We are hesitant to draw strong conclusions beyond the plausibility of sophistication from our commitment valuation data. First, given the ex-post parameter estimates, our elicitation procedure clearly was not optimized for fine price differentiations. Second, it is possible that subjects largely followed the money in the elicitation, preferring either commitment or flexibility depending on which option provided additional payment. A direct experiment precisely identifying βb is a clear next step that research in this vein should take.

3.5 Between Subjects Replication Exercise

A key contribution of our data is the documentation of limited present bias in the domain of money and more substantial present bias in the domain of work. One interpretation is that models of dynamic inconsistency are validated when tested in their relevant domain (consump-

51The assumption of constant per task reservation value is important. With convex costs an individual should have a lower reservation value for the first task than the sixtieth. We opt to present the average valuation recognizing the possibility that valuations could be either higher or lower. AppendixD analyzes the value of commitment demand at a wide range of potential per task valuations to provide sensitivity analysis. 52 If individuals are fully sophisticated, monetary valuations for commitment should be close to those observed. Naturally, evaluating βˆ > β lowers the value of commitment and for βˆ = 1 commitment should be worth exactly zero. In AppendixD we analyze specific values of βˆ and corresponding valuations for commitment under various assumptions for the transformation of V to dollars. This analysis also considers all allocations, not only those at one interest rate. Clear from this exercise is that under the assumption of no additional benefits to flexibility, only in extreme cases should commitment be worth more than a dollar.

37 tion) and that choices over fungible monetary payments cannot easily speak to such models’ predictions. However, in our within-subjects study, several design choices were made that might muddy this interpretation. First, subjects faced different interest rates and forms of budget constraint for effort and for money.53 Second, the delay lengths for money were three to six weeks, while the delay lengths for effort were only one week. Third, subjects always completed their effort allocations prior to completing their monetary allocations. Fourth, present bias is identified for effort from only a dynamic choice, while present bias is identified for money from a combination of static and dynamic choices.54 Fifth, for effort one allocation was chosen to be the allocation- that-counts from the initial and subsequent allocations with an asymmetric probability, while for money each allocation could be the alllocation-that-counts with equal probability. Further, the Week 4 monetary choices were paid separately from the Week 1 choices. Though each design choice has a natural motivation, including our desire to replicate prior exercises, one could potentially imagine them influencing the degree of dynamic inconsistency.55 To alleviate these concerns, we conducted a between subjects replication exercise. 200 subjects, again from the UC Berkeley Xlab subject pool, were randomized into two conditions: one in which allocations were made for money and one in which allocations were made for greek transcription. In both conditions subjects selected into a four week study on decision-making over time and were informed that their earnings would be approximately $60 if all aspects of the study were completed. The main goal of the replication exercise is to keep allocation decisions

53 That is, the constraint for effort was of a present value form, et + Ret+k = 50, while the constraint for money was of a future value form, P ct + ct+k = 20. 54That is, for effort to identify present bias one compares the Week 1 allocations over Weeks 2 and 3 to the Week 2 choices over Weeks 2 and 3. For money to identify present bias one compares the Week 1 allocations over Weeks 4 and 7 to the Week 4 choices over Weeks 4 and 7, the Week 1 allocations over Weeks 1 and 4 to the Week 1 allocations over Weeks 4 and 7, and the Week 1 allocations over Weeks 1 and 4 to the Week 1 allocations over Weeks 1 and 7. 55The specific rationale for each choice, respectively: first, we expected substantially more curvature for effort than money, which suggests different interest rates to avoid corner solutions. Second, we organized the monetary choices around dates the subjects would come to the lab to equalize transactions costs. Third, our primary focus was the effort choices, hence we sought to ensure theses data were collected. Fourth, we wished to replicate the standard static evidence on present bias in money and benefited from an opportunity in Week 4 to additionally generate dynamic evidence. Fifth and sixth, we did not wish to burden the subjects with another, potentially complicated, procedure for determining which monetary decision would be implemented.

38 identical, with the only difference being whether allocations are over money or effort. Mirroring our effort study, in Week 1 of the replication exercise subjects make allocations over Weeks 2 and 3. In Week 2, subjects again make allocations over Weeks 2 and 3. All allocations are made on a study website either in the lab in Week 1 or on any computer with internet access in Week 2. In Week 2, one of the Week 1 or Week 2 decisions is chosen at random, with each having equal probability, and the corresponding allocation is implemented. For both effort and money, allocations are made using budgets of the form,

P a2 + a3 = m.

Where a2 refers to an allocation of either effort or money to Week 2 and a3 refers to an allocation of either effort or money to Week 3. For both effort and money P ∈ {0.66, 0.8, 0.91, 0.95, 1, 1.05, 1.11, 1.25, 1.54}, covering the interest rates used for both money and effort from our initial experiment. For money m = $20 and for effort m = 60 tasks, such that units are easily matched by dividing by three. Following our prior study, minimum payments of $5 and minimum work of 10 tasks are implemented in Weeks 1, 2, and 3. We attempt to put precise time stamps on both the completion of tasks and the collection of money. For effort, subjects are told they must complete their tasks from the chosen allocation on a study website between 9 am and 6 pm on the relevant day in Weeks 2 and 3. For money, subjects are told they must collect their payments from the chosen allocation at the UC Berkeley Xlab between 9 am and 6 pm on the relevant day in Weeks 2 and 3. To make the Week 2 allocations as immediate as possible, subjects are additionally told in advance they will have to either complete their Week 2 tasks or collect their Week 2 funds within two hours of making their Week 2 allocations. AppendixG has the full study instructions. If subjects complete all aspects of the study, including collecting their money or completing their tasks on each relevant date within the relevant time window, they are eligible for a completion payment paid in the fourth week of the study. For effort, the completion payment is $60 with a non-completion payment of $5. For money, the completion payment is $30 with a

39 non-completion payment of $5. All payments, including those from monetary allocations, are made in cash at the Xlab by a single research assistant who remained in place from 9 am to 6 pm on the relevant dates. All 200 subjects began the study on Thursday April 17, 2014. Of these a total of 194 completed the study on Thursday May 1, 2014, with 95 from the effort condition and 99 from the money condition. In this between subjects design, we can directly compare present bias across conditions. Figure8 plots the number of tasks in Panel A (out of 60) and amount of money in Panel B (out of $20) allocated to Week 3 for each level of P . Separate series are provided for when the allocation is made in Week 1 and in Week 2. Note that because the budget constraints are identical, Week 3 tasks are decreasing in P , while Week 3 money is increasing in P . Note as well that due to the form of the budget, it is the constant-value Week 3 units that are graphed.56 Figure8 closely reproduces our prior within-subject findings. At each value of P , individuals appear present-biased for effort, allocating more effort to the later date when the sooner date is the present. Controlling for P , subjects allocate 2.14 (clustered s.e. = 1.10) more tasks to Week 3 in Week 2 relative to Week 1, F (1, 94) = 3.82, p = 0.05. In contrast, for money mean behavior appears almost perfectly dynamically consistent. Controlling for P , subjects allocate $0.14 (clustered s.e. = 0.12) less to Week 3 in Week 2 relative to Week 1, F (1, 98) = 1.37, p = 0.25. Appendix Table A5 provides a corresponding tabulation of behavior, presenting budget shares and the proportion of choices that can be classified as present-biased.57 Non-parametric replication in hand, we now turn to estimation of aggregate utility parameters. In Table4, we replicate the estimation exercise of Table2 with the new between-subjects data. The parameter values and corresponding conclusions are effectively unchanged. For

56This is in contrast to the prior effort figures where earlier tasks had constant value and were graphed and the prior money figures where earlier money was also graphed for ease of comparison. 57For consistency with Appendix Tables A3 and A4, Appendix Table A5 tabulates budget shares for the sooner date, calculated as (P a2)/m for each allocation. For effort, subjects initially allocate around 52.4% (clustered s.e. = 1.1) of their experimental budget to the sooner work date and subsequently allocate around 48.8% (1.7) to the sooner work date, F (1, 94) = 3.82, (p = 0.05). Twenty-five percent of individual choices are dynamically consistent, 43% are present-biased, and 32% are future-biased. For money, subjects initially allocate around 51.4% (0.7) of their experimental budget to the sooner payment and subsequently allocate around 51.9% (0.6) to the sooner payment, F (1, 98) = 0.85, (p = 0.36). Eighty-three percent of individual choices are dynamically consistent, 10% are present-biased, and 7% are future-biased.

40 Figure 8: Between Subjects Replication Exercise

Panel A: Effort Panel B: Money 20 40 15 35 30 10 25 5 Tasks to Date Later Allocated Dollars Dollars to Date Later Allocated 20 0 15 .6 .8 1 1.2 1.4 1.6 .6 .8 1 1.2 1.4 1.6 P P

(from Pa2+a3=60) (from Pa2+a3=20)

Week 1 Allocation Week 2 Allocation Mean Mean SE

monetary present bias in column (1), we estimate β = 0.997 (clustered s.e. = 0.005), which compares favorably to Table2, column (2), which estimates β = 0.988 (0.009). Similar to our within-subjects conclusion, we fail to reject the null hypothesis of dynamic consistency, β = 1, for money, χ2(1) = 0.50, (p = 0.48). Interestingly, we also find quite similar discount factor and curvature estimates between Table4, column (1) and Table2, column (2). For effort present bias in column (2), we estimate β = 0.892 (0.056), which compares favorably to Table 2, column (3) for greek transcription where β = 0.900 (0.037). Similar to our within-subjects conclusion, we reject the null hypothesis of β = 1 for effort, χ2(1) = 3.73, (p = 0.05). Again, we find quite similar estimates for the auxiliary parameters between Table4, column (2) and Table2, column (3). The analysis again allows us to compare present bias across effort and money, and again we reject the null hypothesis that the β identified for money is equal to that identified for effort, χ2(1) = 3.50, (p = 0.06).58

58 Appendix Tables A8 and A9 provide individual estimates of βe and βm along with a summary of allocation behavior for these subjects. Subjects with no variation in experimental response in a given week are also noted. 16 of 194 non-attriting subjects have no variation in experimental response in one or more weeks and 14 of these

41 Though these ﬁndings closely replicate our prior within-subjects data, it is important to note that the data from this exercise yields somewhat less precise measures and test statistics than our initial study. We hesitate to speculate as to the source of this imprecision, and draw some comfort from the replication of the point estimates from our prior work.

Table 4: Replication Exercise Parameter Estimates

Monetary Discounting Eﬀort Discounting Greek (1) (2)) Present Bias Parameter: β 0.997 0.892 (0.005) (0.056) Weekly Discount Factor: (δ)7 0.998 1.009 (0.001) (0.005) Monetary Curvature Parameter: α 0.952 (0.009) Cost of Eﬀort Parameter: γ 1.774 (0.167)

# Observations 1782 1710 # Clusters 99 95

2 2 H0 : β = 1 χ (1) = 0.50 χ (1) = 3.73 (p = 0.48) (p = 0.05) 2 H0 : β(Col. 1) = β(Col. 2) χ (1) = 3.50 (p = 0.06)

4 Conclusion

Present biased time preferences are a core of behavioral research. The key hypothesis of diminishing impatience through time is able to capture a number of behavioral regularities at odds subjects were in the eﬀort condition. Importantly, the results of Table4 are maintained if we eliminate such subjects with no variation in one or more weeks. See Appendix Table A14 for detail.

42 with standard exponential discounting. Further, the possibility of sophistication provides an important channel for policy improvements via the provision of commitment devices. With the exception of only a few pieces of research, most evidence of dynamic inconsistency is generated from experimental choices over time-dated monetary payments. When those are administered in a way to keep transaction costs constant and uncertainty at bay, recent studies have found limited evidence of dynamic inconsistency. However, such findings may not be appropriate to reject a model defined over streams of consumption. The present study attempts to identify dynamic inconsistency for choices over real effort. We introduce a longitudinal design asking subjects to allocate and subsequently allocate again units of effort through time. A complementary monetary study is conducted for comparison. We document three key findings. First, in choices over monetary payments, we find limited evidence of present bias, confirming earlier work. Second, in choices over effort, we find substantial present bias. Subjects reallocate about 9% less work to the present than their initial allocation. Corresponding parameter estimates generate a similar conclusion. Individuals are estimated to be substantially present-biased in effort choices and significantly closer to dynamically consistent in choices over money. Third, we study commitment demand, documenting that at price zero roughly 60% of subjects prefer commitment to flexibility. A key result is that these commitment decisions correlate significantly with previously measured present bias. Individuals who demand commitment are significantly more present-biased in effort than those who do not. This provides validation for our experimental measures and helps to rule out a variety of potential confounds. Importantly, in our design commitment meaningfully restricts activities. Committed subjects are required to complete more effort than they instantaneously desire. By documenting the link between experimentally measured present bias and commitment demand, we provide support for models of dynamic inconsistency with sophistication. Subjects are potentially aware of their present bias and take actions to limit their future behavior. We view our paper as providing a portable experimental method allowing tractable estimation of intertemporal preferences over consumption (effort) and correlating such preferences

43 with a meaningful, potentially constraining, commitment device. Though the implementation here is with American undergraduates, we feel the design is suitable for field interventions. We draw one conclusion and several words of caution from our findings. Our results indicate that present bias is plausibly identified in choices over effort and, furthermore, is linked to effort-related commitment demand. However, we caution using the estimated parameters at face value as they are for a specific subject pool (self-selected to work for six weeks for final payment in week seven) and a specific task. There may be other decision environments wherein behavior may not be well captured by models of dynamic inconsistency. For example, subjects may wish to get a painful single experience over with immediately or postpone a single pleasure (Loewenstein, 1987).59 Lastly and most importantly, though fungibility issues may be mediated in the present design, the natural problems of arbitrage will still exist if subjects substitute effort in the lab with their extra-lab behavior. The existence and use of such substitutes, like avoiding doing laundry or homework in response to the experiment, will confound our measures in much the same way as monetary studies. Discounting will be biased towards market interest rates, present bias will be exhibited only if such rates change through time, and cost functions will be biased towards linearity. Though our data suggest effort is less fungible than money, one cannot say that extra-lab smoothing opportunities for effort are eliminated. Hence, one should view our measures as lower bounds on the true extent of dynamic inconsistency and the instantaneous cost of tasks. We want to, however, point out that to some extent such fungibility will be present in many dimensions in which time inconsistency has been measured. Ultimately, the best measure of time inconsistency will be one that predicts ecologically relevant decisions across a broad set of environments. This suggests important avenues for future research.

59This suggests a key anticipatory component of intertemporal behavior, potentially mediated by our design’s use of minimum eﬀort requirements and convex decisions.

44 References

Andersen, Steﬀen, Glenn W. Harrison, Morten I. Lau, and Elisabet E. Rutstrom, “Eliciting Risk and Time Preferences,” Econometrica, 2008, 76 (3), 583–618.

, , , and , “Discounting Behavior: A Reconsideration,” Working Paper, 2012.

Andreoni, James and Charles Sprenger, “Estimating Time Preferences with Convex Bud- gets,” American Economic Review, 2012, 102 (7), 3333–3356.

and , “Risk Preferences Are Not Time Preferences,” American Economic Review, 2012, 102 (7), 3357–3376.

Ariely, Dan and Klaus Wertenbroch, “Procrastination, Deadlines, and Performance: Self- Control by Precommitment,” Psychological Science, 2002, 13 (3), 219–224.

Ashraf, Nava, Dean Karlan, and Wesley Yin, “Tying Odysseus to the Mast: Evidence from a Commitment Savings Product in the Philippines,” Quarterly Journal of Economics, 2006, 121 (1), 635–672.

Bauer, Michal, Julie Chytilova, and Jonathan Morduch, “Behavioral Foundations of Microcredit: Experimental and Survey Evidence from Rural India,” American Economic Review, 2012, 102 (2), 1118–1139.

Benjamin, Daniel, James Choi, and Joshua Strickland, “Social Identity and Prefer- ences,” American Economic Review, 2010, 100 (4), 1913–1928.

Brown, Alexander L., Zhikang Eric Chua, and Colin F. Camerer, “Learning and Visceral Temptation in Dynamic Saving Experiments,” Quarterly Journal of Economics, 2009, 124 (1), 197–231.

Chabris, Christopher F., David Laibson, and Jonathon P. Schuldt, “Intertemporal Choice,” in Steven N. Durlauf and Larry Blume, eds., The New Palgrave Dictionary of Economics, London: Palgrave Macmillan, 2008.

45 Cheung, Stephen L., “On the Elicitation of Time Preference Under Conditions of Risk,” American Economic Review, Forthcoming.

Chew, S. H. and Larry G. Epstein, “The Structure of Preferences and Attitudes Towards the Timing of the Resolution of Uncertainty,” International Economic Review, 1989, 30 (1), 103–117.

Coller, Maribeth and Melonie B. Williams, “Eliciting individual discount rates,” Exper- imental Economics, 1999, 2, 107–127.

Cubitt, Robin P. and Daniel Read, “Can Intertemporal Choice Experiments Elicit Pref- erences for Consumption,” Experimental Economics, 2007, 10 (4), 369–389.

Dohmen, Thomas, Armin Falk, David Huﬀman, and Uwe Sunde, “Dynamic inconsistency predicts self-control problems in humans,” Working Paper, 2006.

, , , and , “Are Risk Aversion and Impatience Related to Cognitive Ability,” American Economic Review, 2010, 100 (3), 256–271.

Dupas, Pascaline and Jonathan Robinson, “Why Don’t the Poor Save More? Evidence from Health Savings Experiments,” American Economic Review, 2013, 103 (4), 1138–1171.

Epstein, Larry G. and Stanley E. Zin, “Substitution, Risk Aversion, and the Temporal Behavior of Consumption and Asset Returns: A Theoretical Framework,” Econometrica, 1989, 57 (4), 937–969.

