While no one likes to pay more in insurance premiums than they have to, a fundamental principle of insurance is that, in the end, there must be enough premiums (plus growth) to cover potential future claims (plus overhead and profits for the insurance company). Insurance coverage that is too cheap is actually risky, and coverage that seems expensive may actually be the most secure.

In fact, one of the most significant caveats to considering any form of insurance (or annuity) guarantee at all is if the insurance is not going to lose you money on average, it’s actually something to avoid. In other words, insurance guarantees should never be expected to make money on average for the policyowner, or the insurance company will lose money until it inevitably goes out of business and the guarantee will be gone anyway.

As a result, decisions to purchase insurance and/or seek out guarantees should always be viewed from the perspective of seeking to trade a small known loss to avoid a big unknown loss instead. The goal is not to finish with more money on average, but simply to shift the range of outcomes in a manner that increases the number of small losses and reduces the exposure to big ones that may be unrecoverable. The next time you’re considering a type of insurance or annuity guarantee with a client, make sure you know why and how the coverage and guarantees are expected to lose money… and then decide if the trade-off is worthwhile anyway.

If you can’t figure out how the guarantee will lose you money on average, it’s a strong indicator that either you’re missing a key detail, the guarantee is overpromising something it can’t deliver, or the guarantee itself may be a mirage that the insurance company cannot possibly make good on in the end.


There are, of course, insurance offerings for an astonishing range of potential risks, from the (financial) consequences of death to the cost of a severe medical event. And while the potential consequences of such events could be financially damaging (or outright catastrophic) to any individual, the opportunity to pool together the risks of many people — knowing that some will experience the expensive adverse event, but others will not — creates the possibility to turn a large unknown and potentially destructive risk into a more manageable known cost.

For instance, if there’s a 0.1% chance that a $300,000 house will burn down next year, then on average, one out of every 1,000 people will have their house burn down — which means 999 people will be fine.

To manage this risk, if each of the 1,000 people contributes $300 (the premiums) to a central pool of money, then collectively there will be $300,000 available to make whole the one person whose house burns down, requiring a $300,000 claim. We collectively smooth out the risk for everyone, at a fairly modest cost of $300 per person to insure against a $300,000 risk.

This potential for risk pooling — where small premiums from a lot of people can provide for large payments to cover the risky events of a few, which in turn can be reliably predicted (on average) by relying on the law of large numbers — is the core principle of insurance protection. But in the real world, implementation is slightly more complex.

On the plus side, there is the reality that most risks play out over an extended period of time, which means that not every claim has to be paid directly and immediately with a dollar of premium collected. Instead some premiums can be invested for growth to cover future claims needs.

Yet as the network of people and quantity of potential risks being insured grows, there is a need for some administrative overhead cost to manage the arrangement and the organization that implements it — an additional cost over and above just the payout of claims themselves.

And in today’s modern world, many insurance companies are publicly traded companies owned by stockholders who expect the company to generate profits. So insurance companies ultimately need claims and overhead to be a bit less than what they take in premiums (and generate in growth), so that profits remain.

See the “Insure or Do-It-Yourself?” chart below to understand the core identity formula for an insurance company.

From the client’s perspective, this means that premiums (and the growth thereon) must collectively cover the claims that will be paid on the insurance, the overhead to run the insurance company, and the profits for the insurance company shareholders.


Understanding that equation also highlights the distinction between buying insurance and choosing to self-insure.

When the “insurance company” part of insurance is eliminated, and individuals simply save their own implied premiums (and invest them for growth) to pay their own future costs, the equation changes.

Clients need to cover the costs of any future claims, but not the insurance company overhead or profits — so self-insurance is fundamentally a less expensive proposition than buying insurance, all else being equal. (This also shows the appeal of mutual insurance companies, which eliminate the profits portion of the equation by redistributing the share of profits to policyowners in the form of policy dividends.)

Because purchasing insurance coverage should always require greater premiums than the cost of self-insuring, it’s generally best to only insure what is necessary. That means having larger deductibles if the client can afford it, implicitly self-insuring the smaller risks and just keeping insurance coverage for the big ones.

Nonetheless, some potential losses are just so large that it’s not feasible to self-insure. Think about popular uses of insurance like term life for young people, or homeowner’s coverage. In these cases, buying insurance really does become the most effective route; skipping the coverage to self-insure is really just gambling that a catastrophic event won’t occur.

In these cases, buying coverage will still be more expensive on average, given that the risky event usually won’t happen. But the trade-off should be appealing, especially for potentially large claims with very small probabilities, where the ramifications of being underinsured could be severe but the actual cost of insurance may be modest.


Insuring a risk is expected to be at least a little more expensive than self-insuring, both because self-insurers don’t have to cover the cost of insurance company overhead and profits, and because self-insuring also offers the possibility that clients will never have a “claim” (or risky event) to pay for.

Yet when your clients do purchase insurance, the only policy they should ever buy is coverage that will lose them money on average.

Here’s why. As noted earlier, the essential equation for insurance must balance. Premiums and growth must equal the cost of claims plus overhead and the profits of the insurance company, because that’s all the money there is.

