As I find myself waking up unexpectedly at odd hours, such as 3:30 AM without reason, I often turn to engaging activities to pass the time. This past weekend was no different. By around 5:30 AM, after exhausting my usual collection of phone games including Wordle, Quordle, and others, I sought refuge in a book on the extinction of dinosaurs and decided to revisit a thrilling nine-inning baseball victory by the Texas team. Unfortunately, rather than easing my restlessness, these distractions only ignited my curiosity further.

Amidst my restlessness, I stumbled upon an insightful article from Axios highlighting a dramatic plunge in international travel to the United States that occurred in March. As an avid data enthusiast, I was particularly intrigued by the article's reference to an official government dataset that was previously unknown to me.

Delving into the Customs and Border Protection (CBP) Average Wait Time website revealed a treasure trove of information that can be vital for understanding current travel trends. Although the system is somewhat cumbersome—requiring individual selections for agencies and terminals, and with data available only for the past three years, despite its inception in 2008—the dataset is highly exportable. With a bit of effort, it's possible to analyze the daily processing of both U.S. and non-U.S. travelers at some of the country's busiest international airports. This capability is particularly useful for assessing whether new policies may be influencing potential travelers' decisions to visit the U.S.

After downloading data for eight major airports, I aimed to illustrate how entry numbers were shifting. Recognizing that travel often fluctuates seasonally, I sought to compare the current year to the same period in the previous year, rolling over 30 days to account for any recent trends without getting lost in the daily variances.

The findings were quite striking: the past 30 days at this group of airports indicated a reduction of over 10 percent in the number of foreign travelers being processed. I shared an initial version of this graph on Bluesky, which garnered some attention. This immediate response made me apprehensive, as there is always a risk of systemic underreporting when dealing with very recent data.

My experience with CBP data led me to contemplate the parallels with crime statistics, particularly the early-year data in that realm. Previously, I had expressed my admiration for the Texas Department of Public Safety's Uniform Crime Report website, which provides extensive options for accessing virtually all data Texas reports to the FBI, promptly after local agencies submit their findings.

Notably, the San Antonio Police Department excels in publishing the most recent month's data early in the following month. For instance, the February 2025 data was available at the start of March 2025, allowing for quick comparisons of theft rates year over year. Preliminary calculations indicate an 18 percent decline in theft offenses in San Antonio during the initial two months of 2025 compared to the same timeframe in 2024.

However, caution is warranted when interpreting this data. Utilizing an Ad Hoc Query on thefts in San Antonio from the beginning of the year highlights potential discrepancies. The report reveals that there were 121 thefts reported daily in San Antonio from January 1 to February 18, decreasing to 105 per day from February 19 to 24, and further declining to 88, 67, 75, and just 18 on February 25 through 28, respectively.

It is essential to note that law enforcement agencies have until the following year to submit finalized 2025 data to the FBI, meaning these numbers can undergo adjustments (usually upward). Additionally, the January statistics are often susceptible to revisions, although the changes typically lessen significantly after a month or two.

Theft rates experienced a 5 percent decrease in San Antonio during 2024, and early indicators for January and February suggest a continued decline in 2025. Nevertheless, the uncertainty surrounding these figures calls for analytic caution when estimating the actual decline’s magnitude.

As Carl Sagan famously remarked, extraordinary claims necessitate extraordinary evidence. While asserting that theft rates in San Antonio are markedly declining isn’t necessarily an extraordinary statement, the significant drop reflected in the preliminary data calls for further validation. A thorough examination allows for a better understanding and context regarding the likelihood of a decline thus far this year, ensuring a cautious approach to analysis.

Now, returning to the analysis of air travelers entering the U.S., I was apprehensive about my first foray into this dataset, especially since it suggested a notable and sudden decline. To ascertain whether this dip was a genuine change or merely a reporting anomaly, I replicated the process focusing on U.S. travelers processed by CBP. Assuming that U.S. travelers’ behaviors remain steady, a significant drop in their numbers would likely indicate a data reporting issue.

Examining data from New York's JFK airport, Los Angeles International Airport (LAX), and Miami further supported this theory. Additionally, I looked at Orlando, which revealed some wild fluctuations.

Across all four airports, the data illustrated a consistent trend over the past 30 days: a significant decline in non-U.S. passengers being processed, contrasted by a stable or even slight increase in U.S. passengers. This trend strongly suggests that we are witnessing a genuine decline rather than a delay in data entry, as one would expect similar patterns for both U.S. and non-U.S. travelers if that were the case.

To bolster my findings, I revisited the CBP processing data for JFK the following day. The old data indicated 260,250 entries from March 29 to April 3, 2025, and the updated data confirmed the same figure. The exact match provided strong evidence that the observed trend is based on accurate reporting rather than discrepancies.

It is important to remember certain nuances in the data as it relates to context. For instance, travelers from Canadian airports often clear customs there, suggesting that the current dataset could underestimate the actual decline in passenger numbers. Additionally, some users on Bluesky pointed out that Easter falls later this year, potentially leading to fewer travelers heading to Orlando for Spring Break. However, as Easter 2024 was earlier than in 2023, I remain skeptical of its impact on traveler numbers this year.

The data compellingly indicates a marked decrease in the number of international visitors coming to the United States compared to the same time last year. While the newsletter refrains from discussing political factors, the reasons behind this shift are evident and warrant no additional elaboration here.

From a data analysis standpoint, this situation presents an interesting case study on identifying critical trends and validating their authenticity. As time progresses, this issue deserves close scrutiny, and the CBP dataset offers a valuable tool for monitoring whether the situation improves, worsens, or stabilizes as the year unfolds.

Thank you for reading my insights at Jeff-alytics! Feel free to share this post publicly.