American Policyhol v. Nyacol ( 1993 )

                          FOR THE FIRST CIRCUIT
    No. 92-1949
                          Plaintiff, Appellant,
                      NYACOL PRODUCTS, INC., ET AL.,
                          Defendants, Appellees.
                               ERRATA SHEET
         The opinion of  the Court  issued on February  24, 1993,  is
    corrected as follows:
         On page 16, line 25   insert semicolon after "579"
         On page 18, line 15   change "support" to "supports"
                          For The First Circuit
    No. 92-1949
                          Plaintiff, Appellant,
                      NYACOL PRODUCTS, INC., ET AL.,
                          Defendants, Appellees.
               [Hon. William G. Young, U.S. District Judge]
                          Selya, Circuit Judge,
                      Coffin, Senior Circuit Judge,
                        and Stahl, Circuit Judge.
         Jennifer  S.D.  Roberts,  with  whom  Rackemann,   Sawyer  &
    Brewster, P.C. was on brief, for appellant.
         Catherine  M. Flanagan,  Attorney,  Department  of  Justice,
    Environment  & Natural  Resources  Division, with  whom Vicki  A.
    O'Meara, Acting Assistant Attorney General, and David C. Shilton,
    Attorney,  Environment  &  Natural  Resources Division,  were  on
    brief,   for   Julie   Belaga,   Regional   Administrator,   U.S.
    Environmental Protection Agency, appellee.
                            February 24, 1993
              SELYA, Circuit Judge.  This appeal, which arises out of
              SELYA, Circuit Judge.
    an insurance company's efforts to secure a binding declaration of
    its  rights  and responsibilities  under  an  insurance contract,
    poses  an  intriguing  question:    does  the  "officer  removal"
    statute, 28 U.S.C.   1442(a)(1), permit a  federal official, sued
    only  in her  representative  capacity, to  remove  an action  to
    federal district court?   Because we think that the  statute does
    not confer such a right, and because there is no other cognizable
    basis  for  federal  jurisdiction,  we return  the  case  to  the
    district court with instructions  that it be remitted to  a state
              From  1917  until  1977,  a  succession  of  dye-houses
    occupied  a thirty-five  acre  plot  in  Ashland,  Massachusetts.
    During  these six  decades,  toxic wastes  impregnated the  site.
    Eventually,  the  United States  Environmental  Protection Agency
    (EPA)  and   the   Massachusetts  Department   of   Environmental
    Protection (DEP)  discovered  the pollution  and  documented  its
    source  in the dye manufacturing  processes.  In  early 1982, EPA
    notified Nyacol Products, Inc.  (Nyacol), a producer of colloidal
    silicas at  a portion  of the site,  that it considered  Nyacol a
    potentially  responsible  party  (PRP)  under  the  Comprehensive
    Environmental Response, Compensation, and Liability Act (CERCLA),
    42 U.S.C.    9601-9626, 9651-9662, 9671-9675.1
         1EPA also branded two of Nyacol's officers, Robert Lurie and
    Thomas L.  O'Connor, as PRPs.   Lurie and  O'Connor are named  as
    defendants in the  instant suit.  For ease in reference, we refer
              Pursuant to Massachusetts law,  which deems PRP notices
    equivalent  to  law  suits  for  the  purpose  of  triggering  an
    insurer's  duty to defend, see  Hazen Paper Co.  v. United States
    Fid. & Guar. Co., 
    555 N.E.2d 576
    , 581 (Mass. 1990), Nyacol called
    upon   its   insurance   carrier,  plaintiff-appellant   American
    Policyholders  Insurance Company (American), to defend it against
    EPA's claims and  to indemnify  it for loss,  costs, damages,  or
    other expense related thereto.   American provisionally undertook
    the company's  defense under  its general  liability policy.   In
    December of 1991, American brought suit in a Massachusetts  state
    court seeking a declaration  that it had no obligation  to defend
    or recompense its insureds.   In addition to naming  the insureds
    as parties  defendant,  American  joined  two  other  defendants:
    Julie Belaga,  in her  capacity as EPA'S  Regional Administrator,
    and Daniel S. Greenbaum,  in his capacity as Commissioner  of the
    DEP.    Invoking  the  officer  removal   statute,  28  U.S.C.   
