AtlanticPulse

Research

The Hike-Proof Boom: Americans Are Starting Businesses at a Record Pace, and the Fed Just Raised Rates Into It

· Daily edition

Editorial artwork: The Hike-Proof Boom: Americans Are Starting Businesses at a Record Pace, and the Fed Just Raised Rates Into It

Why the Fed pressed the button

On Wednesday the Federal Reserve raised its benchmark rate by a quarter point, to 3.75%–4.00% — the first increase in three years — and Chair Kevin Warsh explained why in one sentence: "The plain fact is that inflation is too high and has been for too long." The Fed also signaled that a majority of its policymakers expect at least one more hike this year, citing an economy that keeps surprising on the strong side. Retail sales jumped 1.2% in August, the largest monthly gain since March, even as gas prices climbed more than 7% in a month on war-driven energy costs (The Business Journal).

Central banks tighten when demand outruns supply, so the interesting question is where all this demand actually lives. One answer hides in a dataset almost nobody reads for fun: the Census Bureau's Business Formation Statistics, the official tally of applications for new Employer Identification Numbers — the paperwork that precedes essentially every new American business, from food truck to fusion startup. Applications are cheap to file, so they are noisy. But when they move this much, they are telling you something about how many Americans are trying to build things rather than just buy them (Census Bureau BFS).

The record: 129,000 founding papers a week

Here is the headline the hike will have to beat: in the first thirty-one weeks of 2026, Americans filed an average of 129,129 business applications per week. In the same calendar weeks of 2019 — the pre-pandemic baseline — the figure was 72,369. That is a 78.4% increase, measured on identical weeks so seasonal quirks (Americans famously file in January and slack off in July) cancel out entirely (Census Bureau BFS).

This is not a one-month blip. The weekly record shows the now-familiar shape: a modest plateau around 70,000–85,000 a week before 2020, a violent spike that summer — the week ending July 18, 2020, hit 124,000 — and then a step-change that never reversed. The averages for the same early-year window ran 115,241 in 2021, 104,980 in 2022, 112,193 in 2023, 107,316 in 2024 and 112,658 in 2025. Each of those was extraordinary by any pre-pandemic standard. 2026 beat them all.

The comparison gets almost comic at the extremes. All 30 comparable weeks of 2026 came in above their 2019 counterparts — 30 for 30 — and every single one cleared 100,000 filings. The weakest week of 2026 so far (112,090) would have been an above-average week in 2021. The average week of 2026 (129,129) outruns the single busiest week of the entire 2020 pandemic surge (124,000). And five of the eight busiest weekly figures in this six-and-a-half-year record occurred in 2026, topped by the all-time high of 155,020 in mid-January. Whatever the Fed is trying to cool, it is not a flickering flame.

