Invest Like an Insurance Company
A framework for what to do with money you already know you owe -- built from the same balance-sheet discipline that keeps insurers solvent for decades.
01. Bottom Line Up Front
BLUF: An insurance company doesn't invest to maximize return -- it invests to guarantee it can pay a known future liability, and only then asks what the leftover capital can earn. Business owners sitting on a projected tax bill face the exact same problem, and in practice they answer it one of five ways. Four of those five mismatch the money's risk, duration, or liquidity against the certainty of the bill coming due. The fifth -- building a tiered "general account" the way an insurer would -- is the only one that treats the tax reserve as a liability to be matched rather than a lump sum to be parked or gambled.
02. The Question Every Tax Client Eventually Asks
This comes up constantly in client conversations, and it always starts the same way: a business is projected to owe, say, $100,000 in taxes. That number is real enough to plan around but not yet due, and not yet fully locked in -- it can still move as the rest of the year unfolds. The question the client is actually asking, whether they phrase it this way or not, is: what do I do with this money between now and the day I have to send it to the IRS?
Over enough of these conversations, the answers cluster into five recognizable patterns.
| The Answer | The Pro | The Con |
|---|---|---|
| 1. Pay it now, all of it | Done. Zero mental overhead, zero risk of a surprise balance. | Full opportunity cost on the capital, and the final number can still move by the time the year actually closes. |
| 2. Put it into something high-risk and illiquid | If it works, the same capital does double duty and comes out ahead. | "If it works" is doing all the work. A known, short-duration liability has no business funded by a risky, illiquid bet. |
| 3. Leave it in the business, eat the underpayment penalty | Real business sense -- capital keeps working at whatever the business's return on capital actually is. | The penalty is a guaranteed, known cost accepted in exchange for an efficient-but-uncapped use of capital. |
| 4. Park it somewhere safe and liquid until it's due | Duration and liquidity actually match the liability -- the money will be there. | Correct instinct, but treated as a single decision rather than a system -- no layering, no spread being earned on the certainty. |
| 5. Invest it like an insurance company would | Matches risk and duration to the liability's actual timing and certainty and extracts a return on the capital in the meantime. | Requires actually building the system instead of making one lump-sum decision. |
Answers 1 and 2 sit at opposite ends of the same mistake: one over-corrects for certainty by giving up all opportunity cost, the other over-corrects for opportunity cost by giving up certainty. Answers 3 and 4 are both defensible middle grounds -- they at least reason about the tradeoff instead of picking a pole -- but neither one is a system, just a single choice made once and left alone. Only answer 5 treats the tax bill the way it actually behaves: a liability with a known-ish size, a known due date, and a probability distribution around both.
What Answer 2 Actually Looks Like
I've seen this pattern play out many times. These are just two examples that come to mind.
The first time, a client I'll call Jane -- a retired grandmother -- spent about a year selling call options against Amazon stock. Amazon was climbing, so every month's premium became the next month's stake: she rolled December's profits straight into January's position, snowballing the trade for the better part of a year without ever cashing out. Then January did what January sometimes does to a trending stock. Amazon turned, and the position that had funded itself for eleven straight months lost hard enough to erase the run. The tax problem wasn't the loss itself -- it was that every one of those prior months was still a taxable event. Short-term gains from options trading, taxed at her ordinary rate, on money she no longer had, because it was already sitting in the trade that lost it.
The second time, I wasn't the advisor -- just a friend giving an opinion he didn't take. He had a real tax bill coming and decided the best place to park that money was a cryptocurrency-and-gold trading fund promising returns good enough to make the mismatch worth it. I told him at the time this was exactly the bucket-two mistake: a known, short-duration, high-certainty liability funded by a high-risk, illiquid bet. He put in tens of thousands anyway. He lost all of it.
Different people, different assets, ten years apart -- same structural error both times. Neither one was funding a liability. They were both funding a bet, and calling it a tax reserve.
