When someone is arrested and charged with a crime, federal law requires that they be taken "promptly"—usually within 48 hours—to a magistrate who will decide their conditions for release pending trial.¹ Conditions of release always include the promise to appear at trial and other important hearings, but they can include many other things as well, such as curfews, check-ins, drug testing, or dispossession of firearms.² The most common condition of release in the United States today is the payment of a set amount of money, known as a bail bond. If the defendant appears at all scheduled hearings, the bail money is refunded. If not, the money is forfeited to the state or county who filed the charges.³

It is commonly reported that the median bail amount set for a felony charge is $10,000,⁴ but depending on the charges and the jurisdiction, bail amounts can be much, much higher. Many judges rely on a bail schedule—a chart of predetermined bail amounts corresponding to particular charges—to quickly assess bail for individual defendants.⁵ New Jersey uses a uniform schedule across the state, which includes entries like this:

Charge	Degree	Bail Range
Murder	1st Degree	$250,000–$750,000
Manslaughter	2nd Degree	$100,000–$200,000
Kidnapping	1st Degree	$200,000–$400,000
Sexual Assault	2nd Degree	$50,000–$200,000
Lewdness	4th Degree	$1,000–$2,500
Robbery	1st Degree	$100,000–$250,000
Robbery	2nd Degree	$50,000–$100,000
Carjacking	1st Degree	$100,000–$250,000
Criminal Trespass	Disorderly Persons	$500–$1,000
Criminal Trespass	Petty Disorderly Persons	$100–$500
Theft	2nd Degree	$35,000–$75,000
Theft	3rd Degree	$5,000–$20,000
Theft	4th Degree	$1,000–$2,500
Theft	Disorderly Persons	$500–$1,000
Prostitution	2nd Degree	$25,000–$50,000
Manufacturing/Distributing CDS	1st Degree	$100,000–$250,000
Possession of CDS or Analog	3rd Degree	$5,000-$10,000
Simple Assault	Disorderly Persons	$500-$2,500
Shoplifting	Disorderly Persons	$500-$1,000

Other states leave it to counties to decide their own bail schedules, which can get quite detailed. The bail schedule for Alameda County, California, runs to nine pages, for instance. No two bail schedules are exactly alike, and often the same charge might incur widely different bail amounts from one county to the next.⁴

Once the bail amount is set, either by schedule or at the discretion of the judge, the defendant faces a crucial choice: how to “make bail.” Some states permit defendants to pay the amount directly to the court (known as a “cash” bond), or in much rarer circumstances, to pay a portion of the bail or deposit secured collateral with the court. Of course, most people—let alone criminal defendants—do not have $100,000 or more ready to go when demanded. Thus most defendants, whatever their relative level of wealth, turn to a commercial bail bondsmen to secure release.⁷ The only other alternative is to remain in jail pending trial, a process that can take several months (longer for felony trials), or in rare instances, to seek judicial review of the bail amount imposed.

The commercial bail industry may conjure images of Dog the Bounty Hunter stepping in to help out the shops with neon signs down by the courthouse, but the day-to-day work of bail bonding looks more like the mundane operations of a small insurance office—which is what most bond shops are. Typically run by one or two bondsmen often employing family members as support staff, a bail bond shop creates a web of contracts that secure a defendant’s release in exchange for payment and, often, the posting of collateral. Here’s how the web works:

Bail bondsmen are underwritten by large surety conglomerates like Lexington National, American Surety Company, and Bankers Insurance Company. These surety companies are the big players of the bail industry—highly capitalized supercenters of organized information and strategic lobbying, often with tax shelters in the Cayman Islands to ensure maximum secrecy for their operations.⁸ These firms provide the capital, insurance, and licensing for local bond shops to operate, usually in exchange for 10% of a bond shop’s revenues. The bond shop in turn pledges to pay the court the full amount of the bond if there is a forfeiture, and based on the bondsman’s licensing, the court accepts the pledge without further payment. The bondsman then contracts with the defendant on the terms securing the bond and paying the fee. All of these contracts are private agreements and most contain (repeated) confidentiality clauses or separate non-disclosure agreements, making the whole industry highly opaque and difficult to study.

