About
Context & sources
Two things on this page: background context for understanding Pennsylvania's fiscal situation, and methodology for how the dashboard works and where the numbers come from.
What "structural deficit" means
Pennsylvania's General Fund — the state's main operating budget — is on track to spend roughly $50 billion this fiscal year while collecting about $45 billion in revenue. The gap between the two is the operating deficit; the portion that's permanent and recurring rather than papered over with one-time funds is what economists call the structural deficit.
The PA Independent Fiscal Office (IFO) currently puts that figure at $3.9 billion for FY 2025-26, projected to grow to $6.7 billion next year and $8.4 billion by FY 2029-30 if nothing changes. The key word is structural: it persists even in good economic years because spending is growing at ~5%/yr while revenue grows at ~2%/yr — a mismatch that compounds annually.
Restricted funds and the "shadow budget"
The $3.5B "Non-Tax Revenue" line on the revenue chart is the narrow slice of fees, licenses, fines, and treasury earnings that actually enter the General Fund. Alongside the General Fund sit 150+ "special funds" — dedicated pools of fees and earmarked taxes that flow directly to specific purposes by statute and never appear in the revenue chart.
Examples: the Motor License Fund (gas tax + vehicle fees → highways and bridges), the Lottery Fund (proceeds → senior programs), the Public Transportation Trust Fund (→ transit), and the Tobacco Settlement, Environmental Stewardship, Oil & Gas Lease, and Race Horse Development funds, among many others.
Because this money is dedicated, it can't legally be used to close a General Fund deficit. But the legislature periodically rewrites statutes to authorize fund transfers or "fund sweeps" — moving surplus balances out of restricted funds into the General Fund. That's part of what "papered over with one-time funds" means: a fund sweep closes this year's gap without changing next year's math.
A few specific examples. For decades the state funded the State Police out of the Motor License Fund — gas-tax money intended for highways; Act 85 of 2016 began a phase-down. Act 43 of 2017 securitized the Tobacco Settlement Fund — selling ~$1.5B in future payments as bonds for one-time cash, locking in ~$115M/yr in lost health revenue going forward. In PEDF v. Commonwealth (2017, reaffirmed 2021), the PA Supreme Court held that royalties from oil-and-gas leases on state forest land are held in public trust under Article I §27 — the leading legal limit on how far these transfers can go.
The Rainy Day Fund: law, history & how PA compares
Pennsylvania's Budget Stabilization Reserve Fund is legally restricted to "emergencies and downturns in the economy that result in unexpected revenue shortfalls that cannot be dealt with through the normal budget process." The operative word is unexpected. The FY 2026–27 structural deficit is neither unexpected nor caused by an economic downturn — the IFO has been projecting this gap for years. Using the fund to cover it arguably violates the spirit of the restriction, even if a simple legislative majority makes it technically legal. The Commonwealth Foundation argues the statute also requires a two-thirds supermajority to withdraw — that interpretation is contested but unresolved.
PA currently covers ~53.6 days of General Fund operations from its reserve — above the 50-state median of 49.1 days. After a $4.6B drawdown, that falls to roughly 15 days.
| State | Withdrawal rule | Key guardrail |
|---|---|---|
| Texas | 2/3 supermajority of legislature required | High bar prevents routine or election-year access |
| California | Prop 2 (2014): automatic deposits when revenues spike; withdrawals only during recessions | Constitutionally restricted — cannot fill structural gaps |
| North Carolina | Restricted to temporary revenue shortfalls; recurring program use prohibited | Explicit ban on funding ongoing structural imbalances |
| Virginia | Revenue Stabilization Fund: formula-driven deposits, tied to revenue volatility | Withdrawals linked to measurable economic conditions, not political convenience |
| Pennsylvania | Simple majority; "unexpected shortfalls" language only — interpretation contested | No explicit bar on use for predictable, structural deficits |
Sources: Pew Charitable Trusts state reserve research · City & State PA · Commonwealth Foundation · PA Fiscal Code · Ballotpedia
PA compared to peer states
States with similar population or geographic proximity. Figures are approximate structural balances — not budgeted totals — sourced from Pew Fiscal 50 (FY 2023–24). PA is highlighted.
