Michael Q. McADAMS and Farrell D. Odom, on behalf of themselves and all others similarly situated, and Berry Boyd et al., Plaintiffs, Appellants, v. MASSACHUSETTS MUTUAL LIFE INSURANCE COMPANY and MassMu-tual Benefits Management, Inc., Defendants, Appellees, 391 F.3d 287

LYNCH, Circuit Judge. Michael Q. McAdams and Farrell D. Odom, former general agents for the Massachusetts Mutual Life Insurance Company (“MassMutual”), filed suit against MassMutual on behalf of themselves and all other similarly situated MassMutual general agents. The claims are Massachusetts state law claims and jurisdiction is premised on diversity.

Plaintiffs participated in a non-qualified deferred compensation plan; they allege that the plan was managed by MassMutual in violation of the contract underlying the plan, the implied covenant of good faith and fair dealing in the contract, and a fiduciary duty running from MassMutual to plan participants. They also allege violations of Massachusetts General Laws

Chapter 93A, which prohibits unfair and deceptive trade practices. Specifically, McAdams and Odom allege that MassMutual unlawfully assessed a tax charge against participants’ deferred compensation earnings that was designed to fully offset the tax costs to MassMutual from running the plan. Under the contract, they argue, no such charge can be assessed, and even if such a charge could be assessed, the charge assessed by Mass-Mutual was, in reality, far too high merely to offset MassMutual’s tax costs.

Mass-Mutual admits that such a tax charge was assessed but argues that the contract allowed it and that the amount of the charge was reasonable. After discovery, the district court dismissed all of McAdams’s and Odom’s claims on

summary judgment and also, after denying a motion for class certification, dismissed all of the actions of thirty related plaintiffs. Because the undisputed record shows that MassMutual’s actions remained well within the bounds of discretion permitted to MassMutual by the contract and by the underlying covenant of good faith and fair dealing, we now affirm the district court. I. We recount all facts in the light most favorable to the party opposing

summary judgment, McAdams and Odom in this case. ■ Dwan v. City of Boston, 329 F.3d 275, 277 (1st Cir.2003), McAdams and Odom worked as general agents for MassMutual in Texas; they ran and owned fairly large (forty- to eighty-person) agencies that sold insurance and other products for the company. Both of their general agencies sold deferred compensation plans to their customers.

These general'agents were not classified as employees of MassMutual and in fact were closer to independent contractors than to employees — for example, the general agents did not receive a salary from Mass-Mutual and did not send profit and loss statements for their general agencies to MassMutual, although MassMutual did pick up certain expenses, such as rent. A. The Deferred Compensation Plan MassMutual first established a deferred compensation plan for general agents in 1967; this first plan allowed for accumulation of deferred amounts of earned income at a fixed interest rate.

The plan at issue here was created in 1970. The new plan, unlike the old, provided an option, should MassMutual wish to provide it, of investing deferred compensation in stocks (a higher-risk, higher-return option) rather than investing in a fixed income account. Mc-Adams first began deferring compensation with the plan in 1983; Odom in 1986. All of the related plaintiffs (whose claims in the putative class have been joined in this consolidated appeal) were general agents who deferred compensation under this plan. At least 117 general agents were participants in this plan at some point.

The deferred compensation plan was created and advertised as a perk for the general agents. As one MassMutual employee noted in a communication with the general agents, “There is no purely corporate reason for [this] plan.

It is provided for the benefit of our General Agents solely.” Like most deferred compensation plans, the general purposes of this one were to provide a convenient “vehicle to save additional funds for retirement” and, most importantly, to “shelter current contributions from current income taxation ... and to allow assets to accumulate tax free.” Personal income tax would only need to be paid by participants on this compensation when it was actually disbursed to the general agents (upon retirement or death); in the meantime the deferred compensation could accumulate income on the investment free and clear of the personal income tax.

Because the general agents are not “employees”- as that term is used by the IRS, MassMutual could not create a “qualified” deferred compensation plan for the general agents but instead had to create a “non-qualified” plan. This had important tax consequences. Money can accumulate in qualified plans, which are considered separate entities from the employer, without any adverse tax consequences for the employer at any time, but the same is not true of non-qualified plans.

The money invested in non-qualified plans is considered part of the employer’s assets and thus the employer has to pay tax on the funds invested in the plan (although the employer may get a tax deduction later, when the deferred compensation is actually disbursed to the general agents).

Article V of the plan’s standard contract, which was signed by MassMutual and by all the plan participants (including McAdams and Odom), provided that Mass-Mutual could choose two different ways of dealing with compensation deferred under the plan: it could either invest such compensation separately from its general reserves and assets (say, in a trust or mutual fund) or it could commingle this compensation with its general reserves and assets.

Article V explains how investment earnings would be credited to general agents’ accounts under each of these two choices. The first paragraph of the

Article notes that if MassMutual invests the deferred compensation separately, then it shall credit to the General Agent’s account from time to time whatever net earnings the assets so invested have produced.

Net earnings from separate investments for purposes of this Agreement shall mean gross earnings, including but not limited to, interest, dividends, realized and unrealized appreciation or loss in the value of such investments, less all expenses and taxes attributable to the making, carrying and liquidation of such investments, including but not limited to, an appropriate accrual of taxes for unrealized appreciation in the value of the investments. [MassMutual] alone shall determine the amount of net earnings that will be credited to the General Agent’s account; and such determination of net earnings, including the determination as to charges for expenses and taxes, shall be final.