The demand the Fed keeps bumping into: weekly business applications, 2019–2026

Line chart of weekly U.S. business applications from January 2019 through August 2026, showing a pandemic spike in mid-2020, an elevated plateau through 2021–2025, and 2026 running highest.
Weekly U.S. business applications (not seasonally adjusted), week ending Jan 12, 2019 through week ending Aug 1, 2026. The 2026 segment is the highest sustained stretch in the series; weekly figures are preliminary and subject to revision. Sources: U.S. Census Bureau — Business Formation Statistics (weekly and historical quarterly files).
Download exact data
View exact chart values
weekendapps
2019-01-1281070
2019-01-1975100
2019-01-2670650
2019-02-0280590
2019-02-0980790
2019-02-1683210
2019-02-2373790
2019-03-0282540
2019-03-0977530
2019-03-1680860
2019-03-2385430
2019-03-3084950
2019-04-0683760
2019-04-1375470
2019-04-2071270
2019-04-2770080
2019-05-0475240
2019-05-1171550
2019-05-1871890
2019-05-2570030
2019-06-0155910
2019-06-0869720
2019-06-1566610
2019-06-2267180
2019-06-2967520
2019-07-0649980
2019-07-1363050
2019-07-2064530
2019-07-2763490
2019-08-0364290
2019-08-1065350
2019-08-1766360
2019-08-2467780
2019-08-3163090
2019-09-0752570
2019-09-1464250
2019-09-2167470
2019-09-2866350
2019-10-0565400
2019-10-1263360
2019-10-1957100
2019-10-2666430
2019-11-0264170
2019-11-0962150
2019-11-1657090
2019-11-2363440
2019-11-3046830
2019-12-0755110
2019-12-1462970
2019-12-2164910
2019-12-2842290
2020-01-0431010
2020-01-1191540
2020-01-1879650
2020-01-2575690
2020-02-0185520
2020-02-0883810
2020-02-1582130
2020-02-2275760
2020-02-2982570
2020-03-0784750
2020-03-1479290
2020-03-2165420
2020-03-2854070
2020-04-0455650
2020-04-1159360
2020-04-1856960
2020-04-2562920
2020-05-0267900
2020-05-0969990
2020-05-1672790
2020-05-2375840
2020-05-3067800
2020-06-0674790
2020-06-1383240
2020-06-2088690
2020-06-27103700
2020-07-04114050
2020-07-11113100
2020-07-18124000
2020-07-25117340
2020-08-01118080
2020-08-08114650
2020-08-15112460
2020-08-22104410
2020-08-29102150
2020-09-05101990
2020-09-1279890
2020-09-1997820
2020-09-2695590
2020-10-0391580
2020-10-1087870
2020-10-1782230
2020-10-2491540
2020-10-3188780
2020-11-0781280
2020-11-1479100
2020-11-2184280
2020-11-2864240
2020-12-0578910
2020-12-1286940
2020-12-1985470
2021-01-0259920
2021-01-09111200
2021-01-16143290
2021-01-23121270
2021-01-30126670
2021-02-06123400
2021-02-13119610
2021-02-20102740
2021-02-27116070
2021-03-06119940
2021-03-13121760
2021-03-20123900
2021-03-27125920
2021-04-03122140
2021-04-10117300
2021-04-17130300
2021-04-24130330
2021-05-01134880
2021-05-08127450
2021-05-15115800
2021-05-22123210
2021-05-29118080
2021-06-0596160
2021-06-12109320
2021-06-1988620
2021-06-26123150
2021-07-03101070
2021-07-1084270
2021-07-17101860
2021-07-2499020
2021-07-3196660
2021-08-0797090
2021-08-1495550
2021-08-2195240
2021-08-2895810
2021-09-0495540
2021-09-1180390
2021-09-1897050
2021-09-2596480
2021-10-0293420
2021-10-0992790
2021-10-1687310
2021-10-2373640
2021-10-30113160
2021-11-0688690
2021-11-1374950
2021-11-20104000
2021-11-2769630
2021-12-0482210
2021-12-1191210
2021-12-1891180
2021-12-2569750
2022-01-0142420
2022-01-08113260
2022-01-15112980
2022-01-22106420
2022-01-29117480
2022-02-05112120
2022-02-12114030
2022-02-19111640
2022-02-26112620
2022-03-05118700
2022-03-12115680
2022-03-19113980
2022-03-26110120
2022-04-02105500
2022-04-09107250
2022-04-16108600
2022-04-23120420
2022-04-30110370
2022-05-07105420
2022-05-14101140
2022-05-21103460
2022-05-28101180
2022-06-0484540
2022-06-11104370