    1442(a)(1),Belaga transferredthe actionto federaldistrict court.2
    to the  company and  the individual defendants,  collectively, as
    "Nyacol"  or "the  insureds."  We  note, moreover,  that although
    EPA,  invoking a  theory of  successor liability,  notified these
    three defendants  that they  might be  liable for  EPA's response
    costs  (past and future), as  well as for  cleanup costs, EPA has
    not yet sued  to compel payment of these expenses or otherwise to
    enforce its claimed rights.
         2The officer removal statute provides in pertinent part:
                   A civil action  or criminal  prosecution
              commenced in a State court against any of the
              following persons may be  removed by them  to
              the  district court of  the United States for
              the district and division embracing the place
              wherein it is pending:
              On  February  20, 1992,  DEP agreed  to  be bound  by a
    declaration  of  rights  in  American's lawsuit  insofar  as  the
    judgment resolves insurance coverage  issues.  The case proceeded
    against Belaga and the  insureds.  On May 12,  1992, the district
    court granted Belaga's motion to  dismiss, reasoning that a  suit
    against  her, in her official capacity, was really a suit against
    EPA  and that,  since EPA  had never  sued  Nyacol or  brought an
    enforcement proceeding  against it, American could  articulate no
    justiciable  controversy  with  EPA.   The  court  entered  final
    judgment for Belaga,  see Fed. R. Civ. P. 54(b), and remanded all
    other  parties and claims to  the state court.   American appeals
    from the dismissal order.
              While  the  parties  vigorously  debate   an  insurance
    company's  ability,   consistent  with  Article  III's   case  or
    controversy requirement, to join EPA in a coverage dispute before
    EPA has brought  an enforcement action against the  insureds, our
    discussion seeps into vastly different legal ground.  The impetus
    behind this  diversion lies  in  the Supreme  Court's opinion  in
    International  Primate  Protection  League  v.  Administrators of
    Tulane Educ. Fund, 
    111 S. Ct. 1700
     (1991).  Concerned about  the
    implications  of  Primate  Protection League  for  federal  court
                     (1) Any  officer of the  United States
              or any agency thereof, or person acting under
              him, for any act under color of such office .
              . . .
    28 U.S.C.   1442(a)(1) (1988).
    jurisdiction, we requested supplemental  briefing on whether this
    action  was  properly  removed  to federal  court.    Both  sides
    responded that  removal was valid  under 28  U.S.C.    1442(a)(1)
    because of Belaga's status as a federal officer.
              Notwithstanding this accord, we must pursue the matter.
    Litigants cannot confer subject matter jurisdiction by agreement.
    See  Insurance Corp. of Ir.  v. Compagnie des  Bauxites, 
    456 U.S. 694
    ,  702  (1982); California  v. LaRue,  
    409 U.S. 109
    ,  113 n.3
    (1972).   Because a federal court is  under an unflagging duty to
    ensure  that it has jurisdiction  over the subject  matter of the
    cases  it proposes to adjudicate,  we are obliged  to address the
    propriety of  removal as a  threshold matter even  though neither
    party  has  raised a  question in  that  regard.   See Mansfield,
    Coldwater  & Lake Michigan  Ry. Co.  v. Swan,  
    111 U.S. 379
    , 382
    (1884)  (stating  that  a federal  appellate  court,  on  its own
    motion,  must "deny its own jurisdiction, and, in the exercise of
    its  appellate  power, that  of all  other  courts of  the United
    States,   in  all   cases  where   such  jurisdiction   does  not
    affirmatively appear in the record").
              The officer  removal statute, 28 U.S.C.    1442 (a)(1),
    quoted supra note  2, is  designed to allow  federal officers  to
    remove  actions   to  federal  court  that   would  otherwise  be
    unremovable.   See Willingham  v. Morgan,  
    395 U.S. 402
    ,  406-07
    (1969)  (stating  that  the section  covers  all  cases in  which
    federal officers,  sued in  state court, "can  raise a  colorable
    defense arising out of their duty to enforce federal law").3   In
    Primate  Protection  League,  the  Supreme Court  held  that  the
    reference to "any agency"  of the United States contained  in the
    officer removal statute did not stand alone, but constituted part
    of a possessive phrase modifying the  noun "officer."  See 111 S.
