Calming the storms: Collaborative public management, hurricanes katrina and rita, and disaster response

Full text

Rma
Chapter
6
vVrs
Calming
the
Storms: Collaborative
Public
11’
v
Management
Hurricanes
Katrina
and
Rita,
and
Disaster
Response
vs
Alisa
Hicklin,
Laurence
I
Oloole
Jr
N
Kenneth
i.
Meler,
and
Scott
E.
Robinson
As
scholars
have
explored
the
relationship
between public
management
and
organizational
performance,
a
considerable
body
of
work
has
identi
fled
interorganizational
collaboration
as
an
effective
strategy
to
improve
performance.
These studies
show
a
number
of
benefits
that
can
he
linked
to
the
presence
of
a
more
collaborative
public
manager.
As
managers
build
relationships
with
oiler
groups
in
the
interdependent
environment,
these
links
often
jesuit
in
higher
levels of
support
for
the
organization,
joint
ventures
in
pursuing
policy
goals,
avenues
for
the
acquisition
of
additional
resources,
and
opportunities
to
proactivelv
address
some
possible
threats
to
the
organization
and
its
programs.
Given
Ue
evidence,
it
would
seem
that
networking
to
build
collaborative
relationships
is
a
managerial
acUv
ity
with
very
few
drawbacks—aside
from
[lie
opportunity
costs
necessarily
However,
most
of
this
evidence
examines
collaboration
at
times
when
the
organization
is
funcuoningas
it
typically
operates—carningout
its
core
involved.
functions
and
addressing
somewhat
predictable
problems.
Researchers
in
public
management
know
much
less
about
patterns
of
collaboration
in
times
of
sharp,
unpredictable
organizational
crisis.
This
gap
in
our
knowl
edge
is
problematic
for
two
reasons.
First,
times
of
short-term
crisis
often
have
long-term
effects
on
the
organization and
the
people
involved
As
we
have
seen
in
a
number
olmajor
crises
(the
September
11,2001,
terrorist

96
Why
Public
Managers
Collaborate
Calming
the
Storms
97
attacks;
Hurricane
Katrina;
etc.),
effective
management
can
help
an
or-
ganization
to
overcome
the crisis
and
even
grow
stronger, but
poor
man-i
aget.s
can
causc
lasting
and
irreparable
damage. Second,
the
imniediacy{
required
in
responding
to
a
crisis
may
change
the
ways
that
managersiew
collaboration.
Though networking
can
be
expected
to
be
benefIcial
for
problem
solving,
there
are
reasons
to
expect
that
building these
collabo-j
rative
relationships
during
times
of
disruption
and
distraction
could
be
yen’
time
consumingand
costly.
In
times
of
externally imposed
crisis, when
the organization
must
respond
quickly
to
a
major
environniental
shock,
what explains
the
extent
of
collaboration
with
others
in
the
interdepen
dent environment?
In
particular,
do
established
patterns
or
styles
of
man
agement
externally
contribute
to
the
development
of
interorganizational
I
collaboration
during
ci-isis
periods?
These
questions
speak
to
the
broader
issue
of
the
determinants
of
collaboration.
Do
managerial
choices
shape
collaborative
results?
Is
the
decision
to
engage
in
collaboration
strategic?
Is
it
problem-specific?
Or
could
collaboration
develop
as
a
product
of
a
more
diffuse
management
style
emphasizing external
interactions
with
others?
This
chapter
ad.
dresses
these
questions
by
drawing
on
recent
work in
public
management
to
develop
lwpotheses
about
what
could
drive
the
development
of
collabo-.
rative
ties
in
response
to
a
major
organizational
shock.
These
hypotheses
are
tested
in
a
natural-experiment
design
by
investigating
the response
during
the
aftermaths
of
Hurricane
Katrina
and
Hurricane
Rita,
when
school
districts
not
directly
hit
by
the
hurricane
nonetheless
had
to
re
spond
quickly
when
bombarded
with
an influx
of
displaced
students.
WICKED
PROBLEMS
AND
COLLABORATIVE ACTION:
DISASTERS
AND
THEIR
PUBLIC
MANAGEMENT
Partnerships, interorganizational
programs., arid
collaboratives
are
all
the
rage.
The
interest
in
these
sorts
of
patterns
is
not
confined
to
the
United
States
but
is
also
visible
in
the
United
Kingdom,
other
Westminster
set
tings
(Lowndes and
Skelcher
2001;
Rhodes
1997,
2002;
Stoker
2004;
Sullivan
and
Skelcher 2002),
continental
Europe
(Bogason
and
Toonen
1998;
Van
Bueren,
Klijn,
and
Koppenjan 2003;
Kiekert,
Klijn,
and
Kop
penjan
1997a,
1997b;
Mijn
1996; Raab
2002),
and elsewhere,
Many forces
have
driven
this
upsurge
in
attention
(O’Toole
1997), even
though
one
can
question
whether
such
artangements
are
particularly
new
or
have
become
unusually
visible
in
recent
years
(Hall
and
O’Toole
2000,
2001).
The
importance
of
public
management
to
successful
collaboration
is
a
theme
that
has
been
emphasized
by
several
scholars
(e.g.,
.-aiioff
and
McGuire
2003a; Meier
and
O’Toole
2003;
O’Toole 1997;
O’Toole
and
Meier
2003;
Provan
and
Milward
1991).
Among
the reasons
wiw
interorgani
zaflonal
arrays
have
been adopted
as
a
means
of
executing
public
purposes,
and
why
public
management
can
be
a
key
element
in
their
successful
operation,
is
the
prominence
of
so-called
wicked
problems
(Rittel
and
Webber
1973)
on
the
public
agenda.
When
the
kinds
of
issues
demand
ing
policy
and
management
attention cannot
be
neatly
compartmental
ized
in
one
sector—and
one public
organization—but instead
span
fields,
sectors,
specialties,
and
extant
institutional arrangements,
new
and
often
collaborative
cross-organizational
forms
may
be
the
preferred
structural
choice.
Many
of today’s
most
pressing
challenges,
from
homeland
secu
rity
to
HIV/AIDS
to
climate change,
exhibit
wicked-problem
features.
One
of
the
most
obvious
of
such
challenges
is
governmental
response
to
natural
disasters
(Comfort
2006).
Often
appealing
without
warning,
disasters
like
earthquakes,
floods,
major
storms,
and
wildfires
can
unleash
devastating
forces
that
cause
massive
destruction
and
loss
of
life,
along
with
severe
disruption
and
dislocation
in
the
lives
of
many.
Natural
disas
ters
also
touch
upon
many
pulley
hells
and
governmental
responsibilities
simultaneously.
Such
events
are
no
respeeters
of
jurisdiction,
and
they
can
be
considered
shocks
to
multiple
social, ecological,
and
physical
sys
tems
simultaneously—with
reverberations
that
can
reach
across
time
and
even
huge
distances.
Consider,
for
example,
the
tsunami
in
South
Asia
in
December 2004,
or
the
prospect
of
significant
melting
in
the
Greenland
ice
field
that
many
experts
anticipate
in
the coming
years.