Thus, the

Article explicitly indicates that when MassMutual invests money deferred under the plan in separate accounts, Mass-Mutual is allowed to deduct any expenses, including taxes, that those investments cause. The purpose here seems to be to ensure that MassMutual does not lose money by administering any separate investments under the plan: if these investments lead to any expenses or taxes, these will be paid for by the plan participants and not by the plan administrator, Mass-Mutual. The third and last paragraph of

Article V discusses the crediting of earnings to deferred compensation accounts when this compensation is not separately invested but instead is kept within the company’s general reserves and assets. MassMutual will then “credit interest on that portion ... of the account not represented by separate investments of [MassMutual] assets. Such interest shall be credited to the account from time to time, in lieu of net earnings from separate investments of [MassMutual] assets, at a rate determined from time to time by the Company.” The

interpretation of this provision lies at the heart of the current dispute between Mc-Adams, Odom, and MassMutual. B. MassMutual’s Administration of the Plan MassMutual chose to invest the deferred compensation separately for only a few years early on, from 1970 through 1975; during this period the money was invested in a separate trust administered by a bank. From 1976 onward, MassMutual ceased making separate investments of deferred money. Instead, participants were generally offered two options. First, they could “shadow” certain equity accounts (Oppenheimer funds, for example).

In this shadow option, the rate of return on the participants’ accounts for a given year would track whatever was the rate of return for the shadowed account. In other words, MassMutual would pretend the money had actually been invested in the shadowed account even though it had not been. Second, participants could instead choose a fixed interest option, in which their accounts would be credited with a fixed rate of return every year. This fixed rate seems to have been designed to be •equivalent with the return on long-term, money-based (low-risk) accounts (like bonds or certificates of deposit).

In 1983, MassMutual began imposing a tax charge on the earnings of deferred compensation that used the shadow account option. This tax charge varied from 1983 to 1993, but tended to hover around 30% (although it was much lower in a few years). From 1994 to 2001, the rate of the tax charge on this option remained constant at 34.16%. MassMutual began in 1983 taking into account tax charges on the credited earnings of deferred compensation placed in the fixed account option; a formal tax charge for this option was put in place in 1987.

This formal rate was 40% for 1987, 34% for 1988 through 1993, and 35% for 1994 through 1999. These tax charges reduce the rate of earnings on deferred money: the tax charge is multiplied by the gross rate of earnings (either the periodic rate of return on the shadowed account or the fixed interest rate determined by MassMutual) to determine a net earnings rate that is actually credited to the general agents’ accounts. Thus, a nine percent gross rate of return would be reduced to about six percent after the tax charge.

If a general agent chose the shadow option and the shadowed fund were to have lost money during a given period, then the result , of the tax charge would actually be to reduce the agent’s loss. No tax charge is assessed against the principal; only the rate of earnings is reduced by a tax charge. The rate of the tax charge has been roughly equal to MassMutual’s marginal tax rate.

MassMutual has stated many times that the purpose of this tax charge is roughly to offset the tax costs to MassMutual from running the plan and to place it in a revenue-neutral position relative to where it would be if it instead paid compensation to the general agents immediately. It is true that MassMutual is later able to deduct compensation to its general agents, but only when it actually pays out this compensation.

The deferred compensation plan, by forcing MassMutual to keep general agent compensation in its coffers longer than it otherwise would, burdens it with two additional tax costs: 1) it gets a deduction on the principal only a number of years down the road, rather than immediately, and 2) it is now forced to pay taxes on any earnings that are accumulated on this extra principal. The tax charge is designed to offset these tax costs to Mass-Mutual. The undisputed facts show that Mass-Mutual’s tax charges have, in a rough sense, correctly achieved this stated goal of tax neutrality from MassMutual’s perspective.

A November 13, 1986 memorandum from Douglas Jangraw, an actuary for MassMutual who was heavily involved with calculating the tax charge, states that because MassMutual ultimately gets a deduction for “interest credited” to participant accounts (when that money is actually paid out), “the current tax charge is probably excessive and should either be eliminated or reduced.” But this analysis was either a mistake or, as Jangraw explains in an affidavit and in his deposition, reflected the results of a partial analysis that considered only the tax cost due to credited interest and not the tax cost due to the principal.

Several other memoranda written by Jangraw, on October 21, 1986, December 4, 1986, January 10, 1987, and December 1, 1993, all concluded and clearly demonstrated that the tax charge was appropriately calculated when both tax costs to MassMutual (the principal cost and the earnings cost) are considered.

A January 26, 1995 memorandum from Jangraw stated that the tax charge, although it did not make a general agent any worse off from a tax perspective than if she had taken her compensation immediately rather than deferring it, did “essentially put[ ] the participant in a neutral position relative to where he or she would have been had they taken their compensation immediately.” But an attachment to this memo, along with two other documents, a 1994 fact sheet sent from Mass-Mutual to its participating general agents and the “written version” of a 1994 discussion between a MassMutual officer and a group of general agents, clarifies this remark considerably.

The 1994 fact sheet notes that the tax-charged deferred compensation plan would not always be better for employees than taking compensation immediately; however, it would be better whenever “[the general agent’s] marginal tax rate is greater than the 34.16% tax charge.” In other words, it would generally be better whenever the total of the general agent’s marginal personal federal income tax rate, marginal personal state income tax rate, and marginal payroll tax rate were greater than the rate of the tax charge. This seems to have been a common situation.

The tax charge was a rough calculation that reflected some simplification; it is undisputed in the record that certain things were not considered in the calculation. First, the charge did not take into account the fact that unrealized gains in equities were not taxed until those gains were realized (the stocks were actually sold). There was a reason for that. Three memoranda from Jangraw — dated March 15, 1993, January 26, 1995, and April 26, 1995 — offer several reasons why unrealized gain had not historically been factored into the calculations.