Frederick, Shane, George Loewenstein, and Ted O’Donoghue, “Time discounting and time preference: A critical review,” Journal of Economic Literature, 2002, 40 (2), 351–401.

Gine, Xavier, Jessica Goldberg, Dan Silverman, and Dean Yang, “Revising Commit- ments: Time Preference and Time-Inconsistency in the Field,” Working Paper, 2010.

Giordano, Louis A., Warren K. Bickel, George Loewenstein, Eric A. Jacobs, Lisa Marsch, and Gary J. Badger, “Mild opioid deprivation increases the degree that opioid

46 dependent outpatients discount delayed heroin and money,” Psychopharmacology, 2002, 163, 174–182.

Halevy, Yoram, “Strotz Meets Allais: Diminishing Impatience and the Certainty Eﬀect,” American Economic Review, 2008, 98 (3), 1145–1162.

, “Time Consistency: Stationarity and Time Invariance,” Working Paper, 2012.

Harrison, Glenn W., Morten I. Lau, and Melonie B. Williams, “Estimating individual discount rates in Denmark: A ﬁeld experiment,” American Economic Review, 2002, 92 (5), 1606–1617.

Kaur, Supreet, Michael Kremer, and Sendhil Mullainathan, “Self-Control and the Development of Work Arrangements,” American Economic Review, Papers and Proceedings, 2010, 100 (2), 624–628.

Kirby, Kris N., Nancy M. Petry, and Warren K. Bickel, “Heroin addicts have higher discount rates for delayed rewards than non-drug-using controls,” Journal of Experimental Psychology: General, 1999, 128, 78–87.

Knutson, Brian, Charles M. Adams, Grace W. Fong, and Daniel Hommer, “An- ticipation of Increasing Monetary Reward Selectively Recruits Nucleus Accumbens,” The Journal of Neuroscience, 2001, 21 (RC159), 1–5.

Kreps, David M. and Evan L. Porteus, “Temporal Resolution of Uncertainty and Dynamic Choice Theory,” Econometrica, 1978, 46 (1), 185–200.

Laibson, David, “Golden Eggs and Hyperbolic Discounting,” Quarterly Journal of Eco- nomics, 1997, 112 (2), 443–477.

Lawrance, Emily C., “Poverty and the Rate of Time Preference: Evidence from Panel Data,” Journal of Political Economy, 1991, 99 (1), 54–77.

47 Loewenstein, George F., “Anticipation and the Valuation of Delayed Consumption,” The Economic Journal, 1987, 97 (387), 666–684.

McClure, Samuel, David Laibson, George Loewenstein, and Jonathan Cohen, “Sep- arate neural systems value immediate and delayed monetary rewards,” Science, 2004, 306, 503–507.

, , , and , “Time discounting for primary rewards,” Journal of Neuroscience, 2007, 27 (21), 5796–5804.

Meier, Stephan and Charles Sprenger, “Present-Biased Preferences and Credit Card Bor- rowing,” American Economic Journal - Applied Economics, 2010, 2 (1), 193–210.

Miao, Bin and Songfa Zhong, “Separating Risk and Time Preferences,” Working Paper, 2012.

O’Donoghue, Ted and Matthew Rabin, “Doing it Now or Later,” American Economic Review, 1999, 89 (1), 103–124.

and , “Choice and Procrastination,” The Quarterly Journal of Economics, 2001, 116 (1), 121–160.

Read, Daniel and Barbara van Leeuwen, “Predicting Hunger: The Eﬀects of Appetite and Delay on Choice,” Organizational Behavior and Human Decision Processes, 1998, 76 (2), 189–205.

Samuelson, Paul A., “A Note on Measurement of Utility,” The Review of Economic Studies, 1937, 4 (2), 155–161.

Shapiro, Matthew D., “The Permanent Income Hypothesis and the Real Interest Rate: Some Evidence from Panel Data,” Economics Letters, 1984, 14 (1), 93–100.

48 Solnick, Jay V., Catherine H. Kannenberg, David A. Eckerman, and Marcus B. Waller, “An Experimental Analysis of Impulsivity and Impulse Control in Humans,” Learn- ing and Motivation, 1980, 11, 61–77.

Strotz, Robert H., “Myopia and Inconsistency in Dynamic Utility Maximization,” Review of Economic Studies, 1956, 23, 165–180.

Sutter, Matthias, Martin G. Kocher, Daniela Glatzle-Ruetzler, and Stefan T. Trautmann, “Impatience and Uncertainty: Experimental Decisions Predict Adolescents’ Field Behavior,” American Economic Review, 2013, 103 (1), 510–531.

Tanaka, Tomomi, Colin Camerer, and Quang Nguyen, “Risk and time preferences: Experimental and household data from Vietnam,” American Economic Review, 2010, 100 (1), 557–571.

Thaler, Richard H., “Some Empircal Evidence on Dynamic Inconsistency,” Economics Let- ters, 1981, pp. 201–207.

Voors, Maarten, Eleonora Nillesen, Philip Verwimp, Erwin Bulte, Robert Lensink, and Daan van Soest, “Violent Conﬂict and Behavior: A Field Experiment in Burundi,” American Economic Review, 2012, 102 (2), 941–964.

Wooldridge, Jeﬀrey M., Econometric Analysis of Cross Section and Panel Data, MIT Press: Cambridge, MA, 2002.

Zeldes, Stephen B., “Consumption and Liquidity Constraints: An Empirical Investigation,” The Journal of Political Economy, 1989, 97 (2), 305–346.

49 APPENDIX - NOT FOR PUBLICATION

A Theoretical Structure and Identiﬁcation Appendix

In the intertemporal allocation of effort and money, discounting and additional parameters can be identified at either the aggregate or individual level under various structural assumptions. In the following three appendix subsections we describe our theoretical environment, explore the demand for commitment, demonstrate which experimental variation provides identification of specific parameters of interest, and lay out methodology for estimation. A fourth subsection presents estimation details for monetary discounting.

A.1 Eﬀort Discounting

A.1.1 Allocation Timing

In the working over time experiment, subjects allocate eﬀort to an earlier date, et, and a later date, et+k, subject to the intertemporal budget constraint described in (1). Subjects make allocations at two points in time, one at time s < t, and one at time t. The allocation-that- counts is randomly implemented from time s with probability p and from time t with probability

60 s∗ 1 − p. Let et,s be the allocation of eﬀort to time t chosen at time s. Let et,t be the allocation

s∗ of eﬀort to time t forecasted to be chosen at time t from the perspective of time s. That is, et,t captures what an individual at time s believes they will optimally choose at time t.

A.1.2 Preferences

To develop our theory, we assume an instantaneous cost function, c(e), for eﬀort, e, that is time separable, stationary, and of an expected utility form with respect to the probability that an allocation is implemented. To aid our development and foreshadow our empirical

60We abstract from the fact that subjects make multiple allocations. Given the assumed separability over time and in probabilities, this abstraction is innocuous.

1 implementation we also make a functional form assumption for the shape of c(·). We assume

c(e) = (e + ω)γ, where γ > 1 represents the stationary parameter on the convex instantaneous cost of eﬀort function. The additive term ω in the cost function could be interpreted as a Stone-Geary minimum or as some background level of required work. Such parameters are used in monetary discounting studies (Andersen et al., 2008; Andreoni and Sprenger, 2012a), and are either taken from some external data source on background consumption or estimated from experimental choices. For simplicity, we interpret ω as the required minimum work of the experiment and set ω = 10 for our eﬀort analysis.61 We assume discounting follows the quasi-hyperbolic partially sophisticated form proposed by O’Donoghue and Rabin(2001). For two periods, t and t + k, discounting, D(t, t + k), is captured by   βδk if k > 0 D(t, t + k) =  1 if k = 0.

The parameter β captures the degree of present bias while the parameter δ captures long run discounting. β = 1 nests the standard model of exponential discounting. From period s < t, the discounted costs of eﬀort at times t and t + k can be written as

t−s γ t+k−s γ βδ (et,s + ω) + βδ (et+k,s + ω) .

61Andreoni and Sprenger(2012a) provide estimates for ω based on non-linear least squares techniques and analyze the extent to which different assumptions for ω influence remaining parameter estimates. Though utility curvature and discounting are sensitive to varying assumptions for ω, present bias, β, is largely unaffected. Andersen et al.(2008) also provide some sensitivity analysis.

2 Eliminating common terms, the decision problem at time s can be written as

γ k γ minet,s.et+k,s p · [(et,s + ω) + δ (et+k,s + ω) ] +

s∗ γ k s∗ γ (1 − p) · [(et,t + ω) + δ (et+k,t + ω) ]

s.t. et,s + R · et+k,s = m,

∗ ∗ which yields the intertemporal Euler equation satisﬁed by the optimal allocation, (et,s, et+k,s),

∗ et,s + ω γ−1 1 1 ( ∗ ) k = . et+k,s + ω δ R

s∗ s∗ Note that the forecasted allocation, (et,t, et+k,t), and the probability of implementation, p, do not feature in the intertemporal Euler due to the assumed separability. Similarly, the decision problem at time t can be written

∗ γ k ∗ γ minet,t.et+k,t p · [(et,s + ω) + βδ (et+k,s + ω) ] +

γ k γ (1 − p) · [[(et,t + ω) + βδ (et+k,t + ω) ]]

s.t. et,t + R · et+k,t = m,

∗ ∗ with corresponding Euler equation satisﬁed by the optimal allocation, (et,t, et+k,t),

∗ et,t + ω γ−1 1 1 ( ∗ ) k = et+k,t + ω βδ R

∗ ∗ The prior allocation, (et,s, et+k,s), and the probability of implementation do not feature in the intertemporal Euler. Any diﬀerences in allocations between time s and time t are delivered by the present bias term, β.62

62A recent discussion of non-expected utility behavior in intertemporal settings has demonstrated that apparently present-biased behavior can be delivered by deviations from expected utility (see, e.g., Halevy, 2008). Under discounted expected utility, allocations over two periods should depend on the ratio of probabilities with which the allocations are realized. In two important conditions Andreoni and Sprenger(2012b) demonstrate in the monetary domain that if sooner and later payments are paid independently with probability 0.5, behavior deviates from the common ratio counterpart of all payments being certain. Under expected utility and atempo-

3 Combining our Euler equations we have

e∗ + ω 1 1 ( t,D )γ−1 = (7) ∗ 1D=t k et+k,D + ω β δ R where D ∈ {s, t} represents whether the allocation decision was made at time t or time s. Note

∗ that for β < 1, an allocation made at time t at a given R will have a lower value of et,D than an allocation made at time s. A present-biased individual allocates less work to time t at time t than they did at time s. Naturally, the prediction that dynamically inconsistent behavior depends only on β relies on the assumption of a stationary cost function. Changes in the cost function through time could easily lead to diﬀerences in allocations between time s and t. Such changing costs could be delivered by a variety of sources. For example, there could be permanent shocks to the cost function, perhaps due to a misforecasting of task diﬃculty. There could also be temporary shocks due to some random events that impose time constraints or leave subjects more tired and exhausted than they normally are. In sectionC we address these concerns directly and provide evidence that such possibilities are unlikely to drive observed behavior.

A.1.3 Partial Sophistication

We allow for the fact that individuals may be partially sophisticated with respect to their own present bias. The nature of sophistication follows that of O’Donoghue and Rabin(2001), where βb captures the belief an individual has on his future present bias: βb = β represents full sophistication, βb = 1 represents full naivete, and βb ∈ (β, 1) represents partial sophistication. ral applications of prospect theory, the deviations cannot be rationalized. An intuition for the eﬀect is that the independent payment probabilities give subjects the opportunity to hedge through time. Cheung(Forthcoming) and Miao and Zhong(2012) demonstrate the importance of this intuition, as they show in the Andreoni and Sprenger(2012b) setup that when one makes the two 0.5 realization probabilities perfectly correlated behavior is closer to the expected utility benchmark. In our environment, the implementation probability applies equally to both the sooner and later work date, creating perfect correlation through time. Hence, the eﬀects of Andreoni and Sprenger(2012b) are unlikely to be present. Additionally, because the same implementation probability applies to both work dates, any non-linear treatment of p or 1 − p must be applied equally, and so drop out of marginal conditions in exactly the same way that undistorted probabilities do. Further potential concerns with respect to the asymmetry of p and 1 − p in the design are addressed in our replication exercise where initial and subsequent allocations are implemented with equal probability. See section 3.5 for detail.

4 This means that allocations at time t, forecasted at time s < t are

s∗ s∗ ∗ γ k ∗ γ (et,t, et+k,t) = argmin p · [(et,s + ω) + βδb (et+k,s + ω) ] +

s γ k s γ (1 − p) · [(et,t + ω) + βδb (et+k,t + ω) ]

s s s.t. et,t + R · et+k,t = m.

s∗ s∗ If βb ∈ (β, 1] an individual’s forecasted allocation, (et,t, et+k,t) will not accord with their actual

∗ ∗ subsequent allocation, (et,t, et+k,t). Note the sophistication parameter, βb is absent from the Euler formulations above. This is by construction both in the theory and the experimental design. An individual at time s may forecast a level of present bias at time t but is incapable of controlling behavior at that point in time. More importantly, this forecasted present bias at time t does not inﬂuence his behavior at time s. The only actions available to the time t self is to complete the time s allocation with probability p, complete the time t allocation with probability 1 − p, or opt out of the experiment, foregoing $90. Given the high penalty, an individual at time s can appropriately forecast the third action will not be taken. The individual is aware that he cannot control the second action. Hence, he optimizes according to his time s preference as above with βb absent from the formulation. The parameter βb will be important for our analysis of commitment in which an individual at time s may indeed control time t behavior.

A.2 Commitment

In the second block of the experiment subjects are oﬀered a probabilistic commitment device. The commitment device favors the initial allocations made at time s over the subsequent allocations made at time t by changing the time s implementation probability from p to 1 − p (i.e. from 0.1 to 0.9). Recall that intertemporal Euler equations and allocations are independent of implementation probabilities. Hence, the value of commitment can be arrived at by comparing discounted

5 costs. An individual prefers to commit if the discounted costs of the chosen allocation at time s are smaller than the discounted costs of the forecasted allocation for time t at time s.63 The value of commitment is given as

t−s s∗ γ k s∗ γ ∗ γ k ∗ γ V = (1 − 2p) · βδ ·{[(et,t + ω) + δ c(et+k,t + ω) ] − [(et,s + ω) + δ c(et+k,s + ω) ]}.

Note that the value of commitment, V , depends upon both actual allocations and forecasted allocations at time s. Hence, the value of commitment depends upon the degree of sophistica-

s∗ s∗ ∗ ∗ tion. Clearly, for naive individuals with βb = 1, (et,t, et+k,t) = (et,s, et+k,s). Actual and forecasted allocations are identical and the value of commitment is zero.

For sophisticated individuals, βb ∈ [β, 1), actual allocations and forecasted allocations at

s∗ s∗ time s diﬀer. By the deﬁnition of the minimum from the perspective of period s,(et,t, et+k,t)

∗ ∗ yields higher discounted costs than (et,s, et+k,s). This implies that the value of commitment should be positive provided p < 0.5, as in the experiment. As βb diverges from 1, the value of commitment increases. AppendixD provides further detail and corresponding simulated values. The extent of commitment demand, when combined with parametric measures for discounting and costs, can be informative for the extent of sophistication. Naturally, there may be intrinsic benefits to flexibility. These unmodeled benefits to flexibility could have many sources including future uncertainty in costs or task difficulty.64

63The inequality between discounted costs

t−s ∗ γ t+k−s ∗ γ t−s s∗ γ t+k−s s∗ γ (1 − p) · [βδ (et,s + ω) + βδ c(et+k,s + ω) ] + p · [βδ (et,t + ω) + βδ c(et+k,t + ω) ] < t−s ∗ γ t+k−s ∗ γ t−s s∗ γ t+k−s s∗ γ p · [βδ (et,s + ω) + βδ c(et+k,s + ω) ] + (1 − p) · [βδ (et,t + ω) + βδ c(et+k,t + ω) ], reduces to the inequality,

∗ γ k ∗ γ s∗ γ k s∗ γ (et,s + ω) + δ c(et+k,s + ω) < (et,t + ω) + δ c(et+k,t + ω) , provided p < 0.5 as in the experiment. Subtracting the discounted costs one arrives at the value of commitment,

t−s ∗ γ t+k−s ∗ γ t−s s∗ γ t+k−s s∗ γ V = {p · [βδ (et,s + ω) + βδ c(et+k,s + ω) ] + (1 − p) · [βδ (et,t + ω) + βδ c(et+k,t + ω) ]} − t−s ∗ γ t+k−s ∗ γ t−s s∗ γ t+k−s s∗ γ {(1 − p) · [βδ (et,s + ω) + βδ c(et+k,s + ω) ] + p · [βδ (et,t + ω) + βδ c(et+k,t + ω) ]}.

64Note that in the presence of such factors even sophisticated present-biased subjects may have low or even negative values for commitment. Hence, it is critical that our design elicits the demand for both ﬂexibility and

6 The value of commitment, V , is measured in the same units as the discounted costs of eﬀort. A potential shortfall of our design is that our experiment does not measure V directly but rather measures its translation into dollars. Hence, we provide potential bounds on V based upon assumptions for the transformation of V to dollars.