2022-06-1899790
2022-06-2588520
2022-07-0296700
2022-07-0976660
2022-07-1696310
2022-07-2398050
2022-07-3095190
2022-08-0691890
2022-08-1393570
2022-08-2093440
2022-08-2795610
2022-09-0396330
2022-09-1080800
2022-09-1796800
2022-09-2495650
2022-10-0193570
2022-10-0894620
2022-10-1589560
2022-10-2292690
2022-10-2992760
2022-11-0588230
2022-11-1290390
2022-11-1984230
2022-11-2665820
2022-12-0379430
2022-12-1089660
2022-12-1788720
2022-12-2484560
2022-12-3133560
2023-01-0786900
2023-01-14129260
2023-01-21103990
2023-01-28122230
2023-02-04123120
2023-02-11118530
2023-02-18113860
2023-02-25110790
2023-03-04124440
2023-03-11123370
2023-03-18128420
2023-03-25123730
2023-04-01119510
2023-04-08112700
2023-04-15112000
2023-04-22116310
2023-04-29117960
2023-05-06115400
2023-05-13108120
2023-05-20111090
2023-05-27110220
2023-06-0389960
2023-06-10110260
2023-06-17110980
2023-06-24105890
2023-07-01109610
2023-07-0884740
2023-07-15111690
2023-07-22110950
2023-07-29106660
2023-08-05105280
2023-08-12105480
2023-08-19104590
2023-08-26103870
2023-09-02105890
2023-09-0991510
2023-09-16106950
2023-09-23106260
2023-09-30105200
2023-10-07104400
2023-10-1496170
2023-10-21105980
2023-10-2899930
2023-11-0495510
2023-11-1195600
2023-11-1895280
2023-11-2575850
2023-12-0287210
2023-12-0998090
2023-12-1692740
2023-12-2392040
2023-12-3034820
2024-01-0696990
2024-01-13125350
2024-01-20106050
2024-01-27120000
2024-02-03131500
2024-02-10122570
2024-02-17108850
2024-02-24111400
2024-03-02117380
2024-03-09115630
2024-03-16116210
2024-03-23120700
2024-03-30119830
2024-04-06109860
2024-04-13110300
2024-04-20109960
2024-04-27115110
2024-05-04109390
2024-05-11109020
2024-05-18102010
2024-05-25100600
2024-06-0185780
2024-06-08104280
2024-06-15103150
2024-06-2296150
2024-06-29100960
2024-07-0678990
2024-07-1396550
2024-07-2096270
2024-07-2794590
2024-08-0393890
2024-08-1094470
2024-08-1797920
2024-08-2499330
2024-08-31100690
2024-09-0783870
2024-09-1497670
2024-09-2196910
2024-09-2895020
2024-10-0590840
2024-10-1291670
2024-10-1993000
2024-10-2691900
2024-11-0295220
2024-11-0993840
2024-11-1691610
2024-11-2393570
2024-11-3074720
2024-12-0781060
2024-12-1493210
2024-12-2191030
2024-12-2859230
2025-01-0411060
2025-01-11139820
2025-01-18137180
2025-01-2599930
2025-02-01112590
2025-02-08113810
2025-02-15111750
2025-02-22103700
2025-03-01114360
2025-03-08118140
2025-03-15131020
2025-03-22131530
2025-03-29121170
2025-04-05119670
2025-04-12117920
2025-04-19115660
2025-04-26120580
2025-05-03113150
2025-05-10116760
2025-05-17108200
2025-05-24109580
2025-05-3192330
2025-06-07106520
2025-06-14109800
2025-06-2191250
2025-06-28122210
2025-07-05109430
2025-07-12100560
2025-07-19105650
2025-07-26105980
2025-08-0293220
2025-08-0998930
2025-08-16124280
2025-08-2385930
2025-08-30121120
2025-09-06109960
2025-09-13105330
2025-09-20120070
2025-09-27111790
2025-10-04112660
2025-10-11104810
2025-10-1899740
2025-10-25108320
2025-11-01115160
2025-11-08127510
2025-11-15123990
2025-11-2295250
2025-11-2980900
2025-12-06109910
2025-12-13101540
2025-12-20100130
2025-12-2790330
2026-01-0331680
2026-01-10122810
2026-01-17155020
2026-01-24128620
2026-01-31141830
2026-02-07129610
2026-02-14128630
2026-02-21118030
2026-02-28129840
2026-03-07133020
2026-03-14137910
2026-03-21134060
2026-03-28130790
2026-04-04125270
2026-04-11130810
2026-04-18129340
2026-04-25132030
2026-05-02132870
2026-05-09131810
2026-05-16130580
2026-05-23125220
2026-05-30112090
2026-06-06131630
2026-06-13123810
2026-06-20121310
2026-06-27121000
2026-07-04134970
2026-07-11116370
2026-07-18128620
2026-07-25138230
2026-08-01117750

