What Really Draws Skilled Migrants? New Fuzzy Math Reveals the Hidden Policy Lever

Every government wants the same prize: the engineers, doctors, researchers, and entrepreneurs who can move anywhere and choose where to build a life. Countries spend billions on tax breaks, fast-track visas, and glossy recruitment campaigns to win them. Yet a new study published in Frontiers in Political Science suggests that the levers policymakers reach for first may not be the ones that actually move the system — and that the factor skilled migrants say matters most to them is, in large part, a symptom of something else.

Enes Olgun and Furkan Baki Zaralıoğlu of Istanbul Medipol University set out to answer a deceptively simple question: what determines whether a country attracts internationally mobile talent? The literature offers six familiar answers — economic attractiveness, career opportunities, work–life balance and livability, institutional trust and governance quality, cultural compatibility, and migration policies and incentives. Each has its own explanatory tradition, from push–pull wage-differential accounts to human capital theory, from the quality-of-life and amenity literature to institutionalist accounts of governance, from cultural-distance studies to the policy and regulatory strand. What the literature has never agreed on is which of these is most decisive, and the two authors argue that the disagreement is not accidental. Because the determinants have been examined largely in isolation and treated as mutually independent, existing rankings cannot separate the factors that drive the system from those that merely register its outcomes — a distinction on which policy design depends.

The methodological core of the study is an integrated spherical fuzzy DEMATEL–AHP framework, a mouthful that hides a genuinely novel piece of decision mathematics. DEMATEL, the Decision-Making Trial and Evaluation Laboratory, recovers the causal structure among criteria: it asks which factors influence which, aggregates those judgments into a total-relation matrix, and classifies each criterion into a cause group (net influencer) or effect group (net recipient). AHP, the Analytic Hierarchy Process, elicits relative importance through pairwise comparisons. The innovation here is to run both within a spherical fuzzy environment, a generalization of fuzzy set theory that admits three independent degrees — membership, non-membership, and hesitancy — subject to a unit-sphere constraint. In plain terms, when a panelist says a factor has high influence, the mathematics records agreement, disagreement, and uncertainty simultaneously rather than collapsing them into a single number. The authors also introduce a defuzzification anchored at the No-Influence point of the linguistic scale, so that a unanimous judgment of no influence maps onto a zero entry, preserving the zero semantics of the classical crisp scale while retaining the representational flexibility that distinguishes spherical fuzzy sets from their intuitionistic predecessors.

Judgments were elicited from a purposively sampled panel of ten internationally mobile professionals working in academia, engineering, human-resource management, and medicine, each having completed a degree in one country, migrated, and worked in the destination country for at least three years. Panelists were based in six countries — Czechia, Germany, the Netherlands, Oman, Qatar, and the United States — and supplied thirty directed influence judgments and fifteen pairwise importance comparisons each. Crucially, the authors did not treat all experts as equally reliable. A distance-based expert weighting procedure derived stage-specific weights from consensus proximity, rewarding panelists whose judgments sat closer to the group consensus. Because the two stages elicit different relations — perceived influence and preferential importance — consensus alignment was assessed separately within each stage, preventing disagreement about causal influence from contaminating the aggregation of preferential importance.

The results deliver a finding that should unsettle any talent-attraction strategy built on a single ranking. In terms of stated importance, economic attractiveness ranks first with a weight of 0.209, followed by career opportunities (0.178), work–life balance (0.170), and institutional trust (0.169), with cultural compatibility (0.139) and migration policies (0.135) closing the ranking. But in terms of causal role, the ordering is almost inverted. Migration policies and incentives exert the largest net influence in the system, at +1.548, and institutional trust follows at +1.379. The three determinants the panel rates most important — economic attractiveness, career opportunities, and work–life balance — all belong to the effect group. Economic attractiveness combines the highest prominence in the system, 13.786, with a moderately negative net effect of −0.458: it is heavily involved in the causal structure, but predominantly as a recipient.

The strongest directed links in the total-relation matrix tell a coherent story. The four links that exceed the threshold by the widest margin all originate in institutional trust or migration policy: institutional trust into economic attractiveness (1.403), migration policy into career opportunities (1.396), migration policy into economic attractiveness (1.384), and institutional trust into career opportunities (1.377). The quality of institutions and the design of migration policy shape the economic, career, and livability conditions that internationally mobile professionals experience, rather than the reverse. This does not contradict the literature documenting that skilled migrants sort on earnings and wage differentials; it re-situates that evidence. Earnings differentials are likely to be the channel through which institutional and policy conditions become observable to a prospective migrant, which may explain why economic attractiveness dominates stated preferences. The economic advantage a destination offers is, in substantial part, a reflection of the institutional and regulatory environment that shapes it.

The authors are careful about what this causal language means. The DEMATEL indices summarize influence as perceived and reported by the panel; they are not effects identified from realized migration flows. What the cause–effect classification recovers is where a group of experienced and, in the majority, disciplinarily qualified observers locate the leverage in the system — a claim about a shared model of how the determinants act upon one another, not an estimate of how much a given intervention would move a given outcome. The recommendations should be read as an ordering of where to intervene first, not as predictions of the return to intervention. Within that scope, the policy implications are pointed. Institutional trust and governance quality occupies the most actionable position in the importance–influence map, combining above-average importance with the second-largest net influence and the highest total influence exerted on other criteria, 7.395. Measures directed at the rule of law, corruption control, administrative predictability, and the reliability of contract enforcement are not adjacent to talent policy but constitutive of it. Migration policy, meanwhile, presents the sharpest contrast: the panel assigns it the lowest stated importance, yet it exerts the largest net influence. The most plausible reading is that experts evaluate migration policy as a precondition rather than an attraction — visa regimes and residence rules are not what draw a skilled individual to a country, but they determine whether the other determinants can operate at all. Its low standing in stated preferences should be read as a measure of its invisibility to those it affects rather than of its inefficacy. Unlike institutional quality, migration policy is a fast-moving instrument that can be redesigned within a legislative cycle, though whether a redesign realizes the leverage the panel attributes to it depends on policy quality, administrative capacity, and political feasibility.

The robustness analysis is unusually thorough for a study of this kind. Replacing the consensus-based weights with uniform weights leaves the results essentially intact: the Spearman correlation between the two SF-AHP weight vectors is 0.943, with a single adjacent reversal involving the near-tied pair of work–life balance and institutional trust. On the influence side, the prominence ranking is identical and the cause–effect classification changes only for cultural compatibility, whose net effect moves from −0.015 to +0.104 — both within rounding distance of zero. Repeating the computations with a conservative anchoring constant changes no classification and no prominence rank. A leave-one-out analysis, recomputing the entire pipeline ten times with one panelist omitted each time, shows that economic attractiveness is ranked first in all ten runs, while institutional trust and migration policies constitute the cause group in all ten runs and migration policies carry the largest net influence in nine of them. The causal structure proves markedly more stable than the importance ranking. A further check partitions the panel by competence: five panelists hold disciplinary expertise in political science and five come from medicine, engineering, and human-resource management. Running the influence stage separately on each stratum returns the same cause group in both. The influence-based weights obtained through the DEMATEL-based Analytic Network Process also correlate strongly with the preference-based weights, at 0.943, so the reported ranking is reproducible from a derivation that does not pass through the pairwise preference judgments at all.

One qualification deserves attention. Agreement on the importance ranking itself is limited: Kendall’s W across the individual importance orderings equals 0.168 and does not reach conventional significance, while agreement in the influence stage is more than twice as high and significant at 0.398. The authors show that this heterogeneity is structured rather than diffuse — panelists resident in EU settlement regimes display substantial and significant agreement at W = 0.677, while the pooled figure results from combining internally coherent groups ordered differently. The aggregate should therefore be read as the center of a genuinely dispersed set of views, and the causal structure carries the more robust part of the message. The panel is purposive rather than representative, the analysis is framed at a general level without distinguishing destination contexts, and the cause–effect classification expresses net perceived influence rather than econometric identification. Replication with larger panels stratified so that destination regime and disciplinary background are crossed rather than aligned would sharpen the picture considerably.

What the study ultimately contributes is a framework that separates what matters from what moves the system, converting a ranking into a policy sequence. A study that measured importance alone would rank economic attractiveness first and stop; the two-stage design shows that intervening there addresses the symptom. The three determinants in the effect group — career opportunities, which absorb more influence than any other criterion at 7.211, work–life balance, and economic attractiveness — merit systematic monitoring rather than direct targeting, since they register whether interventions elsewhere in the system are working. Cultural compatibility, near-neutral in influence and last in prominence, is better understood as a buffer that conditions the cost of adaptation after the decision to move has been taken. For governments competing in the global talent market, the message is uncomfortable but actionable: the fastest way to improve the economic conditions that skilled migrants say they want is to strengthen the institutions and policies that make those conditions possible in the first place.

Subject of Research: Decision Science and Migration Policy

Article Title: Determinants of international talent attraction: an integrated spherical fuzzy DEMATEL–AHP framework with stage-specific expert weighting

Article References: Olgun, E., & Zaralioǧlu, F. B. (2026). Determinants of international talent attraction: an integrated spherical fuzzy DEMATEL–AHP framework with stage-specific expert weighting. Frontiers in Political Science, 8, Article 1959067. https://doi.org/10.3389/fpos.2026.1959067

Image Credits: AI Generated

DOI: 10.3389/fpos.2026.1959067

Keywords: talent attraction, skilled migration, spherical fuzzy sets, DEMATEL, AHP, multi-criteria decision making, institutional trust, migration policy, Determinants, international, talent, attraction