Research and analysis

Evaluation of the County Lines Programme

Published 14 August 2025

Applies to England and Wales

Executive summary

The report evaluates the County Lines (CL) Programme’s impact on various aspects of crime, including serious violence and acquisitive crime, law enforcement activities, safeguarding referrals, drug misuse hospitalisations, and County Lines Programme metrics.

Using a difference-in-differences methodology, the analysis compares changes in outcomes between areas affected by the Programme and those unaffected, before and after its implementation.

The evidence suggests that, following the implementation of the County Lines Programme, in exporter forces:

The CL Programme led to a decrease in serious violence measured as any weapon use hospitalisations (sharp weapons and firearms)^{[footnote 1]},^{[footnote 2]}

Results show a 21.6% decrease or, on average, 33 fewer quarterly cases of hospitalisations for weapon use as a direct implication of the CL Programme. This decrease is driven by a significant decline of 18.9% in sharp weapon hospitalisations. The reduction in weapon use hospitalisations was seen immediately after the start of the Programme.

Additionally, the coefficients for the effects of the CL Programme on drug related homicides and lethal barrel discharges are also negative, yet the effects are not statistically significant. The lack of statistical significance of the effects is primarily due to the low precision of the estimates, which is expected for these measures given the low volumes and volatility of the data.

There has been a 19% increase in police-recorded violence after CL Programme implementation. Improved police recording practices and heightened police presence in the exporter areas may have contributed to this as part of the CL Programme deployment. These channels require further exploration.

In absolute value, there were, on average, 3,650 more quarterly police-recorded violent crimes in the exporter areas after implementing the CL Programme. This effect is statistically significant and, therefore, attributable to the CL Programme. These results are driven by increases in incidents recorded as violence with injury and violent disorders. In terms of dynamics, the effects are increasing over time, taking close to one year to materialise.

On this matter, it should be noted that (1) improved police recording practices largely driving increases in recorded violent crime^{[footnote 3]}, and (2) increases in and heightened police presence could be also taking place in exporter forces as a result of the CL Programme with a focus on violent crime.

Negative changes in acquisitive crime were not statistically significant and therefore not attributable to the CL Programme, with the exception of theft from vehicles.

Whilst the overall impact on acquisitive crime was not statistically significant, there was a significant decline in theft from vehicles by 15.4% (or 985 per average quarter) observed in exporter areas, with this being directly attributable to the Programme.

There has been a 25.1% increase in the number of law enforcement activities due to the CL Programme, mostly measured through an increase in police recorded drug possession offences.

This is equivalent of 1,077 more quarterly offences for drug possession and drug trafficking, on average, which can be interpreted as better targeting and more successful law enforcement activities following the launch of the CL Programme.

Positive changes in NRM safeguarding referrals were not statistically significant and therefore not attributable to the CL Programme. The volatile pattern in safeguarding referral data following the launch of the CL Programme makes it difficult to draw firm conclusions about its impact.

As additional data is gathered on the CL Programme, particularly around safeguarding referrals, the volatile patterns are likely to stabilise, which will in turn allow for a more accurate estimate of the impact of the CL Programme

The CL Programme has resulted in a 14.4% decrease in drug misuse hospitalisations (or 40 fewer quarterly drug misuse hospitalisation cases on average).

The dynamic effects indicate that the effect is immediate, that it persists and is even more pronounced over time.

As expected, there has been a statistically significant effect on the total number of lines closed and on individuals arrested and charged.

While the impact on closures seems to consistently increase over time, for arrests and charges, the trend flattens around 18 months post-policy implementation. This could be due to data coverage, and further analysis of this dimension could be crucial in the future.

In the ‘top’ importer forces:

Overall, the effect on top importer areas due to implementing the CL Programme in the exporter areas was relatively minimal. Hence, it can be concluded that the effect of the Programme is likely to be rather localised to the exporter areas.

Estimates across the various outcomes are not statistically significant, indicating that the CL Programme activity and funding did not directly impact these ‘top’ importer areas for such crime categories. However, the signs of the estimates are in line with what we would expect and there is a sizable decrease in acquisitive crime following Programme implementation, with a statistically significant reduction in theft of vehicles which can likely be attributed to the CL Programme.

1. Introduction

1.1 Background

‘County lines’ is a term used to describe gangs and organised criminal networks involved in exporting illegal drugs (primarily crack cocaine and heroin) into one or more importing areas (within the United Kingdom), using dedicated mobile phone lines or other forms of “deal line”. They are likely to exploit children and vulnerable adults to move and store drugs and money. They will often use coercion, intimidation, violence (including sexual violence) and weapons. A key objective of His Majesty’s Government is to tackle county lines.

This project:

• implements quantitative techniques using a multi-tiered geographical approach to define intervention and control areas, splitting them between exporters, top importers, and other areas; in this way, it can account for spillover effects by using data and intel to separate the ‘top’ importers where the largest spillover impacts may occur from the mostly unaffected areas as a pure comparison group
• uses causal regression analysis that, under the validity of its research design (for example, a strong counterfactual), provides a solid and robust assessment of the CL Programme
• this evaluation includes information on several measures of serious violence, Programme metrics, acquisitive crime, and concurrent investments

The approach intends to arrive at more definitive conclusions regarding the County Lines Programme’s influence on drug-related crimes and drug-related harms.

The research’s further discoveries are anticipated to bolster our knowledge foundation on this subject, contribute to shaping policy decisions concerning county lines, and provide valuable input for upcoming funding requests during the upcoming Spending Review.

1.2 Aim and research questions

The impact evaluation focuses on reviewing, updating, and expanding the understanding of the County Lines Programme. This evaluation will analyse the Programme’s effect on serious violence, acquisitive crime, law enforcement activities, safeguarding referrals, drug misuse hospitalisations, and county lines metrics.

The broad research questions this report intends to answer are the following: What is the impact of the County Lines Programme on crime levels? The broad research question is materialised in the following concrete research questions:

• what is the impact of the County Lines Programme on serious violence?
• what is the impact of the County Lines Programme on acquisitive crime?
• what is the impact of the County Lines Programme on law enforcement activities?
• what is the impact of the County Lines Programme on safeguarding referrals?
• what is the impact of the County Lines Programme on drug misuse hospitalisations?
• what is the impact of the County Lines Programme on CL metrics?

2. Data and variable definitions

The datasets required for this analysis have been chiefly provided through the Home Office. All information is sanitised or aggregated, so no individual information is identifiable. The spatial unit of observation is a Police Force Area, and the frequency of observations is a quarter.

