A System Dynamics Examination
of the
Use of Performance Enhancing Drugs
John Voyer, Ph.D.
Cade Blackburn
Troy Goddu
Brian Sprague
School of Business
University of Southern Maine
96 Falmouth St.
Portland, ME 04104-9300
Corresponding author:
John Voyer
207-780-4597
voyer @usm.maine.edu
Abstract
Athletes face tremendous pressure to perform, and, when conventional means prove
insufficient for performance improvement, some turn to performance enhancing drugs (PED’s).
The present paper uses system dynamics to examine one example: the use of anabolic
androgenic steroids in Major League Baseball (MLB), which operates in the United States and
Canada. The authors provide an explanation of a detailed causal loop diagram of the problem,
along with a stock and flow model, based on the Bass Diffusion Model, of part of the problem.
They provide a few policy recommendations based on model runs.
Key Words: Performance Enhancing Drugs, System Dynamics, Major League Baseball, Bass
Diffusion Model
A System Dynamics Examination
of the
Use of Performance Enhancing Drugs
Athletes face tremendous pressure to perform, and, when conventional means prove
insufficient for performance improvement, some turn to performance enhancing drugs (PED’s).
The present paper uses system dynamics to examine one example: the use of anabolic
androgenic steroids in Major League Baseball (MLB), which operates in the United States and
Canada.
Steroids are a controlled substance, and MLB (and other sports) bans them because of the
effect they have on the user. Positive effects in the short term include an increase in lean
muscle mass, strength, and the ability to train longer and harder. This usually translates into a
better athlete. For example, the use of steroids took center stage during the home run race in
the late 1990’s between Mark McGuire and Sammy Sosa. It was widely believed that both men
used steroids to improve hitting performance. Although the race between these two sluggers
was exciting, the notion that somehow they were cheating drew great ire from fans at all levels
of the game. This cheating not only reflects poorly on the moral integrity of the players
choosing to cheat, but it also forces non-users who wish to remain competitive to contemplate
cheating. This reinforcing escalation is at the heart of the PED problem in MLB.
Another problem with use of PED’s is that no one knows the long-term negative effects. We do
know the major side effects of steroid use include an increased risk of cancer, increased risk of
heart and liver disease, jaundice, fluid retention, reduction in HDL-C (“good cholesterol”), high
blood pressure, changes in blood coagulation, increased risk of atherosclerosis, swelling of the
soft tissues of the extremities (edema), and obstructive sleep apnea. Side effects specific to
men can include testicular atrophy or the shrinking of the testicles, reduced sperm count,
infertility, baldness, and the development of breasts. (Drug Enforcement Administration, 2004;
Hartgens & Kuipers, 2004) It is arguable whether medical intervention can reverse these
effects. If users inject the drug, they are at risk for infection with HIV if they share needles, and
there is a chance of getting hepatitis if they use dirty needles. A recent study by the American
Heart Association showed that the left ventricle, the heart's main pumping chamber, was
significantly weaker during contraction (systolic function) in participants who had taken
steroids compared to a group of similar non-steroid users. (Baggish, et al., 2010) These
documented health risks are rather clear-cut, but they take time to develop. Given the short
careers of professional athletes, and the intense competition they experience to make it to
MLB, it is very difficult for a player to look past the immediate financial and social gains and see
the long-term risks of taking steroids.
The challenge for MLB in eradicating steroid use has largely been self-inflected. Though MLB
has listed steroids as a banned substance since the early 1990’s, the league and the players
union only agreed on random testing in 2003. (Mitchell, 2007) Though theirs was a
controversial assertion, some pundits suggest that use of steroids was rampant. Some said as
many as 60 percent of players were using, and the league’s players, owners, and the members
2
of the press who covered them, reinforced a culture of selective ignorance. More recently,
resistance to eradication of PEDs has arisen from the players’ union, which refuses to allow its
members to submit to tests that involve drawing blood. This has given rise to the use of human
growth hormone (HGH). (Mitchell, 2007) Therefore, in essence, one PED has replaced
another.