A.3 Identiﬁcation

From the intertemporal Euler equation, (7), identiﬁcation of discounting and the cost function is straightforward. Rearranging and taking logs yields

et,D + ω log(β) log(δ) 1 log( ) = · (1D=t) + · k − ( ) · log(R), (8) et+k.D + ω γ − 1 γ − 1 γ − 1 which is linear in the key experimental parameters of whether allocations are made at time t, 1D=t, and the log transform, log(R). In our implementation, variation in log(R) delivers identification of the cost function, γ; the allocation being made in Week 1 (D = s) rather than Week 2 (D = t) delivers identification of present bias, β; and the delay length, k = 7 days, gives identification of the discount factor, δ.65 In order to estimate discounting and cost function parameters from aggregate data, we assume an additive error structure and estimate the linear regression implied by (8). To be specific, the regression equation is, for k = 7,

et + ω log( )i = η0k + η1 · (1D=t)i + η2 · log(R)i + i, et+k + ω

and we recover the parameters of interest as β = exp(ˆη1/ − ηˆ2) and γ = 1 + 1/ − ηˆ2. Note ˆ that δ = exp(ˆη0/ − ηˆ2) is recovered from the constant as only one delay length was used in the experimental design. The parameters of interest can be recovered from non-linear combinations of regression commitment to assess the possible presence of such factors. 65Of course, with only one delay length of seven days considered in the experiment, we have limited conﬁdence that our estimate of δ can be extrapolated to arbitrary delay lengths.

7 coefficients with standard errors calculated via the delta method. One important issue to consider in estimation is the potential presence of corner solutions. We provide estimates from two-limit tobit regressions designed to account for the possibility that the tangency condition implied by (8) does not hold with equality (Wooldridge, 2002). Estimating (8) is easily extended to the study of individual parameters. To begin, (8) can be estimated at the individual level.66 However, with limited numbers of individual choices it is helpful to consider alternative, more structured approaches. In particular, we allow for heterogeneous discounting across individuals, but assume all individuals have the same cost function. Consider a vector of fixed effects (1j)i which take the value 1 if observation i was contributed by individual j. This leads to the fixed effects formulation

et,D + ω log(δ) (log(δj) − log(δ)) log(β) log( )i = · k + · (1j)i · k + · (1D=t)i et+k,D + ω γ − 1 γ − 1 γ − 1 (log(β ) − log(β)) 1 + j · (1 ) · (1 ) − · log(R) , γ − 1 D=t i j i γ − 1 i

where δ, β refer to sample means, and δj, βj refer to individual-specific discounting parameters. With an additive error structure this is easily estimable.67 The individual fixed effect interacted with the decision being made in the present provides identification of the individual-specific

βj. In AppendixB we conduct simulation exercises under various correlation structures for the true parameters of interest and document that the implemented estimation methods perform well both at the aggregate and individual level.

A.4 Monetary Discounting

Our methods for recovering monetary discounting parameters at both the aggregate and individual level closely follow those for eﬀort. Following most of the literature, we abstract from standard arbitrage arguments for monetary discounting and assume laboratory administered

66Broadly similar conclusions are reached when estimating (8) at the individual level, however, parameter precision is greatly reduced and substantial estimate instability is uncovered in some cases. 67We allow both β and δ to vary across individuals such that the implemented regression is a standard interaction with both level and slope eﬀects.

8 68 rates are the relevant ones. In particular, for monetary payments, ct and ct+k, allocated subject to the constraint (2), we assume a quasi-hyperbolic constant relative risk averse utility function,

α 1D=t k α U(ct,D, ct+k,D) = (ct + ω) + β δ (ct+k + ω) . (9)

Where D ∈ {s, t} refers to the same notation as before for when the allocation decision is made. The utility function is assumed to be concave, α < 1, such that ﬁrst order conditions provided meaningful optima. Here, the parameter ω is a background parameter that we take to be the $5 minimum payment of the monetary experiment.69 Maximizing (9) subject to the intertemporal budget constraint (2) yields an intertemporal Euler equation similar to that above for eﬀort. Taking logs and rearranging we have

ct,D + ω log(β) log(δ) 1 log( ) = · (1D=t) + · k + ( ) · log(P ). (10) ct+k,D + ω α − 1 α − 1 α − 1

Again, assuming an additive error structure, this can be estimated at the aggregate or individual level via two-limit Tobit. Discounting and utility function parameters can be recovered via non- linear combinations of regression coeﬃcients as above with standard errors estimated again via the delta method.

B Simulation Appendix

This appendix focuses on two questions related to the estimation strategies laid out in Ap- pendixA. First, we examine the extent to which the implemented estimators identify the true

68The assumptions that individuals narrowly bracket time-dated experimental payments, treat money effectively as consumption, and ignore extra-lab arbitrage have been standard in the literature. One prominent exception to this tradition is Harrison et al.(2002), who measure and account for extra-lab borrowing and savings opportunities. 69Andreoni and Sprenger(2012a) provide detailed discussion of the use of such background parameters and provide robustness tests with differing values of ω and differing assumptions for the functional form of utility in CTB estimates. They provide estimates for ω based on non-linear least squares techniques and analyze the extent to which different assumptions for ω influence remaining parameter estimates. Though utility curvature and discounting are sensitive to varying assumptions for ω, present bias, β, is largely unaffected. Andersen et al.(2008) also provide some sensitivity analysis.

9 parameters of interest, β, δ and γ at the aggregate and individual level. As our individual estimates restrict γ to be constant across subjects, this exercise is conducted under various correlation structures for β and γ to understand the sensitivity of our parameter estimates to this restriction. Further, the correlation structure also helps to investigate the sensitivity of identifying β via a non-linear combination involving γ in the aggregate estimates. Second, we investigate the sensitivity of aggregate and individual estimates to uncertainty. Subjects may make allocations in Week 1 that minimize their discounted expected cost in future weeks given the potential realizations of future parameters. This uncertainty may be subsequently resolved in Week 2, such that subjects minimize their discounted cost at specific realizations of key parameters. As the minimizer of the expectation need not be the expectation of the minimizer, such issues can lead to inconsistencies between initial allocations and subsequent allocations. To explore the extent to which this issue hampers identification of present bias, we conduct simulations under several uncertainty structures. Our procedure for conducting the first simulation exercise is straightforward. We draw 100 samples of 80 individuals with underlying true parameters drawn from distributions centered roughly around our aggregate estimates. That is, for each sample β is drawn from a normal distribution with mean 0.9 and standard deviation 0.2; δ is drawn from a normal distribution with mean 0.99 and standard deviation 0.2; and γ is drawn from a normal distribution with mean 1.6 and standard deviation of 0.2. We introduce five correlation structures for the relationship between β and γ, ρ(β, γ) ∈ {−0.75, −0.25, 0, 0.25, 0.75}. For simplicity and to focus attention on the sensitivity of present bias we assume ρ(β, δ) = 0 and ρ(δ, γ) = 0 when drawing each sample. For each of these correlation structures we conduct two key analyses. First, for every sample we estimate the aggregate parameters, βb, bδ and γb. The empirical distribution of βb over the 100 samples is summarized by the empirical mean, βb, the empirical standard deviation, s(βb).

Similar values summarize the empirical distributions of bδ and γb. We investigate the extent to which the estimated values for βb correspond to the underlying data generating process by

10 comparing βb to the true mean β of 0.9. We also provide a measure of type I error in the form of the probability of rejecting β = 0.9 from each of our 100 drawn samples, 0.9 ∈/ CI(β), and a measure of type II error in the form of the probability of rejecting β = 1, 1 ∈/ CI(β). Table A1, Panel A provide these analyses. With zero correlation structure we precisely estimate all parameters close to the true underlying distribution. We reject the truth with probability around 0.10 and remain powered to reject β = 1. With extreme negative correlation of ρ(β, γ) = −0.75, this precision is largely unaffected, though with extreme positive correlation of ρ(β, γ) = 0.75 the aggregate estimator falters. We begin to overestimate the extent of present bias and reject the truth with frequency. This exercise documents the sensitivity of our aggregate estimates to extreme correlation structures. Next, we focus on individual estimates. Table A1, Panel A provides the results. In each sample of 80 observations, we estimate individual parameters based on the fixed effects regression described in sectionA. We consider the median and mean level of the individual estimate med βbi, βbi and βbi , and the correlation between the true draw of βi and the estimated value βbi, r(βi, βbi). For each of the 100 samples, we construct a correlation coefficient, and present the average value. Across correlation structures, we estimate broadly correct average and median values. Importantly, even when the accuracy of the level of behavior deteriorates due to extreme negative correlation between β and γ, we find the correlation between the true βi and βbi remains above 0.9. This indicates that the individual estimates remain capable of identifying differences across individuals in present bias, providing a solid foundation for our individual analysis. The remainder of Table A1 analyzes the effect of uncertainty. We focus on uncertainty in γ realized only in Week 2. Hence the Week 1 allocations are made under uncertainty that is resolved in Week 2. To operationalize this exercise we again have β and δ drawn from the distributions above in advance. However, we assume that in Week 1, subjects do not know their true γ but optimize subject to the knowledge that γ is drawn from a normal distribution with mean 1.6 and standard deviation of σ. We consider five values of σ ∈ {0, 0.05, 0.1, 0.2}.

11 In Panel B, we provide aggregate and individual analysis.. Though the aggregate estimates and error rates are unaﬀected for the lower value of uncertainty, as parametric uncertainty is increased, we begin to overestimate β and reject the truth with frequency. A similar pattern is observed in the individual estimates. Importantly, the presence of parametric uncertainty greatly reduces the correlation between between the true βi and βbi which drops below 0.3 in the more extreme case. A natural question is why parametric uncertainty leads towards upward-biased estimates of β, pushing away from present bias. Intuitively, a subject with parametric uncertainty attempts to avoid situations of high work under extremely convex cost functions that are rarely realized. As this encourages subjects to spread their initial allocations, we estimate a more convex cost function. When the uncertainty is realized, they allocate less evenly over time on average, but the cost function is required by the estimator to remain constant. This change in behavior in Week 2 winds up being captured partially in the form of an increased β in our parameter space of interest.

Table A1: Simulation Exercises

Aggregate Estimates Individual Estimates Panel A: Simulations: δ ∼ N(0.99, 0.22), β ∼ N(0.9, 0.22), γ ∼ N(1.6, 0.22) Correlation Structure: r(β, γ) ∈ {−0.75, −0.25, 0, 0.25, 0.75} med N βb s(βb) 0.9 ∈/ CI(β) 1 ∈/ CI(β) γb bδ βbi βbi r(βi, βbi) r(β, γ)=0 80x100 .8828 .0242 11% 95% 1.552 .9955 .9080 .9077 0.971 r(β, γ)=-0.25 80x100 .8884 .0235 11% 98% 1.552 .9960 .9113 .9029 0.965 r(β, γ)=-0.75 80x100 .9169 .0235 13% 86% 1.537 .9955 .9359 .9071 0.931 r(β, γ)=+0.25 80x100 .8712 .0228 19% 96% 1.556 .9957 .8997 .9116 0.971 r(β, γ)=+0.75 80x100 .8541 .0265 45% 96% 1.545 .9953 .8872 .9103 0.964 Panel B: Simulations: δ ∼ N(0.99, 0.22), β ∼ N(0.9, 0.22), γ ∼ N(1.6, σ2) Uncertainty Structure: σ ∈ {0, 0.05, 0.1, 0.2}, Unrealized at Initial Allocation med N βb s(βb) 0.9 ∈/ CI(β) 1 ∈/ CI(β) γb bδ βbi βbi r(βi, βbi) σ = 0 80x100 .8800 .0202 13% 94% 1.601 .9957 .9044 .9017 0.995 σ = 0.05 80x100 .9001 .0287 7% 92% 1.608 .9949 .9336 .9122 0.824 σ = 0.1 80x100 .9593 .0369 26% 17% 1.632 .9952 1.022 .9539 0.590 σ = 0.2 80x100 1.186 .0823 98% 58% 1.736 .9957 1.367 1.164 0.325

12 C Discussion of Potential Confounds

Our effort discounting data address several key confounds present in monetary studies, such as fungibility and arbitrage issues. In this appendix section we address whether we can attribute the observed behavior for effort choices to dynamic inconsistency. Foremost, the ability to predict commitment demand based on present-biased allocations gives a degree of confidence that present-biased allocations are driven by dynamic inconsistency. In the following, we discuss four additional hypotheses that can generate time inconsistent effort allocations. These are (unanticipated) permanent shocks to the cost function of performing the tasks, unanticipated shocks to the cost function in Week 2, general uncertainty in cost functions, and simple mistakes. Though none of these explanations would predict a correlation between time inconsistency and commitment demand, we can also address these hypotheses directly. First, subjects may make present-biased allocations in Week 2 not because they are present- biased, but because their cost function for the tasks changed permanently. Maybe upon returning to the tasks they find them to be more or less difficult than they previously envisioned. For example, this could be because they have an injury that makes typing harder, have a bigger and better (or smaller and worse) screen at home than in the lab, which makes the tasks less (more) onerous, etc.70 Though we do attempt to give subjects a sense of the tasks, this is a plausible and critical confound. Our environment is able to address this confound as changes to perceived cost functions are separable from time preferences. The shape of the cost function is identified from changes in the value of R. Because both initial allocations and subsequent allocations are made at various interest rates, the cost function is identified at multiple points and time. In Appendix Table A11, we estimate cost functions and discounting parameters at each point in time. We do not find evidence that cost functions change over time.71 This lends

70We see this channel as distinct from the role of uncertainty, as such changes in difficulty need not have been forecasted. 71The analysis of Appendix Table A11 can be conducted separately for committing and non-committing subjects to examine if those individuals identified to be dynamically inconsistent in their commitment choice have varying cost functions or varying discounting parameters over time. For committing subjects the weekly discount factor measured in Week 1 is 1.082 (s.e. = 0.051), while the weekly discount factor measured in Week 2 is 0.900 (0.037). This difference is significant at the 1% level, χ2(1) = 6.38 (p = 0.01). For committing subjects

13 credence to the notion that changes in cost functions are not driving the observed behavior.72 Second, subjects may reallocate fewer tasks to the present due to an unforeseen, local shock that resulted in an increase in the cost function in Week 2 only. This could be because the subject is unusually busy in Week 2 because of a surprise exam, or finds himself unusually exhausted and hence unusually irritated with the length of work to be done. There are several ways to address this concern. First, a simple way in which subjects may find it unusually difficult to complete the work in Week 2 is if they log on to the experimental website so late, just prior to midnight, that they have only a very limited opportunity to complete their tasks. We can check for this hypothesis because we recorded the time at which subjects made their allocations. The median subject completed their allocations in Week 2 with 10.3 hours remaining before the imposed midnight deadline. Only 4 of 80 subjects completed their allocations in Week 2 with less than 2 hours remaining before the imposed midnight deadline and 0 of 80 completed their allocations with less than 1 hour remaining. We therefore do not find evidence that a physical time constraint is a driving force in the allocations. However, subjects logging on later may indeed be those who experienced an unanticipated shock in costs (even if their timing does not entail a physical constraint). We therefore examine whether subjects who log on to our experimental website later in the evening of their Week 2 work date exhibit more present bias. Individuals who log on with less than 4 hours before midnight (20 percent of our sample) are no more present-biased and have virtually identical allocation behavior as others.73 the cost function parameter measured in Week 1 is γ = 1.739 (0.184), and in Week 2 is γ = 1.519 (0.121). This difference is not significant at conventional levels, χ2(1) = 2.53 (p = 0.11). This indicates that for subjects separately identified as present-biased through their commitment choice, changing behavior through time is more clearly linked to changing discounting parameters and not changing cost functions. No differences in either discounting or cost functions are observed between Weeks 1 and 2 for non-committing subjects. 72Note that if cost functions would change over time, and this were the unique driver for changes in allocations between Week 1 and Week 2, we would observe a specific pattern of allocations. If an individual moved from having an almost linear cost function to a very convex one, the corresponding allocations would shift from being very price sensitive to limited price sensitivity. When initial allocations asked for lot of work to be done in Week 2, we would indeed see a change that amounts to a reduction of work in Week 2. However, for allocations that asked for little work in Week 2, we would see an increase in work to be done in Week 2. This is not what we observe. The data show a universal reduction of work to be allocated in Week 2. 73Subjects logging on with more than 4 hours before midnight allocate an average of 23.80 (s.d = 15.91) tasks to the sooner work date in Week 2, while subjects logging on with less than 4 hours allocate 25.43 (14.06). Even

14 As a final way to assess whether some subjects may have had unusual shocks to their cost function (and whether these are subjects that generate our results of present-biased allocations), we can find a proxy for the costs of the tasks in Week 2. Specifically, we examine the amount of time it takes subjects to complete their minimum work in Week 2. Minimum work took the median subject around 18 minutes to complete. Those subjects who take longer than 25.7 minutes (20 percent of our sample) are no more present-biased and have virtually identical allocation behavior as others.74 Naturally, these analyses may not give a fully satisfactory response to the potential confound presented by forecasting error and boredom. If indeed such a possibility is the source of our present-biased data patterns, a final question is whether or not such a hypothesis delivers the observed correlation between present-bias and commitment demand. We believe the answer to this question to be no. A third class of explanations which can generate a pattern of present-biased behavior in the absence of time inconsistency concerns uncertainty in cost functions. When making initial allocations, subjects do so under a different informational environment than when making their subsequent allocations. There could be uncertainty for initial allocations, which is partially resolved when allocations are again made one week later. Several aspects of uncertainty warrant attention. First, individuals may carry preferences for the resolution of uncertainty (Kreps and Porteus, 1978; Epstein and Zin, 1989; Chew and Epstein, 1989). Unlike monetary designs, in our effort experiment such a preference may more naturally lead to a future bias. without accounting for multiple observations this difference is not significant, t(798) = 1.19, p = 0.24. Subjects logging on with more than 4 hours before midnight have budget share differences between Weeks 1 and 2 of -0.049 (0.21), indicating they allocate around 5 percent less of the budget of tasks to the sooner work date in Week 2 than they allocated in Week 1. Subjects logging on with less than 4 hours have budget share differences between Weeks 1 and 2 of -0.052 (0.20). Even without accounting for multiple observations this difference is not significant, t(798) = 0.15, p = 0.88. Note however, that in general, subjects that log in later may be more present-biased, as they do everything a little later. And indeed, if we instead cut at the median log-in time, 10.3 hours before midnight, marginally significant differences are observed indicating that present-biased individuals may be logging in later. However, such individuals do not appear to be those particularly close to the deadline. 74Subjects taking less than 25.7 minutes allocate an average of 24.11 (s.d = 15.43) tasks to the sooner work date in Week 2, while subjects taking more than 25.7 minutes allocate 24.15 (16.11). Even without accounting for multiple observations this difference is not significant, t(798) = 0.03, p = 0.98. Subjects taking less than 25.7 minutes have budget share differences between Weeks 1 and 2 of -0.049 (0.20), indicating they allocate around 5 percent less of the budget of tasks to the sooner work date in Week 2 than they allocated in Week 1.Subjects taking more than 30 minutes have budget share differences between Weeks 1 and 2 of -0.053 (0.22). Even without accounting for multiple observations this difference is not significant, t(798) = 0.23, p = 0.82.

