Somerset Regional Water v. ( 2020 )

PRECEDENTIAL
UNITED STATES COURT OF APPEALS
FOR THE THIRD CIRCUIT
_______________
No. 19-1874
_______________
IN RE: SOMERSET REGIONAL WATER
RESOURCES, LLC
LARRY L. MOSTOLLER; CONNIE J. MOSTOLLER,
Appellants
_______________
On Appeal from the United States District Court
for the Western District of Pennsylvania
(D.C. No. 3:18-cv-00204)
District Judge: Honorable Marilyn J. Horan
_______________
Submitted Under Third Circuit L.A.R. 34.1(a)
on November 14, 2019
Before: AMBRO, KRAUSE, and BIBAS, Circuit Judges
(Opinion Filed: February 11, 2020)
_______________
Aurelius P. Robleto
Robleto Law
Three Gateway Center
401 Liberty Avenue, Suite 1306
Pittsburgh, PA 15222
Counsel for Appellants
Courtney S. Schorr
McGuireWoods
260 Forbes Avenue, Suite 1800
Pittsburgh, PA 15222
Counsel for Appellee
_________________
OPINION OF THE COURT
_________________
BIBAS, Circuit Judge.
When a lender insists on collateral, it expects the collateral
to be worth something. Larry Mostoller’s company was in
bankruptcy and about to fold. Its largest creditor was willing to
lend another $1 million to keep it afloat, but only if
Mr. Mostoller pledged a forthcoming personal tax refund as
collateral. Everyone who negotiated the deal expected that the
refund would amount to roughly $1 million—the net amount
owed to Mr. Mostoller based on his company’s substantial
2015 losses, which he could use to offset his taxable income in
2013, 2014, and 2015.
2
But now that Mr. Mostoller has the loan and the tax-refund
check, he urges a reading of the agreement that he never men-
tioned during negotiations: that he pledged as collateral his re-
fund on taxes that he paid for 2015 alone, excluding any refund
on his 2013 and 2014 taxes. Yet he admits that his reading
would make the collateral worthless.
The bankruptcy court rightly rejected Mr. Mostoller’s
novel reading of the agreement. Its description of the collateral
was ambiguous, so the court enforced it as the parties under-
stood it: to produce a million-dollar refund. Without that secu-
rity, the lender would never have made so risky a loan. And
because Mr. Mostoller owned almost the entire refund sepa-
rately from his wife, the court properly rejected his argument
that his pledge was unenforceable without her consent. So like
the District Court before us, we will affirm.
I. BACKGROUND
A. The bankruptcy
Mr. Mostoller solely owned Somerset Regional Water Re-
sources, LLC (the Debtor), a water-transportation business that
serviced oil and gas wells. The Debtor used to be profitable.
But when oil prices plummeted in mid-to-late 2014, the oil and
gas industry suffered. The Debtor and its customers were no
exception. Its losses mounted and its balance sheet plunged
into the red.
The Debtor’s largest creditor was Somerset Trust Com-
pany, to which it owed more than $3 million. The Trust’s loans
were secured by a blanket lien on most of the Debtor’s assets
and a personal guarantee by Mr. Mostoller.
3
In late 2015, the Debtor faced a severe cash-flow shortage
and the likely termination of one of its leases. To stop the
bleeding, it voluntarily petitioned for reorganization under
Chapter 11 of the Bankruptcy Code.
B. The emergency loan, its collateral, and the default
Chapter 11 lets struggling companies reorganize so that
they can exit bankruptcy, keep operating, and pay as much as
possible to their creditors. In re Armstrong World Indus., Inc.,
432 F.3d 507, 518 (3d Cir. 2005). But the Debtor faced a dire
liquidity crisis; it stood little chance of surviving a Chapter 11
reorganization without an immediate cash infusion. And be-
cause the Debtor was overleveraged, it would find it hard to
attract new lenders in what little time it had.
The Trust, however, had a unique incentive to lend more: if
a new loan could keep the Debtor afloat, it would more likely
be able to repay the Trust in full. Still, to encourage new lend-
ing, the Debtor would have to pledge to the Trust substantial
new collateral. But the Debtor had already pledged most of its
assets to the Trust as security; it had little left to offer.
Thus, the Debtor’s management turned to Mr. Mostoller to
see if he would pledge some personal assets to secure a loan to
save his business. To entice the Trust to lend more money, the
Debtor’s Chief Restructuring Officer proposed that Mr. Mostol-
ler “assign his interests in the net proceeds of [an anticipated]
federal tax refund.” 3 Ohio App. 1179.