2.1 Outcome variables

We observe 6 primary constructs for which we assess the impact of the County Lines Programme: serious violence, acquisitive crime, law enforcement activities, safeguarding referrals, drug substance use hospitalisations, and County Lines metrics. These serve as our main dependent variables and have been constructed in the following manner:

1. Serious violence data: The data on serious violence is analysed in 4 different components:

   • (i) police-recorded violent crimes (PRC): this variable is the sum of violence without injury, violence with injury, violent disorders, and threats to kill
   • (ii) drug-related homicides
   • (iii) weapon use hospitalisations: this includes information on hospitalisations due to assault with sharp weapons and firearms
   • (iv) lethal barrel discharges
2. Acquisitive crime: This variable is the sum of police-recorded crimes on domestic burglary, theft from the person, robbery of personal property, theft from a shop^{[footnote 4]}, theft of vehicles, theft from vehicles, and vehicle interference.
3. Law enforcement: This variable is the sum of police-recorded crimes involving drug possession, trafficking, and weapons possession.
4. Safeguarding referrals: This variable is the sum of total National Referral Mechanism (NRM) referrals per force and quarter, which includes NRM county lines referrals and all other NRM referrals
5. Drug misuse hospitalisations: This variable is the sum of hospitalisations due to substance abuse^{[footnote 5]}. Specifically, any hospitalisation because of drug poisoning by illicit drugs makes up this metric.

6. County Lines Programme Metrics: includes:

   • (i) total number of line closures
   • (ii) total number of arrests and charges

Table 1 below explains each dimension’s definitions in more detail, the data used to construct each dimension, the timespan covered, and data issues.

Table 1: Main regressors

Name	Type	Variable	Data source (‘Q’ refers to calendar years)
1. Serious violence	Outcome	i. Police-recorded violent crimes - Sum of violence without injury, violence with injury, violent disorder	PRC data (2016Q2 - 2023Q2)
		ii. Drug-related homicides	Home Office (2018Q2 - 2023Q4)
		iii. Weapon use hospitalisations - The sum of sharp weapons and firearms	NHS Digital (2018Q2 - 2024Q1)
		v. Lethal barrel discharges	Home Office (2016Q2 - 2023Q4)
2. Acquisitive crime	Outcome	Sum of: domestic burglary, theft from the person, robbery of personal property, theft from a shop, theft of vehicles, theft from vehicles, and vehicle interference.	PRC data (2016Q2 - 2023Q2)
3. Law enforcement	Outcome	Sum of possession of drugs, trafficking of drugs, and possession of weapons.	PRC data (2016Q1 - 2023Q1)
4. Safeguarding referrals	Outcome	All NRM referrals (includes county lines referrals and all other NRM referrals)	UKDA (2014Q1 – 2023Q3) Home Office (2019Q1 - 2023Q3)
5. Drug misuse hospitalisations	Outcome	Sum of hospitalisations because of drug poisoning by illicit drugs.	NHS Digital (2018Q2 - 2024Q1)
6. CL Programme Metrics	Outcome	i. Total Line Closures	Home Office (2019Q1 - 2023Q3)
		ii. Total arrests and charges	Home Office (2019Q1 - 2023Q3)

2.2 Dependent variables

The dependent variables of this analysis are divided into 2 categories: main regressors and control variables.

On the main regressors, the treatment definition of exporters and ‘top’ importers comes from the Organised Crime Group Mapping (OCGM)^{[footnote 6]} data output provided by the Home Office, intel from the forces, and Surge funding data. The methodology section presents a more thorough explanation of these definitions. Moreover, the Home Office provided information on county lines funding from January to March 2020 to October to December 2023.

Regarding control variables, we include information on Violence Reduction Unit (VRU) funding and GRIP funding in our primary dataset, obtained from the Home Office. These regressors are included to isolate the effects of the County Lines Programme from those stemming from the VRU and GRIP Programmes, which operate in similar areas and have similar objectives.

Table 2 below explains each dimension’s definitions, data, and timespan.

Table 2: Main regressors

Name	Type	Variable	Data source
Exporter/Importer	Treatment variable	Treatment: Exporters Metropolitan Police, West Midlands Police, Merseyside Police, and Greater Manchester Police Treatment: ‘Top’ Importers Lancashire, Cheshire, West Yorkshire, Northumbria, Cumbria, Essex and Kent Control: All other forces	Home Office October 2023. The datasets used comprise: OCGM Data Exporter forces’ intelligence Surge funding data
CL Programme funding	Regressor	County lines funding	Home Office (2020 Q4 - 2023 Q4)
Other funding data	Control	VRU funding GRIP funding	Home Office (2019 Q4 - 2023 Q2) Home Office (2019 Q2 - 2023 Q2)

2.3 Master data

The master dataset is a balanced panel dataset, in which the observation unit is a police force (N=43), and the time-frequency is a quarter. As outlined in Tables 1 and 2, the length of time coverage depends on the variable under observation.

The dataset contains 198 variables gathered from the sources discussed previously. These variables have been further processed to construct our primary outcomes, the regressors, and the controls to be included in the final estimation.

Data issues

Regarding the construction of the dataset, it must be noted that there were some concerns about the quality of the data received, which resulted in some adjustments to the master dataset. These are as follows:

First, the PRC data showed negative values for specific forces and quarters. This concern was shared with the Home Office. As negative values are not possible to trace back to a specific force, quarter, and type of crime (all of which are important to the present analysis), it was decided to drop earlier years in the PRC data. This is because the early years are those that are mainly suffering from this problem. There are 409 negative values for different offence groups recorded between 2012 and 2015, as reflected in Table 3 below. Consequently, the PRC data will only be considered from 2016 onward.

Second, the VRU funding data also showed inconsistencies. There were duplicate registries, as well as missing ones, and there was no straightforward procedure on how to fix these issues. The Home Office was consulted, and the agreement was to use the annual FY amount for each force, which is accurate. This was further split up evenly into quarter-level data. As long as the funding start date for each force is correct, we believe such imputation would not introduce significant disruptions to the policy assessment. With those adjustments made, the master dataset is completed.

Table 3: Negative values in PRC data

Year	Total negative values	Max value	Min value
2012-13	212	-1	-268
2013-14	104	-1	-13
2014-15	77	-1	-11
2015-16	16	-1	-58
2016-17	12	-1	-2
2017-18	0
2018-19	5	-1	-2
2019-20	0
2020-21	0
2021-22	0
2022-23	0
2023-24	0

Source: PRC data

3. Methodology

This evaluation consists of a series of regression models, where the Programme’s causal effects will be identified using a difference-in-differences strategy. This approach was selected due to its effectiveness in quasi-experimental settings where there is no random assignment to treatment (in this case, the CL Programme). We can accurately estimate the Programme’s impact by comparing changes in outcomes between units affected by the Programme (the treatment group) and those unaffected (the comparison group) before and after Programme implementation.

To assess the impact of the County Lines Programme, as already explained, we define 3 different population groups and 2 treatment groups. The first population group consists of the exporter forces. This is also the first treatment group. The second population group consists of the top importer forces. This is the second treatment group. The third population group comprises all other forces, which County Lines less obviously impacts. This is the comparison group. The definition of each group and which forces are included in each is presented in the following sub-section.

