Working title (insurance)

My family recently bought a duplex in Chicago, after years of living in Japan. This exposed me to Relatable Banking Influencer Content. One facet of it is the largest bill you’ll ever get from the insurance industry for the most inscrutable reason, which I thought would be interesting to cover.

Every time you transact in property, you will notice a variety of ticky-tack transactional frictions added to a (hopefully) itemized list. The largest, by a substantial margin, are agent commissions, which have come under substantial scrutiny for their set-by-a-disciplined-cartel character.

Next up on the list is a bundle of services around “title.”

The rest is a mix of government fee passthroughs and Obvious Nonsense, such as a $125 “water processing fee,” $55 for a wire transfer where that number is just made up, etc. But if I were to go through each of the 16 line items summing up to $1,400, we’d be here all day.

So, let’s talk about the title industry.

What is “title,” anyway?

Ownership is a bundle of property rights, which exclude others from using a thing, and which are hopefully (for the owner) enforceable by the legal system in the case where the larger societal system fails to agree on reality. That is the very orthodox, first-year law school answer, at any rate.

Title in real property is the aggregate of rights, commitments, and contracts which make up an ownership claim. And here it gets into gloriously wonky real estate operational trivia about the difference between an easement versus an encroachment, the justiciability of restrictive covenants written by societies far less enlightened than the present, and similar. But in common usage, you can round title to “who owns this address and how do we know?”

You might reasonably think that you know about title because you can look it up in a database, probably maintained by the government. Here you run into a fascinating historical detail.

Distributed versus centralized database design in property rights

Most people assume ownership is recorded in some sort of government database, in the same sense that your bank balance is recorded in some sort of bank database. If you assume this, you’re right… for many places in the world.

For example, if you wonder who owns a particular tiny sliver of Tokyo, you can hire a judicial scrivener to go ask the government, and in a fairly deterministic fashion they will bring you a piece of paper saying that the Legal Affairs Bureau’s records show one Patrick McKenzie as very definitely owning it. That piece of paper suffices as proof of title for almost all purposes in Japan. Courts, lenders, and the ward office will all treat it as one step below holy writ.

The United States, perhaps surprisingly, is not operationally capable of producing that piece of paper. There is no government body in the United States which will confidently say that, as of this instant, Patrick owns this property to the exclusion of all others. Serious professionals who work in or adjacent to the real estate industry understand this incapacity of the United States and organize their lives around it.

As a broad sketch of varied practice over the 50 states, the relevant government body (here, the Cook County Recorder of Deeds) does not record ownership but rather records certain private transactions. Current ownership is not an independent fact; current ownership is the sum of all compounding transactions since time not-quite-immemorial. (Cryptocurrency enthusiasts might see a parallel: blockchains typically don’t record balances. Software operating on top of the blockchain probabilistically estimates balances by being aware of all transactions that happened since genesis.)

At some point in the very near future, it will be a matter of the public record that I bought a property from a particular seller, and that a bank filed a lien against that property due to me taking out a mortgage.

Users of the database will infer that, since the last few entries in the database were that seller buying the property from someone else, recording a mortgage, and extinguishing that same mortgage on full payment, and there are no other recent entries, that I very probably own the property.

There is an important difference between “very probably owns” and “certainly owns.”

A quick digression for privacy-minded buyers

I lied. I don’t actually own any property in Chicago. My wife and I are beneficiaries of a land trust. The land trust actually owns the property. It is contractually obligated to allow us to live there, receive all rents and other benefits which derive from ownership, and pay us when the trust decides to sell the property, which it will eventually do. The trustee cannot independently decide to sell the property; it must, under law and contract, faithfully execute our directives.

“That sure sounds like ownership, Patrick.” Oh yeah, it’s designed to be equivalent to ownership in every way except basically one.

While there are other reasons to use them, the dominant use case for land trusts is mild privacy preservation. Because maintaining records regarding real estate implicates the public trust, in much of the United States, those records are public records. When this required actually schlepping down to the county clerk’s office to review yellowing papers or microfiche, the fact of the records being public was an interesting bit of operational minutiae for practitioners but had very little impact on owners.

But we have computer systems these days maintaining the records, and also vast secondary ecosystems of data brokers who ingest public records at scale and collate them with other information about people, searchable by other identifiers. As a consequence of this, in Chicago (and many other American cities), if you know a homeowner’s name (or address, or phone number, or…) you can have the full text of their mortgage, address, purchase price, monthly payment, etc etc, with 30 seconds of effort. No login or reason is required.

Many people react quite negatively when they learn this. If you are currently reacting negatively, I express no judgment.