What Answer 3 Actually Depends On
Answer 3 -- eating the underpayment penalty to keep capital in the business -- deserves a harder look than "bad idea," because it isn't always one. The underpayment penalty rate has always tracked the Fed funds rate; it just happened to sit around 3-4% during the near-zero interest rate years. At that price, a lot of owners had a legitimate case: their capital was earning more than that inside the business, so the penalty was just the cost of a trade that paid for itself. The same mechanism has since pushed the penalty to 7-8% depending on the quarter. A decision that was rational a few years ago can be a mistake today at the same business's same return on capital -- answer 3 isn't a fixed rule, it's a comparison that has to be re-run whenever rates move.
What Answer 1 Actually Gets Right
It's also worth being honest that not everyone leaning toward answer 1 is making a capital-efficiency error. Some business owners pay early because carrying an open balance is a genuine source of stress, and the relief is worth more to them personally than the opportunity cost. That's a real tradeoff, not a mistake -- it just isn't a capital-efficiency argument, and shouldn't be defended as one. Everything from here forward is about the capital that's actually available to be optimized, not the portion someone needs to pay off just to sleep at night.
03. Why Insurers Solve This For a Living
An insurance company is, structurally, nothing but a very large, very disciplined version of the same problem. It collects premiums today against claims it knows -- statistically, not individually -- it will have to pay later. It does not ask "what's the highest-returning investment available?" It asks a narrower, more useful question first:
"What does this money need to do?"
That question forces asset-liability matching before anything else. Capital needed soon goes into highly liquid, low-volatility instruments, where the job is availability, not yield. Capital needed further out can trade some of that liquidity for a higher expected return, because time is on its side. The rule underneath all of it: investment risk should be matched to the timing and certainty of the liability the investment is meant to fund -- never expose a short-term, high-certainty obligation to long-term investment risk.
Two more insurer habits matter here more than the rest of their playbook. First, they separate return of capital from return on capital, and answer the first question before the second -- can I get this money back at all, before asking what it might earn. A projected tax bill is exactly the kind of obligation where "can I get this back" has to be answered with near-certainty, which is why answer 2 above fails on its own terms even before you get to whether the bet pays off. Second, insurers protect against ruin above everything else, because a loss large enough to threaten the liability itself isn't a bad quarter -- it's the whole system failing at the one job it exists to do.
04. The Tax Reserve General Account
Applied to a projected tax bill, the insurer's capital stack doesn't need all five of its usual buckets -- a tax reserve is a short-to-medium duration liability, not a lifetime income stream, so the growth and opportunistic buckets mostly don't belong here. What it does need is the same tiering logic, sized to how far out the payment actually sits:
Near-Term Layer
The portion of the projected liability that's essentially locked in -- what you'd owe if nothing else changed between now and the deadline. Held liquid, low-volatility. Its job is to be there, not to earn.
Contractual-Income Layer
The portion tied to a payment date still months out. This capital can sit in short-duration, high-certainty instruments -- the kind with a known maturity date and a known payout -- earning a real spread while it waits, without taking on the wrong kind of risk for a fixed, dated obligation.
Variance Buffer
The gap between the projection and what the number could realistically become if the year runs hot. Sized from the same estimate that produced the $100,000 figure in the first place, and held with the same discipline as the near-term layer -- this is what keeps answer 4's "single lump-sum decision" from turning into an unpleasant surprise in April.
The point of naming three layers instead of one lump sum isn't complexity for its own sake. It's that a single "safe and liquid" decision, made once, earns nothing on the portion of the money that genuinely has months of runway before it's needed -- and an insurer would never leave that spread on the table. Layering the reserve by actual time-to-due-date is what turns answer 4 into answer 5.
05. Tax Sherpa Integration
This framework only works if the underlying projection is accurate enough to size the layers against. A tax reserve built on a stale or padded estimate isn't matched to anything real -- it's just answer 1 wearing a spreadsheet.
Tax Sherpa builds the quarter-by-quarter projection this framework depends on, and pairs it with the entity-structure and deduction-stack work covered in Tax Optimization Blueprints -- so the number being reserved against is already the smallest accurate number it can be, before a single dollar gets tiered into a general account.
Get a Projection Worth Building a Reserve Around
Before you decide where the money sits, make sure the number itself is right. Let's build the projection, then the reserve.
Schedule a Tax Sherpa Session →