The decision to incorporate 2 treatment groups stems from the multifaceted nature of the County Lines Programme. We aim to capture the distinct roles and dynamics within the Programme’s operation by distinguishing between exporter and top importer forces. Including “top” importers allows us to account for the potential spillover effects that may be expected in the areas that do not receive direct CL funding but where there is evidence that a large proportion of county lines operate. This nuanced approach allows for a comprehensive assessment of the Programme’s effects on different population segments. Additionally, including a comparison group comprising forces largely unaffected by county lines enables us to isolate the Programme’s causal impact from other external factors.

3.1 Treatment definition

To assess the impact of the County Lines Programme, we define 3 different population groups and 2 treatment groups.

The first population group consists of the 4 exporter forces. These are defined as the forces with the most lines that receive specific County Lines funding, aligning with previous definitions of exporter forces. The defined group of exporter forces is as follows:

• Metropolitan Police
• West Midlands Police
• Merseyside Police
• Greater Manchester Police^{[footnote 7]}.

This is also the first treatment group.

The second population group consists of the top importer forces. These are defined as the importer areas with the most significant amount of county lines activity, making them most likely to be impacted by spillover effects from the County Lines Programme.

The definition of the top importer forces is not as straightforward as the exporters. As all other forces are indeed importers, this definition aims to identify which are the top ones and, therefore, the ones that might be most affected by the CL Programme. To define areas with the highest impacts, we rely on several sources. The first source is the OCGM data, which is a snapshot of the number of drug-related OCGs recorded in each PFA. It identifies OCGs moving drugs between importer and exporter areas, and it is helpful in determining which forces receive the most lines overall. The second source is the Surge Funding data, indicating CL-funded police activity in those importer areas. The third source is intelligence information provided by the exporter forces.

We use 3 different data sources to define top importers for several reasons. Firstly, as the definition of top importers is unclear, using a composite definition alleviates concerns about biasing the definition using only one source. Second, none of the 3 sources used (OCGM, Surge Fund, or intelligence) provides an ideal source for defining this variable. For example, the OCGM data provides a detailed image of OCGs operating across forces. Still, it is a contemporary and static picture, which might not accurately portray the situation when the CL Programme started. Thirdly, these sources are not always consistent regarding the relevance of importer forces. For that reason, using contemporaneous OCGM data alongside past data on the Surge Fund and intelligence from the forces helps alleviate potential concerns or caveats of the existing data to define top importers.

From the latest information, we understand the top importer forces are:

• Lancashire
• Cheshire
• West Yorkshire
• Northumbria
• Cumbria
• Essex
• Kent

These forces were identified as top importer areas due to critical breaks in the OCGM data and the Surge Funding data distribution. Lancashire, Cheshire, West Yorkshire, Northumbria, and Cumbria stand out as the top importer forces overall, with closer values of incoming lines among themselves. The following ranked force has a substantially lower number of incoming lines. This definition also largely matched the information from the intel from forces. Moreover, the inclusion of Essex and Kent stems from the Surge Funding data, which directly indicates CL-funded police activity in those importer areas. Of the top 10 forces, Kent and Essex stand out in this dataset, as well as the intel received from forces. For those 2 reasons, they are also included in this analysis’s set of top importer forces. This is the second treatment group.

The third population group consists of all other forces, which will be mostly unaffected by county lines. This is our comparison group.

3.2 Estimating equations

We propose the following 2 regression models for the primary outcome of interest. In these models, we will isolate the impact of the County Lines Programme on (1) exporter forces, and (2) importer forces.

(1) Crime_it = α + β₁ Exporter_it+ γX_it + δ_i + λ_t + ε_it

(2) Crime_it = α + β₂ Importer_it+ γX_it + δ_i + λ_t + ε_it

Where _i indicates the spatial unit (police force) and _t indicates the time unit (quarter). The outcome variable, Crime, will be measured using a set of variables: serious violence, acquisitive crime, law enforcement activities, safeguarding referrals, drug substance use hospitalisations and CL Programme metrics.

The main independent variables, Exporter and Importer, will define the treated units Exporter_it is an indicator variable that takes value one in exporter forces after receiving the county lines funding. Additionally, Importer_it is an indicator variable that takes value one in the top importer forces after the county lines funding was received in their main exporter force. In both specifications, X_it represents control variables including VRU and GRIP funding. δ_i and λ_t are fixed effects for police force and quarter respectively.

In equation 1, we keep populations groups 1 and 3, and β₁ will reflect the causal effects of the County Lines Programme on the main outcomes of crime: serious violence, acquisitive crime, law enforcement, safeguarding referrals, and drug use hospitalisations for exporter forces. In equation 2, we keep populations groups 2 and 3, and β₂ will reflect the causal effects of the County Lines Programme on the main outcomes of crime for the top importer forces.

4. Results

4.1 Baseline estimates

Exporter forces

The estimated effects of the County Lines Programme for exporter forces are presented in Table 4^{[footnote 8]}.

The CL Programme has likely resulted in a reduction in weapon use hospitalisations of 33 quarterly admissions, on average. However, police recorded violent offences have increased since the implementation of the CL Programme.

Columns (1. i) to (1. iv) present results for the different measures of serious violence.

On one side, column (1. i) shows an average quarterly increase of 3,650 police-recorded violent crimes in exporter areas following the CL Programme. This corresponds to a significant increase in police-recorded violent crimes by 19%^{[footnote 9]} in exporter forces after the implementation of the CL Programme. This increasing effect is primarily driven by violence with injury and violent disorders, as indicated in Table A1.1 in the Appendix. Given the research design, these effects are largely attributed directly to the Programme’s impact^{[footnote 10]}.

The direction of the effect, that is, an increase in police-recorded violent crimes, may be explained by one or more of the following hypotheses:

1. An improvement in recording, reporting and policing activity^{[footnote 11]}. The increased recorded crime might not necessarily indicate a rise in violent incidents. Instead, it could reflect improvements in reporting and recording practices. While changes in reporting are a natural effect of changes in police presence, changes in recording are likely driven by changes in guides and enforcement by His Majesty’s Inspectorate of Constabulary and Fire and Rescue Services (HMICFRS). On this matter, different enforcement practices could be taking place in exporter and importer forces, with the effects being particularly driven by changes in recording practices in the exporter forces.
2. The Programme likely resulted in a greater police presence in exporter areas, leading to more proactive policing and, consequently, a higher number of reported and recorded violent incidents. Increased patrols and operations could have exposed and intervened in more situations that previously might have gone unreported.
3. Changes in police behaviour might stem from an increased perception of security for police officers (for being in larger teams, for example, double crewing), which would lead to more enforcement actions.
4. Increases in PRC violent offences may as well be due to increased gang violence.

Such hypotheses, specifically improved recording practices at force level, require further exploration moving forward.

On the other side, Columns (1. ii) to (1.iv) indicate decreases in different measurements of serious violence, proxied by drug-related homicides, weapon use hospitalisations and lethal barrel discharges. These metrics remain our best measures of serious violence, unaffected by enforcement activity and police recording practices.

