Do I need to tap into my retirement fund? |
If you are on the opposite end of the spectrum and have no savings to speak of, and no sufficient earnings to pay down debt, that is the point at which you might consider dipping into a retirement account to pay off your debt (let's pretend that loans collateralizing your 401(k) do not exist). Similar to the savings account example, your retirement account might be earning 7% a year vs. your debt costing 18% a year. All you need to do is make sure your foregone net interest costs offset the taxes and penalties from distributing prematurely. The important metric is the time period over which you don't pay off the debt. At some point after you hold the debt long enough (maybe in 10 years), the dollar amount of interest costs will be so high that you will have wished you dumped the retirement savings to pay that debt off. So here is a very basic decision framework that will tell you, for every dollar of debt, the payback time above which you should be touching that nest egg:
- Let a = retirement account balance
- Let b = % penalty from early distribution (assume 40% for taxes and 25% for penalties, total of 65%)
- Let y = % annual yield on retirement account (7%; and assume post-tax to keep the formula simple)
- Let i = % annual interest on credit card debt (18%)
- Let d = amount of debt
- Let n = number of years of not paying off debt above which you should simply take retirement funds to pay the debt (this is the variable that we hope to solve for)
- Conceptually, to distribute retirement funds early, the following equation must be true: [d(1+i)^n-d] - [a(1+y)^n-a] < ba (this states that the foregone interest costs over the given time period are less than the early distribution penalty; in other words, DISTRIBUTE!)
- Remember that if you take the early distribution of "a," you receive (1-b)*a; this is thus the amount of debt you can pay off using the distribution proceeds, meaning d = (1-b)*a
- Resubstituting and simplifying, you get a(1-b)[(1+i)^n-1]-a[(1+y)^n-1] < ba
- Reduce this formula to: [(1+i)^n-1]-b[(1+i)^n)-1]-[(1+y)^n-1] < b
- Then: (1+i)^n - b(1+i)^n - (1+y)^n < 0
- Then: (1-b)(1+i)^n < (1+y)^n
- Take the log of both sides to get: LN((1-b)(1+i)^n) <> LN((1+y)^n)
- Then: LN(1-b) + n*LN(1+i) < n*LN(1+y)
- Ultimately: LN(1-b) / (LN(1+y) - LN(1+i)) < n
- In spreadsheet form, enter "b" into C2, "y" into C3, and "i" into C4, then calculate using the formula =+LN(1-C2)/(LN(1+C3)-LN(1+C4))
- Using the assumptions from 2-4, we have n > 10.73 years
This post is dedicated to Antonio Cromartie, whose spending habits and child support payments (despite his large salary) make him a candidate to evaluate the retirement account tradeoff. You're welcome for the formula.
F-One