    Ct. at 1705.  Put another way, the statute is to be read as  if a
    second  "of"  appeared  in  the text  immediately  following  the
    disjunctive "or."   Hence, section 1442(a)(1)  permits removal by
    an "officer  of . .  . any agency  [of the United  States]," and,
    conversely, does not permit removal by the federal agency itself.
    See id. at 1709.
              Although  Primate Protection  League  makes it  crystal
    clear  that EPA, as a federal agency,  cannot remove an action to
    federal  court  under  color  of section  1442(a)(1),  this  case
    presents  a variation on the  theme:  it  requires that we decide
    the  closely  related,  but nonetheless  different,  question  of
    whether,  for purposes  of the  officer  removal statute,  a suit
    brought against  an executive  of an  agency, exclusively  in the
    executive's  official capacity,  constitutes  a suit  against  an
    "officer,"  thereby permitting  removal under  the statute,  or a
    suit against an "agency," thereby precluding such removal.  It is
         3This  is not  to say  that every  case in  which a  federal
    officer  is a defendant is removable.  Rather, removal is limited
    to   situations  in   which   the   officer's  removal   petition
    demonstrates the existence  of a  federal defense.   See Mesa  v.
    489 U.S. 121
    , 136 (1989).
    to this inquiry that we now turn.4
              Generally, a  suit against an officer  in the officer's
    official capacity  constitutes  a suit  against the  governmental
    entity  which the  officer heads.   For  example, in  Kentucky v.
    473 U.S. 159
      (1985), the Court dwelt on  the distinction
    between  suits against a person  in an individual,  as opposed to
    official, capacity.   The Court explained  that while individual-
    capacity  actions  "seek  to  impose personal  liability  upon  a
    government official," id. at 165 (emphasis supplied), payable out
    of personal assets,  see id. at 166,  an official-capacity action
    is, "in  all respects other  than name, to  be treated as  a suit
    against  the  entity."    Id.   Phrased  differently,  "official-
    capacity suits  generally represent only another  way of pleading
    an action  against an entity  of which an  officer is an  agent."
    Monell v. New York City Dep't of Social Servs., 
    436 U.S. 658
    , 690
    n.55 (1978).   Thus, a string of Supreme Court cases holds that a
    suit against a government officer in his or her official capacity
    binds the agency  or other governmental  entity, not the  officer
    personally.  See, e.g., Graham, 473 U.S. at 166; Brandon v. Holt,
    469 U.S. 464
    ,  471  (1985); cf.  Larson  v. Domestic  &  Foreign
    Commerce Corp.,  
    337 U.S. 682
    , 687  (1949) (explaining that,  in
    deciding  whether  a  suit  against a  government  officer  is an
    official-capacity  or  individual-capacity  suit,   "the  crucial
         4The  officer  removal  statute  was first  enacted  in  its
    present  form  in  1948.    Its  legislative  history  is  of  no
    significant assistance in respect to the inquiry at hand.
    question is whether the relief sought . . . is relief against the
              The  character  of  an  official-capacity suit  is  not
    transformed simply because jurisdiction is in issue.  In Loeffler
    v. Frank, 
    486 U.S. 549
     (1988), a case in which the plaintiff sued
    the Postmaster General in  his official capacity, the Court  held
    that acts of a  government officer in his official  capacity "are
    always chargeable" as  acts of the agency for purposes  of a sue-
    and-be-sued clause.   Id. at 563  n.8; see also FHA  v. Burr, 
    309 U.S. 242
    , 249-50  (1940) (holding that a suit against the head of
    an agency in  his official capacity was  indistinguishable from a
    suit against the agency itself for purposes of  a sue-and-be-sued
    clause  and  that a  waiver of  immunity  as to  the  agency head
    necessarily waived the agency's  immunity).  Similarly, even when
    pondering  jurisdictional  and quasi-jurisdictional  issues, this
    court has consistently accepted and applied the principle that an
    official-capacity  suit  against a  government  officer  is fully
    equivalent  to a suit against  the agency.   See, e.g., Northeast
    Fed. Credit  Union v. Neves,  
    837 F.2d 531
    , 533 (1st  Cir. 1988)
    ("Where  .  . .  claims are  made  against a  government official
    acting purely in a representative role, the suit must be regarded
    as one against the sovereign."); Culebras Enters. Corp. v. Rivera
    813 F.2d 506
    , 516  (1st Cir.  1987) (observing  that, for
    Eleventh  Amendment   purposes,  an  official-capacity   suit  is
    "tantamount to a suit against the [governmental entity]").