The present paper will address some of the fundamental forces driving steroid use in the MLB.
Though there is no readily apparent easy answer, we do make some policy recommendations
for MLB, which we suggest as starting points for applying leverage in the right places.
Reference Modes
The time horizon for this study is roughly 60 years, starting in 1950 and ending in 2010. The
challenge with this particular topic is that nobody actually admits to using steroids (save those
few looking to make money in a book deal after their careers have collapsed). These reference
modes are our general estimate given anecdotal evidence, as well as, the Mitchell Report,
which was a rather comprehensive look into the use of PEDs in MLB.
In the Figure 1, we estimate that growth of steroids in MLB has an exponential shape starting in
the 50’s and growing through 2002. MLB instituted mandatory testing in 2003 and fines and
suspensions become more severe in 2005. What this chart does not show is the possible
growth of HGH use, which is still not tested, and for which investigators have discovered very
little evidence.
Performance is very challenging to
measure. For this research, we selected
home runs.’ The challenge with home
runs is that the measure excludes a
population of players, pitchers, whom
many considered equally likely to have
been using steroids. If this is the case,
then stronger pitching may have offset
stronger hitting, thereby keeping home
runs at the normal historical average.
1950 1959 1968 1977 1986 1995 2004 Given the McGuire and Sosa race, and
7 : given how Barry Bonds (who was widely
Figure 1. Performance Enhancing Drug Users thought to be using PED’s) broke Hank
Aaron’s lifetime home run record, we
selected this performance statistic because of the ease of understanding, and the national
coverage it garners. Figure 2, Total Home Runs, does show growth, but it fails to reflect that
PED Users
800
200
* Home runs in baseball occur when a batter hits a pitched (i.e., thrown) ball over the fence in fair territory ona
baseball field. The batter who does so may then run around the game’s four bases at his leisure, arriving back at
“home plate,” which gives this hit its name. Obviously, home runs are much likelier when batters are stronger.
3
the number of teams and the roster sizes have changed throughout the 60-year period. Still,
there is a general rise in total league home runs that could suggest steroid use.
6000 Not surprisingly, if overall system
5000 [an performance increases, it is a logical leap to
i8 A ia suggest that average player performance
increases. This is important for our model
because one of the driving forces for PED use
2000 escalation is the perception of comparison
1000 between oneself and another player. Figure
0 err 3 shows average home runs per player over
BRRBSSSERRSSRRSSS the 60-year period. This statistic controls
see ess Ns somewhat for the change in number of
Figure 2, “Tetel Home Runs it MAB, 19502008 players and number of games over the sixty-
year period, and it shows growth in number
a Dar, of home runs per player from 1992 to 2001.
4000 A fe At very specific points in time, MLB instituted
3000 carey —— policy measures to thwart steroid use. The
2000 measures primarily consisted of testing
1000 standards and intervals, and rules governing
0 EE the punishment for incremental positive
EI 3 3 g g § § 5 g 8 g § a g 3 5 tests. MLB started testing in 2003 and raised
Figure 3 Average Home Runs per Batter, 1950-2010 the testing protocol and punishments in
2005. (ESPN, 2007) This last policy measure
800 is where we think players began looking at
other PEDs, like HGH, for alternatives to
anabolic steroids.
600
Figure 4, “Potential PED Users,” is useful in
200 that it suggests a large number of the
| susceptiable population actually began using
CC a CC a rT steroids. When we examine the model, it will
become evident that there are two
reinforcing loops that are free to run for
many cycles before any balancing loops come
into play. This tends to reflect the rumored usage rate of 60 percent. We will begin our
examination of dynamics of this situation with the full causal loop diagram, which we show in
Figure 5. We will follow that with discussions of subsections of and important loops in that
diagram.
