First quarter GDP components generated extreme movements in both directions. The net result came in below our expectation. Other economic data as usual have been mixed. Non-economic events as usual have been discouraging. None of this is significantly at odds with our expectations, and our forecast is little changed from two months ago.
NIPA Data
The advance release of 2026Q1 of national income and product account data put top line real output growth at 2.0%. (See chart.) This was well below our March outlook (the first using Q4 data) for growth at 3.3%.
One aspect in our miss was that the bounce-back in government spending from the Q4 shutdown was less than we expected. It added only 0.7% to GDP in Q1, after cutting 1.0% from Q4 output.
But this shortfall was over-shadowed by other aspects of the Q1 data. To begin with, there were large movements in all four investment categories. Spending on investment in goods (equipment and intellectual property) grew at a 14.2% rate, more than three times its Q4 rate. The growth reflected AI induced spending on information processing equipment (continuing a surge that began in early 2025) and on software. This strength was partly offset by contraction in construction spending (both business structures and residential). The negative growth here stretches back to early 2024. For Q1 it was estimated at -7.4%, well below our March forecast of -2%. But this weakness was swamped by the AI boom. Overall investment added 1.3% to GDP growth in Q1. This was more than consumption which added 1.1%.
But neither of these were the largest factor in the first quarter growth picture. We had anticipated modest growth in both exports and imports. Instead, both surged – with growth in exports at 12.9% and imports at 21.4%. Together these added $99 billion to the net export deficit and cut 0.9% from GDP growth.
Another perspective is a comparison of the past two quarters to the preceding three. GDP slowed significantly – from 2.5% in the first period to just 1.2% in the recent two quarters. Consumption slowed (in line with our expectations), but only enough to cut GDP growth by 0.3%. Investment was a major and stable contributor to growth in both periods (weak construction notwithstanding). The government sector has been flat, although with a little shutdown weakness in the recent period. Inventories shifted from modest shrinkage to modest growth. Together these four components, which comprise the domestic economy, account for GDP growth of 2.2% in the earlier period, falling just slightly to 2.0% in the past half-year. The rest of the drop in GDP came from the massive swing in trade, mainly from the import side of the ledger.
To summarize, there is a good deal of noise in the Q1 data. But the foundation of the economy – spending by business on investment and by households on consumption – is holding together. Indeed, the former has been consistently strong since the beginning of 2025. Government spending, on the other hand is weak. And trade has been an unpredictable wild card.
Recent Monthly Data
Monthly indicators have been totally indecisive. Some reports point up; some point down; some contain elements of both.
An example of the latter is the data from the Bureau of Economic Analysis on household income and spending. Consumption was up in both February and March, with growth matching that over the previous year. But real disposable income was down in both months. Its March level is basically unchanged from April of last year. And consumer sentiment fell in April, matching its lowest readings going back to 1960. We will return to this confusing situation later.
After a good month in February, industrial production decreased in March. The index has produced only slow growth during the past year. Its business equipment subcomponent, however, has been strong – consistent with the strong AI investment mentioned above.
Housing sector data for March showed a large increase in starts and a large decrease in permits. Grasping at straws: It might be that starts reflect decisions made somewhat earlier, while permits are more contemporaneous, and hence perhaps influenced by the war with Iran.
In the labor market, the job openings number was roughly unchanged in March. New claims for unemployment benefits through April continued at low levels. Continuing claims have declined since the end of last year to levels not seen since before the pandemic. These are both positive indicators.
The employment report for April was a slightly toned-down replay of the month before. The establishment survey had private sector job growth of 123k. This was down from the (revised) 190k in March but still exceeded expectations. Excepting a modest 9k gain in construction, nearly all the gain was in service sectors, especially in health care, transportation, and retail trade. The federal government – as it has in all but one month since February 2025 – again shed workers.