Clearly,
natu
ml
disasters
can
require
responses
that
integrate
efforts
and
organizational
actnThes
from
many
fields,
often
in
thtricate
fashions.
Multiple
levels
of
government,
multiple
agencies
of
government,
multiple
governments
at
the
same
level,
as
well
as
multiple organizations
in
the
private
and
not-
for-profit
sector,
may
all
need
to
be
mobilized
and
may even
be
required
to
work
closely
with
each
other
in
tight
patterns
of
coordination
if the
myriad
issues
generated
by
major
disasters
are
to
be
addressed.
Some
of
these
types
of
collaboration,
involving
certain
of
these
organi
zations.
can
and should
he
anticipated
and
planned
for
in
advance.
This
is
one
of
the
principal premises
underlying
efforts
at
disaster
preparedness,
as
called
for
by
national
policy,
as
well
as
the
plans
and
policies
in
many
states.
Still,
if
the
scale
of
the
disaster
is
great
and
especially
if the
timing
cannot
be
anticipated.
some
of
the resulting
needs,
including needs
for
collaboration,
cannot
be
programmed
in
advance—certainly
not
in
intricate
detail.
In
such
circumstances,
coordinated responses
may have
to
be
mobi
lized
quickly
and
under
pressure,
and public
managers
can
be
called
upon
to
mesh multiple
streams
ofintncate
effort—virtually
overnight.

98
Why
Public
Managers
Collaborate
Calming
the
Storms
99
When
such
efforts
fail,
the
costs
can
be
enormous.
Witness
the
devas
tatthgly
ineffectual
responses
at
virtually
all
levels
and
by
many
individuals
and organizational actors
during
and
after
the
September
2005
hunicanes
on
the
U.S.
Gulf Coast.
Hurricane
Katrina and
1-lurricane
Rita
exposed
huge
weaknesses
in
the
systems
of
disaster
response
in
the United
States,
the
states
of Mississippi
and
Louisiana, and
communities
like
New
Or
leans.
Researchers
will
be
probing
the
experiences
related
to
these
disas
ters
for
some time
to
cnrne.
Still,
some
systematic
investigation
can
be
undertaken
even
now,
at
least
on
some
of
the
salient
questions.
Of
particular
interest
here
is
the
question
of
whether
and
why
some
organizational
responses
triggered
by
Katrina
and
Rita
were
marked
by
substantial interorganizational,
collabo
rative
activity
in
the
interests
of
addressing
unexpected
disruption,
while
others
produced
very
little.
We
focus
specifically
on
one
discrete
slice
of
the
overall
picture
and
explore
in
particular
whether
prior
patterns
of
active
networking
by
top
managers
helped
to
facilitate
a
more
vigorous
collabo
rative
response
to
unexpectedly
disruptive
shocks
in
the service
deliven
system.
In
doing
so,
we
build
on
earlier
work
on
public
management,
collabo
rative
processes,
and
performance
(OToole
and
!\leier 1999).’
We
adapt
that
work
to
the
realm
of
disaster
response,
a
field
in
which
specialists
have
understandably
paid
attention
to
collaborative
patterns.
In
research-
lug
organizational
response
to
disasters,
scholars
have
long
seen
the
im
portance
of
interorganizaUonal
coordination.
Wenger,
Quarantelh,
and
Dynes
(1986,
10—11),
for
instance,
found
that
many
organizations
dis
cussed
collaboration
as
agoal
of
their
efforts
but
seldom
included
coordi
nation
in
theiractual
activities.
Drabek
showed
that
coordination
was
part
of
the
emergence
of
a
multiorganizational
network
in
the
cases
of
disas
ter
response
that
he studied
(Drabek
et
al.
1982;
Drabek
1983,
1985). More
recent
work
has focused
on
the
role
of
intergovernmental
networks
in
responding
to
disasters.
Schneider
(1995)
examined the
coordination
of
federal
and
state
relief
in
multiple
natural
disasters,
A
key
factor
in
the
perceived success
of
government
efforts
was
clarity
about the
division of
responsibility
between the
levels of
government.
Even
today,
the
division
of
responsihility
is
unclear,
as
the
severely
bungled
disaster
response
in
New
Orleans
has made
apparent.
Later
work
has
refocused
attention
on
the
emergent
nature
of
orga
nizational
networks
(Comfort
1999).
In
researching
organizational
pre
paredness
in
Saint
Louis,
Gillespie
and
Streeter
(1987)
found
that
an
organization’s
structure,
the
environment
in
which
it
operates,
and
its
history
with
emergencies
influenced
its
preparedness
efforts.
In
addition,
the
quality
of
its
interorganizational
relations
was
an
important
contribu
tor
to
its
emergency
preparedness.
In
light of
this
research,
Tierney
and
her
colleagues suggested
that research
on
emergency
preparedness
net
works
is
a
“particularly
promising approach”
(Tierney,
Undell,
and
Perry
2001,
60).
This set
of
studies
also
suggests
that
public
management
may
he
a
critical
element
in
this
field,
as
it
is
for
collaboration
and
coordina
tion
in
other
policy
sectors.
Explaining
the
emergence
of
collaboration,
therefore,
is
a
key
ques
tion
generally
in
wicked-problem
contexts
and
takes
on
particular
salience
in
settings
where
collaboration
is
needed
in
unexpected
and
often
wide
spread
fashion.
Specialists
on
disaster
management
recognize
the
issue
as
central.
We
examine
one
portion
of
the
topic
here
by
taking advantage
of
extensive
prehurricane
and
posthurricane
research
conducted
in
one
set
of
contexts:
Texas
school
districts.
EDUCATING
AND
ASSISTING
EVACUEES
F
Collaboration
by
public
managers
and
their
organizations
is
obviously
an
expected
response
to
natural
disasters,
which
typically
overwhelm
those
immediately
affected
and
also
impose
major,
and
often
long-term,
exter
nalities
on
others,
including
other
communities.
During
September
2005,
1-lurricane
Katrina
and
Hurricane
Rita
provided
vivid
examples
These
disasters
caused untold
devastation
and
also
triggered
the
movement
of
millions of
evacuees
to
other jurisdictions,
indeed
to
other
parts
of
the
country.
Even
before
the
landfall
of
Flunicane
Kattina,
thousands
of
people
fled
the
Gulf Coast
and
parts
of
Northeast
Texas
also
in
the
hurricane’s
path,
with
many
traveling
west
to
unaffected
cities
across
Texas,
which
organized
to
accommodate
them.
One
set
of
effects
on
Texas
local
governments and
public
managers
occurred
in
public
school systems, which
had
to
take
on
the
challenge
of
handling
many
additional,
high-need
students
with
almost
no
notwe.
These systems
were
faced
with
the
unique
task
of
absorbing
a
large
num
ber of
students
who
had
diverse,
extensive
needs
reaching
well
beyond
what
school
districts
are
normally
expected
to
address.
This
complexity
was
further
compounded
by
the
fact
that
many
decisions
and
constraints
in
the
arena
of
public
education
are
framed,
in
efftct,
at
one
remove,
because
the
bulk
of
the
regulations,
curricula,
and
operating
procedures
implemented
locally
are
set
at
the
state
level.