Figure 4. Potential PED Users
Average Player
Performance Goal
Average Player
serene Dover
Testing for one drug
leads to adoption of new
PED Escalation loop
drugs
‘Adoption from
Perceived Peformance
Governing Body+ ‘aap
Intervention
Gp
Testing Leads to
Fines & Suspensions
‘Adoption Rate
+ Word of Mouth
‘Adoption from
Talking to PED Users
New
+ Players
+ Retired
Players 2
+ Abandonment Rate Contact Rate vers ———B™Potential PED Users
* an (@s
Natural Potential PED User
AbandonmentRate Replenishment
Poemenant
Abstainer Rate Permanent
Percieved RS Abstainers
Health Risks
to Health Danger
Prolonged PED Use
Leads to Health
Problems
Players with
Heath Problems +
Figure 5. Overall Causal Loop Diagram
‘Average Player
Performance Goal
‘Average Player
Performance
‘Aggregate Player
Performance
Percieved Average
RI Player Performance
Gap
PED Escalation loop
Improved
Performance per New
PED User
PED Users *
id Adoption from
Perceived Peformance
ap
‘Adoption a
Figure 6. Loop R1 — PED Escalation
+
——
PED Users
aw +
F
‘Adoption Rate -_
+ Word of Mouth
Adoption from
“alking to PED Users
New
+ Players
Retired ¥
Players ——t Potential PED Users
+A\
Contact Rate
Figure 7. Loop R2- Word of Mouth Escalation
Governing 8ody i)
Intervention Users
@s
Testing Leads to
Fines & -
Suspensions Natural
‘Abandonment Rate
‘Adoption Rate
Potential PEO
+ Abandonment Rate Users
Permanent
Permenant Abstainers
4 Abstainer Rate
Figure 8. Loop B1 — Testing, Fines, and Suspensions
As we will discuss later, the underlying
structure of this problem is the Bass
Diffusion Model (Sterman, 2000), where
the stock non-using players adopt PEDs
and flow into the stock of users. Several
loops in the diagram illustrate how this
happens.
In the R1 loop (Figure 6), “PED
Escalation,” the key driver is players
comparing themselves to the average
player’s performance. In this case,
players leave the pool of potential users
and become users—if they perceive
themselves as falling behind the average
performance level.
As in the Bass Diffusion Model, players
may also begin using PEDs if they find
out from fellow players that the drugs
can actually enhance performance. In
this scenario, which we show in the
“Word of Mouth” loop in Figure 7,
Potential Users interact with PED Users
at some given contact rate. Each contact
contributes to new adoption, providing
more users to the system, which drives
up both the R1 and the R2 loops.
The key to loop B1, “Testing Leads to
Fines and Suspensions” (Figure 8), and to
the system’s behavior, is the delay
between PED Users and Governing Body
Intervention. In our estimates, it took
MLB forty years from the start of our
time horizon to acknowledge steroids as
a banned substance. Another thirteen
years went by before it implemented
testing.
The “Potential PED User Replenishment” roe Us
loop (R3, Figure 9) is straightforward:
some players who abandon do not quit
permanently. Like smokers addicted to
nicotine, some PED users who quit may (oa
miss the feeling of power and the VY
improved performance that allegedly was ate “talent Nt OS houses
comes with PED use. Those players again 7
become Potential PED Users, and some
elect to use again. (A fraction of those
who abandon choose to _ abstain
permanently, which we depicted earlier
in boldface along the bottom of the
diagram in Figure 8.)
‘Adoption Rate
New
Players
Notural
‘Abandonment Rate
Figure 9. Loop R3 - Abandonment and Replenishment
We do not have any data on users
PED: Switching switching from steroids to HGH, as
* Ase) depicted in Figure 10, the reinforcing
Sig tor one AE loop R4. However, the Mitchell Report
sea? lal (Mitchell, 2007) hints at switching, and
i anecdotal evidence suggests it is a real
7 possibility. The switching that occurs in
euaeitne ole other sports, like cycling, also presents a
Anseryention. basis for comparison. Other than the
Figure 10. R4— PED Switching selective ignorance we have already
mentioned, this is a significant point of
policy resistance. Though we did not present it in our model, the collective bargaining
agreement between the league and the players’ union, the Major League Baseball Players
Association (MLBPA), really helps facilitate this by taking a very strong stance against blood
testing. Until MLB and the MLBPA can come
New
to terms on a more pervasive testing culture, eee, ved
PED switching will continue, and will likely .
have adverse health effects on the players Potential PED Pr adoption Rate
Users
they are trying to protect.