First, the effect on the tax charge is “not that significant”: the January 1995 memo stated that studies had shown that the tax charge percentage would change by only about 3%, from 34.16% to “roughly 31%.” This change would have only a very small impact on the net interest rate credited to the general agents’ accounts. Jangraw also stressed that calculating this sort of gain would add quite a bit of administrative complexity to the calculation of the tax charge; the reduction would depend on the specific characteristics of funds and on the length of deferral for individual general agents.

The tax charge has also never taken into account a certain tax that the company was sometimes subject to, the add-on tax. This is a special tax, assessed between 1984 and 2000 and paid by mutual fund companies, based on surplus equity. The liabilities established by the plan reduce the amount of the surplus and therefore theoretically might justify a lower tax charge.

A 1987 internal committee (including Jangraw, Margaret Sperry, and four other MassMutual employees) concluded that the tax charge was generally correctly calculated, except for the add-on tax factor, which the committee felt should be included in the tax charge calculations. Some calculations in that report suggested that reflecting the add-on tax might result in a 1.50% increase in the net return rate of earnings that general agents received.

There was testimony from Sperry and Jangraw that these calculations were “hypothetical” and were based on an “assumption” that Jangraw said he did not remember and Sperry said was probably not backed by any data. Sperry stated that the add-on tax was zero for 10 of the 17 years between 1984 and 2000, and was less than one percent for five more years. So only in two years did the add-on tax exceed one percent. Jangraw testified that a conscious decision was made not to adopt the committee’s recommendation by including the add-on tax, but he does not remember why.

The answer may be simplification of the calculation. One simplification adopted by the company worked clearly to the advantage of the general agents. MassMutual did not reduce net rates of return for state taxes. Aside from the tax charge, the evidence demonstrates that for the fixed interest rate option but not the shadow account option, MassMutual also reduced the gross rates of returns that it posted to the general agents’ accounts by two other, very small charges. First, there was an administrative charge designed to offset Mass-Mutual’s costs of administration.

This ran at 20 basis points, resulting in a two-tenths of one percent reduction in the net fixed interest rate. For all non-qualified benefit plans that MassMutual administered, this apparently resulted in an annual provision of about $200,000 for the approximately five people who spent the majority of their time working on these plans, which seemed “quite reasonable” to Jangraw in a December 1,1993 memorandum.

Second, there was a charge which had been called conflicting things: it was sometimes referred to as a “profit charge” and sometimes as a “risk charge.” Jan-graw and another actuary (and senior vice-president) with MassMutual, Isadore Jer-myn, both explained that the purpose of this charge is not to allow MassMutual to make a profit, but rather to offset the risk of rapid asset loss associated with holding additional capital; this offset was required by the National Association of Insurance Commissioners.

The amount of this risk or profit charge was also quite small, running at 40 basis points before 1993 and 50 basis points after 1993. Jangraw concluded in his December 1, 1993 memorandum, after performing some calculations and considering the “risk-based capital requirements associated with these [deferred compensation] accounts,” that the “40-50 basis point [risk] charge does not seem unreasonably high.” Thus the total effect of the expense and risk charges was only to subtract six- or seven-tenths of one percent from the gross rate of return.

Significantly, for the fixed interest rate option, the rate was then increased above its otherwise net value (after tax, expense, and risk charges) by a subsidy. The amount of this subsidy has varied widely, but it hovered between 254 basis points and 48 basis points between 1987 and 1999. It would have approximately at least canceled out the expense and risk charges in all but one year, 1999. It was common for insurance companies with deferred compensation plans for general agents to impose tax charges.

Some 1994 research found that the rate of return that MassMutual credited on its fixed interest rate option was about the same as the rates of return for four of the five insurance companies that offered similar plans (the other company offered a substantially higher rate of return). MassMu-tual employees were attuned to standard industry practice and considered it when setting their rates of return: this was one purpose for the subsidy.

In 1999, MassMutual discontinued the tax charge due to complaints from general agents, but only for prospective contributions; for equity reasons, contributions made before 1999 would continue to be reduced by the amount of the tax charge. C. MassMutual’s Efforts to Inform Agents about the Tax Charge There is no evidence that MassMutual ever tried to hide the fact that it was assessing a tax charge; there is, in fact, considerable evidence that it communicated what it was doing to the general agents. This is very clear from 1993 onward.

Throughout 1993 and 1994, the tax charge was a constant topic of discussion between MassMutual and the general agents at the various meetings of the general agents. In 1993, a MassMutual employee wrote a letter to all general agent participants in the deferred compensation plan, explaining the development of the credited rates of return. The letter clearly identifies the tax charge as an issue, explains its rationale, and suggests that general agents “check with their investment/financial advisors” to see if the plan is beneficial to them.

In 1994, the company sent two letters, on July 15 and December 20, discussing the tax charge in question and answer format. The July 15 memorandum answered the question, “What is the tax charge? How is it determined?,” explained why a more complex, fund-specific type of tax charge was not used on the shadow account option, and noted that in the event of losses on a shadowed account, the effect of the tax charge would actually be to reduce the losses.

The December 20, 1994 memorandum went further; it included explanations of the three issues described above and also explained precisely when participating in the plan was likely to be worthwhile for agents. In general, the memo emphasized, both through words and through a tabular illustration, that participation was likely to be a good idea whenever the rate of the tax charge was lower than the participant’s own, personal marginal tax rate (including federal income, state income, and payroll taxes).

The memorandum also stressed that “[i]n the final analysis, the decision to participate or not, should be made after fully considering all of your tax and retirement planning strategies.” The evidence on MassMutual’s communications before 1993 with its general agents is considerably sparser.