With
most
evacuees
having
migrated
across state
lines,
the
interstate
dimension produced
additional
management
challenges
that
were
not
able
to
be
resolved
without
cross
jurisdictional
efforts.

100
Why
Public
Managers
Collaborate
Calming
the
Storms
101
Indeed,
numerous
aspects
of
this
nicked-problem
challenge
called
for
these
public
organizations
to
collaborate
with
other
actors—police,
Ore,
and
first
responders;
nonprofit
and
relief
organizations;
other
school
sys
tems;
governmental
relief
and
welfare
organizations;
business
organiza
tions;
and
local,
community,
and
religious
organizations.
\Vithhi
days
after
Katnna,
it
had become
obvious
that
people
displaced
from
the most
heavily
affected
areas
of
the
Gulf
Coast
would not
be
able
to
return
to
their
origi
nal
homes
any
time
soon.
These
victims
were forced
to
look
for
long-term
housing—and
children
in
these
families
would
not
be
able
simply
to
wait
out
the
displacement.
They
needed
schooling,
and
more.
To
compound
the
situation,
parts
of
East
Texas had
to
evacuate
as
the second
storm,
Hurricane
Rita,
flooded
many
areas.
A
result
was
the
long-term
displace
ment
of
many
students
into
school
districts
in
Texas
just
weeks
after
the
start
of
the
new
school
year.
In
November
2005
we
initiated
the
administration
of
a
mail
survey
directed
at
top
managers—superintendentsof
Texas
school
districts.
The
survey,
administered
in
three
waves
between
November
and
January,
achieved
a
47.7
percent
response
rate
(N
=
600), Data
were
collected
on
a
number
of
issues
related
to
how’
the districts
responded
to
the
sudden
influx
of
these
students
and
how
the
unexpected perturbation
to
the
edu
cational
system
affected
the district’s
own
emergency planning. The
scale
of
the
externally
generated
shock
varied
considerably among
Texas
school
districts.
Many
districts
took
in
only
a
handful
of
students
with
little
dis
ruption,
while
oilier
districts
felt
a
considerable
impact.
;ith
larger
dis
tricts
Onding
themselves
required
to
deal
overnight
with
as
many
as
3,500
new
students
and smaller districts receiving
enough
evacuees
to
raise
their
enrollments
considerably.
Those
districts
that
absorbed
a
substantial
number
of
evacuated
stu
dents needed
to
integrate
them
into
their
classrooms
as
soon
as
possible,
a
challenge
that
in
turn
required
the districts
to
address
a
number
ofad
ditional
needs,
some directly
related
to
education
and
others
important
but
more indirectly
connected
to
the
core
task.
\1aiy
superintendents
reported
that
their
districts
provided
a
number
of
goods
and
services
to
the
evacuated
students.
Sonic
elements
provided
were
directly
related
to
the
educational
process
(crafting
orientation
programs,
providing
text
books,
openingnewbuildings, hiringadditional
teachers),
and
others
were
not (offering
health
care,
shelter,
food,
Federal
Emergency
Management
Agencyinformation,
etc.).
Not
surpnsingly,
addressing
this
broad
array
of
student-centered
needs
led
superintendents
to
look
to
other
groups
and
organizations
in
the
community that
might
be
able
to
assist
in
ameliorat
ing
some
of
the
problems
and
providing
relevant
services.
COLLABORATION
IN
EMERGENCY
RESPONSE
The
decision
of
many
superintendents
to
look
outward
to
other
groups
and
organizations
to
aid
in
the
response
is
unsurprising,
given
the
diver
sitv
of
needs that
emerged.
However,
districts
varied
considerably
hi
the
extent
to
which
they
were
stimulated
to
collaborate
with
other
actors.
This
variation
is
interesting, because
it
touches
on
a
number
of
issues
related
to
managerial
networking
and
interorganizational
collaboration.
In
par
ticular,
it
allows us
to
consider
the
determinants
of
collaboration.
General
treatments
of
networking
and
of
collaboration
often
point
to
the
emergence
of
interorganizational
and
intergovernmental
patterns
of
interdependence,
including
that,
because
of
expectations,
governments
address
tendentious
wicked
problems
(Agranoff
and
McGuire 2003b;
JOhn
2005;
Lynn,
Heindch,
and
Hill
2001;
O’Toole
and
Meier
1999).
One
would
expect
more
complex
problems
to
require
more
innovative
and
compre
hensive
managerial
approaches
and
organizational
strategies,
but
we
have
little
systematic
empirical
evidence
to
support
the
asserted
link
between
heightened
complexity
and collaboration
(Klijn
2005).
Indeed,
as
the
ensuing
discussion
suggests, sonic
of
the
literature
might
encourage
an
expectation
that environmental
shocks might
sometimes
actually
reduce
external
connections.
Most
instances
of
the
emergence
of
networked
pat
terns
in
the
public
sector
reflect
incremental
changes
that
organizations
and
their
contexts undergo
over
extended
per
ods
of
time.
However,
the
challenges
faced
by
public
organizations
and
managers
from
a
major,
unexpected
shock
to
the
system
could
stimulate
fundamentally
different
dynamics
than
those triggered
by
small
Iluctuationsin
organizabonal
pro
cesses,
especially when
this
disruption
is
not
something that
the
organi
zation
has
faced
before.
We
explore
this
type
of
situation here.
Whereas
the
research
literature
on
collaboration
and networks
clearly
argues
that
wicked
problems
are
likely
to
encourage
more
interorganiza
tional ties
and
more
externally
oriented
networking,
other
arguments
suggest
that
matters
may
not
be
so
straightforward.
In
his
discussion
of
how
public
organizations
react
to
major
shocks
from
the
environment,
for
example,
Kaufman (1985)
argues
that
such
units
can
be
expected
to
re
spond
by,
in
effect,
either
expanding
or
contracting.
In
a
decision
to
cx
pand,
the
organization
deals
with
environmental
forces
by
joining
with
externaI
actorsJ
in
confederal
systems
or
federations”
(see
also
Thomp
son
1967),
whereas
a
decision
to
contract,
or
insulate,
would
prompt
a
reduction
of
exchanges
across
boundaries
in
an
effort
to
satisfy
most
needs
and
wants
internally”
(Kaufman
1985,
43). Although Kaufman builds
on
this
logic
to
predict
that
the
vast
majority
of
organizations
are
incapable

Calming
the
Storms 103
lilt
vvny
ruDlic
ivianagers
toiiaoorate
of
dealing
with
these
kinds
of
shocks—a
notion
that has
very
little
empiri
cal
support—we
can
apply
this
basic
logic
to
consider
individual
manage
rial
choices
and behavior.
