GiveDirectly Update - May 2013

Published: June 2013


Plans for future transfers

  • GiveDirectly currently has $3 million in available funds.
  • It intends to use $1 million to set up operations in a new country and begin providing cash transfers to recipients in that country. It started transferring funds in May 2013.
  • It intends to use $2 million to support additional cash transfers in Kenya, the country in which it has operated since its inception. It will distribute some of these funds on a village-level, i.e., to all residents of a village instead of selecting only residents living in mud and thatch-roof houses. The rest of these funds will be targeted at residents of mud and thatch-roof households.
  • GiveDirectly projects it will have the capacity to use $10 million in 2013.

Progress with ongoing transfer sets

  • GiveDirectly has continued to provide cash transfers to recipients in ongoing transfer sets. It has distributed approximately $616,000 thus far in 2013.
  • GiveDirectly continues to follow up with recipients to identify potential problems and understand how recipients report using funds. It has shared this data with us; we discuss the data below.

Funds available

As of March 2013, GiveDirectly had $3 million available, $2 million of which it had designated for future transfers in Kenya, and $1 million of which it had designated for use in scaling its model to a second country. It expects both the additional Kenya transfers as well as the second country operations to begin around mid-2013.1

GiveDirectly projects that it will have the capacity to conduct approximately $10 million-worth of transfers in the next fiscal year (more).2

Future plans

Progress in planning future sets of transfers


GiveDirectly has allocated $2 million of its available funds to additional transfers in Kenya that are planned to begin in mid-June 2013.3

It has not yet selected villages for these transfers, but told us that it will do so following a similar process as was described in our full review. GiveDirectly noted that if it conducts an urban pilot, there would be additional considerations made in choosing locations.4

Some of the $2 million of future transfers will follow GiveDirectly's traditional model of targeting mud and thatch-roof households, but for the rest of the funds, GiveDirectly told us that it is planning to try a new model whereby it will enroll all households in a village to receive transfers. GiveDirectly also said that it is planning to randomly select future villages in Kenya out of the pool of eligible villages, so that it will be possible to design a cluster-RCT of the village-level transfers, if there is sufficient funding.5

If future transfers are made to villages in Western Kenya, GiveDirectly has said that the transfer size and schedule will remain the same ($1000, disbursed in an initial, small transfer and two larger transfers), but if the region changes or urban areas are selected, the transfer size and schedule will be re-assessed.6

Second country

GiveDirectly has allocated $1 million to provide cash transfers in a second country. This funding came from a $2.4 million Google Global Impact Award that GiveDirectly received, of which $190,000 was designated specifically to underwrite the fixed costs of setting up operations in a second country.7

GiveDirectly told us that in choosing the second country, it considered whether there was a mobile money provider accessible to the very poor, how costly it would be to operate in the country, how politically stable the country is and how common corruption is in government affairs. The ease of moving staff between Kenya and the second country was also a factor, as GiveDirectly's current COO will be overseeing the work in both places.8

As of May 2013, GiveDirectly had identified a second country, but asked that we keep the name of the country confidential until it has been formally announced. GiveDirectly expects to be registered as an NGO in this country by the end of June 2013. GiveDirectly has selected a district and sub-county within the country for its first set of transfers. In choosing this location, GiveDirectly considered poverty rates/density, security, proximity to Western Kenya, cell phone network coverage, and mobile money penetration.9 As of June 2013, GiveDirectly was planning to target only mud and thatch households in the second country, though noted that this may change if experiments with whole village transfers go well.10

GiveDirectly reported that it is determining the size of transfers for the second country by comparing PPP-adjusted consumption data for mud and thatch households between Kenya and the second country. If these figures are roughly equivalent, GiveDirectly will likely use the same transfer size ($1000). GiveDirectly also said that it will determine the schedule for transfers based on pilot results, specifically regarding the accessibility and protocols of the country's mobile payment system.11

Expanding staff capacity

GiveDirectly has hired 2 additional staff members to help run its U.S.-based operations:12

  • Joy Sun (bio13) was hired as the Director of Strategic Initiatives, a full-time position that involves keeping GiveDirectly organized and legally compliant, as well as working with the GiveDirectly board.
  • Sharon Harvey was hired as Outreach Coordinator, a part-time position that involves communications with donors.

GiveDirectly told us that it is planning to hire 2 new staff members to manage its operations in the field: one to work in Kenya and the other in the second country. GiveDirectly expects that these field managers will each have the capacity to manage $5 million-worth of transfers per year, with the support of the current COO, who will split her time between countries. This would bring GiveDirectly's total capacity to approximately $10 million-worth of transfers in the next fiscal year.14

Room for more funding

In March 2013, GiveDirectly told us that its projected capacity for fiscal year 2013 was $10 million, and that it currently had access to $3 million.15

While we believe that GiveDirectly's goal is ambitious, we would feel comfortable with it reaching its goal of raising $10 millon in 2013. Nonetheless, we plan to carefully watch GiveDirectly's ability to grow.

