I respectfully dissent.
In my opinion, what happened here is an affront to the integrity and independence of the judicial branch of government as much as it is a breach of faith with the destitute citizens who put their trust in a contract with Los Angeles County. In a single stroke, the county and state government managed to violate two fundamental constitutional provisions—separation of powers1 and the contract clause.2
To provide context for many of the issues which divide me from my colleagues in this case, I begin with a brief review of general assistance and its place within the overall scheme of welfare in this state.
The societal obligation to support the poor and destitute is not an invention of the welfare state. It has been a recognized feature of Anglo-American law since the time of Henry VIII, when the government began backing up the church and its admonitions to help the poor. “The [law of] 27 Hen. VIII.C.25. contains the first provision by which particular districts are directed to support their poor, so that none of them of very necessity shall be compelled to go openly in begging.” (1 Nolan, A Treatise of the Laws for the Relief and Settlement of the Poor (1st ed. 1805) p. 4.) While the English government initially only imposed a “trivial penalty” for failure to live up to this obligation, it soon became more persuasive in its encouragement of generosity toward the poor. Early on, Parliament passed a law providing that “[T]he collectors for the poor, on a certain Sunday in every year, immediately after divine service, were to take down in writing, what every person was willing to give for the ensuing year, and if any should be obstinate and refuse to give, the minister was gently to exhort him; if still he refused, the minister was to certify such refusal to the bishop of the diocese, and the bishop was to send for and exhort him in like manner; if he stood out against the bishop’s exhortation; then the bishop was to certify the same to the *1214justices in sessions, and bind him over to appear there: And the justices, at the said sessions, were again gently to move and persuade him; and, finally, if he would not be persuaded, they were to assess him what they thought reasonable towards the relief of the poor.” (Id. at p. 5.)
During the reign of Elizabeth the First, the English enacted what became known as the Elizabethean Poor Laws, which replaced the church-centered program with a nationally mandated scheme of locally administered and financed assistance for the indigent. (1 Nolan, a Treatise of the Laws for the Relief and Settlement of the Poor, supra, at pp. 4-22.) A few decades later, many of the American Colonies followed suit. “The Plymouth Colony adopted such provisions in 1642, Virginia in 1646, Connecticut in 1673, Massachusetts in 1692, and so on. The colonial poor laws not only were patterned after the Elizabethan legislation, but often retained many of its specific provisions, in some cases in toto” (Trattner, From Poor Law to Welfare State (1984) p. 18.) And there was no lack of poor people to be aided. By the late Colonial period “. . . poor relief consumed on the average anywhere from 10 to 35 percent of all municipal funds, the single largest annual outlay.” (Id. at p. 32.)
Public support of the poor remained largely a local responsibility until the Great Depression spawned the New Deal. Franklin Delano Roosevelt proclaimed “ ‘Modern society acting through its government owes the definite obligation to prevent the starvation or dire waste of any of its fellow men and women who try to maintain themselves but cannot.’ ” Aid to jobless citizens, he declared, “ ‘must be extended by government, not as a matter of charity, but as a matter of social duty.’ ” (Trattner, From Poor Law to Welfare State, supra, at p. 262.) On August 14, 1935, a federal law implementing that philosophy, the Social Security Act, became law. In addition to old age security and unemployment insurance for the short-term unemployed, this omnibus legislation included a federal program to support blind and disabled citizens and a joint federal-state program of financial aid for mothers with dependent children. (Id. at pp. 270-275.)
The New Deal reforms of the 1930’s transferred most of the responsibility for supplying public support to poor people from local government to the federal government or to programs operated as federal-state partnerships. The major exception was the residual category of able-bodied adults without dependent children and so long out of work they were ineligible for unemployment insurance. These citizens remain the responsibility of local government under a system which in many states, including California, retains many vestiges of the Elizabethan Poor Laws.
In California, as in many states, this residual program is called General Assistance. It has existed in California in some form since 1855, only five *1215years after statehood. (Mooney v. Pickett (1971) 4 Cal.3d 669, 677 [94 Cal.Rptr. 279, 483 P.2d 1231].) While in almost half of the nation’s jurisdictions state governments have “taken control of’ General Assistance from their county governments, California remains one of those in which county governments continue to fund and administer the program.3 But it is not an optional responsibility for those governments. In 1932, the California Supreme Court declared counties have a mandatory duty to support the poor. (San Francisco v. Collins (1932) 216 Cal. 187, 190 [13 P.2d 912].) Moreover, counties must follow general state guidelines the Legislature enacts.
Among other things, these state laws establish the standards for the assistance levels the counties are required to provide eligible recipients. At the time appellants filed their lawsuit and the parties negotiated the settlement agreement and the court entered judgment, state law required counties to set monthly General Assistance (GR) grants at actual subsistence levels as determined by current studies (e.g., Welf. & Inst. Code, §§ 17000, 17001, 10000, Whitfield v. Board of Supervisors (1991) 227 Cal.App.3d 451 [277 Cal.Rptr. 815]).4 Recent state legislation has amended and thereby lowered minimum GR grant levels counties are expected to maintain (e.g., Welf. & Inst. Code, §§ 17000.5, 17000.6). But it is in the Legislature’s attempt to retroactively cancel preexisting settlement agreements and court judgments dealing with this subject that it raised the constitutional issues which divide this court.
I. The Settlement Agreement and Ensuing Stipulated Judgment Constituted a Contract Between the County Government and the Plaintiff Class for Purposes of the Contract Clause.
The majority opinion has accepted the county’s position the settlement agreement the parties negotiated and signed, and the court reduced to judgment, was not a “contract” for purposes of the contract clause. Before discussing what is required to qualify as a “contract” protected under the *1216contract clause, it will be helpful to sketch in the background facts surrounding this settlement agreement.
A. The Settlement and Judgment Constituted a Contract, in This Instance a Contract to Pay Money (Not an Injunction or Declaratory Relief).
At the time these welfare recipients filed their lawsuit in December 1990, Los Angeles County already had a long history of noncompliance with Welfare and Institutions Code section 17000 et seq. The county had not conducted the studies that statute required for almost a decade. As a result, in the mid-1980’s the GR welfare recipients had been compelled to file a lawsuit, Blair v. Board of Supervisors (Super. Ct. L.A. County, No. C56184), seeking county compliance with Welfare and Institutions Code section 17000 et seq. The parties ultimately settled this case in 1987. As a result, the county conducted a study of the actual costs of minimum subsistence and raised the GR grant level to $312 a month effective July 1, 1988. But then, once again, the county resumed its previous pattern of failing to conduct the required studies or to update the GR grant levels on a regular basis. Thus, two and one-half years later, when the welfare recipients filed the instant action the GR grant levels still stood at the July 1, 1988, level—$312 a month.
Appellants filed their complaint in Elizabeth Mendly et al. v. County of Los Angeles (No. BC017558) on December 20, 1990. The Mendly lawsuit sought to remedy the loss of income GR recipients had suffered and continued to suffer as a result of the county’s failure to study the current costs of minimum subsistence and to raise the GR grant level since 1987. The plaintiff class included “all present, future and former recipients of GR in the County since July 1, 1989.” Their complaint prayed for several forms of relief.
To take care of the current year, the complaint asked the court to “issue a peremptory writ of mandate compelling defendants immediately to raise the monthly GR grant amounts to subsistence levels” and “immediately to make an accurate, objective and factual determination of the actual monthly costs of minimum subsistence for food, shelter, clothing, transportation and personal needs in the county.” For future years, plaintiffs asked the court to order the county “to adjust the monthly GR grant amounts on an annual basis by amounts sufficient to enable indigent GR recipients to meet their basic minimum subsistence needs after having conducted an annual determination...."
The Mendly complaint then sought a preliminary and permanent injunction enjoining the county “from providing inadequate monthly GR grant *1217levels to indigent GR recipients.” It also prayed for a declaration the county had violated and continued to violate “the statutory requirements for providing general relief in California,” failed to set adequate GR grant levels, and failed to make required studies and to adopt adequate standards.
Finally, the complaint prayed for retroactive monetary damages which accrued since July 1989. The body of the complaint alleged:
“As a direct result of the above-alleged acts and omissions of defendants, plaintiffs and other similarly situated County GR recipients are being and have been injured since July 1, 1989, because they have been receiving inadequate GR benefits, which were and are substantially less than they were and are legally entitled to receive.”
Paragraph 6 of the “Prayer for Relief’ reads:
“6. That this court award plaintiffs, and all other similarly situated recipients of GR in the County since July 1, 1989, retroactive benefits from July 1,1989 until the time this action is concluded, in an amount equal to the difference between the actual amount of GR benefits provided and the level of general relief which should have been provided to members of the class by defendants, plus prejudgment interest; . . ,”5
By the time discovery was completed appellants were asking the GR grant level be raised immediately from $312 to approximately $420, a figure their evidence suggested would be sustained by any properly conducted study of subsistence costs. (This appears reasonable given the fact the federal poverty line for a single person had risen to approximately $550 a month and represented a national average rather than a figure adjusted for high-cost living areas such as Los Angeles. Moreover, a similar lawsuit filed against rural Humboldt County had settled almost two years earlier with the county agreeing to pay GR recipients $393 a month.)
Evidently concerned about the county’s potential exposure should the lawsuit go to trial, on March 12, 1991—three months before the scheduled hearing on the peremptory writ of mandate—the board of supervisors instructed its senior officials to attempt to settle the case. Negotiations stretched over three months virtually to the date of the scheduled writ hearing. Two of the county’s five supervisors, the county counsel, and the chief administrative officer, among others, personally participated in the negotiating sessions with plaintiffs’ lawyers.
*1218During the period negotiations proceeded, the county’s budget crisis was well known and frequently discussed. Indeed it was during this time the chief administrative officer submitted the county’s proposed annual budget accompanied by a 13-page letter to the board warning it reflected a $277.5 million shortfall and he anticipated the state would experience a $12.6 billion shortfall. These concerns shaped.the negotiations and led the county’s representatives to stress the need for an agreement which would fix GR payment levels “for at least a five year period, to achieve fiscal stability.”
On June 11, 1991, the supervisors were able to announce at a press conference they had managed to settle the litigation. They also made the point the settlement agreement was much more favorable financially to the county than what it risked had the case gone to trial.
Given its potential exposure, the county had every reason to be pleased with the terms of the settlement. In return for a guaranteed minimum of $341 a month for the next five years, plaintiffs surrendered their claim to prospective payments of as much as $420 a month. (The Los Angeles County officials should have been especially pleased since, as mentioned above, similar negotiations two years earlier in a far smaller, less costly area, Humboldt County, resulted in an agreement the monthly GR grants in that county be raised to a level $52 higher than what Los Angeles County agreed to pay in this settlement.) Of equal significance, the plaintiffs gave up their claim for retroactive benefits, thus reducing the county’s financial exposure by over $100 million.6
On June 12, 1991, the parties filed a “Stipulation for Settlement and Entry of Judgment” accompanied by a 12-page “Judgment.” The trial court conducted a fairness hearing and on July 30, 1991, filed an “Order Approving Settlement and Final Judgment.” The court approved “as fair, reasonable, and adequate” and without change the “settlement agreement in this case embodied in the ‘Stipulation for Settlement and Entry of Judgment’ previously filed.” (Italics added.) The court then ordered judgment entered “in conformity therewith.”
*1219The terms of this agreement and judgment are significant for what they are and what they are not. Much of the rationale of the majority opinion is built around the notion the plaintiffs obtained a judgment for injunctive and declaratory relief as a result of this settlement. True, the Mendly plaintiffs sought injunctive and declaratory relief (as well as retroactive damages) in their complaint. They asked the court to order the county to conduct annual studies of subsistence levels and to annually adjust GR grants to meet those levels. But that is not what they got in their settlement. Instead they gave up the annual studies and grant adjustments in return for what is really a money judgment. The county agreed to pay a sum equal to $341 a month (or the AFDC payment level if it is higher) times the number of eligible GR recipients for a five-year period in order to avoid imposition of an injunction requiring it to conduct subsistence studies and adjust grant levels upward during those same five years.