| State | Pop. | Gov. | Legislature | Balance | % of GF |
|---|---|---|---|---|---|
| Pennsylvania ★ | 13M | D | Split | -3.9B | -8.7% |
| Ohio | 11.8M | R | R | +1.5B | +6.0% |
| New Jersey | 9.3M | R | Split | -4.0B | -7.5% |
| New York | 19.6M | D | D | -7.5B | -6.7% |
| Maryland | 6.2M | D | D | -3.0B | -12.0% |
| Virginia | 8.7M | D | Split | +0.5B | +1.7% |
| North Carolina | 11M | D | R | +3.0B | +10.0% |
| Michigan | 10.1M | D | D | +0.5B | +1.9% |
| Georgia | 11.1M | R | R | +2.5B | +7.8% |
| Illinois | 12.7M | D | D | -5.0B | -9.4% |
| Florida | 22.6M | R | R | +3.0B | +7.7% |
★ = Pennsylvania. Gov. and legislative control as of January 2026; 2025 elections may have updated NJ and VA. All figures approximate — verify against Pew Fiscal 50 for current data.
The revenue side: where PA stands
Closing a structural deficit has two levers: spend less or collect more. The spending side gets most of the attention in Harrisburg. The revenue side has its own structural story — PA's tax base is narrower than most peer states, for reasons that are partly constitutional, partly political, and partly a matter of policy choices that compound over time.
PA fully exempts pensions, Social Security, 401k, and IRA distributions from state income tax. Most peer states tax at least some retirement income. As PA's population ages, this exemption grows automatically — it's a structural revenue leak, not a one-time policy. Estimated cost: $2–3B/yr in foregone revenue if taxed at PA's existing 3.07% rate with low-income protections.
PA is the 2nd-largest natural gas producer in the US — behind only Texas. Texas collects $10–15B/yr in oil and gas production taxes. PA collects a flat "impact fee" per well: ~$200M/yr total, regardless of production volume or price. A standard 5% severance tax is estimated to raise $500M–$1B/yr at current production, higher in strong price years. Source: PA Budget and Policy Center.
PA's constitution has a "uniformity clause" that courts have interpreted to require a single flat income tax rate for all filers. A progressive rate structure — where higher earners pay a higher percentage — would require a constitutional amendment. Voters rejected one in 2004. This is a hard constraint, not a policy choice, but it limits how much revenue can grow with economic prosperity.
How other states tax income — and what they do (or don't) exempt. PA highlighted.
| State | Type | Top rate | Retirement income |
|---|---|---|---|
| PA ★ | Flat | 3.07% | Fully exempt |
| OH | Progressive | 3.99% | Taxed (with credits) |
| NJ | Progressive | 10.75% | Partial exemption |
| NY | Progressive | 10.9% | Taxed |
| MD | Progressive | 5.75% | Partial exemption |
| VA | Progressive | 5.75% | Partial exemption |
| NC | Flat | 4.5% | Partial exemption |
| MI | Flat | 4.25% | Partial exemption |
| GA | Flat | 5.49% | Partial exemption |
| IL | Flat | 4.95% | Fully exempt |
| FL | None | — | N/A |
| TX | None | — | N/A |
Sources: Tax Foundation State Individual Income Tax Rates · NCSL Retirement Income Tax · PA IRRC
Data centers in PA: scale, power, and who captures the value
Pennsylvania leads all states in planned data center growth — a 188% increase in announced facilities. That's not a trend; it's an industrial transformation happening right now, concentrated in specific counties, with real fiscal consequences that aren't fully visible in press releases or public debate.
Salem Township (1,200 acres rezoned as Special Data Center Overlay District, May 2024) + Falls Township near Morrisville. 15 data center buildings over 10 years. 1,250 jobs projected.
Carlisle. 1.35 GW capacity across three 700-acre campuses. Joint venture: Pennsylvania Data Center Partners + PowerHouse Data Centers. Direct fiber peering to Ashburn, VA. First 300 MW by Q2 2027.
3,200 acres on a decommissioned coal plant site. 4.5 GW natural gas plant with seven hydrogen-enabled turbines. Built specifically to power data centers. 10,000 construction jobs; 1,000 permanent. First power 2028.
Upper Macungie Township. 2.6M sq ft across three buildings on 194 acres. Zoning approvals in progress (2026). Part of the I-78 corridor overflow from Northern Virginia.