A taxpayer is entitled to a refund if he pays more taxes than
he has to. See 26 U.S.C. § 6402(a) (2012). And Mr. Mostoller
had overpaid over several years. As an S Corporation, the
4
Debtor was a pass-through entity for tax purposes. See 26
U.S.C. § 1363(a) (2012). Its taxable income and losses passed
through to Mr. Mostoller, its sole owner. See id.
§ 1366(a)(1)(A). He filed his taxes jointly with his wife in
2013, 2014, and 2015. In 2013 and 2014, when the Debtor was
thriving, the Mostollers had paid millions of dollars in federal
taxes on that income.
But by 2015, the business was struggling. Under a provi-
sion of the Internal Revenue Code in effect at the time, he could
file amended 2013 and 2014 tax returns to carry back the
Debtor’s 2015 losses, which would offset his taxable income
for those two years and trigger a refund. 26 U.S.C. § 172(a),
(b)(1)(A)(i) (2012), repealed in relevant part by The Tax Cuts
and Jobs Act, Pub. L. No. 115-97, tit. I, § 13302(b), 131 Stat.
2054, 2122 (2017). Because the Debtor had lost millions of
dollars in 2015, the Debtor, the Trust, and their advisors ex-
pected that Mr. Mostoller would get a net tax refund of close
to $1 million. And the parties understood that Mr. Mostoller
could pledge this amount as collateral for an emergency loan.
In the hasty negotiations that followed, the parties reached
an agreement. Mr. Mostoller was involved in the negotiations
and signed the agreement. In paragraph 6 of that agreement, he
pledged as collateral “any rights or interest in the 2015 Federal
tax refund due to him individually, but attributable to the oper-
ating losses of the Debtor.” 2 Ohio App. 56–57 ¶ 6. In exchange, the
Trust would lend the Debtor $1 million.
Without that valuable collateral, the Trust would not have
lent the Debtor more money. The tax refund was “a central part
of [the] collateral package” and was “insisted upon by [the]
5
Trust.” 3 Ohio App. 1068. But the agreement left open the details
about executing the tax filings needed to trigger the expected
refund. The parties expected that an accountant would handle
these details later. Soon after the parties struck the deal, the
bankruptcy court approved the agreement and entered it on its
docket as a consent order (the Loan Order). 2 Ohio App. 51–75.
But even this cash infusion could not save the Debtor. It
soon defaulted on the emergency loan. Without a new source
of financing to keep the business afloat, the Debtor converted
its bankruptcy from a Chapter 11 reorganization to a Chapter 7
liquidation.
C. The tax-refund dispute
Right after the Debtor defaulted on the emergency loan,
Mr. Mostoller tried to hang onto the collateral that he had
pledged. At first, he apparently refused to file his 2015 tax re-
turn and amended 2013 and 2014 tax returns, which were
needed to generate the tax refund. So the Trust moved to com-
pel him to do that. At the hearing on this motion, Mr. Mostoller
told the bankruptcy court that he had filed those tax returns.
And he testified that he “agree[d] that [the] Trust gets half of
the tax refund, minus the federal taxes due,” with the other half
going to his wife. 3 Ohio App. 969. The Trust later agreed to that
proposal.
But when the tax refund came, Mr. Mostoller tried to keep
all of it for himself. His accountant received the $1.12 million
refund check from the IRS and followed the bankruptcy court’s
order by promptly depositing it with the court. Yet when the
6
Trust moved to claim Mr. Mostoller’s share of the pledged col-
lateral, he cross-moved, seeking the entire refund.
In briefing on those motions, Mr. Mostoller argued for the
first time that paragraph 6 of the Loan Order was limited to any
tax refund owed to him because of income offset in the 2015
tax year and did not include any refund from income offset in
prior tax years. In the alternative, he argued that because he and
his wife owned the refund as tenants by the entirety under
Pennsylvania law, and because his wife had not signed the
Loan Order, the Trust could not seize the refund proceeds.
In response, the Trust argued that paragraph 6 is ambigu-
ous, that extrinsic evidence from the negotiations showed that
refunds derived from offsetting the Mostollers’ 2013 and 2014
income against the Debtor’s 2015 losses were included, and
that the Mostollers’ property interests in the refund were sepa-
rate. In support, the Trust submitted affidavits from the Trust’s
Senior Vice President, the Debtor’s Chief Restructuring Of-
ficer, and a lawyer who helped the Debtor negotiate the emer-
gency loan. Each affiant had taken part in the negotiations and
maintained that paragraph 6 encompassed refunds derived
from the offset of taxable income from prior years, not just
from 2015. At an evidentiary hearing, each affiant testified to
the same.