Column (1. iii) indicates that the CL Programme led to a statistically significant decrease of 33 quarterly weapon-use hospitalisations in exporter forces. This effect is of sizable magnitude (21.6%). This result is mainly led by an average decrease of 25 quarterly sharp weapons hospitalisations (or 18.9%), highlighted in Column (1. i) of Table A2.1 in the Appendix^{[footnote 12]}. The effect observed can be directly attributed to the launch of the CL Programme in the exporter forces.

Columns (1. ii) and (1. iv) indicate no statistically significant effects of the CL Programme on drug-related homicides or lethal barrel discharges. On this matter, however, the negative sign of the estimated effects is in line with what was expected, and the magnitude of the effect is also non-negligible. The lack of statistical significance of the effects is primarily due to the precision of the estimates (with large standard errors and confidence intervals) rather than from a complete null effect of the CL Programme. This suggests that there might also be a reduction of these measurements after the implementation of the CL Programme, but that the estimates are not accurate enough to find significant effects and attribute the change accordingly.

Overall, results on serious violence indicate a reduction in hospitalisations and increases in police-recorded violence. The most logical reason for harmonising both results is related to enforcing and recording crimes in exporter forces. This is indicative of improved policing, recording, and reporting activity. Even if it is not possible to formally test this hypothesis, it seems the most plausible one.

Negative changes in acquisitive crime were not statistically significant and therefore not attributable to the CL Programme, with the exception of theft from vehicles.

Regarding other outcomes, Column (2) indicates a decline in acquisitive crimes following the impact of the County Lines Programme for exporter forces, although not statistically significant at traditional significance levels (95% confidence interval) - with the exception of theft from a vehicle. Nonetheless, as with drug-related homicides and lethal barrel discharges, the magnitude of the effect is sizable, that is, an average quarterly decline of 3,990 acquisitive crimes in the exporter areas. The statistical non-significance is mostly attributable to the precision of the estimates (large standard errors) rather than to a true insignificant effect. The negative sign of the impact is as expected. Further, Table A3.1 in the Appendix disaggregates this effect by its components, and it shows negative signs for all estimated coefficients and statistical significance for thefts from vehicles (a decline of 15.4% after implementing the CL Programme).

The CL Programme has led to a quarterly increase in the number of law enforcement activities, mostly measured through an increase in police recorded drug possession offences.

Regarding Law Enforcement activities, Column (3) of Table 4 below shows that the County Lines Programme had a positive and statistically significant effect on police recorded crime such as drug possession. Concretely, the Programme led to an average quarterly increase of 1,077 activities, which represents a 25.1% increase with respect to the pre-Programme values. Further, Table A4.1 in the Appendix shows that most of the increase in this dimension came from drug possession offences, which increased by 615 cases after implementing the Programme in the exporter forces. The positive effect for trafficking is not significant at traditional confidence levels, but it is not far from it, indicating a somewhat noisy signal there could also be an effect on these law enforcement activities.

The number of NRM safeguarding referrals has increased since the start of the Programme, indicating an average increase of 641 quarterly referrals. However, these positive changes in safeguarding referrals were not statistically significant and therefore not attributable to the CL Programme.

The results on safeguarding referrals in Column (4) of Table 4 below indicate no statistically significant effect of the CL Programme. In this case, the sign indicates a positive coefficient on safeguarding referrals, as expected given CL Programme has safeguarding as one of its core functions. Moreover, the coefficient is sizable, indicating an average increase of 461 quarterly total NRM safeguarding referrals after the launch of the CL Programme in the exporter areas. However, the estimates have large standard errors, leading to no statistically significant effect.

The Programme has likely had a direct impact on reducing drug misuse hospitalisations.

Finally, column (5) shows that the CL Programme resulted in a significant decline on drug misuse hospitalisations in exporter forces. The results indicate an average decline of 40 quarterly cases of drug misuse hospitalisations in the exporter areas as a result of the Programme, which translates to a 14.4% reduction.

As expected, the CL Programme metrics indicate significant effects on the total number of lines closed and on individuals arrested and charged.

Estimates in columns (6.i) to (6.ii) show that, on average, the Programme led to the average quarterly closure of 18 County Lines in the exporter areas after its launch (specific to each force). The value of 18 is the average effect of the CL Programme by quarter and force. Once aggregated by quarter (for all forces) or by force (for all quarters), the value adds up to the total number of closures. The significant coefficients suggest that the effect on county line closures, arrests, and charges can be directly attributed to the Programme’s launch.

Table 4: Exporter forces

	(1.i) Police recorded violent crime	(1.ii) Drug related homicides	(1.iii) Weapon use hospitalization	(1.iv) Lethal barrel discharges	(2) Acquisitive crime	(3) Law enforcement	(4) Safeguarding referrals	(5) Drug misuse hospitalisations	(6.i) County line metrics closures	(6.ii) County lines metrics arrests and charges
Exporter	3,650.65*	-0.84	-33.32***	-44.59	-3,990.82	1,077.11***	41.34	-40.81**	18.51***	120.38***
	[1,804.05]	[0.86]	[7.86]	[30.40]	[2,895.58]	[246.28]	[61.42]	[18.44]	[4.28]	[9.12]
Observations	1,044	828	828	1,116	1,044	1,044	1,080	828	540	540
R-squared	0.98	0.87	0.92	0.94	0.98	0.98	0.84	0.89	0.84	0.90
Force FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Mean	19,250	8	154	157	28,072	4,286	173	283

Notes:

1. The table represents the coefficient β₁ obtained from estimating Equation 1 for exporter areas. The unit of observation is at force and quarter level. Standard errors in square brackets clustered at force level. ‘***’,‘**’,‘*’ indicate significance at 1%, 5% and 10% respectively.

Importer forces

Overall, the estimates below do not show statistically significant spillover effects of the CL Programme on importer forces across all key outcomes. These results indicate that most of the impacts on the exporter forces are localised and have relatively limited spillovers to top importers.

The estimated effects of the County Lines Programme for importer forces are presented in Table 5, following the same structure as Table 4.

Limited spillover effects were observed in the importer force areas. For example, the statistically significant effects on serious violence (both in police recorded violent crimes and weapon use hospitalizations) in exporter forces do not translate into the importer forces, suggesting localisation of the County Lines model. It is empirically not possible to isolate the impact further.

Nonetheless, the results show that the signs of the estimated effects are in line with what was expected. Concretely, even if the estimates are not statistically significant, the coefficients are negative for weapon use hospitalisations, drug use hospitalisations and, acquisitive crime (which is also of sizable magnitude), and positive for law enforcement activities. Regarding hospitalisations, Table A2.2 in the Appendix points towards a negative coefficient in hospitalisations both for sharp weapon use and firearms. Regarding acquisitive crime, Table A3.2 in the Appendix points towards the Programme leading to 9.2% reduction in thefts of vehicles. In other words, on average, there were 62 fewer vehicle thefts per quarter in the importer areas as a result of the CL Programme.

