(dissenting). This case concerns the statutory duty to bargain about the calculation of retirement benefits under the public employment relations act (PERA), MCL 423.201 et seq. I agree with the majority that the calculation of retirement benefits is a mandatory subject of collective bargaining, that the calculation of retirement benefits is covered by the *93parties’ collective bargaining agreements (CBAs), and that the term “actuarial equivalent” is unambiguous. Without evidence of a mutually agreed upon intentional practice that modified this unambiguous contract term, I would also agree with the majority about the outcome here. But I conclude that the parties’ 24-year intentional practice of using a very specific formula for achieving “actuarial equivalence” amended the contract and requires bargaining anew before a unilateral change may be made to that practice. Therefore I respectfully dissent and would affirm the Court of Appeals’ judgment on this basis.
When there is a statutory duty to bargain under the PERA, the analysis from Port Huron Ed Assoc v Port Huron Sch Dist1 applies when an unfair labor practice (ULP) is alleged. A public employer may defend against an ULP charge by fulfilling the statutory duty to bargain and memorializing the terms of that bargain in a CBA.2 When an issue is covered by the CBA, the parties’ past practice may amend the CBA, such that a public employer is nevertheless bound to bargain under PERA before making a unilateral change to that practice. When contract language is ambiguous or silent “there need only be tacit agreement that the practice would continue.”3 When the agreement is unambiguous with respect to the term affected by a conflicting practice, more is required. The contract language will control:
[U]nless the past practice is so widely acknowledged and mutually accepted that it creates an amendment to the contract.
*94While, to be sure, parties to a contract may modify it by a later agreement,... the conduct relied upon to show such modification must be unequivocal and the terms of modification must be definite, certain, and intentional.[4]
As this Court explained in Detroit Police Officers Ass’n v Detroit, when applying the Port Huron analysis:
The [Port Huron] majority approvingly cited a case that stated that the parties’ agreement to modify the contract can be deduced from their course of conduct if it is unequivocal and the terms of modification are definite, certain, and intentional. Further, the majority indicated that the party seeking to supplant the contract language must prove that the other party intentionally chose to reject the negotiated contract and knowingly acted in accordance with the past practice.[5]
Thus, if the parties’ course of conduct shows that they intentionally chose to modify a provision in the CBA because their past practice contradicted the plain meaning of that provision, a party to the CBA cannot later rely on the plain meaning of that provision and ignore the past practice.
Because I agree with the majority that the term “actuarial equivalent” is unambiguous, the charging parties in this case must meet a higher standard of proof to show that the parties’ practice amended that contract term. They have done so. As the evidence of the parties’ mutual agreement regarding the specific actuarial formula to be used to calculate retirement benefits is longstanding and substantial, I would hold that the charging parties have “submit[ted] proofs illustrating that the parties had a meeting of the minds with respect to the new terms or conditions— *95intentionally choosing to reject the negotiated contract and knowingly act in accordance with the past practice.”6 We have guidance on whether the evidence relating to the past practice meets the higher standard given our past decisions. This case lies somewhere between the facts of Port Huron and Detroit Police. In those cases we addressed whether past practice modified unambiguous contract language, and we came to opposite conclusions.
In Detroit Police, this Court held that the past practice modified the unambiguous language of the contract.7 The contract at issue there provided that the board of trustees would determine whether an incapacitation resulted from the performance of duty. However, the parties’ longstanding practice was inconsistent with this language, such that the medical director would make the determination and the medical board’s decision on the issue was subsequently binding on the board of trustees.8 In 1991, the board of trustees attempted to recapture its authority to make the duty determination by unilaterally passing a resolution, resulting in the ULP charge.