For his part, Mr. Mostoller maintained that paragraph 6 is
unambiguous, though he chose at the last minute not to testify.
7
D. Procedural history
After supplemental briefing, the bankruptcy court held for
the Trust on all grounds. Somerset Tr. Co. v. Mostoller (In re
Somerset Reg’l Water Res., LLC), 592 B.R. 38 (Bankr. W.D.
Pa. 2018). It credited the Trust’s witnesses and the “over-
whelming evidence” of the parties’ intent in holding that “the
deal was to pledge the entirety of the refund generated by the
[Debtor’s] 2015 operating losses.” Id. at 62. So it read para-
graph 6 of the Loan Order to include refunds from 2013
through 2015—that is, the full amount at issue. Id. at 57–60. It
also rejected the Mostollers’ claim that they owned the refund
as tenants by the entirety, holding that under federal tax law
their interests were divisible. Id. at 63–64. After factoring in
the Trust’s concession that Mrs. Mostoller could keep half of
the refund, the bankruptcy court ordered the release of the re-
maining $536,894 held in escrow to the Trust. See id. at 60, 64.
On appeal, the District Court affirmed.
The Mostollers now appeal to our Court. They raise three
arguments: first, that the bankruptcy court lacked subject-
matter jurisdiction to decide the dispute; second, that para-
graph 6 of the Loan Order is unambiguously limited to the re-
fund from only the 2015 tax year; and third, that they owned
the whole $1.12 million refund as tenants by the entirety, so
the Trust cannot seize any of those funds.
On appeal, “we ‘stand in the shoes’ of the District Court”
and apply the same standard of review. In re Glob. Indus.
Techs., Inc., 645 F.3d 201, 209 (3d Cir. 2011) (en banc) (quot-
ing IRS v. Pransky (In re Pransky), 318 F.3d 536, 542 (3d Cir.
2003)). We review the bankruptcy court’s legal determinations
8
de novo, its factual findings for clear error, and its discretion-
ary decisions for abuse of discretion. Schepis v. Burtch (In re
Pursuit Capital Mgmt., LLC), 874 F.3d 124, 133 n.14 (3d Cir.
2017). And because the bankruptcy court heard testimony from
witnesses who participated in the negotiations, we give “due
regard to the opportunity of that court to judge first-hand the[ir]
credibility.” Fellheimer, Eichen & Braverman, P.C. v. Charter
Techs., Inc., 57 F.3d 1215, 1223 (3d Cir. 1995) (internal quo-
tation marks omitted).
II. THE BANKRUPTCY COURT HAD JURISDICTION
TO DECIDE THE DISPUTE
The Mostollers argue that the bankruptcy court lacked
subject-matter jurisdiction to decide the dispute over the tax
refund. Bankruptcy courts have limited statutory jurisdiction
under the Bankruptcy Code and limited constitutional jurisdic-
tion under Article III. Stern v. Marshall, 564 U.S. 462, 473–
74, 482 (2011). Unless the parties consent, bankruptcy courts
have jurisdiction to enter final judgments only in “core pro-
ceedings.” 28 U.S.C. § 157(b), (c)(1); Stern, 564 U.S. at 474–
75; see Wellness Int’l Network, Ltd. v. Sharif, 135 S. Ct. 1932,
1949 (2015).
Here, the bankruptcy court entered a final judgment on the
parties’ cross-motions and ordered the Mostollers to give up
half of the tax refund. They argue that the court exceeded its
statutory and constitutional jurisdiction because the refund dis-
pute is not a core proceeding. See Stern, 564 U.S. at 482 (ana-
lyzing Bankruptcy Code and Article III jurisdiction sepa-
rately); In re Millennium Lab Holdings II, LLC, 945 F.3d 126,
135–36 (3d Cir. 2019) (same). Not so.
9
First, the dispute falls within the bankruptcy court’s statu-
tory jurisdiction over core proceedings. While the bankruptcy
court held that it had statutory jurisdiction under several provi-
sions of 28 U.S.C. § 157(b)(2), we need discuss only one here.
Section 157(b)(2)(D) confers jurisdiction over “orders in re-
spect to obtaining credit.” That is the issue here: the reach and
scope of paragraph 6 of the Loan Order, a court-approved
agreement that gave the Debtor a new loan. Because the Loan
Order “authorized the agreement[ ] at issue in this case,” the
parties “assert[ ] rights that were established in connection with
one of the [b]ankruptcy [c]ourt’s core functions—here, the ap-
proval of [the Debtor’s] requests for more credit.” KeyBank
Nat’l Ass’n v. Franklin Advisers, Inc., 600 B.R. 214, 228
(S.D.N.Y. 2019) (citing 28 U.S.C. § 157(b)(2)(D)); see also
Billing v. Ravin, Greenberg & Zackin, P.A., 22 F.3d 1242,
1245 n.1 (3d Cir. 1994) (“Post-petition transactions are more
likely to be core proceedings.”).
