EEFS Conference 2026 — Seville

Productivity Effects of Hiring Foreign Experts

Evidence from Swedish Matched Employer-Employee Data (1996–2015)

Daniel Halvarsson

The Ratio Institute, Stockholm | daniel.halvarsson@ratio.se

Background & Motivation

"Attracting and retaining talent is essential for innovation, economic growth, and competitiveness. Countries that do not join the global competition for highly skilled workers risk falling behind."

— OECD, Talent Attractiveness (2023)

The dominating narrative in the policy debate have resulted in favorable visa regimes, tax incentives, and fast-track permits.

The argument goes:

High-skilled migration drives innovation (patents, R&D)
It alleviates talent shortages in knowledge-intensive sectors
It fosters international networks and technology transfer

What about productivity?

Does hiring foreign experts actually translate into productivity gains inside the firm? What does the evidence say?

What Do We Know? The research Gap

1. Aggregate & Regional Studies

Large literature links migration to innovation, exports, and wage outcomes (Hunt & Gauthier-Loiselle, 2010; Kerr & Lincoln, 2010; Hatzigeorgiou & Lodefalk, 2015).
Foreign-born and STEM workers raise productivity and wages at the city/state level (Peri, 2012; Peri et al., 2015).

2. Firm-Level — But Not TFP

Swiss reform: positive effects on firm sales but mixed on labor productivity (Ruffner & Siegenthaler, 2017).
Danish experts: positive coworker wage effects, but wages proxy for productivity (Malchow-Møller et al., 2019).

3. Total factor productivity

Mitaritonna et al. (2017): skilled migrants raise TFP in French manufacturing. But limited to average regional effects and uses the STEM definitions.

"Less work, until recently, has examined whether these linkages translate to the firm setting: does hiring more skilled immigrants improve firm-level innovation, productivity, and performance?"

— Glennon (2024, Journal of Economic Perspectives)

This paper: Estimates the direct effect of hiring foreign experts on firm TFP, using rich Swedish employer-employee data using an event-study designs.

Conceptual Framework: How Can Foreign Experts Affect the Firm?

Y_{it} = \underbrace{A_{it}}_{\text{TFP}} \; \Big(\, s_{it} \; F\!\big(\underbrace{I_{it}}_{\text{domestic}},\; \underbrace{U_{it}}_{\text{foreign}}\big) \Big)^{\alpha} \; K_{it}^{\,\beta}

$Y_{it}$ = value added; $s_{it}$ = average skill level; $F(I_{it}, U_{it})$ = labor aggregate of domestic and foreign workers; $K_{it}$ = capital stock; $A_{it}$ = TFP (Ruffner & Siegenthaler, 2017).

Hiring foreign experts ($\uparrow U_{it}$) can affect the firm through four channels:

1. Labor input $F(I_{it}, U_{it})$

More workers $\Rightarrow$ more output (provided foreign labor is not a perfect substitute for domestic workers). If substitutes, crowding out (Borjas & Doran, 2012). If complements, positive growth effects.

2. Skill composition $s_{it}$

Hiring highly qualified experts raises the average skill level of the workforce, increasing effective labor input beyond the number of employees.

3. Capital $K_{it}$

Indirect effects through complementarity with existing machinery and equipment, or direct effects if foreign labor attracts foreign investment.

4. TFP $A_{it}$ ← this paper's focus

The residual: international networks, management & organization, brand, technology adoption, absorptive capacity. These cannot be attributed to labor or capital alone.

Definition of Foreign Experts

Foreign workers refers to labor outside Sweden that migrates to Sweden for work, including Swedish returnees and migrants from countries both inside and outside the EU/EEA.

A foreign worker is classified as an expert if:

They have employment in the same year as their most recent year of arrival to Sweden, and
their reported average monthly income or salary at some point during employment exceeds two price base amounts (approx. 89,000 SEK / ~$10,500 USD in 2015).

The income threshold corresponds to the level at which foreign workers automatically qualify for the Swedish Expert Tax relief (comparable to Malchow-Møller et al., 2019, for Danish experts).

Why an income-based definition?