One employee, Robert Massaro, testified that he began working with the compensation committee of the general agents’ association in 1986; thereafter, he would refer issues dealing with the tax charge back to the company when general agents would raise them “[f]rom time to time.” There was some more specific evidence that the tax charge was the subject of discussion between some general agents (perhaps only retired general agents) and company employees at a conference in the late 1980s. A letter sent from MassMutual to Odom’s general agency in 1989 referenced this conference and explained the tax charge.

In a letter, McAdams stated that he became aware of the tax charge a “few years” after 1983. A 1990 letter from MassMutual to McAdams in March 1990 noted in passing that “we adjust the unit values for taxes” and this indicated to Mc-Adams that a tax charge was applied. Mc-Adams certainly understood the tax charge by 1991 or 1992, when he discussed the issue at length with MassMutual’s CEO at a general agents’ conference. Finally, the company always sent periodic balance statements to each of its participants; these statements did not include any reference to the rate or dollar value of the tax charge, however.

They merely showed the total gain or loss on a participant’s account. II. McAdams and Odom filed suit against MassMutual in U.S. District Court for the District of Massachusetts on December 28, 1999, on behalf of themselves and all others similarly situated. They alleged that MassMutual’s use of the tax charge and related charges constituted a breach of contract, a breach of the covenant of good faith and fair dealing, a breach of fiduciary duty, fraud, negligent misrepresentation, and a violation of Massachusetts General Laws

Chapter 93A,

section 11. MassMutual moved to dismiss the complaint for failure to state a claim. Senior Judge Freedman granted this motion in part and denied it in part on December 13, 2000. McAdams v. Mass. Mut. Life Ins. Co., No. Civ. A.99-30284-FHF, 2000 U.S. Dist. LEXIS 22068 (D.Mass. December 13, 2000). The court held that on the contract count, the “language, at best [for MassMu-tual], presents an ambiguity with regard to whether the Agreement authorizes Mass-Mutual to deduct the tax charge from funds invested in the general reserves and assets.” Id. at *12.

The court emphasized that the agreement expressly provides for a deduction of taxes where deferred accounts are separately invested, but the portion of the agreement dealing with situations where the deferred accounts are not separately invested but instead are placed in MassMutual’s own general reserves and assets does not contain this express language. The court also noted the principle that ambiguous language is to be construed against the drafter, in this case MassMutual. Id.

The court held that plaintiffs’ breach of covenant of good faith and fair dealing count also had to stand, because the allegation that MassMutual.breached the contract and “capitalized on [its] position as the administrators of the [p]lan would amply justify recovery.” Id. at *14. The court refused to dismiss the fiduciary duty count as well, holding that McAdams and Odom had alleged enough facts to withstand dismissal of the fiduciary claim because “the defendants maintained the dominant voice in the management of the deferred compensation incomes.” Id. at *17. The district court did , dismiss the

Chapter 93A count because

Chapter 93A does not apply to “purely private transactions.” This plan was a purely private transaction because it was never offered to the general public and only applied to a small subset of general agents. Id. at *20-*21. The district court also dismissed the fraud and negligent misrepresentation counts as pled without the requisite specificity of fact. Id. at *19-*20. The court granted leave to amend the complaint, but McAdams and Odom did not do so. McAdams and Odom then moved.tq certify the class; certification was denied by Judge Freedman on May 15, 2002. Mc-Adams v. Mass. Mut. Life Ins. Co., No. Civ.

A.99-30284-FHF, 2002 WL 1067449, 2002 U.S. Dist. LEXIS 9944 (D.Mass. May 15, 2002). On June 13, 2002, thirty-five former general agents moved for leave to join or intervene in the McAdams/Odom suit. Judge Freedman denied that motion on March 26, 2003. On April 7, 2003, thirty of these thirty-five filed individual actions in district court based on the same basic legal theories. After a period for discovery, MassMutual moved for

summary judgment on all remaining claims, and District Court Judge Ponsor granted the motion in its entirety on March 26, 2004. The court focused on what it saw as the central issue, the problem of contract

interpretation. It held that the language in the general reserves

section of the contract authorizing MassMutual to credit “interest ... to the account from time to time ... at a rate determined by the Company” was unambiguously “a broad grant [of discretion] to MassMutual.... [It] gives the defendants expansive discretion to determine the rate in any rational manner it sees fit.” The express mention in the contract of deductions for taxes and expenses where money is separately invested and the absence of this express language in the

section dealing with situations where the money is kept in MassMutual’s general reserves did not properly lead to an application of the expressio unius est exclusio alterius maxim to deny MassMutual’s ability to deduct taxes from these types of accounts because the “glaring omission of any limitation on [MassMutual’s] discretion ... trump[ed] any attempt to attribute significance to a lack of an itemization of factors.” The court concluded that “[a]bsent demonstrable bad faith, fraud or egregiously unreasonable behavior,”, MassMutual could do whatever it wanted with the rate of return on the deferred funds.

The court concluded that the claims for breach of fiduciary duty and breach of the covenant of good faith and fair dealing also failed. The court noted that it was “questionable” whether MassMutual was a fiduciary, but even if it was, neither fiduciary duties nor the duties imposed by the covenant of good faith and fair dealing were violated. Plaintiffs had offered only a “difference in opinion” as to how to calculate the rate of return; there was no showing that MassMutual’s method was a result of bad faith.

