In the prisoner interviews, traffickers told us how they assess risks and adjust trafficking activity to ameliorate those risks. Self-reports are subjective, of course, and prisoners may not be entirely truthful. Therefore we also used statistical analysis to objectively estimate how traffickers responded to the threat posed by interdiction. Specifically, we identified special counterdrug enforcement operations and events in the source, transit and arrival zones, and we sought to learn whether those programs had a predictable effect on two measures of drug trafficking:

• Illicit drug prices in the U.S., at both the wholesale and retail levels
• Displacement of drug trafficking activity from one transportation route and method to another

The analysis required several steps. The first step was to identify factors that might affect trafficking behavior. Most important for our purposes, those factors were counterdrug activities that "turned on" and then "turned off" during defined periods. We sought to identify the correlation between our measures of drug trafficking and such periods of active enforcement. We examined a total of 21 counterdrug enforcement operations and events for their effect on cocaine trafficking activity. These deterrent events span the time period of 1991 to 1999 and represent:

• Interdiction operations in source, transit and arrival Zones
• Major adjustments to interdiction tactics (e.g., the implementation of a shoot-down/forcedown policy in Peru and Colombia)
• Investigative operations (e.g., Zorro II); and
• The arrest or death of major drug traffickers

The deterrent events considered in this study (listed in Table 1) were identified through a review of public and limited access materials. The proposed list of deterrent events (with operation start and end dates, affected geographic areas, and affected transportation modes) was critiqued by points of contact at Customs, Coast Guard and ONDCP.

Table 1 Deterrent Events
Event Name	Dates	Description
Arrival & Transit Zone Enforcement Operations:
Hard Line Zorro II Gateway White Shark I White Shark II Frontier Shield Border Shield Gulf Shield White Shark III Brass Ring River Sweep Frontier Lance Conjuntos I Conjuntos II Two Dozen	Feb 1995–Jul 1997 Dec 1995–May 1996 Mar 1996–Feb 1998 Mar 1996–Mar 1997 Sep 1996–Oct 1996 Oct 1996–Dec 1996 Mar 1997–Ongoing Mar 1997–Ongoing Jul 1997–Sep 1997 Feb 1998–Jul 1998 Feb 1998–Ongoing Mar 1998–May 1998 Jun 1998 Dec 1998 Feb 1999	Interdiction operation targeting smuggling across the SWB Investigative operation targeting organizations moving drugs across SWB Interdiction operation targeting all modes of smuggling through Puerto Rico Interdiction operation targeting vessels from Mexico to Gulf coast of Texas Interdiction operation targeting vessels from Mexico to Gulf coast of Texas Interdiction operation targeting vessels to Puerto Rico and Eastern Caribbean Interdiction operation targeting vessels from Mexico to S. California Interdiction operation targeting vessels from Mexico to Gulf coast of Texas Interdiction operation targeting vessels from Mexico to Gulf coast of Texas Interdiction operation targeting smuggling through SWB ports of entry Customs operation targeting Haitian freighters in Miami River Interdiction operation targeting vessels to Jamaica and Hispanola Land & sea interdiction operation off Panama's Caribbean coast Land & sea interdiction operation off Panama's Caribbean coast Interdiction operation targeting aircraft and vessels into the Bahamas and Florida
Source Zone Enforcement Operations:
Support Justice III Support Justice IV Stand-down Shoot-down	Nov 1991–Apr 1992 Jan 1993–Apr 1994 May 1994–Dec 1994 Mar 1995–Nov 1995	Early operation to suppress movement of coca products in South America Follow-on operation to SJ III U.S. ceases radar support in South America pending legal review of Peru's and Colombia's force-down policy U.S. resumes radar support to Peru and Colombia. Peru earnestly implements force down policy
Trafficker Arrests/Deaths:
Arrests/Deaths Orejuela Brothers	Various Jun 1995–Aug 1995	Various arrests or deaths of major Colombian and Mexican traffickers Arrest of the Orejuela brothers

