A century of divided government

How to read this analysis

Current law, proposals, scheduled events and forecasts are labelled separately. Election outcomes are recorded as of 6 November 2026: projected control and dated partial counts, not certified results — see the results baseline. Each forecast states the strongest case against it and the evidence that would change it; the forecast register lists the defined events that will be scored.

Policy outlook for 2027–2030. Evidence cutoff: November 6, 2026. Historical events are documented; the forward judgments are conditional editorial forecasts.

The election changes who can demand information, write a congressional agenda, bargain over appropriations and challenge the executive branch. It does not give the winners control of every institution that determines what Americans experience. A hospital appointment, a housing development, a farmer’s export contract and a school’s special-education service each depend on a different combination of law, money, personnel and private decisions. Political forecasting becomes useful when it follows those combinations.

The Associated Press has called Democratic control of both incoming congressional chambers. Donald Trump remains president. Xavier Becerra is the projected winner of California’s governorship, while official counting and certification continue. Those are the starting conditions for this series, not a claim that all races or final seat totals have been certified.¹²³ The closest recent institutional comparison for Washington is 2007–2008, when Democratic congressional majorities served alongside a Republican president. It is a comparison of decision structures, not a prediction that the same laws, crises or personalities will recur.⁴

Our central judgment is that oversight, executive-legislative bargaining and the terms of implementation will change more readily than the entire statutory framework. Important legislation remains possible. The question is which proposals can assemble the required coalition, survive a presidential veto or receive presidential approval, obtain financing and reach the institutions responsible for delivery. History helps identify those tests. It cannot supply a predetermined outcome.

Begin with the institutions inherited from the last century

The institutional record reaches back more than a century. The Budget and Accounting Act of 1921 established the General Accounting Office, now the Government Accountability Office. Its early focus on the legality of spending later expanded into program evaluation and assistance to congressional oversight.²² That history matters because an opposition majority needs reliable information as well as votes. An investigation that cannot establish what an agency spent, withheld or delivered has a weaker foundation for legislation and enforcement.

The United States entering the late 1920s did not have today’s federal social-insurance system, Medicare, Medicaid or the modern congressional budget process. The Social Security Act became law in 1935. Medicare and Medicaid followed in 1965. The Congressional Budget Act created the Congressional Budget Office in 1974, and the agency began operating in 1975.⁵⁶⁷ The significance of these dates is that today’s disputes take place inside institutions built during earlier conflicts.

Once a program exists, a new administration can inherit beneficiaries, payroll systems, contracts, tax provisions, court interpretations and state agencies. Its options differ from those available before the program was created. Expanding an existing reimbursement stream may be administratively easier than constructing an entirely new payer. Tightening an eligibility condition may be easier to legislate than it is to implement accurately. Repealing a program can require decisions about transition costs and people already relying on it.

That is why an election should be compared with the current-law baseline. If a payment change was scheduled by an earlier statute, its arrival after January does not demonstrate that the new congressional majority caused it. If an expiring provision disappears because no agreement is reached, inaction has changed the practical outcome. If an agency has discretion within a statute, a new enforcement priority may matter without a new act of Congress. Each mechanism needs its own forecast.

The same distinction matters in California. A governor succeeds to operating departments, spending commitments, state constitutional rules and local implementation arrangements. A campaign can announce a goal immediately; changing county software, expanding a clinical workforce or financing a water project takes additional steps. A serious outlook should measure those steps separately and assign credit or responsibility only when the evidence supports it.

Historical episodes reveal mechanisms, not a party formula

Several episodes spanning the last century illustrate different paths to change. The table records institutional lessons to investigate. It does not treat the existence of a law as proof that the law achieved every stated objective.