Moreover, the court emphasized the “transparency” of MassMutual’s conduct, and the fact that McAdams and Odom continued to defer even after they were well aware of the tax charge. Finally, the court concluded that McAdams’s own claims would be dismissed regardless because of a release he signed with Mass-Mutual in 1998. On May 5, 2004, after having issued an order to show cause why

summary judgment should not be granted, the district court granted

summary judgment against the thirty general-agent plaintiffs in the related cases as well. All of the plaintiffs filed timely appeals, and the appeals of the thirty related plaintiffs were consolidated with the appeal of McAdams and Odom by this court. The plaintiffs have not raised the issue of fraud or negligent representation before us; this claim is waived. The claims of breach of contract, breach of covenant of good faith and fair dealing, breach of fiduciary duty, and violation of

Chapter 93A are before us. III. All of the claims were denied either on a Rule 12(b)(6) motion to dismiss or on

summary judgment; our review is de novo. Rolling v. Am. Power Conversion Corp., 347 F.3d 11, 13 (1st Cir.2003); Martin v. Applied Cellular Tech., 284 F.3d 1, 5 (1st Cir.2002). A. Breach of contract Whether a contract is ambiguous is an issue to be determined by the court. Alison v. Byard, 163 F.3d 2, 6 (1st Cir.1998). Ordinarily, contracts are construed by a court “as a matter of law” unless there are material disputes as to extrinsic facts bearing on the correct

interpretation. Fenoglio v. Augat, Inc., 254 F.3d 368, 370 (1st Cir.2001) (quoting Principal Mut. Life Ins. Co. v. Racal-Datacom, Inc., 233 F.3d 1, 3 (1st Cir.2000)). This is true even in “close cases”; the jury does not become involved when words and context alone are used, but only when extrinsic evidence is át issue. See Fishman v. LaSalle Nat’l Bank, 247 F.3d 300, 303 (1st Cir.2001). Words are not viewed in isolation within a contract. ■ The meaning of a contract cannot be understood merely by “isolating words and interpreting them as though they stood alone.” Starr v.

Fordham, 420 Mass. 178, 648 N.E.2d 1261, 1269 (1995) (quoting Boston Elevated Ry. v. Metro. Transit Auth., 323 Mass. 562, 83 N.E.2d 445, 451 (1949)). Contracts must, in other words, be read as a whole. Id. Even if a contract might arguably appear ambiguous from its words alone, the decision remains with the judge if the alternative reading is inherently unreasonable when placed in context. Contract

interpretation under Massachusetts law depends heavily on context and recognizes that words can have different 'meanings in different contexts. See id.; Shea v. Bay State Gas Co., 383 Mass. 218, 418 N.E.2d 597, 600 (1981). Thus, agreements should be construed “with reference to the situation of the parties when they made it and to the objects sought to be accomplished.” Starr, 648 N.E.2d at 1269 (quoting Bryne v. City of Gloucester, 297 Mass. 156, 8 N.E.2d 170, 171 (1937) (internal quotation marks omitted)).

This sort of contextual evidence is not the kind of evidence that is intended to be barred by the parol evidence rule if no ambiguity is found; rather, context should be used as a tool to find unambiguous meaning in the first place. See Robert Indus., Inc. v. Spence, 362 Mass. 751, 291 N.E.2d 407, 409-10 (1973); see also SAPC, Inc. v. Lotus Dev. Corp., 921 F.2d 360, 361 n. 2 (1st Cir.1990). On the basis of such context, this court has often affirmed entry of

summary judgment in contract cases. See Nat’l Tax Inst., Inc. v. Topnotch at Stowe Resort & Spa, 388 F.3d 15, 18-20 (1st Cir.2004); Fishman, 247 F.3d at 302-03. Here, importantly, we see no material disputes as to extrinsic facts. The contract may be interpreted on the face of the document itself against a contextual understanding of the nature of various types of deferred compensation plans. Plaintiffs argue the contract is not ambiguous and it does not permit the assessment of any tax charges when separate investments are made.

They also argue that if there are' any ambiguities, they must be construed against MassMu-tual. Only as a last resort do they argue that if there are any. ambiguities, and they cannot be resolved against MassMutual, then a: jury should resolve them. The precise language being construed is contained in paragraph three of

Article V and provides: If during any period '..., [MassMutual] ... makes no separate investments of its property, ... it shall then credit interest, on that portion or all of the account not represented by separate investments of [MassMutual] assets. Such interest shall be credited to the account from time to time, in lieu of net earnings from separate investments of [MassMutual] assets, at a rate determined from time to time by [MassMutual]..:.

This language must be understood in the context of paragraph one of that article, which governed separate investments: If [MassMutual] invests its property separately ..., it shall credit to the General Agent’s account from time to time whatever net earnings the assets so invested have produced.

Net earnings from separate investments for purposes of this Agreement shall mean gross earnings, including but not limited to, interest, dividends, realized and unrealized appreciation or loss in the value of such investments, less all expenses and taxes attributable to the making, carrying, and liquidation of such investments, including but not limited to, an appropriate accrual of taxes for unrealized appreciation in the value of the investments. [MassMutual] alone shall determine the amount of net earnings that will be credited to the General Agent’s account; and such determination of net earnings, including the determination as to charges for expenses and taxes, shall be final.

The district court found paragraph three to be unambiguous: “It is a broad grant to MassMutual to determine, as it sees fit, the rate of interest credited to the accounts of the general agents.” More specifically, the court determined that the clause requiring the company to credit interest “at a rate determined from time to time by [MassMutual]” permitted Mass-Mutual to deduct expenses and taxes in determining that interest. We agree with this narrower formulation.

Specifically, we reject the plaintiffs’ argument that in the face of an explicit contractual grant of authority to the company, the company could not deduct expenses and taxes unless the contract permitted it to do so in haec verba.

Plaintiffs point to the term “interest,” ignoring the phrase that interest is to be credited “at a rate determined from time to time by the company.” “Interest” has been defined as “[t]he compensation fixed by agreement or allowed by law for the use or detention of money ... esp., the amount owed to a lender in return for the use of borrowed money.” Black’s Law Dictionary 829 (8th ed.2004). The term does not preclude deduction of expenses incurred in determining the rate of return. Here the deferred compensation is invested with the common assets of the company.