Why PA? Northern Virginia (the world's largest data center concentration) is running out of power connections — Dominion Energy has a multi-year backlog. The I-78/I-81 corridor offers cheap Marcellus gas for power, proximity to the Northeast fiber backbone, and low-latency connections to Ashburn. PA isn't being chosen despite its costs; it's being chosen because of its gas.
Data centers are projected to drive 30 GW of new peak load in PJM by 2030 — the regional grid that covers PA and 12 other states. PA is expected to supply 80% of the new generation needed to meet that demand. The consequences are already showing up in electricity prices:
The capacity price spike is effectively a regional electricity surcharge on households and businesses to build generation that serves data centers. PA ratepayers pay this whether or not data centers are located in PA. What PA does control is whether the state captures fiscal value from hosting the facilities and producing the gas.
Data centers create a distinctive fiscal split between state and local government — one that's rarely discussed clearly:
The state is forgoing $517M/yr in sales tax to generate property tax revenue that flows entirely to local school districts and municipalities. Whether that's a good trade depends on whether you think data centers would come to PA without the incentive — and whether PA's competitive position vs. Virginia and Georgia requires matching their packages. The incentive has no mandatory reporting, so the answer isn't currently knowable from public data.
PA produces 19% of US natural gas from the Marcellus Shale — roughly one-third of all US shale gas output at ~26.6 billion cubic feet per day. Data centers require 24/7 dispatchable power; natural gas is the default backstop even for operations claiming renewable energy targets. Homer City is the clearest illustration: a purpose-built 4.5 GW gas plant being constructed specifically to power data centers.
More data center investment in PA — and more data center investment nationally running on gas — means more demand for Marcellus production. PA captures ~$200M/yr from that production via the flat impact fee. Texas captures $10–15B/yr from comparable extraction via percentage-based severance taxes. The data center boom is accelerating the underlying demand that makes the severance tax gap larger each year.
Hyperscale data centers use water to cool servers — typically 1–5 million gallons per day for a large facility, with 70–80% evaporated rather than returned to waterways. That's a real number worth knowing. Here's what similar-or-larger industrial facilities use:
| Facility type | Daily water use | Notes |
|---|---|---|
| Hyperscale data center | 1–5M gal | 70–80% evaporated. Liquid cooling uses more water but less electricity than older air-cooled systems — a deliberate tradeoff. |
| Nuclear power plant | 25–60M gal | 5–12× more than a data center. Nearly all evaporated through cooling towers. PA has 5 operating nuclear units. |
| Semiconductor fab (chip factory) | 3–5M gal | Makes the chips data centers run. Comparable scale — TSMC's Arizona fab uses ~4M gal/day. |
| Coal or gas power plant | 10–50M gal | PA still operates both. Homer City's new gas plant will require cooling water on top of what the data centers it powers draw. |
| Paper / pulp mill | 10–20M gal | Long-established PA industry; higher use with more direct discharge to waterways. |
| 18-hole golf course | 100–300K gal | Less than a data center per facility. PA has ~700 courses. Included for scale reference. |
Each generation of AI chips delivers roughly 2–3× more computation per watt: NVIDIA A100 → H100 → B200. Water and energy per unit of actual compute — not per facility — have declined consistently as newer hardware ships. Microsoft and Google both publish annual water intensity numbers and both show declining gallons-per-computation. The per-query cost of AI is falling fast even as total volume grows.
A single facility's permit doesn't reveal what happens when four facilities draw from the same watershed in a dry summer. Cumulative withdrawal in a single basin is a different question from individual permit compliance — and it's not clear from public data whether PA's PADEP is tracking it in aggregate for data centers specifically.
Data centers are among the least labor-intensive structures in the economy per dollar invested. The distinction between construction employment and permanent operations employment matters and is frequently blurred in economic development announcements:
Workers for a 1 GW facility over ~18 months. Homer City projects 10,000 construction jobs for its combined gas + data center build. These are real jobs — but temporary.
Per 100 MW of capacity. A 1 GW hyperscale facility typically employs 200–300 people permanently. Amazon's $20B investment projects 1,250 "high-skilled" permanent jobs across multiple facilities over a decade. (Source: Brookings Institution)
Brookings (2024) found county employment rises 11% during construction, then settles at a 4–5% permanent increase — real economic activity, but concentrated in the construction and information sectors, not broadly distributed across the labor market.
