The Tide and the Gate: The President, the Party, and the Growth Gap

The question arrives every election season, and it usually arrives as a taunt: which party is better for the economy? For seventy years the honest answer has been that the data seem to favor one side, and the argument has therefore been about whether the data mean anything. That argument deserves care, because the strongest evidence for the claim and the strongest reasons to doubt it come from the same study, and that study’s own conclusion is not the one either side quotes.

So let me lay out what the record shows, what it cannot show, and why the most interesting finding in the literature is one that neither party has an incentive to advertise.

I. The finding

The reference work is by Alan Blinder, a former vice chairman of the Federal Reserve, and Mark Watson, a Princeton econometrician. Presidents and the US Economy: An Econometric Exploration appeared in the American Economic Review in 2016 and covers sixteen complete presidential terms, from Truman’s elected term through Barack Obama’s first (Blinder and Watson, 2016).

Its central numbers are these.

Over a typical four-year term, the economy grew 18.5 percent when a Democrat held the White House and 10.6 percent when a Republican did. The volatility was the same either way, so the Democratic terms were faster without being riskier.

The recession record is more lopsided still. Of the 49 quarters the National Bureau of Economic Research classifies as recessionary between 1949 and 2013, eight came under a Democratic president and forty-one under a Republican one.

Unemployment fell by 0.8 percentage points on average during Democratic terms and rose by 1.1 points during Republican ones. Corporate profits were higher under Democrats, as were stock returns, though the stock result is not statistically significant, because prices are volatile enough to swamp a gap that large.

The authors take the obvious objection seriously, that sixteen terms is a small sample from which to infer anything. They test it three ways: with standard errors clustered by term, with Newey-West errors that allow for correlation over time, and with a permutation test that randomly reassigns party labels to the sixteen four-year blocks of data. Under random assignment, an absolute gap at least as large as the observed one arises about once in a hundred draws. On its own terms the correlation is real, and it is not the artifact of one observation.

Then the paper says something neither side repeats.

II. The authors’ own verdict

Blinder and Watson spend most of their paper hunting for the mechanism, and they cannot attribute the gap to policy.

Four factors together explain most of it: oil shocks, productivity, defense spending, and the growth of foreign economies. Their conclusion is that the analysis “does not attribute any of the partisan growth gap to fiscal or monetary policy.” Their measures suggest, if anything, that fiscal stabilization was “somewhat more stabilizing under Republicans,” and that budget deficits were larger under Republican presidents, which is the opposite of what the deficit-spending story predicts. Deficits should have helped Republicans, and the growth went the other way.

The productivity number is the one that deserves the most attention. Total factor productivity, the efficiency with which an economy turns labor and capital into output, grew 1.89 percent a year under Democratic presidents and 0.84 percent under Republican ones. Productivity is not a lever in the Oval Office. It is closer to a fact about technology, workforce, and timing, and much of it belongs to the private economy rather than the state.

The authors also test the reverse-causality story, that Democrats merely happened to be elected at moments when the economy was already poised to grow. It fails in their data. Republicans inherited higher growth from their predecessors, 4.25 percent in the prior term’s final year against 1.94 percent for Democrats, and the partisan gap shows up largest in the first year of a term, not the fourth. Professional forecasters did not see it coming: the Survey of Professional Forecasters predicted essentially identical growth, 3.1 percent under Democrats and 3.2 percent under Republicans, and what arrived was 3.5 and 1.0.

They also ask whether Congress, rather than the president, is the variable that matters. It is not. Growth under Republican presidents stayed below three percent regardless of which party controlled the House and the Senate.

Their closing summary is the fairest sentence in the paper, and it is not a partisan one. Having accounted for oil, productivity, defense, and the rest, “The rest remains, for now, a mystery of the still mostly unexplored continent.”

III. The recession record, read honestly

The recession count is the most quotable part of this and the least understood, so it is worth reading rather than repeating.

From the NBER’s own chronology, counting the recessions that began between 1953 and 2020, ten of eleven began while a Republican held the presidency. The single exception is 1980, under Jimmy Carter (NBER, 2023).

That is a striking fact. It is also a weak instrument. What begins a recession is usually a shock rather than a policy: the 2001 recession followed the collapse of the dot-com bubble and the September 11 attacks; the 2007 recession was a financial crisis that began in mortgage securitization; the 2020 recession was a pandemic. A president is charged with the timing of such events the way a captain is charged with the weather.

Two Democrats also hold the other end of that ledger. The 1948 recession began under Truman, and the 1980 recession began under Carter, whose proximate cause was a credit-control policy error at the Federal Reserve rather than anything the White House chose. The count is real. The inference that a president’s party causes recessions is not supported by it.

IV. Three problems with the strong reading

Fairness requires naming what this correlation cannot carry.

The sample is small and the pattern is fragile at its edges. Because the distribution of growth rates is skewed, the median gap is roughly half the mean, and the headline number is doing more work than the average term. The gap also shrinks steadily as the sample rolls forward: 4.07 points ending after Eisenhower, 3.12 after Nixon and Ford, 2.41 after the first Bush, and 1.79 across the full sample. Extend the data back before 1947 and the picture inverts, because from 1875 to 1947 growth was higher under Republicans once Franklin Roosevelt’s terms are excluded. A finding that can be reversed by removing one president is a claim about a particular era, not a law of economics.