              We see no reason  to forsake the general rule    that a
    suit against a government officer in his or her official capacity
    is   a  suit   against  the   agency      when  considering   the
    appropriateness of  removal under 28  U.S.C.    1442(a)(1).   The
    theme of Primate Protection League, logically extended, serves an
    esemplastic purpose,  shaping the officer removal  statute into a
    rational whole.  In  that case, the Supreme Court  explained that
    when a suit for monetary damages is brought against an officer in
    his or  her individual  capacity, the complicated  questions that
    arise as to that officer's immunity support a protective grant of
    removal jurisdiction.  See Primate Protection League, 111 S.  Ct.
    at 1708.   By contrast, agencies do  not need the prophylaxis  of
    federal removal because determining an agency's immunity,  unlike
    determining   an  individual's   immunity,  is   a  "sufficiently
    straightforward"  proposition.   Id.    That  rationale  strongly
    favors  treatment  of  official-capacity suits  for  purposes  of
    removal in  the same manner as  suits against the agency.   After
    all, because a suit  against an officer in her  official capacity
    cannot bind the officer personally,  see, e.g., Brandon, 469 U.S.
    at 471, no issues of immunity can possibly arise that differ from
    those arising in a suit directly against the agency.
              Consistency    is    the   touchstone    of   statutory
    interpretation.  If we were to hold that a suit nominally against
    EPA is not removable, as  Primate Protection League demands,  but
    then  go on to hold that a  suit nominally against EPA's regional
    administrator  in her  official capacity    a  suit that  seeks a
    judgment  binding on the EPA   is nonetheless removable, we would
    spawn a  glaring interpretive inconsistency and,  in the bargain,
    impugn  the Primate  Court's reasoning.   Given  the identity  of
    juridical interest  that exists  between a government  agency and
    its executive officer when the latter is sued only in  his or her
    official capacity, the  fact that  the agency may  not remove  an
    action under  the officer removal statute  compels the conclusion
    that  an official-capacity  defendant is  likewise  disabled from
    initiating   removal   thereunder.5      See   generally  Primate
    Protection  League, 111 S. Ct. at 1708 (explaining that access to
    removal  under  section  1442(a)(1)  does  not  turn  on  a "mere
              Our conclusion is  fortified by Judge  Posner's opinion
    in Western Secs. Co. v. Derwinski, 
    937 F.2d 1276
     (7th Cir. 1991).
         5Ordinarily, the  question of  whether a complaint  names an
    officer  in  a  personal, as  opposed  to  an official,  capacity
    requires  little  more  than a  glance  at  the  pleadings.   If,
    however,  a  federal official,  reasonably  believing  himself or
    herself to be sued individually, attempts to remove under section
    1442(a)(1),  any   dispute  as  to  the   officer's  status  will
    necessarily  be resolved  by a  federal court  in the  context of
    assaying its own  jurisdiction.   See 14A Charles  A. Wright  et.
    al.,  Federal   Practice  and   Procedure     3730,   at  499-500
    (explaining that a defendant  wishing to remove need only  file a
    notice of removal, with the result that the propriety of removal,
    if challenged at  all, will "be tested later in the federal court
    by a  motion to remand").  Thus, our holding that an officer sued
    in an official  capacity may not  remove the  action in order  to
    obtain a  federal forum for  resolution of the  underlying merits
    will in no way deprive  the officer of access to a  federal forum
    for  determination of  whether  the suit  is  in fact  an  action
    against him or her personally.   By the same token, if an officer
    is sued in both individual and official capacities, we see no bar
    to  removal under  section 1442(a)(1).   See  El Gran  Video Club
    Corp. v. E.T.D., Inc., 757 F.  Supp. 151, 155 & n.3 (D.P.R. 1991)
    (observing that federal courts have removal jurisdiction over all
    pendent claims that are not "separate and independent").   
