In a blog post on March 5, Jeremy Neufeld of the Niskanen Center argued that our estimate of the possible number of birth tourists was far too high. In a response, we defended our number, and he responded on March 9.
Neufeld was correct all along in identifying our error. We thank him for his attentiveness.
The exchanges primarily revolve around two key issues. To arrive at our estimate of the possible number of birth tourists, we first subtracted the number of births to foreign-born women living in the United States as shown in the American Community Survey (ACS) from births to foreign women as reported by the Centers for Disease Control (CDC) on birth certificates. The difference represents our baseline estimate of the number of people who may have had a child and then left the country. Neufeld argues that, first, the number of foreign-born mothers in the ACS was higher than we estimated, so the number of potential birth tourists is lower, and second, that most of the remainder are likely long-time residents, not birth tourists, who he thinks could have left the country soon after giving birth, creating the difference between the CDC and ACS figures.
Taking Neufeld's second criticism first, his observation that nearly all of the difference (93 percent) between ACS and CDC data could be explained by long-time immigrants who left shortly after giving birth did not, and still does not, make sense to us. As discussed at length in our response on March 9, "We know from the Census Bureau's Current Population Survey that few young children with immigrant mothers leave each year. This should come as little surprise as one of the primary reasons people come to America is so that their children will have the benefit of growing up in the United States."1 Thus his idea that almost all of the difference between the ACS and CDC could be explained by well-established immigrants having a child and then quickly leaving the country seems extremely unlikely.2
But Neufeld's other criticism, dealing with the number of foreign-born people in the ACS, was correct. In our March 9 response to his original post, we thought he had made a mistake by using the wrong definition of the foreign-born from the American Community Survey (ACS). Upon closer examination, he is right: The CDC definition of foreign-born mothers does, in fact, include new mothers who were born in the outlying U.S. territories of Puerto Rico, American Samoa, the Northern Mariana Islands, Guam, and the U.S. Virgin Islands. Our analysis in contrast used the citizenship question and did not include these new mothers as part of the foreign-born. This meant that our estimate of the number of foreign-born mothers in the ACS was too low.3 Thus when we subtracted ACS births from CDC births, the result was an estimate that was too high.
However, it turns out that newly released data recently from the CDC renders this issue moot.
The new Natality for 2016 - 2018 (expanded) data at the CDC website now allows users to generate tables that show exactly how many births there were to mothers who themselves were born in outlying territories. These individuals can now be easily subtracted from the old CDC definition of the foreign-born, making for more accurate estimates. Even better, it is now possible to calculate the number of births to foreign-born mothers by month. This also allows us to more accurately match birth figures from the ACS.4 As best as we can tell, the 2016 to 2018 expanded tool that allows this detailed analysis was only recently added to the CDC website.
Table 1 reports the new numbers, which we use to re-create our prior estimate. We first take the number of births to foreign-born mothers (old CDC definition) in the second half of 2016 and the first half of 2017. Second, we subtract from the old CDC definition all births to mothers from outlying U.S. territories, which can now be identified in CDC data by month. As reported in row four, the total number of births to foreign-born mothers in the CDC data is 877,416 when those from outlying territories are excluded.
Table 1. Births to Foreign-Born Mothers from CDC Data
|Second Half of 2016||First Half of 2017||Total|
|Births to foreign-born mothers using old CDC definition1||471,984||429,313||901,297|
|Births to mothers born in P.R. or outlying territories in CDC data2||12,360||11,521||23,881|
|Births to mothers whose country was unreported3||4,200||3,672||7,872|
|Births to foreign-born mothers in CDC data excluding outlying territories4||459,624||417,792||877,416|
|Births to foreign-born mothers in CDC data excluding outlying territories and apportioning mothers with unreported nationality5||460,590||418,637||879,227|
|Births to foreign-born mothers in CDC data excluding outlying territories and assuming all mothers with unreported nationality are foreign-born||463,824||421,464||885,288|
1 All mothers born out of the 50 US states and Washington, D.C.; includes those born in the outlying U.S. territories of Puerto Rico, American Samoa, the Northern Mariana Islands, Guam, and the U.S. Virgin Islands.
2 Births to mothers born in outlying U.S. territories.
3 These mothers have a reported place of birth of "unknown" or "not stated" in CDC data.
4 Takes births to all mothers born outside of the United States shown in row 1 and subtracts births to mothers from outlying U.S. territories.
5 Takes totals from prior row and adds 23 percent of mothers with an unreported place of birth, based on the distribution of foreign- and U.S.-born mothers in CDC data.
In addition to these births, there were 7,872 new mothers who did not indicate their country of birth on their birth certificates. We do know from the place of birth question on the ACS that new mothers who do not report a place of birth are all foreign-born.5 Therefore, it seems likely that a significant share of such mothers in the CDC data are also foreign-born. Table 1 makes different assumptions about the place of birth of these cases: 1) none are foreign born; 2) the share foreign-born is the same as the share foreign-born overall; 3) all are foreign born.6 This is a relatively small population we had not taken into account in our prior analysis. Since we include those mothers in the ACS who do not report a country of birth in our analysis it is reasonable to make some provision for these mothers in the CDC data.7
Following the same approach as before, Table 2 subtracts the number of women in the ACS who had a child in the prior year from the number of new foreign-born mothers based on birth certificate records as reported by the CDC. Both the citizenship question and the place of birth question in the ACS now produce the same number of births to foreign-born women (845,896), once those women born in outlying U.S. territories are excluded. Again, the assumption is that those found in CDC data, but not in ACS data, should be an indication of the number of people who had a child and then immediately left the country.