Table 5: Importer forces

	(1.i) Police recorded violent crime	(1.ii) Drug related homicides	(1.iii) Weapon use hospitalization	(1.iv) Lethal barrel discharges	(2) Acquisitive crime	(3) Law enforcement	(4) Safeguarding referrals	(5) Drug misuse hospitalization
Importer	-77.81	-0.09	-0.45	-0.82	-465.44	38.89	5.21	-4.16
	[271.75]	[0.19]	[1.87]	[2.29]	[333.52]	[67.45]	[13.06]	[4.51]
Observations	1,131	897	897	1,209	1,131	1,131	1,165	897
R-squared	0.97	0.35	0.83	0.80	0.97	0.94	0.74	0.88
Force FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Mean	8,664	2	34	20	7,696	1,034	30	134

Notes:

1. The table represents the coefficient β₂ obtained from estimating equation 2 for importer areas. The unit of observation is at force and quarter level. Standard errors in square brackets clustered at force level. ‘***’,‘**’,‘*’ indicate significance at 1%, 5% and 10% respectively.

4.2 Dynamic estimates

Figures 1 to 7 show dynamic estimates of the results shown in Tables 4 and 5 for exporter and importer forces, respectively, with respect to comparison forces^{[footnote 13]}. These graphs present a decomposition of the effects shown in the previous tables over time.

Their purpose is two-fold.

1. These figures allow us to assess the validity of the parallel trend assumption^{[footnote 14]}, which is crucial to the current research design. If the estimated effects are non-significant in periods before the policy implementation, it supports the validity of the research design. In simple words, if the point estimates (denoted by navy dots) are close to zero in the figures before the intervention, then the research design is valid.
2. The graphs show how the CL Programme’s effects materialise over time. This helps assess the speed by which the effects occur. For example, graphs show whether the CL Programme effects are immediate or if they take long to appear. Additionally, the graphs also help in understanding whether effects are stable over time, if they fade, or if they increase with time. These differences are important from a policy perspective.

Other points are worth explaining before moving into the dynamic analysis:

1. Across all graphs, the horizontal axis of indicates the time since treatment, that is, quarters before and after the treatment, with zero (vertical dashed line) being the time at which funding was received. The vertical axis measures the magnitude of the average effect in absolute values.
2. The navy dots represent the coefficient estimates from the regression exercise. If these points are close to zero (the dashed horizontal line), it means that the CL Programme affected the analysed metric in a relatively small extent.
3. The faded blue area represents the confidence intervals (that is, the upper and the lower bound) of the estimate. Tight/narrow confidence intervals indicate precision whereas wise confidence intervals indicate lower precision. If the confidence intervals contain the value of 0 (the dashed horizontal line), the estimated effects are not statistically significant at 95% levels.

4.3 Analysis

First, it is worth noting that all figures broadly show no statistically significant estimates for the pre-period analysis (the confidence interval contains the value of 0). This indicates that the parallel trend assumption holds in this setting, validating the research design, that is, exporter forces (or importer forces) are comparable to the control areas.

Second, on the matter of results, Figure 1 indicates that the significant increase in police recorded violent crimes in exporter areas is not immediate (as indicated by the point estimate close to zero at time zero in Figure 1a). Instead, it increases over time, and most of its effect occurs around one year (4 quarters later) after the Programme deployed in that specific exporter force. It remains on the increasing trend thereon. This can likely be explained by an improvement in enforcement/reporting activity over time. For importer forces, the effects are precise zeros for the post-period (as indicated by the blue dots corresponding to zero). This means that there were limited spillover effects in the importer areas from the CL funding received in the exporter areas.

Figure 1: Effect of the CL Programme on police recorded violent crime - exporter and importer forces

Figure (1. i): Exporter forces

Figure (1. ii): Importer forces

Notes:

1. Effect of county line funding on police recorded violent crime. The x axis represents the quarters before and after the launch of CL funding Programme and the y axis represents the average change in crime (in levels) in the treated areas as compared to the control areas over time. Panel (1. i) and (1. ii) represent variation for exporters and importers, respectively, using equation 1 and equation 2. The blue dots represent the point estimates, and the bounded area represents the confidence interval at 95%.

Figure 2 shows that for drug-related homicides, the estimated effects are also close to zero in the pre and post-period for both exporter and importer forces. Moreover, the confidence intervals contain the value of 0. Hence, the results need to be interpreted with caution. However, the direction of the effect is negative (below the horizontal dashed line at zero), as shown in the regression tables.

Figure 2: Effect of the CL Programme on drug-related homicides – exporter and importer forces

Figure (2. i): Exporter forces

Figure (2. ii): Importer forces

Notes:

1. Effect of county line funding on drug-related homicide. The x axis represents the quarters before and after the launch of CL funding Programme and the y axis represents the average change in crime (in levels) in the treated areas as compared to the control areas over time. Panel (2. i) and (2. ii) represent variation for exporters and importers, respectively, using equation 1 and equation 2. The blue dots represent the point estimates, and the bounded area represents the confidence interval at 95%.

Figure 3 shows that weapon use hospitalisations reflect different exporter and importer forces patterns. The effects for exporter forces are more precise, and the reduction in weapon use hospitalisations is immediate after the start of the Programme as indicated by the downward trend in the point estimates since treatment. The importer areas’ graph indicate null and less precise results.

Figure 3: Effect of the CL Programme on weapon use hospitalisations – exporter and importer forces

Figure (3. i): Exporter forces

Figure (3. ii): Importer forces

Notes:

1. Effect of county line funding on weapon use hospitalisations (sharp weapons and firearms). The x axis represents the quarters before and after the launch of CL funding Programme and the y axis represents the average change in crime (in levels) in the treated areas as compared to the control areas over time. Panels (3 .i) and (3. ii) represent variations for exporters and importers, respectively, using Equation 1 and Equation 2. The blue dots represent the point estimates, and the bounded area represents the confidence interval at 95%.

Figure 4 also indicates that the effect on lethal barrel discharges is immediate for exporter forces (as indicated by point estimates below the horizontal dashed line at zero immediately after period 0), and also growing over time. In the case of importer forces, estimates are consistently close to zero.

Figure 4: Effect of the CL Programme on lethal barrel discharges – exporter and importer forces

Figure (4. i): Exporter forces

Figure (4. ii): Importer forces

Notes:

1. Effect of county line funding on lethal barrel discharges. The x axis represents the quarters before and after the launch of CL funding Programme and the y axis represents the average change in crime (in levels) in the treated areas as compared to the control areas over time. Panel (4. i) and (4. ii) represent variation for exporters and importers, respectively, using equation 1 and equation 2. The blue dots represent the point estimates, and the bounded area represents the confidence interval at 95%.

