13 April 2021
The catalysing effect of innovations can produce enormous impact, especially if incubated at a relatively early stage. While a successful innovation will eventually require many years and attract many funders in achieving its full scale, GIF’s strategy is to help the innovation get it off the ground and on that trajectory.
In deploying scarce resources, we can rank prospective investments according to the expected impact per dollar to help choose those that are most cost-effective.
In order to effectively determine if the costs and benefits of a potential investment show a good value for the resources invested, GIF and many other funders seek to compare projected results against a standard like unconditional cash transfers to households. This process, known as ‘benchmarking,’ allows us to determine whether an investment will generate more or less impact than simply transferring cash to beneficiaries. This is especially helpful for a patient investor like GIF which is heavily focused on long term impact.
U.S. Agency for International Development summarises the reasons many donors benchmark their results against the projected use of cash transfers, which is that evidence shows cash transfers may have significant positive impacts in an extremely cost-effective way.
To determine the value for money of a potential investment and therefore be able to compare it, GIF starts with the total long-term impact expected to be catalysed (adjusted for risk) and divides the credit for this impact between GIF and any co-funders in proportion to funding to compute our share of impact. Then to determine the catalytic impact per dollar, we can divide GIF’s share of impact by the cost of GIF’s funding.
We compare the ultimate benefit that we expect to catalyse with our investment – recognising that, along the way, subsequent investors will contribute to the cost of scaling up.
GIF’s nearly unique ability to invest in both for-profit and non-profit innovations, and to use grants as well as debt, convertible debt, and equity, complicates the calculation somewhat. A grant may offer larger impact than a loan or equity investment of the same amount, but a loan is likely to be repaid, allowing the funds to be reinvested for further impact.
Below you will find a description of our current approach to determining innovations’ value as compared to the value generated through cash transfers, describing first how we forecast and quantify the impact we expect to catalyse and then second, how we calculate investment cost in a way that puts grants and risk capital on a common footing. This leads to a description of how we benchmark the resulting catalysed impact and cost ratio.
We found that on a portfolio basis, we expect that a dollar of GIF investment will catalyse ultimate benefits to low-income people that are 180 times greater than the benefit that such a person would experience from just receiving an extra dollar.
Our methodology is continually refined and improved, which means the calculations here will continue to be updated.
As part of its Practical Impact methodology, GIF forecasts the annual flow of impact ten years after investment assuming the innovation scales successful, and then adjusts that forecast by the judged probability that the innovation is successful. This yields the risk-adjusted ten-year impact forecast PYI10 , expressed in in GIF’s impact metric units: PYI (person-years of income-equivalent.) (See Box 1 for a brief description.) 
GIF uses a time horizon of ten years because it typically takes that long for an impactful innovation to scale to reach a significant portion of its addressable market. Such innovations will typically continue to expand for some time after that.
For present purposes, we need to impute a long-term time-path of the innovation’s impact, based on the snapshot PYI10. Our assumptions are shown in Figure 1. We assume that
All impact is then discounted back to the present using a discount rate of 10 percent. (This is a conservative, relatively high discount rate compared to most commonly-used social rates of discount.)
Where GIF has co-invested with another funder, GIF is attributed a share of total impact proportional to GIF’s share of funding. However, no impact is allocated to funders who contribute before GIF’s entry or after GIF’s exit.
Box 1 — Practical Impact in a nutshell
GIF builds up its impact measure as follows. Start with the number of low-income people who will benefit, in 10 years, if the innovation scales. Adjust by the depth of benefit, reckoned as the relative gain in consumption or standard of living. Adjust again by the likelihood of success. (See figure below). Do this in a good-enough way, aiming for the right order of magnitude, not spurious accuracy. Revise over time as risks are resolved and impacts measured.