The PAX campus, the Amazon facilities, Homer City — these aren't being built to store files. They're the physical infrastructure for AI workloads that require enormous compute, and that compute is enabling things that weren't possible at scale before:
Whether any individual company's product delivers on its promises is a separate question from whether the underlying capabilities are real. The infrastructure is being built regardless of where PA sits in the debate — the fiscal and environmental choices are about whether PA captures the upside while managing the costs, not about whether to permit the technology.
- —Act 25 has no mandatory reporting. How much has the exemption actually cost, and which companies have claimed it?
- —Would the Amazon and PAX investments have come to PA without the incentive, given the natural gas advantage already exists?
- —PJM projects PA supplying 80% of new regional generation. What does that mean for PA electricity rates specifically, and who is modeling it?
- —Homer City's 4.5 GW gas plant is sized to power data centers. How does that fit with PA's clean energy commitments, and who approved it?
- —PA DEP issues individual water withdrawal permits, but there's no public dashboard tracking cumulative data center withdrawals by watershed. As more facilities come online in the same river systems, who is aggregating the total draw?
Sources: Spotlight PA · Data Center Dynamics · PJM Interconnection 2025 Long-Term Load Forecast · Brookings Institution · PA General Assembly (SB 939) · PowerHouse Data Centers / Pennsylvania Data Center Partners · Goldman Sachs · DOE/Lawrence Berkeley National Laboratory · Marcellus Coalition
All states at a glance: surplus vs. structural deficit
The states not in the comparison table above, grouped by approximate fiscal position. These fluctuate — states near zero move between categories year to year.
Florida · Texas · North Carolina · Georgia · Ohio · Tennessee · South Carolina · Utah · Colorado · Idaho · Montana · Wyoming · North Dakota · South Dakota · Nebraska · Iowa · Kansas · New Hampshire · Indiana · Missouri · Oklahoma · Arkansas · Mississippi · Alabama · Alaska
Illinois · California · New York · New Jersey · Connecticut · Hawaii · Massachusetts · Minnesota · Oregon · Washington · New Mexico · Rhode Island · Delaware · Kentucky · Louisiana · Maryland · Michigan · Arizona · Nevada · Wisconsin · West Virginia
Approx. FY 2023–24. Source: Pew Charitable Trusts Fiscal 50. States in the comparison table above are not repeated here.
What surplus states do differently — spending & revenue
Limits how fast appropriations can grow — typically tied to population + inflation or personal income growth. Prevents the structural mismatch from compounding silently in years when revenue grows slower than spending. PA could adopt this by statute without a constitutional amendment.
Texas requires a 2/3 legislative vote to withdraw from its rainy day fund. The high bar makes election-year drawdowns politically costly. PA currently requires only a simple majority, and the 'unexpected shortfall' language is contested — a supermajority rule would be the cleaner fix.
Virginia's Revenue Stabilization Fund automatically deposits money when revenues spike above forecast and withdraws only when economic conditions deteriorate. Removes the political discretion that lets structural deficits accumulate quietly in good years.
NC and OH have both reduced income tax rates in recent years — but tied the cuts to surplus benchmarks. Rates only drop when the state hits a revenue target first. PA has debated income tax reform without the trigger mechanism that makes cuts fiscally responsible.
PA is the 2nd-largest natural gas producer in the US and collects only a flat 'impact fee' (~$200M/yr). Texas and Wyoming fund significant portions of their budgets from percentage-based severance taxes on oil and gas production. A PA severance tax is estimated to raise $500M–$1B annually at current production volumes, rising with natural gas prices. Source: PA Budget and Policy Center.
GA and NC apply moderate flat rates with fewer large exemptions. PA fully exempts all retirement income — pensions, Social Security, 401k, and IRA distributions — a growing gap as the state ages. Most peer states tax at least some retirement income. Taxing retirement income at PA's existing 3.07% flat rate (with low-income protections) is estimated to raise $2–3B annually.
The spending-side items are statutory or constitutional fiscal rules — any can be proposed as a standalone bill. The revenue-side items range from politically difficult (retirement income) to constitutionally constrained (progressive rates) to consistently stalled despite bipartisan economic logic (severance tax).
Where this is headed — two schools of thought
Most budget analysts — left and right — largely agree on the diagnosis: states baselined recurring programs on one-time federal COVID money, reserves are being spent at the top of the economic cycle instead of saved for the bottom, and annual "balanced budget" rules are easy to satisfy with one-time fixes that leave next year's math untouched. The disagreement is about the cure. Two composite views:
"The failure is structural, so the fix must be mechanical — rules that hold in good years, when discipline is politically hardest."