The making of investments itself involves costs and any sensible rate of return accounts for those costs. The fact that this is an unqualified plan means that the company has additional taxes it must pay (unlike a qualified plan). The decision to account for those taxes in setting the rate of interest is within the literal language affording the company the discretion to set the rate of interest.

The reasonableness of this reading is reinforced by the fact that the company is explicitly granted the authority in the first paragraph to deduct from net earnings, when funds are invested separately, “all expenses and taxes attributable to the making, carrying and liquidation of such investments.” The clause in paragraph three gives the company even broader authority. It would be unreasonable to interpret that broader grant of authority in paragraph three as not encompassing powers explicitly authorized in the narrower grant of authority in paragraph one.

In this situation, the principle of expressio unius est exclusio alterius has no place. Nothing in industry custom operates to render this clear reading suspect. Most others in the industry follow the same practices as MassMutual. At least one company does not, but we do not know its precise contractual arrangements. The canon of construing an ambiguous contract against the drafter does not change the result. First, there is no ambiguity.

Second, as said in National Tax Institute, [T]he canon is a qualified one ... a default rule that arguably has more force where the parties differ in sophistication or where standard forms are used (e.g., insurance contracts). In any event, the canon has little to do with actual intentions and should only be used, as a last resort, if other aids to construction leave the case in equipoise. Nat’l Tax Inst., 388 F.3d 15, 18. Here, the parties are extremely sophisticated and there is no need to turn to a last resort canon. B.

Implied Covenant of Good Faith and Fair Dealing McAdams and Odom respond that MassMutual’s assessment of the tax charge was a breach of the implied covenant of good faith and fair dealing in two senses: that there could be no consideration of taxes at all in the setting of the interest rate and that the particular rates set violate the covenant. Under Massachusetts law, every contract implies such a covenant. Anthony’s Pier Four, Inc. v. HBC Assocs., 411 Mass. 451, 583 N.E.2d 806, 820 (1991). The purpose of the covenant is not to add terms to a contract; indeed, it may not do so.

The Massachusetts Supreme Judicial Court recently reinforced this understanding of the covenant. In UNO Restaurants, Inc. v. Boston Kenmore Realty Corp., 441 Mass. 376, 805 N.E.2d 957 (2004), the court held that a directed verdict should enter against a claim of breach of the covenant. The court stressed that [t]he covenant may not ... be invoked to create rights and duties not otherwise provided for in the existing contractual relationship, as the purpose of the covenant is to guarantee that the parties remain faithful to the intended and agreed expectations of the parties in their performance.

Id. at 964; see also Sparks v. Fid. Nat’l Title Ins. Co., 294 F.3d 259, 274 (1st Cir. 2002) (“An implied covenant ... imposes on the parties the obligation ... to act in good faith to accomplish the purposes of their agreement. It does not, however, add new substantive obligations to their contractual undertakings.”) (citation omitted). This contract, read in light of the covenant of good faith and fair dealing, constrains MassMutual’s contractual discretion somewhat. See 2 E.

Allan Farnsworth, -Farnsworth on Contracts § 7.17, at 365 (3d ed.2004) (covenant is often used-to limit discretion where the terms of . the contract give one side discretion over another). There is obviously a requirement that the company act in good faith; the lack of good faith can be inferred from “unreasonable[ness] under all the circumstances.” Nile v. Nile, 432 Mass. 390, 734 N.E.2d 1153, 1160 (2000); see Krapf v. Krapf, 439 Mass. 97, 786 N.E.2d 318, 324-25 (2003).

While, as MassMutual urges, an “arbitrary and capricious” use of discretion would certainly violate the covenant, the Massachusetts courts have also used the language of “unreasonableness,” which is the usage we employ. See Nile, 734 N.E.2d at 1160. The core of the covenant is to ensure that one party does not deprive another of the “fruits of the contract.” Anthony’s Pier Four, Inc., 583 N.E.2d at 820 (quoting Druker v. Roland Wm.

Jutras Assocs., Inc., 370 Mass. 383, 348 N.E.2d 763, 765 (1976) (internal quotation marks omitted)) (emphasis added). “The covenant is preserved so long as neither party injures the rights of another to reap the benefits prescribed by the terms of the contract.” UNO Rests., 805 N.E.2d at 964 (emphasis added). The general agents argue that such deprivation is precisely what happened here.

They argue that by assessing the tax charge, MassMutual was wiping out the whole point of the deferred compensation plan, which was to allow the general agents to invest money tax free until it was actually distributed to participants. The agents believe that the company’s imposition of an annual tax charge based on its corporate tax liability completely frustrated this purpose and deprived the agents of the benefits of this type of plan. This argument is misplaced - because it focuses merely on the phrase “deferred compensation plan” and not on the contractual language of the second sentence of paragraph three.

We have interpreted the contract as giving MassMutual discretion to consider taxes, expenses and other costs in setting the rate of interest. A proper exercise of this discretion cannot deprive the agents of the fruit of the contract. As UNO Restaurants makes clear, the covenant cannot be used to contradict clear contractual terms. The agents’ first argument on the covenant claim was thus resolved in our discussion of the contract claim. There remains the possibility that Mass-Mutual might have exercised its discretion to set a rate so unreasonably as to violate the covenant.

Plaintiffs make exactly that argument. McAdams and Odom argue that unreasonableness is shown by the fact that MassMutual was told that its basic methodology for calculating the tax charge was wrong and resulted in an excessive charge on several occasions; its continuing use of the methodology showed bad faith. But the evidence indisputably shows that the basic method was, after several minor simplifications, correct; the occasional criticisms of it were themselves based on errors.