Think of one Practical Impact unit as meaning: one person got a one-time benefit equal to 100% of her annual income (or consumption). This unit is called a person-year of income-equivalent (PYI). If 20 people each received a benefit equivalent to 5% of their annual consumption, that would also be reckoned as one PYI. There are conversion rates for gains in health and education, for reductions in mortality, and improvements in women’s and girls’ agency.
For grants, the cost is the amount of the grant.
Risk capital includes debt, convertible debt, and equity. The starting point for computing the cost of a risk capital investment is the multiple of invested capital:
MOIC = current financial valuation/ total investment amount.
For new investments, MOIC is set equal to 1. Where there is a subsequent financing round, the new valuation may be applied. In the absence of a financing round, we use judgment in determining the holding value of the investment. For debt investments, the value is the amount of the loan plus interest accrued and paid back to date. An adjustment is made if there is a significant chance of default.
The MOIC is not adjusted for the time between initial investment and current valuation and so is not, strictly speaking, a return measure.
Cost =rXT +(1-MOIC)X
The first term, rXT, is the opportunity cost of tying up the capital.
Where MOIC <1, (1-MOIC)X is the currently anticipated loss of capital, and this counts as a cost.
Where MOIC>1, there is a profit, and this reduces the cost.
Ex ante, all investments will have a positive cost. Note that ex post, the cost of an individual investment could be negative if MOIC is significantly greater than 1. For this reason it is preferable to calculate $/PYI for comparing individual deals – where lower numbers represent greater efficiency. At the portfolio level it is possible to aggregate to an overall PYI/$ figure.
We can express investment efficiency in terms of the cost per unit of impact in PYI/$
unit of impact catalysed/cost
To assess the efficiency of a GIF investment, a benchmark is to compare the investment’s social benefits with the benefit that would result from a direct cash transfer with the same cost. This benchmark is facilitated by the definition of a PYI: it is the increase in well-being if a person received, on a one-time basis, additional income equivalent to 100 percent of their annual consumption. By asking how much does it cost to ‘buy’ a PYI via a cash transfer, we can create a benchmark against which a GIF investment (or the portfolio) can be compared.
PYI is a relative measure: it is expressed relative to the current consumption level and therefore the cost of ‘buying’ a PYI depends on the recipient’s consumption level. If we assume a recipient living at $5 per day at purchasing power parity (PPP), and a market dollar per $PPP exchange rate of 0.4, then it would cost $730 to obtain a PYI via cash transfer. Another way of looking at it is to take the reciprocal:
$730/PYI is equivalent to 1 PYI/$730 or .00137 PYI/$
In other words, the benchmark for the portfolio is 1.37 PYI/$1000.
We calculated the portfolio $/PYI ratio using costs for the entire portfolio, including unsuccessful investments that are not expected to generate future impact, and investments which may be impactful but for which it was not possible to make an ex ante impact forecast. Because we include the latter, the calculated ratio is an underestimate.
We found that on a portfolio basis, we expect that a dollar of GIF investment will catalyse ultimate benefits to low-income people that are 180 times greater than the benefit that such a person would experience from receiving an extra dollar.
It’s important to understand the strengths and limitations of this benchmarking analysis. The expected impacts are discounted for risk and for occurring in the future. They are apportioned between GIF and its co-investors. But no portion is allocated to follow-on investors. The argument is that these benefits would not happen at all without GIF’s current intervention.
The benchmark is a conceptual one, not directly comparable to actual cash transfer programs. For instance, in supporting Give Directly’s program for Uganda refugees, GIF is supporting the theory that the cash transfers will lead to investment and sustained improvements in livelihoods.
The tool attributes all future impact gains to GIF and its contemporaneous co-investors, on an argument that the current investment is catalytic or essential for all future impact. A tool under development looks in more depth at the increment in future impact due to GIF’s current involvement. That will result in an alternative measure of efficiency that will be closer to a traditional benefit/cost ratio.
 The procedure described here could be undertaken using social benefits reckoned in dollars. GIF prefers to measure impact in PYI because it treats income, health and mortality gains more equitably between people with different income levels.