Second, because this dispute could have arisen only in
bankruptcy, the bankruptcy court’s exercise of jurisdiction did
not offend Article III. See Stern, 564 U.S. at 499 (holding that
a bankruptcy court’s exercise of jurisdiction poses no constitu-
tional problems if “the action at issue stems from the bank-
ruptcy itself or would necessarily be resolved in the claims al-
lowance process”). Without the court’s Loan Order, the Debtor
most likely could not have gotten the emergency financing it
needed to try to survive Chapter 11 reorganization. And once
the court entered the Loan Order, it “plainly had jurisdiction to
interpret and enforce it[ ]” against its signatories, including
Mr. Mostoller. Travelers Indem. Co. v. Bailey, 557 U.S. 137,
151 (2009). As a leading treatise recognizes, “[t]here has never
10
been any doubt” about bankruptcy courts’ jurisdiction over
“matters of administration,” like entering and enforcing “or-
ders in respect to obtaining credit.” 1 Collier on Bankruptcy
¶ 3.02[3][a] (16th ed. 2019).
The bankruptcy court thus properly exercised core-
proceeding jurisdiction over the tax-refund dispute under 28
U.S.C. § 157(b)(1)–(2). The District Court in turn had jurisdic-
tion under 28 U.S.C. §§ 157(a) and 158(a). And we have juris-
diction under 28 U.S.C. §§ 158(d)(1) and 1291. Having settled
this threshold issue, we proceed to the merits.
III. THE BANKRUPTCY COURT RIGHTLY FOUND
AMBIGUITY AND CONSTRUED IT IN FAVOR OF THE TRUST
The crux of this appeal is the phrase “the 2015 Federal tax
refund due to [Mr. Mostoller] individually, but attributable to
the operating losses of the Debtor” in paragraph 6 of the Loan
Order. 2 Ohio App. 56–57 ¶ 6. The parties dispute whether that
phrase is ambiguous and, if so, which side’s reading controls.
The Loan Order is a consent decree formalizing the parties’
agreement, so we interpret it as a contract. McDowell v. Phila.
Hous. Auth., 423 F.3d 233, 238 (3d Cir. 2005). The parties
agree that Pennsylvania law governs.
The bankruptcy court found that paragraph 6 is ambiguous
because it is “subject to multiple reasonable interpretations.”
592 B.R. at 52. After considering documentary evidence and
credible testimony from the negotiators, it resolved that ambi-
guity by adopting the Trust’s reading. It rejected the Mostol-
lers’ contrary reading, which it found would lead to
11
unreasonable results and which Mr. Mostoller never expressed
throughout the loan negotiations. All of that reasoning is
sound.
A. Paragraph 6 of the Loan Order is ambiguous
Whether an agreement is ambiguous is a question of law,
so we review de novo. Pacitti ex rel. Pacitti v. Macy’s, 193
F.3d 766, 773 (3d Cir. 1999) (applying Pennsylvania law). Un-
der Pennsylvania law, “[a] contract is ambiguous if it is rea-
sonably susceptible of different constructions and capable of
being understood in more than one sense.” Schwab v. Penn-
summit Tubular, LLC (In re Old Summit Mfg., LLC), 523 F.3d
134, 137 (3d Cir. 2008) (quoting Hutchison v. Sunbeam Coal
Corp., 519 A.2d 385, 390 (Pa. 1986)). Paragraph 6 is ambigu-
ous because it is subject to at least three reasonable, competing
readings:
First, “2015 Federal tax refund” could mean a tax refund
paid to the Mostollers in the 2015 calendar year. Natural per-
sons generally must use the calendar year as their tax year and
do not file their tax returns until the next year. See 26 U.S.C.
§ 441(g) (2012). So a tax refund issued in 2015 would be for
the 2014 tax year. None of the parties urges us to adopt this
reading, but it is reasonable on its face.
Second, that phrase could mean a refund received at any
time on only the Mostollers’ 2015 taxes (which they paid later
on), triggered by the Debtor’s losses in the 2015 tax year. The
Mostollers urge us to adopt this reading. And the bankruptcy
court agreed that it was reasonable on its face. Because the
Debtor’s profits were their chief source of income, and the
12
parties expected the Debtor to incur millions of dollars in
losses in 2015, this reading would produce a relatively small
tax refund.