Captures a broader group than STEM-only definitions: includes managers, business leaders, and other key personnel — not just scientists and engineers.
This is important because foreign experts may influence firm performance not only through technical knowledge, but also through management, organization, and the implementation of new business practices.
These broader channels are likely reflected in TFP rather than innovation metrics alone.

Data: Swedish Matched Employer-Employee Registers

The analysis draws on linked administrative microdata from Statistics Sweden (SCB), covering the full Swedish private sector (excl. financial) from 1996 to 2015.

Individual-Level (LISA + Wage Statistics)
Age, gender, region of birth
Year of arrival (most recent migration)
Education level
Annual wage income (full population)
Monthly salary (stratified sample)
Occupation (SSYK codes)
Employer linkage (workplace ID)

Firm-Level (FDB — "Företagens ekonomi")
Value added, sales
Number of employees (by skill level)
Capital stock (tangible assets)
Intermediate inputs (materials)
Industry affiliation (2-digit SNI)
Export status

Key advantage: The employer-employee link allows us to track exactly which firm hires which foreign expert, when they arrive, and to observe the firm's full production accounts (value added, capital, labor by skill) before and after the hire — enabling TFP estimation and DiD analysis at the firm level.

Profile of Foreign Experts (1996–2015)

The definition isolates 2,329 unique foreign experts distributed across 1,278 private-sector firms.

Demographics: Mean age of 41.7 years; 10% are female.
Global Pool vs. Returnees: 53.1% are Swedish-born return migrants. The remaining 47% comprise arrivals from the EU27, Nordic neighbors, Asia, and North America.
Occupational Density: Concentrated in leadership and professional roles:
- Managerial Categories: 55.8%
- Advanced Higher Education Roles: 21.8%

Top Sectors Employing Experts	%
Wholesale Trade (excl. motor vehicles)	18.7%
Computer Programming & IT Consultancy	9.7%
Head Office Activities & Business Consulting	6.7%
Manufacture of Machinery & Equipment	4.5%
Manufacture of Motor Vehicles	3.6%

Treated Firms and Comparison Group

Treatment group: Firms that for the first time hire a foreign expert directly from abroad, 2001–2015 (1,278 firms).

Comparison group: Firms in the same 2-digit sectors that never hire a foreign expert during the period.

A selection problem: Firms hiring experts are on average much larger, more capital-intensive, more internationalized, and already more productive:

Characteristic (pre-hiring)	Treated	Other firms
Value Added (1,000 SEK)	163,313	9,997
Employees	226	16
Exporting	69%	21%
Share domestic experts	3.1%	0.4%

Is this a problem? Not necessarily. Higher productivity levels in treated firms are fine — the DiD approach compares changes in productivity around the hiring event. The critical assumption is parallel trends: absent the hire, productivity would have evolved similarly in both groups.

To make this assumption more plausible, the model includes:

Firm and year fixed effects — absorb all time-invariant firm characteristics and common shocks.
Matching controls (predetermined at $T_0 - 1$):
- Size class (Eurostat: micro / small / medium / large)
- Export status
- Proportion of domestic experts
- Growth in productivity and average labor costs ($T_0-2$ to $T_0-1$)

Empirical Methodology

Step 1: Estimating firm-level TFP

Starting from a Cobb-Douglas production function for value added (in logs):

y_{it} = \beta_0 + \beta_K k_{it} + \beta_L l_{it} + \beta_H h_{it} + \underbrace{\omega_{it} + \eta_{it}}_{a_{it}}

where $l_{it}$, $h_{it}$ = low- and high-skilled labor; $k_{it}$ = capital; $\omega_{it}$ = productivity observed by the firm; $\eta_{it}$ = unanticipated shocks.

Problem: OLS is biased because firms observe $\omega_{it}$ and adjust inputs accordingly (simultaneity).

Solution: Proxy variable methods (Wooldridge, 2009) — use intermediate inputs (materials) to proxy for $\omega_{it}$, estimated separately by 2-digit sector. TFP is then the residual: $\hat{a}_{it} = y_{it} - \hat{\beta}_K k_{it} - \hat{\beta}_L l_{it} - \hat{\beta}_H h_{it}$.