Same weeks, different country: Jan–Aug application averages, 2019–2026

Bar chart of the average number of weekly U.S. business applications in the first week of January through early August of each year, 2019 to 2026, rising from about 72,000 in 2019 to about 129,000 in 2026.
Average weekly U.S. business applications in identical calendar windows (weeks ending early January through Aug 10), so seasonality cancels. 2026 is the highest on record: 129,129 per week, +78.4% versus the same weeks of 2019. Sources: U.S. Census Bureau — Business Formation Statistics (weekly and historical quarterly files).
Download exact data
View exact chart values
yearavgweeklynweeks
20197236931
20208326031
202111524131
202210498031
202311219331
202410731631
202511265831
2026 YTD12912930

Thirty for thirty: 2026 weeks against the same weeks of 2019

Two-line chart comparing U.S. weekly business applications in 2026 against the same numbered weeks of 2019, with the 2026 line consistently far above the 2019 line across all 30 matched weeks.
Week-by-week matched comparison, weeks ending early January through Aug 1: every one of the 30 comparable weeks of 2026 (upper line) exceeded the same week of 2019 (lower line). Data not seasonally adjusted; identical calendar windows. Sources: U.S. Census Bureau — Business Formation Statistics (weekly and historical quarterly files).
Download exact data
View exact chart values
weeknumy2019y2026
181070122810
275100155020
370650128620
480590141830
580790129610
683210128630
773790118030
882540129840
977530133020
1080860137910
1185430134060
1284950130790
1383760125270
1475470130810
1571270129340
1670080132030
1775240132870
1871550131810
1971890130580
2070030125220
2155910112090
2269720131630
2366610123810
2467180121310
2567520121000
2649980134970
2763050116370
2864530128620
2963490138230
3064290117750

Every region, same direction

Nor is this a one-region story driven by population shuffling. Across nine large states spanning the map, average weekly applications in 2026 ran far above the same weeks of 2019: Arizona +105.9%, Texas +91.2%, North Carolina +89.1%, Florida +87.9%, California +72.3%, Washington +71.4%, Georgia +67.3%, Illinois +63.3% and New York +44.3%. The Sun Belt is pulling hardest — Arizona has effectively doubled its founding pace — but the slowest grower on the list is still up nearly half from 2019 levels (Census Bureau BFS).

It is worth being careful about what these figures are not. They are not seasonally adjusted counts of open businesses, and the nine states here are a large-sample comparison rather than a full 50-state league table. But the direction is unanimous, which is the part that matters for a policymaker looking for weak demand: the part of the economy where people commit their own names, savings and evenings to starting something is the strongest it has been in the life of this dataset.

Nine states, one direction: application growth vs 2019

Horizontal bar chart ranking nine large U.S. states by the percent change in average weekly business applications between the same calendar weeks of 2019 and 2026, led by Arizona at about 106 percent and trailed by New York at about 44 percent.
Percent change in average weekly business applications, 2026 year-to-date versus identical calendar weeks of 2019, nine large states (weekly data, not seasonally adjusted). Arizona has roughly doubled; even last-place New York is up 44%. Sources: U.S. Census Bureau — Business Formation Statistics (weekly and historical quarterly files).
Download exact data
View exact chart values
stateavg2019avg2026pctvs2019
Arizona15703232105.9
Texas62741199391.2
North Carolina2153407189.1
Florida81411529487.9
California75391299072.3
Washington1426244571.4
Georgia3566596467.3
Illinois2414394263.3
New York4683675644.3

The fine print: applications are promises

Now the skeptic's turn, and it is a good one: a business application is a form, not a firm. Some applications are side hustles that never hire; a portion are filed by people protecting a name before they know what it does. The Census Bureau has long tracked how many applications become actual businesses within four quarters — and the history is sobering. In 2005, about 20.5% of applications turned into a real business within a year. By 2019 the conversion rate had fallen to roughly 9%. In other words, even before the pandemic, America was filing a ballooning number of intentions and delivering a much thinner slice of companies (Census Bureau historic file).

There is a catch for our moment, though: the published conversion series in the current historical file ends with 2019-Q4, because formations necessarily lag applications by at least a year and the quarterly series here has not caught up. So nobody — including us — can yet say whether the 2021–2026 application boom is producing a proportionate wave of real companies. The most recent independent read on the mix is suggestive rather than definitive: a report by the AI firm MyUser, circulated by ANI, estimated that 36.3% of new U.S. companies in the first half of 2025 had a single founder, up from 23.7% in 2019, which it attributed to AI tools absorbing work that used to require staff (Indias News/ANI). If that is right, today's application flood may be a flood of one-person enterprises — exactly the kind of activity a rate hike is least able to stop and least likely to cool.

