Gun violence remains a serious problem in the USA, with over 11 000 gun homicide victims and some 467 300 victims of non-fatal firearm crime.1 The annual medical, criminal justice, and other government and private costs of gun violence have been estimated to be >$170 billion.2 There are also some 300 million privately owned firearms, including roughly 100 million handguns, in the USA.3 Given the prevalence of guns, some observers suggest that gun control restrictions on commerce and possession of firearms are futile in limiting criminal access to firearms.4

States vary greatly in the nature of their gun sales laws and gun dealer regulation policies and procedures.5 A growing body of empirical evidence suggests that more restrictive state-level firearm sales laws and better regulation of gun dealers may reduce the illegal diversion of guns from lawful commerce to criminals.6 In addition to having strong gun sales laws, California closely regulates firearms sales by in-state retail dealers. Bureau of Alcohol, Tobacco, Firearms, and Explosives (ATF) crime gun trace data were used to analyse the impact of California's more restrictive firearms commerce laws and regulatory regimes relative to legal and regulatory schemas in other US states on a key measure of the illegal diversion of firearms to criminals—the amount of time between a firearm's first legal purchase and its ultimate recovery by a law enforcement agency following its use in crime, known as time-to-crime.

State firearms commerce laws and dealer regulation

Federal, state and local governments regulate commerce in firearms and the possession and use of firearms. Most jurisdictions occupy the middle ground between laissez-faire and prohibition, seeking to preserve legitimate uses of guns while minimising their use as an instrument of criminal violence.7–10 A primary purpose of federal law is to prevent lax firearms controls in one state from undermining more restrictive regulations in another state. State gun laws and regulations vary considerably. A series of research studies have found that crime guns tend to migrate from states with less restrictive firearms sales laws to states with more restrictive laws.11–14 A recent study used 2009 state-level ATF crime gun trace data to examine the cross-sectional association between state gun laws and the per capita rate of crime guns exported to criminals in other states, controlling for potential confounders.6 States that exported the most crime guns per capita (Mississippi, West Virginia and Kentucky) had none of the state gun laws examined. States that exported the fewest crime guns per capita (New York, New Jersey, Massachusetts and California) had strong dealer oversight, regulated private sales and had other stringent firearms commerce laws.

More restrictive state firearms sales laws do seem to reduce the ease through which criminals illegally acquire guns from in-state dealers. A study that used aggregate ATF crime gun trace data from 53 US cities for the years 2000–2002 examined the association between state gun regulations and the diversion of guns to criminals.15 The illegal diversion of firearms to criminals was measured as the number of guns recovered by police within 1 year of retail sales for those guns where the legal retail purchaser was not the criminal possessor. Strong regulation and oversight of gun dealers was associated with 64% less diversion of guns to criminals by in-state gun dealers. The study also found that regulation of private handgun sales and discretionary permit-to-purchase licensing were also associated with lower levels of diversion of guns sold by in-state gun dealers. Similarly, when Missouri repealed its permit-to-purchase law, the number of crime guns with very short sale-to-crime intervals increased sharply.6

California has firearm-related laws and regulations that are relatively more stringent than those in most other states. Californian residents are limited to one handgun purchase per month and a 10-day waiting period before purchased firearms can be acquired. With rare exceptions, private-party transfers must be routed through a licence retailer. Since 2003, a purchaser's right thumbprint is required on all transactions and the purchaser's name, date of birth, address and unique ID number are entered by ‘swiping’ a government-issued photo identification card (California driver's licence or military ID card). These regulations are designed to reduce the potential for purchasers providing false identification to firearms dealer. Beyond those convicted of a felony or domestic violence misdemeanour and other federal law prohibitions, California has adopted broader exclusions, which include persons convicted of any violent misdemeanour, such as simple assault and battery, persons served with temporary domestic violence restraining orders and persons admitted for emergency psychiatric evaluation on grounds that they pose a danger to themselves or others.