More than government-sponsored enforcement efforts affect trafficking decisions and illicit drug prices. For example, the attractiveness of certain vectors and conveyance mode combinations may vary with the season. Our analyses introduced "control" variables to avoid confusing the effects of enforcement operations with other factors, such as weather, seasonality, and so on. Specifically, control factors considered include:

• Major weather events (e.g., hurricanes, tropical storms, and severe El Nino effects). These are detailed in Appendix F.
• Seasonality
• Month-to-month correlation of price data

Deterrent event data were complemented with annual source and transit zone interdiction force laydown data from Customs and Coast Guard. We were unable to obtain this type of information from the Department of Defense, therefore, we used DoD annual interdiction funding levels as published in the annual National Drug Control Strategies as an indicator of DoD interdiction force laydown trends.

The approach used to evaluate the effect of specific deterrent events varied depending upon the nature of the event. For example, since source zone operations affect all transit and arrival zone routes and transportation modes, it was not possible to consider them in the displacement analysis (which evaluates whether interdiction operations displace drug trafficking activity from one transportation route/mode to another).

Effect of Deterrent Events on Cocaine Prices

The objective of this analytic approach is to identify the factors that affect cocaine prices in the U.S. We are specifically interested in the impact of various types of counterdrug enforcement operations, but as explained earlier, we control for a variety of other non enforcement-related factors. We used transfer function models to conduct our analysis (details are provided in Appendix E). Put simply, the full price effect of successful enforcement operations/events may unfold over months as markets adjust to higher costs and shortages. Rather than assuming that successful enforcement is instantaneous, a transfer function allows for more realistic incremental adjustments.

Approximating Cocaine Prices

We examined the effect of deterrent events on four levels of cocaine prices—national retail, national wholesale, southwest border retail and southwest border wholesale. The consideration of multiple levels of prices allowed us to evaluate the possibility that different forms of enforcement operations will affect different categories of prices. These price series are presented in Figures 2 and 3.

It is important to recognize that estimating cocaine prices is not a matter of simple algebra. First, the data that supports analyses of cocaine prices (STRIDE) is not collected randomly. As such, there is a great deal of bias in the raw data. How data analysts deal with this bias, if they do at all, has a large impact on the resulting estimates. Secondly, the STRIDE data are wrought with grossly outlying values that may be data entry errors (e.g., $1 million for 1 pure gram of cocaine). How analysts deal with such errors also affects the resulting estimates.

Abt Associates has worked with the STRIDE data for over a decade, and we have had the benefit of improving our methods over these years. The price-series used here is an adaptation of the price-series recently provided to ONDCP. The basic methodology is reported in Johnston, Rhodes and Kling (2001). That report provides a quarterly series. The adaptation used here provides a monthly series. The southwest border prices do not appear in the price-series reports, but are based on applying a similar methodology to cocaine purchases from Texas, Arizona, New Mexico and California.

Consistent with data about the purchasing habits of hardcore drug users, the retail price series is based on purchases of $100 and less. Purchases of around $40 dominate this series. The wholesale price is a construct consisting of purchases of over 100 pure grams, which averages to 417 pure grams at 82% purity.

Methodology

We began our analysis by creating a baseline model for use in all price series evaluations. This baseline model includes control variables representing month-to-month correlation in cocaine price data, seasonality, weather⁹, and source and transit zone air and marine interdiction force laydown trends.

We then commenced the process of introducing the individual deterrent events into this baseline model. Recognizing that there would be a delay in the effect of different types of operations/actions on cocaine prices in the U.S., model selection involved choosing both a deterrent event as well as its optimal (from a statistical standpoint) delay. Delay ranges considered were four to six months for the five source zone events, and zero to two months for all other deterrent events. Deterrent events meeting standard statistical criteria of significance were allowed into the model.

The results for each of the four price series are presented in Table 2. Specifically, for each price level, this table identifies whether a deterrent event had a statistically significant effect on cocaine prices in the U.S. If so, the table indicates whether that effect was to increase prices (as indicated by a 'positive' entry) or to decrease prices (as would be indicated by a 'negative' entry). Deterrent events found to have no statistically significant effect are indicated with an entry of 'none'.