Episode	Documented event	Mechanism relevant to forecasting
1921	The Budget and Accounting Act established GAO.	Independent scrutiny of spending is an institutional resource that survives changes in party control.
1935	Social Security Act established major federal benefit and assistance arrangements.	Creating an institution changes the baseline inherited by later governments.
1947	Congress overrode President Truman’s veto of the Taft–Hartley Act.	A sufficiently broad congressional coalition can overcome presidential opposition; a simple majority cannot do the same.
1956	Congress and President Eisenhower enacted the interstate highway legislation.	Agreement on an objective still requires a workable financing arrangement and allocation of responsibilities.
1965	Medicare and Medicaid were established.	Different administrative designs produce different federal-state bargaining relationships.
1974–1975	The congressional budget framework and CBO were established.	Independent information and budget procedures are instruments of institutional power.
1986	The Tax Reform Act became law during divided congressional control.	Negotiated packages can combine changes that would face different coalitions if considered separately.
1996–1997	Welfare legislation and the Balanced Budget Act changed major benefit structures; CHIP was created in 1997.	Divided government can produce substantial policy change in different ideological directions.
2007–2008	Energy legislation and emergency financial stabilization legislation became law.	An opposition Congress and president can agree when interests overlap or the cost of inaction changes.
2011 and 2023	Major budget and debt-limit agreements were enacted.	Deadlines and disagreement costs shape bargaining, without guaranteeing which side obtains its preferred terms.

The underlying records are available from the National Archives, Senate, House, Social Security Administration, CMS and published statutes.⁵⁶⁸⁹¹⁰¹¹¹²¹³¹⁴¹⁵¹⁶ The lesson is broader than “divided government works” or “divided government fails.” Governments can legislate while also investigating one another, postpone one dispute while resolving another, and pass measures whose consequences unfold years later.

The 1947 example: distinguish majority power from override power

The Senate’s record shows the Taft–Hartley veto overridden in both chambers in June 1947.⁸ That episode demonstrates a constitutional possibility, not an assumption that today’s majority possesses the necessary votes. For a current proposal, the relevant evidence would be member positions and actual roll calls across party lines. A large election victory does not by itself establish an override coalition on Medicaid, tariffs, immigration or executive authority.

The forecasting error would be to move directly from “the opposition controls Congress” to “the opposition controls legislation.” Congress has multiple forms of power. Setting a hearing, refusing consent to a proposal, attaching conditions to a negotiating position and enacting a replacement statute are different events. Their political and legal thresholds should never be collapsed into one measure of strength.

The 1956 example: agreement needs a financing design

The Senate Historical Office describes how the eventual interstate highway law emerged from competing financing proposals and congressional work, with a Highway Trust Fund central to the enacted arrangement.⁹ The contemporary relevance is practical. Infrastructure promises become more credible when they identify recurring revenue, the public contribution, project selection and delivery responsibilities.

A 2027 agreement on housing, transmission or transportation could fail at any of those stages. Conversely, a narrower financing arrangement might deliver more than a much larger rhetorical commitment. A forecaster should therefore watch the structure of an agreement: who bears construction risk, who pays for operation, which approvals remain outstanding and whether funding can legally be used for the proposed purpose. Announced totals alone are weak evidence of delivery.

The 1974 example: information itself changes bargaining

CBO’s institutional history describes conflict over executive impoundment and Congress’s effort to establish independent budget capacity.⁷ That history is relevant to oversight because executive-legislative conflict is partly a contest over facts: what money was appropriated, what was obligated, what was withheld, which assumptions produced a cost estimate and whether an agency met its legal duties.

An investigation can change the available evidence even if it does not immediately change a statute. It can reveal a contractual problem, establish a more accurate cost baseline or support a later appropriations condition. Yet disclosure is not the same as a remedy. A useful forecast should specify whether it predicts document production, a public finding, a funding decision or a durable legal constraint. Success at the first stage does not guarantee the last.

The last forty years provide closer comparisons

The post-1986 period is especially useful because the institutional environment is closer to the present, although still substantially different. The House’s historical party-control tables provide a disciplined way to identify comparisons. They show, for example, a Democratic House and Republican Senate under Ronald Reagan before 1987; Republican congressional control during much of Bill Clinton’s presidency; Democratic control of both chambers during George W. Bush’s final two years; and different divided configurations under Barack Obama, Donald Trump and Joe Biden.⁴

The 1986 Tax Reform Act demonstrates that a negotiated package can become law across divided institutions.¹⁰ It does not establish that another comprehensive tax reform is likely in 2027. To make that forecast, we would need evidence of overlap on the tax base, distribution of burdens, treatment of businesses, budget effects and transition rules. The historical event supplies a question—where could a package create an acceptable bargain?—rather than an answer.