    There, the Seventh Circuit, acting sua sponte, refused to allow a
    federal administrator sued in his  official capacity to remove an
    action to federal court  under section 1442(a)(1).  Acknowledging
    that,  in  the  aftermath  of Primate  Protection  League,  suits
    against  federal   agencies  cannot  be   removed  under  section
    1442(a)(1), see id. at  1278, Judge Posner wrote that  "while the
    suit in this case  is nominally against the Administrator,  it is
    against   him  in  his  official  capacity  and  such  suits  are
    considered  to be against the  government itself."   Id. at 1279.
    Thus, the  court concluded that, absent an  alternative basis for
    jurisdiction,  dismissal for want  of subject matter jurisdiction
    would be required.  See id.6
              We  agree  with   the  Seventh  Circuit   that  Primate
    Protection League,  logically extended,  mandates that a  federal
    officer  sued  solely in  his or  her  official capacity  may not
    remove a suit  to federal court  under the aegis  of 28 U.S.C.   
    1442(a)(1).   In the case  before us, this holding  draws the sap
    from  the tree:  American's  suit, brought against  Belaga in her
    official capacity  and seeking no relief  against her personally,
    is in reality a suit against the agency.  It  necessarily follows
    that, because EPA itself could not have removed this action under
    section  1442, see Primate Protection League, 111 S. Ct. at 1708,
    Belaga's  attempt to  remove under  the same  statutory provision
    must fail.
         6In Derwinski  the court  retained jurisdiction  because the
    plaintiff's action arose  under federal law.  See  Derwinski, 937
    F.2d at 1280; see also 28 U.S.C.    1331, 1441 (1988).
              Notwithstanding  that  this  action was  infelicitously
    removed  under  28  U.S.C.     1442(a)(1)  and  that the  parties
    steadfastly   disclaim  any   independent   basis   for   federal
    jurisdiction,7 we  inquire whether any other  toehold for federal
    court  jurisdiction  exists.    Belaga's notice  of  removal  did
    mention 28 U.S.C.   1441   a statute which permits removal of any
    suit that originally  could have been brought in  federal court.8
    See, e.g., Merrell Dow Pharmaceuticals Inc. v. Thompson, 
    478 U.S. 804
    , 808 (1986).  Thus, principles of equity, as well as the law,
    compel us  to  explore whether  American's  declaratory  judgment
    action  falls  within  the  federal  district  court's   original
    jurisdiction.  We proceed to run that gauntlet.
         7The parties' appellate filings ground  removal jurisdiction
    solely in the officer removal statute.   In answer to our express
    inquiry  anent  jurisdiction,  Belaga  replied  that  the  United
    States, on her behalf, removed the action pursuant to 28 U.S.C.  
    1442(a)(1) and contended  that such removal was proper.  American
    averred that, apart from the officer removal statute, it was "not
    aware of any other  basis for federal jurisdiction."   Nyacol, by
    electing  not  to participate  in  this  appeal, has  effectively
    consented to the appropriateness of a state forum.
         8The statute provides in pertinent part:
              Except as otherwise expressly provided by Act
              of Congress,  any civil  action brought  in a
              State court  of which the  district courts of
              the United States have original jurisdiction,
              may  be  removed  by  the  defendant  or  the
              defendants,  to  the  district court  of  the
              United  States for the  district and division
              embracing  the place  where  such  action  is
    28 U.S.C.   1441(a) (1988).
              In this instance, the presence of original jurisdiction
    hinges  upon the existence vel  non of a  federal question,9 that
    is,  the existence of an action  "arising under the Constitution,
    laws,  or treaties  of  the United  States."   28  U.S.C.    1331
    (1988).   Ordinarily, a claim arises under federal law within the
    meaning  of section 1331 if a  federal cause of action appears on
    the face of a well-pleaded  complaint.  See Gully v.  First Nat'l
    Bank  in  Meridian,  
    299 U.S. 109
    ,  113  (1936);  Louisville  &
    Nashville  R.R.  v.  Mottley, 
    211 U.S. 149
    ,  152  (1908).   For
    purposes  of  determining  the  existence   of  federal  question
    jurisdiction  in  a  declaratory  judgment  action, however,  the
    Supreme Court directs  our attention  away from the  face of  the
    complaint  and focuses  it instead  on the  law that  creates the
    underlying cause of action:
              Where   the  complaint   in  an   action  for
              declaratory  judgment  seeks  in  essence  to
              assert   a   defense  to   an   impending  or
              threatened  state  court  action, it  is  the
              character  of the threatened  action, and not
              of  the defense, which will determine whether
              there is federal-question  jurisdiction . . .
    Public  Service Comm'n v. Wycoff  Co., 
    344 U.S. 237
    , 248 (1952);
    see  also Franchise  Tax  Bd. v.  Construction Laborers  Vacation
    463 U.S. 1
    , 15-16 & n.14 (1983).
              Here, American's declaratory judgment action represents
    a mounted offensive on two related fronts:  the suit  is not only
         9The  parties  to the  lawsuit  are not  of  totally diverse
    citizenship  and,  apart  from   the  possibility  of  a  federal
    question,  there  is no  other  arguable  foundation for  federal
    an attempt to preempt impending actions by the insureds (seeking,
    presumably, to secure a  defense against charges of environmental
    harm  and to secure indemnification with  respect to loss, costs,
    damages, and  expense associated  therewith), but  it is also  an
    attempt  to foreclose  governmental  authorities from  trying  to
    reach insurance  proceeds to satisfy  as-yet-unrealized judgments
    under CERCLA.  Thus,  the underlying causes of  action, howsoever
    visualized, are in the  nature of claims to enforce  an insurance
    contract, the  adjudication of which will  require interpretation
    of  various  policy provisions  and contract  terms.   A  suit is
    customarily deemed to arise under the law that gives birth to the
    cause of action.   See American Well Works Co.  v. Layne & Bowler
    241 U.S. 257
    , 260 (1916).  Applying that approach, this case
    is  based  solely  upon,  and,  therefore,  arises  under,  state
    law.10  See CPC  Int'l, Inc. v. Northbrook Excess  & Surplus Ins.
    962 F.2d 77
    ,  97-98 (1st  Cir. 1992)  (divining applicable
    state law  to  interpret  a  pollution  exclusion  clause  in  an
    insurance policy); A. Johnson & Co. v. Aetna Cas. & Sur. Co., 
    933 F.2d 66
    ,  70-74 (1st Cir. 1991)  (same); Ryan v. Royal  Ins. Co.,
         10We  note  in passing  that, were  American to  assert that
    CERCLA preempts state insurance actions, such an assertion, while
    mentioning   federal  law,  would   be  insufficient   to  confer
    jurisdiction because federal law  would come into play only  as a
    defense.    See Franchise  Tax Bd.,  463  U.S. at  13-14; Nashoba
    Communications, Inc. v. Town  of Danvers, 
    893 F.2d 435
    ,  440 (1st
    Cir. 1990); see also Hudson Ins. Co. v. American Elec. Corp., 
    957 F.2d 826
    , 830  n.4 (11th  Cir.), cert.  denied, 113  S.  Ct. 411
    (1992).  We except, of course,  the rare case, discussed infra p.
    15-16, where federal law so completely  displaces state causes of
    action in a particular area that all such claims are "necessarily
    federal in character."  Metropolitan Life Ins. Co. v. Taylor, 
    481 U.S. 58
    , 63-64 (1987).
    916 F.2d 731
    , 734-35 (1st  Cir. 1990) (similar);  In re Acushnet
    River & New Bedford  Harbor, 
    725 F. Supp. 1264
    , 1278-81 (D. Mass.