- ·Require budgets to balance on recurring revenue vs. recurring spending — not just annual cash
- ·Cap spending growth to a formula (population + inflation, or personal income)
- ·Automate reserve deposits; require supermajorities to withdraw
- ·Set pension funding minimums that can't be skipped
Evidence cited: NC, OH, GA, VA, and TX run surpluses under rules like these.
The critique: rules are blunt and procyclical — they force cuts in recessions, and Colorado's TABOR (the strictest version) squeezed school funding until voters suspended parts of it.
"The deficit reflects an eroding tax base and rising needs, not overspending — austerity rules would lock in underinvestment."
- ·Update the tax base to match the modern economy: services, retirement income (with low-income protections), gas extraction at a percentage
- ·Treat education and health spending as investment with long-run returns, not just cost
- ·Keep budget flexibility for recessions instead of formula straitjackets
Evidence cited: MN and MA combine high service levels with mostly balanced budgets through broad, modern revenue bases.
The critique: "trust the legislature" has a poor track record — without binding rules, new revenue tends to get absorbed into new spending and the structural gap reappears.
- —One-time money should be separated from recurring money on the face of the budget
- —Reserves are for recessions, not election years
- —Voters should be able to see the structural gap before it becomes a crisis
These are composite schools of thought drawn from state fiscal policy research — not WTPPPA platform positions. The overlap — honest accounting and visibility — is the part this dashboard exists to serve, whichever cure you favor.
Where each number comes from
- Live deficit ticker
- Calculated from IFO's $3.9B annual structural deficit projection, ticked up second by second based on fraction of the fiscal year elapsed (PA fiscal year runs July 1 – June 30). The number is "deficit accrued so far this fiscal year." Source: PA IFO.
- Revenue and spending breakdowns
- FY 2025-26 General Fund totals (~$45B revenue, ~$50B spending) from PA IFO Monthly Revenue Update categories and the FY 2025-26 Enacted Budget published by the Office of the Budget. Per-category splits are approximate; dollar totals reconcile with IFO figures.
- Federal funding context + cliff scenarios
- Compiled from USAspending.gov, KFF Medicaid State Indicators (FMAP rates), and the Office of the Budget federal funds reporting. Cliff scenarios are modeled estimates based on current program sizes — actual cuts depend on specific federal legislation.
- Federal money by county
- Pulled from the USAspending.gov API for the most recent complete federal fiscal year. Important caveat: totals include all federal obligations — contracts, grants, direct payments, loans, and insurance — not just "aid." Large federal contractors and research institutions (Pitt, UPMC, Penn, federal labs) significantly inflate metro-county totals. This is not a measure of state aid to your county.
- Personal impact calculator
- Anchored on IFO's own "$1,500 per family of 4" figure for closing the FY 2025-26 deficit. We scale per capita by household size, then sum projected annual shares using IFO's growth curve ($3.9B → $6.7B → $8.4B), extrapolated linearly beyond FY 2029-30.
- Rainy Day Fund countdown
- Counts down to the IFO-projected depletion window (FY 2026-27) based on the current ~$7B balance and the proposed $4.6B drawdown rate.
- Latest from PA IFO
- Scraped automatically from ifo.state.pa.us by a GitHub Actions cron job that runs every Monday at 06:00 UTC. New publications appear within ~7 days of release. Last refreshed: June 8, 2026.
What's auto-updated vs. seed data
Limitations and caveats
- —This dashboard scopes to the General Fund. Pennsylvania also operates 150+ restricted special funds outside the GF; see "Restricted funds and the shadow budget" above.
- —Per-category dollar splits are approximate; absolute totals reconcile with IFO and Office of the Budget figures.
- —"Cliff scenarios" are modeled estimates to show order-of-magnitude impact, not forecasts of specific cuts.
- —The personal impact calculator uses per-capita scaling — PA's flat income tax makes this roughly linear, but exemptions and credits vary across households.
- —State comparison figures (above) are approximate FY 2023–24 structural balances from Pew Fiscal 50; verify before citing.
A note on AI
We used Claude, an AI assistant from Anthropic, to help write code and draft copy for this dashboard. The data is not AI-generated — every number comes from a public PA source we cite.
The energy used during development was small — roughly comparable to streaming a few hours of video.