As in March (and February) the household survey was disappointing. The April labor force was down 92k, following -396k in March and a weak +18k in February. The participation rate fell by a tick in each of these months. Its 61.8% reading in April was the lowest since October 2021 when the economy was digging out of the pandemic shutdown. Unemployment rose in April to 4.34% from March’s 4.26%, but rounding put the reported rate at 4.3% in both months.
Overall, a month with some numbers to cheer optimists and others to worry pessimists. But nothing to ignite either euphoria or despondency.
Baseline Forecast
As shown in the chart, our revised forecast is similar to March. During the upcoming year (ending with 2027Q1) we expect output growth to average 2.1%, essentially the same as the past five quarters of data. This is accounted for by solid growth in consumer spending and in investment. In terms of growth rates, consumption is just above 2% in the data period and just below in the forecast period. Investment growth rises from 4.5% to 4.6%. In both periods the investment growth comes entirely from strength in equipment and intellectual property – averaging 7.1% during the next year. This offsets continued shrinkage in construction.
The other three components of GDP roughly offset one another. Government shifts from very slightly negative to a small positive. The former is a shutdown effect; the latter is mostly due to growth in defense spending. Trade shifts from a flat deficit to a rising deficit. The latter is a more normal situation given the imbalance between U.S. saving (including the government deficit as negative saving) and investment. Last, inventories move from decrease to increase – also more normal for a growing economy.
Discussion
A lot has happened over the past two months, and not much has changed. At least regarding evaluation of our forecast. In March we wrote “We think [our] outlook is optimistic. The relatively smooth path shown in Figure 3 is wildly unrealistic.” No change needed.
We laid out four sources of uncertainty: government disfunction, trade development from changes in the tariff regime, possible change at the Federal Reserve as Warsh takes over from Powell, and the war in Iran. All still relevant, and with little, if any, resolution of their surrounding uncertainty.
The year-ahead growth we expect rest on consumer spending and a continuation of strong AI related investment. Neither is certain.
Starting with consumption: As we mentioned above, it has been showing growth in the face of disposable real income that is flat even with income tax cuts from the Big Beautiful Bill. And despite historically low consumer confidence.
We think K-effects are part of the story here. Those in the lower fork of the K are probably more exposed to real income shrinkage due to cost-of-living effects from higher energy and food prices. Anecdotal data suggests their spending is under pressure.
The upper branch of the K is likely less affected by pressure on income. Spending by the more affluent rests on wealth as well as income – currently a support for continued spending. As this is written the S&P 500 has just hit 7400 for the first time. A recent study from the New York Fed supports this view. Dependence of consumption on a relatively small group of households also explains how aggregate consumption spending can be healthy even though the households’ average sentiment is terrible.
Turning to investment: The story here is, of course, the incredible boom driven by the AI revolution. The first question has to be: Should we have used “bubble” rather than “boom” in the previous sentence? At present we think the answer is: No. Six months ago, we were more on the fence; now we think “boom” is the correct description at least looking out well into next year. The main impetus is that we think the driving force behind AI is shifting from the supply-side to the demand-side. That is: From firms who are developing this “really neat cutting-edge” technology that (they say) will enable “really neat things” to firms that are using the technology to actually do new things (and also to do old things a lot more efficiently).
There are, of course, significant risks. For one thing there is the possibility (we think a low probability) that AI has some fatal flaw. Or that it will face regulations that will hamper its development. But at the top of our list are potential financial difficulties. Financial markets have a long history of moving to extremes. The stock market boom mentioned above rest to a disturbing extent on a handful of companies up to their eyeballs in AI. They might handle a market correction, but what about smaller firms farther down the AI food chain? Problems here could slow the AI rollout and with it the investment growth in our forecast. Financial market problems could also impact the upper branch households who are supporting aggregate consumer spending.
And, of course, the war continues with little sign that a resolution is near. Here we think the risk to the U.S. economy is indirect. Europe and the Far East are far more at risk from the closure of Hormuz. But if they experience significant problems, we will be affected.