An Experimental Evaluation and Land Lottery in Nepal

How to cite

Holland, Abraham. 2018. "An Experimental Evaluation and Land Lottery in Nepal." AEA RCT Registry. AEARCTR-0002664. https://doi.org/10.1257/rct.2664-4.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Abraham Holland*

* Primary investigator

Registry ID
AEARCTR-0002664
DOI
https://doi.org/10.1257/rct.2664-4.1
First registered
Last updated
Status
On going
Trial dates
2017-07-18 – 2020-04-30
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/2664

Abstract

What is the role of community preferences in driving the viability of village relocation programs following natural disasters? Nepal’s post-2015 earthquake village resettlement program, an effort to shift high-risk communities to safer, previously unsettled, locations provides a setting in which to examine this question. A key challenge in this program is how to fairly and transparently allocate new private residential space to households, while maintaining individual and overall community satisfaction with the relocation effort. In response, the government plans to administer land lotteries in affected villages. This paper focuses on understanding potential drivers behind the popularity of different lottery types and seeks to understand its implications for community engagement efforts supporting the plot allocation process.

Experimental design

There are two randomized instruments that are central to the study's design. The first gains experimental traction in this situation by leveraging the insights of previous field experiments on social preferences. In particular, randomizing whether individuals make decisions in a private (anonymous) or public (non-anonymous) context can identify the effects of external social concerns on decision-making. To accomplish this goal, I use a `lab in the field' Lottery Preference Game (LPG).

The LPG is a private-value cash lottery game where groups of participants are randomized into anonymous or non-anonymous contexts. They are asked to choose which type of lottery they would like to play, either a `Box-Choice' Random Dictatorship (RD) or a `Box' lottery, and the entire group plays the lottery type preferred by a randomly-selected individual. This ensures incentive compatibility and causally estimates anonymity's effect on preferences. The LPG group composition is also randomly determined. With this in mind, group composition / individual characteristics are also a potential source of experimental variation. Pending additional funding and sufficient statistical power estimates, an additional treatment arm may be added to this experiment. In that event, this submission will be amended.

The second experiment is conducted by randomizing the order of information in a household-based Lottery Preference Survey (LPS). This survey solicits individual preferences over a set of three possible lottery allocation schemes for managing residential plot distribution. Within this survey there are two distinct sources of experimental variation. The first is whether potentially relevant information on possible preference heterogeneity within their community is presented before or after the enumerator has solicited their preferences for the three lottery types. The second source of experimental variation is in randomizing both the order of explanation and pair-wise rankings solicited prior to asking for a complete preference ranking over all three lottery types.

A third experiment based on the random variation in household location induced by actual land lotteries conducted in these communities is currently being submitted for financial support. Pending successful funding, there will be a separate registry entry created for that experimental study component.

Design details

Randomization method
Randomization was accomplished by a computer/tablet.
Randomization unit
The randomization for the LPS is done at the household-level. For the LPG, participants are randomized at the household-level into groups and then each group (e.g. each set of 3 game participants) is randomized into either an anonymous or non-anonymous game context. Additional randomization of the prize value and display order, as well as a further randomization of the choice order on the LPG player response forms is also implemented at the player and group-round level.
Clusters
In the case of the LPG, the planned number of clusters is 210. This may change depending on pending funding applications. In this event, I will update this application.
Observations
For the LPG the planned number of observations is 640 individual participants. For the LPS, the planned number of observations is 650 households/respondents.
Arms
The sample size by treatment arm is expected to be balanced. For the LPS this would mean ~325 households/treatment arm and for the LPG this would be 105 groups/treatment arm.
Primary outcomes
The key outcome of interest in both the LPS and LPG experiments is support of choice-based lottery mechanisms. In the LPG case, this is a Random Dictatorship (RD). In the LPS case there are actually two types of Random Serial Dictatorships (RSDs). Both the general level of support for RSDs (choice-based approaches), as well as relative preferences is an outcome of interest. Additionally, this outcome is of interest both inter-personally (across treatment arms), but also intra-personally (within the LPS control arm). Similarly, within the LPS respondents are also given the option of a purely administratively-driven non-lottery allocation mechanism. The degree of support for this administrative approach is also a key outcome of interest.
Secondary outcomes
A secondary outcome of interest in the LPG is whether or not the anonymity protocols functioned appropriately, to this end we collected several outcomes related to whether or not the players knew one another. Other additional outcomes include respondent self-reported and assessed understanding of the game, whether the respondents viewed the pre-game training as helpful, whether respondent choices over prizes were payoff maximizing, whether player preferences over prizes differed between the two lottery approaches, and whether respondents reported knowing people caused them to behave differently. Within the LPS, secondary outcomes of interest include evidence of strategic voting, lottery type comprehension, support for voluntary land exchange, support for possible land exchange services, willingness to pay to improve their land allocation, reported confidence in choosing `good' land, willingness to move to the new location, and whether or not the respondent thought the preference heterogeneity intervention was potentially applicable to their context. A third survey instrument, a Location Preference Survey (LCPS) provides another set of outcome variables. This instrument solicits the respondent's preferences over potential locations in the proposed (new) settlement. As such, another set of secondary outcome variables that are of interest relate to the number and types of plots households find desirable, as well as the justification/explanation(s) for why a given household prefers/dislikes a particular plot.

Cited by

No citation data available for this registration yet.

Keywords