Table 2. Estimating the Possible Number of Birth Tourists
|Births to foreign-born mothers in CDC data, excluding outlying U.S. territories1||Births to foreign-born mothers in the ACS, excluding U.S. outlying territories2||Baseline estimate of possible birth tourists3||Assumes some of the difference in CDC and ACS data is due to U.S. citizens who left the country shortly after giving birth4|
|Assumes none of those with unknown nationality are foreign-born||877,416||845,896||31,520||20,488|
|Assumes 23% of those with unknown nationality are foreign-born||879,227||845,896||33,331||21,665|
|Assumes all those with unknown nationality are foreign-born||885,288||845,896||39,392||25,605|
1 Figures are from last three rows of Table 1.
2 Based on citizenship question in the ACS. To match CDC data it includes those born abroad of American parents, naturalized citizens and non-citizens, but not those born in outlying territories.
3 Subtracts second column from first column.
4 Assume 35 percent of baseline estimate may be due to foreign-born American citizens who may have left country shortly after giving birth.
The difference between births in the CDC and ACS data can be seen as our baseline estimate of the number of possible birth tourists. Table 2 shows this number is 31,520, 33,331,or 39,392 depending on how mothers who do not report a country of birth are allocated. Our highest baseline estimate of 39,392 is certainly substantial, though it is still significantly lower than our prior baseline of 51,327. If, as before, we reduce these figures to reflect the possibility that some of the difference between ACS and CDC data is due to foreign-born American citizens who left the country, it leaves us with 20,488, 21,665, or 25,605. This means that there were possibility 20,000 to 25,000 birth tourists in the second half of 2016 and the first half of 2017.
The CDC online data tool now allows for more detailed analysis than was possible when we first developed our estimate last year. This has improved our ability to estimate birth tourism.
That said, it is still the case that our new estimate range should not be seen as hard data points. As we previously stated, our estimate should be seen as a "rough approximation, based on limited data". As we further made clear in our original post back in December, the estimates come with a number of important caveats, which we listed at the time. These caveats all still hold true with our new estimates.
1 The Current Population Survey Annual Social and Economic Supplement, unlike the ACS, asks individuals about their mother's place of birth. The number of four-year-olds in 2018 who were U.S.-born children with a foreign-born mother was just 2.2 percent less than the number of one-year-olds in 2015. The difference was less than 17,000 and well within the margin of error of the Current Population Survey, which is a much smaller survey than the ACS. This indicates that immigrants with young U.S.-born children seldom take their children and leave the country.
2 Neufeld's defense of this assumption is that most births to foreign-born mothers in the ACS data are to well-established mothers. However, our birth tourist estimate is based on people not found in the ACS, but who show up in CDC data. The CDC data does not include any information about citizenship status or length of stay in the United States of new foreign-born mothers, so it not clear why he thinks this is a good reason to assume almost all the difference in the two data sources is due to longtime residents leaving. He does observe that because we allow for the possibly that about one-third of those not in the CDC data could have been citizens who left, it is reasonable to assume that nearly all (93 percent) of them are citizens or longtime residents who left shortly after giving birth. But he provides no additional justification for why this is possibly the case, despite its seeming implausibility.
3 The Census Bureau, in the citizenship question on surveys such as the ACS and Current Population Survey, defines all persons born in the United States, born in outlying territories, or born abroad of American parents as the "native-born", while only naturalized citizens and non-citizens constitute the "foreign-born". The CDC does not ask new mothers about citizenship and relies solely on the mother's place of birth reported on birth certificates to define the foreign-born. As a result, published reports from the CDC considered those born abroad to American parents and those born in outlying territories as foreign-born. The inclusion of those born to Americans abroad with the foreign-born is unavoidable by the CDC since only place of birth is recorded. But including those born in outlying territories never made sense because all of these mothers were American citizens when they were born. In our prior estimate, when we calculated births to foreign-born mothers in the ACS we did include mothers born abroad to American parents with the foreign-born, but not those mothers born in outlying territories. However, it would have been possible to include those born in outlying territories had we used the place of birth question as Neufeld argues. As we will see, the CDC now allows researchers to separate out those mothers born in outlying territories, making it possible to use the citizenship question, or the place of birth question from the ACS for that matter, to compare births in the ACS and CDC.
4 Our prior estimate required us to assume that the distribution of births to foreign-born mothers in the last half of 2016 and first half of 2017 was the same as that for all births. This is necessary because the ACS reflects the population at mid-year, while the fertility question asks respondents if they had a child in the last year, so for 2017 it reflects births in the second half of 2016 and the first half of 2017.
5 We know this because there is a citizenship question on the ACS. All of the new mothers in the ACS whose place of birth is reported as "other and n.e.c. [not elsewhere classified]" are not American citizens based on the citizenship question.
6 Excluding those who did not report a place of birth in the CDC data, 23 percent of the all new mothers were foreign-born. If we add 23 percent of mothers with an unknown reported place of birth to the foreign-born mother figures it means there are 879,227 births to foreign-born mothers in the CDC data.
7 We could exclude those in the ACS who do not report a place of birth, but there is no need to do so since by using the citizenship question we know these individuals are all foreign-born, as they all indicate they are not American citizens.