Federal Reserve Economic Data

The value of the US dollar can influence trade flows by changing the relative prices of exports and imports. A stronger dollar tends to make imports cheaper for Americans and US goods more expensive abroad, which can put upward pressure on the trade deficit.

In practice, though, the relationship between the dollar and the US trade balance is far from consistent.

Our FRED graph above shows two measures:

• A trade balance ratio, defined as (exports − imports) ÷ (exports + imports), which expresses the US trade position relative to the total value of trade flows.
• An exchange rate measure, which in this case is the trade-weighted US dollar index, which reflects the nominal value of the US dollar against a broad basket of currencies for goods trade.

When we plot these two series together, we see that the relationship varies over time.

• In 2014-16, the dollar strengthened considerably and the trade balance ratio (exports minus imports, divided by total trade) weakened, consistent with the idea that a stronger dollar can reduce net exports by making US goods more expensive abroad and imports cheaper at home.
• In both the 2008-09 recession and 2022-23, the trade balance ratio improved alongside a stronger dollar, suggesting that other forces, such as collapsing import demand during a downturn or shifts in commodity prices, were the dominant drivers.

Trade policy can also affect both the dollar and the trade balance in ways that break the usual pattern.

• The 2018-19 tariff increases on a broad set of imports, especially from China, affected relative prices and sourcing decisions directly. They may have contributed to a stronger dollar through capital inflows, while at the same time reducing certain import volumes.
• In 2025, across-the-board tariffs and targeted increases on specific goods could again influence the trade balance through price and sourcing effects that do not operate primarily through exchange rate changes. They have the potential to shift both import volumes and export competitiveness, sometimes reinforcing and other times counteracting the influence of the dollar.

These episodes underscore that the link between the dollar and the trade balance is not systematic. Exchange rates are just one factor in shaping trade outcomes. Domestic demand, global growth, commodity price swings, and trade policy all play a role. And in recent years, tariffs and other trade measures have been especially relevant.

How this graph was created: Search FRED for and add “Trade Weighted U.S. Dollar Index: Broad, Goods” (DTWEXBGS) to the graph on the left axis. From the “Edit Graph” tab, add “Exports of Goods and Services” (EXPGS) and “Imports of Goods and Services” (IMPGS) as Line 2. To do this, enter the formula (a-b)/(a+b) in the Line 2 tab. Finally, change the starting date to “2006-01-01.”

Suggested by Ana Maria Santacreu.

Does it always feel like your flight is full?

It makes sense that more people experience full flights because there are, by definition, fewer people on less-crowded flights. The other reason is that, yes, flights are indeed mostly full.

Our FRED graph above shows “load factors” for US airlines: that is, the percentage of seats sold.

The red line shows clear seasonal patterns: If you don’t like crowds, avoid flying in June and July and instead fly in January and February.

The blue line shows the same series, but removes the regular seasonal patterns. Here we can see how the load is trending within a year without having to compare with the same month in the previous years. And we see that the de-seasonalized load is fairly constant over time.

Our second FRED graph, below, shows seasonal data for domestic flights (blue line) and international flights (red line). There’s little difference between the two series, except for the period right after the pandemic. In particular, it doesn’t look like there’s much room for airlines to arbitrage between domestic and international flights during the year if the same planes could be used for both.

How these graphs were created: Search FRED for and select one of the load factor series. Click “Edit Graph” and use the “Add Line” tab to search again for the other series. Use the “Format” tab to change the settings of the second line. Proceed similarly for the second graph.

Suggested by Christian Zimmermann.

Every month, the Bureau of Labor Statistics (BLS) releases data on total nonfarm employment in two forms: seasonally adjusted and not seasonally adjusted.

• Seasonally adjusted (SA) employment data have had the effects of seasonal changes removed, such as the typical increase in hiring during the holiday season. This allows us to more clearly see the business cycle trends in employment.
• Not seasonally adjusted (NSA) employment data are the raw employment levels at a given time.

Our FRED graph above tracks the past two years of SA and NSA employment levels, showing how the number of workers rises and falls throughout the year. Looking at the latest initial data for July 2025, we can see the NSA number of jobs declined by over 1 million, while the SA number increased by 73,000.

With almost 160 million workers, the BLS cannot count each job every month. They use a sample of data from the Current Employment Statistics (CES) survey and a model to estimate hiring and layoffs by new firms that arose and former firms that went out of business since the previous survey. The BLS website provides more information on this process.

Because of this estimation, revisions to employment data are common. The BLS receives additional responses after their initial release of monthly data and adjusts as seasonal factors are more accurately calculated for the year. Check this FRED Blog post for more insight into BLS revisions.

Our next two graphs come from ALFRED: They show the revisions to the May and June SA and NSA employment numbers between the release of the data on July 3, 2025, and the next release of the data on August 1, 2025. The changes in SA employment were revised down by 258,000. This feeds into the total employment for June, which decreased from 159,724,000 to 159,466,000, or a –0.16% change in total employment.

To see which portion of the revision was driven by late-survey responses and which portion came from revisions to the seasonal adjustment, we can use the NSA revisions to decompose this number. Looking at the NSA revisions in the change in employment below, May and June were revised down by a combined 182,000. Taking this number to the seasonally adjusted number and subtracting it, we can see a downward revision of 76,000 came from the new calculations of seasonal adjustment. Both of these revisions were in the same direction, but sometimes they offset one another, which is why we subtract the NSA from the SA numbers.

A second and final revision for the June numbers will occur on September 5, when the first revisions for the July numbers will also be available. The May numbers were finalized in the August release.

How these graphs were created: First graph: Search FRED for “All Employees, Total Nonfarm” and click the first link to get the seasonally adjusted numbers. Then click the blue “Edit Graph” and open the “Add Line” tab. Search for PAYNSA and click the first result for the not seasonally adjusted numbers and then go to the format tab and click customize to change the line style to dash and the color to red. Then change date range to the last two years. Next two graphs: Go to ALFRED and search for All Employees, Total Nonfarm and click the first series labeled Monthly, Seasonally Adjusted. Use the “Edit Graph” panel to change the units to Change, Thousands of Persons. Select Bar 2 and do the same. Repeated this process for the Monthly, Not Seasonally Adjusted series.

Suggested by John Fuller and Charles Gascon.