California also regulates firearm retailers to a unique degree. In addition to a federal firearms licence, all persons engaged in the business of selling firearms must possess a state Certificate of Eligibility and be named on the state's Centralized List of firearms retailers. With rare exceptions, firearm sales must be processed electronically, using software developed by the California Department of Justice and in most cases on computers supplied by the department. A sale cannot be accepted for processing unless the transaction data are complete and consistent. The California Bureau of Firearms maintains its own corps of seven field representatives who serve as inspectors. Retailers are inspected in their first year of operation and approximately every 2 years thereafter, on average. Inspections may occur more frequently if there has been a history of rule violations in the past, and less frequently if the retailer has previously been in compliance. Scheduling of inspections also takes into account the retailer’s sales volume and any referrals from law enforcement agencies or the general public. Repeated and wilful violations can lead to a licensee being removed from the Centralized List. This robust commitment to dealer regulation and enforcement of state firearms sales laws distinguishes California from other states, such as Massachusetts,12 that have strong state gun laws but lack well-developed regulatory and enforcement regimes.

Data and analytical framework

The ATF firearm trace data used in this article were collected through a US National Institute of Justice grant to study illegal gun markets. This kind of research is no longer possible since the 2003 passage of the Tiahrt Amendment that greatly restricts the use of such data by non-law enforcement personnel. As such, this study advances our understanding of the impact of state firearms sales laws and dealer regulations on the illegal diversion of guns by analysing the actual characteristics of specific traced crime guns, including interstate and intrastate movement patterns, rather than being limited to analysis of aggregate state-level and city-level summaries of ATF firearms data.

The Gun Control Act of 1968 (GCA) established a set of requirements that allows any given firearm to be traced from its manufacture or import to its first sale by a retail dealer. GCA mandates that each new firearm, whether manufactured in the USA or abroad, must be marked with a unique serial number. In addition, GCA requires all federally licensed firearms dealers, including manufacturers, importers, distributors and retail dealers, to maintain records of all firearms transactions. Firearms traces can be unsuccessful for a variety of reasons such as local police incorrectly completed the trace request form, the firearm was too old to trace (pre-1968 manufacture) or the gun had obliterated serial numbers.

ATF trace data can provide policy-relevant insights on illegal gun market dynamics when conclusions are based on careful analyses that are coupled with clear acknowledgements of the data limitations.16 ,17 ATF and academic analyses of firearms trace data typically focus on a critical dimension of the illegal firearms market: the time between a firearm's first sale at retail and its subsequent recovery by a law enforcement agency, most often in connection with a crime (‘time-to-crime’). Law enforcement investigators consider a traced firearm with short time-to-crime, defined as recovery within 3 years of first retail sale, as possibly having been recently and illegally diverted from a retail outlet.18

Our analysis examined 840 863 ATF trace requests for all recovered crime guns in 50 US states between 2003 and 2006. Trace requests for firearms recovered in California accounted for 11.1% of the total (93 292 of 840 863). ATF successfully traced 58.7% (493 688 of 840 863) of these recovered crime guns to a first retail purchaser. Short time-to-crime firearms represented 16% of the traced firearms recovered in California relative to 28% of the traced firearms recovered in other US states. Moreover, the mean time-to-crime was 12.2 years for California traced firearms versus 9.8 years for traced firearms recovered in other US states (a 26% difference).

These data were further restricted to examine traced firearms where the purchaser and possessor were identified as different individuals, indicating these guns changed hands at least once between their first retail purchases and recovery by law enforcement agencies. Of the 493 688 firearms traces with time-to-crime data, there were 228 305 cases (46.2%) where the purchaser and possessor were identified as different individuals. The other firearm traces involved possessors and purchasers identified as the same person or sufficient information on the crime gun possessor was not available.