Figure 5 shows the estimated effects of acquisitive crimes. In the case of exporters, the effects also occur quickly, immediately after the policy implementation (observed through the immediate jump downward of the point estimate). In the case of importers, a very similar pattern arises to the exporter forces. However, there is less precision than in the case of exporter forces. In both cases, the most considerable impacts appear in the first quarters after the policy implementation, and they seem to fade closer to zero for post-periods further away from policy implementation.

Figure 5: Effect of the CL Programme on acquisitive crime – exporter and importer forces

Figure (5. i): Exporter forces

Figure (5. ii): Importer forces

Notes:

1. Effect of county line funding on acquisitive crime. The x axis represents the quarters before and after the launch of CL funding Programme and the y axis represents the average change in crime (in levels) in the treated areas as compared to the control areas over time. Panels (5. i) and (5. ii) represent variations for exporters and importers, respectively, using equation 1 and equation 2. The blue dots represent the point estimates, and the bounded area represents the confidence interval at 95%.

Figure 6 shows results for law enforcement activities. In both cases, positive effects for exporter and importer forces are concentrated at a peak in quarter one after policy implementation. In contrast, the impact for subsequent quarters is much smaller and closer to zero.

Figure 6: Effect of the CL Programme on law enforcement – exporter and importer forces

Figure (6. i): Exporter forces

Figure (6. ii): Importer forces

Notes:

1. Effect of county line funding on law enforcement. The x axis represents the quarters before and after the launch of CL funding Programme and the y axis represents the average change in crime (in levels) in the treated areas as compared to the control areas over time. Panel (6. i) and (6. ii) represent variation for exporters and importers, respectively, using equation 1 and equation 2. The blue dots represent the point estimates, and the bounded area represents the confidence interval at 95%.

Figure 7 indicates that the effect on all NRM safeguarding referrals is cumulative for exporter forces(7.i), as the effect increases over time after the policy implementation. For importers(7.ii), even if the post-policy coefficients are also positive and statistically significant (different than the horizontal dashed line), the pre-trend analysis is not as clean as for other outcomes^{[footnote 15]}, making this specific result less reliable.

Moreover, there is no clear effect of the CL Programme on CL safeguarding referrals specifically until 18 months into programme implementation. This is indicated by the point estimates close to zero in Figure 7.iii, for the first 5 coefficients after programme implementation. Around 6 quarters post implementation, there is a sharp increase in CL safeguarding referrals. This later increase in CL safeguarding referrals is “diluted” into the average effect, as for all previous quarters the effect is very close to zero. These dynamics and effects could be due to data coverage, but it is not possible to fully assess the existing data. Further analysis of this dimension could be crucial in the future.

Figure 7: Effect of the CL Programme on safeguarding referrals – exporter and importer forces

Figure (7. i): Exporter forces

Figure (7. ii): Importer forces

Figure (7. iii): CL safeguarding referrals – Exporter forces

Notes: Effect of county line funding on safeguarding referrals. The x axis represents the quarters before and after the launch of CL funding Programme and the y axis represents the average change in the number of referrals in the treated areas as compared to the control areas over time. Panels (7. i) and (7. ii) represent variations for exporters and importers in the total number of NRM referrals, respectively, using equation 1 and equation 2. Panel (7.iii) shows the trend in CL Programme safeguarding referrals over the period. The blue dots represent the point estimates, and the bounded area represents the confidence interval at 95%.

Figure 8 shows the effect of drug misuse hospitalisations. For exporters, the CL Programme has an immediate and persistent negative effect, as observed by the declining trend of the point estimates (below the horizontal dashed line at zero). Moreover, this group’s estimates are precise (narrow confidence intervals). In the case of importer areas, the results are not statistically significant and relatively flat in time.

Figure 8: Effect of the CL Programme on drug misuse hospitalisations – exporter and importer forces

Figure (8. i): Exporter forces

Figure (8. ii): Importer forces

Notes: Effect of county line funding on drug substance misuse hospitalisations. The x axis represents the quarters before and after the launch of CL funding Programme and the y axis represents the average change in crime (in levels) in the treated areas as compared to the control areas over time. Panels (8. i) and (8. ii) represent variations for exporters and importers, respectively, using equation 1 and equation 2. The blue dots represent the point estimates, and the bounded area represents the confidence interval at 95%.

Figure 9 portrays the estimated effects of the County Lines Programme on its metrics. While the impact on closures seems to increase over time, in the case of arrest and charges, the effects dilute around quarter six post-policy (that is the growth rate decreases over time).

Figure 9: Effect of the CL Programme on county line metrics – exporter forces

Figure (9. i): Closures

Figure (9. ii): Arrest and charges

Notes:

1. Effect of county line funding on county line metrics. The x axis represents the quarters before and after the launch of CL funding Programme and the y axis represents the average change in crime (in levels) in the treated areas as compared to the control areas over time. Panels (9. i) and (9. ii) represent variation for exporters using equation 1. The blue dots represent the point estimates, and the bounded area represents the confidence interval at 95%.

5. Concluding remarks

The evaluation of the County Lines Programme reveals a multifaceted impact on various aspects of crime, highlighting both successes and challenges in addressing drug trafficking and associated criminal networks.

The Programme has shown efficacy in reducing serious violence, particularly in exporter areas, evidenced by a reduction in hospitalisations due to weapon use, especially sharp weapons. The observed decrease in serious violence due to weapon use suggests a very positive outcome attributable to the County Lines Programme. The simultaneous increase in police-recorded violence may warrant further investigation into potential shifts in reporting practices and law enforcement priorities. However, nuanced considerations emerge when examining spillover effects on importer forces and changes in police-recorded crimes. The limited spillover effects observed in areas with importer forces underscore the localised nature of County Lines operations.

These findings emphasize the importance of employing rigorous quantitative methodologies, such as the difference-in-differences strategy used in this analysis, to evaluate the effectiveness of crime prevention initiatives comprehensively. This methodology estimates the causal impact of the Programme without needing to create an exact counterfactual, as required in randomized control trials or matching, provided the assumption of parallel trends holds.

In terms of future research for the CL Programme and its evaluation, the following points are key to explore:

1. The role of changes in reporting practices across forces should be analysed. There have been differential rises in the overall police-recorded violence in different forces depending on their approach to crime recording and/or their baseline recording levels. Overall, this finding needs further exploration in CL context to understand whether alongside improvements in recording practices and law enforcement priorities, there has been a genuine increase in violence as a result of disrupting CL gangs/OCGs.
2. Data gathering across all CL Programme metrics and other impact variables must continue. As many estimates on this assessment lack certain precision (or put differently, have wide confidence intervals), a longer timeframe and more data could alleviate this concern.
3. When possible, data quality should be monitored and validated.

As we move forward, it is imperative to continue refining strategies to combat county lines activities. This involves leveraging insights from rigorous evaluation methodologies and incorporating a holistic approach that addresses law enforcement efforts and the underlying social and economic factors driving criminal behaviour.