Other than the provisions the county “shall” pay these fixed payments to eligible recipients the settlement/judgment does not order the county to do or refrain from doing anything. Nor does the settlement/judgment contain the language of an injunction or of declaratory relief. It does not enjoin the county to comply with state law or to do anything else. Nor does it prohibit the county from failing to comply with state law.
One can imagine a settlement of this lawsuit which indeed would have been in form and substance a judgment for injunctive and declaratory relief. For instance, the county and plaintiff welfare recipients could have stipulated to entry of a judgment which declared the county had failed to comply with state statutes requiring subsistence studies and grant adjustments in the past and prohibited it from continuing to disobey those statutes in the future. Or the parties could have agreed to a stipulated judgment which affirmatively enjoined the county to conduct the statutorily required studies and adjust grant levels accordingly in 1991-1996. In either of these examples, the parties would have created the kind of agreement/judgment on which the majority bases many of its arguments.
But this agreement/judgment does not grant declaratory or injunctive relief. Instead it substitutes an award of specific sums of money to members of a defined class. The key language reads:
“3. For fiscal year . . . 1991-92, commencing on July 1, 1991, and ending on June 30, 1992, the County general relief cash grant for a single person household shall be the greater of $341 per month or an amount not less than the monthly amount allowed on the date of enactment of the State of California FY 1991-92 Budget for a single person household under the Aid *1220To families With Dependent Children (‘AFDC’) grant (herein ‘AFDC alternative amount’). The general relief cash grants for two through ten or more person households shall be the greater of the respective amounts set forth in the General Relief Basic Budget Table, attached as Exhibit A hereto and incorporated herein by this reference, or the AFDC alternative amounts for the corresponding size households.” (Subsequent paragraphs apply this same formula to the four succeeding fiscal years. Another provision allows the county at its own option to raise the GR grants above $341 (or the then existing AFDC level) if through inflation these payment levels fall below the “equivalent value of $312 on July 1,1991.” During the last three fiscal years covered by the agreement, the plaintiffs have the option to terminate the agreement if and only if the grant amount the county provides is less than the inflation adjusted value of $312 on July 1, 1991.)
The $341 payment level for the five-year period not only substitutes for the injunctive and declaratory relief the plaintiff welfare recipients originally sought, but also includes an increment attributable to the plaintiffs’ surrender of their claim for retroactive damages. For paragraph 16 of the judgment provides: “All members of the class of plaintiffs . . . hereby waive any and all rights they may have to receive any retroactive GR benefits. . . .”
Significantly, the agreement specifically addressed the issue of which changes in the law would affect the terms of the agreement.
“Changes in State or Federal Law
“17. Nothing in this judgment shall prohibit defendants from implementing any other policies or practices relating to the general relief program required by changes in state or federal law. In the event the California State Legislature abolishes all obligations under Welfare and Institutions Code, section 17000, or takes control of or otherwise assumes the County’s obligation under Welfare and Institutions Code, section 17000, the .terms in this judgment will automatically expire.” (Italics added.)
This is a feature of the settlement agreement in this case which is seldom found in contracts, public or otherwise, which the courts have been called upon to review under the contract clause. In most instances, the contract is silent on this issue and the court must impose a promise on government not to allow changes in the law to impair its contracts with its citizens. (See discussion, post, at pages 1228-1239.) Here, however, the county itself expressly promised to honor the grant levels set forth in this agreement despite future changes in state or federal law, unless those changes were so dramatic as to “abolish all obligations” under the GR statute or to transfer *1221the county’s entire obligation to the state or to constitute a state “takeover” of the general relief program.
The county is allowed to implement “other policies or practices . . . required by changes in state . . . law,” that is, those not related to GR grant levels. (Italics added.) (For instance, consistent with the agreement, the county could implement statutory changes in eligibility criteria or work rules, etc.) But as to the monthly grant levels of $341 (or the AFDC alternative amount) this clause of the agreement constitutes an explicit promise to honor the terms of this agreement despite the type of statutory change which occurred in this case.7 As to future amendments reducing the county’s financial obligations, they are not to have any effect since the contract only relieves the county of its responsibilities under the settlement agreement if the Legislature “abolishes all obligations.” It would have been inconsistent with the basic purpose of this agreement—to provide a fixed level of GR grants for a five-year period—for it to include a clause allowing the GR grants to decrease if the law changed to reduce the county’s payment obligations. Achievement of this fixed, predictable grant level obviously required appellant welfare recipients to give up increases adherence to the then existing law might otherwise entail and the county to give up decreases changes in the law might otherwise allow. Only the clearest, most specific language in the contract could authorize the county to reduce GR grants below the level fixed in that contract merely because the state amended its law to permit counties to pay less than that amount. There was no such clear, specific language. Indeed, as observed above, the language of the agreement *1222only relieves the county of its obligation to pay GR grants at the contract amount if the Legislature “abolishes all obligations” not if it merely reduces those obligations.
B. The Settlement Agreement (and Appended Judgment) Easily Qualifies as a Contract Protected Under the Contract Clause.
It is apparent under well-settled principles that the settlement agreement reached in this case was a contract for purposes of the contract clause. Moreover, it did not lose its protection under that constitutional provision merely because it was embodied in a judgment.
The contract clause embraces an expansive definition of the term “contract.” As the United States Supreme Court pointed out over 70 years ago: “It has long been settled by decisions of this court that the word ‘contracts’ in § 10 of Article I of the Constitution is used in its usual or popular sense as signifying an agreement of two or more minds, upon sufficient consideration, to do or not to do certain acts.” (Crane v. Hahlo (1922) 258 U.S. 142, 146 [66 L.Ed. 514, 517, 42 S.Ct. 214].)
The settlement agreement in this case easily satisfies all of these elements. It represented the “agreement of two or more minds”—the plaintiffs and the county. Both sides provided “sufficient consideration”—the county raised the GR grant level by 10 percent and guaranteed it for five years while the plaintiffs accepted a level almost 25 percent below what they sought and surrendered their claim for retroactive payment even of the 10 percent increase they were receiving. They also gave up for five years their statutory right to annual studies of subsistence needs and adjustments in grant levels to match those needs. These acts of forbearance clearly represent adequate “consideration” for the settlement agreement in this case. (Warren Nat. Bank v. Suerken (1920) 45 Cal.App. 736, 740 [188 P. 613]; 1 Witkin Summary of Cal. Law (9th ed. 1987) Contracts, § 214, p. 223 and cases cited.) Indeed the forbearance consideration plaintiffs supplied here is the form of consideration supporting most settlement agreements arising out of litigation. The plaintiffs give up a part of their claim in return for defendants’ agreement to pay them a lesser sum of money without the time, expense, and risk of a trial. If the plaintiffs’ forebearance consideration in this case is insufficient to support a settlement agreement, it is difficult to imagine what settlement agreement would qualify as a contract enforceable under the laws of California.8
Finally, the agreement called for the parties “to do or not to do certain acts”—the county was to pay the increased grants for five years and to do so *1223despite changes in state or federal law and the plaintiffs were to terminate their case, surrender their claims, and refrain from pursuing any claims for higher GR grants during that same five-year period.
That certainly sounds like “contract... in its usual or popular sense” to me. Not surprisingly, settlement agreements have been treated as just that. “A settlement agreement is in the nature of a contract and is therefore governed by the same legal principles applicable to contracts generally.” (In re Marriage of Hasso (1991) 229 Cal.App.3d 1174, 1180 [280 Cal.Rptr. 919]; Folsom v. Butte County Assn. of Governments (1982) 32 Cal.3d 668, 677 [186 Cal.Rptr. 589, 652 P.2d 437] [“Compromise agreements are, of course, ‘governed by the legal principles applicable to contracts generally’ ”]; Adams v. Johns-Mansville Corp. (9th Cir. 1989) 876 F.2d 702, 704 [“Under California law, settlement agreements are governed by general principles of contract law.”]; Varwig v. Leider (1985) 171 Cal.App.3d 312, 315-316 [217 Cal.Rptr. 208].)
The majority appears to argue, as did the county, that the simple process of turning this settlement agreement into a judgment somehow removed it from the category of “contracts” and thereby denied it protection under the contract clause. As a class action, this agreement was required to be reviewed for fairness, principally to the members of the class, and approved *1224by the court. But did that render it less of a contract? The majority quotes language from some California decisions suggesting consent judgments have the attributes both of contracts and of judgments and are more like contracts for some purposes and more like judgments for others. (Maj. opn., ante, at pp. 1206-1207.) It does not follow from this observation, however, that consent judgments are not contracts for the purpose relevant to this case— inclusion under the protective umbrella of the contract clause. The majority fails to go the next step and ask whether consent judgments are enough like contracts to qualify for protection under the contract clause. The answer to that question is a resounding yes.
“ ‘[I]t would be passing strange if a consent decree, which possesses all the attributes of an ordinary contract plus the additional element of judicial approbation, were to be accorded some inferior status.’ ” (Smith v. Sumner (9th Cir. 1993) 994 F.2d 1401, 1406, italics added.) Indeed in California, . . the rule is well settled that insofar as a judgment, or a part thereof, has been rendered pursuant to the expressed consent of the parties, it will be regarded as a contract between the parties, to be construed as any other contract; . . .” (Robinson v. Robinson (1962) 208 Cal.App.2d 213, 218 [25 Cal.Rptr. 143]; Pardee Construction Co. v. City of Camarillo (1984) 37 Cal.3d 465, 471 [208 Cal.Rptr. 228, 690 P.2d 701] [“. . . a consent judgment is in the nature of a contract . . . , subject to interpretation and construction”]; Vogel v. City of Cincinnati (6th Cir. 1992) 959 F.2d 594, 598 “[A consent decree, although in effect a final judgment, is a contract founded on the agreement of the parties. [Citations.] It should be construed to preserve the position for which the parties bargained.”]; 6 Witkin, Cal. Procedure (3d ed. 1985) Proceedings Without Trial § 254, p. 552.)
If there ever was a judgment which was the substantial equivalent of a settlement agreement—and a contract within the meaning of the contract clause—this is the one. In fact, the “settlement agreement” document is only two pages long. It says the parties agree the trial court may enter the judgment after holding a fairness hearing on that judgment and personally approving the stipulated settlement agreement in a related case. All of the terms of the settlement agreement actually are set forth in the judgment, which is a dozen pages long. The “settlement agreement” concludes with the provision “[t]his settlement agreement, the terms listed in the judgment, and the exhibits attached thereto, constitute the entire agreement among the parties . . . .” (Italics added.) In this case the terms of the judgment are the terms of the settlement agreement. The settlement agreement clearly is a contract. As a contract it qualifies for whatever protection the contract clause affords any other contract.
The only grounds the majority tenders for considering this settlement agreement (incorporating the terms of the judgment) to lack the qualities *1225of a contract for purposes of the contract clause is that it arose out of a case seeking injunctive and declaratory relief. According to this reasoning, since a trial court is authorized to modify or dissolve an injunction where the ends of justice would be served, somehow the consent judgment is no longer a contract for purposes of the contract clause. (Maj. opn., ante, at pp. 1206-1207.) As emphasized in the case the majority quotes, a trial court is empowered to modify or dissolve an injunction for these reasons “even though the judgment providing the injunctive relief is predicated upon stipulation of the parties.” (Welsch v. Goswick (1982) 130 Cal.App.3d 398, 404-405 [181 Cal.Rptr. 703].)
But does any of this mean the settlement agreement (incorporating the terms of the appended judgment) in this case fails to qualify as a contract for purposes of the contract clause? The problems with the majority’s position are several.
To begin with, of course, this is not a case where the settlement agreement or judgment (to the extent there is a difference here) constituted a “judgment providing . . . injunctive relief . . . predicated upon stipulation of the parties” as was the situation in Welsch v. Goswick, supra, 130 Cal.App.3d 398, 404-405, and like cases. As highlighted earlier, this is not a stipulated injunction where the county agreed to a judgment enjoining it to abide by the provisions of certain statutes or to do certain acts which those statutes require. Instead it is a contract in which the county agreed to pay money at agreed levels to plaintiffs in order to avoid imposition of an injunction requiring it to comply with a law which would have entailed the conduct of subsistence studies and payment of different and probably higher sums to plaintiffs. Just as it might in any money judgment arising out of a class action, the county promised to pay certain sums of money to an identifiable class of plaintiffs. (See ante, at pp. 1219-1220.)