This strikes me as similar to many questions about privacy rights. The range of human preferences is wider than anticipated. Framing influences perception quite a bit (“Anyone on Twitter can figure out your children’s exact walk to school” sounds different than “Your property tax payment is a public record” despite being the same physical database entry). Our laws have (as a descriptive not normative statement) not been updated in the wake of technological progress.

And so, I pay a very boring company $300 a year for a two-page contract that makes them our trustee. They have that contract in a filing cabinet. It is (assuming competent execution, always a risky assumption in the real estate industry) not cross-referenced in any databases. You can get them to show it to you, but you’ll need a court order, and fighting that court order is basically their reason for existing.

Many people, when they learn about land trusts, immediately assume that something extremely hinky is going on. Not so much; this is an extremely common way for savvy people to own property. It is in no way a loophole. The same polity which told its elected representatives that it wants property records to be public also told its elected representatives that it wants to exempt the rich, powerful, and savvy from that requirement. (That is a commentary on the American political system, certainly.)

Anyhow, should you want to avail yourself of this the next time you buy property, just tell your real estate attorney “What’s the privacy option in this state? Land trust or something?” They do this all the time. Or you can choose to have your full-text mortgage publicly available and automatically imported into hundreds of data sets. Whichever you prefer.

(As long as I’m adding I-grew-up-discussing-real-estate-quirks-at-the-dinner-table-sorry-not-sorry notes, another use case for land trusts is that judgments against individuals are more difficult to enforce against property held by a trust. LLCs are also commonly spun up to act as legal firewalls for that reason. Ask your friendly neighborhood real estate lawyer; this is common knowledge in that community of practice. Like most complex systems underlying how the world works, it is very understandable by mortal minds, and people who tell you otherwise are lying to you.)

High confidence and complete confidence are different

Perhaps the digression about land trusts has helped convince you that, if someone tells you they own a property and want to sell it to you, that claim might be more difficult to verify than you’d naively expect. It is also a claim that can sometimes be falsified well after a transaction.

Suppose a person lives in a community property state. One day, in the throes of passion, they swear their undying love and devotion in front of a justice of the peace, perhaps in a commercial establishment in a jurisdiction well-known in popular fiction for facilitating other-than-considered vows of this nature. That passion wanes along with the alcohol, and everyone involved just tries to forget this incident. A year later, they purchase a property, using their own money and a mortgage. And then, in the future, they sell their property, without asking for permission from their spouse, because socially speaking they are not married.

Does the new buyer actually own the property? No, they do not, because that property has been fraudulently transferred to them, against the interests of the spouse with a 50% claim to it by law. Does a person purchasing it from them actually own the property? Again, no, they do not.

You might sensibly object that no database reasonably available to these innocent buyers recorded the fact of the out-of-state marriage. The law does not care, and remediation at this point will be extensive and expensive.

(If it sounds implausible that marriages are not trivially searchable: the marriage is equally legally valid if conducted overseas. For example, when an American marries a Japanese person in Japan, the right U.S. government agency to register that fact with is no one at all. My wife and I joke about our unlicensed marriage, but it is absolutely valid in the U.S., and rights under it are enforceable by U.S. courts, because of the principle of comity. Comity doesn’t care that your SQL query returned zero records.)

“Undiscovered marriage torpedoes a real estate deal after-the-fact” sounds far-fetched, I know, I know. Every real estate lawyer has variants of this story in particular and another few dozen with similar effect. Partially they’re deployed tactically to drum up additional work for real estate lawyers. And partly they’re only slightly fictionalized versions of real cases where the full details are recorded for posterity by court reporters. (In the category of particularly historically well-attested-to title disputes, a particular family lost their home three times due to title defects. The family was forced to migrate as a result of these disasters. The young son, perhaps scarred by them, later went on to practice law here in Illinois. He is better known for other work.)

So we have a system for remediating title defects.

Title insurance and title searches

Our first procedural countermeasure is that one hires a professional to diligently conduct “title searches.” And, indeed, someone is certainly going to bill you for doing this work, and for a non-obvious risk transfer incident to doing that work.

But simply querying the database harder will not, and cannot, shake out all title problems. (Back in the days of yellowing paper and microfiche, knowing how hard your title searcher searched was actually consequential. Now that a twelve-year-old can do a physically equivalent search, the competence distribution is… slightly narrower than it used to be.)

For all those edge cases that no amount of searching can derisk, there exists title insurance. It is a specialized insurance policy which says that, if there is an undiscovered defect in title, the insurance company will pay for the expensive and painful remediation, up to (and inclusive of) simply refunding the entire purchase price of the property to the insured buyer/lender. It is critical to understand that title insurance is effectively mandatory since almost all purchases are financed. Lenders will require a policy be purchased, and they are themselves similarly obligated to require this, due to the supply chain for mortgage financing.