6. Glossary

• CL - County lines
• DEAA - Drugs, Safeguarding and Abuse Analysis unit, Home Office
• FY - Financial Year
• GRIP- (formerly Surge) hotspot policing Programme
• LSOA - Lower Super Output Area
• NHS - National Health Service
• NRM- National Referral Mechanism
• OCGM - Organised Crime Group Mapping
• ONS - Office for National Statistics
• PRC - Police Recorded Crime
• VRU - Violence Reduction Unit
• UKDA - UK Data Archive

Appendix 1: Additional tables

Table A1.1: Disaggregated effect on police recorded violent crime for exporter forces

	(1) Police recorded violent crime	(1.i) Violence without injury	(1.ii) Violence with injury	(1.iii) Violent disorders
Exporter	3,650.645*	474.52	3,149.61**	26.51*
	[1,804.05]	[563.83]	[1,271.91]	[14.20]
Observations	1,044	1,044	1,044	1,044
R-squared	0.98	0.99	0.97	0.64
Force FE	Yes	Yes	Yes	Yes
Time FE	Yes	Yes	Yes	Yes
Mean	19,250	8,856	10,370	23

Notes:

1. The table represents the coefficient obtained from estimating equation 1 for exporter areas. The unit of observation is at force and quarter level. Standard errors in square brackets clustered at force level. ‘***’,‘**’,‘*’ indicate significance at 1%, 5% and 10% respectively.

Table A1.2: Disaggregated effect on police recorded violent crimes for importer areas

	(1) Police recorded violent crime	(1.i) Violence without injury	(1.ii) Violence with injury	(1.iii) Violent disorders
Importer	-77.81	-123.47	46.10	-0.43
	[271.75]	[133.21]	[246.36]	[0.86]
Observations	1,131	1,131	1,131	1,131
R-squared	0.97	0.97	0.96	0.32
Force FE	Yes	Yes	Yes	Yes
Time FE	Yes	Yes	Yes	Yes
Mean	8,664	3,720	4,940	4

Notes:

1. The table represents the coefficient obtained from estimating equation 2 for importer areas. The unit of observation is at force and quarter level. Standard errors in square brackets clustered at force level. ‘***’,‘**’,‘*’ indicate significance at 1%, 5% and 10% respectively

Table A2.1: Disaggregated effect on hospitalisations for exporter areas

	(1) Weapon use hospitalization	(1. i) Sharp weapon	(1. ii) Firearms
Importer	-33.32***	-25.56***	-4.46
	[7.86]	[4.63]	[3.24]
Observations	828	828	551
R-squared	0.92	0.92	0.91
Force FE	Yes	Yes	Yes
Time FE	Yes	Yes	Yes
Mean	154	135	21

Notes:

1. The table represents the coefficient obtained from estimating equation 1 for exporter areas. The unit of observation is at force and quarter level. Standard errors in square brackets clustered at force level. ‘***’,‘**’,‘*’ indicate significance at 1%, 5% and 10% respectively.

Table A2.2: Disaggregated effect on hospitalisations for importer areas

	(1) Weapon use hospitalization	(2) Sharp weapon	(3) Firearms
Importer	-0.45	0.34	-0.75
	[1.87]	[1.64]	[0.70]
Observations	897	897	608
R-squared	0.83	0.82	0.42
Force FE	Yes	Yes	Yes
Time FE	Yes	Yes	Yes
Mean	34	31	4

Notes:

1. The table represents the coefficient obtained from estimating equation 2 for importer areas. The unit of observation is at force and quarter level. Standard errors in square brackets clustered at force level. ‘***’,‘**’,‘*’ indicate significance at 1%, 5% and 10% respectively.

Table A3.1: Disaggregated effect on acquisitive crime for exporter forces

	(1) Acquisitive crime	(1.i) Domestic burglary	(1.ii) Theft from person	(1.iii) Robbery of personal property	(1.iv) Theft from a shop	(1.v) Theft of vehicle	(1.vi) Theft from a vehicle	(1.vii) Vehicle interference
Exporter	-3,990.82	-1,030.29	-792.54	-571.16	-396.31	-79.77	-985.41**	-135.35
	[2,895.58]	[733.64]	[610.80]	[541.15]	[368.99]	[199.96]	[457.00]	[139.32]
Observations	1,044	1,044	1,044	1,044	1,044	1,044	1,044	1,044
R-squared	0.98	0.97	0.93	0.96	0.97	0.99	0.98	0.97
Force FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Mean	-28,072	6,077	3,301	2,517	5,377	3,004	6,377	1,418

Notes:

1. The table represents the coefficient obtained from estimating equation 1 for exporter areas. The unit of observation is at force and quarter level. Standard errors in square brackets clustered at force level. ‘***’,‘**’,‘*’ indicate significance at 1%, 5% and 10% respectively.

Table A3.2: Disaggregated effect on acquisitive crime for importer forces

	(1) Acquisitive crime	(1.i) Domestic burglary	(1.ii) Theft from person	(1.iii) Robbery of personal property	(1.iv) Theft from a shop	(1.v) Theft of vehicle	(1.vi) Theft from a vehicle	(1.vii) Vehicle interference
Importer	-465.44	-129.77	-29.66	-16.31	-138.26	-62.95**	-74.81	-13.69
	[333.52]	[124.26]	[28.39]	[16.57]	[130.63]	[29.67]	[107.23]	[31.48]
Observations	1,131	1,131	1,131	1,131	1,131	1,131	1,131	1,131
R-squared	0.97	0.94	0.91	0.94	0.93	0.95	0.94	0.89
Force FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Mean	7,696	1,775	369	276	2646	683	1,567	380

Notes:

1. The table represents the coefficient obtained from estimating equation 2 for importer areas. The unit of observation is at force and quarter level. Standard errors in square brackets clustered at force level. ‘***’,‘**’,‘*’ indicate significance at 1%, 5% and 10% respectively

Table A4.1: Disaggregated effect on law enforcement for exporter forces

	(1) Law enforcement	(1.i) Possession of drugs	(1.ii) Trafficking of drugs	(1.iii) Possession of weapon
Exporter	1,077.11***	615.94**	222.50	238.68
	[246.28]	[236.51]	[158.84]	[190.12]
Observations	1,044	1,044	1,044	1,044
R-squared	0.98	0.97	0.86	0.92
Force FE	Yes	Yes	Yes	Yes
Time FE	Yes	Yes	Yes	Yes
Mean	4,286	2,906	522	858

Notes:

1. The table represents the coefficient obtained from estimating equation 2 for importer areas. The unit of observation is at force and quarter level. Standard errors in square brackets clustered at force level. ‘***’,‘**’,‘*’ indicate significance at 1%, 5% and 10% respectively.