The general agents next stress that the refusal to reduce the tax charge due to the add-on tax and the tax treatment of unrealized gain shows MassMutual’s bad faith. The evidence is undisputed that MassMu-tual made these two simplifications because it understood that they cut down on the company’s administrative costs without having a significant impact on the tax charge rate. And not all of the simplifications ran against the general agents: the failure to consider state corporate income taxes when setting the tax charge probably aided general agents considerably.

The agents also unsuccessfully argue that the expense and risk/profit charges on the fixed interest option violate the covenant of good faith and fair dealing. In theory these are, like the tax charge, reasonable attempts to allow MassMutual to break even and avoid losses from administering the deferred compensation plan. In practice, the assessment of these charges has not been unreasonable; the charges have been approximately correct after some sensible simplifications.

At any rate, these two charges have been quite small and usually have been far more than offset by the subsidy added by MassMutual to fixed interest rate returns. We note finally that a lack of transparency, with other evidence, may be important evidence in supporting a breach of the covenant of good faith and fair dealing. Bad faith may be shown partially by the fact that a party is trying to hide what she is doing from the other party in order to disadvantage that party.

For example, where a party uses discretionary rights under a contract pretextually, as cover for a hidden and illicit motive, an inference of bad faith may be warranted. See Anthony’s Pier Four, Inc., 583 N.E.2d at 820-21; Starr v. Fordham, 420 Mass. 178, 648 N.E.2d 1261, 1266 (1995) (finding breach of covenant on partnership contract where partners unfairly reduced one partner’s share by fabricating a list of negative factors and by “selectfing] performance criteria in order to justify the lowest possible payment to the plaintiff’). Here, Mass-Mutual was reasonably transparent about what it was doing and why.

The 1998 and 1994 communications marked a substantial attempt to inform the general agents about the workings of the tax charge and about whether it would still be a good idea for the agents to participate in the plan. Even before 1993, MassMutual was not trying to hide anything; there is evidence that it tried to communicate the tax charge to the general agents on some occasions in the 1980s. Based on the evidence, no reasonable finder of fact could conclude in the plaintiffs’ favor on the covenant claim. C.

Breach of Fiduciary Duty Under Massachusetts law, the question of whether one party owes fiduciary duties to another is a question of fact. Indus. Gen. Corp. v. Sequoia Pac. Sys. Corp., 44 F.3d 40, 44 (1st Cir.1995). Key factors in this fact-specific inquiry include one party’s lack of sophistication relative to another on the relevant issues, and whether one party has granted another party a great deal of discretion. Patsos v. First Albany Corp., 433 Mass. 323, 741 N.E.2d 841, 849-51 (2001).

We need not decide whether there is a disputed issue of fact concerning the existence of a fiduciary relationship between MassMutual and any or all of the general agents, because even if fiduciary duties existed, there was no breach. The core fiduciary obligation in this sort of context is a duty on MassMutual to exercise any discretionary power that it has in good faith, with prudence and after serious consideration, and reasonably according to the purposes of the agreement and standard fiduciary principles. See Ventura v. Ventura, 407 Mass. 724, 555 N.E.2d 872, 875 (1990); Briggs v.

Crowley, 352 Mass. 194, 224 N.E.2d 417, 421-22 (1967); Copp v. Worcester County Nat’l Bank, 347 Mass. 548, 199 N.E.2d 200, 203 (1964). This obligation was met here: the tax charge was both created and applied in a manner that was well thought out and reasonable in light of the plan’s purpose, and there was no bad faith. Sometimes, there may be a breach of a fiduciary duty if the fiduciary fails to disclose certain information. Puritan Med. Ctr., Inc. v. Cashman, 413 Mass. 167, 596 N.E.2d 1004, 1010 (1992).

MassMutual has satisfied any such obligation by making, as already explained, considerable efforts to pass information about the tax charge on to the general agent participants. D.

Chapter 93A Massachusetts General Laws,

Chapter 93A,

section 11 protects against unfair and deceptive trade practices. A dispute must involve a “commercial transaction” to fall into the reach of the statute. Szalla v. Locke, 421 Mass. 448, 657 N.E.2d 1267, 1270 (1995). The Massachusetts Supreme Judicial Court has interpreted this to include only transactions that are “offered generally by a person for sale to the public in a business transaction” and to exclude transactions that “are principally ‘private in nature,’ ” such as a dispute between employer and employee. Manning v. Zuckerman, 388 Mass. 8, 444 N.E.2d 1262, 1265-66 (1983).

MassMutual argues that the dispute at issue here is private in nature, and thus no 93A claim can lie; plaintiffs counter that such a claim is stated because they are independent contractors and not employees. Whether an independent contractor can recover for a 93A violation is unclear under Massachusetts law. See Schinkel v. Maxi-Holding, Inc., 30 Mass.App.Ct. 41, 565 N.E.2d 1219, 1224-25 (1991). The issue may hinge not on the label of “independent contractor,” but on a fact-specific, case-by-case analysis into the type of relationship that the independent contractor has with the company at issue.

See Linkage Corp. v. Trs. of Boston Univ., 425 Mass. 1, 679 N.E.2d 191, 207 (1997) (corporation that worked for a university as, according to the contract, independent contractor rather than employee or agent, could maintain 93A action against university); Benoit v. Landry, Lyons & Whyte Co., 31 Mass.App.Ct. 948, 580 N.E.2d 1053, 1053 (1991) (real estate salesman could not use 93A even if he was an independent contractor). Because the state law is unclear, we decline to decide the 93A claim on this ground.