Because it arose out of a class action lawsuit, the court had to approve the terms of this contract and the contract is embodied in a judgment the court can readily enforce. But even in its enforcement provisions, this settlement agreement does not speak in terms of injunctive relief. It is not an injunction directed at the county and requiring it to do certain things or to refrain from doing other things. Instead it speaks in terms of contract enforcement: “The Court retains jurisdiction to enforce the terms of this judgment through the full period any terms of this judgment are in force and effect.” (As we have seen, the terms of the judgment are the terms of the settlement contract.) The enforcement mechanism the settlement agreement provides likewise does not refer to the court using contempt or other sanctions to punish the county’s failure to adhere to a provision of an injunction. Rather it is *1226concerned with enforcement of the terms of a contract (which happens to be embodied in a judgment) and enforcing those terms on both parties. “Before either party seeks to enforce the terms of this judgment [which are the terms of the settlement agreement], the following procedures shall apply: (1) the moving party shall give written notice to all attorneys of record for the other party of the term or terms sought to be enforced and the reasons for enforcement; (2) the parties shall meet and confer and attempt to resolve any differences in good faith; (3) if agreement is reached, the party against whom enforcement is sought shall have 30 days from the date of notice to cure the alleged default in performance or implement the agreement; and, (4) if no agreement is reached, the moving party may seek to enforce the judgment on 15 days notice to the other party.”
There would be problems with the majority’s analysis even if this settlement agreement were considered a “judgment providing .' . . injunctive relief . . . predicated upon stipulation of the parties.” (See Welsch v. Goswick, supra, 130 Cal.App.3d at pp. 404-405.) This is not a case where the trial court itself actually modified or dissolved an injunction after concluding changes in circumstances required it to do so in the interests of justice. If so, we would be reviewing the reasonableness of the trial court’s determination there were changed circumstances and the modification/dissolution truly served the interests of justice. No, what the trial court did here was uphold the constitutionality of the Legislature’s “dissolution” of the consent judgment. The question is whether the Legislature had the power under the contract clause to do so, not whether the trial court had the authority to modify or dissolve the judgment. Indeed it is difficult to see any logical connection between a trial court’s theoretical power to modify or dissolve an injunction and any stipulated (or consent) judgment’s status as a contract for purposes of the contract clause.
Furthermore, contracts of all types are subject to reformation or rescission or may prove unenforceable in appropriate circumstances. Yet that does not mean contracts in general are ineligible for protection under the contract clause. As far as this characteristic is concerned, a stipulated judgment for injunctive relief which is subject to modification or dissolution in appropriate circumstances is no different from any other contract. Accordingly, the characteristic the majority finds so crucial is irrelevant to the applicability of the contract clause.
In sum, a settlement agreement including one incorporated in a consent judgment is one of the clearest examples of a “contract” within the meaning of the contract clause. Unlike many of the statutes and general policies which have been deemed to create “contracts” with individual citizens *1227warranting protection by the contract clause9 this settlement agreement (incorporating the terms of the judgment) was the product of face-to-face negotiations between high-ranking county officials and representatives of the appellant welfare recipients. The county officials made the promises embodied in this agreement/judgment directly to this specific, defined class of citizens in exchange for other promises from those citizens. If this does not constitute a contract “in its usual or popular sense” it is difficult to imagine what would. There is nothing in logic or common sense which would strip this contract of its protection under the contract clause merely because it is later reviewed and approved by a judge or even assuming the judge may retain some power to modify or dissolve the contract in certain unusual circumstances.
II. The Contract Clause Guarantees the Government’s Own Contractual Obligations to Its Citizens in This Case.
The federal contract clause appears in article I, section 10 of the United States Constitution and reads simply, “No State shall. . . pass any . . . Law impairing the Obligation of Contracts . . . .” (U.S. Const., art. I, § 10, italics added.) The analogous California guarantee found in article I, section 9 of our state’s Constitution reads, “A . . . law impairing the obligation of contracts may not be passed.” (Cal. Const., art. I, § 9.) Since the federal guarantee so clearly applies to the actions of the State of California and its constituent entity, Los Angeles County, I will concentrate my discussion on the United States Constitution.
The first, most obvious feature of the federal contract clause, is that it only limits state governments and not the federal government. Accordingly, cases defining the due process constraints on retroactive actions by the federal government have little bearing on the issue before this court. They simply do not involve, nor do they interpret, the contract clause. “[The U.S. Supreme Court has] never held . . . that the principles embodied in the Fifth Amendment’s Due Process Clause are coextensive with prohibitions existing against state impairments of pre-existing contacts. . . . [W]e have contrasted the limitations imposed on States by the Contract Clause with the less searching standards imposed on economic legislation by the Due Process Clauses.” (Pension Benefit Guaranty Corp. v. R.A. Gray & Co. (1984) 467 U.S. 717, 733 [81 L.Ed.2d 601, 613, 104 S.Ct. 2709], italics added.) (Significantly, among these due process cases defining the lesser limits on *1228the federal government and not the higher standards imposed on state governments is System Federation v. Wright (1961) 364 U.S. 642 [5 L.Ed.2d 349, 81 S.Ct. 368], the decision on which the majority relies so heavily. See discussion, post, at pages 1237-1239.)
Over the past 200 years, the Supreme Court has divined two distinct limitations on state governments in the language of the contract clause. First, the clause purports to limit the effect new state legislation may have on existing contracts between private parties. Second, the clause stands as a prohibition against state governments reneging on their own public contracts with their citizens.
In recent years, courts and commentators frequently have referred to the contract clause as a dead letter. Like substantive due process and other constitutional constraints on government regulation of private affairs, the contract clause has fallen into disfavor. But the shadow of that disfavor only falls on the first limitation described above—the impairment of private contracts incidental to the enactment of new regulatory laws.10 Under prevailing interpretations, the contract clause prohibition must defer to a legislative determination the impairment of private contractual rights is justified by legitimate societal goals.11 The impairment of the contract involved in this case is not a member of this disfavored class, however. The welfare recipients here did not have a private contract with some other private persons, their landlords for example, which was somehow “impaired” to their detriment because the state enacted a new law regulating relationships between landlords and tenants. No, the contract here was between these private citizens and their government. The government then enacted legislation specifically designed to impair its contractual obligation to those citizens. As will be discussed below, the contract clause prohibition against this form of state legislation is alive and well. Moreover, the many decisions involving incidental impairments of private contracts have little to tell us about the scope of this prohibition against impairment of public contractual *1229obligations. (See, e.g., Annot., State’s Exercise of Police Power as Constituting Impairment of Obligation of Private Contract in Violation of Contract Clause (Art. I, § 10, cl. 1) of Federal Constitution—Supreme Court Cases (1979) 57 L.Ed.2d 127 and cases cited.)
The first and most important case to highlight this distinction was United States Trust Co. v. New Jersey, supra, 431 U.S. 1. In that case, the States of New Jersey and New York issued bonds in order to improve and operate a port. The states passed a statute containing a covenant pledging none of the bond proceeds would be used to finance mass transit. Over a decade after the initial sale of these bonds the New Jersey Legislature passed a law which diverted the pledged revenues to what it found to be important public purposes, including the expansion of mass transit. The bondholders sued, arguing the new legislation impaired the contract between them and the state which was implicit in the bond instruments they purchased.
The New Jersey trial court and the New Jersey Supreme Court upheld the new legislation on grounds the state’s police powers justified the impairment of the state’s implied contract with its bondholders. However, for the first time in many years, the United States Supreme Court invoked the contract clause to declare a state law unconstitutional and reversed the state court’s decision. I quote at length from this opinion because it refutes directly many of the arguments on which the majority bases its decision in the instant case.
“The trial court concluded that repeal of the 1962 covenant was a valid exercise of New Jersey’s police power because repeal served important public interests in mass transportation, energy conservation, and environmental protection. [Citation.] Yet the Contract Clause limits otherwise legitimate exercises of state legislative authority, and the existence of an important public interest is not always sufficient to overcome that limitation. ‘Undoubtedly, whatever is reserved of state power must be consistent with the fair intent of the constitutional limitation of that power.’ Blaisdell, 290 U.S., at 439. Moreover, the scope of the State’s rqserved power depends on the nature of the contractual relationship with which the challenged law conflicts.
“The states must possess broad power to adopt general regulatory measures without being concerned that private contracts will be impaired, or even destroyed, as a result. Otherwise, one would be able to obtain immunity from state regulation by making private contractual arrangements. . . .
“Legislation adjusting the rights and responsibilities of contracting parties must be upon reasonable conditions and of a character appropriate to the *1230public purpose justifying its adoption. [Citations.] As is customary in reviewing economic and social regulation, however, courts properly defer to legislative judgment as to the necessity and reasonableness of a particular measure. [Citations.]
“When a State impairs the obligation of its own contract, the reserved-powers doctrine has a different basis. The initial inquiry concerns the ability of the State to enter into an agreement that limits its power to act in the future. As early as Fletcher v. Peck, the Court considered the argument that ‘one legislature cannot abridge the powers of a succeeding legislature.’ 6 Cranch, at 135. It is often stated that ‘the legislature cannot bargain away the police power of a State.’ [Citation.] This doctrine requires a determination of the State’s power to create irrevocable contract rights in the first place, rather than an inquiry into the purpose or reasonableness of the subsequent impairment. . . .
“In deciding whether a State’s contract was invalid ab initio under the reserved-powers doctrine, earlier decisions relied on distinctions among the various powers of the State. Thus, the police power and the power of eminent domain were among those that could not be ‘contracted away,’ but the State could bind itself in the future exercise of the taxing and spending powers. Such formalistic distinctions perhaps cannot be dispositive, but they contain an important element of truth. Whatever the propriety of a State’s binding itself to a future course of conduct in other contexts, the power to enter into effective financial contracts cannot be questioned. Any financial obligation could be regarded in theory as a relinquishment of the State’s spending power, since money spent to repay debts is not available for other purposes. Similarly, the taxing power may have to be exercised if debts are to be repaid. Notwithstanding these effects, the Court has regularly held that the States are bound by their debt contracts.[12]
“Of course, to say that the financial restrictions of the 1962 covenant were valid when adopted does not finally resolve this case. The Contract Clause is *1231not an absolute bar to subsequent modification of a State’s own financial obligations. As with laws impairing the obligations of private contracts, an impairment may be constitutional if it is reasonable and necessary to serve an important public purpose. In applying this standard, however, complete deference to a legislative assessment of reasonableness and necessity is not appropriate because the State’s self-interest is at stake. A governmental entity can always find a use for extra money, especially when taxes do not have to be raised. If a State could reduce its financial obligations whenever it wanted to spend the money for what it regarded as an important public purpose, the Contract Clause would provide no protection at all.” (United States Trust Co. v. New Jersey, supra, 431 U.S. 1, 21-26 [52 L.Ed.2d at pp. 109-112], fns. omitted, italics added.)
“Mass transportation, energy conservation and environmental protection are goals that are important and of legitimate public concern. Appellees contend that these goals are so important that any harm to bondholders from repeal of the 1962 covenant is greatly outweighed by the public benefit. We do not accept this invitation to engage in a utilitarian comparison of public benefit and private loss. . . . Thus a State cannot refuse to meet its legitimate financial obligations simply because it would prefer to spend the money to promote the public good rather than the private welfare of its creditors.” (United States Trust Co. v. New Jersey, supra, 431 U.S. at pp. 28-29 [52 L.Ed.2d at pp. 113-114].)