When managers
are
confronted
with
a
large-scale
disruption,
they
may
be
faced
with
a
decision
abotit
whether
to
be
proactive
and
externally
ori
ented
in
addressing
the
disruption
or
to
become
much
more
insular,
in
an
effort
to
shut
out
external
perturbations.2
These
two
options,
on
their
face,
lead
to
competing expectations. If
a
manager
chooses
to
connect
with
other
actors,
we
could
expect
more
external
networking
and
more
build
ing of
collaboration,
whereas
a
decision
to
buffer
and
insulate
would
result
in
few-er
collaborative
relationships
in
an
effort
to
protect
the
organiza
don
(for
recent
investigations
of
buffering
and
internal
protective
responses
by
managers,
see
Meier
and
O’Toole
2008,
forthcoming).
One
question,
therefore,
is
whether
wicked-problem
shocks
in
natural
disaster
settings
stimulate
or
inhibit
collaboration.
Given
the
preponderance
of
the
theo
retical
argtiments,
our
expectation
is in
the
former direction, particularly
given
sufficient
organizational
capacity:
Hypothesis
1:
Controlling
for
organizational
capacity,
districts
which
receive
a
larger
number
of
evacuees
(as
a
proportion
ofthe
regular
stu
dent
body)
will
be likely
to
engage
in
more
collaborative
relaiionships.
Another issue
has
to
do
with
managenal
patterns
of
interaction—
networking—as
such
patterns
estahhsh themselves
prior
to
an
unexpected
crisis
period.
Do
these
shape
collaborative
interorganizational
arrays
dur
ing
the stressful
postdisaster
period?
NcR
orking
can
include
a
variety
of
managerial
functions, including
efforts
to
form
longer-term
cooperative
relationships
and
to
block
potentially
threatening
influences
(O’Toole
and
Meier
1999;
Khjn
2005;
Meier and
O’Toole
-2008).
These
options suggest
that managers
could
be
working
with
other
organizations
either
to
lever
age
resources
and
support
or
to
sort
out
jurisdiction
and
responsibility
to
avoid
being
overwhelmed
by
events.
MANAGERIAL
NETWORKING
Examining
collaboration
in
response
to
an
organizational
shock
offers
the
opportunity
to
ask
questions
about the
nature
of
managerial
networking.
In
particular,
it
is
possible
to
consider
how
patterns
of
managerial
net
working
may
be
related
to
interorganizational
collaboration
in
response
to
disasters.
How
might
externally
oriented managerial
behavior,
pre
dIsaster,
be
related
to
the
exLont
of
organizational collaboration achieved,
postdisaster?
Three
somewhat
simplified
alternative
possibilities can
be
sketched:
(1)
Managers
network
to
deal
with
particular
problems,
so
pat
terns
of
networking
behavior prior
to
an
unexpected
disaster
should
be
essentially
unrelated
to
the
development
of
organizational
collaborations.
postthsaster.
(2)
Managers
networkefficiently,
so
they
can
be
expected
to
build
on
their
extant
interactions
selectively
in
response
to
a
disaster,
thus
minimizing
transaction
costs
necessitated
when
building
interorganiza
tional
ties
de
novo. (3)
Managers
develop
an
external
networking
style
or
habit
of
behavior,
and
the general
level
of
networking
activity
predisaster,
should
thus
be
related
to
the
extent
of
interorganizational
collaborations
developed
to
deal
with
unexpected
environmental
shocks,
postdisaster.
Each
pnssibility
leads
to
an
hypothesis.
We
shall
first
sketch
the
causal
logies in
a
bit
more detail
and
then
outline
the
corresponding
three
addi
tional
hypotheses.
First,
as
Kaufinan (1985)
suggests,
networking
could
be
a
problem-
F
specific
response,
one
in
which
managers
network
only
when
triggered
to
do
so
by
the
problem-solving
requirements
immediately
at
hand.
If
net
working
is
largely
a
problem-specific
managerial
behavior,
there
should
belittle
or
no
relationship
between
past
levels
of
networking
and
eollabo
rative
efforts
following
an
environmental
shock, when
controlling
for
the
extent
of
the
shockitself
and
the
extent
of
organizational
capacity
present
in
the system.
Thus:
Hypothesis
2:
The
level
of
managerial
networking
activity
in
noncrisis
times
will
be
unrelated
to
the
extent
of
interorganizational
collabo
ration
during
crisis
periods
A
second
possibility
is
suggested
by
some
of
the
literature
on
the
emer
gence
and
maintenance
of
networks,
which
points
to
the
large
transac
tion
costs
involved
in
setting
up
collaborative
partnerships
(Agranoff
and
MeGuire
2003a;
Bardach
1998;
Khjn
2005).
Here,
networks
are
viewed
as
complex
relationships that
take
considerable
time
and
effort
to
build
and
maintain.
This
logic
would lead
to
the
expectation
that
managers
will
be
more
likely
to
build
collaborations
in
response
to
an
organizational shock
in
instances
for
which
the
transaction
costs
are
relatively
low.
Stated
dii
ferently,
managers
who
network
in
noncrisis
times
will
be
more
likely
to
build
interunit
collaborations
with
the same
interaction partners
in
re
sponse
to
organizational
shocks
because
working
with
extant
relation
ships
lowers
the
transaction
costs
incurred
when
building
and
tapping
collaborative
relationships.
This
possibility
will
be
evaluated
by
testing
the
following:

iu’*
vvny
ruOlic
Managers
Collaborate
Calming
the
Storms
105
Hypothesis
3:
Collaborations
are more
likely
in
response
to
disasters
in
instances
for
which
managers
have
developed
a
history
of
inter
action
with
these
actors/organizations.
Finally,
a
third
possible
explanation
is
that
managerial
networking
could
be
a
fundamental (learned
or
innate)
part
of
an
individual’s
mana
gerial
style;
ffso,
levels
of
networking
will
be
relatively stable;
and
inter-
organizational
collaborations
to
deal
with
environmental
shocks
are
more
likely
to
be
developed,
in
general,
if
top
managers’
established
network
ing
style
involves
more
activity
and
involvement
externally.
Previous
work
provides some
evidence
in
support
of
this
expectation
(Meier
and
O’Toole
2005),
because
a
manager’s
level
of
networking
in
one year
is
found
to
be
a
significant
predictor
of
his
or
her
level
of
networking
at
a
later
time.
If
networking
is
stable across
Lime,
it
could
also
be
a
stable
component
of
management
strategy’
to
build
collaboration
in
both
crisis
and noncrisis
times,
Managers
who
networkin
noncrisis
times
will
be
more
likely
to
build
collaboration
in
response
to
organizational
shocks,
but this
networking
activity
will
not
necessarily
beJbcuscd
on
those
orqanizutions
with which
they have
established
relationships.
Formally
stated:
Hypothesis
4:
Organizational collaboration
with
others,
including
particular
actors/organimtions,
will
be
a
function
of
the
top
managers
overall
propensity
to
engage
in
networking,
not
a
function
of
a
his-
j
ton’
of
collaboration
with
that
particular
actor/organization.
DATA
AND
METHODS
Some
of
the
data
for
this
chapter
are
drawn
1mm
two
sun’evs
of Texas
superintendents.
