During the mid-1930s, FDR pursued an aggressive set of policies including various actions intended to raise wages, as well as an undistributed profits tax. These actions were widely seen as anti-business, a view reinforced by FDR’s frequent attacks on the “economic royalists”.
In the second half of 1937, the US economy fell into a deep secondary depression, despite the fact that it had not yet recovered from the severe 1929-33 slump. The Roosevelt administration blamed the downturn on a lack of investment in the business sector, asserting that there was a “capital strike” motivated by hatred of New Deal policies. In fact, the slump was mostly caused by various New Deal policies, which pushed up wages at a time when monetary policy was reducing prices.
I was reminded of this event when I saw the following story:
After signing an executive order granting Canada and the US another temporary tariff reprieve, the US president blamed “globalist” nations and corporations for market-wide declines and shrugged off spooked markets.
The Treasury secretary also chimed in:
“There’s going to be a natural adjustment as we move away from public spending to private spending,” Bessent said Friday on CNBC. “The market and the economy have just become hooked and we’ve become addicted to this government spending, and there’s going to be a detox period.”
It is possible that this sort of lagged effect might be true for the overall economy, but it is certainly not true for the financial markets, which are forward looking. Policy initiatives that produce short-term pain and an even greater long-term benefit should be a positive for the stock market.
That’s not to say that markets won’t bounce back—as stock prices are almost impossible to predict. Think of a scenario where a policy initiative was put forward that other things equal might be expected to reduce stock prices by 10%. Also assume that the market thought there was a 50% chance that the initiative would be quickly reversed, with no damage done. In that case, you might expect stock prices to fall by roughly 5%. But that would merely represent the initial reaction, as more information came in stocks would either fall further, or (if the initiative was reversed) would regain lost ground.
On a related note, the Atlanta Fed has been forecasting a drop in real GDP during Q1. The media suggests that this forecast is based on a recent surge in imports, particularly gold imports. That may be true, but if so it is quite odd. A $20 billion surge in gold imports to beat expected tariff increases would not be expected to have any effect on actual GDP. The import category would move $20 billion in a negative direction while gold inventories would move $20 billion in a positive direction. If the media reports are correct (and I have no reason to doubt them), this suggests the government does not know how to measure GDP. It suggests that they are treating imports as a negative, but not applying an equal positive to investment (or consumption.) Why would they do this?
PS. Greg Mankiw is of course correct, and I am almost certain that Kevin Hassett knows that. I understand that we must all make compromises in our careers, but surely there are limits . . .
READER COMMENTS
Craig
Mar 10 2025 at 9:55pm
If I were to say something to the effect that car prices increased 10% and thus contributed to inflation, you might say something along the lines of ‘that’s a relative price change’ but mathematically how the CPI is constructed the car component to the extent its weighted in the index has a mathematical impact on the M/M model and I think that is what is happening here. The Atlanta Fed GDP Now report is here:
https://www.atlantafed.org/-/media/documents/cqer/researchcq/gdpnow/realgdptrackingslides.pdf
On page 2 one can immediately see the forecast drop off the cliff, right? Well, why? Ok, we need to read a bit further and now please refer to page 4 of the report which has the various columns, the data contributing to the GDP Now forecast. So as they get data that data moves their needle and one can easily see that all of a sudden the import delta spikes while at the same time exports drops a bit. So I would suggest that mathematically that is driving their forecast.
Scott Sumner
Mar 10 2025 at 11:23pm
Yes, but the surge in imports should be matched by a surge in inventories, unless the imports are simply being dumped in the trash.
Craig
Mar 11 2025 at 12:41am
Well, perhaps they’re trying to replace the gold in Fort Knox so they can’t credit inventories? 😉
My speculative hypothesis is that gold and oil are down a bit last 30 days so perhaps the book value of the inventory is lower?