    1989)  (certifying various  questions  regarding the  substantive
    interpretation  of  an  insurance  policy  to  the  Massachusetts
    Supreme Judicial Court in order to determine an insurer's duty to
    cover cleanup  costs), aff'd in  part and rev'd in  part on other
    grounds sub nom. Lumbermens  Mut. Cas. Co. v.  Belleville Indus.,
    938 F.2d 1423
     (1st Cir. 1991), cert. denied, 
    112 S. Ct. 969
    (1992); see also  Hudson Ins.  Co. v. American  Elec. Corp.,  
    957 F.2d 826
    , 828  (11th Cir.) (holding  that an insured's  potential
    suit to recover insurance proceeds for its liability under CERCLA
    arises   under   "the   applicable   state   law   governing  the
    interpretation of insurance contracts"), cert. denied, 
    113 S. Ct. 411
     (1992).  Federal  jurisdiction cannot take root in  this arid
              In  an abundance  of  caution, we  take one  additional
    step.   There may  exist unusual  circumstances wherein  a state-
    created cause of action can be deemed to arise under federal law.
    For  example, when  a plaintiff's  state-created right  to relief
         11We note that CERCLA itself does not provide a direct cause
    of  action against a responsible  party's liability insurer.  See
    Port  Allen Marine Servs., Inc. v.  Chotin, 
    765 F. Supp. 887
    , 889
    (M.D.  La.  1991)  (dismissing claims  brought  directly  against
    carrier  because CERCLA "does not create a direct right of action
    against  [PRP's] insurers");  cf. 42  U.S.C.    9608(c) (allowing
    direct action  against guarantors in limited  circumstances).  It
    is,  therefore,  abundantly  clear  that  American's  declaratory
    judgment complaint  anticipates future coercive actions  that not
    only will  be entirely governed  by state law,  but also will  be
    initiated only through state-created mechanisms.
    "necessarily depends  on resolution of a  substantial question of
    federal  law,"  Franchise Tax  Bd., 463  U.S.  at 28,  or  when a
    parallel "federal  cause of  action completely pre-empts  a state
    cause of action," id. at 24, the suit is considered a creature of
    federal law.  We test these waters.
               The  latter proposition  (complete preemption)  can be
    summarily dismissed.   Structurally, CERCLA provides "no parallel
    federal  cause of action  for the recovery  of insurance proceeds
    for CERCLA-created liability."  Hudson, 957 F.2d at 830.  Rather,
    its  provisions,  read objectively,  choreograph  a  pas de  deux
    wherein CERCLA-driven suits to  collect insurance proceeds are to
    be brought under state  law.  See,  e.g., 42 U.S.C.    9607(e)(1)
    (preserving "any agreement to insure, hold harmless, or indemnify
    a  party" for CERCLA  liability).  We  turn, then,  to the former
    proposition (necessary dependence on a federal-law question).  
              Conceivably, American might argue that a suit to compel
    it  to defend and/or indemnify  its insureds is  one that, though
    created by  state law, necessarily  turns on federal  common law.
    However, such an argument  amounts to a call for  the application
    of a uniform federal rule of decision to govern interpretation of
    an  insurance  policy's  scope   of  coverage  vis-a-vis   CERCLA
    liability.  We  decline to heed  that call in  the face of  solid
    precedent  pointing  in  the  opposite  direction.    The  massed
    authority  for  treating insurance  coverage questions  in CERCLA
    cases as peculiarly  matters of state law  pervades the courts of
    appeals.   See, e.g., Northbrook,  962 F.2d at  79; Northern Ins.
    Co. v. Aardvark Assocs., Inc., 
    942 F.2d 189
    , 192 (3d Cir.  1991);
    Liberty  Mut. Ins. Co. v.  Triangle Indus., Inc.,  
    957 F.2d 1153
    1157 (4th Cir.), cert. denied, 
    113 S. Ct. 78
     (1992); FL Aerospace
    v. Aetna Cas.  & Sur. Co.,  
    897 F.2d 214
    ,  219 (6th Cir.),  cert.
