Disbursement timing tool
Disbursement timing tool

This is a tool, based on my donation timing model (here, see especially section 3), for helping you determine the schedule on which you think philanthropists trying to maximize their impact should disburse their assets, as a function of your beliefs about some input parameters.

For inputs that roughly capture my own beliefs, at least as of when I made this tool, click here.

#states:   –   +  
State Now = 1
Criticality ? 1
Annual discount rate ?
Annual interest rate ?
Diminishing returns rate ?
Annual transition probability to: ?
State 1
COMPUTE Invalid inputs Loading...


This tool currently omits several variables relevant to the disbursement timing problem. For a more thorough model and list of considerations, see the paper draft linked above. I hope to make this tool more sophisticated over time. For now, though, I think its recommendations are a decent first approximation.
×
Criticality

This is a scale parameter (previously dubbed “hingeyness”) capturing how easy it is to do good with a unit of spending in each state, relative to how easy it is to good by spending now.

It must be a positive number.
×
Annual discount rate

This is the annual probability by state that invested assets will become valueless to you, as might happen in the event that the assets are expropriated, that a catastrophe destroys them, or that the assets' inheritors abandon your values for values orthogonal to your own. A rate of .01, for instance, implies that the expected lifespan of a value-aligned fund is 100 years. If you have a positive rate of pure time preference, this should be added into the discount rate.

It must be a positive number.
×
Annual interest rate

This is the interest rate you expect invested assets to earn in each state. A rate of .07, for instance, implies that invested assets will double in value roughly every ten years.

It can be any number.
×
Diminishing returns rate

This is the rate at which spending more quickly at a given time produces less marginal impact per unit spent. A rate of 0.5, for instance, implies that impact scales with the square root of the spending rate; a rate of 1 implies that impact scales with the natural logarithm of the spending rate.

It must be a positive number.
×

Each cell (row, column) of the table below is, roughly, the annual probability of moving to state (row), given that you start in state (column).

Technically, we are assuming that when in state (column), you are exposed to a constant Poisson process of annual rate (row, column) of moving to state (row). A cell therefore represents something more like the annual probability of moving to state (row) given that you start in state (column) and don’t move to another state first that year. So the columns can sum to a value greater than 1. But all entries must be nonnegative.