Title insurance has been called an expensive racket. A wag might say that this is grossly unfair. To rackets.

Why? Well, it comes down to how title insurance is priced, sold, and purchased.

To understand that, three magic insurance words you should know: frequency, severity, loss ratio. Frequency is the rate of occurrence of claims. Severity is the cost of claims contingent on claims happening. Loss ratio is the total amount of paid-out claims divided by collected premiums.

Title insurance has extraordinarily low frequency for insurance products. However, when it does pay, the severity can be very high. Title insurance defenders will tell you that the reason title insurance is expensive is because the insurance company is promising to literally buy you a house in event of a problem.

Title insurance defenders are dissimulating, though, because the actual loss ratio on title insurance policies is laughably low. This number is exhaustively tracked by insurance regulators, and floats around the 5% region. And so, of the $4,000 or so that I paid in title insurance, the underwriter expects to pay out $200 in losses.

A high loss ratio means an insurance policy is inexpensive relative to the actual risks it insures; a low loss ratio means the opposite. Title policies are among the most expensive insurance policies issued for any risk whatsoever.

Now you might ask “What is a typical insurance loss ratio?” These are not unknowable numbers; they’re some of the most accurate figures captured by capitalism, with a combination of financial institutions and government regulators obsessed over quarter-to-quarter variations in them. Let me quote a couple of representative examples: fire, 65%. Workers’ comp, 48%. Medical profession liability, 56%. Auto liability, 76%. Homeowner, 82%. Even travel insurance, which is legendarily a poor option for customers (for reasons), pays substantially more out in claims than title insurance does.

So why does this policy cost 10-20X as much as other comparable insurance risks?

One very quirky risk transfer and a statistical artifact

We mentioned that title insurance is bound fairly tightly with conducting a title search. In theory, one needs to chain that title search backwards for a few hundred years, at which point there will be an entry that sounds something like “ceded by the king of Spain to the United States” or “acquired by right of conquest.” (These are absolutely real facts that appear in title records. If you want to pay six figures, you can get a degree in philosophizing that all property is based on a theft, late-stage capitalism, etc etc. Few people who work in title insurance have that sort of degree.)

In practice, most title searches are strictly limited. My transaction obligated the searcher to do backbreaking labor and laboriously read twenty four months of transactions (i.e., two transactions) which would take a non-specialist thirty seconds to find using an online publicly available portal. They were not required to read several dozen transactions going back to the digitization of records (which gets you to almost when I was born) or to try to reconstruct what happened to Chicago title records in 1871.

For diligently reading two search results, the searcher was paid $260. ... Or were they?

Yes, according to an invoice. Not really, according to the title industry. But yes again, in reality.

The reason that the title industry says the $260 is not actually earned solely for reading two records is that there is a complex contractual risk transfer happening incident to the search and determining insurability of the title. They, acting as the agent of the insurance underwriter (this would be called a “carrier” in most insurance industries but in title “underwriter” is used to mean “the insurance company” and not “a specific professional at the insurance company”), represent and warranty that they’re making commercially reasonable efforts to avoid “on-record” title flaws (i.e., failing to read those search results accurately).

The title insurance industry expects there to be three main categories of claims.

One, vanishingly unlikely (as a percent of claims) but extremely evocative, is a “historical defect,” where the king of Spain (or, in Illinois, far more likely an Indian tribe or the federal government) has a justiciable concern about the original transfer of land to private ownership.

The far more likely type of claim is “off-record” flaws, where someone has ownership but that isn’t reflected in the searched records. The above ownership-by-undiscovered-marriage scenarios are examples of off-record claims. There is an infinite universe of fact patterns that can result in them, though; this is substantially why title insurance exists.

Then there are “on-record” flaws, where… somebody goofed, and nobody caught it before the transaction closed. There is a clear indication in the search results that the seller lacks legal right to sell the property. Maybe the mortgage isn’t paid and arrangements haven’t been made. Maybe they haven’t gotten a lien released. Maybe they are in the midst of an unresolved divorce. Maybe there is a charming historical anomaly on the deed. (Some anomalies are of a variety that are presumptively void in present-day America.)

In the case of an on-record flaw, where the searcher (who is also usually the agent of the title insurance company) “dun goofed,” in theory the title insurance company can put the claim back on their agent. In theory, that would result in “the system” paying a claim without that claim showing up in the loss ratio. In theory, this means that title insurance isn’t as expensive as it looks.

In practice, this basically never happens. But it’s a nice theory on why title agents deserve to get paid 80-85% of the insurance premium. (These are the standard numbers in Illinois. In some states, it goes as high as 95%. This is, I rush to add, done in the clear light of day. It is definitely not a kickback. A kickback requires someone involved to feel shame.)