The major component of loop B2 (Figure 11), +
C , ‘Abandonment Rate PED Users
as in B1, is the long delay. Where players are * Lt)
so focused on the short term payout, the protsiaed PED Une
longer term dangers of prolonged PED use get venteved ‘problems
brushed aside. Unfortunately, it is going to Hela ‘.
take a very high profile health issue, like "Bhan cinee”
death, to snap players out of their money fog Players with +
Heath Problems
and realize PEDs are very, very dangerous.
Figure 11. Loop B2 - Heath Problems
System Dynamics Model
We converted part of the causal loop diagram into a system dynamics model, which we show in
Figure 12. This model is a variant of the Bass Diffusion Model (discussed at length in Sterman,
2000). The model we used is very rough, and has several weaknesses. The biggest weakness is
that it uses a fixed number for Players in the System, when we know that MLB baseball
expanded greatly over the sixty years of this time horizon. It also uses a constant for Home
Runs per Player per Year, when we know that this average crept up during the period in
question. Therefore, results from the present study should be taken as preliminary.
In this model, Potential PED Users adopt them and become PED Users. We focused primarily
on adoption rates as driving by perceived performance gaps and word of mouth. We created a
user life cycle stock and flow, and looked at drivers of both adoption and abandonment.
Finally, we examined the effect of use on performance, specifically home runs, through a co-
flow. Because of the long delay, we excluded health risks as a driver of permanent abstention.
We will cover some of the key highlights here before we move on to a policy discussion.
Home Runs per
Player per Year —~C-——-—=m|_ Total Home
Increase in Runs Dacieaps' hi
Home Runs Homa Rune Average Home
~__ Runs per Player
Natural Rate of
Stopping PED Use
MLB Testing __». Positve Test.
Results Stopping
Postive Test
Results Leads to
Stopping PED Use
Rate of Stopping
PED Use
Players in the
System
Potential
ome = PED users me pmee
Abstaining Adoption
Rate Rate AR
R
Word of
Mouth
Abandonment
percentage
Adoption from ‘Adoption from
Peformance Talking to Users
‘Gap
+ ‘Adoption
Fraction i
Rate of Adoption from
Perceived
Performance Gap
Contact
Rate
Figure 12. A System Dynamics Model of the Performance Enhancing Drugs (PED) Problem
PED Users Firstly, as expected, the usage rates starting at
800 time zero (1950) grow exponentially until the
first and second MLB interventions, which
occur in 2000 and 2005, as shown in Figure 13.
‘
= 400
Pe Secondly, the stock of potential PED users
200 drains quite quickly. We first thought this was
° excessive, but given some of the anecdotal
1950 1956 1962 1968 1974 1980 1986 1992 1998 2004 2010 evidence, we have reason to suspect this
‘Time (Year) Fy . :
PED Users : MLB testing re pattern is not entirely off base. By the mid to
Figure 13. Number of PED Users late 1990s, our model suggests only around
100 out of 750 players were not using (see Potential PED users
Figure 14). 800
Finally, when we examine our model’s 600
output relative to real data (actual league
home runs per year), which we show in 5 0
Figure 15, we are encouraged to see we
captured the general trend. However, we
clearly missed on a few points, including 6
the oscillation that appears, and that the 1950 1956 1962 1968 1974 1980 1986 1992 1998 2004 2010
‘Time (Y ear)
number of teams and players has Potential PED users : MLB testing
increased over the period. Figure 14. Potential PED Users
7000
6000 —
5000
4000
——Total HR in MLB
3000
—— Model Tot HR
2000
1000
(ee LAAAAASAAABAAAASAASANLAAGASAAAALAADBLAALALASAASSAALALLLALIGS
Ot+yHMNDOCOHTHRAVOTHAWO
RAKHSSERARRHADASSST
DBAARA HSBARARHHRASSS
SAASSRAASAAAAAAAAN
Figure 15. Total Home Runs, Model vs. Actual Data
Policy Analysis
When looking at the policies MLB has implemented to combat PED use, two things stand out.