This same theme has been carried forward in subsequent Supreme Court opinions. Where the government is a party to the contract, the level of scrutiny is necessarily greater because its self-interest is involved. “[I]mpairments of a State’s own contracts would face more stringent examination under the Contract Clause than would laws regulating contractural relationships between private parties. . . .” (Allied Structural Steel Co. v. Spannaus (1978) 438 U.S. 234, 244, fn. 15 [57 L.Ed.2d 727, 736, 98 S.Ct. 2716].) “When a State itself enters into a contract, it cannot simply walk away from its financial obligations.” (Energy Reserves Group, Inc. v. Kansas Power & Light Co. (1983) 459 U.S. 400, 412, fn. 14 [74 L.Ed.2d 569, 581, 103 S.Ct. 697].)13
The California Supreme Court, if anything, has proved more vigorous in holding this state’s governments to their contractual obligations with their *1232citizens. In the leading case of Sonoma County Organization of Public Employees v. County of Sonoma, supra, 23 Cal.3d 296, decided two years after United States Trust Co., our high court applied the same contract clause protection to enforce government’s obligations to its employees. The state legislation in question arose during another of California’s recurring budget crises, this one occasioned by the passage of Proposition 13. Because Proposition 13 drastically reduced the property tax revenues which funded local government, they had to look to the state government for relief. The Legislature responded by assuming greater financial responsibility for some local programs and giving grants-in-aid to local governments. As part of this package, the state enacted a law prohibiting county and city governments from paying their employees a more generous cost-of-living increase than the state was giving its employees. The state government did not consider it fair to spend part of its money to fund higher compensation increases for local employees than the state was giving its own workers.
This new state law posed a constitutional issue under the contract clause because it ran counter to multiyear collective bargaining contracts several counties and one city had signed with employee unions before the state Legislature sought to limit local wage and salary raises. In a unanimous opinion the California Supreme Court struck down this state law on grounds it impaired the collective bargaining agreements between the state’s local governments and their employees.
“There can be no doubt that [the state law in question] impairs the obligations entered into under these [collective bargaining] agreements. The Legislature could hardly have expressed itself more clearly in this regard, for it declared null and void any provision[14] of ‘a contract, agreement, or memorandum of understanding between a local public agency and an employee organization or an individual employee which provides for a cost of living wage or salary increase’ in excess of the increase provided for state employees.” (23 Cal.3d at pp. 304-305, italics in original.)
“In applying [the federal contract clause] standards to the impairment of petitioners’ contracts, we assess, first, the severity of the impact of section *123316280 upon the contract rights of petitioners. . . . [Ujnlike the situation in . . . cases ... in which contractual benefits were either postponed ... or enforcement remedies were altered . . . , in the present case the right of petitioners to a wage increase for the 1978-1979 fiscal year is irretrievably lost. Thus here, . . . there was a ‘severe, permanent, and immediate change’ in petitioners’ rights under the contract [citation] and it cannot be said that ‘[the] measure taken . . . was a mild one . . . hardly burdensome’ to petitioners. [Citation.]” (23 Cal.3d at pp. 308-309.)
“Since the statute accomplishes a severe impairment of petitioners’ contractual rights, the ‘height of the hurdle the state legislation must clear’ is elevated and ‘a careful examination of . . . [its] nature and purpose’ is required. [Citations.]
“[S]ection 16281 declares that the salary limitation was intended by the Legislature to alleviate the ‘fiscal crisis’ created by the passage of Proposition 13, and to provide for maintaining essential services, and that the measure would allow local government to continue services at a higher level than would otherwise be the case, would promote full employment, and prevent layoffs.” (23 Cal.3d at p. 309.)
“In assessing this contention, we must recall the admonition in United States Trust Company that complete deference to a legislative assessment of reasonableness and necessity is not required where, as here, government is attempting to modify governmental financial obligations. Here respondents rely upon the existence of a fiscal crisis as justification for the impairment, but they have not met their burden of establishing that such a crisis existed.” (23 Cal.3d at p. 310, fn. omitted.)
California courts likewise have invoked the contract clause to strike down statutory limitations on judicial cost-of-living raises (Olson v. Cory (1980) 27 Cal.3d 532 [178 Cal.Rptr. 568, 636 P.2d 532]; Olson v. Cory (1983) 35 Cal.3d 390 [197 Cal.Rptr. 843, 673 P.2d 720]) and pension payments to firefighters (United Firefighters of Los Angeles City v. City of Los Angeles (1989) 210 Cal.App.3d 1095 [259 Cal.Rptr. 65]). These California cases make it clear the contract clause requires our state and local governments to honor their contracts not only to those who lend money to government but to others to whom they have contracted to pay out money in the future.15
*1234Appellants have an equal claim to the protection of the contract clause for the settlement agreement they signed with Los Angeles County. That agreement created present and future government obligations just as binding, and by any measure more important to the private citizens involved, as New Jersey’s covenants with its bondholders in United States Trust Co. or the cost of living clauses embodied in the collective bargaining agreements upheld in Sonoma County.
The impairment of appellants’ contract is far more serious here than it was in either United States Trust Co. or in Sonoma County. Appellants in this case suffered more than a 14 percent cut in monthly income when the state voided their agreement with Los Angeles County. While the bondholders in United States Trust Co. lost some of the security supporting their investments, the market price of the bonds did not decrease as a result of that impairment. Nor did the county employees in Sonoma County stand to lose as high a percentage of their income if held to the state employees’ average cost of living increase as do the welfare recipients in this case.
The state’s justification16 for this obvious impairment is subject to the same high level of scrutiny as the impairments in United States Trust Co. and *1235in Sonoma County. This means the Constitution requires us to critically examine the policy reasons the government asserts in support of its impairment and seriously question any statistics and other factual claims offered to support those policies. In Sonoma County, for example, the state government tendered the policy justification it would prove unduly costly to allow local governments to honor the salary increase commitments they had made with their employees as a result of contract negotiations. The California Supreme Court looked behind the statistical projections the state government presented in support of this justification and found they were based on unrealistic assumptions and invalid calculations. The court concluded the state’s figures overstated the financial burden of compelling local governments to live up to their contractual obligations.
In the instant case, the government did not argue countervailing public policies outweighed and thus justified the impairment of appellants’ contractual rights. Instead the government relied on other arguments discussed below, e.g., the contract clause does not apply at all to agreements embodied in judgments and, in any event, the police power grants government an absolute right to renege on prior contracts.
The reason the state did not rely heavily on the justification argument is apparent when one analyzes what appears to be the only public policy consideration which could be thrown into the balance against appellants’ contractual rights. That justification is similar to what the state advanced to support its cost-of-living cap in Sonoma County—a budget crisis. The California Supreme court found this an insufficient reason to allow the government to dishonor its financial obligation to its employees. (23 Cal.3d at pp. 309-312.) The claim would be even less persuasive here for several reasons.
First, allowing appellants the welfare increases the county had contracted to provide would not have had significant budgetary consequences for the county. For appellants, the extra $29 a month would have made a great deal of difference—whether they had food for the last few days of each month, or whether they could scrape together enough to keep a roof over their heads. But even in the aggregate these increases would have represented only seventeen-hundredths of 1 percent (.17 percent) of the county’s overall budget. Such a small item can neither precipitate nor cure a budget crisis. If *1236minimal financial burdens such as this could justify the abandonment of a government’s solemn contractual obligations with its citizens, then no contract would qualify for protection under the contract clause.
Second, the county voluntarily assumed new financial obligations which imposed more serious budgetary increases during the time this same budgetary crisis existed. Most significantly, the board adopted an extraordinary pension increase which raised its own pension benefits and those of senior county employees by thousands of dollars a year.17 The annual incremental cost to the county of this pension increase is double the incremental cost had the county continued to live up to its contractual obligations to general welfare recipients. To find a budgetary crisis supplies enough justification for government to renege on its contractual obligations to its poorest citizens while that same government finds the budgetary crisis does not prevent it from dramatically bolstering its own compensation would make a mockery of the contract clause.
Third, the budget crisis was in full bloom at the time the settlement agreement was being negotiated and signed. Indeed these financial problems prompted the county to bargain for a fixed, predictable level of general assistance for a full five years into the future. This formed a central term of the settlement agreement the parties signed. In Sonoma County, on the other hand, the parties negotiated the collective bargaining agreements at issue before Proposition 13 passed, some of them years before. Thus, the initial budget crisis that initiative caused was a new development which arguably represented such a dramatic change in circumstance that it justified cancellation of the cost-of-living salary increases included in those contracts. Yet the Supreme Court did not deem the passage of Proposition 13 after the collective bargaining agreements were signed a changed circumstance justifying impairment of cost-of-living clauses in Sonoma County. A fortiori, the continuation of the budget crisis which already existed when the settlement agreement was signed by the parties in this case fails to warrant impairment of that agreement. Government cannot negotiate and sign an agreement with some of its citizens in the shadow of a budget crisis and then use that crisis as justification for reneging on the agreement. To find such justification sufficient to avoid a contractual obligation would also make a mockery of the contract clause.
*1237The majority opinion relies principally on System Federation v. Wright, supra, 364 U.S. 642 to support the contrary view. In that case, railroad labor unions and a railway company agreed to a consent decree which prohibited the union from discriminating against nonunion employees. (The decree also granted monetary awards to the nonnion employees which were untouched by later developments.) At that time the federal Railway Labor Act prohibited “closed shops” and the consent decree enforced this law. Several years later, the federal law changed and authorized “closed shops.” The union sought relief from the prohibitions of the consent decree. The United States Supreme Court, in a split decision, ruled the trial court was free to disregard the consent decree and would not run afoul of the due process clause in doing so.
System Federation is easily distinguished. To begin with, it does not involve the contract clause at all. Both the original “anti-closed shop” law and the superseding law permitting “closed shops” were acts of Congress not subject to the contract clause but only to the lesser scrutiny of the due process clause. (See ante, at pp. 1227-1228.)
Second, System Federation involved a change in the law which incidentally affected a consent decree between two private parties, rather than one which impaired the government’s financial obligations to its citizens. As discussed earlier, the former is subject to lesser scrutiny than the latter, even when the contract clause is implicated. (See ante, at p. 1228.) The instant case, unlike System Federation, involves government impairment of its own financial obligations to its citizens and therefore demands close scrutiny of the justification for the impairment.
But at a still more fundamental level, and as emphasized earlier in this dissent, the consent decree in System Federation is very different from the settlement agreement (incorporating the terms of the judgment) in this case. In the System Federation decree the defendants accepted an injunction against imposing a “closed shop” on plaintiffs. In effect, they agreed to comply with the then existing law. So it made sense that when the law changed to permit “closed shops” the injunction should be dissolved.
In Mendly, on the other hand, the county did not consent to an injunction requiring obedience to then existing law. Instead it agreed to pay money at agreed levels to plaintiffs in order to avoid imposition of an injunction requiring it to comply with the law. It is as if the parties in System Federation had reached a financial settlement of their dispute in which they paid money to plaintiffs in return for being relieved of responsibility for obeying the “anti-closed shop” law. None of the rationale of the System *1238Federation opinion would apply to support retroactive reversal of that financial settlement. Accordingly, System Federation has nothing to say about the Mendly agreement and the validity of the Legislature’s attempt to retroactively declare that contract “null and void.”
The consent decree in System Federation fully enforced the law as it existed at the time rather than accepting as fair a financial compromise arrived at by the parties. To put it another way, this was not a consent judgment adopting a settlement agreement, e.g., a contract, worked out between the union and the nonunion members. Rather it was a very different sort of consent decree, one in which one party agrees to abide by the prohibitions of a law it formerly was trying to avoid or ignore. So when the law later changed 180 degrees, the agreement to abide by the prohibitions of the old law no longer made sense. (True, the monetary element of the decree did represent a compromise agreement. But significantly the Supreme Court did not rule the statutory change rendered that portion of the decree unenforceable in System Federation.) This is far different from the instant case.
Another distinction arises because the consent decree in System Federation prohibited actions which had become lawful. To continue that prohibition of now lawful acts meant enforcing “an instrument of wrong.” (364 U.S. at p. 647 [5 L.Ed.2d at p. 353].) The consent decree in the instant case neither prohibits a lawful act nor requires an unlawful act. Neither Assembly Bill No. 2883 nor any other statute makes it unlawful for Los Angeles County, or any other county, to pay the GR recipients at the rate the county agreed to in its settlement agreement, that is, $341 a month.
Finally, System Federation unlike the instant case did not involve a contract which expressly addressed the possibility of statutory changes and their effect on the terms of the contract. Here, as will be recalled, the contract specified the county would be relieved of its obligations to pay $341 a month only if the state completely eliminated or totally absorbed the county’s statutory obligation to provide general assistance. (See ante, at pp. 1220-1222.) I am not at all convinced System Federation would have been decided as it was had the consent decree contained language which expressly addressed the effects of a change in the law authorizing closed shops. Suppose, for example, the consent decree had said that particular union and that particular railway could only adopt a “closed shop” if the law was later changed to mandate rather than merely permit closed shops throughout the railroad industry. Thus, the decree would have contemplated the possibility and effects of statutory change and specifically provided the type of change which in fact did occur was not to affect the enforceability of the terms of the decree. Under those circumstances, it is not at all clear the System *1239Federation majority would have ruled the anticlosed shop provisions of the decree should be stricken. Notably, those are the circumstances present in the instant case.
For all of these reasons, System Federation and similar cases simply have nothing to teach us in this case. Here government seeks to renege on its own financial obligations, obligations it voluntarily and indeed happily contracted to assume. At the time it entered into that contract, the government regarded it far preferable to have a fixed, predictable obligation for a five-year period than to risk changed circumstances which might affect that obligation. Now that one of those circumstances, the law, has changed in ways allowing but not requiring lower payments, government wants to dismantle the contract it made with its poorest citizens. This the contract clause does not permit.18
III. The Legislation Declaring This Existing Judicial Judgment Null and Void Is Unconstitutional as a Violation of Separation of Powers
I already have discussed why the fact this settlement agreement had been reduced to judgment did not strip away the constitutional protection the contract clause confers on contracts between government and its citizens. (See ante, at pp. 1215-1239.) Now we consider whether the fact this was a judgment, not just a contract, gives it further protection under another constitutional principle—separation of powers.
A. The Legislative History of Assembly Bill No. 2883—the California Legislature Declares a Judgment of a California Court to be “Null and Void.”
The ink was barely dry on the judgment incorporating the settlement agreement between the county and appellant welfare recipients when the county began lobbying the state Legislature to get it out of the promises it made in that agreement. Lobbyists for the same board of supervisors which had authorized and participated in the negotiations that produced the agreement sat down with legislators and their staffs and devised legislation which declared that agreement was “null and void.” Using the special access a subdivision of state government traditionally enjoys with the state government itself, Los Angeles County persuaded the California State Legislature to get it off the hook by cancelling an agreement and voiding a court judgment its highest officials had voluntarily agreed to.
*1240This effort began in early 1992, after the state Legislature in its 1991 session added section 17000.5 to the Welfare and Institutions Code. This section allowed—but did require—counties to set their general assistance levels without conducting the studies Welfare and Institutions Code section 17001 required. Section 17000.5 told counties they “may adopt a general assistance standard of aid” which is as little as 62 percent of the federal poverty line or the payment level set for the AFDC program (at that time $327 a month.) Counties likewise were told they “may annually adjust” general assistance levels by as little as “an amount equal to any adjustment” in AFDC payments. But this time out the Legislature, which was well aware of the Mendly settlement as well as a couple of similar agreements in other counties, took care to reassure that “[n]othing in this section is intended to abrogate preexisting settlements.” (Welf. & Inst. Code, former § 17000.5, subd. (c).)
The record in this appeal reflects that during 1992 Los Angeles County mounted a concerted campaign to eliminate Welfare and Institutions Code section 17000.5, subdivision (c) and to undermine the settlement agreement and ensuing judgment which had been entered less than a year earlier, on July 30, 1991. The California State Association of Counties proposed legislation to that effect. Los Angeles County’s legislative representative played a prominent role in meetings between members of the Assembly and lobbyists discussing the amendment. According to attendees, discussion focused almost entirely on the Mendly settlement and the money Los Angeles County would save if the Legislature would free it from the terms of the July 30, 1991, judgment.
The county succeeded in its legislative campaign when the Legislature adopted Assembly Bill No. 2883 (Stats. 1992, ch. 721). This law deleted subdivision (c) from Welfare and Institutions Code section 17000.5, thereby eliminating the express protection for “preexisting settlements.” But it went further by declaring “null and void . . . provisions of any agreement, including a court-ordered stipulated judgment that requires a county to provide a monthly general assistance grant greater than [the levels provided AFDC recipients].” (Italics added.) (Only Los Angeles County and one or two other counties were affected by this legislation.)
Almost immediately after passage of Assembly Bill No. 2883, Los Angeles County reduced its monthly general assistance grants from the $341 required by the July 30, 1991, judgment to $293. The county administrative officer projected annual savings of $23.4 million, representing less than one-fifth of 1 percent (.17%) of the county budget.
*1241B. Legislation Purporting to Nullify an Existing Court Judgment Violates Separation of Powers and Is Itself “Null And Void.”
While some might question the morality of the county’s behavior in seeking to obtain legislation cancelling a settlement and judgment to which it had so recently and happily agreed, the real problems are constitutional, not moral. In addition to its violation of the contract clause, Assembly Bill No. 2883 is perhaps the most conspicuous example extant of a legislative incursion on the preserve of another branch of government, in this case the judicial branch. As such, it violates the constitutional scheme of separation of powers. This, of course, is something the Constitution cannot and does not tolerate. (United States v. Klein (1872) 80 U.S. (13 Wall.) 128 [20 L.Ed. 519]; McCullough v. Virginia (1898) 172 U.S. 102 [43 L.Ed. 382, 19 S.Ct. 134]; Mandel v. Myers (1981) 29 Cal.3d 531 [174 Cal.Rptr. 841, 629 P.2d 935]; Serrano v. Priest (1982) 131 Cal.App.3d 188 [182 Cal.Rptr. 387].)
The California Constitution provides: “The powers of state government are legislative, executive, and judicial. Persons charged with the exercise of one power may not exercise either of the others except as permitted by this Constitution.” (Cal. Const., art. III, § 3, italics added.)
The wisdom of this division of governmental powers is especially evident when it comes to the judicial branch. Citizens look to the courts to decide disputes and interpret laws strictly on the merits and without regard to the parties’ voting strength or political influence. The judicial branch is structured, its procedures designed and its judges sworn to guarantee that objectivity and freedom from political control. But if the legislative branch can declare judgments the judicial branch renders to be “null and void,” those essential virtues of the branch of government chosen to exercise judicial power are lost. For then a political branch, the Legislature, can exercise the true judicial power—deciding according to its own wishes disputes originally submitted to the courts and interpreting, not just enacting, the laws of the state.
Fortunately, the courts have seldom found it necessary to invoke the separation of powers doctrine to preserve the integrity of the judicial power. Apparently, this concept is so firmly ingrained in our governmental institutions, the legislative and executive branches rarely have made a run on the decisions of the third branch. While this is a tribute to their loyalty to the separation of powers doctrine, it means there are few judicial precedents to set the boundaries of the constitutional protection of judicial independence.
It is when the judiciary has rendered judgment in favor of a politically disfavored class of litigants that the Legislature is most tempted to intrude *1242on its fellow branch’s territory. During the Reconstruction Era, the disfavored class, not unexpectedly, was former Confederates. Years earlier, during the Civil War, President Lincoln had issued a proclamation granting pardons and restoring property rights to all confederates who swore allegiance to the United States. Thereafter, many of these pardoned citizens filed claims for return of their property with the recently established Court of Claims. The court honored these claims, consistent with President Lincoln’s wartime proclamation.
The Reconstruction Congress was angered by these judgments in favor of the nation’s former enemies. So in 1870, Congress passed legislation forbidding the Court of Claims from admitting any pardon as evidence to support a claim but allowing any admissions in the pardon application to be used to prove the applicant’s disloyalty. (Appropriation Act of July 12th, 1870 (16 Stat. 235).) Then Congress went one step further, targeting judgments the Court of Claims already had rendered which were on appeal in the Supreme Court. This portion of the act said: “And that in all cases where judgment shall have been heretofore rendered in the Court of Claims in favor of any claimant on any other proof of loyalty . . . , the Supreme Court shall, on appeal, have no further jurisdiction of the cause, and shall dismiss the same for want of jurisdiction.” (Ibid.)
One of those caught up in this reversal of federal policy was a Mr. Klein, the administrator of a former Confederate’s estate. During the war the federal government had taken possession of a large quantity of cotton the former Confederate owned. Under a wartime law, the seized cotton was sold and the proceeds turned over to the United States Treasury. The former Confederate had taken advantage of President Lincoln’s offer, sworn allegiance, and obtained a pardon. He and subsequently his estate sued the federal government for the money the treasury department had received when it sold the cotton the government had captured. The former Confederate’s estate won in the Court of Claims, before the 1870 act went into effect. The government appealed on December 11, 1869, just months before the act became law.
This brought the constitutionality of the act of 1870 squarely before the United States Supreme Court. If the high court followed that law, it had no recourse but to dismiss Mr. Klein’s claim—not the government’s appeal but the claimant’s cause of action—because that claim was based on a presidential pardon. This the Supreme Court refused to do, explaining: “The court is required to ascertain the existence of certain facts and thereupon to declare that its jurisdiction on appeal has ceased, by dismissing the bill. What is this but to prescribe a rule for the decision of a cause in a particular way? In the *1243case before us, the Court of Claims has rendered judgment for the claimant and an appeal has been taken to this court. We are directed to dismiss the appeal, if we find that the judgment must be affirmed, because of a pardon granted to the intestate of the claimants. Can we do so without allowing one parly to the controversy to decide it in its own favor? Can we do so without allowing that the legislature may prescribe rules of decision to the Judicial Department of the government in cases pending before it?
“We must think that Congress has inadvertently passed the limit which separates the legislative from the judicial power, [f] It is of vital importance that these powers be kept distinct. . . . [f] Congress has already provided that the Supreme Court shall have jurisdiction of the judgments of the Court of Claims on appeal. Can it prescribe a rule in conformity with which the court must deny to itself the jurisdiction thus conferred, because and only because its decision, in accordance with settled law, must be adverse to the government and favorable to the suitor? This question seems to us to answer itself.
“We repeat that it is impossible to believe that this provision was not inserted in the appropriation bill through inadvertence; and that we shall not best fulfill the deliberate will of the legislature by Denying the motion to dismiss and Affirming the judgment of the Court of Claims; which is [f] Accordingly Done.” (United States v. Klein, supra, 80 U.S. 128, 146-147, 148 [20 L.Ed. at pp. 525-526].)
By comparison with what happened in the instant case, the Reconstruction Congress was quite subtle in its attempt to overturn a series of judgments favoring a politically disfavored group. The legislators only reached back as far as those judgments still on appeal and only directed the appellate court to reverse those judgments if it determined they were based on a certain form of evidence, a pardon, rather than other proof of the claimant’s loyalty. One can only speculate how the Supreme Court would have reacted and the language it might have used if Congress had been so bold as to declare these Court of Claims’ judgments to be “null and void.”
Today the politically disfavored class is not made up of the losing rebels in a recent civil war, but the poorest of the poor amidst an economic recession. But the legislative motive is the same, to deprive these citizens of the gains the judicial branch granted them through judgments rendered under *1244then existing law. In both cases, the government sought to reverse judgments which cost government money. And in both cases, the Legislature amended the substantive law which would govern future cases and then took aim at preexisting judgments. Except Congress was less ambitious. It did not even try to strike at those judgments which already had survived appeal. The California Legislature was not so modest. It took the extraordinary step of declaring judgments which were almost two years old and which had never been appealed to be “null and void.”
If what Congress did in United States v. Klein is invalid as a violation of the separation of powers doctrine, a fortiori, what the California Legislature did here—at the behest of and for the benefit of its governmental subdivision, Los Angeles County—must be reversed as an obvious, indeed violent, intrusion on the power of the judicial branch. For, if today, the courts allow the legislative branch to declare judicial judgments “null and void” just because those judgments happen to help politically unpopular people, not too long in the future there will be no independent judicial branch at all.