In finding that the past practice modified the unambiguous contract language, this Court found the following facts to be important: (1) the board of trustees meeting minutes from prior years accepting decisions by the medical board as final and binding; (2) the city attorney’s admission to the past practice; (3) the disapproval by the board’s attorney of the resolution and reference to the ‘current and well established practice’ of the medical director making the decisions; (4) board member testimony that from 1983 to 1990 medical *96board decisions were regarded as final and binding; and (5) evidence that the board developed forms for use by the medical board, which expressly indicated that the medical board was to make a “duty-connectedness finding.”9
In contrast, in Port Huron this Court held that the past practice did not modify the unambiguous language of the contract. Port Huron concerned the proration of health insurance benefits for teachers hired midyear. A 1978 CBA had provided in unambiguous terms that such benefits would be prorated for midyear hires.10 In the 1987-88 school year, the district hired 8 teachers midyear, prompting the school district to inform the new hires that their benefits would be prorated per the contract language.11 This Court affirmed the Michigan Employment Relations Commission’s determination that the school district’s payments of full insurance benefits for midyear hires prior to 1987 was inadvertent, and that the teacher’s association had not presented sufficient proof that “the district knowingly paid employees hired midyear insurance benefits for the entire summer in disregard of contract language to the contrary, with the intent that such payment would supplant the agreement.”12 Notably, the hearing officer found that “the district’s failure to prorate benefits before 1988 was simply a mistake or oversight.... There was no evidence the district was aware it had not followed the express language of the agreement.”13
*97The facts regarding the practice at issue in this case are far closer to Detroit Police than to Port Huron. The past practice of calculating optional pension benefits using a 100% female/0% male mortality table was certainly not a mistake and it was used for over two decades.14 At least one of the trustees of the retirement commission opposed the unilateral adoption of the new mortality table because of this longstanding practice. Additionally, an actuarial study was conducted in 1993 and no action was taken to change the practice of using 100% female/0% male mortality tables as a result, even though the purpose of the study was to review the system’s actuarial assumptions for calculating employer contributions and employee optional benefits. This continued use of the 100% female/0% male tables after the previous study, given that formula’s uneasy relationship with the term “actuarial equivalence,” underscored the intentional commitment to that formula by the parties.
The 100% female/0% male mortality table has never achieved actuarial equivalence; in fact, the parties selected it to accomplish other goals. Thus, while actuarial equivalence may have an unambiguous meaning, the application of that table was contrary to the plain meaning of that term, and the employers cannot now rely on the term’s plain meaning when it is convenient or beneficial. Specifically, the 1982 actuarial study indicated that only a 100% female/0% male blend would be *98able to provide female employees with the same benefits they had received in the past, under gender-based tables which the parties agreed they could no longer use, because any other blend would reduce benefits to female employees.15 The retirement commission did not want that result. These same issues were on the table in 1993 when the parties studied the actuarial tables again. That no change was adopted following the results of this 1993 study, even though the mortality table was still derived from the 1971 data and assumptions designed not to achieve actuarial equivalence but rather to give female employees the same benefits they had received under gender-based mortality tables, is further evidence that the practice of using the 100% female/0% male mortality table was deliberate and agreed on.
There is additional evidence that the practice of using the specific actuarial 100% female/0% male tables was intentional and not inadvertent. For example, the 1982 actuarial report that originally led the retirement commission to adopt the 100% female/0% male table contained the following statement:
COMMENT C: The Retirement System Ordinance provides that an optional benefit will be “the actuarial equivalent” of the standard benefit. The Retirement Commission could adopt a rule stating that for purposes of determining amounts of optional benefits, the actuarial equivalent will be based upon a stipulated interest rate and unisex mortality table. This could eliminate the need for an ordinance change.
The report proposed a solution to the problem presented by the ordinance’s stated goal of “actuarial *99equivalence” and the adopted practice’s departure from that goal. The retirement commission took the study’s recommendation and amended the retirement ordinance in 1982 to reflect the newly adopted 100% female/0% male actuarial assumptions:
For purposes of determining actuarial [sic] equivalent Retirement Allowances, the Retirement Commission is currently using a 7V2% interest rate and a blending of male and female rates based on the 1971 group annuity mortality table projected to 1984 with ages set back 2 years.[16]
This extremely specific language amended the retirement ordinance because the actuaries and the retirement commission trustees realized that the 100% female/0% male table they were committed to using posed a problem with respect to the trade definition of the term ‘actuarial equivalent.’ In other words, when the 1982 retirement commission decided to adopt a practice that would not achieve actuarial equivalence, it voluntarily amended the retirement ordinance to reflect this understanding. Likewise, the employers’ decision to apply the 100% female/0% male table was a “ ‘definite, certain, and intentional’ ” action.17 Deliberate action is evidence that the practice of using a 100% female/0% male table was “widely acknowledged and mutually accepted.”
