Coming to the U.S. by Air:

A Lower Bound To Immigration Estimates

By Edwin S. Rubenstein

President, ESR Research

As the native population growth rate has fallen, net foreign immigration has come to constitute a larger and larger share of overall U.S. population growth. The size of this net inflow is not known with precision, however, and this constitutes the largest unknown in American population statistics. Despite the uncertainty, there has emerged a consensus among demographers and public policy analysts on its probable range.

After dropping to relatively low levels during the period 1931 to 1970, legal immigration to the United States has increased significantly in recent decades, primarily as a result of a change in immigration law in 1965. (see Table 1) According to the U.S. Immigration and Naturalization Service (INS), 7.6 million people attained legal permanent residence in the United States in the fiscal year 1991 to 1998 period. That comes to an average of 845,000 annually -- equaling the peak levels of immigration early in the Twentieth Century.

Table 1. Legal immigration to the United States: 1901 to 1998

Years	Legal Immigrants	Average per Year
1901-10	8,795,386	879,539
1911-20	5,735,811	573,581
1921-30	4,107,209	410,721
1931-40	528,431	52,843
1941-50	1,035,039	103.504
1951-60	2,515,479	251,548
1961-70	3,321,677	332,168
1971-80	4,493,314	449,331
1981-90	7,338,062	733,806
1991-98	7,605,068	845,008

Source: 1998 INS Yearbook.

At the end of the 1990s about 950,000 legal immigrants per year wer coming into the U.S., according to the INS. If this figure is correct, immigration accounts for more than one quarter of the country's total population increase. Consequently, accurate estimates of the total U.S. population are highly dependent upon accurate statistics on legal immigration.

How accurate is the immigration figure? Twenty years ago population researcher Daniel R. Vining, Jr. tested the validity of consensus immigration figures by estimating a plausible “lower bound” to net U.S. immigration.(footnote 1) Vining focused on one component of the net inflow of persons to the United States, namely, their net flow by commercial airlines, which, assuming that net migration by land, sea, and by clandestine and noncommercial air is positive, provides a lower limit for total U.S. immigration.

A lower bound is only interesting, of course, if it proves surprisingly high. The official U.S. government tally of arriving and departing air passengers certainly satisfies this criterion. In 2000, for example, the excess of arriving over departing international air passengers was 4.718 million, according to the U.S. DOT. (Table 2.) For the entire decade (1990 to 2000) it averaged 3.689 million per annum. When Vining looked at this data in the late 1970s, he found the excess to be about 1 million.

Table 2: International Air Passenger Arrivals to and From the United States (Thousands)
Year	Arrivals (A)	Departures (D)	Total (A + D)	Net (A - D)	Retention Rate (A-D)/A
1960	2,566	2,336	4,902	230	9.0%
1970	9,874	9,096	18,970	778	7.9%
1975	12,646	12,053	24,699	593	4.7%
1980	20,262	19,256	39,518	1,006	5.0%
1985	24,156	22,487	46,643	1,669	6.9%
1990	36,414	34,046	70,460	2,368	6.5%
1991	35,464	33,286	68,750	2,178	6.1%
1992	38,927	36,211	75,138	2,716	7.0%
1993	41,558	38,254	79,812	3,304	8.0%
1994	43,818	40,349	84,167	3,469	7.9%
1995	46,910	43,026	89,936	3,884	8.3%
1996	49,853	45,785	95,638	4,068	8.2%
1997	54,315	49,684	103,999	4,631	8.5%
1998	56,181	50,863	107,044	5,318	9.5%
1999	57,785	53,856	111,641	3,929	6.8%
2000	62,217	57,499	119,716	4,718	7.6%
Ann. Avg., 1990-2000	47,586	43,896	91,482	3,689	7.7%
SOURCE: U.S. DOT, National Transportation Statistics 2001, Tables 1-37 and 1-38; Air Carrier Profile (Appendix A). <http://www.bts.gov/publications/nts/index.html>

Interestingly, while the number of international passengers has risen by over 4-fold, the percentage difference between arriving and departing international passengers, which Vining calls the “retention rate,” has hardly changed at all: it was 7.8% in the 1970s, and was 7.7% in the 1990s. The constancy seems to imply that the impact of commercial air travel on U.S. immigration has risen in lock step with the number of airline passengers coming into the U.S.

The gap between international arrivals and departures (4.7 million in 2000) exceeds even the largest estimates made by demographers and government population experts of total net immigration into the United States. Part of the problem appears to be attributable to a systematic undercount of departures on flights on flights leaving the United States – but only a part. Vining found a systematic undercount of passengers departing on charter flights in the government statistics. The problem still exists: In contrast with most other countries, the U.S. has no exit controls at airports. While the DOT and INS (and, presumably, its successor agency) assures that the proper forms are filled out on flights arriving in the U.S., both chartered and scheduled, because all incoming passengers must pass through immigration and customs and because the federal agencies want their own counts to tally with those of the air carriers, they are less careful about recording outbound passengers. Outbound charter flights in particular are cited as not complying with the paperwork, and thus creating a large undercount of outbound international passengers.

Moreover, DOT rules regarding its basic international air traffic report (the T-100 report) had, until recently, exempted air carriers that operate aircraft with 60 seats or less. Therefore passenger volume in particular markets – particularly the Caribbean and certain Canadian markets – is understated due to the large amount of service provided by small aircraft.

Statistical coverage of international airline traffic was the goal of Federal legislation passed in the wake of 9/11. The Enhanced Border Security and Visa Entry Reform Act (HR 3525) that passed Congress last May requires U.S. citizens traveling abroad to reveal their identities and travel plans. For the first time government officials will monitor everyone who comes or goes from the United States as part of an overall strategy to enhance national security. Expanded monitoring is a first step toward a long-delayed automated entry/exit system for tracking visitors at U.S. ports of entry.

Although electronic tracking of all arriving and departing air passengers was to start on January 1, 2003, INS did not publish the proposed new rule in the Federal Register until February. The air passenger data requirement will not formally take effect until the final rule is published, sometime after the 30-day comment period. In addition, since there is normally a six-month lag before airline passenger data are publicly released, comprehensive data on international arrivals and departures will not be available until later this year.

Vining's "solution" to flawed government statistics was to use air passenger statistics independently collected by individual airports as well as by the International Air Transport Association (IATA) on air passenger travel to and from the U.S. over the North Atlantic – statistics that there is no reason to believe undercount departures relative to arrivals. The non-government data suggested a 50 to 60 percent smaller gap between arrivals and departures. Vining reported that government data on scheduled international flights, which in contrast to the charter flight data appear to be unbiased, were also consistent with this estimate.

Unfortunately, recent figures on arrivals and departures on scheduled international air flights are not readily available from DOT or the IATA. However, if we apply the retention rate for scheduled flights reported by Dr. Vining (3.8%) to current air passenger traffic volumes, we get an estimate of 2.4 million net arrivals in 2000 instead of the 4.7 million reported by the government. A more conservative approach would simply cut Vining’s retention rate in half, giving us an estimated million 1.2 million net arrivals in 2000.

Even this lower figure for airborne immigration is substantially above recent INS estimates of total net immigration to the U.S., which, of course, includes net migration by sea and land as well. The large disparity indicates either increasingly large scale underreporting of departing air passengers (even on scheduled flights) or a gross underestimate of U.S. immigration rates. Clearly further research is needed to resolve the question.

1. Daniel R. Vining, Jr., “Net Migration by Commercial Air: A Lower Bound on Total Net Migration to the United States,” Research in Population Economics, Volume 4, pages 333-350, 1982.