It bears emphasizing that the agreement between a criminal defendant and a local bondsman is, in theory, a privately negotiated contract. The terms can include anything the parties “voluntarily” agree to. Even if a court does not require weekly check-ins or drug testing as a condition of release, a bondsmen might require these things as a condition of securing the bond. In addition to these conditions, the bondsman charges a premium as well as processing fees. Most jurisdictions place a legal limit on these charges, usually around 10% to 15% of the face amount of the bond. In other words, if the court sets bond at $100,000 a bondsman will pledge that amount to the court in exchange for the defendant paying a nonrefundable premium of $10,000 plus fees. Some jurisdictions permit bondmen to offer financing agreements and charge additional interest on the premium if defendants cannot pay all at once, although these arrangements can be controversial within the industry.⁹ While in theory these are freely negotiated agreements, they are often conducted by defendants or their families under conditions of high stress and urgency, where the alternative to a failed negotiation is the indefinite detention of the accused.

Besides the premium, the most important term in the bond contract is the collateral. Bondsmen will rarely take on the full risk of a forfeiture. Instead, they usually demand co-signers and/or other security, including liens on bank accounts, automobiles, and real estate. It’s in that last category that the bail process becomes more transparent. Any interest in real estate—like a bail lien—must be recorded to be protected in case of sale or foreclosure. Bail bondsmen protect their security by filing public notice of the lien, like this:

Note that in the example, the lien specifies the original amount of the bond—$100,000—and adds an additional debt of $7,500—likely the unpaid portion of a $10,000 premium. The lien also names the defendant and gives a little information about the court that charged the defendant. In this instance, as in most, the defendant is not the same person as the owner of the property but is likely a close family member.

If the defendant makes all scheduled court appearances, the court will “exonerate” the bail—i.e., dissolve the bondsman’s pledged liability. So long as the premium has been paid in full, the bondsman should then enter a release or satisfaction of the lien.

As criminal trials can take years to resolve, bail liens can burden a property for a very long time. Moreover, many bond shops are slow or neglect to enter releases on the record. Sometimes a release is not entered until a homeowner goes to sell the property years after trial proceedings have ended only to discover that the bondsman’s lien is still there. Across the Cities of Bail database, only 40% of liens have a release recorded. While many liens may still be running because proceedings are not resolved, many more are simply the result of bail bondsmen’s oversight.

If a defendant fails to appear for a critical hearing or trial, the court can start the process of forfeiting the bond, but forfeiture does not happen immediately. Most jurisdictions give an initial warning and extension of time, and some jurisdictions give especially favorable terms to commercial bail bondsmen—such as six months’ forbearance—to see if the defendant returns. During this period bail bondsmen might hire specialized “recovery agents,” otherwise known as bounty hunters, who specialize in tracking runaway or missing defendants in exchange for a portion of the bond premium.

If the defendant re-appears within the grace period, the court might reassess the bond and then start the process over. If too much time elapses without an appearance, the court can forfeit the bond. Forfeiture is a complex and highly opaque process within the courts. One exposé contends that some courts never hold bail bondsmen to their pledges, functionally turning all bond premiums into pure profits for the industry. Other jurisdictions may forgive the bondsman’s debt depending on the circumstances, like the re-arrest of the defendant in another jurisdiction.

In any event, a failure to appear is usually a violation of the contract between the bondsman and the defendant, giving the bondsman the right to foreclose on mortgaged property regardless of what actions the court decides to take. That is, even if the bondsman loses nothing and bears no risk, the defendant’s home (or their parents’) may still be foreclosed. That does not necessarily happen often. Bail bondsmen rely heavily on the threat of foreclosure to make contact with missing defendants and pressure them to appear. In many cases that pressure can be more valuable to the industry than a house sold at foreclosure prices.