15 Subjects desiring to resolve uncertainty in their subsequent allocation choices could, in prin- ciple, choose to complete their tasks immediately when the present is available. Second, our discounting estimates do not account for subjects’ potential uncertainty on their own parameters, such as uncertainty with regards to the future costliness of the task. Though the weekly parameter estimates provided in Table A11 help to alleviate some concerns, a deeper problem may exist. Subjects may make allocations in Week 1 that minimize their discounted expected cost in future weeks given the potential realizations of future parameters. This uncertainty may be subsequently resolved in Week 2, such that subjects minimize their discounted cost at specific realizations of key parameters. As the minimizer of the expectation need not be the expectation of the minimizer, such issues can lead to inconsistencies between initial allocations and subsequent allocations. To explore the extent to which this issue hampers identification of present bias, we conduct simulations under a variety of uncertainty structures in Appendix B. Uncertainty, unresolved at initial allocation and realized at the time of the subsequent allocation, does bias our estimates of β both at the aggregate and individual level. However, the direction of bias is generally upward in the parameter regions of interest, leading to less estimated present bias.75 Importantly, a subject with future uncertainty would benefit from flexibility, such that even if present bias was delivered by uncertainty of some form one would not expect a correlation between present bias and commitment demand. Fifth, present-biased allocations of effort may be a simple decision error. Hence, present bias, or any dynamic inconsistency, may be an unstable phenomenon. The two blocks of our experiment speak to this possibility. Subjects have two opportunities to exhibit present-biased allocations. Indeed, present-biased behavior in Block 1 and Block 2 is significantly correlated.76 At the allocation level, a subject who is present-biased in Block 1 is 58% more likely than others

75 Intuitively, subjects with unresolved uncertainty on future parameters seek to avoid the extreme possibilities of working under a very convex cost structure that is only rarely realized. This leads initial allocations to be frequently lower than subsequent allocations, particularly at higher interest rates. AppendixB provides greater detail. 76 Though the behavior is significantly correlated when examined as indicators for present bias, future bias and dynamic consistency; the budget share differences are not significantly correlated through time. This may be due to the sheer volume of data with budget share differences equal to zero and the relative lack of stability for future-biased behavior.

16 to be present-biased in Block 2, F (1, 79) = 6.94, (p = 0.010).77 Additionally, an individual who is dynamically consistent in Block 1 is 85% more likely to be dynamically consistent in Block 2 than others F (1, 79) = 50.88, (p < 0.01).78 This discussion helps to clarify some of the potential confounds for our observed effects. We view it as unlikely that present-biased allocations of effort are driven by unanticipated permanent or temporary shocks, uncertainty, or decision error. Further, that present bias over effort exhibits stability and predicts commitment demand gives confidence that our observed effects are generated by dynamic inconsistency.

D Commitment Value and Sophistication

We analyze the relationship between commitment valuations and sophistication by calculating

t−s s∗ γ k s∗ γ ∗ γ k ∗ γ V = (1 − 2p) · βδ ·{[(et,t + ω) + δ c(et+k,t + ω) ] − [(et,s + ω) + δ c(et+k,s + ω) ]}. at the estimated parameter values from Table2, column (3) of γ = 1.6, δ = 1, and β = 0.9 under various assumptions for the value of βb. Diﬀering values of βb deliver diﬀerent forecasted

s∗ s∗ allocations (et,t, et+k,t) and hence diﬀerent values of V . As βb diverges from 1, forecasted allocations diﬀer more dramatically from initial allocations and the value of commitment grows. For each value of V we calculate the equivalent number of tasks as

T γ = V. 77Test statistic from OLS regression of binary indicator for a present-biased allocation in Block 2 on matched indicator for present-biased allocation in Block 1 with standard errors clustered on the subject level. The estimated constant is 0.218 (s.e. = 0.030) and the coefficient on Block 1 present bias is 0.128 (s.e. = 0.049). 78 Test statistic from OLS regression of binary indicator for a dynamically consistent allocation in Block 2 on matched indicator for a dynamically consistent allocation in Block 1 with standard errors clustered on the subject level. The estimated constant is 0.400 (s.e. = 0.041) and the coefficient on Block 1 dynamic consistency is 0.342 (s.e. = 0.048). Interestingly, somewhat less precision is found for future biased allocations. An individual who is future-biased in Block 1 is 54% more likely to be future-biased in Block 2 than others F (1, 79) = 3.07, (p = 0.08). Test statistic from OLS regression of binary indicator for a future-biased allocation in Block 2 on matched indicator for a future-biased allocation in Block 1 with standard errors clustered on the subject level. The estimated constant is 0.162 (s.e. = 0.025) and the coefficient on Block 1 future bias is 0.088 (s.e. = 0.050).

17 It is useful to go through the calculation similar to that in the main text, solving for T given a set of parameters. Consider a subject with parameter values γ = 1.6, δ = 1, β = .9, and ω = 10 (our maintained assumption), optimizing at R = 1 with m = 50 and the experimental

∗ ∗ implementation probability of p = .1. Optimization at time s yields et,s = et+k,s = 25. A

s∗ s∗ subject with βbe = .9 perceives that she will choose et,t = 21.9 and et+k,t = 28.1. V can then be calculated as

V = (0.8) · 0.9 ·{[(21.9 + 10)1.6 + (28.1 + 10)1.6] − [(25 + 10)1.6 + (25 + 10)1.6]} = T 1.6

Solving for T yields 1.32 tasks. This calculation does not take into account the fact that subjects make 10 allocations at time s. Hence, the value of commitment should be expressed as the expectation of V across these 10 allocations, 10 X 1 T γ = E[V ] = V . 10 i i=1 For simplicity, we ignore the fact that the elicitation of commitment demand entails a second stage price list randomization procedure. Using this calculation and the parameters above, it is possible to solve for the equivalent number of tasks, T , for a given value of βb. In Table A2, we calculate T for various values of βb at the parameter values noted above. For each T we also provide the monetary value of commitment at a wide range of per-task values, w. Note that as βb decreases, the value of commitment increases and that only in the extremes do commitment valuations exceed one or two dollars.

18 Table A2: Commitment Values, βb and Per Task Valuations Value of Commitment Given Diﬀerent βb Equiv. # of Tasks Monetary T w = $0.10 w = $0.20 w = $0.30 w = $0.40 w = $0.50 (˜$6/hour) (˜$12/hour) (˜$18/hour) (˜$24/hour) (˜$30/hour) βb = 1 0 $0.00 $0.00 $0.00 $0.00 $0.00 βb = .9 1.2 $0.12 $0.23 $0.34 $0.46 $0.57 βb = .8 2.9 $0.29 $0.59 $0.88 $1.18 $1.47 βb = .7 5.3 $0.53 $1.06 $1.59 $2.12 $2.64 βb = .6 8.2 $0.82 $1.64 $2.46 $3.28 $4.10 βb = .5 11.7 $1.17 $2.35 $3.52 $4.69 $5.87

19 E Additional Tables and Figures

E.1 Tabulations of Dynamic Consistency

The following tables provide tabulations of each experimental interest rate or task rate for money and eﬀort and two measures of dynamic inconsistency. First, the average budget share to either the sooner work date or the sooner payment is contrasted across allocation timing and, second, the proportion of subjects who are present-biased, dynamically consistent, and future-biased is provided. Separate tables are provided for the ﬁrst block of the experiment, the full set of experimental data and the replication exercise.

20 Table A3: Present Bias By Interest Rate

Panel A: Eﬀort Choices R Initial Subsequent t-test Proportion Proportion Proportion Budget Share Budget Share (p-value) Present-Biased Dynamically Consistent Future-Biased (1) (2) (3) (4) (5) (6) 0.5 0.787 0.761 1.76 0.294 0.444 0.263 (0.180) (0.219) (p=0.08) 0.75 0.717 0.690 1.70 0.356 0.363 0.281 (0.206) (0.245) (p=0.09) 1 0.541 0.489 3.65 0.237 0.656 0.106 (0.134) (0.183) (p<0.01) 1.25 0.324 0.250 4.12 0.388 0.444 0.169 (0.239) (0.222) (p<0.01) 1.5 0.289 0.222 3.67 0.369 0.425 0.206 (0.242) (0.226) (p<0.01) Overall 0.532 0.482 3.86 0.329 0.466 0.205 (0.286) (0.311) (p<0.01) Panel B: Money Choices P t 6= 0 t = 0 t-test Proportion Proportion Proportion Budget Share Budget Share (p-value) Present-Biased Dynamically Consistent Future-Biased (1) (2) (3) (4) (5) (6) 0.952 0.924 0.923 0.07 0.073 0.813 0.113 (0.228) (0.189) (p=0.94) 1 0.774 0.813 1.32 0.200 0.660 0.140 (0.368) (0.323) (p=0.19) 1.11 0.102 0.148 1.86 0.180 0.733 0.087 (0.259) (0.300) (p=0.06) 1.25 0.051 0.087 1.97 0.113 0.853 0.033 (0.177) (0.239) (p=0.05) 1.429 0.053 0.077 1.40 0.100 0.847 0.053 (0.182) (0.228) (p=0.16) Overall 0.381 0.410 1.87 0.133 0.781 0.085 (0.461) (0.458) (p=0.07)

Notes: Panel A tabulates initial and subsequent budget shares for sooner tasks for each R in eﬀort. Each row calculates from 160 initial allocations (one each for tetris and greek at each task rate) and 160 subsequent allocations. Paired t-tests with 159 degrees of freedom presented. Panel B tabulates t 6= 0 and t = 0 budget shares for sooner payments for each P in money. Each row calculates from 75 t 6= 0 allocations (one at each interest rate in the Week 4 vs. Week 7 prospective choices) and 150 t = 0 allocations (one at each interest rate in the Week 4 vs. Week 7 actual and Week 1 vs. Week 4) choices. Paired t-tests with 149 degrees of freedom presented. Overall tests in both panels come from regression of budget share on allocation timing with standard errors clustered on individual level. Test statistic is t-statistic testing the null hypothesis of no eﬀect of allocation timing, which controls for multiple comparisons.

21 Table A4: Present Bias By Interest Rate, Full Data Set

Panel A: Eﬀort Choices R Initial Subsequent t-test Proportion Proportion Proportion Budget Share Budget Share (p-value) Present-Biased Dynamically Consistent Future-Biased (1) (2) (3) (4) (5) (6) 0.5 0.796 0.768 2.95 0.291 0.494 0.216 (0.179) (0.207) (p<0.01) 0.75 0.729 0.694 3.11 0.375 0.384 0.241 (0.208) (0.240) (p<0.01) 1 0.533 0.494 3.87 0.231 0.653 0.116 (0.152) (0.181) (p<0.01) 1.25 0.293 0.260 2.74 0.297 0.509 0.194 (0.232) (0.235) (p<0.01) 1.5 0.244 0.222 1.81 0.278 0.525 0.197 (0.234) (0.231) (p=0.07) Overall 0.519 0.488 3.90 0.329 0.466 0.205 (0.301) (0.311) (p<0.01) Panel B: Money Choices P t 6= 0 t = 0 t-test Proportion Proportion Proportion Budget Share Budget Share (p-value) Present-Biased Dynamically Consistent Future-Biased (1) (2) (3) (4) (5) (6) 0.952 0.924 0.923 0.07 0.073 0.813 0.113 (0.228) (0.189) (p=0.94) 1 0.774 0.813 1.32 0.200 0.660 0.140 (0.368) (0.323) (p=0.19) 1.11 0.102 0.148 1.86 0.180 0.733 0.087 (0.259) (0.300) (p=0.06) 1.25 0.051 0.087 1.97 0.113 0.853 0.033 (0.177) (0.239) (p=0.05) 1.429 0.053 0.077 1.40 0.100 0.847 0.053 (0.182) (0.228) (p=0.16) Overall 0.381 0.410 1.87 0.133 0.781 0.085 (0.461) (0.458) (p=0.07)

Notes: Panel A tabulates initial and subsequent budget shares for sooner tasks for each R in eﬀort. Each row calculates from 320 initial allocations (one each for tetris and greek at each task rate in each round) and 320 subsequent allocations. Paired t-tests with 159 degrees of freedom presented. Panel B tabulates t 6= 0 and t = 0 budget shares for sooner payments for each P in money. Each row calculates from 75 t 6= 0 allocations (one at each interest rate in the Week 4 vs. Week 7 prospective choices) and 150 t = 0 allocations (one at each interest rate in the Week 4 vs. Week 7 actual and Week 1 vs. Week 4) choices. Paired t-tests with 149 degrees of freedom presented. Overall tests in both panels come from regression of budget share on allocation timing with standard errors clustered on individual level. Test statistic is t-statistic testing the null hypothesis of no eﬀect of allocation timing, which controls for multiple comparisons.

22 Table A5: Present Bias By Interest Rate, Replication Exercise

Panel A: Eﬀort Choices P Initial Subsequent t-test Proportion Proportion Proportion Budget Share Budget Share (p-value) Present-Biased Dynamically Consistent Future-Biased (1) (2) (3) (4) (5) (6) 0.666 0.337 0.318 1.09 0.400 0.305 0.295 (0.203) (0.232) (p=0.27) 0.8 0.385 0.361 1.25 0.421 0.242 0.337 (0.196) (0.223) (p=0.21) 0.909 0.432 0.405 1.21 0.453 0.189 0.358 (0.191) (0.219) (p=0.23) 0.952 0.458 0.418 1.96 0.474 0.221 0.305 (0.175) (0.220) (p=0.05) 1 0.522 0.507 0.58 0.337 0.358 0.305 (0.153) (0.211) (p=0.56) 1.053 0.581 0.554 1.18 0.463 0.200 0.337 (0.194) (0.235) (p=0.24) 1.111 0.618 0.566 2.37 0.474 0.211 0.316 (0.195) (0.239) (p=0.02) 1.25 0.658 0.603 2.29 0.400 0.253 0.347 (0.210) (0.255) (p=0.02) 1.538 0.727 0.667 2.38 0.400 0.305 0.295 (0.220) (0.280) (p=0.02) Overall 0.524 0.489 1.95 0.425 0.254 0.322 (0.229) (0.260) (p=0.05) Panel B: Money Choices P Initial Subsequent t-test Proportion Proportion Proportion Budget Share Budget Share (p-value) Present-Biased Dynamically Consistent Future-Biased (1) (2) (3) (4) (5) (6) 0.666 0.932 0.935 0.39 0.091 0.879 0.030 (0.174) (0.182) (p=0.70) 0.8 0.929 0.945 1.77 0.081 0.859 0.061 (0.172) (0.153) (p=0.07) 0.909 0.917 0.917 0.03 0.101 0.828 0.071 (0.186) (0.194) (p=0.97) 0.952 0.908 0.901 0.55 0.152 0.788 0.061 (0.190) (0.206) (p=0.58) 1 0.621 0.695 2.37 0.232 0.667 0.101 (0.336) (0.302) (p=0.02) 1.053 0.105 0.084 1.05 0.081 0.828 0.091 (0.240) (0.212) (p=0.30) 1.111 0.086 0.088 0.25 0.081 0.869 0.051 (0.207) (0.226) (p=0.81) 1.25 0.064 0.064 0.07 0.071 0.869 0.061 (0.162) (0.182) (p=0.95) 1.538 0.062 0.048 1.71 0.020 0.879 0.101 (0.171) (0.150) (p=0.09) Overall 0.514 0.520 0.92 0.101 0.829 0.070 (0.451) (0.456) (p=0.36)

Notes: Panel A tabulates initial and subsequent budget shares for sooner tasks for each P in eﬀort. Each row calculates from 95 initial allocations and 95 subsequent allocations. Paired t-tests with 94 degrees of freedom presented. Panel B tabulates initial and subsequent budget shares for sooner payments for P in money. Each row calculates from 99 initial allocations and 99 subsequent allocations choices. Paired t-tests with 98 degrees of freedom presented. Overall tests in both panels come from regression of budget share on allocation timing with standard errors clustered on individual level. Test statistic is t-statistic testing the null hypothesis of no eﬀect of allocation timing, which controls for multiple comparisons. 23 E.2 Individual Estimates

We contrast initial and subsequent allocations for work and for money within subjects for the 80 subjects in the primary study sample and the 75 subjects with complete monetary data. Estimates of present bias for each subject are also provided. Corresponding allocations and estimates also provided for between subjects replication study.

