Trump Closed the Representation Gap on Immigration

Donald Trump has focused on the immigration issue more and restricted unauthorized immigration more severely than any other President in recent history. His policies are often framed as extreme and unpopular. Under this reasoning, the immigration policy of Donald Trump, as well as his repeated election as President of the United States, is a puzzle.

In this post, I provide evidence that Donald Trump’s immigration policy is neither extreme nor unpopular and that his election is therefore not a puzzle from the perspective of simple issue voting models. For decades, US citizens have demanded a reduction in immigration flows. Donald Trump is the first president to deliver a bundle of immigration policies that is close to what the median voter wants — he is filling the representation gap. Accordingly, his immigration policy is comparatively popular. Hence, there exists a plausible potential explanation for his election victory: many US voters vote for the candidate who is closest to their own immigration views, and for most US citizens, this closest candidate is usually Donald Trump.

Measuring (anti-) immigration attitudes

Immigration attitudes are multidimensional, such that a comprehensive understanding requires the combination of many individual measures. Still, this post relies mainly on a single survey item, which should be seen as a general caveat. The reason for this simplification is data availability. My goal is to track public opinion through several US administrations, which requires comparability, and only a few measures of immigration attitudes are available throughout.

I decided to focus on one attitude measure: the cleanest long-run trackable on US immigration opinion is Gallup’s question, asked since 1965: “In your view, should immigration be kept at its present level, increased, or decreased?“ It is a standard reference for researchers because it has the longest series, balanced response options, and asks about direction-of-change relative to current policy rather than about specific bills or events. Other items exist (Pew “should be reduced”, ANES restriction items, GSS levels), but most are shorter, policy-specific, or unbalanced.

With this caveat in mind, my post should be read as describing one plausible model for what is going on regarding US immigration policy, a model that should be tested more rigorously by others. I may do so in the future myself, but such a test is beyond a Substack post. Nonetheless, it appears that the model I propose here — Trump just filled a representation gap — does pass a reasonable empirical test.

The US representation gap on immigration

Figure 1: US public opinion on immigration. Note: Three response shares (decrease / satisfied with present level / increase). Shaded bands mark presidential administrations. Own calculation based on Gallup data.

Figure 1 shows the immigration attitudes of a representative sample of US citizens over time, based on the Gallup survey item. This figure is well-known and cited frequently as informative about US public opinion. One can see that the red line is usually above the green line, meaning that the share of people wanting a decrease in immigration is larger than the share wanting an increase. I will use the difference between these two shares as a measure of how much US citizens lean toward a stricter immigration policy. For instance, if a large share of people desire lower immigration and few want higher immigration, this indicates an overall net desire for less immigration.

Since the Gallup item uses actual policymaking as its reference category, I interpret a net desire for less immigration as a representation gap. Promising a policy that fills such a representation gap need not necessarily increase one’s chances of getting elected, but under many simple and reasonable attitude distributions and election models, it does. This post is supposed to test whether such a simple and reasonable model can help to explain why Trump got elected.

Figure 2: Net desire to decrease US immigration. Note: Net desire to decrease is the share saying “decrease” minus the share saying “increase” (Gallup). Shaded bands mark presidential administrations. Own calculation based on Gallup data.

Figure 2 visualizes the net desire for lower immigration. Two findings stand out:

First, the public has wanted less immigration than it got, by a wide margin, for nearly the entire post-1995 period. Across the 31 Gallup waves from 2001 to 2025, 29 carry a positive net desire to decrease. The two exceptions, in 2020 and 2021, may be related to the post-COVID border closure.

Second, the gap closes most under both Trump administrations. Net desire to decrease falls to single digits under Trump-1 (mean +3.5) and again under Trump-2 (+4.0). All other administrations have net desires for lower immigration in the double digits.

Table 1 shows average values by administration and also adds the share of citizens who are satisfied with the current magnitude of immigration flows. The net desire measure ignores these people by construction. As can be seen, the share of citizens who are satisfied with the current magnitude of immigration flows is highest under the two Trump administrations. Hence, compared to other administrations, Trump’s immigration policy is not only closest to the median citizen, but it also appears comparatively popular. Notably, the two Trump administrations are within roughly two percentage points of each other on every component.