In
late
2005
and
early
‘2006,
after
the
hurricanes,
we
administered
the
Survey
of
Emergency
Preparedness
and
the
Impact
of
Hurricanes
Katrina
and
Rita
on
Texas
Public School
Districts. This
ques
tionnaire,
described
briefly
earlier
in
the
chapter,
collected
data
from
school
district
superintendents
throughout
Texas
on
how
the
hurricane
evacuees
alfected
their
school
districts,
including
the
extent
of
the
im
pact,
the
naturc
of
the
district’s
response,
the
patterns
of
collaboration
in
response, and
how
the
hurricanes
affected
the district’s
own
emergency
planning.
The
data
from
this
postliurricane
survey
have
been
con]bined
with
data
from
an
earlier,
prehurricane
survey
of
Texas
public school
district
supenntcndents
that
we
administered
in
Januarv
2005.
This
prchurdcane
survey,
one
in a
series
of
several
implemented
regularly
starting
in
2000
by
Meier
and
O’Toole,
collected
data
on
managenal
strategies
and
behav
ior
of
supeLintendents,
with
an
emphasis
on
their
networking
activity.
The
districts
ranged
widely on
avatiety
of
dimensions,
includingstudent
com
position
(race,
ethnicity,
etc.),
resources,
setting
(urban,
rural,
suburban),
r
and
performance.
The
response rate
for
the
prehurricanc
survey
was
58.0
percent
(N
=
729).
In
combining
the
data
from
the
prehurricane
and
[
posthurricane
surveys,
we
had
considerable
overlap,
with
450
districts
F
responding
to
both
suneys.
MI
nonsurvcv
data
were
drawii
from
the Texas
[
Education
Agency.
Our
analyses
were
aimed
at
two
general
objectives: (1)
seeing
whether
the
size
of
the
environmental
shock
helps
to
explain
the
extent
ofcollabo
ration
developed
in
school
districts
follow’ing
the
arrival of
evacuees,
while
controlling
for
the organizational
capacity
of
tile
district; and
(2)
deter
mining which
of
the
several
possible
causal
relationships
between
earlier
managerial
networking
and
postdisaster
school
district
collaboration
seems
[
to
be
supported
by
the
evidence.
For the
second
objective,
the
availability
of
data from
the
two
surveys
provided
an
unusual
opportunity
to
execute
a
natural-experiment
design.
Dependent
Variables
The
dependent
vanables
for
this study
measure
the
extent
to
which—and,
for
sonic
of
the
analyses,
whether—school districts
collaborated
with
other
organizations
as
a
part
of
their
efforts
to
respond
to
the
influx
of
displaced
students
in
their districts.
The
top
managers
were
asked
which
of
the
fol
lowing
t;pes
of
organizations
they
collaborated
with
to
provide
for
displaced
students:
police,
fire,
and
first responder’s;
nonprofit
and
relief
organiza
[ions;
other
school
districts;
government
relief
and
welfare
organizations;
business
organizations;
and
local,
community,
and
religious
organizations.
The
primary
dependent
variable
is
the
total
number
of
types
of
organiza
tions
that
the
school
district
worked with
in
response
efforts,
ranging
from
o
to
6.
Because
1-lypotheses
3
and
4
explore
the
extent
to
which
superinten
dents
were
strategic
(or
influenced
by
transaction
costs)
in
choosing
col
laborative
partners,
we
examined
in
particular
links
with
two
individual
types
of
nodes
in
the
environment
of
the
school
districts:
other
school
districts and
business
organizations.
These
were
chosen
for
attention
because
they
represent
actors,
the
interaction
with
whom
at
the individual
level
(i.e.,
with
other
supenntendents
and
local
business
leaders,
respec
tively)
the
superintendents
had
also
been
surveyed
about
in
the
prehur
ricane
period.
Each
of
these
dependent
variables
is
dichotomous,
with
a
a
II
I
I

Calming
the
Storms
107
106
Why
Public
Managers
Collaborate
“1”
representing
when
superintendents
collaborated
with
that
particular
group.
Independent
Variables
Two
variables were
used
in
all
the models
used
to
test
the
hypotheses.
First,
we
constructed
a
variable
to
represent
the
size
of
the
shock
to
the
organization. The
measure
is
the
number
of
displaced
students
absorbed
by
the
district,
as
reported
on
the
posthunicane
survey, divided
b’s’
the
total
enrollment
prior
to
the
hurricanes
and multiplied
by
100.
This
variable,
‘total
evacuees,”
represents
the
amount
of’
students
absorbed,
as
a
per
centage
of
the
regular
student
body.
It
was
used
in
particular
to
test
Fly
pothesis
I
(that
the
size
of
a
shock,
or
problem
severity,
leads
to
more
collaboration).
In
the
first
and
all
other
estimations
reported
here,
we
also
controlled
for
the
overall
size
of
the
school
district,
because
many larger
districts
have
greater
administrative
capacity
than
small
districts.
To
control
for
size,
we
included
the
logged
enrollment
of
the
district.
To
test
whether
collaboration
in
response
to
the
influx
of
evacuees
is
unrelated
to
the general
level of
networking
behavior
of
the
top
man
agers
during
more
stable
and
routine
times
(Hypothesis 2),
we
devel
oped
from
the
preliurricane
survey
a
measure
of
managerial
networking
prior
to
the
onset
of
the
unanticipated
shocks
to
the
school
districts.
We
followed
earlier
work
by
Meier
and
O’Toole
(e.g,
2001)
and
asked
re
spondents
to
report,
on
a
6-point
scale
ranging
from
daily
to
never,
how
often
they
interacted
with
each
of
several
external
actors..
In
the
pre
hurricane
survey,
we
asked
about
interactions
with
seven
external
par
ties:
teacher
associations,
parent
groups,
local
business
leaders,
other
superintendents,
federal
education
officials,
state
legislators,
and
the
Texas
Education
Agency.
A
composite
managerial
networking
scale
was
created
using
factor
analysis.
All
four
items
loaded
positively
on
the
first
factor,
producing
an
eigenvalue
of
1,76;
no
other
factors
were
statisti
cally
signilicant.
Factor
scores
from
this
analysis
were
then
used
as
a
measure
of
managerial
networking,
with
higher
scores
indicating
a
greater
networking
orientation.
Because both
surveys
asked
about
interactions
with
two
particular
external
actors—business
leaders/organizations
and
other
superinten
dents/school
districts—we explored
Hypotheses
3
and
1
by
compar
inginteractions
with
each
group
before
and after
the
hurricane-induced
displacement.3
Descriptive
statistics
for
these
variables
are
displayed
in
table
6.1.
Table
6.1
Descriptive
Statistics
Period
and
Measnre
N
Mean Mw.
Max.
After
Katrina
Total
collaboration
Collaboration
with
other
school
districts
Collaboration
with
business organizations
Evacuees
as
a
percentage
of
previous
enrollment
560
0.793
0
7.10
Logged
enrollment
560
7190
3.912
11.969
Before
Katrina
Total
networking
405
—0.0005
—1.804
2.827
Netvor[dng
with
other
supenutendents
420
3.911
2
6
Nenvorking
with local
business
leaders
423
3.783
1
6
METHODS
1*
Because
the
primary
dependent
variable—collaboration
to
assist
with
the
[
evacuees
and
their
challenges—is ordinal,
we
used
multiple
estimators
in
the
analysis.