24 Table A6: Individual Estimates Subjects 1-45

Eﬀort Choices Monetary Choices

Subject # Mean Initial Mean Subsequent βe Mean Initial Mean Subsequent βm Budget Share Budget Share Budget Share Budget Share (1) (2) (3) (4) (5) (6) 1 .5 .5 1 .4 .3 1.046 2 .496 .476 .952 .527 .537 .999 3 .522 .516 .983 .4 .3 1.046 4 .534 .498 .86 .309 .354 .979 5 .516 .402 .731 .4 .4 1 6 .514 .512 .995 .4 .4 1 7 .514 .514 1 .4 .4 1 8 .576 .472 .742 . . . 9 .514 .5 .959 . . . 10 .5 .5 1 .4 .4 1 11 .5 .52 1.078 .4 .4 1 12 .5 .5 1 .4 .4 1 13 .628 .486 .664 .4 .4 1 14 .462 .512 1.143 .3 .456 .934 15 .542 .414 .683 .4 .4 1 16 .5 .5 1 .4 .4 1 17 .5 .4 .725 .4 .4 1 18 .468 .404 .796 .4 .3 1.046 19 .582 .296 .437 .4 .3 1.046 20 .504 .48 .884 .4 .4 1 21 .51 .424 .739 .4 .4 1 22 .482 .344 .665 .4 .4 1 23 .5 .51 1.012 .294 .376 .964 24 .832 .512 .426 .4 .415 .999 25∗ .56 .518 .815 .4 .3 1.046 26 .52 .404 .731 . . . 27 .648 .514 .698 .2 .407 .914 28 .642 .254 .319 .4 .513 .961 29 .588 .51 .724 .2 .3 .956 30 .552 .5 .851 .4 .4 1 31 .51 .498 .961 .3 .4 .956 32∗ .432 0 .249 .4 .4 1 33 .5 .5 1 .4 .4 1 34 .526 .478 .842 .4 .5 .957 35 .496 .4 .723 .2 .35 .935 36∗ .516 .522 1.045 .4 .7 .88 37 .666 .5 .638 .3 .35 .978 38 .792 .746 .879 .457 .5 .983 39 .514 .54 1.072 .51 .45 1.02 40 .516 .516 1 .4 .4 1 41 .536 .51 .945 .206 .428 .91 42 .6 .6 1 .4 .4 1 43 .514 .512 .995 .4 .666 .897 44 .522 .514 .976 .531 .532 1 45 .514 .51 .991 .4 .3 1.046

Notes: Tabulates initial and subsequent budget shares for both eﬀort and money and corresponding present bias estimates for subjects 1-45 of 89 non-attriting subjects and 84 subjects with complete monetary data. Nine subjects excluded from primary sample marked with ∗. Subject 25 provided no variation in response in Weeks 4 or 5. Subject 32 provided no variation in Week 2. Subject 36 provided no variation in Week 5.

25 Table A7: Individual Estimates Subjects 46-89

Eﬀort Choices Monetary Choices

Subject # Mean Initial Mean Subsequent βe Mean Initial Mean Subsequent βm Budget Share Budget Share Budget Share Budget Share (1) (2) (3) (4) (5) (6) 46 .512 .36 .654 .509 .743 .91 47∗ .34 .27 .818 .4 .3 1.046 48 .508 .518 1.024 .4 .4 1 49∗ .514 .516 1.006 .4 .4 1 50∗ 1 .844 .603 .4 .3 1.046 51 .59 .692 1.174 0 .2 .913 52 .548 .5 .979 .4 .4 1 53 .506 .38 .686 .403 .306 1.044 54 .5 .5 1 .4 .4 1 55 .514 .514 1.003 .4 .147 1.121 56 .354 .444 1.32 .201 .256 .976 57 .896 .502 .291 .4 .4 1 58 .512 .504 .983 .203 .405 .92 59 .474 .48 1.027 .4 .4 1 60 .24 .21 .876 .4 .4 1 61 .5 .6 1.379 .4 .4 1 62 .518 .5 1.086 .4 .8 .842 63 .57 .5 .788 .4 .45 .979 64 .55 .6 1.174 .2 .4 .914 65 .5 .4 .725 .2 .3 .956 66 .604 .524 .788 .4 .4 1 67 .42 .4 .957 . . . 68 .546 .552 1.019 0 .471 .814 69∗ .4 0 .238 .4 .4 1 70 .47 .5 1.132 .6 .4 1.092 71 .46 .476 1.092 .6 .5 1.045 72∗ .22 0 .429 .539 .851 .873 73 .512 .516 1.005 .2 .198 1.001 74 .5 .5 1 .4 .4 1 75 .44 .51 1.203 .4 .35 1.023 76 .544 .506 .802 1 .5 1.236 77 .594 .626 1.114 .2 .2 1 78 .52 .5 .964 .4 .4 1 79 .53 .518 .973 .52 .5 1.007 80 .72 .056 .14 .9 .959 .973 81 .542 .508 .912 .499 .409 1.041 82∗ .754 1 2.629 .2 .1 1.047 83 .5 .49 .973 . . . 84 .512 .5 1.107 .4 .4 1 85 .504 .538 1.091 .4 .4 1 86 .46 .504 1.12 0 .303 .87 87 .514 .458 .935 .4 .4 1 88 .51 .51 1.003 .4 .4 1 89 .5 .5 1 .4 .5 .957

Notes: Tabulates initial and subsequent budget shares for both eﬀort and money and corresponding present bias estimates for subjects 51-89 of 89 non-attriting subjects and 84 subjects with complete monetary data. Nine subjects excluded from primary sample marked with ∗. Subject 47 provided no variation in response in Week 5. Subject 49 provided no variation in Week 5. Subject 50 provided no variation in Week 1. Subject 69 provided no variation in Week 2. Subject 72 provided no variation in Weeks 2, 4 or 5. Subject 82 provided no variation in Weeks 2 or 5.

26 Table A8: Replication Study Individual Estimates Subjects 1-50

Eﬀort Choices Monetary Choices

Subject # Mean Initial Mean Subsequent βe Subject # Mean Initial Mean Subsequent βm Budget Share Budget Share Budget Share Budget Share (1) (2) (3) (4) (5) (6) 1 .444 .556 1.502 1 .499 .555 .965 2 .528 .774 2.126 2 .444 .444 1 3 .556 .552 .958 3 .55 .495 1.036 4 .463 .5 1.12 4 .492 .532 .978 5∗ .072 0 .753 5 .522 .527 .997 6∗ .578 .258 .369 6 .555 .555 1 7 .575 .556 .888 7 .555 .666 .932 8 .5 .5 1 8 .497 .502 .997 9∗ .258 .606 2.835 9 .555 .555 1 10 .509 .501 .98 10 .499 .499 1 11 .542 .423 .703 11 .489 .499 .991 12 .484 .523 1.129 12 .555 .555 1 13 .427 .25 .586 13 .555 .555 1 14 .502 .502 1 14 .555 .555 1 15 .579 .593 1.043 15 .301 .444 .909 16 .556 .444 .666 16 .499 .499 1 17 .693 .553 .647 17 .555 .555 1 18∗ .499 .504 1.002 18 .499 .499 1 19 .665 .603 .84 19 .507 .438 1.044 20 .509 .362 .644 20∗ .497 .499 .999 21 .526 .31 .527 21 .345 .363 .989 22 .558 .441 .674 22 .555 .555 1 23 .494 .487 .976 23 .5 .501 1 24 .423 .475 1.144 24 .499 .499 1 25 .508 .53 1.076 25 .499 .555 .965 26 .502 .506 1.022 26 .499 .499 1 27 .402 .39 .963 27 .499 .499 1 28 .5 .5 1 28 .499 .499 1 29 .529 .553 1.072 29 .555 .418 1.097 30 .283 .486 1.824 30 .499 .501 .999 31 .551 .486 .827 31 .444 .555 .931 32∗ .498 .83 2.87 32 .499 .488 1.006 33∗ .509 1 6.087 33 .499 .555 .965 34 .598 .621 1.103 34 .444 .555 .931 35 .518 .505 .959 35 .499 .499 1 36 .527 .407 .727 36 .499 .499 1 37 .604 .659 1.181 37 .555 .499 1.036 38 .814 .798 .906 38 .555 .555 1 39 .509 .52 1.04 39 .499 .499 1 40 .528 .537 .98 40 .555 .555 1 41 .507 .506 .999 41 .444 .555 .931 42∗ .52 0 .153 42 .499 .555 .965 43∗ .747 0 .071 43 .494 .555 .96 44 .561 .504 .818 44 .499 .499 1 45∗ .68 0 .091 45 .555 .555 1 46 .517 .598 1.422 46 .666 .452 1.148 47 .486 .518 1.157 47 .499 .499 1 48 .5 .502 1.005 48 .499 .499 1 49 .568 .742 1.776 49 .444 .444 1 50 .488 .51 1.064 50 .499 .555 .965

Notes: Tabulates initial and subsequent budget shares and corresponding present bias estimates for both effort and money for between subjects data for first 50 of 99 money subjects and first 50 of 94 effort subjects

27 Table A9: Replication Study Individual Estimates Subjects 51-99

Eﬀort Choices Monetary Choices

Subject # Mean Initial Mean Subsequent βe Subject # Mean Initial Mean Subsequent βm Budget Share Budget Share Budget Share Budget Share (1) (2) (3) (4) (5) (6) 51 .505 .741 2.438 51 .503 .481 1.013 52 .493 .493 1 52 .465 .536 .961 53 .632 .591 .863 53 .555 .555 1 54 .5 .502 1 54 .555 .555 1 55 .411 .333 .805 55 .493 .534 .977 56 .588 .542 .866 56∗ .699 .337 1.199 57 .296 .334 1.112 57 .555 .555 1 58 .538 .526 .924 58 .555 .555 1 59 .574 .514 .826 59 .555 .555 1 60 .502 .499 .985 60 .555 .555 1 61 .574 .586 1.027 61 .555 .499 1.036 62 .685 .654 .831 62 .555 .555 1 63 .794 .528 .369 63 .555 .555 1 64 .448 .46 1.043 64 .499 .499 1 65 .554 .53 .892 65 .555 .555 1 66 .493 .491 .992 66 .444 .499 .965 67 .5 .498 1 67 .555 .555 1 68 .505 .516 1.03 68 .562 .617 .971 69 .574 .632 1.188 69 .499 .499 1 70 .433 .348 .777 70 .555 .545 1.009 71 .489 .499 1.028 71 .444 .555 .931 72 .595 .551 .885 72 .476 .515 .973 73 .496 .516 1.053 73 .555 .555 1 74 .512 .176 .353 74 .72 .749 .976 75 .462 .476 .994 75 .444 .555 .931 76 .49 .603 1.39 76 .499 .555 .965 77∗ .67 .67 1.011 77 .444 .444 1 78 .572 .553 .983 78 .555 .555 1 79 .755 .273 .231 79 .555 .555 1 80 .503 .512 1.016 80 .499 .499 1 81∗ .283 0 .308 81 .777 .666 1.073 82 .477 .503 1.076 82 .503 .512 .995 83 .634 .341 .424 83 .499 .499 1 84∗ .344 .501 1.577 84 .384 .477 .953 85∗ .667 .667 1 85 .555 .555 1 86 .517 .509 .982 86 .499 .499 1 87 .497 .48 .955 87 .444 .499 .965 88∗ .516 .597 1.321 88 .444 .499 .965 89 .505 .504 .995 89 .555 .555 1 90 .528 .539 1.033 90 .533 .533 1.006 91 .503 .401 .753 91 .289 .281 1.005 92 .641 .506 .677 92 .444 .555 .931 93 .514 .495 .951 93 .555 .555 1 94 .571 .518 .862 94 .499 .499 1 95 .581 .243 .364 95 .499 .499 1 96 .472 .383 1.048 97 .499 .499 1 98 .444 .444 1 99 .666 .555 1.073

Notes: Tabulates initial and subsequent budget shares and corresponding present bias estimates for both effort and money for between subjects data for subjects 51-99 of 99 money subjects and first 51-94 of 94 effort subjects

28 E.3 Estimates Including Nine Subjects With Limited Eﬀort Alloca-

tion Variation

We re-conduct the primary aggregate analysis including 9 subjects with limited variation in their eﬀort allocation choices.

Table A10: Parameter Estimates Including 9 Additional Subjects

Monetary Discounting Eﬀort Discounting (1) (2) (3) (4) (5) All Delay Three Week Delay Job 1 Job 2 Combined Lengths Lengths Greek Tetris Present Bias Parameter: β 0.975 0.988 0.870 0.848 0.858 (0.008) (0.008) (0.045) (0.042) (0.040) Weekly Discount Factor: (δ)7 0.988 0.979 0.996 1.014 1.002 (0.002) (0.003) (0.034) (0.035) (0.032) Monetary Curvature Parameter: α 0.976 0.977 (0.006) (0.005) Cost of Eﬀort Parameter: γ 1.666 1.580 1.621 (0.122) (0.101) (0.109)

# Observations 1680 1260 890 890 1780 # Clusters 84 84 89 89 89 Job Eﬀects Yes

H0 : β = 1 χ2(1) = 9.09 χ2(1) = 2.12 χ2(1) = 8.41 χ2(1) = 13.39 χ2(1) = 12.23 (p < 0.01) (p = 0.15) (p < 0.01) (p < 0.01) (p < 0.01)

H0 : β(Col. 1) = β(Col. 5) χ2(1) = 11.45 (p < 0.01)

H0 : β(Col. 2) = β(Col. 5) χ2(1) = 13.79 (p < 0.01)

29 E.4 Estimates For Eﬀort Discounting By Week

We re-estimate parameters by week and test the null hypothesis of equality of discount rates identiﬁed from initial allocations and subsequent allocations.

Table A11: Parameter Estimates By Week

Eﬀort Discounting (1) (2) (3) (4) Week 1 Week 2 Week 4 Week 5 Initial Allocations Subsequent Allocations Initial Allocations Subsequent Allocations

Weekly Discount Factor: (δ)7 0.998 0.898 0.939 0.892 (0.029) (0.025) (0.019) (0.024) Cost of Eﬀort Parameter: γ 1.668 1.521 1.463 1.528 (0.126) (0.097) (0.074) (0.092)

# Observations 800 800 800 800 # Clusters 80 80 80 80 Job Eﬀects Yes Yes Yes Yes

7 7 H0 :(δ) (Col. 1) = (δ) (Col. 2) χ2(1) = 7.02 (p < 0.01) 7 7 H0 :(δ) (Col. 3) = (δ) (Col. 4) χ2(1) = 4.10 (p = 0.04)

Notes: Parameters identified from two-limit Tobit regressions of equation (6) assuming β = 1 for effort discounting. Parameters recovered via non-linear combinations of regression coefficients. Standard errors clustered at individual level reported in parentheses, recovered via the delta method. Effort regressions control for Job Effects (Task 1 vs. Task 2). Tested null hypotheses 7 7 are equal discounting in Weeks 1 vs. 2 and Weeks 4 and 5, H0 :(δ) (Col. 1) = (δ) (Col. 2) and 7 7 H0 :(δ) (Col. 3) = (δ) (Col. 4).

30 E.5 Full Eﬀort Data Set Tables Figures

We reconduct all analyses using Block 1 and Block 2 data to identify eﬀort discounting parameters.

Table A12: Parameter Estimates: Full Eﬀort Data Set

Monetary Discounting Eﬀort Discounting (1) (2) (3) (4) (5) All Delay Three Week Delay Job 1 Job 2 Combined Lengths Lengths Greek Tetris Present Bias Parameter: β 0.974 0.988 0.927 0.927 0.927 (0.009) (0.009) (0.022) (0.021) (0.020) Weekly Discount Factor: (δ)7 0.988 0.980 0.978 0.979 0.977 (0.003) (0.003) (0.022) (0.022) (0.021) Monetary Curvature Parameter: α 0.975 0.976 (0.006) (0.005) Cost of Eﬀort Parameter: γ 1.566 1.510 1.537 (0.090) (0.081) (0.084)

# Observations 1500 1125 1600 1600 3200 # Clusters 75 75 80 80 80 Block Eﬀects Yes Yes Yes Job Eﬀects Yes

H0 : β = 1 χ2(1) = 8.77 χ2(1) = 1.96 χ2(1) = 11.1 χ2(1) = 11.9 χ2(1) = 13.94 (p < 0.01) (p = 0.16) (p < 0.01) (p < 0.01) (p < 0.01)

H0 : β(Col. 1) = β(Col. 5) χ2(1) = 5.46 (p = 0.02)

H0 : β(Col. 2) = β(Col. 5) χ2(1) = 8.61 (p < 0.01)

Notes: Parameters identified from two-limit Tobit regressions of equations (4) and (6) for monetary discounting and effort discounting, respectively. Parameters recovered via non-linear combinations of regression coefficients. Standard errors clustered at individual level reported in parentheses, recovered via the delta method. Effort regressions control for Block Effects (Weeks 1,2,3 vs. 4,5,6) and Job Effects (Task 1 vs. Task 2). Tested null hypotheses are zero present bias, H0 : β = 1, and equality of present bias across effort and money, H0 : β(Col. 1) = β(Col. 5) and H0 : β(Col. 2) = β(Col. 5).