The 1996 welfare law replaced the former AFDC structure with TANF; the following year’s Balanced Budget Act established CHIP and made other major health-policy changes.¹¹¹² These episodes warn against assuming that an agreement under divided government will necessarily preserve a program exactly as either party originally preferred. A bargain can expand one form of assistance while restructuring another. Its distributional effects must be read in the enacted text and implementation rules, not inferred from the word “bipartisan.”

The 2007–2008 configuration is particularly relevant because both congressional chambers were Democratic while the president was Republican. The Energy Independence and Security Act was enacted in December 2007; emergency financial stabilization legislation followed in October 2008.¹³¹⁴ These are examples of important legislation under that arrangement. They are not evidence that a financial crisis is necessary for all cooperation or that the present Congress will follow the same policy agenda.

The 2011 Budget Control Act and 2023 Fiscal Responsibility Act illustrate another mechanism: negotiations conducted around consequential fiscal deadlines.¹⁵¹⁶ A deadline can increase the value of agreement, but it can also encourage temporary arrangements, narrow bargains or escalating demands. Predicting the direction of a negotiation requires identifying each participant’s alternative if no agreement is reached. Describing one side as having “more leverage” is incomplete until the analysis explains leverage over which decision, at what moment and at what cost.

What political science adds—and what this series has not done

David Mayhew’s Divided We Govern examines lawmaking and investigations across different party-control arrangements. His Yale research materials make underlying lists and datasets available. George Tsebelis’s work on veto players offers a framework for studying how actors whose agreement is necessary constrain policy change.¹⁷¹⁸ These are useful intellectual foundations because they direct attention to institutions and coalitions rather than a single election-night narrative.

This series has not replicated those studies, estimated a new century-long statistical model or calculated a validated probability that any particular 2027 bill will pass. Citing a scholar does not turn a qualitative judgment into the scholar’s conclusion. Nor does the existence of a historical dataset establish that the dataset measures today’s desired outcome. Counting major enactments is different from measuring household affordability, agency compliance or the distribution of benefits across counties.

The appropriate use of this literature is to improve the questions. How many actors must agree? How far apart are their preferences on the specific issue? Can a coalition be assembled by changing the scope, funding or timing? Are the institutions operating under the same procedural rules as the historical comparison? Which actors can change implementation independently? The answers can differ sharply across sectors even under the same partisan map.

Five errors that would make a post-election forecast unreliable

Selecting only the history that resembles a desired outcome. A writer expecting progress can collect examples of productive compromise; a writer expecting paralysis can collect failures. Both collections can be accurate and misleading. A stronger comparison identifies the relevant class of decisions first—for example, negotiated health financing under divided government—and then examines success, failure and partial agreement within that class.

Treating the election as the cause of everything that follows. Prices, hiring, migration and investment can change because of interest rates, international conditions, demographics, technology or an older statute. A new administration may influence those forces without fully controlling them. A decline in a household cost after inauguration does not by itself establish the administration’s contribution; a rise does not identify which institution could have prevented it.

Using the wrong unit of analysis. A national employment measure cannot establish conditions in a particular Fresno neighborhood. An enacted state appropriation is not a county expenditure. A gross agricultural production value is not farm profit. A policy can benefit a statewide average while imposing costs on a particular group. Forecasts should name the geography, affected population and relevant denominator.

Ignoring the default. If a benefit expires automatically, keeping it unchanged may require legislation. If a regulation remains in force, changing leadership may not displace it. If a deadline applies unless relief is approved, an application for relief does not establish a new deadline. Much apparent political surprise comes from failing to describe what happens when decision-makers do nothing.

Equating passage with delivery. A law can create an entitlement, authorize a program or direct an agency to act. Funding, procurement, staffing and technical systems can still determine the result. Each stage should have its own evidence. A dashboard that counts announcements and completed projects in the same column makes accountability weaker precisely when it appears to make progress easier to see.