In a situation where claims are expected to exceed the premiums and growth — in other words, where the insurance buyer expects to make money on average by claiming on the insurance or guarantee — the equation can only balance in one way (given normal fixed overhead): The “profits” must go negative.

In other words, the line can only balance if the insurance company has a sustained and ongoing financial loss and keeps bringing its own money to the table to cover those losses.

Yet there’s a longer-run problem with such a path. If the insurance company incurs losses indefinitely by paying out more than it receives in premiums (plus asset growth thereon), the insurance company will eventually run out of money to make the payments. The model is inherently unsustainable.

In other words, buying an insurance policy that is expected to come out ahead for the policy owner is the equivalent of buying a policy from a company that is expected to go out of business. Which, of course, means that the financial outcome for the policy is not actually likely to be positive at all.

Notably, the key point here is about averages, not individual claims. Clearly at least some individuals will receive more in claims than they contribute in premiums; with high-impact, low-probability coverage like term life or homeowner’s insurance, any claim could be many multiples of a particular policyholder’s premium.

On average, however, policy buyers should expect to receive less than they pay in premiums (plus growth), because they either never have a claim or have only small claims that amount to less than their payments.
And again, there’s a key difference between potentially getting more in claims than were paid in premiums — because the risky event happens to you — and expecting on average to get more in claims than are being paid in premiums (plus growth). 

The former represents the essential principle of insurance, while the latter is an unsustainable path toward bankruptcy (for the insurance company) and a potential total loss (for the policyowner).


While the key tenet of buying insurance is that it all should result in a financial loss on average (or else the insurance policy is unsustainable), that doesn’t mean all insurance is a losing proposition. It simply means that insurance should be targeted specifically for those situations where the risk would be so severe and catastrophic that self-insuring is either not feasible or just too disruptive to other goals.

Think about protecting younger workers who haven’t had enough time to save up a cushion to protect against earnings loss due to death or disability, for instance. Or clients whose house might burn down: Even if they could afford to replace it, it’s just far easier to manage a small premium than pay for a new house out of pocket, if it became necessary.

So it makes sense to buy insurance where the premiums will result in a known small loss, if that small loss on average is still a better outcome than the possibility of the big, disruptive loss. This means that buying insurance is not about improving a client’s financial situation on average, but about narrowing the range of possible outcomes and eliminating the worst “tail” events.

See the “Risk Trade-off” chart below. The blue line shows the self-insuring client’s potential wealth, while the red line shows how the results shift with insurance. The outcomes are slightly worse in almost all situations, with a slightly lower probability of gains and a slightly larger likelihood of losses — but the small number of extreme events in the left (negative) tail are eliminated.

This is the core trade-off of insurance: a slightly worse average outcome in exchange for avoiding a small number of improbable but potentially catastrophic events.


There’s a corollary to all this: In addition to buying insurance policies that you expect to lose money, it’s just as important to avoid insurance that is expected to be a winning proposition. After all, insurance companies that consistently pay out more than they take in will eventually end in ruin; those who get paid last will find out there’s no money left.

Thus, if some form of insurance product, solution or guarantee appears to improve your results across the board by giving generous benefits for a very modest cost, it really only means one of a few things: Either you’re missing a key cost, factor or outcome; the guarantee is just overpromising something it can’t deliver; or the guarantee is accurate — which would mean the product guarantee (or the entire company) is doomed to failure in the long run.

Thanks to insurance regulation, that last scenario is unlikely; insurance company defaults are now blissfully rare. Nonetheless, the basic principle remains the same: Be very wary of any insurance guarantee that implies your client will come out better most of the time — whether it’s a variable annuity with a guaranteed living benefit rider, generating “income” from equity-indexed universal life insurance, or a study that professes certain situations where long-term care insurance has a positive expected value for the buyer.

Often there are additional costly scenarios you’re missing that will make the outcome less favorable and allow the insurance company to survive. Sometimes what looks like insurance protecting against a loss is actually just a guarantee to return your own principal in the first place; while in other situations insurance companies hope you don’t realize what you’re giving up in forgone interest and opportunity costs. Or the policy itself is at risk and can’t back the guarantees being provided.

Analyzing an insurance or annuity guarantee and concluding that it has a positive expected value on average generally means either that you’re wrong, or you’ll lose (or both). That’s a time to run, not a time to sign on the dotted line.

Ultimately, the bottom line about insurance is just this: If your insurance policy is expected to improve the outcome in all scenarios for a low cost, and you can’t proactively identify how the average customer will lose money (and the insurance company will make money), you shouldn’t buy the insurance or annuity guarantee.

The real time to buy the insurance guarantee is when you can identify what the “extreme” loss could be, understand how much less you’ll finish with on average by buying insurance instead and find that trade-off appealing enough to pay the implicit and explicit costs of insurance.

Michael Kitces, CFP, is a Financial Planning contributing writer and a partner and director of research at Pinnacle Advisory Group in Columbia, Md. He’s also publisher of the planning industry blog Nerd’s Eye View. Follow him on Twitter at @MichaelKitces.

Read more: