Digging for gold

Stephen Stanley | March 7, 2025

This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors. This material does not constitute research.

Merchandise imports have surged higher in December and January.  The broad narrative is that retailers and other firms have been pulling their orders forward to get finished goods and inputs into the country before tariffs could take effect.  While this dynamic is likely behind the widening in the trade gap, there is a more specific phenomenon taking place that complicates the GDP implications of the move.  A significant portion of the import jump reflects shipments of gold into the US, and the Bureau of Economic Analysis excludes gold from its export and import tallies when calculating GDP.  As a result, the record widening in the trade gap seen in January will not have as much of an impact on first quarter real GDP as it might appear at first glance.

Measuring net exports in GDP

Just as a basic review, the BEA’s tally of GDP is actually added up by measuring spending, not production.  The key components of final demand include consumer spending, housing activity, business investment, and government outlays. To get from domestic spending to domestic production, government statisticians have to net out trade flows.

The net exports component of GDP reflects exports minus imports.  So, exports add to GDP and imports subtract. To the extent that the BEA is attempting to measure domestic output, statisticians have to, for example, take imported goods and services out of consumer spending. Even though robust imports may be a sign of strength in the US economy, from a purely arithmetical standpoint, higher imports mean lower GDP.

Gold

There are certain adjustments made to the raw data to get to the BEA’s treatment of GDP.  According to the BEA’s handbook of methods, the vast majority of international transactions for gold relate to investment rather than business or industrial use. As a result, these flows are considered financial investment rather than output.  BEA goes through a complicated set of calculations that involve estimating the value of domestic gold production using data from the US Geological Survey and then estimating industrial use of gold—for jewelry, specialty coins, or dental applications, for instance—from data gleaned from the World Gold Council. The rest of international gold flows, as reflected in the monthly trade figures, are stripped out.

Ordinarily, this ex-gold calculation represents a small and relatively stable adjustment to the headline monthly trade figures.  However, in recent months, there have been unusual flows in the gold market.  A Reuters article published last month reviewing the January Swiss customs data noted that gold exports from Switzerland to the US tripled from December to January to a record high.  The article suggested that gold investors were concerned that tariff hikes might apply to gold deliveries into the U.S. and thus wanted to move their holdings into the US ahead of any new levies.

Quantifying the impact of gold flows on GDP

It is difficult to precisely forecast the impact that gold flows will have on the GDP calculations because the BEA is using some data on gold production and consumption that are not widely available.  In addition, the Census Bureau monthly trade data introduce another layer of complication.  There is a category of merchandise exports and imports labeled “nonmonetary gold.”  I always assumed that this would include all or at least the vast majority of the gold flows.  We have learned over the last two months that this is not the case.

To be sure, imports of nonmonetary gold have risen in recent months.  Imports of “nonmonetary gold” averaged about $1.3 billion per month in 2023 and through the first 11 months of 2024.  However, they jumped to $2.2 billion in November, $3.2 billion in December, and $3.8 billion in January.

However, this proved to be the tip of the iceberg.  A separate line item labeled “finished metal shapes” has seen the bulk of the action in December and January.  Imports of that line item averaged $3.6 billion in 2023 and $3.5 billion in the first 11 months of 2024.  However, they spiked to $13.75 billion in December and then ballooned to $34.2 billion in January, accounting for $29.7 billion of the $47.6 billion cumulative surge in imports in December and January while “nonmonetary gold” accounted for $1.6 billion.

Of course, there is more than just gold bars in the “finished metal shapes” category, as it includes all sorts of industrial steel and aluminum products such as pipes.  This makes it difficult to gauge exactly how to translate the monthly trade readings into a GDP-consistent net exports estimate.

Luckily, we have some insight from the BEA source data table for the fourth quarter.  Since this spike in “finished metal shapes” began in December, we can see how the BEA figures evolved through the end of last year.  The detailed source data published with the last set of fourth quarter GDP numbers (on February 27) offer some insight into the gold adjustment.  In this set of numbers, gold imports went from $3.2 billion in October, a relatively normal level compared to prior months, to $4.6 billion in November and to $13.2 billion in December.  For reference, the sum of “nonmonetary gold” and “finished metal shapes” in December in the monthly trade data added up to about $16 billion.

The monthly Balance of Payments basis merchandise exports and imports figures showed a widening of the nominal trade gap from an annualized $1,235.6 billion in the third quarter to $1,302.0 billion in the fourth.  However, excluding exports and imports of gold, the widening was much narrower, from $1,241.7 billion to $1,259.7 billion.  That is, what in the monthly data appeared to be a $66 billion deterioration in the nominal trade gap in the fourth quarter was narrowed to only $18 billion once gold flows were excluded.

The January figures published today showed imports on a Balance of Payments basis surging by $46.2 billion.  About $21 billion of that increase was accounted for by our two line items.  The rise in imports in January was certainly more broad-based than in December and likely reflected firms in a variety of industries pulling their orders forward to avoid tariffs.

Let us assume for simplicity’s sake that the first quarter averages for goods and exports will be exactly the same as the January readings.  In that case, the quarterly annualized nominal merchandise trade deficit would be $1,881 billion, a deterioration of close to $600 billion from the fourth quarter.  That would clearly tank GDP for the quarter, almost irrespective of anything else that happened in the economy.

However, if we assume that all but $4 billion of January “finished metal shapes” imports were gold (in December, it was less than $3 billion, so this seems to me to be a generous assumption), then the widening of the nominal trade deficit excluding gold would be about $190 billion, still a big number but less than one third as large.

Of course, to get to a first quarter estimate, one also has to make assumptions about what will happen in February and March as well as translate the nominal data into real terms.  In addition, we have to add in services trade flows.  I am willing to assume that the surge in import flows will not be sustained through March.  At the moment, I am eyeballing a real trade deficit widening in the first quarter of around $60 billion annualized.  There is certainly a risk that the actual result ends up being much larger, but I see that as a reasonable guess for now.

I would also note, as I often do, that a surge in imports usually finds its way into inventories.  Especially given that inventories increased by only about $10 billion in real terms in the first quarter, the odds favor a substantial bounce in inventory accumulation, which may offset most of the net exports drag.  At the moment, I have a first quarter real GDP advancing at about a 1.3% annualized pace, but this is mainly because I look for real final domestic demand to cool to a 1.4% annualized gain, not because of the net exports drag.  If you forced me to assume a $120 billion annualized deterioration in real net exports, then I would probably add another $40 to $50 billion in inventories, largely offsetting the bigger trade drag.

Just to introduce one alternative view, the Atlanta Fed GDPNow estimate for the first quarter has gained considerable notoriety in recent days, as it has plunged deep into negative territory.  While going deeply into the GDPNow methodology is a whole other story, suffice it to say that the projections for future months are derived from mechanical regression equations.  It has also become evident that the GDPNow equations do not account for the gold nuances described above.  As a result, the latest Atlanta Fed GDPNow estimate is -2.4%, with net exports accounting for a massive $256 billion annualized drag and inventories only providing a $25 billion offset.  So, they have net exports subtracting almost four percentage points from real GDP growth in the first quarter.  Hopefully, this offers some additional insight on why financial market participants are not supposed to put much weight on the stunningly weak GDPNow estimate.