31 Table A13: Monetary and Real Eﬀort Discounting by Commitment: Full Eﬀort Data Set

Monetary Discounting Eﬀort Discounting Commit (=0) Commit (=1) Commit (=0) Commit (=1) (1) (2) (3) (4) Tobit Tobit Tobit Tobit Present Bias Parameter: β 0.999 0.981 0.989 0.880 (0.010) (0.013) (0.018) (0.031) Weekly Discount Factor: (δ)7 0.978 0.981 0.911 1.032 (0.003) (0.005) (0.030) (0.029) Monetary Curvature Parameter: α 0.981 0.973 (0.009) (0.007) Cost of Eﬀort Parameter: γ 1.485 1.579 (0.123) (0.116)

# Observations # Clusters 28 47 33 47 Block Eﬀects Yes Yes Job Eﬀects Yes Yes

H0 : β = 1 χ2(1) = 0.01 χ2(1) = 2.15 χ2(1) = 0.34 χ2(1) = 15.12 (p = 0.94) (p = 0.14) (p = 0.56) (p < 0.01)

H0 : β(Col. 1) = β(Col. 2) χ2(1) = 1.29 (p = 0.26)

H0 : β(Col. 3) = β(Col. 4) χ2(1) = 9.35 (p < 0.01)

Notes: Parameters identified from two-limit Tobit regressions of equations (4) and (6) for monetary discounting and real effort discounting. Parameters recovered via non-linear combinations of regression coefficients. Standard errors clustered at individual level reported in parentheses, recovered via the delta method. Commit (=1) or Commit (=0) separates individuals into those who did (1) or those who did not (0) choose to commit at a commitment price of zero dollars. Effort regressions control for Block Effects (Weeks 1,2,3 vs. 4,5,6) and Job Effects (Job 1 vs. Job 2). Tested null hypotheses are zero present bias, H0 : β = 1, and equality of present bias across commitment and no commitment, H0 : β(Col. 1) = β(Col. 2) and H0 : β(Col. 3) = β(Col. 4).

32 E.6 Replication Exercise Additional Tables

Table A14: Replication Exercise Parameter Estimates, Restricted Sample

Monetary Discounting Eﬀort Discounting Greek (1) (2)) Present Bias Parameter: β 0.995 0.934 (0.004) (0.035) Weekly Discount Factor: (δ)7 0.989 1.077 (0.005) (0.032) Monetary Curvature Parameter: α 0.955 (0.009) Cost of Eﬀort Parameter: γ 1.733 (0.169)

# Observations 1746 1458 # Clusters 97 81

2 2 H0 : β = 1 χ (1) = 1.13 χ (1) = 3.60 (p = 0.29) (p = 0.06) 2 H0 : β(Col. 1) = β(Col. 2) χ (1) = 3.07 (p = 0.08)

Notes: Parameters identified from two-limit Tobit regressions of equations (4) and (6) for monetary discounting and effort discounting, respectively. Parameters recovered via non-linear combinations of regression coefficients. Standard errors clustered at individual level reported in parentheses, recovered via the delta method. Chi- squared tests used in last two rows. Sample restricted to those individuals with positive variation in experimental response in both weeks of replication exercise. Excluded subjects noted with ∗ in Appendix Tables A8 and A9

33 Figure A1: Real Eﬀort Discounting Behavior: Full Eﬀort Data Set

Greek Transcription Tetris 40 30 20 Tasks to Date Early Allocated 10 .5 1 1.5 .5 1 1.5 R

(from et+Ret+k=50) Initial Allocation Subsequent Allocation Mean Mean SE

Gr aphsbytk

Figure A2: Individual Estimates of Present Bias: Full Eﬀort Data Set .6 .4 Fraction .2 0 .5 .75 1 1.25 1.5 Work Present Bias .6 .4 Fraction .2 0 .5 .75 1 1.25 1.5 Monetary Present Bias 1.2 1.1 1 .9 .8

Monetary Present Bias Present Monetary .5 .75 1 1.25 1.5 Work Present Bias

34 Figure A3: Commitment Demand: Full Eﬀort Data Set

Panel A: Commit (=0)

Greek Transcription Tetris 40 30 20 Tasks to Date Early Allocated 10

.5 1 1.5 .5 1 1.5 R

(from et+Ret+k=50) Initial Allocation Subsequent Allocation Mean Mean SE

Gr aphsbytk

Panel B: Commit (=1)

Greek Transcription Tetris 40 30 20 Tasks to Date Early Allocated 10

.5 1 1.5 .5 1 1.5 R

(from et+Ret+k=50) Initial Allocation Subsequent Allocation Mean Mean SE

Gr aphsbytk

35 F Instructions

F.1 Week 1 Eﬀort Instructions Welcome: Thank you for participating in our experiment. We will begin shortly.

Eligibility for this study: To be in this study, you need to meet these criteria: You must be willing to participate for six consecutive weeks. Participation will require your presence on six consecutive Thursdays for at least 10 minutes per week for an average of 60 minutes. Weeks 1 (today) and 4 will occur in the xlab. Weeks 2,3,5, and 6 will occur at any computer that has access to the Internet. You must be willing to receive your payment from this experiment as one single completion payment at the end of the study. Payments will be made one week after the ﬁnal session, on Thursday, March 22. You will return to the xlab to receive this payment. If you do not meet these criteria, please inform us of this now.

Informed Consent Placed in front of you is an informed consent form to protect your rights as a subject. Please read it. If you would like to choose not to participate in the study you are free to leave at this point. If you have any questions, we can address those now. We will pick up the forms after the main points of the study are discussed.

Anonymity Your anonymity in this study is assured. Your name will never be recorded or connected to any decision you make here today. Your email will be collected in order to send reminder emails. After the study, your email information will be destroyed and will not be connected to your responses in the experiment.

1 Rules Please turn your cell phones off. If you have a question at any point, just raise your hand. Please put away any books, papers, computers, etc. There will be a quiz once we have finished with the instructions. If it is clear that you do not understand the instructions when we review your answers, you will be emailed and removed from the study. Your Earnings If you complete all six weeks of participation, a completion payment of $100 will be provided. You may receive additional earnings during the experiment. If you choose to end your participation before the six weeks are complete, please report this to study administrators, and you will receive a minimum payment of $10. All payments will be made one week after the final session, on Thursday March 22. You will return to the xlab to receive this payment.

Jobs In this study there are two types of jobs, Job 1 and Job 2. These jobs will be completed over time. Some portion of the jobs may be completed sooner, and some portion of the jobs may be completed later depending on your choices and chance. Importantly, some tasks for each job must be completed in each week. That is, as mentioned before, your participation is required in each of the six consecutive weeks of the study.

Job 1: In Job 1 you are asked to transcribe letters from a greek text. Greek text will appear in the Transcription Box on your screen. For each letter you will need to ﬁnd and select the corresponding letter and enter it into the Completion Box on your screen. One task is one row of greek text. For the task to be complete your accuracy must be 80% or better.

2 Job 2: In Job 2 you are asked to play a tetris game. Blocks of different shapes drop from the top of the task screen into a box. Each block is made up of four small squares arranged to make a larger square, an L-shape or a column. As the blocks fall they can be rotated (by pressing the up arrow key), moved horizontally (by pressing the left and right arrow keys), or moved down more quickly (by pressing the down arrow key). Your goal is to fill a entire horizontal line with parts of the blocks. When a horizontal line is filled, that line is ”destroyed,” moving the rest of the placed pieces down by one square. If a line remains incomplete, another line must be finished above it. The more lines that stand incomplete, the higher the blocks above them stack, reducing the space in which falling shapes can be manipulated. Eventually the blocks reach the top of the screen and the game ends. One task will be 4 lines of blocks completed. If a game ends before a task is complete, the completed lines will be counted in the subsequent game.

Practice: We will now spend a few minutes practicing both jobs on the computer. Before we continue, you will be asked to register using your email by clicking ”register” once you open the experiment. Make sure that you enter a valid email address.

3 The Experiment Timeline Now that you’ve tried Job 1 and Job 2, let’s consider the timeline of the study. Along the way we will discuss a few important details of how the study works.

Note: Minimum Work for each week In each week (including today), you are required to complete a minimum number of tasks of both Job 1 and Job 2.

Today (Week 1): Once your minimum work is complete, you will be asked to make a series of 5 decisions for each job. In these decisions you are asked to allocate tasks between one week from today (Week 2) and two weeks from today (Week 3). You will make 5 decisions for both Job 1 and Job 2. In each decision you are free to allocate your tasks as you choose. Note that this allocation decision does not include the minimum work for each week, which you must also complete. You will choose by moving a slider to your desired allocation. We will now practice on the computer.

4 Task Rates: In the example decision above every task you allocate to Week 2 reduces the number of tasks allocated to Week 3 by one. This is what we will refer to as a 1:1 task rate. The task rate will vary across your decisions. For example, the task rate may be 1:1.5, such that every task you allocate to Week 2 reduces the number of tasks allocated to Week 3 by 1.5. Or, the task rate may be 1:0.5, such that every task you allocate to Week 2 reduces the number of tasks allocated to Week 3 by 0.5. For simplicity, the task rates will always be represented as 1:X, and you will be fully informed of the value of X when making your decisions. Please practice with the diﬀerent allocations using the computer.

5 Week 2 (One Week From Today): Week 2, one week from today, will occur online. You will receive an email with instructions on how to access the website with the jobs. You will again complete your minimum work. You will be asked again to make 5 allocation decisions for each job. Exactly one of your 20 total allocation decisions will be implemented. That is, we will implement one decision from Week 1 for Job 1, or one decision from Week 2 for Job 1, or one decision from Week 1 for Job 2, or one decision from Week 2 for Job 2. We will discuss how this allocation decision is chosen shortly. We refer to this allocation decision as the ”decision-that-counts.” The tasks you allocated to Week 2 in the decision- that-counts must be completed. If you do not return or do not complete the tasks in Week 2, you cannot complete the study, and you will receive only the minimum payment of $10. In order for your tasks in Week 2 to be counted, they must be submitted by midnight on February 16th, 2012.

6 Week 3, Two Weeks From Today: Week 3, two weeks from today, will occur online. You will receive an email with instructions on how to access the website with the jobs. You will again complete your minimum work. Then, you must complete the tasks you allocated in the decision-that-counts. If you do not return or do not complete the tasks in Week 3, you cannot complete the study, and you will receive only the minimum payment of $10. In order for your tasks in Week 3 to be counted, they must be submitted by midnight on February 23rd, 2012.

Choosing the Decision-That-Counts: To summarize: In Week 1 (today), you will make 5 allocation decisions for both Job 1 and Job 2 for diﬀerent task rates. In Week 2, you will also make 5 allocation decisions for both Job 1 and Job 2 for diﬀerent task rates. Therefore, you will make 20 total allocation decisions. As stated above, we will choose only one of these decisions as the decisions-that-counts. That is, we will either implement one decision from Job 1 or one decision from Job 2, but not both.

There are three stages to determine the decision-that-counts.

1. First, we will choose if the decision-that-counts will come from Week 1 or Week 2. To do this, we will pick a random number from 1 to 10. If the number is 1, then the decision- that-counts will come from your Week 1 allocations. If the number is 2,3,4,5,6,7,8,9 or 10, then the decision-that-counts will come from your Week 2 allocations. Therefore, the decision-that-counts will come from Week 1 with a 10 percent chance and the decision- that-counts will come from Week 2 with a 90 percent chance.

2. Second, we will choose if the decision-that-counts will come from Job 1 or Job 2. To do this we will pick a second random number from 1 to 2. If the number is 1 then the decision-that-counts will come from Job 1. If the number is 2, then the decision-that- counts will come from Job 2. Therefore, the decision-that-counts is equally likely to come

7 from Job 1 and Job 2.

3. Third, we will choose the decision-that-counts from the 5 allocations you made in the chosen week and the chosen job. To do this, we will pick a third random number from 1 to 5. Therefore, within the chosen week and chosen job, every allocation is equally likely to be chosen as the decision-that-counts.

For example, consider the following allocation examples. Imagine that your allocations were shown in the following diagram for Weeks 1 and 2. Now, imagine that we determine the decision-that-counts.

Week 1 Allocations

Week 2 Allocations

1. Following the ﬁrst step above, we would ﬁrst generate a random number from 1 to 10 to determine whether the Week 1 or the Week 2 allocations will be implemented. If the

8 number is 1, then the decision-that-counts will come from your Week 1 allocations. If the number is 2,3,4,5,6,7,8,9 or 10, then the decision-that-counts will come from your Week 2 allocations. Imagine the number is 7, such that your Week 2 allocations will be implemented.

2. Following the second step above, we would generate a random number from 1 to 2 to determine the job of the decision-that-counts. If the number is 1 then the decision that counts will come from Job 1. If the number is 2, then the decision that counts will come from Job 2. Imagine the number is 1, such that the your Job 1 allocations will be implemented.

3. Following the third step above, we would generate a random number from 1 to 5 to determine the decision-that-counts from your Week 2 allocations for Job 1. Imagine this number is 3 such that the decision-that-counts would then be third allocation decision from your Week 2 allocations for Job 1

In Week 2, you would be required to complete 27 tasks of Job 1 and in Week 3 you would be required to complete 23 tasks of Job 1.

Note that these tasks will be in addition to the minimum work that you will be required to complete for both jobs in both weeks.

REMEMBER: EACH DECISION COULD BE THE DECISION-THAT-COUNTS SO TREAT EACH DECISION AS IF IT WAS THE ONE DETERMINING YOUR TASKS.

9 Recap:

• You will be participating in a six week study that requires participation one day per week on six consecutive weeks.

• You will receive a completion payment of $100 at the end of the study by check one week after Week 6. You will return to the xlab on March 22, 2012 to receive this payment.

• If you choose to no longer participate, or do not complete the jobs you chose, you will receive only a minimum payment of $10 by check one week after Week 6. You will return to the xlab on March 22, 2012 to receive this payment.

• There are two possible jobs in the study. Job 1 is transcription of greek letters. Job 2 is a tetris game.

• In each week, you will be asked to complete minimum work for each job.

• In Week 1, today, you will be asked to make a series of allocation decisions for both Job 1 and Job 2. You will allocate tasks to Weeks 2 and 3 at various task rates.

• In Week 2, you will again make allocation decisions.

• One of your allocation decisions will be chosen at random as the decision-that-counts and your allocation will determine the tasks that you complete in Weeks 2 and 3.

• One of your Week 1 allocations will be implemented with 10 percent chance while one of your Week 2 allocations will be implemented with 90 percent chance.

• Weeks 4, 5, and 6 will mirror Weeks 1, 2, and 3. In Week 4 you will make allocation decisions. In Week 5, you will again make allocation decisions and one of your allocation decisions will be chosen at random as the decision-that-counts. Your allocation will determine the jobs that you complete in Weeks 5 and 6.

10 • One week after week 6, you will receive your completion payment of $100. You will return to the xlab on March 22, 2012 to receive this payment.

11 Consent Now that we have explained the study, you are free to leave if you would like to choose not to participate in the study. Otherwise, please sign the consent form and we will pick these up now.

Minimum Work Now you will complete your minimum work for each job for this week. For each job, we ask that you complete 10 tasks.

Reminder of Timeline Today you will be asked to make a series of 5 allocation decisions for both Job 1 and Job 2. In these decisions you are asked to allocate tasks between one week from today (Week 2) and two weeks from today (Week 3). In each decision you are free to allocate your tasks as you choose. The allocations do not include the minimum amount of work for each job. You will choose by moving a slider to your desired allocation.

Allocations In the sliders on the screen, you will be asked to make 5 allocations for Job 1. Then, you will be asked to make 5 allocation decisions for Job 2. Remember each decision could be the decision-that-counts, so please make each decision as if it were the one that determines your tasks.

12 F.2 Week 1 Money Instructions

Thank you for completing your allocations. On the following screens we would like to ask you several additional questions allocating money over time. Your decisions in this portion of the study are completely unrelated to your allocations over Job 1 and Job 2 and will be paid separately. You must be willing to receive your payment for this study by cash provided to you in the xlab by Professor Ned Augenblick of the Haas School of Business. You will be required to return to the xlab on the dates indicated to complete the study and so your choice of payments will not require you to arrive any extra times.

Earning Money To begin, you will be given a $10 thank-you payment, just for participating in this study! You will receive this thank-you payment in two equally sized payments of $5 each. The two $5 payments will come to you at two different times. These times will be determined in the way described below. In this portion of the study, you will make 15 choices over how to allocate money between three possible dates: 1) February 9th (today - week 1), 2) March 1st (three weeks from today - week 4) 3) March 22nd (six weeks from today - week 7). Note that these are all days that you will be in the xlab. In each decision, you will allocate money between two of these dates. In the first set of five decisions, you will allocate money between week 1 (today) and week 4. In the second set, you will allocate money between week 1 (today) and week 7. In the third set, you will allocate money between week 4 and week 7. This means you could be receiving payments as early as today, and as late as the week 7. Once all 15 decisions have been made, we will randomly select one of the 15 decisions as the decision-that-counts. We will use the decision-that-counts to determine your actual earnings.

13 Note, since all decisions are equally likely to be chosen, you should make each decision as if it will be the decision-that-counts. When calculating your earnings from the decision-that-counts, we will add to your earnings the two $5 thank you payments. Thus, you will always get paid at least $5 at the chosen earlier time, and at least $5 at the chosen later time.

IMPORTANT: All payments you receive will be paid in cash in the xlab. On the scheduled day of payment, you will come to the xlab for the regular schedule of the study. Hence, you will not be asked to make any special arrangements to receive payment from this portion of the study. You will receive your payment from Professor Ned Augenblick.

On your desk are two envelopes: one for the sooner payment and one for the later payment. Please take the time now to write your participant ID on them.

14 How It Works: In the following three screens you are asked to make 15 decisions involving payments over time. Each row is a decision and is numbered from 1 to 15. Each row will feature a series of options. Each option consists of a sooner payment AND a later payment. You are asked to pick your favorite option in each row by moving the slider to your desired location. You should pick the combination of sooner payment AND later payment that you prefer the most. Note that there is a trade-oﬀ between the sooner payment and the later payment. As the sooner payment goes down, the later payment goes up. All you have to do for each decision is choose which combination of sooner and later payment you prefer the most by moving the slider to that location. Once all 15 of your decisions are complete, we will choose one at random to be the decision- that-counts. Your chosen allocation will be implemented.

