What if we saved instead of borrowing?

Smith County borrowed $251 million from 2018 through 2025 — road bonds nearly every year, then a new courthouse and parking garage in 2023. What if it had saved first and paid cash instead? This page replays that, using the county’s own bond schedules and the real yearly returns of Treasury bills, bonds, and stocks. Updated 2026-09-26.

Play it — pick a start year and a stretch of market history

Showing: saving starts in 2000; markets follow what really happened. Each plan buys the same things on the same dates.

Smith County — $251 million borrowed

What was borrowed for

2018	Road bonds, Series 2018	$11,320,000
2019	Road bonds, Series 2019	$11,320,000
2020	Road bonds, Series 2020	$7,125,000
2021	Road bonds, Series 2021	$7,425,000
2022	Road bonds, Series 2022	$17,270,000
2023	Courthouse parking garage bonds	$18,170,000
2023	New courthouse bonds	$152,335,000
2024	Road bonds, Series 2024	$6,675,000
2025	Road bonds, Series 2025	$19,375,000

Borrow what actually happened

$410 million taxpayers pay · $159 million of it interest

$63.7 million paid through 2026. $346 million still due, the last in 2048.

Save first, in Treasury bills allowed today

$224 million taxpayers pay · $8.6 million a year for 26 years

Every project paid in cash on the same date. Nothing owed. $186 million less than borrowing.

Save first, in 60% stocks and 40% bonds not allowed for county money

$110 million taxpayers pay · $4.2 million a year

Pick this amount ahead of time and it came up short in 12 of 73 stretches of market history since 1928 — once by $61.8 million.

Save first, in stocks not allowed for county money

$84.3 million taxpayers pay · $3.2 million a year

Pick this amount ahead of time and it came up short in 25 of 73 stretches of market history since 1928 — once by $126 million.

Invest and borrow against it not allowed for county money

$224 million taxpayers pay · $8.6 million a year into stocks

Borrow just as it really did, and pay the bonds from the fund. By 2026 the fund holds about $1.17 billion while $346 million of bond payments are still due — about $822 million ahead. For comparison, the same money in stocks paying cash for each project would hold about $784 million.

Payments through 2026 are estimated from each bond’s own interest rate (about $33.7 million of interest). Every payment from 2027 on is exactly as the county prints it. Sources: Smith County Annual Comprehensive Financial Report, FY ended Sept 30, 2024, Note VIII, p. 48 — original amount of each bond, 2018–2023 · Smith County FY2025-26 Adopted Budget, p. 184 — 2024 and 2025 bond payments (original amounts rebuilt from them) · Smith County FY2026-27 Budget (posted as “FY 2027 Proposed Budget”), pp. 27–30 — every payment still due, by year, from Oct 1, 2026

Starting today — the next round, not the last one

Smith County owes $346.3 million in principal and interest on the bonds above, through 2048 — that gets paid no matter what. The live question is the next road program. From 2018 to 2025 the county borrowed about $10 million a year for roads.

Keep borrowing: a $10 million, 20-year bond at 4.5% costs about $0.77 million the first year. Borrow again every year and the payments stack up to about $15.5 million a year by 2046 — and stay there.
Pay cash instead: $10 million a year from the first year — about 3 cents per $100 of property value, on top of the old bond payments until they end in 2048.
Borrowing is cheaper to taxpayers until about 2054. From then on, paying cash saves about $5.4 million every year.
For comparison: Tyler’s city government states it owes $0 in debt paid by property taxes.

Sources: Smith County FY2026-27 Budget (posted as “FY 2027 Proposed Budget”), pp. 27–30 · City of Tyler FY2026-27 Budget, cover page

“Invest it and borrow against it” — why counties can’t, and who can

Wealthy people borrow against their stocks so they never have to sell them and pay capital-gains tax. Governments pay no income tax, so that part doesn’t apply. What’s left is a bet: that investments earn more than the loan costs. In the table above, most of the gain comes from investing early — not from the borrowing.

Texas law lets a county invest only in government-type securities, bank CDs, and funds “invested exclusively in obligations approved by this subchapter” (Gov’t Code §2256.014). No stocks.
Arkansas law limits county and city spare cash to securities “having a maturity of not longer than five (5) years,” bank CDs, pools, and repurchase agreements (§19-1-504). No stocks.
Federal law won’t let tax-exempt bond money be invested at a higher return than the bond pays; extra earnings go to the U.S. Treasury (26 U.S.C. §148).
The warning: in 1994, Orange County, California used borrowed money to stretch a $7.6 billion investment pool to more than $20 billion. It lost about $1.7 billion and filed for bankruptcy on December 6, 1994 (SEC order).
The version that works is run by the state: approved Texas school bonds are “guaranteed by the corpus and income of the permanent school fund” (Education Code §45.052), so districts borrow at top-rated rates — borrowing against investments, with a state-sized fund behind it (texaspsf.org).

What this doesn’t show — the honest limits

Saving first means paying for years before anything gets built. Borrowing gets the building now.
People who move here later use a building they never saved for. Spreading the cost over the years people use it is the fair argument for bonds.
A growing pot of money has to survive many elections without being spent on something else.
Money left in residents’ pockets while a county borrows isn’t worth nothing to them.
All figures are in dollars of their own year, not adjusted for inflation. Only the bonds listed above are counted.

Market returns: Aswath Damodaran, NYU Stern — Historical Returns on Stocks, Bonds and Bills (histretSP.xls), retrieved 2026-09-26 — data page. Calendar-year returns. stock = S&P 500 including dividends; tbill = 3-month U.S. Treasury bill; tbond = 10-year U.S. Treasury bond. Returns run through 2025; 2026 is counted with no market growth. This page computes and cites; it doesn’t recommend a plan. Spot an error? Report it.