A worked forecast: congressional oversight in 2027

Consider the proposition that the incoming majority will intensify oversight of executive administration. The authority is relatively direct: congressional organization and committee processes provide channels for hearings, investigations and information demands. The direction of attention is therefore more forecastable than whether a particular investigation will produce compliance, survive a legal dispute or lead to legislation.

The relevant event should be concrete. For example: By June 30, 2027, committees controlled by the new majority will have conducted public oversight proceedings examining the implementation or funding of at least one major domestic-policy program. This is a deliberately modest directional forecast. It can be resolved from committee records. It should not be advertised as proof of unusual predictive skill, because ordinary congressional behavior already provides a strong baseline.

A more ambitious forecast would identify the program, requested records, legal dispute, proposed remedy and deadline. That requires more evidence. The probability of a hearing should not be reused as the probability of a binding remedy. Senate procedure also matters: ending debate on most contested legislation generally requires a larger coalition than a simple majority, while some categories operate under different procedures.¹⁹ Historical analogies do not remove those procedural distinctions.

For a household, the practical consequence may emerge through a later agreement or an administrative response rather than the hearing itself. For a provider or business, the immediate consequence might be record preservation, scrutiny of procurement or a change in agency attention. These are plausible transmission channels, not claims that every investigation produces a measurable improvement.

A worked forecast: California promises and the budget constraint

California’s November 2026 fiscal assessment is a reminder that favorable revenue news and recurring affordability are different questions. The Legislative Analyst’s Office describes an underlying structural shortfall despite the measures used to balance the enacted budget.²⁰ The incoming administration’s ability to expand services depends on which revenues recur, which commitments continue and what federal financing remains available.

The defensible near-term forecast is therefore about selection and implementation: the administration is more likely to prioritize a limited set of funded measures, changes within existing programs and administrative improvements than to deliver every expensive campaign objective simultaneously. That is an editorial inference from a constraint, not a forecast that all spending must decline. A budget can expand one priority while reducing another, rely on temporary measures or move a cost to another payer.

The strongest contrary case is sustained revenue growth, an unexpectedly favorable federal agreement or an effective financing design that changes the constraint. Those developments would warrant revision. Merely repeating a promise would not. The January proposal, May revision and enacted budget supply progressively stronger evidence because they identify actual choices and financing assumptions. Implementation reports later show whether the spending authority became service capacity.

Turning forecasts into a record that can be judged

A credible forecast has a defined event, a deadline, a starting assessment, the evidence supporting it and a rule for revision. The original wording must remain available. If an author changes “a regulation will be issued” into “officials will discuss regulation” after the first event fails to occur, the record no longer measures forecasting performance.

Probability estimates can improve decision-making when they express uncertainty honestly, but numerical precision is not evidence of calibration. Research by Barbara Mellers and colleagues found that training, collaboration and tracking improved performance in a geopolitical forecasting tournament.²¹ It supports a disciplined process. It does not establish an accuracy rate for this publication or justify assigning a precise percentage to every broad political judgment.

For this first edition, qualitative confidence is appropriate where no validated model exists. A scheduled deadline under enacted law belongs in a milestone register. A judgment that officials will negotiate an amendment belongs in a forecast register. A possibility conditional on recession, war or a different 2028 election belongs in a scenario. Keeping those categories separate makes later evaluation possible.

Revision should be tied to observable developments: a bill text, committee vote, court order, approved waiver, appropriation, administrative contract or reliable outcome series. The reason for revision should be published. Better information is a legitimate reason to change an assessment; silently changing the event being predicted is not.

The 2028 boundary and the global audience

The governing arrangement for 2027–2028 should not be projected automatically through 2030. The 2028 presidential and congressional elections can alter the actors whose agreement is required. A longer outlook should therefore branch into conditions: continued divided control, aligned control, or a different division of institutional power. None should be presented as the result of an election that has not occurred.

Foreign governments and international businesses face the same problem of distinguishing durable law from temporary policy. A congressional hearing, an executive announcement, an enacted appropriation and a long-term statutory commitment carry different implications. A negotiated policy with broader institutional support may be more durable, but durability still depends on its legal design, funding and future political incentives. Historical precedent can identify those dimensions without guaranteeing continuity.

