Economy back to growing

The Bureau of Economic Analysis announced today that seasonally adjusted U.S. real GDP grew at a 3.0% annual rate in the second quarter. I have some concerns, but it looks better than many economists had been anticipating.

Quarterly real GDP growth at an annual rate, 1947:Q2-2025:Q2, with the historical average since 1947 (3.1%) in blue. Calculated as 400 times the difference in the natural log of real GDP from the previous quarter.

With the new numbers the Econbrowser recession indicator index is up to 11.7%. This primarily reflects the drop in GDP that we observed in the first quarter. The index offers an assessment of where the economy was as of 2025:Q1. Since we started reporting this measure in 2005, Econbrowser reports the index with a one-quarter lag to allow for data revisions and to aid the algorithm in pattern recognition. Though up slightly, the latest value of 11.7% is not alarming.

GDP-based recession indicator index. The plotted value for each date is based solely on the GDP numbers that were publicly available as of one quarter after the indicated date, with 2025:Q1 the last date shown on the graph. Shaded regions represent the NBER’s dates for recessions, which dates were not used in any way in constructing the index.

The story in the second-quarter GDP report was the flip of the first quarter. In Q1 there was a surge in imports (which other things equal means lower GDP) as businesses built up inventories of imported goods in anticipation of tariffs. In Q2, imports fell dramatically (which accounts for much of the strength in GDP in Q2) with firms drawing down those inventories. Both residential and nonresidential fixed investment were weak in the second quarter.

Inventories are also the wild card in watching the effects of the tariffs on inflation. Some businesses may still be setting prices based on the historical cost of goods, meaning some of the inflationary effects are yet to come. Still, let us count our blessings that the actual current average effective tariff rate is lower than originally threatened.

As of this morning, anyway.

19 thoughts on “Economy back to growing

  1. Paweł Skrzypczyński

    Thank you for mentioning my estimate of the average effective tariff rate Professor Hamilton!

    1. James_Hamilton Post author

      Thanks for providing the public service Pawel! I’ll be using it all the time.

  2. baffling

    so with import surge in Q1 drawing down gdp, and the drop in imports during Q2 pulling it back up, do you think a realistic estimate today would be to calculate GDP using the average of those two periods? and did the differences effectively cancel one another out, or was the drop or increase greater? from a personal perspective, economic activity did not increase in a way that doubles gdp growth compared to a quarter ago. it actually seems like we have less economic activity than earlier in the year. that said, we are concluding around $150k in home upgrades, so money is still being spent. although it does not look like the new car will be purchased any time soon. my 2017 acura will have to serve a little while longer…

    1. rjs

      baffling, re: do you think a realistic estimate today would be to calculate GDP using the average of those two periods?
      yes, that’s probably as close as we can get…. i’m on record that the GDP prints for both quarters would be inaccurate, because increased imports in the first quarter weren’t showing up in the inventory data until a month or two later…see here: https://econbrowser.com/archives/2025/04/economy-stumbles-into-2025
      if your goal is to estimate the quarterly change in the output of goods and services, you cannot subtract imports that haven’t been added to another GDP component in the same quarter… unfortunately, we don’t have data that captures that nuance…

  3. Macroduck

    Real final sales to domestic purchasers rose 1.1% in Q2 (SAAR):

    https://fred.stlouisfed.org/graph/?g=1KYRz

    That’s precariously close to stall speed. For H1, the pace of rise was 1.8% vs 3.0% in 2024, 2.7% in 2023 and 2.2% in 2022. The last time growth in core GDP was this slow was in Q3, 2022. As I recall, 2022 was when right-wing recession cheerleaders were claiming the U.S. was in recession. Wonder where they are today?

    Real GDP rose by 1.1% (SAAR) in H1 of this year, also right around stall speed and the slowest of any 2-quarter period so far in this expansion:

    https://fred.stlouisfed.org/graph/?g=1KYSd

    Normally, it would be too early in a presidential administration to attribute much of any fluctuation in economic activity to the nee President’s policies. In part, that’s because policy works gradually, in part, because the current budget is still from the previous President’s term.

    The felon-in-chief’s extraordinary efforts to change the way the economy runs, his mostly illegal DOGE cuts and staff reductions, the increased uncertainty caused by his actions and his undermining of the rule of law all suggest that the slowdown in growth so far this year is at least partly attributable to the felon’s actions. The further suggestion is that a continuation of his pattern of behavior will cause a further slowing of the economy.

    Keep in mind that the economy was already slowing during the felon’s first term, before Covid hit. That slowing was evident in exports and manufacturing output, suggesting that tariffs were a reason for the slowdow. Back then, the Fed cut rates. No wonder, then, that the felon is demanding another Fed bailout for his policies.