To examine whether California's firearm-related laws and regulations impacted the distribution of time-to-crime of recovered firearms in that state relative to other US states, we grouped states according to the presence or absence of firearms purchase and registration laws in the state. States were classified as having no purchase and registration laws, purchase-only laws, registration-only laws or both purchase and registration laws based on the Vernick and Hepburn study of state and federal gun laws.19 Firearms purchase laws require prospective gun buyers to first obtain a licence or permit before purchasing a firearm. Registration laws mandate permanent records of each gun sale that are kept by states in centralised locations. As of 1999, six states had purchase laws only (Iowa, Illinois, Minnesota, Montana, North Carolina and Nebraska), two states had registration laws only (California and Maryland) and five states had both purchase and registration laws (Connecticut, Hawaii, Massachusetts, Michigan, New Jersey and New York).

California, which has purchase but not registration legislation, was kept separate because it is the focus of this research and has other relevant firearms laws and regulations that provide for a more stringent regulatory context in terms of the potential trafficking of firearms. These dichotomous independent variables were included in the analysis as a series of dummy variables. To investigate the relationship between state firearms laws and in-state/out-of-state gun purchases, a set of seven legal context/place of purchase dummy variables was constructed by interacting the four purchase and registration law contexts with the two place of purchase conditions. The reference category for this group of dummy variables is in-state firearm purchases in states with no purchase and no registration laws.

The rapid diversion of guns from legal commerce to recovery in crime by law enforcement agencies may not only be dependent on the place of purchase and the legal context of state firearms laws but also on other characteristics of local illegal firearms markets.14 ,17 ,20–22 For that reason, a series of gun trafficking indicators based on specific characteristics of the licensed firearm dealer, first retail purchaser, crime gun possessor, purchaser–possessor relationship and recovered crime gun were included in the multivariate analysis. These same firearm commerce-based gun trafficking indicators have been shown to be significant predictors of time-to-crime in previous empirical analyses.11 Dealer control variables included the number of traces to a dealer, the number of multiple gun sales sold by a dealer, a dummy variable for whether the dealer was a pawnshop and a dummy variable for whether a dealer was in business as of 2006 is included as a control variable. Dummy variables were used to control for the gender and age of crime gun purchasers and possessors. Other purchaser control variables included the number of traces to an individual purchaser and the number of traces from a purchaser's home residence zip code. Purchaser/possessor relationship indicators included the difference in age between purchaser and possessor, the distance between a purchaser's and possessor's home, and a dummy variable for purchasers with the same last name but different first name from a possessor.

A dummy variable was used to indicate whether the recovered crime gun was a semiautomatic pistol or another type of firearm. Finally, a modest number of jurisdictions engage comprehensive firearms tracing practices and submit guns recovered in a broader range of circumstances to ATF for subsequent tracing. Traced firearms recovered in domestic disputes, found on the street and seized for health and safety reasons are likely to be older firearms. Therefore, we included a dummy variable indicator indicating comprehensive tracing practices associated with these types of recovered firearms.

Descriptive statistics for the dependent and independent variables are listed in table 1. The distributions of variables that represent counts of events were smoothed through natural log transformations.23 The number of cases for the multivariate analysis drops slightly to 225 392 traces due to a small amount of missing information on several explanatory variables. The level of missing data was larger for the gun trafficking indicator derived from the geographic distance between the home zip code of the first retail purchaser and the home zip code of the gun possessor (8.7%; 19 765 of 225 932 cases). To prevent dropping these cases from the empirical analysis, a dummy variable was included in the analysis to control for missing information on this particular indicator.24

Data that measure lifetimes or the length of time until the occurrence of an event are generally called survival data.25 ,26 In this analysis, we are interested in modelling the length of time between the first retail sale of a firearm and its subsequent recovery following use in crime by law enforcement. Survival data are often censored. Of the 225 392 firearm traces in the analysis, 17 627 firearms with time-to-crime >20 years were censored (7.8%). These cases were right censored at 20 years because federally licensed dealers are not required to maintain records on the sale and purchase of firearms beyond that time. Cox proportional hazards models were used to analyse the time-to-crime (failure time) for traced crime guns included in this multivariate analysis.