For California and the Central Valley, international effects return through trade, agricultural demand, energy prices, research partnerships, migration and supply chains. A national foreign-policy forecast becomes locally useful when it identifies those channels and the institutions that can respond. It becomes misleading when a global headline is converted directly into a precise county employment or household price prediction without a model and supporting data.

The judgment for this edition

History supports a disciplined expectation of active bargaining, stronger opposition oversight and uneven policy change across sectors. It also supports the possibility of significant agreements when decision-makers share an objective, face a consequential default or can assemble a package acceptable to the necessary coalition. Neither paralysis nor comprehensive reversal should be treated as inevitable.

The test for the next two years is concrete: which legal changes occur, which receive financing, which institutions implement them and which people experience a measurable difference? The accompanying sector articles apply that sequence to healthcare, housing, work, technology, family support, energy, rights and international affairs. The Fresno scorecard applies it to the local institutions where national and state policy becomes a renewal notice, a construction permit, a utility bill or an available appointment.

The value of this outlook will depend on preserving that chain of evidence. Readers should be able to see what was known, what was inferred, what could make the inference wrong and how the assessment changed when new information arrived.

Sources and reading notes

Source notes

Numbered notes support the sentence that carries the number. A source supports the claim it is cited for, with its own date and scope; it is not an endorsement by the issuing institution.

Associated Press, Calling control of Congress in the 2026 midterm elections, November 5, 2026. Supports called chamber control, not a certified final seat total. ↩
KQED election results, California governor, accessed November 6, 2026. AP projection and live count; preserve the date of any numeric snapshot. ↩
California Secretary of State, November 3, 2026 general election. County canvass and statewide certification dates are distinct from media calls. ↩
Office of the Historian, U.S. House, Party Government Since 1857. Institutional control, not a measure of legislative productivity or public welfare. ↩↩
National Archives, Social Security Act, 1935. ↩↩
National Archives, Medicare and Medicaid Act, 1965. ↩↩
Congressional Budget Office, History. Establishment of the budget institutions and CBO; historical description, not a current cost estimate. ↩↩
U.S. Senate, Vetoes by President Harry S. Truman. Records the June 1947 Taft–Hartley override. ↩↩
Senate Historical Office, Congress Approves the Federal-Aid Highway Act, and National Archives, National Interstate and Defense Highways Act, 1956. ↩↩
Internal Revenue Service historical statistics, Tax Reform Act of 1986 discussion, hosted by the U.S. Government Publishing Office. Historical law and tax-system context; no causal economic estimate is asserted here. ↩↩
Social Security Administration, 1996 Welfare Amendments. ↩↩
Balanced Budget Act of 1997, Public Law 105-33; CMS, Medicare and Medicaid Milestones, 1937–2015. Historical enacted provisions, including CHIP creation. ↩↩
Energy Independence and Security Act of 2007, Public Law 110-140, December 19, 2007. ↩↩
Public Law 110-343, October 3, 2008, including the Emergency Economic Stabilization Act. ↩↩
Budget Control Act of 2011, Public Law 112-25, August 2, 2011. ↩↩
Fiscal Responsibility Act of 2023, Public Law 118-5, June 3, 2023. ↩↩
David R. Mayhew, Divided We Govern, second edition; Yale, associated datasets and materials. The present article has not replicated the study or re-estimated its results. ↩
George Tsebelis, Veto Players: How Political Institutions Work, Princeton University Press. Used as an analytical framework; this article does not claim a new empirical estimate from the book. ↩
U.S. Senate, Filibusters and Cloture. The general legislative cloture rule has exceptions and should not be applied indiscriminately to nominations or special procedures. ↩
California Legislative Analyst’s Office, November 2026 fiscal assessment, Report 5211, November 5, 2026. The distinction between recurring commitments and the enacted budget’s balance is central to the inference. ↩
Barbara Mellers and colleagues, Psychological Strategies for Winning a Geopolitical Forecasting Tournament, Psychological Science, 2014. Evidence for forecasting practices; not a validation of this series. ↩
U.S. Government Accountability Office, History. Establishment in 1921 and evolution of the audit and evaluation role; this is institutional history rather than a forecast of a particular investigation. ↩

What a Century of American Government Can—and Cannot—Tell Us About 2027