The catch: fewer applications became businesses before the pandemic

Line chart from 2005 to 2019 showing the share of U.S. business applications that became actual businesses within four quarters, falling from about 20.5 percent in late 2005 to about 9 percent by late 2019.
Share of U.S. business applications that turned into an actual business within four quarters (formations within four quarters divided by applications four quarters earlier, both seasonally adjusted, spliced historical series). The published series in this file ends with 2019-Q4; recent-year conversion is not yet available here. Sources: U.S. Census Bureau — Business Formation Statistics (weekly and historical quarterly files).
Download exact data
View exact chart values
quarterappslagform4qrate
2005-Q357477711787820.5
2005-Q458320311384719.5
2006-Q161066411175618.3
2006-Q261809410796917.5
2006-Q363038210350616.4
2006-Q464287110178015.8
2007-Q167080610000614.9
2007-Q26696619839814.7
2007-Q36522519564614.7
2007-Q46514909292614.3
2008-Q16508118543513.1
2008-Q26617108300512.5
2008-Q36692537837011.7
2008-Q46780397575311.2
2009-Q16395397269411.4
2009-Q26498927182511.1
2009-Q36361027262411.4
2009-Q46305167135311.3
2010-Q16005597266512.1
2010-Q25969917259412.2
2010-Q36074737369412.1
2010-Q45961117201112.1
2011-Q16081517326712
2011-Q26149287148011.6
2011-Q36194147289911.8
2011-Q46213427144911.5
2012-Q16527846836910.5
2012-Q26176697068911.4
2012-Q36312856854010.9
2012-Q46353767329211.5
2013-Q16186716907411.2
2013-Q26439447057111
2013-Q36192857082011.4
2013-Q46603387196410.9
2014-Q16398196969010.9
2014-Q26365837212511.3
2014-Q36460907261411.2
2014-Q46601017492911.4
2015-Q16507337552411.6
2015-Q26760907274310.8
2015-Q36697317256410.8
2015-Q46926327082610.2
2016-Q17049127513810.7
2016-Q26876967224010.5
2016-Q37031587605410.8
2016-Q46910047434010.8
2017-Q17268477572110.4
2017-Q27129707632610.7
2017-Q37560517593410
2017-Q47500177690510.3
2018-Q17754318183310.6
2018-Q27907258198410.4
2018-Q37907398084610.2
2018-Q4819236799459.8
2019-Q1859520784089.1
2019-Q2874271794499.1
2019-Q3872926775238.9
2019-Q4862674777939

What it means for the hike — and what would prove this wrong

For the Fed, this picture cuts both ways. Record applications help explain the resilience officials keep citing: new firms eventually rent storefronts, buy equipment and hire, which supports exactly the demand-side strength that keeps inflation sticky. It also suggests the mechanism rate hikes rely on — making future plans more expensive to finance — is running into a wave of ventures whose upfront costs are low and whose financing needs are modest. That is a description, not a forecast: seven years of low-rate and high-rate regimes alike have now coincided with application totals far above the old normal, so the record alone cannot tell you what rates are doing to it.

For readers, the payoff is a dashboard with two dials. The first is the weekly application pace itself: if 2027 begins and filings are still clearing 120,000 a week, the demand story the Fed is fighting is intact. The second is the conversion data, whenever the Census Bureau publishes the pandemic-era cohorts — the single number that will reveal whether this was a golden age of American company creation or the largest pile of unfinished paperwork in statistical history. Given everything else in this week's news, betting on the paperwork seems safe either way.

Sources and methods

Analysis uses the pinned Census Bureau Business Formation Statistics snapshot (built 2026-08-23); weekly observations run through the week ending 2026-08-01 and are preliminary and subject to revision.

Weekly series are not seasonally adjusted. All 2019-vs-2026 comparisons use identical calendar windows (weeks ending in the Jan 7–Aug 10 span), so seasonal patterns affect both years equally; a separate February-onward check (+77.6%) confirms the January spike does not drive the result.

The seasonally adjusted national quarterly file in this snapshot ends with 2020-Q4 and the business-formations (conversion) series ends with 2019-Q4; claims about 2021–2026 therefore rely on the weekly record, and no post-2019 conversion rate is asserted.

Nine large states were compared (those extracts chosen for this analysis), not all 50; state figures use unadjusted weekly counts.

The analysis is descriptive. No causal effect is claimed between the Fed's rate decision, the Iran war, and application growth.

The digest reports same-day events; this research uses the most recent official data snapshot, which necessarily lags the news cycle by about six weeks.

Research completed 2026-09-17, for the September 16, 2026 news edition. The available source data may cover earlier periods; see the observation periods and source vintages above.

Read the related National / Federal News edition →