Third, paragraph 6 could refer to a tax refund owed to the
Mostollers because of the Debtor’s 2015 losses, even if the
Mostollers used those losses to offset their 2015 income and
income from past years. In other words, the phrase “2015 Fed-
eral tax refund . . . attributable to the operating losses of the
Debtor” might refer to any refund paid because of losses that
the Debtor incurred in 2015. 2 Ohio App. 56–57 ¶ 6. The Trust ad-
vances this reading. And the bankruptcy court agreed that it
was reasonable on its face too. That reading would include re-
funds owed because of the Debtor’s multi-million-dollar losses
in 2015, which the Mostollers could use to offset their taxable
income from 2015 as well as from 2013 and 2014 (the two
years within the carryback period). See 26 U.S.C.
§ 172(b)(1)(A)(i) (2012).
Because paragraph 6 is subject to each of these reasonable,
competing readings, we agree with the bankruptcy court that it
is ambiguous.
B. The Trust’s reading best resolves that ambiguity
To resolve a contract’s ambiguity, we look to extrinsic or
parol evidence—that is, “[e]vidence relating to a contract but
not appearing on the face of the contract because it comes from
other sources, such as statements between the parties or the cir-
cumstances surrounding the agreement.” Extrinsic Evidence,
Black’s Law Dictionary (11th ed. 2019). Whether the ambigu-
ity is patent or latent, we may look to parol evidence all the
13
same. Zuber v. Boscov’s, 871 F.3d 255, 258 (3d Cir. 2017) (cit-
ing Kripp v. Kripp, 849 A.2d 1159, 1163 (Pa. 2004)). And we
may use parol evidence “to show both the intent of the parties
and the circumstances attending the execution of the contract.”
Osial v. Cook, 803 A.2d 209, 214 (Pa. Super. Ct. 2002).
We now turn to the bankruptcy court’s reading of the parol
evidence. “[T]he resolution of conflicting parol evidence” is a
question of fact that we review for clear error. In re Old Sum-
mit, 523 F.3d at 137 (quoting Hutchison, 519 A.2d at 390). The
bankruptcy court found that the parol evidence of the parties’
intent “overwhelming[ly]” favors the Trust’s reading. 592 B.R.
at 62. We agree.
1. The parol evidence of the Loan Order negotiations. The
Trust submitted affidavits and later live testimony from three
witnesses who helped negotiate the loan: the Trust’s Senior
Vice President, the Debtor’s Chief Restructuring Officer, and
a lawyer for the Debtor. We defer to the bankruptcy court’s
finding that each of these witnesses was credible. Fellheimer,
57 F.3d at 1223. Their testimony about the negotiations made
three things clear:
First, without valuable new collateral, the Trust would not
have lent the Debtor more money. That makes sense. Because
of the Debtor’s financial distress, it was at great risk of default.
To secure such a risky loan, someone needed to put up valuable
new collateral.
Second, everyone expected that, after offsetting other tax-
related liabilities, Mr. Mostoller could pledge close to $1 mil-
lion of his tax refund as collateral. This understanding was
14
based on preliminary estimates from an accountant who had
prepared the Debtor’s and the Mostollers’ taxes. Given the fast
pace of negotiations, those estimates were the best the parties
could get.
Third, the parties chose not to specify the details of how
they would execute filing the Mostollers’ new and amended
tax returns. Because the negotiations were moving so fast, they
left those specifics to a professional accountant. So neither the
negotiations nor the Loan Order touched on how the Debtor’s
2015 losses would create a refund. But all the parties expected
that the resulting refund would be worth around $1 million.
Thus, the parties understood that the “2015 refund” referred
generally to any refund generated by the Debtor’s 2015 losses.
2. The Trust’s reading tracks the actual refund. Only the
Trust’s reading squares with this extrinsic evidence. The Trust
reads the key phrase to mean any tax refund owed to
Mr. Mostoller because of losses incurred by the Debtor in
2015, even if those losses produced further and larger refunds
after being carried back to the 2013 and 2014 tax years. Other-
wise, the eventual tax refund would fall far short of the parties’
expectations. To show how, we must reconstruct the Mostol-
lers’ tax filings for those three years.
a. Tax year 2015. In 2015, the Debtor incurred roughly
$6.3 million in losses, dwarfing the Mostollers’ other taxable
income. Because it was a pass-through entity, Mr. Mostoller
claimed the Debtor’s losses on his and his wife’s joint personal
tax return. After factoring in these losses, the Mostollers paid
zero taxes for the 2015 tax year. Because these losses more
than offset their other taxable income, they got a refund of
15
about $126,000 for that year. The Mostollers ask us to stop
there; the Trust urges us to march on.