Our reasoning:

The Google cohort is GiveDirectly's most recent transfer set, and it conforms to a "standard" model: mud and thatch households were targeted; transfer size is $1000; and, there are no experimental components.16 This makes it a useful proxy for GiveDirectly's ability to move funds. In the Google cohort, GiveDirectly enrolled 850 recipients17 and has transferred approximately $385,00018 to them thus far; it intends to complete most transfers in August 2013.19 Most of the needed staff capacity is utilized relatively early in the distribution process (selecting villages and enrolling recipients), so most of the work necessary for the Google Cohort is complete. This required the COO and field staff to spend several months. Assuming GiveDirectly is able to hire field managers who can play a similar role to the one the COO played in the past, it is not unreasonable to project that each staffer should be able to manage 5 distributions to 1,000 recipients in the course of a year.

In addition, GiveDirectly intends for some of the future transfers in Kenya to be village-wide rather than targeted at mud and thatch-roof households, which will reduce the steps required for household identification.20

Plans for future research

GiveDirectly is interested in conducting a cluster RCT to study the effects of providing cash transfers to all households in a village. It currently has a $1 million commitment for this kind of study, as well as $300,000 from GiveWell donors that is available for this use.21

GiveDirectly also received a $30,200 grant to extend data collection using mobile phone-based data collection techniques in a sub-sample of participants from the Rarieda RCT. The grant is from USAID via the Policy Design and Evaluation Lab at UCSD. The goals of the project are to generate data on longer-term effects of cash transfers (up to 2 years after completion of the RCT) and how treatment effects unfold over time, as well as to study the potential for using mobile phones as cost-effective, easily adaptable tools for data gathering.22

Progress of ongoing transfers

Status of ongoing transfers

Cohort Number of recipients Dates of transfer process23 Size of transfers ($) Current status (as of 4/26/2013)
Rarieda, group A 31224 May 201125-January 2013 (RCT);26 March 2013-ongoing (follow-up to RCT)27 300 + 70028 RCT data collection is complete, $300 treatment group recipients are receiving "top ups" of $700 sent in two installments. 87% of these recipients received first installment of $350 in March 2013, second installment scheduled for May 2013.29
Rarieda, group B 14030 May 201131-January 2013 (RCT)32 1000 89.9% of recipients had received full transfers by January 1, 2013.33
Rarieda, control group ~500, not yet enrolled34 Has not begun35 1000 Has not begun36
Siaya 19937 June 201238-May 2013 (projected)39 1000 198 out of 199 recipients will have received full transfers by early May 201340
Google (referred to in full review as "Siaya II") 85041 October 201242-end date depends on ongoing participant enrollment43 1000 ~$385,000 has been sent to a total of 809 households44
Nike, group A 3945 July 201246 - February 2014 (projected)47 500 ~$16,000 has been transferred48
Nike, group B 3849 July 201250 - February 2014 (projected)51 1000 ~$19,000 has been transferred52

The above table reflects the status of ongoing transfers as reported to us by GiveDirectly on April 26, 2013. GiveDirectly also sent us disaggregated transfer data for Rarieda group A (top-ups only), Siaya, Google, and Nike (both groups). The disaggregated data sets support the claims made by GiveDirectly about the status of transfer sets.53


According to GiveDirectly's updated financial report from March 2013, the costs of its transfer sets have largely been in line with what GiveDirectly projected in August 2012 (cited in our original review); no large, unexpected costs have arisen. Total projected spending slightly increased for Siaya and Nike cohorts,54 and is slightly lower than expected for Rarieda.55

On average, GiveDirectly has met it's target of transferring at least 90% of funds directly to recipients.56 In the Nike set, GiveDirectly fell short of this target, projecting that it will transfer 80.3% of funds for the project directly to recipients.57 GiveDirectly noted that enrollment costs were higher in this set because it was targeting a narrower demographic (18-19 year old girls). This targeting is unique to the terms of the Nike grant; it does not match GiveDirectly's standard process.58

How transfer funds have been spent

GiveDirectly sent us spending data from follow-up surveys in the Siaya, Nike, and Google transfer sets in April 2013. As these transfer sets were not complete by April, the reported funds spent account for only part of the total planned transfers in each cohort (Siaya data represents 70.1% of total funds, Nike data represents 35.6% of total funds, Google data represents 11.1% of total funds).59

Below we provide aggregated spending statistics based on the disaggregated data for common spending categories.60 For more, see the documents listed in the footnotes for each transfer set.

Note that the units of measurement used for the first table are survey responses, not recipients (surveys are given after each part of a transfer is sent, so data may be gathered from the same recipient more than once). The units of measurement used for the second table are reported amounts of funds spent. All of the survey data is based on self-reports.