Table 2 Effect of Deterrent Events on Cocaine Prices in the U.S. (delay of effect, in months, is in parenthesis)
	National Retail Prices	National Wholesale Prices	SWB Retail Prices	SWB Wholesale Prices
Arrival & Transit Zone Events
Hard Line Two Dozen Zorro II Gateway Frontier Shield Border/Gulf Shield** Brass Ring River Sweep Frontier Lance Conjuntos I Conjuntos II	None None None None None Positive (1) None None None None None	None None None None None Positive (2) None None None None None	Positive (0) None None None None None None None None None Positive (1)	None None None None None None None None None None None
Source Zone Events
Support Justice III Support Justice IV Stand-down Shoot-down	Positive (4) None None Positive (6)	Positive (5) None None Positive (6)	Positive (4) None None None	Positive (5) Positive (5) None None
Trafficker Arrests/Deaths
Arrests/Deaths*** Orejuela Brothers Arrest	None Positive (1)	None Positive (0)	Positive (1) None	None Positive (0)
** Due to the high collinearity between Gulf Shield and Border Shield, the two operations are represented in our model by one combined variable. *** The Arrests/Deaths variable for SWB prices includes only arrests or deaths of major Mexican traffickers

Effects on National Wholesale Prices
Four deterrent events were found to have statistically significant effects on national wholesale prices of cocaine: Operation Support Justice III, Shoot Down, the arrest of the Orejuela brothers, and Operation Border/Gulf Shield. The effect of the two source zone interdictions, Support Justice III and Shoot Down was to increase wholesale prices by $3.53 per pure gram five to six months after initiation of the operation. The effect of Orejuela arrests was to increase wholesale prices by $4.71 in the month of the arrest, and the effect of long-active Border/Gulf Shield was to increase wholesale prices by $4.52 with a delay of two months.

These effects are shown in the accompanying graphs (Figures 4 and 5). Figure 4 shows the effect of the deterrent events only, while Figure 5 shows the final model (which takes into account the effect of monthly correlation, time, and weather events). The continuous line represents wholesale prices. The dashed line represents wholesale prices predicted from the model. The timing of deterrent events is shown at the bottom of the graphs. The lower sequence shows the actual timing of the individual deterrent events, while the upper sequence shifts these by the optimal delay estimated in the model. For example, Support Justice IV actually started in January 1993, but its effect began five months later in May 1993. Thus the interdiction response in the predicted (dashed) line is contemporaneous with Support Justice IV shifted by five months. Also, note that although Shoot Down actually preceded Orejuela arrests by three months, the effect of the latter was realized first. Indeed, wholesale prices were kept elevated for 12 consecutive months, first by Orejuela arrests (three months) and then by Shoot Down (nine months).

We identified a positive month-to-month correlation in the data (suggesting a momentum of prices) for both national wholesale and national retail prices (discussed below), but no seasonal effect could be detected in the final models. In addition, both wholesale and retail prices were affected by weather. Thus wholesale prices in months with hurricanes every day (e.g., April, May and June 1997) are expected to be $4 higher than months with no hurricanes (e.g., the four months preceding April 1997). Similarly, retail prices are expected to be $17 higher in hurricane-saturated months than in hurricane-free months. The effects of weather and month-to-month correlation are evident from the figures depicting the final models for national wholesale and national retail prices. In particular, high prices in the spring of 1997 appear to be largely attributable to the high frequency of hurricanes during that period.

National Retail Prices
The same deterrent events found to be significant in predicting national wholesale prices were similarly found significant in predicting national retail prices. The effect of the two source zone operations, Support Justice III and Shoot Down was to increase wholesale prices four and six months later by $30 per pure gram and $26 respectively. The effect of the Orejuela arrests was to increase wholesale prices by $33, and the effect of long active Border/Gulf Shield was to increase wholesale prices by $30. In both latter cases, the delay was by one month.

