A closer look at food assistance: Examining our work with the World Food Programme

Families worldwide are struggling to cope with a devastating hunger crisis that continues to escalate. The conflict in Ukraine has compounded the emergency, driving up the costs of fertilizer and wheat that many depend on to survive—especially children and families who are already hungry, those living in regions where survival is already precarious.

Hunger is an issue that World Vision has been focused on for a very long time. For more than 30 years, we’ve partnered with the UN World Food Programme (WFP)—including 18 years as their largest implementing NGO partner¹—providing food and cash assistance for the girls, boys, women and men who need it most through programs including general food distributions, cash-based programming and integrated school feeding, among others.

In fiscal year 2022, World Vision provided 67,690 metric tons of food and $68 million in cash assistance, reaching 3,836,601 people in 18 countries through our partnership with WFP. Given the gravity of the hunger crisis and our priority in this area, we’ve begun an in-depth analysis of our food assistance portfolio to better understand its effectiveness and efficiency.

Analysis of our food assistance portfolio began in 2022 and is ongoing. Through this work we’re focusing on two measurements:

1. COST-EFFECTIVENESS ANALYSIS. This method investigates a program to understand how much it costs to achieve a certain result.²
2. COST-BENEFIT ANALYSIS. This method goes a step further and asks whether the economic benefits of a program outweigh its costs.

To understand how much it costs World Vision to deliver food programming, we first determined the average cost per participant and compared that cost across different regions and program areas where we work.

World Vision’s food programs are carried out in diverse settings, each with their own challenges that affect the cost of implementation. Geography, infrastructure, local markets and security concerns all play a role in the resources required to bring assistance to a given community. The average cost per participant is a helpful baseline, allowing us to pinpoint and investigate notable outliers so we can identify challenges and assess areas for improvement.

We approach this metric with caution, because our goal is not to only provide food assistance wherever or however it’s cheapest. World Vision is committed to reaching the world’s most vulnerable girls and boys, which may require going to more remote communities in more fragile settings, where the costs are often higher. Even so, analyzing cost-effectiveness helps to ensure our costs are in line and provides a metric to track over time.

Our analysis revealed that the average monthly cost to provide food for a family of five living in a fragile context was $66 over the past six years. This includes both the food and cash distributed, along with all operational costs attached to the programs. To see how this figure is trending over time, we calculated moving averages over that period, from 2017-2022.

The results were surprising, showing a decrease from $75 to $61 over the six-year span. We expected to see these costs increasing over time, not decreasing, because global inflation and the conflict in Ukraine, among other factors, are believed to be increasing the cost of food assistance worldwide.

So why would our numbers not follow the same trend? Unfortunately, we don’t have the answer yet, but we are committed to continuing our investigation.

What do we currently know?

1. This data represents only fully completed projects. This means projects currently in operation are not part of the sample. This does bias the trend—particularly the final two data points—as unfortunately, there are still projects pending completion from those years. If this is indeed the cause of the bias, we may not be able to include most recent years for trend analysis.
2. These numbers represent what we call “fully loaded costs,” which include everything from the food and operational costs all the way to administrative and fundraising efforts.
3. “Provide food every month” does not equal “feed a family completely every month.” Some programming types—like general food distribution—lean toward covering a family’s food needs but others, such as targeted vulnerable group feeding or integrated school feeding, take a different approach. Until we have a full breakdown of the data, we won’t be able to provide a more precise statement.
4. Over the past two years, we have been shifting more of our food portfolio toward cash assistance—this programming type has many benefits such as greater efficiency, flexibility for families and even anecdotally-reported safety, compared to in-kind food distribution. Because our raw data lacks the level of detail required to run an in-depth analysis of all programming types, we have not yet been able to understand the full implications of this shift.
5. Logistical disruptions, known as pipeline breakages, affect aspects of our programs such as the amount of food distributed and costs of operation, along with a variety of known consequences that are poorly documented in the database. This hinders our ability to understand the cost-efficiency implications.

More investigation is needed to understand these results, particularly as we seek to be data-driven in our partnership with WFP and accurate in our forecasting of food assistance costs across contexts.

Between 2019 and 2021, World Vision ran 174 joint programs with WFP across 24 countries. Using the data available from this span of time, we are working in partnership with a third-party analytics firm to conduct a cost-benefit analysis.

Preliminary results show that overall, World Vision’s food assistance portfolio has been an effective use of funds, bringing a significant positive impact and value.