Although bail systems are broadly the same across the United States, generally each city and county is in charge of its own jail and bail policies, so bail regulation is an intensely local affair. Instead of one “bail system,” the U.S. really contains about 3,000 bails systems. The same can be said about property recording, where each locality has its own preferred terminology—deed of trust, mortgage, lien, etc.—and idiosyncratic ways of maintaining and retrieving records.

Sometimes those idiosyncrasies can frustrate research. Pennsylvania, for instance, totally removes criminal charging data from public view if charges are dropped or a defendant is acquitted. While this well-meaning reform protects the privacy of the exonerated, it also makes it impossible for researchers to know when bail was paid or homes foreclosed on defendants who were later deemed innocent. Other times, the particular array of regulations and data collection can help us better visualize how bail works across the system. That is the aim for the six urban regions studied in Cities of Bail: Albuquerque, New Mexico; Tucson, Arizona; Oakland and San Francisco, California; Jersey City and Newark, New Jersey; Cleveland, Ohio; and Pittsburgh, Pennsylvania.

Conceptualized by a professor at Columbia Law School, Cities of Bail leverages public data from criminal justice records and from recorded property deeds to shed light on the overall legal system of bail. While similar studies could be conducted in any U.S. jurisdiction, the six metro areas represented in this study share several characteristics that made them ideal for a digital project: 1) each urban area occupies a single county (or in the case of California and New Jersey, two counties) that are all subject to the same bail regulations and use similar property recording techniques, 2) each jurisdiction makes its property records available online, either for free or upon payment of a small fee, and 3) records extend back to the year 2000 or before.

In addition, the six urban areas offer useful comparisons and contrasts. Tucson and Albuquerque are in the Southwest, Pittsburgh and Cleveland in the Rust Belt, while Newark/Jersey City and Oakland/San Francisco represent larger metro areas on the coasts. New Jersey and New Mexico engaged in extensive bail reform in 2016 to 2018 while the others did not, making comparisons between those jurisdictions especially stark in the latter period of this study. Because regulations differ in each jurisdiction, the same data is not always available for comparison, but to the extent possible, Cities of Bail aggregates property records, criminal charging documents, and foreclosure notices to paint as complete a picture as possible of how bail works for those who have the property to meet the system’s most extreme demands.

Researchers at Columbia Law School searched online records hosted by county clerks’ offices for liens made out to bail bondsmen or bail bond shops from the year 2000 through the end of 2020. While it is possible that some bail bondsmen wrote liens in their own names or in the name of an entity that would elude searches for “bail” and “bond,” searching known names of bail bondsmen and their families confirmed that liens were typically granted in the name of the bail bond shop, or in rarer cases, the surety corporation as at least one of the parties to the lien. All of the (mostly handwritten) data from the liens, available releases, and other property documents such as lis pendens, was then keyed into a master database.

Where liens identify the defendant whose bail was secured by the lien (which over 90% of the liens do), legal research assistants queried criminal case documents where available in order to collect demographic data on the defendants, charges, and case dispositions.

When property is sold at foreclosure, any liens are satisfied out of the proceeds of the sale and a release is entered on the record. Consequently, foreclosures can be difficult to track in property records—a voluntary sale, a forced sale, or any exoneration of the bond will all have the same result: a recorded satisfaction of the lien. But many cities post foreclosure sale notices in newspapers that are now archived online. Where available, researchers searched every address in the database for a foreclosure sale during the study period and recorded information about the timing and parties to the sale.

The result is a relational database containing over 8,000 documents of liens, releases, criminal docket sheets, and foreclosure notices. Staff at the Center for Spatial Research used this data, as well as demographic data available from the U.S. Census Bureau, to construct the maps, stories, and visualizations for the Cities of Bail project site.

All of the data used in Cities of Bail comes from publicly available sources. But individual identifying information is often stored behind many layers of lookups and fee-based retrieval systems, and records concerning the same individual are never aggregated in the public databases as they in this project. To protect privacy, Cities of Bail does not make personally identifying information, including the full database underlying this study, available on its public-facing pages. If you are a researcher seeking de-anonymized data, please contact the project director.