These effects are depicted in Figures 6 and 7. For wholesale prices, the effects of Shoot Down and Orejuela arrests were strictly adjacent, giving rise to two plateaus of elevated prices lasting 12 months. For retail prices, in contrast, the two interdictions were additive for the month of September 1995, and this combined effect appears to largely account for the unusually large spike in retail prices that month. The final model predicts a still higher price because of the high hurricane percentage in August (48% or 15 days) and September (32% or 10 days), the effect of hurricanes in August on prices in September being explained by the positive month-to-month correlation.

Southwest Border Wholesale Prices
The model indicates that wholesale prices at the South West Border were influenced by the three interdictions with the arrest of the Orejuela brothers having a dynamic effect¹⁰ with no delay, and the two source zone interdictions having non-dynamic effects each with a five month delay.

The effect of Support Justice III and Support Justice IV was to increase wholesale prices five months later by $5.47 per pure gram and $3.14 respectively. The effect of the Orejuela arrests was to increase wholesale prices by $3.84 in the month of the arrest, $2.66 the next month, and $1.86 the following month. Since the last Orejuela arrest was in August 1995, prices began an exponential decline towards their original level starting in September 1995. These effects are shown in Figures 8 and 9.

South West Border Retail Prices
The arrest or death of major Mexican drug traffickers (January 1996, February 1997, and July 1997) and Operation Conjuntos II (December 1998) were found to increase South West Border retail prices one month later by $68 and $95 respectively. These effects are represented in Figures 10 and 11 by spikes in the model at the appropriate months, and in three of the four cases, by coincident spikes in the data.

The effect of Support Justice III was to increase wholesale prices by $59 four months after initiation of the operation, $34 in the next month, and $19 in the following month, approaching a total increase of $137. However, what is more evident from the figure is the subsequent exponential decline starting in September 1992.

Operation Hard Line had a similar dynamic effect while it was in full operation. The initial increase in prices was about $6 approaching a total increase of $46, half of which would be achieved by the fifth month. This effect was attenuated as Hard Line ramped up (February 1995 through December 1995) and ramped down (September 1996 through July 1997). These effects induced an S-shape to the modeled curve from February 1995 through July 1997.

Effect of Deterrent Events on Cocaine Movement

We argued above that successful enforcement efforts should increase cocaine prices, at least until cocaine traffickers adapt. Demonstrating that prices are temporarily sensitive to interdiction events is important, but that demonstration says nothing about how traffickers actual adapt to additional anti-drug operations. This section provides some detail about those adaptations. The objective of this part of the analysis is to determine whether or not traffickers respond to interdiction operations by altering how and where they move drugs through the transit zone and across U.S. borders. Specifically, when a special interdiction operation is focused on a specific vector and transportation mode for a defined period, we would expect two adjustments:

• Traffickers would reduce shipments through the targeted vector and by the targeted mode for the period of operation; and
• They would increase shipments through other vectors and modes for the same period of operation.

Characterizing Trafficker Behavior

Data from the interagency Consolidated Counterdrug Database (CCDB) was used as the indicator of trafficker behavior in the transit zone. This database contains drug smuggling event data dating back to 1991 and supports two primary types of analyses: (1) evaluations of drug flow from areas of production to consumption countries, and (2) assessments of the performance of counterdrug forces against that movement¹¹. For purposes of this study, we used those events, called known events, within the CCDB that support counterdrug performance assessments. Known events are distinguished by (1) seizure or observation of drugs; (2) observation of activity that could not be reasonably attributed to anything other than drug smuggling; and/or (3) highly reliable intelligence. We chose to use known event data for our study because it has been captured since 1991, while event data used to support flow assessments date back to only 1996. Although using only known event data results in a large loss of data points, the value gained by expanding the time span of the study to 1991 outweighs this loss.

Smuggling event data were grouped into ten movement vectors. The vector breakout is consistent with that used by the Interagency Assessment of Cocaine Movement (IACM) and includes: Eastern Pacific, Western Caribbean, Jamaica/Cuba/Bahamas, Hispanola, Puerto Rico/USVI, Lesser Antilles, Southeast U.S., Northeast U.S., Southwest Border, and U.S. West coast. Arrival zone vectors (i.e., the last 4 listed) include only those events where the event moved directly from South America into the U.S., by-passing any transshipment countries. Figure 12 displays these vectors.