Step 2: Average treatment effect of the treated (conditional) — Local Projections DiD

Firms hire experts in different years (staggered treatment). Standard TWFE event studies can produce biased estimates when treatment effects vary across cohorts.

lpdid (Dube et al., 2023) avoids this by running separate cross-sectional regressions for each horizon $h$:

\hat{a}_{i,t+h} - \hat{a}_{i,t-1} = \alpha + \delta_t + \beta_h^{lpdid}\,\Delta D_{it} + \gamma\, x_{i,t-1} + \varepsilon_{it}

where $\Delta D_{it} = 1$ for newly treated firms and 0 for not-yet-treated or never-treated firms.

Intuition: By long-differencing ($t-1$ to $t+h$), firm fixed effects drop out. Each horizon $h$ uses a clean comparison — only firms that have not yet been treated serve as controls, avoiding the "bad comparison" problem of TWFE. The coefficient $\beta_h^{lpdid}$ traces out the dynamic treatment effect.

Results: Effect on Firm TFP

6% – 11%

TFP increase within 2–3 years of hiring

The effect emerges with a short lag and peaks at +11.6% by $T_0+3$ ($p < 0.001$).
The effect persists: +10.8% at $T_0+5$.
Pre-trends: Estimates for $T_0-5$ through $T_0-2$ are statistically indistinguishable from zero ($p \approx 0.14$), supporting the parallel trends assumption.

For comparison, average annual TFP growth in Swedish firms has been around 2% per year (Ekonomifakta).

Event time	(1) No controls	(3) Preferred
$T_0-5$	0.033 (0.022)	0.011 (0.022)
$T_0-4$	0.043* (0.021)	0.025 (0.021)
$T_0-3$	0.019 (0.020)	0.004 (0.020)
$T_0-2$	0.027 (0.017)	0.019 (0.016)
$T_0-1$	reference period
$T_0$	-0.020 (0.017)	0.004 (0.016)
$T_0+1$	0.006 (0.019)	0.038* (0.019)
$T_0+2$	0.029 (0.020)	0.076* (0.020)**
$T_0+3$	0.065** (0.021)	0.116* (0.021)**
$T_0+4$	0.052* (0.023)	0.100* (0.024)**
$T_0+5$	0.053* (0.026)	0.108* (0.026)**

SE clustered at firm level. * $p<0.05$, ** $p<0.01$, *** $p<0.001$. Model (3) adds size, export, domestic expert share, and wage growth controls.

The Effect Is Driven by Small and Medium-Sized Firms

Restricting the sample to SMEs (< 250 employees) strengthens the TFP effect considerably: +14.1% at $T_0+3$.
For large firms, the estimates are small and statistically insignificant.

In a smaller firm, a single expert has more leverage to reshape organization, introduce new technologies, and open international networks. In large firms, the marginal impact of one hire is diluted across hundreds of employees.

Stronger Effects When Excluding Swedish Returnees

Over half (53%) of the experts are Swedish-born individuals returning from abroad. When we exclude them:

The TFP effect increases — reaching +14.6% by $T_0+5$.
Suggests that foreign-born experts do bring knowledge, networks, and organizational perspectives that affect the firm's productivity.
Returning Swedes don't drive the result.

Effects on Coworker Incomes

Does the hiring of a foreign expert benefit the incumbent workforce?

We track log average income of all employees in the firm excluding the expert.
A positive effect of 1.5–2% emerges from $T_0+3$ onward, coinciding with the timing of the TFP gains.
The wage effects are less robust, though, across specifications and should be interpreted with caution.

Consistent with knowledge spillovers raising coworker productivity, and rent-sharing as firms become more profitable.

Conclusion & Policy Implications

Key Conclusions:

Hiring top-tier foreign experts generates large, sustainable firm-level productivity gains (+6% to +11%).
The productivity gains pertains to primarily SMEs.
Gains extend beyond TFP as both labor efficiency and physical capital utilization increase along with productivity.

Policy Relevance:

Supports the case for targeted fiscal incentives (e.g. the Swedish Expert Tax) as channels for productivity growth.
That access to mobile global talent is a driver of firm performance in the treated group of firms.

Policy impact: Cited in SOU 2025:3 "Skatteincitament för forskning och utveckling" — the Swedish Government inquiry reviewing the Expert Tax rules. The study's findings on the importance of international competence for Swedish productivity were used to inform proposed reforms.

Thank you!

Questions & comments welcome.