Since “basically never happens” is a claim about reality that can be measured with numbers, I’ll observe that title insurance agents themselves carry insurance policies. One important genre is errors and omissions insurance, which can cover them if e.g. they goof and actually have to reimburse a purchaser/lender without the title insurance company covering it. Those policies themselves have a price, and that price encodes the information “we’re talking basis points on basis points of risk here.”

So why do title insurance agents actually get paid so richly, directly driving up the cost of title insurance?

How title insurance is sold

In theory, there is a vibrant, functioning market in title insurance, with thousands of agents ultimately backed by dozens of carriers in Illinois. Price should float down to the minimum amount of loss ratio plus administrative costs plus profit required to sustain a vibrant insurance industry.

In practice, nobody shops for title insurance. I write articles like this as my actual literal job and I didn’t shop for title insurance. I used the insurance company nominated by the seller’s lawyer.

Unsurprisingly, the seller’s lawyer nominated the insurance company that she is an agent of. This was disclosed, in an entirely aboveboard fashion, on one of dozens of documents of paper sent back and forth during the buying process.

You are welcome to your estimate of whether anyone saw fit to explain that document as anything other than “sign this to continue.”

The seller’s attorney earned $625 for legal services in connection with the transaction. 80% of $4,000 in title insurance is $3,200. I think you can make a reasonable estimate as to how important that attorney understands having a ~100% attach rate of title insurance to real estate closings is.

Like all industries, real estate is a very small world, particularly since it is conducted hyperlocally. I have many, many dinner table discussions from my father (in commercial real estate in Chicago for most of his career) about this. My attorney, who I found independent of any other party to the transaction, had interacted with the seller’s attorney on numerous occasions. They mutually collaborated to take a straightforward transaction to a speedy and efficient close.

You are welcome to your estimate of how many times my attorney called attention to the title insurance fee, that the number was set by an act of the seller’s attorney, or that this fee could be shopped.

A really good mental model to carry around for analyzing the finance industry is one-shot versus iterated games. Real estate attorneys model (residential, owner-occupied) closings as effectively one-shot with respect to the client but iterated with respect to the other attorney. If one were conspiratorially-minded, one could say unkind words like "conflict of interest" at this point, but this sort of equilibrium doesn't require anyone to act invidiously. The other attorney is a peer running their business in a socially accepted fashion and very likely quite similarly to how you run your own business. You will see them again both professionally and socially. Why make trouble over nothing.

One reason I personally, despite being fairly financially sophisticated, did not shop the quote was that I was unsure the juice would be worth the squeeze. It’s pretty clearly possible to insure this transaction for 1/10th the price; that pricing prevails in other U.S. markets. It was not obvious to me it would actually be offered by anyone serving Chicago. (You can read a lot on this topic in the book The American Title Insurance Industry: How a Cartel Fleeces the American Consumer. I’ll give you one guess as to the thesis of the work.)

Is there anything to be done here?

You, dear reader, are highly likely to transact property many times over the course of your life. There is a specific line item on your disclosure that almost all participants will skip over. You might make the decision to shop around on that item, and in doing so potentially save a few thousand dollars for an hour of work. (Bits about Money is supported by members, some of whom just got excellent ROI.)

On a societal level… title insurance adds up, fairly quickly. The typical American purchaser will reside in a house for 7 years, and get repeatedly cheesed in this fashion.

We could simply decide, as a policy priority, to not structure the industry this way. However, this is a classic political economy problem, with diffuse costs and concentrated benefit. The real estate industry is extremely politically powerful. It is nationally distributed, extremely well-resourced, and staffed by vocal pillars of the community. Those advocates are everywhere and talk to likely voters because it is their job to do so and are extremely well-liked. The same lawyer who quietly cheeses buyers on title insurance will, in about half of their client interactions, write a client a very obvious, very salient, very memorable check for multiple years of the client’s salary.

The title insurance industry extracts a relatively small rake, hidden in the minutiae of a complex transaction that most legislators and regulators don’t truly understand. There is a strong, organized constituency in favor of that rake existing. That constituency is not shadowy forces in smoky backrooms. They are pillars of your community. They are your friends and neighbors.

And they, independently and through their lobby, know how to present this business to the American polity. The industry provides a valuable service, and charges money for it, unabashedly. The price is set by a vibrant, competitive marketplace. Historical infelicities like kickbacks have mostly been replaced by market mechanisms, like controlled business arrangements, which are fully disclosed to the consumer. Of course consumers read and understand disclosures. The state even released model disclosure language which the industry adopted almost universally.

Do I think this equilibrium is likely to change? I would not bet on it over short timeframes. But, if anyone out ever wants to take a serious run at disrupting this industry, I’ll happily write you a check.