First,
the
models were
run using
ordinary
least
squares
(OLS)
analysis;
then
they
were
estimated
as
ordered
logits;
and
finally
they
were
analyzed
via
poisson
regression,
which
is
considered
more
appropri
ate
for
this
type
of
dependent
variable
(Gujarati 2003).
Although
OLS
is
not
the
most
appropriate
estimator,
the
results
were
yen’
similar
across
the
estimators,
and
the
OLS
coefficients
are most
easily
interpretable.
The
models
evaluating
collaboration
with
individual nodes
were
estimated
as
logistic
regressions.
Post-estimation
diagnostics
showed
no
problematic
heteorskedasticity
or
multicollineantv.
FINDINGS
Our
first
hypothesis.
that
managers
will
engage
in
higher
levels
of
collabo
ration
when
faced
with
larger organizational
shocks,
is
supported
by
the
analysis.
Table
6.2
presents
the
three
different
models
(OLS,
ordered
logit,
Poisson
regression),
all
of
which
show
that
the
number
of
evacuees
taken
in
by
the
district
as
a
proportion
of
baseline
enrollment
is
a
significant
predictor
of
the
extent
of
totaL
collaboration.
Still,
although the
size
ofshock
(evacuees)
is
always
significant,
it
is
never
substantively
large.
On
the
basis
of
the
OLS
coelilcients,
a
school
would
have
to
take
on
enough
evacuees
to
constitute
roughlv3.5 percent
of
its
student
population
to
move
the
level
of
collaboration
up
one
unit.
The
size
of
the
shock
helps
to
shape the
extent
I
4
508
2.848
509
0.497
509
0.2-14
06
0
1
0
1

108 Why
Public
Managers
Collaborate
Table
6.2
Collaboration
in
Response
to
Influx
of
Displaced
Students; Dependent
Variable:
Extent
of
Collaboration
(0
to
6)
independent
Variable
OLS
Ordered
Logil
Poisson
Number
of
evacuees
0.276
0.287 0.077
(379)a
(3.57) (3.36)
District
size
0,477 0.510 0.160
(9.28)
(8.63)
(9.05)
Constant
—0.850 —0.222
(2.31)
(1.67)
N
508
508
508
112
0.21
Pseudo
fl2
0.06 0.06
of
collaboration,
built
is
only
part
of
the explanation.
Note, as
well,
that
the
sheer
size
of
the
district,
reflecting
the
capacity
of
the
organization
to
engage
in
various
forms
of
collaborative
activity,
also
contributes
to
the
extent
of
die
result.
For
a
more
complete
test
of
whether
the
size
of
the
shock
is
driving
collaboration,
and
also
to
explore
the possible
influence
of
an
earlier
pat
tern
of
managerial
networking,
we
also
tested
Flypothesis
2,
framed
as
the
null
hypothesis:
thai
a
pattern
of
managerial networking
during
normal
times
does
nothing
to
drive
collaborative
results
when
serious
problems
arise.
To
examine
this
relationship,
we
included
in
the
specification the
measure
of
the
superintendent’s
overall
networking score
as
tapped
prior
to
the hurricanes.4
If
nets
orkngwere
purely
a
problem-specific
response,
we
would
expect
that
the
number
of
evacuees
would
significantly
predict
collaboration
but
that
prior
networking
would
have
no effect.
The
models
given
in
table
6.3
provide
evidence
to
rebut
Hypoihesis
2—thai
collabora
tion
during
crises
is
not
shaped
by
prior
behavioral
patterns
developed,
or
manifested,
in
more stable periods.
The
superintendent’s
overall
level
of
networking
prior
to
the
hurricanes
is
a
significant
predictor
in
all
the
models.
Note
also
that
the
size
of
the
shock
to
the organization,
as
mea
sured
by
the
relative
size
of
the
influx
of
the
pool
of
evacuees,
is
still
a
significant
predictor
of
collaboration.5
In
fact,
the
coefficients
for
the
number
of
evacuees
are
relatively
stable
between
tables
6.2
and
6.3.
We
can
conclude
that
an
established
general
pattern
of
managerial
network-
Calming
the
Storms
109
Table
6.3
Does
onrisis
Networking
Predict
Collaboration
in
Crisis
Times?
Dependent
Variable:
Extent
of’
Collaboration
(0
to
6)
Independent
Variable
OLS
Ordered
Logtt
Pin
SSOn
Networking
0.516 0.156
(4.08)
(4.27) (4.00)
NumberofevaCuecs
0.273
0.300
0.077
(3.14)
(3.13) (2.76)
District
size
0.441
0.474 0.147
(7.29)
(6.85)
(7.09)
—0.148
(0.94)
370
370
006
0.06
Pseudo
R
ingis
clearly
not
the
only
influence
on
postdisaster
collaboration,
but
also
that
superintendents
with
a
history
of
such
interaction
externally
are
more
likely
to
engage
in
higher
levels of
collaboration
in
response
to
the
organi
zational
shock.
What
explains
this
relationship?
The
idea
that
collaborative
relation
ships
would
have
some
element
of
stability
has
been
discussed
in
much
of
the
work
on
interorganizational
relations.
Researchers
often
argue
that
such
interunit
stability
derives
from
the
costs
involved
in
the
time
[
consuming
process
of
establishing
the
relationship,
formalizing
processes
for
shared
decision
making,
and
other
similar
tasks. If
this
explanation
were
to
be
valid,
we
would
expect
that
superintendents
would
turn
to
the
same
external organizations
and
organizational
representatives
that
are
a
part
of
their
developed set
of
relationships
when
they
seek
support
for
han
dling
complicated challenges.
We
tested
for
this
relationship,
as
outlined
in
I-Ivpothesis
3,
with
the
models
displayed
in
table
6.4.
We
analyzed
whether
predisaster
interactions
between
superinten
dents
and
two
other
external
nodes—other
school
distnct superintendents
and
also
members
of
the
business
community—help
to
explain
posthur
ricanc
collaboration
with
other
school
districts and
business
organizations,
respectively.
The
idea
is
to
distinguish the
influence
of
a
habit
or
pattern
of
general
interaction
externally
on
the
part
of
top
managers,
on
the
one
hand,
&om
a
history
of
nodc.specific
interactions
and
exchanges,
on
the
other.
If,
over
time,
networking
is
driven
by
stable
patterns
olinteractions
.4
I
Tfte
I-scores
are included
for
OLS
coellicients.
Z-scores
are
reported
for
the
ordered
Iigits
and
Poisson
estimates.
Constant
N
0.630
(1.44)
370
112
0.22

110
Why
Public
Managers
Collaborate
Calming
the
Storms
111
Table
6.4
Are
Managers
Strategic
in
Collaboration
to
Lower
Transaction
Costs?
Dependent
Variable:
Collaboration
with
Business
Organizations!
