As an organization that values iteration, we very much welcome feedback and comments on our impact model and impact approach - Please email us with any feedback, comments, or questions at: [email protected]

What is Relative Income Change?

Watch this video explaining the model

Relative income change is the difference in a person’s annual earnings relative to their current annual earnings. Relative change is expressed as a percentage change, rather than an absolute dollar amount. An example is a skills training program in Kenya that increases its graduates’ annual incomes by an average of 30 percent relative to what they would have earned without the training program.

Why Use a Relative Income Change Model?

At the GitLab Foundation, we believe that conducting a cost-benefit analysis is one crucial step in determining whether a grantee should receive our support, in order to ensure our dollars are used for maximum impact.

Currently, our cost-benefit model defines ‘benefit’ in absolute dollar increases in income, as outlined in our North Star ROI methodology. This model quantifies income gains as the total present value income gain that is generated due to the Foundation’s grants. The absolute ROI model is indifferent to the starting income level of those impacted. A grant that increases someone’s income by $1,000 per year results in the same quantitative benefits regardless if that person’s baseline level of income is $2,000 per year (Kenya GNI per capita) or $75,000 per year (US GNI per capita). Due to the Foundation’s international work, the model may unintentionally make it easier to work in higher income countries.

The relative income change intentionally focuses the impact of income gains relative to the starting level of a person’s income. The additional $1,000 for someone earning $75,000 a year (1.3% increase) is not as transformative as it is to someone earning $2,000 a year (50% increase). This enables the Foundation to incorporate the law of diminishing marginal utility. This model values a 1% change in income equally, regardless of baseline income. This removes the bias of focusing only on projects that increase absolute dollars, as a 1% change in income will naturally have a larger absolute increase for countries that have higher incomes as this graphic demonstrates:

https://lh7-us.googleusercontent.com/f60-yyErH547q6XlNJXXSgNRoAB9n69-ACh7uVgrhtdhdqIg4p2b9OtEHFPf0T_9CMFFGUkdKqAiM2zfoasa0BCQ51r123UgUJOZYRwprhs_E6cmi-yFTgDiNbfJc1celS5wQzCUecjUMWuWApA-e8U

This relative income change methodology is inspired by the Global Innovation Fund’s methodology as well as IDInsight’s impact model. GiveWell’s cost-effectiveness model also quantifies income changes as a relative percentage change difference, albeit using a more complex methodology.

Relative income gains also provide an easier yardstick (as defined by Global Innovation Fund) to quickly determine the depth of impact that a grantee creates for its clients:

• Annual income changes under 5% are considered perceptible
• A 50% increase in income is considered substantial
• A 100% or more increase in income is considered transformative

A RIC model focuses the Foundation’s grantmaking on organizations that are the most cost-effective at increasing participants’ relative income. We believe, all else equal, it is more important to increase someone’s relative income than their absolute income. Our mission is to improve the most lives per dollar. We started with an absolute income change model as our work began in the United States with a similar population focus. As we have expanded to work in Kenya and Colombia, using an absolute income model creates an unintentional bias which is why we are exploring this alternative model. (See Appendix Table 1 for a detailed example.)

How Do We Model Relative Income Change?

Estimating Percentage Increase in Income:

1. The relative income change model estimates how much income will increase as a percentage relative to what an individual would have earned without the grantee's services. This calculation is determined during the grant application phase after the Impact Measurement & Analytics team reviews a grantee's impact predictions. This percentage income change is translated into “Relative Income Points” per person.
   1. Think of one “Relative Income Point” as meaning one person achieved a one-year income increase equal to 1% of their comparable annual income. If one person increased their annual income by 10% for one year, their total Relative Income Points would be 10. If ten people each increased their annual income by 1% for one year, the total Relative Income Points would also be 10.
   2. The total Relative Income Points for a grant are the total Relative Income Points per person multiplied by the number of people impacted by the grant and the number of years that the people are experiencing the increase in income.

Reference Table:

Calculating Relative Income Change - the “DIL” Threshold

The relative income change metric is the total cost of the grant, divided by the total Relative Income Points, multiplied by the total number of people impacted by the grantee’s program. This results in a cost-effective dollar amount: the cost to increase income by 1% for one person for one year.