Consider if the decision-that-counts was the third decision, and in that decision, you allocated $11 on February 9th and $10.50 on March 1st. Then, on February 9th, we would place $11 along with your $5 minimum payment, making $16.00, into your ﬁrst envelope. This envelope will be given to you on February 9th (today) in the xlab. On March 1st, we would place $10.50 along with your $5 minimum payment, making $15.50, into your second envelope. This envelope will be given to you on March 1st when you return to the xlab. Recall that this will not require you to make any special arrangements to receive payment as you will be returning to the laboratory as part of the regular schedule of the study.

Once your payments have been determined, you will write the amounts and dates on the inside of the two envelopes. When you receive your payments you can guarantee there have been no clerical errors by checking against the amounts and dates you wrote.

15 Remember that each decision could be the decision-that-counts! It is in your interest to treat each decision as if it could be the one that determines your payment.

16 F.3 Week 1 Quiz Quiz Please complete the quiz in order to make sure that you understand the allocation decisions and the timeline of the study.

Participant #

1. How many weeks are you required to participate?

2. In which weeks are you asked to come to the xlab to participate?

3. In which weeks are you asked to participate online and not come to the xlab?

4. Will you have to complete minimum work for each job in each week?

5. You will make allocation decisions for Weeks 2 and 3 both today and in Week 2. What is the percent chance that one of your Week 2 allocations will be implemented?

6. If you face a 1:2 task rate for allocations between Weeks 2 and 3, every task you allocate to Week 2 decreases by how many the number of tasks you allocate to Week 3?

7. You will make allocations for each job. Apart from your minimum work, will you complete any tetris tasks if a transcription job allocation is chosen as the decision-that-counts.

17 F.4 Week 4 Eﬀort Instructions Welcome: Thank you for returning to the experiment. We will begin shortly.

Eligibility for this study: To continue in this study, you need to meet these criteria: You must be willing to participate for three consecutive weeks. Participation will require your presence on three consecutive Fridays for at least 10 minutes per week for an average of 60 minutes. Week 4 (today) will occur in the xlab. Weeks 5 and 6 will occur at any computer that has access to the Internet. You must be willing to receive your payment from this experiment as one single completion payment at the end of the study. Payments will be made one week after the ﬁnal session, on Friday, March 23. You will return to the xlab to receive this payment. If you do not meet these criteria, please inform us of this now.

Your Earnings If you complete all six weeks of participation, a completion payment of $100 will be provided. You may receive additional earnings during the experiment. If you choose to end your participation before the six weeks are complete, please report this to study administrators, and you will receive a minimum payment of $10. All payments will be made one week after the ﬁnal session, on Friday March 23. You will return to the xlab to receive this payment.

18 in each of the six consecutive weeks of the study.

Job 1: In Job 1 you are asked to transcribe letters from a greek text.

Job 2: In Job 2 you are asked to play a tetris game.

19 The Experiment Timeline Note: Minimum Work for each week In each week (including today), you are required to complete a minimum number of tasks of both Job 1 and Job 2.

Today (Week 4): Once your minimum work is complete, you will be asked to make a series of 5 decisions for each job. In these decisions you are asked to allocate tasks between one week from today (Week 5) and two weeks from today (Week 6). You will make 5 decisions for both Job 1 and Job 2. In each decision you are free to allocate your tasks as you choose. Note that this allocation decision does not include the minimum work for each week, which you must also complete. Task Rates: For one example task rate, every task you allocate to Week 6 reduces the number of tasks allocated to Week 5 by one. This is what we will refer to as a 1:1 task rate. The task rate will vary across your decisions. For example, the task rate may be 1:1.5, such that every task you allocate to Week 6 reduces the number of tasks allocated to Week 5 by 1.5. Or, the task rate may be 1:0.5, such that every task you allocate to Week 6 reduces the number of tasks allocated to Week 5 by 0.5. For simplicity, the task rates will always be represented as 1:X, and you will be fully informed of the value of X when making your decisions. Week 5 (One Week From Today): Week 5, one week from today, will occur online and follows week 2 of the experiment. You will receive an email with instructions on how to access the website with the jobs. You will again complete your minimum work. You will be asked again to make 5 allocation decisions for each job. Exactly one of your 20 total allocation decisions will be implemented. That is, we will implement one decision from Week 4 for Job 1, or one decision from Week 5 for Job 1, or one decision from Week 4 for Job 2, or one decision from Week 5 for Job 2. We will discuss how this allocation decision is chosen shortly. We refer to this allocation

20 decision as the ”decision-that-counts.” The tasks you allocated to Week 5 in the decision- that-counts must be completed. If you do not return or do not complete the tasks in Week 5, you cannot complete the study, and you will receive only the minimum payment of $10. In order for your tasks in Week 5 to be counted, they must be submitted by midnight on March 9th, 2012.

Week 6, Two Weeks From Today: Week 6, two weeks from today, will occur online and follows week 3 of the experiment. You will receive an email with instructions on how to access the website with the jobs. You will again complete your minimum work. Then, you must complete the tasks you allocated in the decision-that-counts. If you do not return or do not complete the tasks in Week 6, you cannot complete the study, and you will receive only the minimum payment of $10. In order for your tasks in Week 6 to be counted, they must be submitted by midnight on March 16th, 2012.

Choosing the Decision-That-Counts: To summarize: In Week 4 (today), you will make 5 allocation decisions for both Job 1 and Job 2 for diﬀerent task rates. In Week 5, you will also make 5 allocation decisions for both Job 1 and Job 2 for diﬀerent task rates. Therefore, you will make 20 total allocation decisions. As stated above, we will choose only one of these decisions as the decisions-that-counts. That is, we will either implement one decision from Job 1 or one decision from Job 2, but not both.

The decision-that-counts will be chosen using a similar method to the one used in Week 2. However, this week, you will make a set of new decisions that aﬀect the precise way that the decision-that-counts is chosen. To understand these new decisions, please recall how the decision-that-counts was chosen in Week 2: How the decision-that-counts was chosen in Week 2 We used 3 steps to choose the decision-that-counts in Week 2.

21 1. First, we chose if the decision-that-counts came from the sooner week (Week 1) or the later week (Week 2) allocations. To do this, we picked a random number from 1 to 10. If the number was 1, then the decision-that-counts came from the allocations from the sooner week (Week 1). If the number is 2,3,4,5,6,7,8,9 or 10, then the decision-that-counts came from the allocations from the later week (Week 2). Therefore, the decision-that-counts came from the sooner week with a 10 percent chance and the decision-that-counts came from the later week with a 90 percent chance. This is the part of the choosing the decision-that-counts that you will be able to aﬀect in the new set of decisions this week.

2. Second, we chose if the decision-that-counts came from Job 1 or Job 2. To do this, we picked a second random number from 1 to 2. If the number was 1 then the decision-that- counts came from Job 1. If the number was 2, then the decision-that-counts came from Job 2. Therefore, the decision-that-counts was equally likely to come from Job 1 and Job 2.

3. Third, we chose the decision-that-counts from the 5 allocations you made in the chosen week and the chosen job. To do this, we picked a third random number from 1 to 5. Therefore, within the chosen week and chosen job, every allocation was equally likely to be chosen as the decision-that-counts.

How the decision-that-counts will be chosen in Week 5 In Week 5, the decision that counts will be chosen in a similar way to Week 2 with one important difference. Today, you will make a set of 15 decisions that can affect the first step of the process. In Week 2, there was a 10 percent chance that the decision-that-counts would come from your sooner (Week 1) allocations. In Week 5, based on your decisions, there will either be a 10 percent chance or a 90 percent chance that decision-that-counts will come from your sooner (Week 4) allocations. That is, your decisions will change the likelihood that one of your Week 4 allocations is chosen as the decision-that-counts.

22 For example, in one of the decisions, you will simply be asked to choose which option you prefer: 1) a 10 percent chance that decision-that-counts will come from your Week 4 allocations (and 90 percent chance that it comes from Week 5). 2) a 90 percent chance that decision-that-counts will come from your Week 4 allocations (and 10 percent chance that it comes from Week 5) This decision measures your preference about which choices will be allocated. For example, if you would prefer that one of your week 5 allocations were chosen rather than a week 4 allocation, you should choose the first option. Please take a second to think about this decision. The other decisions measure the strength of your preference about which choices will be allocated. In these decisions, you will make this same decision but with additional payments added to one of the two options. So, for example, you will be asked to choose which option you prefer: 1) a 10 percent chance that decision-that-counts will come from your Week 4 allocations (and 90 percent chance that it comes from Week 5). 2) a 90 percent chance that decision-that-counts will come from your Week 4 allocations (and 10 percent chance that it comes from Week 5) plus $3. For example, if you would very strongly prefer that one of your Week 5 allocations were chosen rather than a Week 4 allocation, you might still choose the first option, even though you could get an extra $3 for choosing the second option. We will choose one of your 15 percentage decisions to be implemented at random. This implemented decision will be used to determine the percentage chance that the decision-that-counts comes from your Week 4 allocations. Furthermore, if your implemented decision includes an additional payment, this additional payment will be added to your final $100 completion check.

23 REMEMBER: EACH DECISION COULD BE IMPLEMENTED SO TREAT EACH DECISION AS IF IT WAS GOING TO BE IMPLEMENTED.

24 Recap:

• You will be continuing in a study that requires participation one day per week on three consecutive weeks.

• You will receive a completion payment of $100 at the end of the study by check one week after Week 6. You will return to the xlab on March 23, 2012 to receive this payment.

• There are two possible jobs in the study. Job 1 is transcription of greek letters. Job 2 is a tetris game.

• In each week, you will be asked to complete minimum work for each job.

• In Week 4, today, you will be asked to make a series of allocation decisions for both Job 1 and Job 2. You will allocate tasks to Weeks 5 and 6 at various task rates.

• In Week 5, you will again make allocation decisions.

• One of your allocation decisions will be chosen at random as the decision-that-counts and your allocation will determine the tasks that you complete in Weeks 5 and 6.

• You will be asked to make decisions about the percentage chance that the decision- that-counts will come from your Week 4 allocations. You will make a series of 15 decisions between (10% Week 4) and (90% Week 4) with additional payments potentially added to the options. One of these decisions will be implemented. If the decision that is implemented includes an additional payment, this will be added to your completion payment.

• One week after Week 6, you will receive your completion payment. You will return to the xlab on March 23, 2012 to receive this payment.

25 Minimum Work Now you will complete your minimum work for each job for this week. For each job, we ask that you complete 10 tasks.

Allocations Today you will be asked to make a series of 5 allocation decisions for both Job 1 and Job 2. In these decisions you are asked to allocate tasks between one week from today (Week 5) and two weeks from today (Week 6). In each decision you are free to allocate your tasks as you choose. The allocations do not include the minimum amount of work for each job. You will choose by moving a slider to your desired allocation.

In the sliders on the screen, you will be asked to make 5 allocations for Job 1. Then, you will be asked to make 5 allocation decisions for Job. Remember each decision could be the decision-that-counts, so please make each decision as if it were the one that determines your tasks. Determining how the decision-that-counts will be chosen in Week 5 On the screen you will be asked to choose between a 10% or a 90% chance that the decision- that-counts comes from today’s allocations (Week 4) rather than the allocations you will make next week (Week 5). In each decision, you are also given an additional payment for choosing one of the two options. Remember each decision could be implemented, so please make the decision as if it was determining the percent chance and your additional payment.

26 F.5 Week 4 Money Instructions

Thank you for completing your allocations. On the following screen we would like to ask you several additional questions allocating money over time. Your decisions in this portion of the study are completely unrelated to your allocations over Job 1 and Job 2 and will be paid separately. You must be willing to receive your payment for this study by cash provided to you in the xlab by Professor Ned Augenblick of the Haas School of Business. You will be required to return to the xlab on the dates indicated to complete the study and so your choice of payments will not require you to arrive any extra times.

Earning Money To begin, you will be given a $10 thank-you payment, just for participating in this study! You will receive this thank-you payment in two equally sized payments of $5 each. The two $5 payments will come to you at two diﬀerent times. These times will be determined in the way described below. In this portion of the study, you will make 5 choices over how to allocate money between two points in time: 1) March 2nd 2) March 23rd Note that these are days that you will be in the xlab. In each decision, you will allocate money between these dates. Once all 5 decisions have been made, we will randomly select one of the 5 decisions as the decision-that-counts. We will use the decision-that-counts to determine your actual earnings. Note, since all decisions are equally likely to be chosen, you should make each decision as if it will be the decision-that- counts. When calculating your earnings from the decision-that-counts, we will add to your earnings the two $5 thank you payments. Thus, you will always get paid at least $5 on March 2st, and

27 at least $5 on March 23nd. IMPORTANT: All payments you receive will be paid in cash in the xlab. On the scheduled day of payment, you will come to the xlab for the regular schedule of the study. Hence, you will not be asked to make any special arrangements to receive payment from this portion of the study. You will receive your payment from Professor Ned Augenblick.

On your desk are two envelopes: one for the sooner payment and one for the later payment. Please take the time now to write your participant ID on them and study time/date on them.

28 How It Works: In the following screen you are asked to make 5 decisions involving payments over time. Each row is a decision and is numbered from 1 to 5. Each row will feature a series of options. Each option consists of a sooner payment AND a later payment. You are asked to pick your favorite option in each row by moving the slider to your desired location. You should pick the combination of sooner payment AND later payment that you prefer the most. Note that there is a trade-oﬀ between the sooner payment and the later payment. As the sooner payment goes down, the later payment goes up. All you have to do for each decision is choose which combination of sooner and later payment you prefer the most by moving the slider to that location. Once all 5 of your decisions are complete, we will choose one at random to be the decision- that-counts. Your chosen allocation will be implemented.

Consider for example the ﬁrst decision. If this was chosen as the decision that counts and your preferred allocation was $11 on March 2st and $10.50 on March 23nd, this would then be implemented. On March 2st, we would place $11 along with your $5 minimum payment, making $16.00, into your ﬁrst envelope. This envelope will be given to you on March 2st in the xlab. On March 23nd, we would place $10.50 along with your $5 minimum payment, making $15.50, into your second envelope. This envelope will be given to you on March 23nd when you return to the xlab. Recall that this will not require you to make any special arrangements to receive payment as you will be returning to the laboratory as part of the regular schedule of the study.

29 Remember that each decision could be the decision-that-counts! It is in your interest to treat each decision as if it could be the one that determines your payment.

G Replication Study Instructions

INSTRUCTIONS

Welcome Thank you for participating in our study. We will begin shortly.

Eligibility and Study Requirements Participation in this study will require activities lasting at least 30 minutes on four consecutive Thursdays, beginning today and ending three weeks from today: Apr-10, Apr-17, Apr-24, May-1. Your ﬁrst activity for this study is choosing amounts of money to be received one week from today, Thursday, Apr-17, and two weeks from today Thursday, Apr-24. You will make nine such choices today and you will make nine such choices one week from today on Thursday, Apr-17. The choices you make today will be referred to as your Week 1 choices. The choices you make one week from today on Thursday, Apr-17, will be referred to as your Week 2 choices. You will make your Week 1 choices today in the laboratory via the study website. Next week, on Wednesday night, you will receive an email with the study website. This link will be active at 9am on Thursday morning. You can then login and make your Week 2 choices using any computer that has internet access. You must make these decisions by 4pm that day. Your second activity for this study is collecting your chosen payments. Payments will be collected at a table setup directly outside of the xlab. The Apr-17th payments must be collected within 2 hours of making your decisions. The Apr-24th payments must be collected between 9am and 6pm. In order to complete this study you must be willing to both choose amounts of money and to collect these amounts on Thursday, Apr-17 and Thursday, Apr-24. If you complete these elements, you will be eligible to receive a completion payment of $30. This can be collected on Thursday, May-1 outside the xlab between the hours of 9 a.m. and 6 p.m. If you do not complete all elements of the study, you will be eligible to receive only a payment of $5. This can also be collected on Thursday, May-1 at the xlab between the hours of 9 a.m. and 6 p.m.

30 If you do not meet or understand the study requirements, please inform us of this now.

Your Earnings All payments will be made by Professor Ned Augenblick and his assistants. All payments will be made in cash. You will receive your payments only in designated locations, at designated times, on designated dates.

Anonymity Your anonymity in this study is assured. Your name will never be recorded or connected to any decision you make here today. Your email will be collected in order to communicate with you during the study. After the study, your email information will be destroyed and will not be connected to your responses in the experiment.

Rules Please turn your cell phones oﬀ. If you have a question at any point, just raise your hand. Please put away any books, papers, computers, etc.

Registration We will now begin the study. Please open the study interface and enter your e-mail address. Make sure that you enter a valid e-mail address as this will be the method by which we contact you throughout the study.

Study Activities We will now discuss in detail the study activities. In this study there are two activities. The ﬁrst activity is choosing payments over time. You will make 9 such choices today, Thursday, Apr-10. The choices you make today will be referred to as your Week 1 choices. You will make 9 such choices one week from today, Thursday, Apr-17. The choices you make one week from today will be referred to as your Week 2 choices. In both your Week 1 and Week 2 choices, you will be choosing an amount of money to be received on Thursday, Apr-17 AND an amount of money to be received on Thursday, Apr-24.