I understand what you’re saying, but by the same token the Atlanta GDP Now model is definitely attributing a fall off in forecasted GDP between Feb 26th and Feb 28th and mathematically that is attributable to ‘net exports’ and primarily imports. The CIPI number does NOT follow so the result is a bit peculiar.
Craig
Mar 11 2025 at 12:58am
Reading the definition of what counts for purposes of CIPI “Change in private inventories (CIPI), or inventory investment, is a measure of the value of the change in the physical volume of the inventories—additions less withdrawals—that businesses maintain to support their production and distribution activities.” {emphasis supplied}
Perhaps gold held as a store of value simply doesn’t count for purposes of CIPI?
Scott Sumner
Mar 11 2025 at 2:07pm
Privately held gold would be consumption.
Craig
Mar 13 2025 at 9:27am
The definition seemed to be making a distinction between ‘consumed’ jewelry and gold in the form held as a store of value like a krugerand.
Thomas L Hutcheson
Mar 10 2025 at 10:44pm
“In fact, the slump was mostly caused by various New Deal policies, which pushed up wages at a time when monetary policy was reducing prices.”
Not to defend FDR, but shouldn’t a central bank take account of shocks and inflate enough to minimize there effect on real income?
It looks like the Euro crisis. Sure, too many Euros flowed into “Southern” economies, as investors did not realize that the elimination of exchange rate risk had increased the investment risk. But ECB should have acted (inflated enough) to prevent these real errors from compounding into causing unemployment of resources.
Scott Sumner
Mar 11 2025 at 2:09pm
FDR pressured the Fed to tighten policy. The Treasury also tightened with its gold sterilization.
Thomas L Hutcheson
Mar 11 2025 at 10:32pm
Perverse! But of course FDR had not had the benefit of reading Freidman & Swartz. 🙂
But I guess you agree that the Fed _should_ not have sterilize gold and _should_ have resisted FDR’s pressure.
Jon Murphy
Mar 11 2025 at 7:23am
I’ve every reason to doubt them. The media are terrible when it comes to understanding and reporting GDP and imports’ impact. I don’t think I’ve ever seen them get it right, constantly reporting imports as a negative to GDP. I see no reason they’d start now.
In other words, which is more likely: the economists of the Fed are miscalculating a basic accounting formula or news reporters are mistaken about what net exports mean?
steve
Mar 11 2025 at 11:37am
Let me second this and add that in general if the topic is science or medicine related you should never accept a media report at face value. Read the original study. (My own biased opinion is that if there is any math involved its too difficult for many reporters.)
Steve
Scott Sumner
Mar 11 2025 at 2:10pm
Good point, but in that case the economists at the Atlanta Fed seem to be making the same mistake.
John Hall
Mar 11 2025 at 8:24am
We’ve been following the gold imports issue since February when the advance trade goods numbers first came out. Goldman was out front as being on top of the issue.
Here’s the TLDR of what I’ve learned: the Census and Balance of Payments calculations of the trade deficit include gold imports. The Balance of Payments trade deficit numbers flow into the BEA’s ITAs (this is like current account, capital account, etc.). However, for the purposes of calculating the NIPA trade numbers (i.e. the ones that show up in GDP, etc.), the BEA removes the impact of gold as calculated from the ITAs. They then add back in their own estimate of gold imports. They argue that ITA gold trade is largely driven by investment flows, not those used for business purposes.
See pages 13 and 14 for more detail:
https://www.bea.gov/resources/methodologies/nipa-handbook/pdf/chapter-08.pdf
I’ve found it best to start from the idea that the government statisticians know what they’re doing.