After satisfying other tax-related obligations, around
$4.9 million of the Debtor’s 2015 net operating losses re-
mained. At the time, the Internal Revenue Code let the Mostol-
lers use those leftover losses to offset taxable income in the two
preceding tax years: 2013 and 2014. See 26 U.S.C.
§ 172(b)(1)(A)(i) (2012). So march on we must, back to 2013
first. See id. § 172(b)(2) (requiring taxpayers to carry losses
back “to the earliest of the taxable years” available under
§ 172(b)(1)(A)(i)).
b. Tax year 2013. In 2013, the Debtor was making money
and the Mostollers had about $490,000 in taxable income. Car-
rying back the Debtor’s 2015 losses on an amended tax return,
the Mostollers were entitled to a refund of about $143,000 for
that tax year. That brought their total refund to roughly
$269,000. Still, more than $4.4 million of the Debtor’s 2015
losses were left to carry over to the 2014 tax year.
c. Tax year 2014. In 2014, the Debtor thrived. The Mostol-
lers earned more than $7.3 million and paid about $2.9 million
in federal income taxes. Through another amended tax return,
the Mostollers used the Debtor’s remaining 2015 losses to off-
set more than half of their 2014 income. That produced a re-
fund of roughly $1.8 million, bringing the total gross refund to
just over $2 million.
d. Adding it all up. The IRS subtracted roughly $900,000
in other tax obligations from those refunds. Ultimately, the
Mostollers got a tax-refund check of about $1.12 million,
16
which matched the parties’ expectations when they executed
the Loan Order. But if paragraph 6 referred to a refund on tax-
able income earned during the 2015 tax year alone, the refund
would have fallen well short of those expectations. Only
through the amended 2013 and 2014 tax returns could the re-
fund reach the anticipated size, one large enough to secure the
million-dollar emergency loan. So only the Trust’s reading of
paragraph 6 tracks the Mostollers’ actual refund, as anticipated
by extrinsic evidence of the parties’ intent.
3. The Mostollers’ reading of paragraph 6 fails. By con-
trast, the parol evidence contradicts the Mostollers’ reading,
which limits the phrase “2015 Federal tax refund” to a refund
of only their 2015 taxes and excludes refund proceeds from
2013 and 2014. As noted, that reading of paragraph 6 would
have foreseeably produced a refund too small to induce and
secure the risky loan.
Worse still, the Mostollers admit that under their reading,
the agreed-upon refund “would be valueless.” Appellants’
Br. 4; accord id. at 15. Their other tax obligations would swal-
low up the 2015-only refund, leaving no value in the collateral
to secure the emergency loan. It strains credulity to think that
such evidently worthless collateral would have enticed the
Trust to make so risky a loan. We reject this commercially un-
reasonable reading. See Starling v. Lake Meade Prop. Owners
Ass’n, 162 A.3d 327, 345 (Pa. 2017) (noting the “prohibition
on any interpretation” of a contract “that leads to an absurd
result”).
To be clear, we do not let parol evidence about ethereal ex-
pectations trump the text. Here, the text itself is ambiguous and
17
is open to several reasonable readings. And “the extrinsic evi-
dence proffered by [the Trust] concerns the parties’ objectively
manifested linguistic reference regarding certain terms of the
contract, rather than merely their [subjective] expectations.”
Bohler-Uddeholm Am., Inc. v. Ellwood Grp., Inc., 247 F.3d 79,
99 (3d Cir. 2001). Thus, the parol evidence helps us to resolve
ambiguity, not to create ambiguity where none exists.
In sum, the Trust’s reasonable reading tracks the parol evi-
dence that the bankruptcy court found credible. And the
Mostollers’ reading conflicts with both the parties’ understand-
ing and basic commercial sense. So we will affirm the bank-
ruptcy court’s reading.
C. The bankruptcy court rightly rejected the
Mostollers’ unexpressed reading
Even if the Mostollers’ reading better tracked the extrinsic
evidence, we would still affirm on another ground. The
bankruptcy court rightly applied the “central principle of
contract interpretation that if a party knew or had reason to
know of the other parties’ interpretation of terms of a contract,
the first party should be bound by that interpretation.” Bohler-
Uddeholm, 247 F.3d at 97, 99 (applying Pennsylvania law);
accord Restatement (Second) of Contracts § 201(2)(b) (Am.
Law Inst. 1981).