Even if the transaction between MassMutual and the general agents was commercial in nature and the claim should have survived a Rule 12(b)(6) motion to dismiss for failure to state a claim, Mass-Mutual’s actions in this case did not violate

Chapter 93A, as amply demonstrated by the

summary judgment record. Plaintiffs do not argue that there is any additional evidence pertinent to this claim that is not on this record. Conduct that is “ ‘in disregard of known contractual arrangements’ and intended to secure benefits for the breaching party” falls afoul of

Chapter 93A. Anthony’s Pier Four, Inc. v. HBC Assocs., 411 Mass. 451, 583 N.E.2d 806, 821 (1991) (quoting Wang Labs., Inc. v. Bus. Incentives, Inc., 398 Mass. 854, 501 N.E.2d 1163, 1167 (1986)). Here, Mass-Mutual did not disregard the contract and behaved reasonably in its application of the contract. A 93A claim is based on unfairness or deception, not a mere “difference of opinion”: “a good faith dispute as to whether money is owed, or performance of some kind is due, is not the stuff of which a c. 93A claim is made.” Duclersaint v. Fed. Nat’l Mortgage Ass’n, 427 Mass. 809, 696 N.E.2d 536, 540 (1998).

MassMutual’s actions were all fair, reasonable, in good faith, and transparent; the general agents’ differences of opinions do not come close to raising a violation of

Chapter 93A. Given the way we have resolved the case, we need not consider the general agents’ motion to certify the class, Mass-Mutual’s statute of limitations defense, or MassMutual’s assertion that McAdams released his own claims. IV. The district court’s grant of

summary judgment on all claims for MassMutual is affirmed. So ordered. Costs are awarded to MassMutual. . The tax charge for the fixed interest rate option has merely equaled MassMutual's federal corporate income tax rate on simple income. MassMutual thus simplified by assuming that all deferred money placed in the fixed interest option was held as cash (rather than bonds or something else). In reality, this money was held as part of MassMutual's general cash reserves, which apparently reflected a mix of asset types.

The tax charge on the shadow option was a bit more complex; the company followed the practice of treating this money as though it was actually invested in a shadowed fund, and so the tax charge on all shadowed accounts equaled an approximate "weighted average” of the federal marginal tax rates on ■ the various types of income found in all of these funds, such as dividend income, capital gains, and bond income, as well as simple income. In treating this income as though it .were actually being taxed at the same rate as the shadowed funds, MassMutual was, of course, using a fiction.

This shadowed money was really held in MassMutual’s general reserves, and thus its effective tax rate was dependent on the mix of assets in those general reserves. But the fiction was not unfair; it treated' general agents approximately the same way they would have been treated if they had actually put their money in the fund that was being shadowed. At any rate, this difference in method of calculation had little impact on the ultimate rate: the rates for the fixed interest and shadow options were generally similar. .

The record has one other piece of evidence which plaintiffs argue demonstrates that the tax charge was incorrectly calculated. An outside consultant hired by MassMutual during a 1992 committee review of deferred compensation plans stated, in a “position paper,” that "we feel that the way the rate is currently calculated results in a rate which is too low.

We do not feel that the ultimate company tax deduction was taken into account when the rate is calculated.” This point, which is made only cursorily and without any supporting documentation in the paper, simply seems to be based on a misunderstanding of the charge, as noted by Margaret Sperry, a manager and lawyer for MassMutual, who corrected the consultant about his mistake. It is, in any event, insufficient to support the inferences needed to withstand

summary judgment. . Jangraw noted in his deposition testimony that the purpose of the plan was to keep the cost to the company "approximately” neutral while at the same time not having too high "an administrative burden on the plan administrators.” There was thus some interest by MassMutual in not making the plan unreasonably complex. . The memo actually noted that the issue of whether the plan was beneficial to the general agents was a bit more complex.

The discussion included "some important qualifiers": 1) "changes in tax rates have a big impact on the financial advantages of [the plan],” 2) because the plan is nonqualified, the general agents would have mere unsecured general creditor status if MassMutual were to declare bankruptcy, and 3) general agents should put as much money as possible into 401(

k) plans before putting money into this plan, because 401(k)’s have more advantageous tax treatment. . The choice of law clause in the agreement states that Massachusetts state law applies to "[a]ll questions pertaining to the construction, validity and effect of the provisions of this Agreement.” In the briefs before this court, both sides have cited Massachusetts law for all of their claims, including the fiduciary duty and 93A claims. Because of this implicit agreement between the parties that Massachusetts law should apply to all claims, we apply it in this diversity case without inquiring further. Hershey v.

Donaldson, Luflcin & Jenrette Sec. Corp., 317 F.3d 16, 20 (1st Cir.2003). . Even where there are ambiguities in the language and there is relevant extrinsic evidence, there is support for the view that under Massachusetts law the judge decides the issue if all of this extrinsic evidence is undisputed, even if the outcome is debatable. See Fishman, 247 F.3d at 303; Atwood v. City of Boston, 310 Mass. 70, 37 N.E.2d 131, 134 (1941).

And of course, judges interpret even ambiguous contracts involving disputed extrinsic evidence where the evidence is so one-sided that no reasonable finder of fact could decide for the alternative

interpretation. Boston Five Cents Sav. Bank v. Sec’y of Hons, and Urban Dev., 768 F.2d 5, 8 (1st Cir.1985). These rules about extrinsic evidence are not necessary to decide this case, because the meaning of this contract is clear after looking at its language and context. . We note that the breach of implied covenant claim alleged here is not the theory recognized in Fortune v. Nat’l Cash Register Co., 373 Mass. 96, 364 N.E.2d 1251, 1255-59 (1977). There is no claim here that plaintiffs' employment was terminated to avoid paying them benefits earned and due.