  4. Macroduck

    Sticking close to the topic at hand, why expect a recession as the result of the felon-in-chief’s policies when estimates of the dead-weight loss from trade restrictions tend to be small? The simple answer is that the transition from the previous set of business arrangements to a future set of arrangements is going to be a beast.

    For those of you who’ve had a bit of economics, think back to the “production possibility frontier”, the level of output that’s possible with a given level of resources, fully employed. Those who aren’t familiar, here’s Wikipedia’s article:

    https://en.m.wikipedia.org/wiki/Production%E2%80%93possibility_frontier

    When resources are shifted from one point on the frontier to some other – which is to say, when we replace imports with domestically produced goods – the first thing that happens is that resources are withdrawn from productive use. The economy runs below full capacity while in transition. Only once all the resources which are put out of use by the transition are put back into use – and are fully trained up in their new activity – does output return to the production possibly frontier.

    This happens all the time. Businesses open and close, people retire, production methods change and so on. Frictional unemployment is a commonly-mentioned example of this phenomenon. The economy typically expands fast enough to absorb this friction. When business arrangements change suddenly, as for instance during petroleum embargoes, the normal pace of growth isn’t enough to outpace the loss of activity due to the shift of resources to new uses.

    What the felon-in-chief is doing is inducing a sudden, large change in business arrangements. That means odds of recession are elevated.

    Why do estimates of dead-weight loss from trade barriers show relatively small losses if recession is the likely result of trade barriers being imposed? Think about the production possibly frontier. The estimate of dead-weight loss reflects a new position on the frontier; it’s a long-term estimate of the loss due to trade barriers. In the short term, resources are put out of use and we produce less than is possible. The upshot is a recession in the short run and lower output in the long run, as represented by dead-weight loss estimates.

    In sum, being slightly poorer in the long run comes with the additional cost of being lots poorer in the short run.

  5. Macroduck

    By the way, the Census Bureau calculates a monthly Index of Economic Activity which combines 15 economic data series. The latest reading, for June, was -1.06, the second lowest since January of 2024. The lowest reading of this expansion was in April, at -1.09. (Hmmm…what happened in April?) For all of Q2, the average reading was -0.71, the lowest in this expansion. The history on this index extend only back to 2005. In that time, only once has the quarterly average dipped this low outside a recession, in 2006.

  6. Macroduck

    Along with all the other interesting stuff in the GDP report, we got data on the housing sector. Ed Leamer famously declare that Housing IS the Business Cycle :

    https://www.nber.org/papers/w13428

    As to Leamer’s point, this cycle may be different. Fixed residential investment certainly looked recessionary in 2022, without NBER declaring recession:

    https://fred.stlouisfed.org/graph/?g=1KZ4n

    Now, if we look just at housing investment after correcting for inflation, housing is looking recessionary again. The curious thing is, fixed residential investment is well off its cycle high in real terms, but not in nominal ones:

    https://fred.stlouisfed.org/graph/?g=1KZ4M

    The “real” series isn’t long enough to show how odd this is, but it’s quite odd. Sad that the housing boom is not covered in the “real” series.

    Anyhow, one reason housing leads the cycle is that it responds so quickly to changes in interest rates. Recently, interest rates have been fairly steady, after rising quite smartly earlier in the cycle. So maybe this time will be different, because housing isn’t reflecting a RISE in rates. Interest rates are holding back housing demand, but not enough to prevent that big rise in prices – in nominal housing investment, relative to real investment. We have a supply-side problem as well as a demand-side problem; sales are slow and prices are high.

    So while there are some pretty troubling hints at recession out there, housing may not be one of them.

    1. David S

      I’m a cult follower of Kevin Erdmann and his position is consistent with your conclusion that housing is okay at the moment. He’s also stake a claim that “high” borrowing rates actually aren’t a major impediment to new housing construction—with some qualifications made for multi-family starts. Housing production is constrained by zoning regulations and mortgage access. I haven’t read Paul Krugman’s recent essay on the topic, BTW.
      (I say “high” because there was robust construction in prior decades despite 30 year notes at 7% and higher)

      Housing is sensitive to some tariffed goods, especially Canadian lumber, although the price pressure is actually minimal in the context of all the inputs to your typical house. Construction labor is very sensitive to ICE activity, but I don’t have any source on how that is impacting pricing and schedules. Who will build all of Stephen Miller’s concentration camps? Presumably, Medicaid recipients who have been slouching around for decades collecting benefits but are now incentivized for some rigorous Greatening of America.