During the negotiations, the Debtor’s and the Trust’s
representatives repeatedly put Mr. Mostoller on notice that
they read paragraph 6 as requiring him to pledge around
$1 million of the tax refund as collateral for the emergency
loan. But Mr. Mostoller never told them of his contrary reading
18
at the time. He kept them in the dark until after the Trust had
relied on the collateral to extend a risky loan. By then it was,
as the Mostollers aptly put it, “too late.” Appellants’ Br. 4.
To prevent a silent party from later ambushing his
unwitting opponents, we reject the silent party’s unexpressed
reading. See, e.g., Bohler-Uddeholm, 247 F.3d at 99; Emor,
Inc. v. Cyprus Mines Corp., 467 F.2d 770, 775–76 (3d Cir.
1972) (applying Pennsylvania law). We agree with the
bankruptcy court that Mr. Mostoller is “seeking to take
advantage of both sides of the coin.” 592 B.R. at 62. So the
court was right to apply this doctrine here.
IV. THE MOSTOLLERS’ INTERESTS IN THE 2015
TAX REFUND ARE SEPARATE
In a last-ditch effort to keep the collateral, the Mostollers
argue that they owned the tax refund as tenants by the entirety
under Pennsylvania law. If that is right, then the Trust could
not seize those funds because only Mr. Mostoller, not
Mrs. Mostoller, signed the Loan Order. “Pursuant to Pennsyl-
vania law, property owned as tenants by the entirety cannot be
accessed by the creditors of only one spouse.” Clientron
Corp. v. Devon IT, Inc., 894 F.3d 568, 575 (3d Cir. 2018).
But Pennsylvania law is only part of this equation. It is fed-
eral tax law that determines who owns what portion of a federal
tax refund and how they own it. And federal tax law provides
that spouses’ ownership of a refund depends on how they
owned the income that generated that refund under state
property law.
19
Under Pennsylvania law, the Mostollers held separate in-
terests in the Debtor’s income because Mr. Mostoller alone
owned the Debtor. So the Mostollers’ interests in the refund
were separate too. And they never merged those separate inter-
ests into entireties interests. Thus, the bankruptcy court rightly
rejected this argument.
A. Federal tax refunds are separately owned if the
income is separately owned
As we shall explain, a mixture of federal and state law gov-
erns ownership of federal tax refunds. We discuss each in turn.
1. Federal tax law. The Internal Revenue Code does not
automatically treat refunds from joint marital returns as jointly
owned. Rather, each spouse owns a portion of the refund
separately, according to his or her share of the tax
overpayment. See 26 U.S.C. § 6402(a) (2012) (authorizing the
IRS to “credit the amount of [any] overpayment . . . against any
[tax] liability . . . on the part of the person who made the
overpayment and . . . [to] refund any balance to such person”
(emphasis added)). So the ownership of the spouses’ income
determines how they own a tax refund on that income. If the
income is separate going in, then the refund is separate coming
out. Merely filing a joint tax return does not change that.
Our sister circuits concur. The Fifth Circuit, for instance,
has held that if the income leading to a tax overpayment
belongs to one spouse, then, even if the two file a joint tax
return, the refund does not belong jointly to both spouses.
Ragan v. Comm’r, 135 F.3d 329, 333 (5th Cir. 1998). As Judge
Higginbotham explained for the court, “[a] joint income tax
20
return does not create new property interests for the husband
or wife in each other’s income tax overpayment.” Id. Because
the income was the husband’s alone under Texas law, the wife
had no interest in the resulting refund. Id.
Nor is the Fifth Circuit alone. In the words of the Ninth
Circuit: “A joint return does not itself create equal property
interests for each party in a refund. Spouses who file a joint
return have separate interests in any overpayment, the interest
of each depending upon his or her relative contribution to the
overpaid tax.” United States v. Elam, 112 F.3d 1036, 1038 (9th
Cir. 1997). Thus, “fil[ing] a joint tax return . . . does not change
the underlying property interests at stake.” Id.; see also
Callaway v. Comm’r, 231 F.3d 106, 117 (2d Cir. 2000) (“[T]he
filing of a joint return does not have the effect of converting
the income of one spouse into the income of another.” (citing
McClelland v. Massinga, 786 F.2d 1205, 1210 (4th Cir. 1986));
Gordon v. United States, 757 F.2d 1157, 1160 (11th Cir. 1985)
(“Where spouses claim a refund under a joint return, the refund
is divided between the spouses, with each receiving a
percentage of the refund equivalent to his or her proportion of
the withheld tax payments.”); cf. Rev. Rul. 74-611, 1974-2
C.B. 399, 399 (“Court decisions have consistently held that a
husband and wife who file a joint return do not have a joint
interest in an overpayment; each has a separate interest.”).