First, the culture of selective ignorance in all levels of the system absolutely contributed to the
exponential growth in steroid use. Second, MLB has been manhandled by the MLBPA in
negotiating real and meaningful testing policies to prevent future types of PEDs from entering
the system. However, the policies have been effective in at least setting the groundwork for
more policies and more testing. And, when faced with the realization of how pervasive steroid
use was, MLB took immediate steps to increase the penalties. In 2002, the penalties consisted
of 5 allowable positive tests, with the fifth positive test a discussion with the commissioner. By
the end of 2005, the league had instituted a new penalty scale (the second one that year)
consisting of three positive tests, with the third positive test a lifetime ban. We think this
newer scale has had a significant effect on steroid use. However, we are not certain it has
really forced PEDs out of baseball, since testing is very limited in the current collective
bargaining agreement.
So, what else can MLB do? The penalty scale could go down to two or one positive test, but
that presents a steep penalty for the real, though small, percentage of tests that generate false
positives. And, in the current Collective Bargaining Agreement environment, we are not sure
how much that would stop
PED Users other PEDs.
am We do think MLB and MLBPA
ies could do a lot more with
education. We_ understand
a 95 most players will be more
By concerned with their short-
425 term payout and ignore any
health risks, but if MLB
-10 implemented mandatory
1250 1760 1200 T ne) te 2000 2010 education classes on the health
‘ime (Y ear)
PED Users : Current MLB policy risks of PED use, it might have
PED Users : More stopping an effect on a few players.
PED Users : More abandonment
Figure 16. Effect of Different PED Reduction Policies
Figure 16 shows the effects of
three policies run in the model:
the current policy, a policy where double the players temporarily stop using after testing (called
“More Stopping”) and a policy where the rate of permanent abstention goes from 5% to 6%
after learning about the adverse long-term health effects (called “More Abandonment”). Figure
16 shows that the improvements from doubling the amount of temporary stopping are modest;
however, increasing permanent abandonment by 20 percent (from 5% to 6%) has a much
greater effect.
Ultimately, we think the real leverage is in renegotiating with the MLBPA to institute a flexible
policy that will adapt to the PED market and allow the league to institute testing in a real and
10
meaningful way. Until then, PED switching will likely continue to drive usage and keep
relatively high the percentage of players using.
Future research
PED usage in Major League Baseball is a big problem for the league, and we think it is worth
refining the current model to make policy analysis more effective. We expect to continue
working to improve and refine the present model.
References
Baggish, AL, Weiner, RB, Kanayama, G, Hudson, JI, Picard, MH, Hutter Jr., AM and Pope Jr. HG.
2010 “Long-Term Anabolic-Androgenic Steroid Use Is Associated With Left Ventricular
Dysfunction.” Circulation: Heart Failure 3: 472-476
Drug Enforcement Administration. 2004 “Steroid Abuse in Today's Society: A Guide for
Understanding Steroids and Related Substances.” Accessed on March 20, 2011 at
http://www.deadiversion.usdoj.gov/pubs/brochures/steroids/professionals/index.html
ESPN.com. 2007 “Mitchell report: Baseball slow to react to players' steroid use.” Accessed on
March 20, 2011 at http://sports.espn.go.com/mlb/news/story?id=3153509
Hartgens, F., Kuipers, H. 2004 “Effects of androgenic-anabolic steroids in athletes.”
Sports Medicine 34(8): 513-554.
Mitchell, GJ. 2007 Report to the Commissioner of Baseball of an Independent Investigation into
the Illegal Use of Steroids and Other Performing Enhancing Substances by Players In Major
League Baseball. Major League Baseball, Office of the Commissioner, 53.
Sterman, JD. 2000 Business dynamics: systems thinking and modeling for a complex world.
Boston: Irwin/McGraw-Hill.
11