If the judicial branch is to avoid this fate, the judiciary must remain ever vigilant. In my opinion, it must make its stand here in this case or there is no end to the mischief. No one’s judgment will be safe.19 At any time it wants, the Legislature will declare “null and void” prior judgments which happen to cost government too much money or to favor citizens who no longer have or never did have political power.
Subsequent opinions of the United States Supreme Court have clarified, but never undercut, the principal holding of United States v. Klein. Indeed the most serious limitation announced in later cases actually reinforces Klein's relevance to the instant case. In United States v. Sioux Nation of Indians (1980) 448 U.S. 371 [65 L.Ed.2d 844, 100 S.Ct. 2716], the nation’s high court upheld a congressional act waiving the res judicata defense which had allowed the government to prevail in prior lawsuits and allowing the plaintiffs in those lawsuits to refile their cases. When some plaintiffs refiled their lawsuits the government challenged this legislation on grounds it violated separation of powers by reopening judgments the judicial branch already had entered.
*1245In replying to this argument, the Supreme Court explained: “[The objection Congress was dictating the rule of decision to a court] is not applicable to the case before us for two reasons. First, of obvious importance to the Klein holding was the fact that Congress was attempting to decide the controversy at issue in the Government’s own favor. Thus, Congress’ action could not be grounded upon its broad power to recognize and pay the Nation’s debts. Second, and even more important, the proviso at issue in Klein had attempted 'to prescribe a rule for the decision of a cause in a particular way.’ [Citation.] The amendment at issue in the present case, however, like the Special Act at issue in Cherokee Nation, waived the defense of res judicata so that a legal claim could be resolved on the merits. Congress made no effort in either instance to control the Court of Claims’ ultimate decision of that claim.” (United States v. Sioux Nation of Indians, supra, 448 U.S. 371, 405 [65 L.Ed.2d 844, 869].)
The instant case obviously involves both of the vices the United States Supreme Court found missing in Sioux Nation. Here the California Legislature was “attempting to decide the controversy at issue in the government’s own favor.” Furthermore, it attempted to “dictate the ultimate decision” through the flagrant tactic of declaring the existing judgment “null and void.” This, under Klein, and reinforced by legal principles discussed in Cherokee Nation and Sioux Nation, it may not do.
The California appellate courts have found more reason in recent years than has the United States Supreme Court to invoke the separation of powers doctrine in order to protect the independence of the judicial branch.
In 1981, the judicial action in dispute was a court order awarding legal fees to the successful plaintiff in a controversial lawsuit challenging the state government’s practice of giving its employee paid time off for Good Friday. The Legislature did not declare that order “null and void” but made it so by refusing to appropriate the funds required to pay the fee award. When the plaintiff appealed to the high court, the justices ruled the Legislature’s refusal to honor an existing judicial order violated separation of powers. (Mandel v. Myers, supra, 29 Cal.3d 531, 547-548.) The Supreme Court explained its rationale in strong language:
“[W]hile the Legislature enjoys very broad governmental power under our constitutional framework, it does not possess the authority to review or to readjudicate final court judgments on a case-by-case basis. The recognition of such legislative authority would completely deprive court judgments of the respect and deference which the Constitution contemplates each branch of government will accord to final actions within the jurisdiction of a *1246coequal branch, and would repose in the Legislature a combination of powers that the constitutional draftsmen specifically intended to forestall.
“Indeed, this fundamental separation of powers principle has particular force in instances, such as the present case, in which the Legislature attempts to void the effect of a court judgment which determines that the government itself is obligated to pay a sum of money to an individual. If the Legislature in such a case were empowered to reexamine the merits of litigation and to ignore a particular judgment whenever it so chose, the myriad safeguards of the judicial process would come to naught and one party to a lawsuit would in effect become both litigant and judge. In our view it is difficult to imagine a clearer example of legislative usurpation of judicial authority.” (Mandel v. Myers, supra, 29 Cal.3d 531, 549, italics added.)
If the Mandel court felt it was difficult to imagine “a clearer example of legislative usurpation of judicial authority” than it had before it in that case, the justices should have seen Mendly v. County of Los Angeles. In Mendly like Mandel the “Legislature attempted] to void the effect of a court judgment which determines that the government itself is obligated to pay a sum of money to an individual.” But in Mandel the Legislature merely attempted to “void the effect of a court judgment” by failing to “implement” the obligation to pay a sum of money the judgment ordered. In Mendly, on the other hand, the Legislature has gone still further by declaring the court judgment itself “null and void.” To match the constitutional threat the Mendly case poses, the 1981 Legislature would have to have declared the Supreme Court’s decision in the underlying case to be null and void—the one which held unconstitutional the paid Good Friday holiday.
The Supreme Court that decided Mandel was not the first to hold the kind of legislative act which occurred in the instant case would be invalid under the separation of powers doctrine. Over 100 years ago, an earlier Supreme Court declared in dictum, “had the Legislature ... by special enactment . . . commanded the courts which had rendered a judgment in favor of plaintiff, ... to set it aside and to enter a judgment for the defendant, such arbitrary attempt would, at once, have been recognized as an abuse not to be tolerated under our free constitution of government.” (Pryor v. Downey (1875) 50 Cal. 388, 405.)
What the Pryor court was convinced would be “recognized as an abuse not to be tolerated” has now become reality in the form of Assembly Bill No. 2883. If anything, it goes a step further. This special enactment does not just command the courts to set aside a judgment favorable to a plaintiff and enter one for the defendant. No, in this case, the Legislature itself purports to *1247accomplish this deed without any court action at all by passing a law unilaterally declaring a prior judgment “null and void.” In Assembly Bill No. 2883, the only role the Legislature leaves to the judicial branch is to affirm this legislative nullification of an existing judgment if and when the unhappy party complains to the courts.
After the Supreme Court’s Mandel decision, the Legislature made a further attempt to defeat another attorney fee award in a controversial case. This one granted an attorney fee award to the successful lawyers in the Serrano v. Priest litigation. This time the lawmakers passed a new law prohibiting payment of court-awarded legal fees in any case, past or future, unless the Legislature had expressly appropriated funds for that purpose. But in Serrano v. Priest, supra, 131 Cal.App.3d 188, the appellate court overturned this statute as applied to the fee award in the underlying Serrano case. The court ruled the Legislature’s attempt to throw this additional hurdle in front of existing court orders violated the principle of separation of powers.
“We do not quarrel with the Legislature’s right to make a statutory amendment which prospectively changes the standards for payment of this type of attorneys’ fee award. [Citation.] We do, however, question the validity of [the Budget Act provisions] insofar as they attempt to readjudicate the August, 1975 judgment [inter alia authorizing an attorney fee award to the prevailing plaintiffs].
“The case at bar represents a ‘specific controversy’ which was assigned to the judicial branch for resolution. (Cal. Const., art VI, § 1.) And ‘the Legislature enjoys no constitutional prerogative to disregard the authority of final court judgments resolving specific controversies within the judiciary’s domain.’ [Citation.]” (Serrano v. Priest, supra, 131 Cal.App.3d 188, 200-201.)
When enacting Assembly Bill No. 2883, as in the Budget Act amendments examined in Serrano, supra, the Legislature sought both to affect future state financial obligations and existing court judgments which imposed such obligations. In Serrano, the existing court judgments embodied judicially enforceable obligations for government to pay fee awards to attorneys, including the judgment awarding fees to the prevailing plaintiffs in tire underlying Serrano v. Priest litigation. In the instant case, the existing court judgments embody judicially enforceable obligations for government to pay defined levels of welfare benefits to General Assistance recipients for a specified five-year period. Here, as in Serrano, the case was properly *1248assigned to the judicial branch for resolution. And, here as in Serrano, the Legislature lacks authority under the constitutional guarantee of separation of powers to interfere with the enforcement of (to say nothing of nullifying) this judgment “resolving [a] specific contro vers [y] within the judiciary’s domain.”
C. Contrary to the Majority Opinion, the Judgment in This Case Qualifies for Protection From Nullification by the legislature Under the Separation of Powers Doctrine.
The majority opinion seeks to deprive the judgment in this case of protection under the separation of powers doctrine on grounds analogous to its claim the contract clause does not apply.20 (Maj. opn., ante, p. 1212.) *1249According to this reasoning, „ „ the judicial authority embodied in the stipulated judgment was the authority to enforce a particular statute.” (Maj. opn., ante, p. 1212.) Once again relying on System Federation v. Wright, supra, the majority argues that when the statute changed the judgment likewise became susceptible to change, including outright nullification at the hands of the Legislature.21 (Maj. opn., ante, at pp. 1207-1210.)
There are at least two problems with the majority’s argument. First, it mischaracterizes the judgment in this case and thus erroneously concludes it does not vest appellants with rights the Legislature is prohibited from altering. But second, even if its characterization of the judgment were correct and the Legislature could retroactively alter the law governing the court’s treatment of the judgment this does not mean the Legislature has the right to unilaterally declare the judgment null and void. That is, assuming the court which entered this judgment is empowered to modify it, and could decide to modify it in the light of the Legislature’s amendments of the general assistance law, does not mean the Legislature is empowered to usurp this judicial power and itself modify or nullify the judgment.
1. The Mendly judgment is not a stipulated injunction merely enforcing compliance with a then existing statute but a monetary settlement not affected by subsequent amendments of that statute.
I already have dealt with the majority’s characterization of the judgment in this case in the context of the contract clause. (See ante, at pp. 1222-1227.) Contrary to the majority’s view, and unlike System Federation, this *1250judgment in neither form nor substance constitutes an injunction enforcing the terms of a then existing statute. To the contrary, the county avoided imposition of an injunction requiring it to comply with the then existing statutory law which required it to conduct subsistence studies and adjust GR grants upward annually. The county did not consent to abide by Welfare and Institutions Code section 17001, as it then existed, thus opening up the argument that when the Legislature amended this statute the “consent decree” in effect was modified to require the county to abide by the new law. Rather under the terms of the judgment in this case the parties are governed by a financial decree providing the plaintiffs with five years of payments at a guaranteed level of $341 per month irrespective of what the county might have been obligated to pay were it to have conducted the studies required by Welfare and Institutions Code section 17000 et.seq. in the past or during that five-year period.
There is another questionable assumption in the majority’s opinion. The opinion seeks to justify legislative nullification of the judgment in this case by characterizing it as a ‘type of declaratory judgment and injunctive decree” which is open to modification. This position is both erroneous and dangerous.
The majority’s characterization is erroneous because the judgment in this case is not a “declaratory judgment and injunctive decree” which instructs one or more of the parties to behave in certain ways. Had this judgment merely enjoined the county to conduct the types of studies and otherwise comply with Welfare and Institutions Code section 17001, it would have fit the model of a “declaratory judgment and injunctive decree.” But the judgment here was different. It was a financial judgment resolving past and future financial claims the plaintiffs had against the county.
Indeed the Mendly judgment is not a “stipulated injunction” in any sense, but rather the typical money judgment one might find concluding the settlement of any class action providing monetary relief to the plaintiff class. Like any such judgment, it provides the defendants “shall” pay each member of the plaintiff class (however many that might turn out to be) a certain sum of money for each period that person is entitled to that sum. (In the instant case, that would be the months a given class member is eligible for General Assistance during the five-year period.) This judgment grants appellant welfare recipients payment levels for the five-year period which are far below what they had reason to expect they would receive if they had won the lawsuit and the county therefore was enjoined to comply with Welfare and Institutions Code section 17001 as it then existed. Moreover, a portion of each monthly payment during those five years represents compensation for *1251the underpayment of welfare benefits during an eighteen-month period prior to the judgment.
When viewed for what it really is, the judgment in this case did constitute a “final judgment” for purposes of the separation of powers doctrine. (Mandel v. Myers, supra, 29 Cal.3d at p. 549.) The judicial authority embodied in the stipulated judgment was not merely the authority to enforce a particular statute any more than any judgment arising from a cause of action based on a particular, then existing, statute can be said to derive its judicial authority from that statute. Yet would anyone seriously argue that because an underlying cause of action is predicated on a statutory right the judgment on that cause of action, once entered, is not final “for purposes of the separation of powers doctrine?” Does the statutory basis of the underlying cause of action mean the Legislature is free to step in and unilaterally modify or nullify the resulting judgment? If so, few judgments would be immune from modification or nullification at the hands of the Legislature, especially ones which imposed financial burdens on government.
If the majority opinion in this case represented the proper construction of the separation of powers doctrine, United States v. Klein itself would have been decided differently. There the plaintiff’s cause of action, his claim for reimbursement from the treasury for his seized cotton, was based on then existing statutes authorizing and enforcing President Lincoln’s proclamation in favor of former Confederates who swore allegiance to the Union. The “judicial authority” of the Court of Claims which rendered judgment in plaintiff’s favor was the authority to enforce those statutes and that executive proclamation. Before that judgment could be decided on appeal Congress had changed the law and purported to tell the judicial branch it could not enforce judgments based on the old law. This the Supreme Court struck down as an unconstitutional incursion into the dominion of the judicial branch, even as it upheld the new statute in its application to future cases which had not gone to judgment. If, on the other hand, the fact the Court of Claims’ “judicial authority” derived from its “authority to enforce a particular statute” meant the judgment was susceptible to retroactive modification or nullification at the hands of the Legislature, the Supreme Court would have endorsed the statute Congress enacted and complied with the statute’s directive it reverse the judgment the Court of Claims had rendered.
2. In my opinion, it would have been an abuse of discretion for the trial court to modify this judgment even if it somehow were susceptible to modification for “changed circumstances" but, in any event, that is not what the trial court did in this case.
The fact the schedule of payments this judgment provided had not yet been completed when the Legislature amended the law in no sense would *1252have required the trial court to modify the judgment. Indeed for reasons suggested above, had the trial court done so in this case it probably would have constituted an abuse of discretion, I conclude Assembly Bill No. 2883 itself is unconstitutional and further question whether the other amendments prospectively reducing the county’s obligations to General Assistance recipients justified any modification of the judgment in this case. To begin with, this was not a “stipulated injunction” which conceivably might be open to modification because of changed circumstances. Even if it were, this case did not present changed circumstances sufficient to warrant modification.
What started out as a lawsuit to compel enforcement of the then existing law turned into a monetary settlement in which the county agreed to pay money at agreed levels to plaintiffs in order to avoid imposition of an injunction requiring it to comply with the law. Thus, this was not a “stipulated injunction” susceptible to amendment because of changed circumstances. Assuming it were susceptible to amendment, the trial court had no more reason to modify this monetary judgment because of a change in the law than it would to modify any other class action judgment for monetary relief where the statute underlying the original cause of action happened to be amended before the schedule of payments to class members was completed. Indeed here there was even less reason to modify since the primary purpose of the agreement for both parties was to establish a fixed, predictable payment level for a five-year period. In agreeing to that fixed payment schedule, both sides assumed the risk something would happen during those five years which would mean the payments deviated from what the law or other circumstances would require. By signing the agreement, appellant welfare recipients risked changes would occur which meant the $341 was far below what they would have received in the absence of the agreement. Meantime the county risked changes which would mean $341 was a lot more than what they would be compelled to pay in the absence of the agreement. These are risks—including the risk of changes in the law—which are inherent in all contracts, including stipulated judgments. As an additional consideration here, the agreement defined a series of changes in the law which would affect the enforcement of the agreement. Yet notably missing is any reference to changes which reduce rather than abolish entirely the county’s obligation to support appellants. As explained earlier, the absence of such a provision is perfectly understandable, since it would run counter to the primary purpose of the contract—to guarantee both parties a fixed, predictable GR grant level for a five-year period. For these reasons, I would have been inclined to reverse the trial court had it attempted to modify this judgment to lower appellants’ GR grants to the levels allowed in the statutory amendments enacted after this judgment was entered.
In any event, the issue of the trial court’s discretion to modify this judgment because of changes in the law is not before us because the trial court did not *1253reach that issue. Instead the court merely affirmed the constitutionality of Assembly Bill No. 2883, and complied with its directive to void the judgment in this case.
3. Even if it would have been within the court’s discretion to modify this judgment, the power to do so is a judicial power to be exercised by the judicial branch and the Legislature’s attempt to declare the judgment “null and void" must fail under the constitutional requirement of separation of powers.
Even assuming it was within the trial court’s discretion to modify this judgment in the light of the amendments to the general assistance code sections, that does not mean the Legislature can constitutionally exercise the court’s discretion and either modify or, as in this case, nullify that judgment. The discretionary power to modify a court judgment, to the extent this power exists, is judicial power to be exercised by the judicial branch and the judicial branch alone. In this case, the Legislature blatantly usurped the judicial branch’s power to decide whether and how much to modify a judgment of one of its courts. By declaring the Mendly judgment and like decisions “null and void,” the Legislature took away the judiciary’s power to enforce, modify, or dissolve what is the ultimate work product of the judicial branch—a court judgment resolving a dispute between contesting parties. “The recognition of such legislative authority would completely deprive court judgments of the respect and deference which the Constitution contemplates each branch of government will accord to final actions within the jurisdiction of a coequal branch, and would repose in the Legislature a combination of powers that the constitutional draftsmen specifically intended to forestall.” (Mandel v. Myers, supra, 29 Cal.3d 531, 549.)
If the Legislature’s action in doing so in this case is allowed to stand, what is there left of judicial power, the independence of the judiciary, or separation of powers in the State of California? Very little, I fear. The judicial branch will have only as much independence and as much power as the legislative branch suffers it to have. In any case where the Legislature deems the judiciary’s decision inconvenient to government, or to those with influence with the Legislature, it will be free to declare the court’s judgment “null and void.”
This I am unwilling to endorse. If the constitutional separation of powers doctrine means anything, it means legislation like Assembly Bill No. 2883 is unconstitutional. Accordingly, I must dissent on this ground, too.
*1254IV. Conclusion
In my view, this is neither a difficult nor a close case. If the contract clause means anything, it means government cannot settle litigation with promises of future payments and then enact legislation reneging on those promises. And if separation of powers means anything it means the Legislature may not declare the judicial branch’s existing judgments to be “null and void.”
An amicus curiae, the Los Angeles County Bar Association, expresses fear the trial court’s opinion, if upheld, will discourage litigants of all sorts from entering into settlement agreements with California governments, state and local. Should the rationale of the majority opinion ultimately prevail, I share that concern. If governments can renege on their settlement agreements with welfare recipients today, why not their settlement agreements with plaintiffs in tort cases and contract actions and property disputes tomorrow. And if the legislative branch can nullify the judicial branch’s judgments favoring welfare recipients today, no one’s judgments—whether the result of settlement negotiations or adjudication, whether favoring plaintiffs or defendants—will be secure tomorrow.
For all these reasons, I would reverse the order of the trial court and require it to enforce the county’s solemn promise to its citizens and to uphold the integrity of the judgment it rendered as part of the independent judicial branch of the California government.
Early in this century, Anatole France observed with some irony, “The law, in its majestic equality, forbids all men to sleep under bridges, to beg in the streets, and to steal bread—the rich as well as the poor.” (France, Crainquebille (1902).) Appellants, many of whom are homeless all or part of the time, undoubtedly have personal experience with this side of the law’s equality. Now they will find out if, in its majestic equality, the law also forbids governments to break their contracts with anyone—the poor as well as the rich, and the welfare recipient as well as the bondholder,22 the government employee,23 and the judge.24 They also will discover whether it protects all judgments the judicial branch enters from legislative nullification—those judgments favoring the poor as well as those favoring the rich. I am *1255convinced the law does both of these things and, accordingly, would declare Assembly Bill No. 2883 unconstitutional and require the county to comply with the judgment of July 30, 1991.
A petition for a rehearing was denied April 25, 1994, and appellants’ petition for review by the Supreme Court was denied July 28, 1994. Mosk, J., and Kennard, J., were of the opinion that the petition should be granted.
See post, at pages 1215-1239.
See post, at pages 1239-1253.
Bensinger, From Public Charity to Social Justice: The Role of the Court in California’s General Relief Program (1988) 21 Loyola L. A. L.Rev. 497, 498, footnote 14 (citing HEW, SSA/Office of Family Assistance, Characteristics of General Assistance in the United States, HEW Pub. No. (SSA) 78-21239 (1978).) According to this article, as of 1978, 22 states, including California, relied on locally administered and funded general assistance programs, 19 states and the District of Columbia were state funded and administered, and 9 were hybrids with the state government supervising locally administered programs funded jointly by state and local government.
In Whitfield, the Court of Appeal held state law required counties to conduct studies of the actual cost of maintaining oneself at subsistence level in that locality and to set monthly GR grants at the level the study revealed. A county could not satisfy this legal requirement by simply paying GR recipients the same amount as recipients of Aid to Families with Dependent Children (AFDC) were paid. (227 Cal.App.3d at pp. 459-461.)
Welfare recipients are entitled to recover past underpayments through a retroactive monetary award. (Mooney v. Pickett (1972) 26 Cal.App.3d 431 [102 Cal.Rptr. 708]; Board of Social Welfare v. County of Los Angeles (1945) 27 Cal.2d 81, 85-86 [162 P.2d 630].)
The plaintiff welfare recipients surrendered this part of their overall claim in paragraph 16 of the agreement which reads in pertinent part:
“Waiver of Retroactive Benefits
“All members of the class of plaintiffs . . . , hereby waive any and all rights they may have to receive any retroactive GR benefits in an amount over that actually paid by the County in FYs 1989-90 and 1990-91 as the result of any alleged deficiency in the General Relief Basic Budget Table figures for FYs 1989-90 and 1990-91. . . .”
As will be noted, the plaintiffs agreed to forego the difference between $312 and $341 for the past fiscal years as well as their right to sue for the still greater deficiencies which might have been payable had the case been fully litigated.
Although the majority does not rely on the point as a rationale for its affirmance, in one section it appears to construe Assembly Bill No. 2883 declaring this settlement agreement (and ensuing judgment) to be “null and void” (see maj. opn., ante, at pp. 1205, 1209) as the state “tak[ing] control of . . . the County’s obligations” for GR. From our earlier discussion of the history and overall framework of GR (see pp. 1213-1215, ante), it is apparent what happened here was not a state government “tak[ingj control of’ or abolishing or assuming the “County’s obligations” for GR within the meaning of this paragraph of the agreement.
The parties to this agreement undoubtedly were aware many other states administer and fund GR differently than does California. In many of these jurisdictions, the state government has “taken control of’ GR from local governments. In some, state government have assumed the financial obligation for providing general relief, an obligation formerly borne by local goverment in those states.
Against this background of possible changes the California Legislature conceivable might adopt, the function of paragraph 17 is clear. Los Angeles County wanted to protect itself from being obligated to continue making payments under this agreement if the Legislature abolished General Relief entirely or changed the basic scheme for general welfare from a county administered and funded program to one which was state administered and funded. If the state assumed the county’s obligations and paid the plaintiffs their GR grants, the county did not want to be responsible for continuing to pay them a $341 a month bonus under this settlement agreement. Assembly Bill No. 2883 fell far short of the kind of fundamental reform of GR the parties contemplated in paragraph 17 of their agreement/judgment.
The majority suggests plaintiffs’ waiver of their claim for retroactive payments was not the relinquishment of a right, but the surrender of nothing and further implies plaintiffs *1223therefore furnished no consideration for the settlement agreement (maj. opn., ante, at pp. 1205-1206). This argument fails for two reasons.
First, plaintiffs indeed gave up something when they surrendered their claim for retroactive damages. Assume plaintiffs had persisted in this lawsuit and established the county indeed had been violating its statutory duty to conduct subsistence studies and raise GR grants during earlier years. If so, they would have been entitled to a monetary award as high as $100 million in retroactive benefits, the incremental increases to which they would have been entitled during those earlier years if the county had obeyed the law. Accordingly, when they waived that monetary claim as part of the settlement agreement, plaintiffs were engaging in an act of forbearance, not an act of charity. Presumably, a portion of the settlement amount of $341 per month represents the perceived value of this particular claim to the parties—the loss of the expected value of the claim to appellants and the avoidance of the expected risk of this claim by the county.