Although the 1982 actuarial study indicated that a 100% female/0% male mortality table was the only way for female retirees to continue to receive benefits at the rate they had been receiving them, the retirement commission could have chosen to adopt a mortality table that featured a different blend at less cost to the system, and which was more likely to achieve “actuarial *100equivalence,” as the actuarial report made clear. Instead, the retirement commission decided to use a system that benefited female employees at a cost to actuarial equivalence.18 If the retirement commission had wanted to retain full discretion to change the actuarial assumptions unilaterally, it would not have amended the ordinance with this language. This action is an even stronger indicator of intent than any cited in Detroit Police: Here, the parties’ course of conduct was unequivocal and the terms of modification were definite, certain, and intentional.
The majority states that the retirement commission has always retained the discretion to elect how to determine actuarial equivalence, and I agree with the majority that § 15 of the Ordinance says as much.19 But *101this is the same provision of the retirement ordinance that specifically provides that the actuarial equivalence is to be determined “using a 7V2% interest rate and a blending of male and female rates based on the 1971 group annuity mortality table projected to 1984 with ages set back 2 years,” which reflects the 1982 amendment to the ordinance and the adoption of the 100% female/0% male mortality table.20 It cannot be disputed that the amended § 15 language refers specifically to “the 1971 Group Annuity Mortality Table projected to 1984, for females, set back two years.”21 This 1971 data set formed the basis for the 100% female/0% male blend the parties began using after this amendment, underscoring that the amended language refers to the 100% female mortality table, as it is the only mortality table that is derived from the 1971 data set.
The retirement commission’s practice of using an agreed upon formula sacrificed both actuarial equivalence and its full discretion. If the commission had *102intended to retain the discretion to unilaterally change the actuarial assumptions for the sake of implementing the plain meaning of the actuarial equivalency requirement, it could have done so without explicitly memorializing this roundabout way “actuarial equivalence” was to be achieved. Moreover, even if the retirement commission retained some degree of discretion under the amended portion of the ordinance’s reference to a “blended” table, the employer’s unequivocal application of the 100% female/0% male table for 24 years meant that the employers sacrificed adherence to the plain meaning of “actuarial equivalence,” and that all parties to the CBAs were bound to the specific 100% female/0% male blend.
The retirement commission’s amendment to the retirement ordinance makes the comparison to Gogebic College Mich Ed Support Personnel Ass’n v Gogebic Community College22 unhelpful. In Gogebic, the Court of Appeals held that the employer had discretion to select the dental insurance carrier, because the contract only indicated what benefits were due employees, not what carrier would provide the benefits. The retirement ordinance in this case did not merely state that employees would be able to elect optional pension benefits, but actually set the means by which those benefits would be calculated and explicitly incorporated the actuarial assumptions to be used in calculating optional pension benefits.
The amendment to § 15 of the retirement ordinance reflects the retirement commission’s adoption of the 100% female/0% male mortality table, and the parties’ application of that table represents a “definite, certain, and intentional” action. Because the parties’ commit*103ment to this intentional formula lasted 24 years and survived a previous actuarial study without change, I would find that the charging parties have shown that the past practice was so widely acknowledged and mutually accepted as to create an amendment to the contract, and that the PERA mandates the parties return to the bargaining table.
CAVANAGH, J., concurred with MCCORMACK, J. VIVIANO, J., took no part in the decision of this case because he was the Chief Judge of the 16th Judicial Circuit Court before his appointment to this Court.Port Huron Ed Assoc v Port Huron Sch Dist, 452 Mich 309; 550 NW2d 228 (1996).