We now join our sister circuits in adopting this rule. Thus,
when the Mostollers got their refund check, each spouse
acquired a separate interest in it proportional to his or her
contribution to the overpayments. If the income was jointly
owned, then the Mostollers had a common interest in the
21
refund. But if it was separately owned by Mr. Mostoller, then
his wife had no interest in the refund when it arrived.
To figure out who owned what (and how), we turn to
Pennsylvania law. See United States v. Nat’l Bank of Com-
merce, 472 U.S. 713, 722 (1985) (“[S]tate law controls in de-
termining the nature of the legal interest which the taxpayer
had in the property. . . . [F]ederal statute[s] create[ ] no property
rights but merely attach[ ] consequences, federally defined, to
rights created under state law.” (internal quotation marks
omitted)).
2. State property law. Under Pennsylvania law, Mr.
Mostoller owned most of the spouses’ income in 2013, 2014,
and 2015, because he separately owned the main producer of
that income: the Debtor. So he also owned most of the refund
when it arrived.
In Pennsylvania, spouses ordinarily own property in one of
three ways: separately, as tenants in common, or as tenants by
the entirety. In the first, only one spouse owns the property; the
other does not. The second means that each spouse possesses
the property but has “separate and distinct” legal title to it. In
re Estate of Quick, 905 A.2d 471, 474 (Pa. 2006). And the third
gives each spouse a joint, singular interest in “ ‘the whole or
the entirety,’ and not a ‘share, moiety or divisible part’ ” of the
property. Clientron Corp., 894 F.3d at 579 (quoting In re Bran-
non, 476 F.3d 170, 173 (3d Cir. 2007)). Creditors of one spouse
can attach separate or common interests of that spouse, but they
cannot attach jointly held entirety interests without the other
spouse’s consent. See id. at 575. To win, the Mostollers must
show that they owned their income as tenants by the entirety.
22
For a tenancy by the entirety, Pennsylvania requires the tra-
ditional common-law elements: a marriage, plus the “four uni-
ties” of time, title, possession, and interest. In re Estate of
Quick, 905 A.2d at 474. To satisfy those unities, the spouses
must (1) have their interests “vest at the same time,” (2) “ob-
tain[ ] their title by the same instrument,” (3) have “an undi-
vided interest in the whole,” and (4) own interests “of the same
type, duration and amount.” In re Estate of Rivera, 194 A.3d
579, 586–87 (Pa. Super. Ct. 2018) (quoting Fenderson v.
Fenderson, 685 A.2d 600, 607 (Pa. Super. Ct. 1996)).
Here, none of the unities was present. Because Mr. Mostol-
ler was the Debtor’s sole owner, he alone had legal title to, pos-
session of, and an interest in its income when it accrued. Thus,
he owned that income as separate property. As the Mostollers
had no other significant source of income, the Debtor ac-
counted for the lion’s share of their taxable income from 2013
to 2015. So under federal tax law, Mr. Mostoller owned most
of the refund separately.
B. The Mostollers never merged their separate
interests into entireties interests
After spouses get a refund, they can change their ownership
of that money under state property law. For instance, spouses
can merge their separate interests into entireties interests over
time, as long as they satisfy the four unities needed for a ten-
ancy by the entirety. Cf. In re Estate of Brose, 206 A.2d 301,
304 (Pa. 1965). But when the IRS issued the Mostollers’ refund
check, Mr. Mostoller owned almost all of it separately through
his sole ownership of the Debtor’s income, even though the
check was made out to both spouses. While the unities of time
23
and title were satisfied by then, the unities of possession and
interest were still missing. See In re Estate of Rivera, 194 A.3d
at 586. And the Mostollers could not have merged their inter-
ests: their accountant immediately deposited the refund check
with the bankruptcy court before they could commingle the
proceeds.
Because their interests both started and remained separate,
Mr. Mostoller validly pledged his share on his own. Thus, the
bankruptcy court properly ordered the Mostollers to turn over
half of the refund to the Trust.
* * * * *
Now that he has his loan, Mr. Mostoller wants to water
down his pledge. He admits that, on his novel reading of the
loan agreement, the promised collateral “would be valueless.”
Appellants’ Br. 4. But the bankruptcy court rightly found that
the Loan Order’s description of the collateral was ambiguous
and that the Trust’s reading tracked the parties’ understanding.
It also rightly held that, under federal tax law, the Mostollers
owned the tax refund separately, so Mr. Mostoller alone could
pledge it. Finally, in doing so, it properly exercised its jurisdic-
tion over core proceedings. The District Court affirmed in all
respects. So too will we.
24