This cost-effectiveness metric can be translated into easier-to-understand ratios. For example, when multiplied by 30, the ratio is the cost to increase someone’s income by 1% for their working life (30 years). If we multiply that value by 100 we get the cost to increase one person’s lifetime income 100%, which we are calling “Double Income over Lifetime,” or DIL. While not every grant will increase participants’ incomes 100%, translating the cost-effectiveness ratio to the cost to “Double Income over Lifetime” creates a useful threshold that can be quickly applied when comparing potential grants.

Note, that we are suggesting a 0% discount rate to simplify the calculation and improve the adoption of this model by other philanthropic organizations. We also believe we should be indifferent about when individuals increase their incomes as it is important to improve incomes throughout one's lifetime.

These cost-effectiveness values can be seen in this cascading calculation for a $500,000 proposed grant:

1. Income Change for One Year per Person = 50% = 50 points
2. Income Change for Lifetime (30 Years) per Person = 50 * 30 = 1,500 points
3. Income Change for 1,000 People for Lifetime = 1,500 points * 1,000 people = 1,500,000 total Relative Income Change Points
4. Cost to Increase Income 1% for One Year per Person = $500,000 / 1,500,000 = $0.33
5. Cost to Increase Income 1% for Lifetime per Person = $0.33 * 30 = $10.00
6. Cost to Increase Income 100% for Lifetime per Person (DIL) = $10.00 * 100 = $1,000

Calculation for the Cost per DIL (Double Income over Lifetime):

Setting A Cost-Effectiveness Threshold

Similar to our North Star goal of generating at least $100 in lifetime earnings increases per $1 dollar spent, we intend to set an ambitious threshold for the RIC model. A preliminary literature review finds that unconditional cash grants, a common philanthropy benchmark, have a DIL of $1,500-$4,000. We have also re-run our predictions for our current grantee set (as of early November 2023) using the RIC model, and found that their median DIL is $1,300, with a mean DIL of $1,600 (see Figure 1, below). Given these data points, our early recommendation is to set our RIC model cost-effectiveness threshold at $1,000 DIL or less.

Figure 1

Cost per Person Depends on Grant Impact on Income

Potential grantees may ask whether their programs must double incomes, or have a cost per person of $1,000 or less, in order to be eligible for GitLab Foundation funding. The answer is no. Rather, the DIL methodology allows flexibility in the cost per person impacted, depending on the percentage of income change predicted in a grant proposal. The lower the lifetime income increase, the lower the cost per person threshold, as shown in the table below:

Our Test-and-Learn Plan

We will not move to using this new impact estimation model immediately; rather, we will research how this new cost-effectiveness model could affect our future grantmaking portfolio, and other effects it could have.

We plan to select grants using both our absolute ROI model and this relative ROI model in 2024, doing side-by-side comparisons of results to understand both the impact and operational implications of changing models.

Some key questions we hope to address during the testing phase include:

1. How easy or difficult is it to communicate Relative Income Change externally and internally, in comparison to our current absolute income model?
2. What are the Relative ROI results for well-known development interventions around the world?
3. How do Relative ROI results differ by high- and low-income countries or populations?
4. How much overlap is there between effective Relative ROI programs and effective absolute North Star ROI programs?
5. Would we positively impact more people with a Relative ROI model?
6. What processes might we need to change, and what capabilities might we need to build, in order to effectively source and vet grants using the RIC model?

We will know that the Relative Income Model is successful if:

1. The model is determined to be easy to communicate internally and externally
2. The model is easier to integrate into application, due diligence, and reporting processes
3. Grantees find the model easier to interpret and report against
4. The model enables us to source a satisfactory number of pipeline opportunities in a variety of countries, regardless of baseline income level
5. The model enables the Foundation to create more transformational change for more people

Appendix

Presentation Link

Video Presentation Explaining the Relative Income Change Model

Acronyms

DIL - Cost to Double Income over Lifetime

RIC - Relative Income Change

RIC Model - Relative Income Change Model

RIP - Relative Income Point

Table 1

The table below demonstrates how two different skills training programs that both receive a $500,000 grant ($615,000 total cost including overhead) from the GitLab Foundation would be prioritized differently depending on if the return on investment model was based on absolute or relative income change:

Using absolute income change analysis results in the U.S-based grant increasing lifetime income $115 for every $1 spent by the Foundation. The Kenyan grant only increases lifetime income by $92 for every $1 spent.

In contrast, using relative income change analysis results in the Kenyan-based grant doubling a person’s lifetime income for $103, which is 24 times more cost-effective than the U.S.-based grant, which costs $2,460 to double a person’s lifetime income.