Table 1: Gallup immigration responses by administration. Note: Mean response shares across all Gallup waves under each administration. Net = % decrease minus % increase, in percentage points. Number of waves per administration in parentheses. Republican administrations shaded red, Democratic blue. Own calculation based on Gallup data.

Where this locates the median voter

If we take the Gallup measure seriously, the median US citizen is more right-wing on immigration than any US administration since Clinton, including both Trump administrations. The two Trump administrations come closest to the median but do not reach it; they sit slightly to the left of it.

Many commentators frame Trump’s immigration policies, particularly those of his second administration, as extreme. “Extreme” may be defined in different ways; for instance, certain policies may be directly assumed to be extreme. In the context of politics, one very natural definition of “extreme” is a policy that is very different from the attitude of the median voter. Under this definition, the immigration policy of Donald Trump is the least extreme policy bundle at least since Clinton, with Clinton himself being the most extreme.

This fact seems very important because it gives a sense of how a bundle of immigration policies looks like that fills the representation gap. Such a bundle is slightly more restrictive than current US policymaking.

If other commentators use the same definition of extreme as I do, their estimate of the median citizen is substantially mistaken. This could explain why they fail to find ways to stop the rise of right-wing populism. Whether or not to accommodate their positions and, if so, by how much one should change one’s own positions, crucially depends on precise estimates of what immigration policy citizens want. If an observer believes that Trump-2 is far from the median voter, his perception is likely so biased that he is not able to propose a popular set of policies.

Different survey items can and do show different patterns. Large shares of Americans want ICE abolished, want DACA preserved, and want family-separation policies reversed. Hence, some results do suggest that most Americans are more liberal on immigration than Trump’s policies. But there are also results suggesting that most Americans are more right-wing than the current immigration policy. Significant shares want stricter workplace enforcement, faster removals of those with criminal records, and reduced asylum admissions. Both directions exist simultaneously in the same population.

Hence, focusing on individual, untested policy-specific items bears the risk of cherry-picking, particularly with such an emotionally loaded topic. As this post suggests, the US media tends to cite survey items that portray the public as more liberal than it really is. This may explain why many commentators overestimate how liberal the public is and, as a result, how extreme Trump is. To rule out such cherry-picking, I use the Gallup item as a neutral, broad, and established measure.

Of course, the fact that Trump’s recent policies fill the representation gap does not automatically mean that they are good. Critics may respond that the median voter is just wrong, which is, I think, the underlying battle populists and mainstream politicians are fighting. This post does not take a stance regarding this fight.

Does opinion respond to actual immigration?

The Gallup item asks Americans whether immigration should be “decreased,” “increased,” or “kept at the present level.” This is a direction-of-change question relative to the status quo flow of immigration. So, the question maps to: is the actual flow too high, too low, or about right relative to what people want? The natural empirical correlate is therefore the flow of immigration over time.

Whether public opinion follows actual immigration matters beyond academic interest. A common position in the policy debate is that public opinion on immigration responds to media portrayal rather than to actual immigration outcomes. If true, Democrats would only need to change their messaging or attack their opponents, not their actual immigration policy. Real immigration numbers would be irrelevant.

To shed some light on whether public opinion responds to actual immigration, I compare the net desire for less immigration with the inflow of legal immigrants (DHS Lawful Permanent Resident issuance, annual) and the net inflow of unauthorized immigrants (Federal Reserve Bank of Dallas Economic Letter, monthly, from 2014). I have argued in a prior post that the unauthorized layer is the one that moves preferences for the case of Germany.

Figure 3: Opinion and flow. Note: Net desire to decrease per Gallup wave on the left axis (Gallup wave dates, not annualised). Right axis in thousands per month: net unauthorized inflow (Dallas Fed monthly micro-data through December 2025, from the March 2026 Economic Letter, net entry = gross entry minus exits, identical to the black series in Dallas Fed Chart 1) and legal permanent residents (DHS, through FY2024 from the OHSS Legal Immigration and Adjustment of Status Q4 release; the full Yearbook 2024 is not yet posted; annual issuance divided by 12 to put it on the same monthly-thousands scale).