Furthermore, in that same agreement, plaintiffs likewise gave up their claim for retroactive payment of the $341-per-month GR grant level they were to receive for the succeeding five years. Once again, this was an act of forbearance which presumably was factored into the GR grant level they were to receive in future years. Instead of a lump sum retroactive payment now and a future grant level of say $330 a month, the settlement agreement provided the plaintiff class five years of future benefits at $341 per month.
Second, even assuming the waiver of retroactive benefits somehow should be construed as a meaningless gesture, the plaintiffs provided other consideration more than ample to qualify this settlement agreement as a contract within the meaning of the contract clause. As discussed in the text above, plaintiffs surrendered their claims to injunctive and declaratory relief, claims which could have required the county to perform subsistence studies and adjust GR grants upward to levels far above what the settlement agreement provided. If any settlement agreement is supported by consideration and thus is enforceable this one is.
See, e.g., United States Trust Co. v. New Jersey (1977) 431 U.S. 1 [52 L.Ed.2d 92, 97 S.Ct. 1505], in which the Supreme Court deemed a statute promising certain bond revenues would not be diverted to certain purposes to create a contract with the bondholders, a contract which was protected by the contract clause.
“ ‘It is the settled law . . . that the interdiction of statutes impairing the obligation of contracts does not prevent the State from exercising such powers as are vested in it for the promotion of the common weal, or are necessary for the general good of the public, though contracts previously entered into between individuals may thereby be affected. This power, which in its various ramifications is known as the police power, is an exercise of the sovereign right of the Government to protect the lives, health, morals, comfort and general welfare of the people, and is paramount to any rights under contracts between individuals.’ ” (Allied Structural Steel Co. v. Spannaus (1978) 438 U.S. 234, 241 [57 L.Ed.2d 727, 734, 736, 98 S.Ct. 2716], italics added.)
Unless the state is a contracting party, courts are to defer to the Legislature’s judgment about the necessity and reasonableness of the legislation. (Energy Reserves Group, Inc. v. Kansas Power & Light Co. (1983) 459 U.S. 400, 412-413 [74 L.Ed.2d 569, 581-582, 103 S.Ct. 697].)
This clearly and explicitly refutes one of the main arguments found in the majority opinion. That opinion quotes language from cases which have nothing to do with government’s own financial contracts with its citizens to the effect government may not contract away its “police power.” (Maj. opn., ante, at p. 1207.) Thus, the majority opinion argues, the contract clause cannot prevent the government from exercising its “police power” to cancel the contract it made with these welfare recipients. (Maj. opn., ante, at pp. 1207-1208.) As the Supreme Court makes clear here in United States Trust Co., government indeed can contract to assume financial obligations with a citizen or citizens. If it does so, the contract clause protects that contract and government cannot renege on its obligation.
This language also applies when state legislation relieves counties of their financial obligations to their citizens rather than the state government’s own obligations to those citizens. (Sonoma County Organization of Public Employees v. County of Sonoma (1979) 23 Cal.3d 296, 314, fn. 17 [152 Cal.Rptr. 903, 591 P.2d 1] [invalidating under contract clause state legislation which would have reduced county and city government obligations to pay cost-of-living raises to their employees].) As the Supreme Court explained in an earlier opinion: “The county is merely a political subdivision of state government, exercising only the powers of the state, granted by the state, created for the purpose of advancing ‘the policy *1232of the state at large, for purposes of political organization and civil administration, in matters of finance, of education, of provision for the poor, of military organization, of the means of travel and transport, and expressly for the general administration of justice.’ [Citations.]” (County of Marin v. Superior Court (1960) 53 Cal.2d 633, 638-639 [2 Cal.Rptr. 758, 349 P.2d 526].)
Significantly, this language closely parallels the language of Assembly Bill No. 2883 which declares “null and void” the “provisions of any agreement, including a court-ordered stipulated judgment that requires a county to provide a monthly general assistance grant greater than” the amounts provided AFDC recipients.
The majority opinion seeks to distinguish Sonoma County on grounds there was statutory authority for counties to set employee salary levels through collective bargaining. I am not sure about the relevance of this observation.
One possibility is that the majority is of the view the contract clause does not protect contracts governments may make with their citizens unless there was expressd preexisting statutory authority to enter into that particular type of contract. If so, many contracts between *1234California governments and their citizens—including but not limited to settlement agreements in tort, contract and property litigation with the state and its subdivisions (such as county governments)—are in jeopardy. (I have been unable to locate any express statutory authority for California governments to enter into settlement agreements with litigants which sue them.) Fortunately for many California citizens, there is no reason to be concerned. The United States Supreme Court has not imposed any such requirement. Indeed the contract clause has been applied to protect many contracts not specifically—or even generally—authorized by any statute.
Another possibility is that the majority is of the view the settlement agreement between the county and appellants was an “illegal” contract because it imposed future obligations on government. While it may be true government is not allowed to contract away its “police power,” that is not true of its financial obligations. More than once, the United States Supreme Court has underscored this distinction and struck down laws which had the effect of reneging on those financial obligations in the name of exercising the state’s “police power.” (See ante, at pp. 1229-1231.)
In my opinion it would be an extraordinary, indeed untenable, position to say a county government can have its supervisors—who combine the county’s legislative, executive and quasi-judicial powers—personally negotiate and then formally approve a settlement contract which diminishes the county’s financial exposure for a five-year period, but then allow the county to renege on that contract on grounds those supervisors lacked the power to enter into such an agreement. Fortunately, I see nothing in the law which would support such an inteipretation of the contract clause.
The majority opinion appears to place the burden on appellant welfare recipients to prove the impairment was not “reasonable and necessary to further an important state interest” rather than imposing the burden on the government to prove the impairment was so justified. The Supreme Court quite logically places that burden squarely on government, not the individuals whose contracts government impaired. For instance, after finding an impairment of the county employee’s contractual rights, the Supreme Court in Sonoma County held “. . . *1235the ‘height of the hurdle the state legislation must clear’ is elevated and ‘a careful examination of . . . [its] nature and purpose’ is required.” (Sonoma County Organization of Public Employees v. County of Sonoma, supra, 23 Cal.3d 296, 309.) Our high court then proceeded to examine the justification the government, not the plaintiffs, was forced to produce in an attempt to clear that heightened hurdle. (See ante, at p. 1233.)
In 1991, the board of supervisors approved a recommendation from its chief administrator to include employee benefits in calculating the salary level on which future pension benefits are based. Employee benefits represent a high percentage of the total compensation package for Los Angeles County employees. Consequently, this amendment to the pension law raised pension payments by 20 percent, estimated to increase pension payments by as much as $30-$50 million a year. (Muir, County Oks Diluted Pension Reform Plan, L. A. Times (Feb. 16, 1994) Metro Desk, p. 1, col. 6.)
Of no precedential value but of some interest, only a month after the Los Angeles County Superior Court rendered the decision on appeal in this court, another superior court, in Humboldt County, ruled the contract clause prevented that county from avoiding its responsibilities under a similar settlement agreement with GR recipients. (Welfare Rights, et al. v. Frank (Super Ct. Humboldt County, 1993, No. 71025).
Assume the Legislature decides it can no longer afford to pay tort judgments resulting from the negligence of government employees. It reimposes “sovereign immunity.” If Assembly Bill No. 2883 does not violate separation of powers, what is to prevent the Legislature from going a step further and also declaring all existing judgments which had been decided under the now repealed tort claims statutes to be “null and void” and any damage awards or structured settlements which had balances not yet paid out thereby cancelled?
If the majority opinion in this court ultimately prevails, appellants will find out that, unfortunately, the law’s “majestic equality” only works one way.
The majority wisely elected not to rely on the county’s main arguments to support its position Assembly Bill No. 2883 passed muster under the separation of powers doctrine. In its brief, the county took the position the Legislature can constitutionally declare existing judgments “null and void” so long as the same legislation changes the substantive law governing future cases. To support this argument the county relies primarily on a pair of recent Ninth Circuit cases (Gray v. First Winthrop Corp. (9th Cir. 1993) 989 F.2d 1564 and Seattle Audubon Soc. v. Robertson (9th Cir. 1990) 914 F.2d 1311, revd. on other grounds (1992) _ U.S. _ [118 L.Ed.2d 73, 112 S.Ct. 1407].)
These Ninth Circuit cases neither support the broad principle the county asserts, nor is that principle consistent with California Supreme Court cases defining separation of powers in this state, nor would that principle support the constitutionality of Assembly Bill No. 2883, even were it a valid test.
Both of the Ninth Circuit cases dealt with the effect of legislative changes on pending cases not judgments which had been rendered in the past. Even as to the pending cases, the legislative branch did not purport to declare some judicial act “null and void.” Indeed in reviewing Robertson, the Supreme Court took pains to emphasize Congress merely changed the legal standards “without directing particular applications [by the courts] under either the old or the new standards.” (_U.S. at p._[118 L.Ed.2d at p. 83, 112 S.Ct. at p. 1413].)
Neither the United States Supreme Court nor California courts have accepted the notion the Legislature can nullify existing judgments based on previous law just because the same legislation changes that previous law. Indeed United States v. Klein, supra, involved just such a statute. The same legislation both changed the law for future claims ex-Confederates might file and purported to require reversal of judgments on appeal before the Supreme Court. As discussed earlier, the nation’s high court found the latter provision to be invalid as a violation of separation of powers despite the fact the law also purported to change the law for all future cases as well. (See ante, at pp. 1242-1243.) The same is true in California. (See, e.g., Serrano v. Priest, supra, 131 Cal.App.3d 188, where the Legislature passed a law requiring specific appropriations for all fee awards, both those embodied in existing judgments and those awarded in future cases.) Despite the fact this statute represented a general change in the law, the appellate court held it could not apply to existing judgments under separation of powers principles.
Finally, even if the county’s reading of the two Ninth Circuit cases represented an accurate statement of the current law, it would not save Assembly Bill No. 2883. The latter did not embody any change in the substantive law. That change had been accomplished almost a year earlier with the enactment of Welfare and Institutions Code, section 17000.5. (See discussion at pp. 1239-1240, ante.) Assembly Bill No. 2883 had one function and one function only—to *1249target and strike down two or three existing judgments, including the one entered in the Mendly case.
For these reasons, I am not surprised at all my colleagues searched elsewhere for some argument which might save Assembly Bill No. 2883 from its obvious failing under the separation of powers doctrine.
Curiously, the majority quotes language from System Federation which authorizes a court to modify its existing judgment to take account of changes in the law rather than language suggesting the Legislature could have unilaterally modified (or nullified) the court’s judgment. (Of course, this is because System Federation contains no language approving congressional interference with judicial orders or judgments.) As the majority argues this point, “. . . the court ‘must ... be free to modify the terms of a consent decree when a change in law brings those terms in conflict with statutory objectives.’ (System Federation v. Wright, supra, 364 U.S. 642, 651 ....’’ (Maj. opn., ante, at p. 1212, italics added.)
System Federation says nothing about the separation of powers issue. Assuming for the sake of argument a court could exercise its judicial power to modify or nullify that court’s judgment, that does not mean the Legislature can exercise its legislative power to modify or nullify that same judgment. Congress did not purport to declare consent decrees or other judgments based on its “no closed shop” law to be null and void when it reversed course and passed a new law authorizing “closed shops.” If it had done so, based on the rationale of United States v. Klein, supra, there is every reason to expect the United States Supreme Court would have ruled Congress had unconstitutionally intruded on the preserve of the judicial branch of government.
United States Trust Co. v. New Jersey, supra, 431 U.S. 1.
Sonoma County Organization of Public Employees v. County of Sonoma, supra, 23 Cal.3d 307; United Firefighters of Los Angeles City v. City of Los Angeles, supra, 210 Cal.App.3d 1095.
Olson v. Corey, supra, 35 Cal.3d 390; Olson v. Corey, supra, 27 Cal.3d 532.