Id. at 317-318.
Id. at 325 (quotation marks and citation omitted).
Id. at 329 (quotation marks and citation omitted).
Detroit Police Officers Ass’n v Detroit, 452 Mich 339, 345; 551 NW2d 349 (1996) (citation omitted).
Id.
Id. at 341.
Id. at 341-342.
Id. at 346-348.
Port Huron, 452 Mich at 312.
Id. at 313.
Id. at 331.
Id. at 314-315. Notably, “[t]he record is unclear whether there were any teachers hired for less than a full year before the 1983-84 school year. From 1983 to 1987, however, eleven teachers were hired midyear.” Id. at *97313. In other words, the practice the charging parties cited may have only been in place a few years.
The relevant time span in Port Huron was at most 10 years, whereas the time span in Detroit Police was potentially as long as 49 years. Although the length of time is not a dispositive factor, it certainly hears on our analysis of whether a past practice is something that an employer merely “knew or should have known,” as opposed to something that is more "definite, certain, and intentional.”
The 1982 actuarial study states: “Instead of being designed to remove all cost of option election from the plan, the factors could be designed to make sure that no participant will receive a lesser benefit than under present procedures. To accomplish this, the present female factors would be used for all future retirants.”
Section 15 of the Macomb County Retirement Ordinance.
Port Huron, 452 Mich at 329 (citation omitted).
The actuarial report specifically acknowledged that adopting a 100% female/0% male blend would probably impose a higher cost to the system. See n 16.
Once a court has determined that an issue is covered by a CBA, the Port Huron analysis directs the court to determine whether that language is ambiguous. In this case, the majority finds that “actuarial equivalent” is the relevant term for ambiguity analysis and that it is unambiguous. Unambiguous language will control unless the past practice is widely acknowledged and mutually accepted such that it amends the CBA itself. However, the majority opinion also finds that the retirement ordinance gives the retirement commission the unilateral discretion to change the actuarial assumptions: thus, no past practice can overcome the language of § 15. The majority conflates the issues: If the majority believes that the discretionary language of § 15 affects the past practice analysis, this language should also be scrutinized for ambiguity. Because the retirement ordinance specifically sets the actuarial assumptions at a definite interest rate and mortality table in § 15, I would find that the § 15 is ambiguous as to whether the retirement commission retains unilateral discretion in selecting actuarial assumptions. When there is ambiguous contract language, the parties’ practice is evaluated under a “knew or should have known” standard. Under that standard, I would find that the respondents-appellants knew or should have known that the language of § 15 sacrificed the retirement commission’s unilateral discretion in favor of setting “actuarial equivalent” to a benchmark that would not conflict *101with its trade definition. Contrary to the majority’s position, when read in full instead of piecemeal, § 15 of the retirement ordinance is ambiguous as to whether the retirement commission has retained unilateral discretion to modify actuarial assumptions. Indeed, when read in full, § 15 is instead evidence that the retirement commission intentionally abandoned actuarial equivalence.
Section 15 of the retirement ordinance. The majority’s view that the adverb “currently” modifying the verb “using” indicates that the retirement commission retained unilateral discretion ignores a full reading of § 15. As previously explained, the entire second sentence of § 15 (“currently” included) is unnecessary if the majority is correct. Because the retirement ordinance sets actuarial equivalence to a set benchmark, I still find that the § 15 does not unambiguously grant the retirement commission unilateral discretion.
The majority also fails to note that § 15 refers explicitly to “the 1971 group annuity mortality table projected to 1984 with ages set back 2 years.” The newly proposed 60% female/40% male table would be based on “the 2000 RP Mortality Table projected for 15 years, with no set back on ages.” The newly proposed table represents a change to the plain language of § 15.
Gogebic College Mich Ed Support Personnel Ass’n v Gogebic Community College, 246 Mich App 342; 632 NW2d 517 (2001).