Percentage of survey responses that reported spending in common spending categories

Transfer cohort Siaya61 Nike62 Google63
% of total transfer funds represented by data 70.1 35.6 11.1
Spending category
Iron Sheet/Home 75.1 63.5 90.2
Food 38.9 34 39
School 19.6 29.4 29.8
Livestock 29.8 9.1 30.4
Clothes 16.4 21.3 21.2
Non-farm business 14.3 17.8 10.1
Farm business 5.6 6.1 18.9

Percentage of self-reported funds spent in common spending categories

Transfer cohort Siaya64 Nike65 Google66
% of total transfer funds represented by data 70.1 35.6 11.1
Spending category
Iron Sheet/Home 55.1 44.6 72.7
Food 3.9 6.4 1.7
School 3.7 13 2.1
Livestock 13.2 3.9 4.4
Clothes 1.6 3.2 1
Non-farm business 6.5 8 1.7
Farm business 1.3 1.3 1.5
Miscellaneous67 14.7 19.6 14.9

Issues in transfer process

GiveDirectly sent us data from follow-up surveys in all transfer sets pertaining to the quality of its process and issues that arose. As these transfer sets were not complete at the time that we received the data, the survey responses are based on partial transfers only.

Below we provide aggregated statistics for a sampling of issues. For more, see the documents listed in the footnotes for each transfer set.

The data below provides the number and percentage of recipients who reported a given issue in any of the follow-up surveys that had been conducted at that time.

Issue Rarieda (RCT) Siaya Google
Oct 201268 May 2013, top-ups only69 Oct 201270 April 201371
Had collected transfer at time of survey72 96.2% (454/472) 96.7% (291/301)73 100% (156/156) 92% (207/224)
Had trouble collecting transfer74 7.5% (31/412) 0% (0/291) 0.6% (1/156) 1.3% (3/225)
Regrets own spending75 1.3% (5/384) Data not collected 0% (0/152) 0% (0/223)
Reports tension/conflict within household76 0.9% (4/463) Data not collected 0% (0/150) 1.8% (4/223)
Reports hearing complaints in community77 35.7% (165/462)78 Data not collected 44% (70/159) 14.7% (33/225)
Reports shouting or angry arguments among people in village79 0.9% (4/458) 0.3% (1/299)80 0% (0/150) 2.7% (6/223)
Reports violence or crime81 0% (0/458) Data not collected 0% (0/151) 2.2% (5/223)
Reports feeling threatened82 Data not collected Data not collected 0.7% (1/137) 0.4% (1/223)
Reports M-PESA agent demanded bribe83 0.5% (2/416) 1.3% (4/299)84 0% (0/151) 0% (0/223)

Note: the table below is separate because the units of measurement for this data are not recipients, but survey responses. Surveys are given after each main part of a transfer is sent, and in these data sets, data is presented from the same recipients on more than one occasion.

Issue Siaya, April 201385 Nike, April 201386
Had collected transfer at time of survey87 99.1% (348/351) 93.4% (197/211)
Had trouble collecting transfer88 2.3% (8/343) 5.3% (11/209)
Regrets own spending89 0.9% (3/337) 1.5% (3/203)
Reports tension/conflict within household90 0% (0/340) 0.5% (1/208)
Reports hearing complaints in community91 45.3% (159/351) 15.6% (33/212)
Reports shouting or angry arguments among people in village92 3.6% (12/338) 1.4% (3/208)
Reports violence or crime93 1.2% (4/341) 0.5% (1/208)
Reports feeling threatened94 1.8% (6/327) 1.4% (3/208)
Reports M-PESA agent demanded bribe95 0.6% (2/341) 0% (0/208)

Anecdotal challenges and issues GiveDirectly has reported:

  • In one village, the village elder visited recipient households, falsely purporting to be collecting ground nuts on behalf of GiveDirectly.96
  • GiveDirectly staff verify recipient information at multiple stages and record any issues. GiveDirectly has shared this information with us in the enrollment databases for all transfer sets. Examples include:97
    • "His GD line was [#]. I remember him mentioning to have registered this number and now he says he tried registering and was told it was already registered. His new GD line is [#]."
    • "Has no ID card. Still has a waiting card. Expects ID card this December. To register when she gets."
    • "[Name] (female) initially was assigned the transfer. Later found out that she was underage. UCT knew this but accidentally included her in the randomization. Changed to male"
  • GiveDirectly also noted multiple challenges faced in conducting follow-up phone surveys:
    • In a few cases, staff are unable to collect spending data because funds have not been spent by the time survey is conducted, or recipients do not remember how funds have been spent (usually very elderly recipients who had relatives make purchases on their behalf).98
    • In a few cases, GiveDirectly has incomplete follow-up survey data, either because calls were dropped, recipients had to leave abruptly during a call, or staff made errors with data entry.99


Clicky Web Analytics