    2. New Deal democrat

      I’ll let you look up the relevant series on FRED, but a big reason the 2022-23 downturn did not produce a recession can be found if you track the following 3 series:
      1. Residential units under construction
      2. All employees residential construction
      3. Residential construction spending deflated by the cost of construction materials

      What you will find out is that the number of units continued to grow, and then stabilize, for a far longer period of time than after previous peaks. Secondly, you will find that there were far fewer workers per unit under construction than in previous expansions, partly explaining #1. Finally, you will find that”real” construction spending did not go down as much as previously, and rebounded quickly.

      In the past 12 months, #1 has cliff-dived, and #3 is under renewed stress as well. Only #2 has not turned down yet – as to which, stay tuned tomorrow.

    1. Macroduck

      Hakimian has “literally” lied, since his claim is that BLS has “literally stopped” collecting CPI data. The graphics he presents show that data collection has been reduced, not stopped; more of an hyperbolic lie, an effort to mischaracterize the data, than an outright lie.

      A quick trip to the BLS websight confirms that Hakamuan’s grapic representation of reduced data collection is roughly correct:

      https://www.bls.gov/cpi/tables/imputation.htm

      Cut staff and you cut the amount of work that can be done, so at some level, this is intentional.

    2. Macroduck

      By the way, the two Fed governors who dissented against leaving rates steady at yesterday’s FOMC meeting were both nominated to the Board by the felon-in-chief. Both favored a rate cut, same as the felon. Two dissents at the same meeting is rare. Whatever the thinking of these two Governors, this gives the impression of politicization of monetary policy.

      1. Anonymous

        Felon in chief aka Epstein’s “nothing to see here” in chief and his cult economists remain miffed that the Fed went off ZIRP in his first occupation of the Oval Office.

        Put a clown in the palace and it becomes a circus.

      2. baffling

        these two governors have a history of tight money policy. very strange they now want to be accommodative, especially if the economy is as strong as trump claims it is. you drop rates upon seeing weakness.

    3. NewDealdemocrat1932@gmail.com

      This claim has been debunked. Here’s an entire thread on the subject:
      https://bsky.app/profile/peark.es/post/3lv7txhigsk2y

      And here is NYT economics correspondent Ben Casselman:
      https://bsky.app/profile/bencasselman.bsky.social/post/3lv7f4xrpo22k
      Money quote:
      “ I am making the stronger claim. I have been in touch with a substantial number of longtime staffers at all levels of these agencies, people I know and trust. Many have concerns about where things might go, but all say that so far there has been no interference.”

      Hope that is helpful.

  7. joseph

    Macroduck: “Two dissents at the same meeting is rare. Whatever the thinking of these two Governors, this gives the impression of politicization of monetary policy.”

    Ya think? The last time there were two formal dissents was over 30 years ago. Also notable that Bowman’s dissent was that rates are two high. Her last dissent was in September saying that rates were too low. I wonder what changed?

    Trump was complaining that lowering rates before the election was interference, and Bowman dutifully echoed that. Now Trump is complaining that rates are two high and Bowman says ditto.

    The field for Fed chair auditions is getting crowded. Bessent, Hassett, Warsh, and now Bowman are all making their loyalist bids to the Dear Leader.

  8. rjs

    i’m late getting to this, but this morning’s construction spending report showed construction spending for all three months of the second quarter was higher than what was reported by the BEA in the advance report for 2nd quarter GDP earlier in the week, figures which can be viewed here: https://www.bea.gov/sites/default/files/2025-07/gdpkeysource-2q25-adv.xlsx .…the annual rate of construction spending for April was revised more than $7.9 billion higher, from $2,145.5 billion to $2,153.44 billion, and annualized construction spending for May was revised $5.7 billion higher, from $2,138.2 billion to $2,143.9 billion….

    in reporting 2nd quarter GDP, the Excel file with key source data and assumptions linked above indicated on line 86 that they had estimated that the annualized value of June’s nonresidential construction would be $2.6 billion less than that of the previously reported May figure, that June’s annualized residential construction on line 109 would be $7.7 billion less than that of the previously reported May figure, and that the value of June’s public construction shown on line 201 would be $0.1 billion more than the previously published May figure…hence, the total of the annualized figures used by the BEA for total June construction in the 2nd quarter GDP report were $10.2 billion less than the previously published May figure…with June construction now reported to be down $7.7 billion from a May figure that was revised $5.7 billion higher, that means that the BEA had underestimated annualized June construction spending by $8.2 billion when reporting 2nd quarter GDP…thus, after averaging the implied revisions to construction spending for the three months of the 2nd quarter, we find the total revised annualized figure for 2nd quarter construction spending would thus be $7.3 billion more in current dollars than the current dollars figures used by the BEA when computing 2nd quarter GDP, implying we’ll see an upward revision of about 0.09 percentage points to the construction components of 2nd quarter GDP when the 2nd estimate is released on the 28th of August, give or take depending on the mix of inflation adjustments to the revised figures…

Comments are closed.