Student
Mar 11 2025 at 10:47am
FWIW, Krugman agreed with the media take this morning:
”You may have seen people citing the Atlanta Fed’s “nowcast” that attempts to infer GDP far in advance of the official numbers, and which is currently showing a sharp decline in the first quarter. But that’s almost certainly a statistical red herring: It’s mostly about a surge of gold importsin anticipation of Trump’s tariffs, which is screwing up the usually helpful Atlanta model.”
https://paulkrugman.substack.com/p/the-economic-excuse-industry-is-booming
Now I have seen you end up being right about the zero lower bound in the past but I think it is an artifact of the nowcast methodology. I am not an expert on GDP or monetary theory, but I do understand forecasting methods (particularly BVARs and such) and i do think it’s an artifact of the approach.
nowcast seems to work by separately forecasting 13 or so major components (for example C, I, G, and net exports) using a mix of econometric techniques. It’s bottom up in that it uses BVARs and bridge equations to mimic the BEAs chain weighting methodology (as I understand it anyways) with a top down factor model approach. In other words… it takes the latest monthly and quarterly data, forecasts the missing values for the current quarter and aggregates into an overall quarterly forecast.
if you look close at the methodology paper, it seems to use available trade data to compute a contribution to GDP via a formula similar to a tornquist index. Usually this works well. However, if on component experiences a large temporary spike, this can cause an issue in the overall forecast.
I think you are right that from a purely accounting perspective, x billion in gold imports shpuld be offset by an equal increase in inventory.
However, the nowcast does not seem to automatically make that adjustment in real time. Once the adjustment for offsetting inventory changes is made… this effect will disappear.
Am in misunderstanding nowcast? Sure like to hear from the creator lol.
John Hall
Mar 11 2025 at 11:50am
You get a lot of stuff right here. The missing piece (that I describe above) is that the data GDPNow relies on for trade is calculated by the Census and the BEA adjusts that data for their measure of gold imports.
Craig
Mar 11 2025 at 12:07pm
Both very informative thanks
Brent Buckner
Mar 11 2025 at 11:13am
When calculating Q1 GDP we expect the US government to discount gold imports that the government expects to not be used in production. However, GDPnow is not enough of a finely-honed instrument that the Atlanta Fed builds in an expectation of this adjustment.
David Seltzer
Mar 11 2025 at 12:08pm
Scott: As you know, current market prices are present value calculations of future streams of expected cash-flows discounted by some stochastic discount factor. (SDF). When tariffs were implemented, market declines wiped out 4.2% in US market capitalization. I suspect the market’s real concern, going forward, is the uncertainty of DJT’s capricious policies affecting the US economy. One can hedge risks in markets, but Knightian uncertainty is virtually impossible to insure against. How does the Atlanta Fed incorporate that uncertainty in there GDP forecasts? Equally important, how do producers and consumers adjust their behavior in terms of individual risk aversion? Personal note: When DJT threatened tariffs during his campaign, like Warren Buffett, I adjusted our family portfolio’s to 90% cash and short term treasuries. Gold and oil make up the remaining 10%. Is stagflation or worse on the horizon?
David Seltzer
Mar 11 2025 at 12:10pm
Errata, should read their. Not there
Scott Sumner
Mar 11 2025 at 2:15pm
Everyone, Thanks for the helpful information. To be clear, I was not taking a firm position as to what had gone wrong with the forecasts. Rather what’s clear is that something is wrong, as gold imports do not affect actual GDP. That’s the only claim I was confident about.
My takeaway from all the comments is that the Atlanta Fed model is flawed.
John Hall
Mar 11 2025 at 4:10pm
I would put it as “the model is not incorporating important information that will likely cause a meaningful forecast error this quarter” but “flawed” isn’t that different from that, I guess.
At the end of the day, the GDPNow estimate is a systematic real-time estimate of current quarter US real GDP growth based on publicly available data. Private forecasters are able to use judgement and less traditional data sources, which will enable them to make better forecasts this quarter. But most of the time GDPNow does pretty good, and we have a good sense of why it is going wrong now. The most recent Census trade numbers are really terrible on the face of it, but it’s not so simple to correct for the issues in a systematic model framework.
Student
Mar 11 2025 at 6:36pm
Both you and Scott make good comments on this. I am not sure how they would fix this under their real-time approach. It is what it is, but I agree that it’s a weakness in the method.