Events were further broken down into two modes—non-commercial air and marine movements. Technically, the marine category covers both commercial and non-commercial marine events. By definition¹², only those commercial (i.e., port-to-port) marine movements that are actionable at sea are considered by the CCDB for performance evaluation purposes. Since this definition inherently excludes certain forms of commercial marine movements (such as cargo containers), we felt that making a distinction between commercial and non-commercial marine events would be somewhat misleading and under-represent commercial marine smuggling events. To the extent we are modeling the effect that marine interdiction operations have on marine smuggling activity in the transit zone, the general criteria of 'actionable at sea' is adequate for our purposes.

Methodology

Our analysis tests whether or not interdiction operations displaced trafficking activity from those vectors/modes against which the operation was targeted to other vectors/modes. To conduct this test, we examined cocaine shipments (using the data set described earlier) through each of 10 vectors by two transportation modes (maritime and air) monthly for the period 1991–1999 (108 months). Ostensibly, this provided 2,160 data points, but in fact, there were no shipments through three of these vector/mode combinations¹³. Therefore, analysis used 1,836 data points.

We considered counts of cocaine shipments in two ways. The first was to include all shipments known through any combination of intelligence or observation by interdiction assets. The second was to consider only those shipments for which there was intelligence (i.e., some form of corroborating information) on the event. The second measurement is a subset of the first. Each measure has relative advantages and disadvantages. The measure based on shipments identified through intelligence or observation is the more comprehensive of the two. The disadvantage to this measure is that observations may be sensitive to the level of interdiction actually conducted in a vector and targeted against the mode. That is, interdiction activity can affect the measurement of drug movements, and thus, we might mistakenly infer that shipments have changed when only the measurement process has changed.

The second method of identifying shipments avoids this measurement problem by limiting measurements to shipments on which there was intelligence. This provides a measure that is relatively immune to changes in interdiction activity levels in the various vector/mode combinations. The disadvantage to this measure is that it is likely sensitive to changes in intelligence resource levels. Moreover, it excludes data from what is already a sparse data set for some vector/mode combinations. Given the mixed advantages and disadvantages of these two measurements, we repeated the analysis using both approaches.

As noted, our analysis attempts to explain the number of shipments through each vector, by each transportation mode, for every period as a function of several variables including special interdiction activities. To test our hypotheses, we used an estimation based on a Poisson distribution. Appendix E describes our analytic approach in detail.

We first sought to establish a baseline against which the effectiveness of interdiction could be measured. The baseline model includes variables that control for the following:

• The paucity of event data in certain vector/mode combinations
• Shifts in shipment frequency across vector/modes that are not related to special interdiction operations
• Seasonality

There is strong evidence from these baseline models that shipments are cyclical, that the trends vary across vectors and modes, and that the vector/mode combinations vary in the number of shipments that are typical. More importantly for our purposes, there is evidence that the baseline model for shipments with intelligence is void of any influence from interdiction operations. That suggests we are 'starting with a clean slate' when we begin to introduce specific interdiction operations into the model. As expected, however, this is not the case with the baseline model for shipments based on intelligence and/or observation.

Interdiction operations were represented by variables that were coded 1 for the vector/mode/time the operation took place. Typically, this was a period of more than one month. For each of these periods, we defined a complement variable coded 1 for every vector/mode that was not the target of the interdiction. For example, during the period when Gateway was operational, the GATEWAY variable was coded 1 for vector/mode combinations that were Gateway targets. During this same period, a variable NGATEWAY was coded 1 for vector/mode combinations that were not Gateway targets. During periods when Gateway was not operational, both GATEWAY and NGATEWAY were coded zero.

