NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
The Endowment Effect and the Take-up of Collateralized Loans
Last registered on September 21, 2017


Trial Information
General Information
The Endowment Effect and the Take-up of Collateralized Loans
Initial registration date
September 20, 2017
Last updated
September 21, 2017 12:49 PM EDT
Primary Investigator
Harvard University
Other Primary Investigator(s)
PI Affiliation
Harvard University
Additional Trial Information
In development
Start date
End date
Secondary IDs
Loans to purchase new assets sometimes use the new assets themselves as collateral, as in car or home loans in the developed world. Other loans—especially in developing countries —instead require using existing assets as collateral. We hypothesize that the endowment effect (Kahneman et al. 1990) causes consumers to dislike placing existing assets at risk by promising them as collateral. This might drive down take-up of loans collateralized using existing assets, relative to collateralizing with new assets (which may not yet have entered the reference point). The proposed project will test this mechanism using a randomized experiment in Kenya. A secondary goal will be to test whether borrowers anticipate that new assets might themselves eventually enter the reference point, becoming subject to an endowment effect. Our findings will shed light on psychological factors affecting demand for financing, and on the welfare implications of loans collateralized using new assets.
External Link(s)
Registration Citation
Kremer, Michael and Gautam Rao. 2017. "The Endowment Effect and the Take-up of Collateralized Loans." AEA RCT Registry. September 21. https://doi.org/10.1257/rct.2448-1.0.
Former Citation
Kremer, Michael, Gautam Rao and Gautam Rao. 2017. "The Endowment Effect and the Take-up of Collateralized Loans." AEA RCT Registry. September 21. http://www.socialscienceregistry.org/trials/2448/history/21659.
Sponsors & Partners

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information
Experimental Details
We endow households with one randomly-selected household or agricultural durable good, and offer them a loan to purchase a second randomly-selected good. We elicit willingness to pay for two types of loans to purchase the second good: standard asset-collateralized loans (SACLs) and new asset-collateralized loans (NACLs). SACLs require borrowers to use existing assets as collateral, while NACLs use the very item being purchased as collateral. These loans are administered by a well-established Kenyan Savings and Credit Cooperative Organization (SACCO).
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
Primary outcome:
Willingness to pay for the loans.

Secondary outcomes:
Willingness to pay for household and agricultural items without a loan
Predicted future willingness to pay for these items using loans
Perceived risk of loan default
Willingness to accept for previously endowed items
Predicted future willingness to accept for items to be endowed
Acceptance or rejection of small-scale, gain-loss lotteries (to measure loss aversion).
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
We will endow each respondent with an item for free, and later elicit their willingness to pay for new items using a loan, experimentally varying whether the previously endowed item or the new item being purchase is used as collateral. We will elicit predictions of future willingness to pay for these loans, to determine whether respondents anticipate their loan preferences prior to being endowed with any item.
Experimental Design Details
In a first session, we will elicit participants’ willingness to pay for a list of agricultural and household items using the Becker-DeGroot-Marschak (BDM) method. For each respondent, we will identify three of these items that they value similarly. Out of these three, one item will be randomly selected to be given to the respondent for free. Then, after several days, the respondent will return for a second session. In this session, we will elicit willingness to pay (again using BDM) for the remaining two items using a loan. One of the items will be paid for using a SACL, with the item endowed in the first session as collateral, while the other item will be paid for using a NACL, with the new item being purchased itself as collateral. Random assignment of which item is initially endowed, which is for sale using a SACL, and which is for sale using a NACL ensures that the average willingness to pay for the assets being purchased and used as collateral is the same in the SACL and the NACL, such that any systematic difference between willingness to pay for the two loans can be attributed to the endowment effect. In the first session, before respondents are endowed with any item, we elicit predictions of future willingness to pay for items using a SACL and a NACL. This allows us to determine whether respondents who are subject to an endowment effect are sophisticated about how such an effect will influence their future loan take-up decisions. However, this survey module will only be completed by a randomly selected subset of all respondents. Randomizing whether respondents are asked to make these predictions will allow us to measure the extent to which having to predict future decisions influences actual future decisions. If the loan prediction module does influence responses in the later loan take-up module, having a group of respondents who did not make these predictions allows us to measure loan demand without such a prediction effect. For a randomly selected half of the respondents who participate in this module, we will offer a small incentive for accurate predictions. Additionally, to estimate the magnitude of the endowment effect over the initially endowed item, we will elicit willingness to accept for the endowed item during the second session, after respondents have owned the item for several days. A difference between willingness to pay for the item in the first session and willingness to accept for the item in the second session indicates that the respondent experiences an endowment effect over the item. Again, we measure sophistication over such an effect by eliciting predictions in the first session, before the item is endowed, of future willingness to accept in the second session. Again, these predictions will be incentivized for a randomly selected half of respondents, as in the loan willingness to pay predictions. Finally, we measure individual-level loss aversion, which may drive the endowment effect, by asking respondent to accept or reject a series of six small-stakes gain-loss lotteries with varying (positive) expected value. We then select one of the six lotteries at random, and implement that lottery if the respondent previously chose to accept it.
Randomization Method
The randomization of goods is conducted in SurveyCTO using randomly generated numbers.
Randomization of prices in BDM procedure done by a draw of lots by the participant themselves
Randomization Unit
Individual-good level
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
700 individuals
Sample size: planned number of observations
700 individuals
Sample size (or number of clusters) by treatment arms
We do not have treatment "arms" as such, since each of the 700 individuals will have one good randomized to their endowment, one good offered for sale, and will reveal their willingness to pay for two loan types.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
KSh 80 (US$0.77) difference in willingness to pay for the two loans.
IRB Name
Harvard University-Area Committee on the Use of Human Subjects
IRB Approval Date
IRB Approval Number
Post Trial Information
Study Withdrawal
Is the intervention completed?
Is data collection complete?
Data Publication
Data Publication
Is public data available?
Program Files
Program Files
Reports, Papers & Other Materials
Relevant Paper(s)