• The benefit-cost ratio is 1.97, which means for every $1 spent on WFP programming, $1.97 in health and educational benefits were produced between 2019 and 2021.³
• 8,546,943 people have been helped through our food assistance portfolio, at a cost of $304,551,463.
• 169,276 disability-adjusted life years (DALYs) have been prevented, at a cost of $1,795 per DALY saved.⁴
• With a net present value of $295,598,142, we know this portfolio has provided more benefits than costs.⁵

Analysis has also revealed that over the past three years, our programs run in partnership with the UN World Food Programme prevented the deaths of 10,841 girls and boys under age five—nearly 10 lives per day.

1. We need to track the risk of pipeline breakages to prevent delays in delivering food assistance.

Within our food portfolio, we know that there are instances of pipeline breakages—disruptions and challenges that prevent us from delivering assistance as planned.

While conducting the cost-effectiveness analysis, we looked to see if these breakages would lead to outliers in the dataset. Early results indicate that this is happening, but further analysis is needed to understand the scope of the issue, and to what degree it is affecting the quality of our interventions.

2. We need to investigate the decreasing cost of food delivery seen in our cost-effectiveness analysis.

This will be challenging and require multiple steps. Our food assistance portfolio is extensive and complicated and our analysis to date has multiple variables, making it difficult to pinpoint issues quickly. For example, the structure of our data does not allow us to easily separate the different programming types for analysis, which is a necessary step in this investigation (see point No.3).

3. We need cost data that is disaggregated by programming type.

We use five main approaches through our partnership with WFP: general food distribution, cash and voucher-based programming, food or cash for assets, targeted vulnerable group feeding and integrated school feeding. Based on the way financial data is tracked, it is difficult to separate the costs associated with each programming type when multiple types are being used in the same geographical region, because of overlaps.

Having cost data per programming type is needed for more in-depth analyses of the cost-effectiveness trend as well as the efficiency of the specific interventions. For example, targeted vulnerable group feeding accounts for a significant portion of benefits; however, we need the cost of these specific programs to determine if they are more cost-effective at reducing malnutrition compared to other interventions, such as general food distribution.

4. Having the change in food consumption scores at the project level would allow us to generate more accurate results.

The food consumption score (FCS) is an index developed by WFP—it aggregates a household’s food group diversity and frequency, weighted according to nutritional value, and helps to classify households according to their caloric intake. It is one of the main metrics used to estimate the portfolio’s impact reported here.

We do not currently have FCS at the project level and instead, have relied on assumptions grounded in country-level data. The analysis showed that even when we changed the current conservative assumption of the FCS impact to a very conservative assumption (thus reducing the estimated impact), the broad conclusions from the CBA results did not change—the benefits were still greater than the costs.

Even so, using an assumption does limit the accuracy of the analysis we can perform. Moreover, with project-level FCS data, we could assess the relative effectiveness of cash transfers and compare them to the effectiveness of general food distributions. We could also consider regional or country variations, allowing for more knowledge about our implementation strengths, and propose shifts of the portfolio where applicable.

These cost-effectiveness and cost-benefit analyses have provided valuable insights, revealing strengths in our food assistance programming while highlighting areas for follow-up.

Going forward, we will:

1. Analyze our food assistance portfolio to identify issues that may be skewing our cost-effectiveness trend lower, such as: outlier projects, changes in the mix of programming types being used, and the implications of our increased investment in cash programming.
2. More thoroughly investigate the consequences of pipeline breakages and how they may affect calculations and benefit accrual going forward.
3. Separate cost data for projects that implement multiple programming types, so we can assess the effectiveness of each intervention.
4. Perform a sensitivity analysis on the CBA models’ critical parameters, to determine how the key assumptions affect the results.
5. Perform a gap analysis to determine additional indicators and research areas that could be collected, to better inform cost-effectiveness analysis or program evaluation for this portfolio. 

¹ Through the global World Vision Partnership

² While WFP does report this value at a high level, World Vision requires more granular detail to understand unique contexts and account for total costs.

³ The benefit-cost ratio is a way to compare the total benefits and total costs of a project. A value greater than 1 means that the benefits outweigh the costs.

⁴ The disability-adjusted life year (DALY) is a measure of the overall burden of disease. It combines years of life lost due to premature death and years of life lost due to time lived in less than full health. One DALY represents the loss of the equivalent of one year of full health.

⁵ Net present value is a way to determine how much money something will make or cost in the future by accounting for the time value of money.

⁶ While mortality refers to death, morbidity is the condition of having a disease or illness.