Continuing with the Operation Gateway illustration, we expected the variable GATEWAY to have a negative correlation with the number of shipments. If shipments moved from vectors/modes that were targeted by Gateway to those vector/modes that were not, then the parameter associated with GATEWAY should be negative while the parameter associated with NGATEWAY should be positive. In fact, because the model's controls for time trends is imperfect, we examine the difference between the regression parameter associated with NGATEWAY and the regression parameter associated with GATEWAY. If this difference is positive and statistically significant, then we infer that Gateway probably caused smugglers to shift to alternative vectors and modes.

We proceeded to add individual interdiction operations to the baseline model to see whether they have an appreciable effect on drug movements. For example, to examine the effects of Gateway, we add GATEWAY and NGATEWAY to the baseline model and test to see if the differences between the NGATEWAY and GATEWAY parameters are positive and statistically significant. We repeated this for each of the border and transit zone interdictions. Operations indicating statistical significance independently were included in a final regression model to ascertain their significance when examined within the context of other operations. The results of these final models are presented in Table 3 below.

Based on this analysis, five of the border interdictions seem to cause traffickers to move shipments from where the interdiction happened to where it did not. We note the apparent effectiveness of Hard Line, Gulf Shield, Border Shield, White Shark III and River Sweep. Regarding transit zone interdiction, Gateway appears to have been effective.

Table 3 Effect of Transit Zone and Border Operations on Displacing Drug Smuggling Activity From Targeted Smuggling Routes and Methods
	Geographic Vectors Affected	Analysis Based on Events With Intelligence	Analysis Based on All Movement Events
Border Operations
Hard Line Brass Ring Gulf Shield Border Shield River Sweep White Shark I White Shark II White Shark III	US Southwest Border Western Caribbean Eastern Pacific US Southwest Border Western Caribbean Eastern Pacific Southeast US US West Coast Southeast US Hispanola Southeast US Southeast US Southeast US	None None Positive Positive Positive None None ****	Positive None Positive Positive Positive None None Positive
Transit Zone Operations
Two Dozen Conjuntos (I and II) Frontier Lance Frontier Shield Gateway	Southeast US Jamaica/Cuba/Bahamas Western Caribbean Hispanola Jamaica/Cuba/Bahamas Puerto Rico/USVI Lesser Antilles Puerto Rico/USVI	None None None None Positive	Positive None None None Positive
**** Indicates that data were too sparse to test the significance of the operation. "Positive"—indicates that the operation had a statistically significant positive effect on displacing drug smuggling activity. "Negative"—indicates that the operation had a statistically significant negative effect on displacing drug smuggling activity. "None"—indicates that the operation had no statistically significant effect on drug smuggling activity.

Conclusions

Based on these analyses, source zone interdiction operations and the arrest or death of major drug traffickers appear to have very significant effects on increasing cocaine prices in the U.S. The impact of the source zone interdiction operations is likely due to success in temporarily reducing the supply of cocaine to the United States. Since cocaine prices are expressed in terms of price per pure gram, the increase in prices (particularly at retail level) are the result of increased 'cutting' by the dealers. This would be a fairly quick and obvious way to deal with a temporary shortage in supply.

The positive effect of operations Hard Line, Border/Gulf Shield and Conjuntos II on cocaine prices implies that these operations had a very real impact as well and either caused traffickers to increase their fees and/or reduced the supply of cocaine to U.S. markets.

Although the remaining transit and arrival zone interdiction operations examined do not have a statistically significant impact on cocaine prices in the U.S., many exhibit an impact on trafficker behavior. Figures 13 14, 15, and 16 display all CCDB movement events for four periods during 1998, by vector. The shading represents the total flow through each vector, and the pie charts represent the distribution of this flow by conveyance¹⁴. The figures clearly show not only switches in geographic vector, but mode of transportation.

This suggests that operations were effective enough to force smugglers to change their transportation routes and/or methods. However, as the counterdrug community has long recognized, drug traffickers' ability to get drugs from South America into the U.S. is limited only by their creativity. Accordingly, it appears the ability of these operations to impact cocaine prices was ameliorated by the availability of alternative transportation routes/methods.

14 It is important to note that these are not the data we used for analysis, as they include both ICPAWG and flow events.

Previous

Contents

Next

Download Adobe Acrobat Reader