The retirement rates source data, found in Reason's "../Reports/extracted inputs/" folder, in files such as "normal retirement tier 1.xlsx", has rows for each single year of age except for the 70-79 age group, which has just one row. The 70-79 row is followed by a final row, for age 80. Reason's data preparation expands this group to have a row for each individual age from 70 to 79, and gives each age the retirement rate for the 70-79 group. (See the function clean_retire_rate_table() in "Florida FRS benefit model.R". I have used the same rule here, in "retirement_rates.R" in the data-raw/systems/frs folder.)
For example, in "normal retirement tier 1.xlsx", in the first column (Regular K12 instructional female), the retirement rate for the entire 70-79 age group is 20%. The retirement rate at age 80 is 100%. After data preparation, each age from 70-79 has a retirement rate of 20%, which is the rate for the 70-79 group.
For consideration: Should we instead have the retirement rate ramp up from 20% to 100% over ages 70-79?
I'm not sure. That's probably a better representation of what happens in the real world. But is it an accurate representation of what the Florida FRS actuary does in valuing liabilities for the system? It is possible that the actuary also uses the same rate for each age from 70-79. Alternatively, the retirement rate tables may just be summary representative tables published by the actuary, who may use more-detailed internal tables for valuation purposes with different rates for each year of age from 70-79. If we're trying to match what the actuary does rather than what happens in the real world, then we'd want to know what the actuary does.
It does not seem like a big issue because I cannot imagine that reasonable alternative retirement rates for people ages 70-79 could have a material impact on valuations or cash flows, but it's worth thought at some point.
@gchen3 simply FYI.
The retirement rates source data, found in Reason's "../Reports/extracted inputs/" folder, in files such as "normal retirement tier 1.xlsx", has rows for each single year of age except for the 70-79 age group, which has just one row. The 70-79 row is followed by a final row, for age 80. Reason's data preparation expands this group to have a row for each individual age from 70 to 79, and gives each age the retirement rate for the 70-79 group. (See the function
clean_retire_rate_table()in "Florida FRS benefit model.R". I have used the same rule here, in "retirement_rates.R" in the data-raw/systems/frs folder.)For example, in "normal retirement tier 1.xlsx", in the first column (Regular K12 instructional female), the retirement rate for the entire 70-79 age group is 20%. The retirement rate at age 80 is 100%. After data preparation, each age from 70-79 has a retirement rate of 20%, which is the rate for the 70-79 group.
For consideration: Should we instead have the retirement rate ramp up from 20% to 100% over ages 70-79?
I'm not sure. That's probably a better representation of what happens in the real world. But is it an accurate representation of what the Florida FRS actuary does in valuing liabilities for the system? It is possible that the actuary also uses the same rate for each age from 70-79. Alternatively, the retirement rate tables may just be summary representative tables published by the actuary, who may use more-detailed internal tables for valuation purposes with different rates for each year of age from 70-79. If we're trying to match what the actuary does rather than what happens in the real world, then we'd want to know what the actuary does.
It does not seem like a big issue because I cannot imagine that reasonable alternative retirement rates for people ages 70-79 could have a material impact on valuations or cash flows, but it's worth thought at some point.