The blue line, legal admissions, is flat across the entire period: roughly 1 million a year regardless of administration, with no co-movement with opinion. The ochre line, net unauthorized inflow, rises sharply during Biden (peaking near 310,000 a month in late 2023, an annual rate of about 3.7 million) and collapses to negative values under Trump-2 (averaging minus 46,000 a month across calendar 2025, an annual sum of about minus 550,000). The black line, net desire to decrease, follows the ochre line: low during the pre-Biden years, where unauthorized flow was modest, rising sharply through Biden as flow rose, then falling sharply under Trump-2 as flow turned negative.

Consistent with the visual pattern, net desire to decrease and Dallas Fed net unauthorized inflow correlate r = +0.68 with a one-year lag (95% CI [+0.13, +0.91], p = 0.022, n = 11). With legal admissions, the one-year lag correlation is essentially zero (r ≈ +0.25, not statistically distinguishable from zero). A first-difference specification at lag-1 (asking whether year-on-year changes in opinion co-move with year-on-year changes in flow) gives r ≈ +0.70 with a similarly wide confidence interval, so the unauthorized-flow relationship does not look like a shared trend alone. None of these results is precisely estimated or causally identified (only 11 annual observations). I added this numerical analysis mainly for completeness.

Still, the visual evidence from Figure 3 appears clear enough for observers to take the hypothesis that public attitudes follow unauthorized immigration seriously. According to this model, most US voters are fine with the current level of legal immigration but think that there is too much unauthorized immigration. They differ in how much unauthorized immigration they find optimal and compare the actual numbers to their preferred number to judge whether the number of unauthorized immigrants is too high, too low, or just right. If the number of unauthorized immigrants increases, more US citizens consider the actual number too high and therefore demand a reduction. This happens with some delay because it takes time until an increase in unauthorized immigration becomes visible. Under this model, Trump’s policy regarding unauthorized immigration comes closest to what the median citizen wants: very little or negative unauthorized immigration. In fact, as Figure 3 shows, Trump-2 is the only administration to achieve sustained negative levels of unauthorized immigration, i.e., net emigration.

The two Trump administrations

Because Trump-2 did in fact strongly decrease unauthorized immigration, which may exactly be what most US citizens want, it makes sense that the representation gap closed and that the share of those satisfied with the immigration flow is now at a record high (Table 1).

However, actual unauthorized immigration does not explain the full pattern. Net unauthorized inflow under Trump-1 averaged about 368,000 a year, on the same order of magnitude as the late-Obama Dallas Fed values (around 380,000 a year in 2014-2016, the late-Obama years our monthly series covers). Under Trump-2, the same series ran at minus 548,000 across the full calendar year 2025. Hence, if only the unauthorized flow mattered, one would expect Obama and Trump-1 to score similarly, below Trump-2. In the following, I will highlight a few factors that may help to explain why this is not the case.

Perceived effort

Under both Trump administrations, executive posture on immigration shifted to credible restriction immediately on inauguration: travel ban, ICE workplace raids, asylum-metering, family separation, the wall rhetoric. Most of these efforts ran into court blocks, congressional non-cooperation, and an executive branch that had not yet learned which institutional levers actually move the flow.

The signal of effort may have registered with the public regardless of whether the aggregate flow moved. Trump-2’s enforcement is the same posture executed with more institutional knowledge: the same restriction intent, but with eight years of learning about which executive authorities to use (§212(f), the asylum-bar rule), which third-country agreements to negotiate, and which procedural choke-points to close. The signal is therefore credible under both Trumps because in both cases the executive is visibly trying, and now also visibly delivering.

Misperceptions

The close resemblance between unauthorized immigration and net desire for decreased immigration suggests that Americans are reacting to a changing reality. However, this does not rule out misperceptions. Trump’s speeches and his media portrayal may have fooled many voters into thinking that immigration numbers were extremely low under Trump-1, which they were not.