    111 S. Ct. 284
      (1990); Aetna Cas. & Sur. Co.  v. General
    Dynamics  Corp., 
    968 F.2d 707
    ,  710 (8th Cir.  1992); Industrial
    Indemnity Ins. Co. v. Crown Auto Dealerships, Inc., 
    935 F.2d 240
    241 (11th  Cir. 1991).  State  courts chime in tune.   See, e.g.,
    Hazen Paper, 555  N.E.2d at 579; Boeing Co. v.  Aetna Cas. & Sur.
    784 P.2d 507
    , 509  (Wash. 1990); Technicon  Elecs. Corp. v.
    American  Home  Assurance Co.,  
    542 N.E.2d 1048
    , 1050-51  (N.Y.
              Case law aside, we  doubt that Congress intended CERCLA
    to  be the springboard  for catapulting federal  courts into what
    has historically been a state-law preserve.  Congress has made it
    plain that  federal legislation  should rarely be  interpreted to
    encroach  on  a state's  regulation  of  insurance.   See,  e.g.,
    McCarran-Ferguson Act,  15 U.S.C.    1012(b) (1988).   Nothing in
    CERCLA  suggests  that Congress  intended  to  deviate from  this
    regimen.  Indeed,  CERCLA's text not only  envisions the bringing
    of  suits  under  state   law  but  specifically  mandates  their
    resolution in accordance with that law.   See 42 U.S.C.   9672(a)
    (stating  that  CERCLA's  insurance subchapter  "shall  [not]  be
    construed to affect . . . the law governing the interpretation of
    insurance  contracts of  any State").   Thus,  CERCLA effectively
    rebuts the claim  that its drafters intended to  transform state-
    law insurance actions into actions arising under federal law. 
              We  think  that  the   situation  at  hand  is  closely
    analogous  to that which confronted  us in Royal  v. Leading Edge
    Prods., Inc.,  
    833 F.2d 1
      (1st Cir. 1987).   There, a  plaintiff
    sought to recover for breach of  a royalty agreement related to a
    copyrighted work.  In an effort to maintain federal jurisdiction,
    he argued that his  case arose under the federal  copyright laws.
    See id. at 2.  We disagreed, pointing out  that when an action is
    brought  to enforce a royalty  contract the action  arises out of
    the contract and not under the copyright statute, even though the
    contract concerns a copyright.  See id. at 4.  By the same token,
    an action brought to enforce the pollution-coverage provisions of
    an  insurance policy  arises  out of  the  policy and  not  under
    federal  environmental law,  even though  any  potential recovery
    under  the  policy  will satisfy  a  CERCLA-generated  liability.
    Accord Hudson, 957 F.2d at 829-30.  In this case, as in Royal, it
    would  be wrong to arrogate unto the federal courts "jurisdiction
    over what  is  essentially a  garden-variety  contract  dispute."
    Royal, 833 F.2d at 5.
              In fine, because the insurance dispute which American's
    declaratory judgment  action anticipates  is a creature  of state
    law  and  cannot be  said to  arise  under federal  law, original
    federal  question  jurisdiction     and,  by  extension,  removal
    jurisdiction under 28 U.S.C.   1441(a)   does not lie.
              Since neither  section 1442 nor  section 1441  supports
    the removal  of American's declaratory judgment  action, there is
    simply no serviceable  hook on which federal  jurisdiction can be
    hung.  We, therefore, go no further. Inasmuch  as the lower court
    lacked subject  matter  jurisdiction, its  order  dismissing  the
    action against EPA is null.  See Insurance Corp. of Ir., 456 U.S.
    at 701 ("The validity of an order of a federal court depends upon
    that court's  having jurisdiction over . . . the subject matter .
    .  . .").     We vacate  the dismissal order and  direct that the
    district court reinstate  Belaga, in her official  capacity, as a
    party and  thereafter return the improvidently  removed action to
    the court from whence it emanated.12
              Vacated and remanded with directions.  No costs.
         12We take  no view of Belaga's claims of sovereign immunity,
    non-justiciability, unripeness  and the  like.  We  are similarly
    noncommittal  as  to  the  effect,  if  any,  of  remand  on  the
    stipulation  entered into between American and  DEP.  Because the
    federal  courts  lack  jurisdiction,  all such  matters  must  be
    presented to, and resolved by, the state courts.