Other
School
Districts
Business
Other
School
Independent
Variable
Organizations
Districts
Previous
networking
with
0175 0120
business
leaders/other
superintendents
(1.34)
(1.02)
Number
of
evacuees
0.341
0.265
(2.70) (2.32)
District
size
0.584
0.192
(5.88) (2.51)
Constant
—6.679
—2.160
(7.31) (2.72)
\!
383
383
Pseudo
112
0.15
0.03
in
individual
relationships,
we
would
expect
that
superintendents
who
regularly
interact
with
lbr
example,
business
leaders
would
be
more
likely
to
turn
to
those
business leaders
and
their
firms
during
crisis times.
The
results
presented
in
table
6.4
do
not
support
this hypothesis.
In
both
the
models,
more
interaction
with
specific
nodes
prior
to
the
hurricanes
ex
plained
none
of
the
valiance
in
collaboration
with
these
nodes after
the
evacuees
had
arrived.
We move
on
to
the
test
of
Hypothesis
4.
We
investigated
the
possibil
ity
that
the
extent
of’
collaboration
with
particular
external
organizations
in
the
wake of
disaster
is
not
about
building
on
preedsting
relationships
but
is,
rather,
partially
shaped
by
managers’ general
styles
of
managing
outward.
To
test
this
hypothesis,
we
added
the general
networking
inca-
sure
to
the
equations
from
the
preceding
analysis
predicting
interaction
with
the
individual
nodes.
Table
6.5
presents
the
results,
which
lend
some
support
to
I-lvpotlwsis
4.
The
logit
model
estimating
collaboration
with
business
organizations
does
not
find
previous
interaction
with
local
busi
ness
leaders
to
be
a
significant
predieawr
of
collaboration,
but
it
does show
managers’
earlier
level
of
overall
networking
to
be
related
to
the
likelihood
of
collaboration
with
such
organizations.
In
a
similar
model seeking
to
predict
collaboration
with
other
school
districts,
prior node-specific
inter
actions
are
unrelated
to
collaboration;
general
networking
style
has
a
posi
tive
direction,
although
the
relationship
to
posthurzieane
collaboration
with
other
school
distncts
is
significant
only
at
the
.10
level
in
a
two-tailed
test.
N
Pseudo
U2
0.18
0.04
CONCLUSIONS
The
results
reported
in
the
preceding
section
derive
from
one
rather
un
usual
set
occircumstances_schol
districts
managing
unexpected
influxes
of
high-need
students
on
short
notice.
Considerable
caution
needs
to
be
exercised,
therefore,
before
the
findings
are
treated
as
generalizable
to
management
involving
other
kinds
of
natural
disasters,
or
to
management
challenges
of
collaboration
more
generally,
because
a
number
of
other
factors
have
‘‘et
to
be
explored.
In
particular,
the
findings
on
collabora
tion
here
are
necessarily
focused
on
the
relatively
short
term;
how
long
such
patterns
are
hkey
to
persist
remains
an
unanswered
question.
Ad
ditionally,
the
collaboration
studied
here
was
largely
voluntary;
managers
chose when
to
collaborate
and
i;ith
whom
they
would
collaborate.
In
situ
ations
where
collaboration
might
be
forced
(possibly
by
the
federal
gov
ernment),
we
may see
very
different
patterns.
For
all
these reasons,
the
results
of
this
study
constitute
a
beginning
for
analysis
rather
than
a
real
conclusion.
The
levels
of
explanation
for
the
extent
of
collaboration
are
relatively
modest.
Further
analysis
is
needed
to
explore the
other
factors
that
contribute
to
why
managers
in
similar
organizations
react
differently
to
the
same
intervention.
And
how
much
difference
such
collaborations
make
in
terms
of
performance
results
is
a
Table
6.5
Is
Collaboration
Influenced
by
Pre-Existing
Relationships
or
Managerial
Style?
Business
Other
School
Independent
Variable
organizatwns
Districts
Composite
networking
0.681
0.297
(3.10) (1.69)
Previous
networking
with
-0.126
-0.033
business
leaders!
other
superintendents
(0.75)
(0.22)
Number
of
evacuees
0.353
0.282
(2.73) (2.43)
Distret
size
0.598 0.163
(5.89)
(2.01)
Constant
—3.679
—1.375
(5.85)
(1.49)
371
371

112
Why
Public
Managers
Collaborate
key
and
thus
lär
unexplored
issue.
We
know
that
these
organizations
faced
a
considerable disruption,
but
we
do
not
know
whether
this
disruption
affected
their
ability
to
early
on
their
core
functions.
Did
the
influx
of
hurricane
evacuees
affect
student
outcomes?
If
so,
could
increased
col
laboration
help
to
share
the
burden
of
the
response and
possibly
lessen
the
impact
on
student
outcomes?
Still,
a
number
of
the findings
from
this analysis
are
interesting
and
instructive,
Organizational
capacity
clearly
makes
a
difference
in
the
ability
to
develop
collaboration
in
multiple
directions.
This
finding
is
unsurprising
but
also
important.
The
size
of
the
unexpected
shock
to
the
organizational
system
also
matters—positively—as
a
stimulus
or
prod
toward
the
devel
opment
of
collaboration.
Wicked-problem
stimuli
trigger
increased inter-
organizational
collaboraiion
for
the
school
districts
dealing
with
many
issues
in
the
wake
of
major
hurricanes.
In
a
net
sense,
at
least,
efforts
to
reach
out
to
partner
with
others trumps
any
temptation
to
hunker
down,
organizationally
and
managerially
speaking. This
finding
is
encouraging
for
those
interested
in
whether
public
management
is
likely
to
be
respon
sive
to
the
increased
challenges
posed
by
complex
issues,
even
Wit
sug
gests
that
pubhc
managers
and
their
organizations
may
have
to
do
some
rather
heay
lifting
to
address
their
responsibilities.
Most
interestingly,
the
findings
given
in
this
chapter
demonstrate
that
public
management matters
for
collaboration.
They
also
provide
some
evidence
regarding
how
management
makes
that
diflérence.
Controlling
for
organizational
capacity
and
the
size
of
a
shock
to
an
organizational
system,
a
top
manager’s
established
style
of
externally
oriented
inter
action
helps
to
explain the
extent
of
interorganizational
collaboration
developed
after
an
unexpected
disaster. Intriguingly,
the
node-specific
interaction
histories
seem
rather
unimportant
in
this
regard,
particularly
when
compared
with
the
general
style
or
habit
of
managerial
networking.
The
combination
of
these
findings
could
speak
to
the
yen’
nature
of
collaboration
and networking. Much
of
the
research
on
collaboration
and
networking
follows
one
of
two
streams.
Collaboration
is
either
studied
as
a
structural
issue
(with
the
study
of
a
‘network”),
or
the
focus
is
on
the
more
behavioral
aspects
of
collaboration
(with
the
study
of
the
manager’s
networking
activity).
Here,
we
find
that
collaboration
is
less
a
function
of
stable
relationships,
and
thus
structural
ties,
and
more
a
product
of
the
individual-level decisions
of
the manager.