Preference change

The responses to the Gallup item are the result of an interaction of perceived flows and underlying preferences. Those underlying preferences are not observed here, and they may have changed between Trump-1 and Trump-2. To explain the pattern, this would require Americans to be now fundamentally more anti-immigration than under Trump-1. While this is possible, I do not think that this is the case because, as Alexander Kustov has shown, such underlying immigration-related preferences are quite stable.

Will the closure last?

Will the closure of the representation gap persist past Trump-2’s term? The Economist editorial of 13 May 2026 (”Asylum in America is all but over. It may never come back.“) suggests that it will. The reason is the institutional configuration Trump-2 has built.

Trump-2’s bid for non-reversibility is two-sided. On one side, the statutory and regulatory ceilings that Trump-1 could not break through are operationally gone: aggressive use of INA §212(f), the DOJ/DHS asylum-bar rule, and a wave of mass third-country agreements together remove the procedural floor that bounded Trump-1’s enforcement. Hence, the previous operational asylum architecture is largely no longer functioning. On the other side, new restriction infrastructure has been built up: the removal apparatus, third-country agreement infrastructure, expanded ICE detention capacity, expanded federal-state cooperation programs. And asylum-side infrastructure has been actively run down: the asylum officer corps cut, USCIS asylum processing scaled back, refugee resettlement effectively suspended, humanitarian parole programs ended.

The asymmetry between dismantling and rebuilding is what produces durability. Restriction-side infrastructure that has been built up persists into the next administration regardless of executive intent. Asylum-side infrastructure that has been gutted requires personnel hiring, training, treaty renegotiation, and litigation work that take years. A Democratic successor inheriting this configuration would face a multi-year reconstruction project before the pre-Trump-2 asylum pipeline could process at scale. That may leave the representation gap closed even without permanent statutory change. The default outcome is now gap-stays-closed rather than gap-reopens.

Europe may be next

In Europe, the same representation gap on immigration persists that had existed for decades in the US. As is often the case, Europe seems to follow the same trajectory as the USA.

Sweden is the clearest single case. The right-wing populist Sweden Democrats, excluded from coalition politics for two decades, are now the parliamentary support for the Kristersson centrist government and have driven a sharp tightening of asylum policy: cash incentives for voluntary return, work-permit requirements raised, asylum recognition rates cut, family reunification restricted. The party that the mainstream refused to deal with until 2022 is now setting the immigration agenda. Net asylum approvals dropped sharply through 2023-2025.

At the EU level, the New Pact on Migration and Asylum (formally adopted April 2024) is in its implementation phase. The asylum-procedure regulation, the screening regulation, and the solidarity mechanism all narrow the procedural slack that asylum-seekers had under the 1999-2013 Dublin/CEAS architecture. The political momentum visible in 2025-2026 is further-right than the New Pact itself: member states are pushing for the kind of third-country processing arrangements that the Pact itself did not contemplate.

Most importantly, the 1951 Geneva Refugee Convention is now openly under pressure. Multiple European leaders have called for its renegotiation or reservation; legal scholars are publishing serious work on what an exit or reformed instrument would look like. The Convention has been the load-bearing international legal anchor for asylum since 1951; if it falls or is fundamentally reformed, the European architecture for asylum follows the US one into a different operational regime.

The deeper pattern across these cases is the same one visible in the US Trump-1-to-Trump-2 sequence. First-time populist governments come in, try a wide range of restriction measures, and run into the same institutional brake the US Trump-1 administration ran into: courts, lawyers, treaty obligations, civil-service resistance, coalition partners. The first attempt produces opinion-closure but limited flow-closure. The populist parties then either return to opposition or share government, learn which institutions actually bind, and come back better prepared. Meloni in Italy fits the pattern (FdI excluded for decades, in government since 2022, escalating institutional moves in the second half of the term). Sweden has already cycled through the first-attempt phase under Kristersson and the new round is now visible in the New Pact implementation. If this reading is correct, the European representation gap closes in the second populist cycle, not the first.