31 Each choice is a series of options. Each option consists of a sooner payment (to be collected on Thursday, Apr-17) AND a later payment (to be collected on Thursday, Apr-24). You are asked to pick your favorite option in each choice by moving a slider to your desired location. In the example sliders on the website, please explore the potential choices. Note that there is a trade-oﬀ between the sooner payment and the later payment. When the sooner payment goes down, the later payment goes up and vice versa. In each choice, the trade-oﬀ will be summarized by an “exchange rate,” and will be expressed as a number 1 : X. This means that if you increase the sooner payment by $1, the later payment will be decreased by $X. In the example, the exchange rate was 1 : 1, meaning that if you increase the sooner payment by $1, the later payment is decreased by $1. Remember, all you have to do for each decision is choose which combination of sooner and later payment you prefer the most by moving the slider to that location. The second activity is collecting payments. You will collect the payments from one of your choices. This means you will collect some amount of money on Thursday, Apr-17 and some amount of money on Thursday, Apr-24. The payment on Apr-17 must be collected within 2 hours of making your decision. The payment on Apr-24 can be collected anytime between 9am and 6pm. Payments will be collected at a table setup outside of the xlab.

The Experiment Timeline We will now discuss the timeline of the study. Along the way we will discuss a few important details of how the study works.

Note: Minimum Payments for Each Week. With the exception of the ﬁnal completion payment date, on each day of study participation (including today), you will receive a minimum payment of $5. These payments are in addition to your chosen payments. This payment will be paid in cash and be added to your experimental earnings. These payments must be collected for successful completion of the study.

Week 1(Today: Apr-10) Today you will make 9 choices. Each choice is a series of options. Each option will consist of a sooner payment (to be collected on Thursday, Apr-17) AND a later payment (to be collected on Thursday, Apr-24). In each choice, the “exchange rate” will be diﬀerent. You will be asked to pick your favorite option in each choice by moving a slider to your desired location. You should pick the combination of sooner payment AND later payment that you prefer the most. Once your 9 choices are complete, you will receive your minimum payment of $5 and depart.

32 Week 2 (One Week From Today: Apr-17) Next week, on Wednesday night, you will receive an email with the study website. This link will be active at 9am on Thursday morning. You must log in to the study website between 9am and 4pm of next Thursday, Apr-17. You will also receive a reminder email on Thursday. You will again make 9 choices. Each choice is a series of options. Each option will consist of a sooner payment (to be collected on Thursday, Apr-17) AND a later payment (to be collected on Thursday, Apr-24). In each choice, the “exchange rate” will be diﬀerent. You will be asked to pick your favorite option in each choice by moving a slider to your desired location. You should pick the combination of sooner payment AND later payment that you prefer the most. Once your 9 Week 2 choices are complete, you will have made a total of 18 choices: 9 Week 1 choices and 9 Week 2 choices. We will then pick one of your 18 total choices at random to be the decision-that-counts. Your earnings will be determined by your decision in the decision-that-counts. You will receive the amounts speciﬁed in the decision-that-counts on the designated dates, Thursday, Apr-17 and Thursday, Apr-24. Recall that these earnings are in addition to your two $5 minimum payments. Thus, you will always pick up a payment on Thursday, Apr-17 and $5 on Thursday, Apr-24 of at least $5.

REMEMBER: EACH DECISION COULD BE THE DECISION-THAT-COUNTS SO TREAT EACH CHOICE AS IF IT WAS THE ONE DETERMINING YOUR EARNINGS.

Consider if in the decision-that-counts your preferred choice was an $11 sooner payment (to be collected on Thursday, Apr-17) AND a $10.50 later payment (to be collected on Thursday, Apr-24). Then, on Thursday, Apr-17 outside the xlab, you would collect $11 along with your $5 minimum payment, making $16.00 in cash. On Thursday, Apr-24 outside the xlab, you would collect $10.50 along with your $5 minimum payment, making $15.50 in cash. You must collect your earnings on Apr-17 within two hours of making your decisions online. You must collect your earnings on Apr-24 between 9am and 6pm.. If you do not collect your payment on either Thursday, Apr-17 or Thursday, Apr-24, you will be removed from the study and forfeit all future payments including your completion payment of $30. You will be eligible only for the reduced payment of $5 at the end of the study. There will be no exceptions to this rule.

REMEMBER:YOU MUST PICK UP YOUR PAYMENTS BETWEEN 9AM AND 6PM.

33 YOU MUST PICK UP YOUR PAYMENT NEXT WEEK WITHIN 2 HOURS OF LOGGING INTO THE WEBSITE. Week 3, (Two Weeks From Today: Apr-24) In Week 3, you will make no decisions. You will receive an e-mail the night before reminding you of the study. You must pick up your payment for Thursday, Apr-24 along with your $5 minimum payment outside the xlab between 9am and 6pm. If you do not pick up your payment, you will be removed from the study and forfeit your completion payment of $30. You will be eligible only for the reduced payment of $5 at the end of the study. There will be no exceptions to this rule.

34 Week 4, (Three Weeks From Today: May-1) In Week 4, you will make no decisions. You will receive an email reminding you to pick up your completion payment outside the xlab between 9am and 6pm on Thursday, May-1. If you have completed all elements of the study you are eligible to receive a $30 completion payment. If you have not completed all elements of the study you are eligible to receive a $5 completion payment.

35 Recap:

• You will be participating in a four week study that requires participation for at least 30 minutes on four consecutive Thursdays.

• On Thursday, April-10, Thursday, Apr-17 and Thursday, Apr-24, you will receive minimum payments of $5. These minimum payments are in addition to your chosen payments from the decision-that-counts. These payments must be collected for successful completion of the study.

• In Week 1, today, Thursday, Apr-10, you will be asked to make 9 choices. Each choice is a series of options. Each option will consist of a sooner payment (to be collected on Thursday, Apr-17) AND a later payment (to be collected on Thursday, Apr-24). In each choice, the “exchange rate” will be diﬀerent.

• In Week 2, one week from today, Thursday, Apr-17, you will again be asked to make 9 choices. These decisions will be made online between 9am and 4pm. Each choice is a series of options. Each option will consist of a sooner payment (to be collected on Thursday, Apr-17) AND a later payment (to be collected on Thursday, Apr-24). In each choice, the “exchange rate” will be diﬀerent.

• You will be asked to pick your favorite option in each choice by moving a slider to your desired location. You should pick the combination of sooner payment AND later payment that you prefer the most.

• You will make 18 total choices: 9 in Week 1 and 9 in Week 2. We will pick one of your 18 total choices at random to be the decision-that-counts.

• Once your Week 2 decisions have been made on Thursday, Apr-17 and the decision- that-counts has been determined, a two hour window will begin. You must collect your earnings for Thursday, Apr-17 from the decision-that-counts outside of the xlab within this two hour window (and between 9 am and 6pm).

• In Week 3, two weeks from today Thursday, Apr-24, you must collect your earnings for Thursday, Apr-24 from the decision-that-counts outside the xlab between 9am and 6pm.

• If you fail to collect your earnings from the decision-that-counts you will be removed from the study and forfeit all future payments. You will be eligible only for the reduced payment of $5 at the end of the study.

• In Week 4, three weeks from today Thursday, May-1, you will pick up your completion payment outside the xlab between the hours of 9 a.m. and 6 p.m. to receive your completion payment. If you have completed all elements of the study you are eligible for a $30 completion payment. If you have not completed all elements of the study you are eligible only for the reduced amount of $5.

36 Consent Now that we have explained the study, you are free to leave if you would like to choose not to participate in the study. Otherwise, please sign the consent form and we will pick these up now.

Allocations In the sliders on the screen, you will be asked to make 9 allocations. Remember each decision could be the decision-that-counts, so please make each decision as if it were the one that determines your payment.

37 INSTRUCTIONS Welcome: Thank you for participating in our study. We will begin shortly.

Eligibility and Study Requirements Participation in this study will require activities lasting at least 30 minutes on four consecutive Thursdays, beginning today and ending three weeks from today: Apr-10, Apr-17, Apr-24, May-1. Your ﬁrst activity for this study is choosing amounts of work to be completed one week from today, Thursday, Apr-17, and two weeks from today Thursday, Apr-24. You will make nine such choices today and you will make nine such choices one week from today on Thursday, Apr-17. The choices you make today will be referred to as your Week 1 choices. The choices you make one week from today on Thursday, Apr-17, will be referred to as your Week 2 choices. You will make your Week 1 choices today in the laboratory via the study website. Next week, on Wednesday night, you will receive an email with the study website. This link will be active at 9am on Thursday morning. You can then login and make your Week 2 choices using any computer that has internet access. You must make these decisions by 4pm that day. Your second activity for this study is completing your chosen work. The work in this study will be completed via the study website and can be completed on any computer that has internet access. The Apr-17th work must be completed within 2 hours of making your decisions. The Apr-24th work must be completed between 9am and 6pm. In order to complete this study you must be willing to both choose amounts of tasks and to complete these tasks on Thursday, Apr-17 and Thursday, Apr-24. If you complete these elements, you will be eligible to receive a completion payment of $60. This can be collected on Thursday, May-1 outside the xlab between the hours of 9 a.m. and 6 p.m. If you do not complete all elements of the study, you will be eligible to receive only a payment of $5. This can also be collected on Thursday, May-1 at the xlab between the hours of 9 a.m. and 6 p.m. If you do not meet or understand the study requirements, please inform us of this now.

Your Earnings The completion payment will be made by check by Professor Ned Augenblick and his assistants. You will receive your payment only in designated locations, at designated times, on designated dates.

38 Tasks The tasks in this study are transcriptions of letters from a greek text. Greek text will appear in a Transcription Box on your screen. For each letter, you will need to ﬁnd and select the corresponding letter and enter it into the Completion Box on your screen. One task is one row of greek text. For a task to be complete, your accuracy must be 80% or better. Each task takes an average student between 40-60 seconds.

Rules Please turn your cell phones oﬀ. If you have a question at any point, just raise your hand. Please put away any books, papers, computers, etc.

39 Study Activities We will now discuss in detail the study activities. In this study there are two activities. The first activity is choosing amounts of work over time. You will make 9 such choices today, Thursday, Apr-10. The choices you make today will be referred to as your Week 1 choices. You will make 9 such choices one week from today, Thursday, Apr-17. The choices you make one week from today will be referred to as your Week 2 choices. In both your Week 1 and Week 2 choices, you will be choosing an amount of work to be completed on Thursday, Apr-17 AND an amount of work to be completed on Thursday, Apr-24. The amount of work on each date is expressed as a number of tasks. Each choice is a series of options. Each option consists of a sooner number of tasks (to be completed on Thursday, Apr-17) AND a later number of tasks (to be completed on Thursday, Apr-24). You are asked to pick your favorite option in each choice by moving a slider to your desired location. In the example sliders on the website, please explore the potential choices. Note that there is a trade-off between the sooner number of tasks and the later number of tasks. When the sooner number of tasks goes down, the later number of tasks goes up and vice versa. In each choice, the trade-off will be summarized by an “exchange rate,” and will be expressed as a number 1 : X. This means that if you increase the sooner number of tasks by 1 task, the later number of tasks will be decreased by X tasks. In the example, the exchange rate was 1 : 1, meaning that if you increase the sooner number of tasks by 1, the later number of tasks is decreased by 1. Remember, all you have to do for each decision is choose which combination of sooner and later tasks you prefer the most by moving the slider to that location. The second activity is completing work. This means you will complete some tasks on Thursday, Apr-17 and some tasks on Thursday, Apr-24. The tasks on Apr-17 must be completed within 2 hours of making your decision. The tasks on Apr-24 can be completed anytime between 9am and 6pm. Tasks will be completed online using any computer that has access to the Internet.

The Experiment Timeline We will now discuss the timeline of the study. Along the way we will discuss a few important details of how the study works.

Note: Minimum Work for Each Week. With the exception of the ﬁnal completion payment date, on each day of study participation (including today), you are required to complete a minimum number of 10 tasks. These tasks are in addition to your chosen numbers of tasks. These tasks must be completed for successful completion of the study.

40 Week 1(Today: Apr-10) Today you will make 9 choices. Each choice is a series of options. Each option will consist of a sooner number of tasks (to be completed on Thursday, Apr-17) AND a later number of tasks (to be completed on Thursday, Apr-24). In each choice, the “exchange rate” will be diﬀerent. You will be asked to pick your favorite option in each choice by moving a slider to your desired location. You should pick the combination of sooner tasks AND later tasks that you prefer the most. Today you will complete your minimum work of 10 tasks, then make your 9 choices, and depart.

Week 2 (One Week From Today Apr-17) Next week, on Wednesday night, you will receive an email with the study website. This link will be active at 9am on Thursday morning. You must log in to the study website between 9am and 4pm of next Thursday, Apr-17. You will also receive a reminder email on Thursday. You will again complete minimum work of 10 tasks. Then, you will again make your 9 choices. Each choice is a series of options. Each option will consist of a sooner number of tasks (to be completed on Thursday, Apr-17) AND a later number of tasks (to be completed on Thursday, Apr-24). In each choice, the “exchange rate” will be diﬀerent. You will be asked to pick your favorite option in each choice by moving a slider to your desired location. You should pick the combination of sooner tasks AND later tasks that you prefer the most. Once your 9 Week 2 choices are complete, you will have made a total of 18 choices: 9 Week 1 choices and 9 Week 2 choices. We will then pick one of your 18 total choices at random to be the decision-that-counts. Your tasks will be determined by your decision in the decision- that-counts. You will complete the tasks speciﬁed in the decision-that-counts on the designated dates, Thursday, Apr-17 and Thursday, Apr-24. Recall that these tasks are in addition to your 10 tasks of minimum work. Thus, you will always login into the website and complete at least 10 tasks on Thursday, Apr-17 and 10 tasks on Thursday, Apr-24.

REMEMBER: EACH CHOICE COULD BE THE DECISION-THAT-COUNTS SO TREAT EACH CHOICE AS IF IT WAS THE ONE DETERMINING YOUR TASKS.

Consider if in the decision-that-counts your preferred choice was 30 sooner tasks (to be completed on Thursday, Apr-17) AND 25 later tasks (to be completed on Thursday, Apr-24). Then, on the study website, on Thursday, Apr-17, you would complete 30 tasks along with

41 your 10 tasks of minimum work, making 40 tasks. On Thursday, Apr-24, on the study website, you would complete 25 tasks along with your 10 tasks of minimum work, making 35 tasks. You must complete your tasks on Apr-17 within two hours of making your decisions online. You must complete your tasks on Apr-24 between 9am and 6pm. If you do not complete your tasks on either Thursday, Apr-17 or Thursday, Apr-24, you will be removed from the study and forfeit the completion payment of $60. You will be eligible only for the reduced payment of $5 at the end of the study. There will be no exceptions to this rule.

REMEMBER:YOU MUST COMPLETE YOUR TASKS BETWEEN 9AM AND 6PM. YOU MUST COMPLETE YOUR TASKS NEXT WEEK WITHIN 2 HOURS OF LOGGING INTO THE WEBSITE. Week 3, (Two Weeks From Today: Apr-24) In Week 3, you will make no decisions. You will receive an e-mail the night before reminding you of the study. You must login into the website between 9am and 6pm and you must complete your tasks from the decision-that-counts for Thursday, Apr-24 along with your 10 tasks of minimum work on the study website. If you do not complete your tasks by 6pm, you will be removed from the study and forfeit your completion payment of $60. You will be eligible only for the reduced payment of $5 at the end of the study. There will be no exceptions to this rule.

Week 4, (Three Weeks From Today: May-1) In Week 4, you will make no decisions. You will receive an email reminding you to pick up your completion payment in the xlab. If you have completed all elements of the study you are eligible to receive a $60 completion payment. If you have not completed all elements of the study you are eligible to receive a $5 completion payment. These payments will be available outside of the xlab on Thursday, May-1 between the hours of 9 a.m and 6 p.m.

42 Recap:

• You will be participating in a four week study that requires participation for at least 30 minutes on four consecutive Thursdays.

• On Thursday, April-10, Thursday, Apr-17 and Thursday, Apr-24, you are required to complete minimum work of 10 tasks. These tasks are in addition to your chosen tasks from the decision-that-counts. These tasks must be completed for successful completion of the study.

• In Week 1, today Thursday, Apr-10, you will be asked to make 9 choices. Each choice is a series of options. Each option will consist of a sooner number of tasks (to be completed on Thursday, Apr-17) AND a later number of tasks (to be completed on Thursday, Apr-24). In each choice, the “exchange rate” will be diﬀerent.

• In Week 2, one week from today Thursday, Apr-17, you will again be asked to make 9 choices. Each choice is a series of options. Each option will consist of a sooner number of tasks (to be completed on Thursday, Apr-17) AND a later number of tasks (to be completed on Thursday, Apr-24). In each choice, the “exchange rate” will be diﬀerent.

• You will be asked to pick your favorite option in each choice by moving a slider to your desired location. You should pick the combination of sooner tasks AND later tasks that you prefer the most.

• You will make 18 total choices: 9 in Week 1 and 9 in Week 2. We will pick one of your 18 total choices at random to be the decision-that-counts.

• Once your Week 2 decisions have been made on Thursday, Apr-17 and the decision-that- counts has been determined, a two hour window will begin. You must complete your tasks for Thursday, Apr-17 from the decision-that-counts on the study website within this two hour window (and between 9 am and 6pm).

• In Week 3, two weeks from today Thursday, Apr-24, you will again login to the website. You must complete your tasks for Thursday, Apr-24 from the decision-that-counts on the study website within two hours of logging in (and between 9am and 6pm).

• If you fail to complete your tasks from the decision-that-counts you will be removed from the study. You will be eligible only for the reduced payment of $5 at the end of the study.

• In Week 4, three weeks from today Thursday, May-1, you will come outside the xlab between the hours of 9 a.m. and 6 p.m. to receive your completion payment. If you have completed all elements of the study you are eligible for a $60 completion payment. If you have not completed all elements of the study you are eligible only for the reduced amount of $5.

43 Consent Now that we have explained the study, you are free to leave if you would like to choose not to participate in the study. Otherwise, please sign the consent form and we will pick these up now.

Minimum Work Recall that, each week, you must complete a mandatory number of 10 tasks. We will not complete those tasks.