Table A4.2: Disaggregated effect on law enforcement for importer forces

	(1) Law enforcement	(1.i) Possession of drugs	(1.ii) Trafficking of drugs	(1.iii) Possession of weapon
Importer	38.89	29.08	15.69	-5.89
	[67.45]	[54.20]	[27.49]	[28.49]
Observations	1,131	1,131	1,131	1,131
R-squared	0.94	0.91	0.84	0.92
Force FE	Yes	Yes	Yes	Yes
Time FE	Yes	Yes	Yes	Yes
Mean	1,034	566	205	262

Notes:

1. The table represents the coefficient obtained from estimating equation 2 for importer areas. The unit of observation is at force and quarter level. Standard errors in square brackets clustered at force level. ‘***’,‘**’,‘*’ indicate significance at 1%, 5% and 10% respectively.

Table A5.1: Disaggregated effect on safeguarding referrals for exporter forces

	(1) Safeguarding (NRM referrals total)	(1. i) NRM referrals with a county lines flag
Exporter	41.34	11.95
	[61.42]	[10.06]
Observations	1,080	1,080
R-squared	0.88	0.81
Force FE	Yes	Yes
Time FE	Yes	Yes
Mean	173	13

Notes:

1. The table represents the coefficient obtained from estimating equation 1 for exporter areas. The unit of observation is at force and quarter level. Standard errors in square brackets clustered at force level. ‘***’,‘**’,‘*’ indicate significance at 1%, 5% and 10% respectively.

Table A5.2: Disaggregated effect on safeguarding referrals for importer forces

	(1) Safeguarding (NRM referrals total)	(1. i) NRM referrals with a county lines flag
Importer	5.21	-0.21
	[13.06]	[1.67]
Observations	1,165	1,165
R-squared	0.74	0.70
Force FE	Yes	Yes
Time FE	Yes	Yes
Mean	28	3

Notes:

1. The table represents the coefficient obtained from estimating equation 2 for importer areas. The unit of observation is at force and quarter level. Standard errors in square brackets clustered at force level. ‘***’,‘**’,‘*’ indicate significance at 1%, 5% and 10% respectively.

Appendix 2: Estimates relative to the counterfactual

Tables A7.1 and A7.2 in this appendix present the estimated effects of the CL programme on all outcome measures for both exporters and importers, as outlined in the main text. In order to facilitate the translation of these effects into percentage changes, 2 reference values are provided. One reference value is that of the variable in question in the corresponding treatment group (exporter/importer) prior to the implementation of the Programme. This is presented in row 6, and the corresponding effect of the CL Programme as a percentage change with respect to this reference value is presented in row 7. Another reference value is that of the variable in the comparison group. This is presented in row 8, and the corresponding effect of the CL Programme as a percentage change with respect to this reference value is presented in row 9.

Table A7.1 Exporter forces

	(1.i) Police recorded violent crime	(1.ii) Drug related homicides	(1.iii) Weapon use hospitalization	(1.iv) Lethal barrel discharges	(2) Acquisitive crime	(3) Law enforcement	(4) Safeguarding referrals	(5) Drug use hospitalization
Exporter	3,650.65*	-0.84	-33.32***	-44.59	-3,990.82	1,077.11***	41.34	-40.81**
	[1,804.052]	[0.86]	[7.86]	[30.40]	[2,895.58]	[246.28]	[61.42]	[18.44]
Observations	1,044	828	828	1,116	1,044	1,044	1,080	828
Mean exporter	19,250	8	154	157	28,072	4,286	173	283
% Change	19	-10	-22	-28	-14	25	24	-14
Mean control	5,752	1	27	19	5,558	999	41	75
% Change	64	-60	-124	-235	-72	108	101	-54

Table A7.2 Importer forces

	(1.i) Police recorded violent crime	(1.ii) Drug related homicides	(1.iii) Weapon use hospitalization	(1.iv) Lethal barrel discharges	(2) Acquisitive crime	(3) Law enforcement	(4) Safeguarding referrals	(5) Drug use hospitalization
Importer	-77.81	-0.09	-0.45	-0.82	-465.44	38.89	5.21	-4.16
	[271.75]	[0.19]	[1.87]	[2.29]	[333.52]	[67.45]	[13.06]	[4.51]
Observations	1,131	897	897	1,209	1,131	1,131	1,165	897
Mean importer	8,664	2	34	20	7,696	1,034	30	134
% Change	-1	-5	-1	-4	-6	4	17	-0.1
Mean control	5,146	1	21	10	4,272	783	28	68
% Change	-2	-9	-2	-8	-11	5	19	-6

1. Throughout the report, we will define weapon hospitalisations as the sum of sharp weapons and firearm hospitalisations. This is done to have a broad assessment of County Lines. ↩

2. Throughout the report, results presented in percentage change are to be interpreted as the change in exporter/importer areas after the implementation of the CL Programme with respect to the situation in the same forces before the implementation of the CL Programme. ↩

3. The nature of violent crime in England and Wales - Office for National Statistics (ons.gov.uk) ↩

4. Theft from a shop was added due to the evidence linking it with drug use, even though it isn’t part of the general HO definition of acquisitive crime. ↩

5. To be specific, these include opium, heroin, opioids, methadone, synthetic narcotics, cocaine, unspecified narcotics, cannabis, LSD, unspecified hallucinogens and psychostimulant. ↩

6. OCGM is the only available law enforcement dataset which provides an index of law enforcement intelligence on organised crime groups (OCGs) within the UK. There are a number of significant data quality issues around OCGM data which means that any findings within this section should be treated with caution. The data quality issues associated with OCGM greatly inhibit the ability to generate an accurate picture of the known serious and organised crime (SOC) threat to the UK, and to make assessments around its scale. As the issues are the result of multiple differing approaches to recording data on OCGM, a cross-system response is required to ensure the data on OCGM adds value at a strategic level. ↩

7. The results of this table and the subsequent ones are robust to the inclusion of COVID-19 incidence per PFA and quarter as a control variable. ↩

8. Greater Manchester Police is considered as an exporter area from 2022 onward, later than the other forces. ↩

9. All percentage changes have been reported in comparison to pre-intervention means. ↩

10. In terms of police recorded violent crimes, the research design is not able to account for potential changes in recording practices, both across time and forces. Qualitative evidence indicates there have been differential rises in different forces, dependent on their approach to crime recording and/or their baseline recording levels. ↩

11. The nature of violent crime in England and Wales - Office for National Statistics (ons.gov.uk) ↩

12. There is a significant decline in sharp weapon hospitalizations for under 25s. Specifically, there are 3.31 less sharp weapon hospitalizations for under 18 (16% reduction) and 11.79 less sharp weapon hospitalisations for ages between 18 and 24 (28% reduction) after the launch of the County Lines Programme in the exporter areas. These results are statistically significant which means the decline can be directly attributed to the launch of the CL Programme. ↩

13. In these graphs, the quarter right before the start of the County Lines Programme is taken as the reference point. ↩

14. The validity of this assumption ensures that the treatment group and the comparison group follow similar trends before the policy’s implementation and are thus comparable. ↩

15. The effects in the pre-period are statistically significant. ↩