We
need
more
research
on
the
extent
to
which
interorganizational
management
is
an
organizational-
level
or
an
individual-level
concept, because these differences
may
af
fect
some
of
the
fundamental
assumptions
made
about
how
organizations
work
together.
[
Calming
the
Storms 113
This finding
also
suggests
that
such
developed
patterns
of
network
ing
not
only
contribute
in
the
short
term
to
performance,
as
earlier
re
search
has
demonstrated
(e.g.,
Meier
and
O’Toole
2001,2003),
but
also
constitute
a
sort
of
investment—a
social-networking
capital,
as
it
were—
that
can
pay
dividends
on
collaboration
in
the
future,
and
in
particu
lar
during
unexpected
crisis
periods.
In
fact,
given
that
administrative
[
systems
are
typically highly
inertial, managerial
networking seems
to
con-
tribute
to
results
in
at
least
three
ways:
short-term
performance
improve
ments,
enhancement
of
the
base
over
time
and
thus
gradual
amplification
of
the
impact
of
networking
over
time,
and
also
estabhshment
of
social
networking capital that
can
be
drawn
on
in
times
of
need
or
to
help
manage
significant
shocks.
All
in
all,
therefore, the
contribution
of
man
agers
to
performance
and
to
networked
collaboration
is
a
topic
that
deserves
considerably more careful
attention
both
theoretically
and
em
pirically.
Further,
if
validated elsewhere,
this
set
of
findings
carries
sig
nificant implications
for
how
practicing
public
managers
might
spend
their
time and
devote
their
attention,
particularly
those
operating
in
systems
that
are
likely
to
be
subjected
to
sizable
and
unpredictable
shocks
from
the
environment.
An
additional
finding
of
interest,
imphcit
in
the
empirical
results
reported
in
this
chapter,
is
that
individual-level
patterns
of
behavior
(managerial
networking)
can
have
organizational
consequences
(inter-
unit collaboration). Though
top
managers
can
he
expected
to
he
rather
influential
in
their
mvii
organizations,
it
is
nevertheless
interesting
to
see
clear
relationships
between
these
two
levels
Whether
the
collabo
rations
in
question
develop
from
the
leadership
and
direct
individual
efforts
of
the
managers
themselves,
or
whether
others
in
the
organiza
tion
observe
the managerial
behavior and
are
thus
stimulated
to
mimic
these
externally
oriented
patterns,
or
whether
perhaps
externally
ori
ented managers
also
invest in
building organizational
processes
and
even
specialized
subunits
to
help
broker
the
development
of
more
formalized
collaboration
is
an
interesting
question—but
one
that cannot
be
answered
with
the
data
at
hand.
What
can
be
said
is
that
these
organizational
systems
are
stimulated
to
initiate
collaboration
when
they
expenence
an
unexpected
and
signifi
cant
wicked-problem
shock,
and
the
presence
of
a
manager
who
chooses
to
engage
in
networking during
normal
times—a
truly
collaborative
pub
lic
manager—also
contnbutes
to
how
these
organizations
respond.
It
may
be
only mildly
consoling
under
the
circumstances,
given
the
massive
costs
borne
in
the
wake
of
Hurricane
Kattina and
Hurricane
Rita,
but
it
is
true
nonetheless:
In yet
another
way,
public
management
matters.
I
I
I

114
Why
Public
Managers
Collaborate
NOTES
1.
The
present
chaptea-
explores
public
management
and
its
links
to
collabora
non.
Additional
work
is
under
way
analyzing
the
relationship
between
inter-
organizational
collaboration
and
performance.
2.
Elsewhere
(e.g.,
in
O’Toole
and
Meier
1999;
we
have
referred
to
these
two
options
in
terms
of
exploiting
environmental
opportunities
or
managerial
buffer
ing
against
environmental
shocks.
In
the
terms
of
the
formal model
we
have
been
working with
for
the
past
several
years,
these
are
[lie
M3
and
[he
“M4”
func
tions,
respectively.
3.
The details
of
the
survey
questions
differed
somewhat
between
prehurricanc
and
posthurricane
surveys.
In
the
former
case,
we
asked
about
the
top
manag
ers’
znteraclzons
with
a
range
ofactors
in
the
environment.
In
the
latter
instance,
we
asked
about
the
collaborations
between the
school
district
and
a
range
of
tspes
of
exiernal
organizations..
The
former-,
therefore,
taps
individual
behavior,
while
the
laiter
has
to
do
with
interorganizational
links.
The
two
should
be
related,
but
because
the
items
ask
for
different
thform;ition,
we
can
expect some
attenuation
in
any
connection
between
the
two
sets
of
responses.
Nonetheless, the
possible
link
between
patterns
of
managerial
behavior and
patterns
of
organizational
col
laboration
‘san
interesting
empirical
question.
4.
For
this
and
subsequent
models,
the
number
of
cases
is
modestly
lower.
These
analyses
include
only
those
school
districts
in
which
superintendents
re
sponded
to
both
surveys.
5.
District
size
also
remains
signilicani,
as it
does
in
all
models
reported here.
Part
II
HOW
PUBLIC
MANAGERS
COLLABORATE
I-
[
How
do
public
managers
collaborate?
The
mechanisms
of
collaboration
arc as
varied
as
the
public
managers
who
do
the
collaboration. Phvatization
is
one
of
the
managerial
tools
at
the
disposal
of
public
managers.
Con
tracting
out in
whole
or
in
part
is
another
tool.
Working
side
by
side
with
the
public
is
another
form of
collaboration.
Each
chapter
in
part
II
exam
ines
how
public
managers
collaborate
through
different
lenses.
How
the
tool of
privatization
is
used
to
navigate
complicated
networks
[
of
service provision
that
require
coflaborative
public
management—both
thtersectorally
with
nonprofit
organizations.,for-profit
businesses,
and
other
public
agencies,
and
intergovemmentallr
with
federal,
state,
and
local
offi
cials
and
institutions—is
the essence
of collaborative public
management,
and
is
the
subject
of
chapter
7,
by
Jeffrey
Brudney,
Chung-Lae
Cho,
and
Deil
Vright.
Their
work
finds
that state
agency
heads
are
extensively
con
nected
through
the
contracting
process
with
mazes
of networks
in
pursuit
of
public
service
delivery.
They
document
that
in
1998
and
-2004,
state
agen
cies
contracted
for
service
deliveiw
with
other
governments,
nonprofits,
and
private firms
at
rates
approximating,
respectively,
60, 70,
and
80
percent.
They
develop
a
summary
statistic
measuring
agency
contracting
perfor
manee
based
on
combinations
of
two
key
factors that are
part
of
agency
contracting:
the cost
of
delivering
services
to
the
public
and
the
quality
of
the
services
delivered.
They
also
directly
explore
the
relationship
between
F
collaboration
and
agency
contracting
and
find
that
state agencies
that
rated
more
highly
on
measures
of
collaboration
are
more
likely
to
invest
a
greater
percentage
of
their
budget
in
contracting.
In
chapter
8,
David
Van
Slyke
examines
the
pivotal
role
of
contract
ing
in
collaborative public
management
through
a
different
lens:
relational