The president does not run the economy. The Federal Reserve sets monetary policy and historically tightens or eases on its own schedule. Congress holds the purse, and a presidential budget is a proposal rather than a command. The world price of oil, the pace of a trading partner’s growth, and the arrival of a pandemic are on no president’s desk. Blinder and Watson found that growth was also faster under Federal Reserve chairs who were first appointed by Democrats, which is either a large coincidence or a sign that the correlation is tracking something bigger than the White House.

Nothing about the claim predicts the next president. A correlation across sixteen terms is a statement about a set of administrations that have already finished. It has no forecasting power about the one now in office, and the paper itself says the gap could not have been forecast by standard methods in real time.

V. The recent decade does not cooperate

If the pattern were a stable law, it should still be visible in the years since the paper’s sample ended. It is not, or at least not cleanly.

Growth figures are the annual change in real gross domestic product, from the Bureau of Economic Analysis series (FRED, 2026).

Two honest readings sit side by side here. Biden’s 6.15 percent in 2021 is the largest figure in the decade, and it is also a reopening artifact: an economy restarting after a pandemic shutdown grows at a rate that reflects how far it fell, not how well it is run. Strip that year out and his remaining three, 2.52, 2.93, and 2.79 percent, are ordinary. Trump’s 2018, at 2.97 percent, is higher than three of Biden’s four years, and then the pandemic erased his fourth.

The pattern that held across seventy years does not hold tightly across the last ten. That is not a refutation of the older data. It is a warning against treating a historical regularity as a rule that a voter can apply to a candidate.

VI. The recalibration

The most direct challenge to the finding is recent and worth weighing, because it takes the “global luck” explanation seriously and tries to measure it.

A 2026 study in Frontiers in Political Science decomposes US growth against a weighted common trend drawn from twenty-seven comparable economies, deliberately excluding the United States so the trend cannot be driven by the outcome being measured. The authors find that global and regional cycles account for more than 94 percent of US growth, and that once that common trend is removed, the Democratic advantage falls to statistical insignificance, with a small-to-medium effect size (Su et al., 2026).

There is a symmetry worth noticing in the same paper. On the residual it isolates, Republican control of the House of Representatives is associated with higher idiosyncratic growth, and the authors’ most favorable configuration is a Democratic president paired with a Republican Congress. If that finding sounds as convenient as the one it replaces, that is the point: a method that reverses the partisan sign is a method whose partisan output should not be trusted in either direction.

The study has real limits, and they should be stated. It is one paper, in a journal with a lighter review reputation than the American Economic Review. It rests on roughly sixty annual observations. Its authors call their own political findings exploratory and concede the statistical power is low. It is not a refutation of Blinder and Watson. It is a recalibration, and the recalibration points the same way the original paper’s own conclusion does: toward luck rather than governance.

VII. What is actually true

Here is the honest state of the evidence, stated so that neither side can quote half of it.

The correlation is real and large, and the dismissive reply that “the sample is too small to mean anything” is wrong. Sixteen terms, tested three ways, produce a gap that survives every standard check.

The causal reading is not supported. The authors of the study that established the correlation attribute the gap to oil, productivity, foreign growth, and timing rather than to any policy a president chooses, and they found that both fiscal and monetary policy pointed, if anywhere, the other way. That is not a talking point in either party’s favor. It is the paper’s finding, and it is inconvenient for the party that most likes to cite it.

The president is one actor among several, and rarely the largest. The Federal Reserve, Congress, the world oil market, and a virus have more to do with the growth rate in any given year than a tax rate or a regulation, which is why the correlation is a fact about history rather than a tool for a voter.

Both failure modes are real and they mirror each other. One side reads a statistical association as a verdict on competence, and the other waves away a large, significant result by calling it noise. The second error is the more defensible of the two, because the evidence genuinely supports the luck explanation, but it is not a refutation of the evidence. The finding is what it is. What it is not is a promise about anyone now running for office.

One distinction is worth carrying out of all of this. If the growth rate is mostly a global tide, then partisan control of the White House explains less about the economy than the argument assumes. What does move a price is the subject of a companion essay: The Corruption Tax argues that graft is the one political variable for which investors reliably charge a premium, not because they object to it on moral grounds but because it raises the risk premium on every asset whose value depends on American rules holding. The tide is luck. Corruption is a fee, and a fee is charged whether or not anyone is watching.

The tide comes in on its own schedule. A president stands at the gate and claims credit for the water.

PRH | huffmanwrites.org | © Philip Huffman

Sources

• Blinder, A. S., & Watson, M. W. (2016). Presidents and the US Economy: An Econometric Exploration. American Economic Review, 106(4), 1015–1045. https://doi.org/10.1257/aer.20140913
• Federal Reserve Bank of St. Louis. (2026). Real Gross Domestic Product [GDPC1; data from the U.S. Bureau of Economic Analysis]. Retrieved September 28, 2026.
• National Bureau of Economic Research. (2023). US Business Cycle Expansions and Contractions. Business Cycle Dating Committee.
• Su, Q., Xu, X., Yu, Y., Fu, T., Chen, G., & Liu, W. (2026). The democratic growth advantage revisited: disentangling global trends from domestic political effects in the United States economy. Frontiers in Political Science, 8. https://doi.org/10.3389/fpos.2026.1856151