Effects of the enhanced nurse licensure compact on interstate mobility and nursing employment

Abstract

Introduction

The United States continues to face persistent nursing shortages, even as many licensed nurses are not employed in nursing roles. The enhanced Nurse Licensure Compact (NLC 2.0), a mutual licensure recognition agreement, is promoted as a strategy to increase workforce supply and flexibility, yet no prior research has assessed whether it helps licensed but nonpracticing nurses return to the profession.

Methods

Using American Community Survey data (2010-2023) and state-level licensure and employment data (2015-2023), we evaluated whether NLC 2.0 adoption is associated with changes in interstate registered nurse (RN) inflows and in the license-employment mismatch—the share of active licenses not linked to payroll nursing employment—across 26 states.

Results

We found no statistically significant effect on interstate migration (+0.27 percentage points; P = 0.67). However, adoption was associated with a 3.30-percentage-point reduction in the share of licensed nurses not employed in nursing (P = 0.054), with the effect strengthening over time. A staggered difference-in-differences robustness check confirmed this pattern.

Conclusion

These findings suggest that while the compact does not draw nurses across state lines, it is associated with more licensed nurses working in nursing within adopting states. Licensure reform may therefore be one component of a broader strategy to support nurses' return to bedside roles.

Key Points.

• NLC 2.0 adoption was not associated with increased interstate nurse migration, suggesting that licensing barriers are not the primary driver of geographic redistribution.

• Compact adoption was associated with a 3.30-percentage-point reduction in the share of licensed nurses not employed in nursing, with effects growing over time.

• Licensure reform may be most effective at activating nurses already residing in adopting states rather than attracting nurses from other states and should be paired with broader workforce retention strategies.

Introduction

The United States faces a persistent nursing workforce challenge. The Bureau of Labor Statistics projects a need for more than 200 000 new registered nurses annually through 2032, yet many states report chronic vacancies and difficulty retaining bedside staff.^1,2 The 2022 National Nursing Workforce Survey found that approximately 100 000 registered nurses left the workforce during the COVID-19 pandemic, and nearly 900 000, roughly one-fifth of the 4.5 million registerd nurse (RN) workforce, reported intent to leave by 2027, citing stress, burnout, and retirement.³ The challenge, however, is not solely producing more nurses: we know that a substantial pool of nurses holds active licenses but is no longer employed in nursing and that nurses' geographic distribution does not match the distribution of patient need.⁴ Addressing both the problem of inactive nurses and their geographic distribution across states is therefore an important policy priority. In this study, we use a two-way fixed-effects (TWFE) difference-in-differences model to determine whether a national licensure compact could help alleviate either or both issues.

State-based licensure regulation, which was designed for a workforce that practices within a single jurisdiction, has become increasingly misaligned with care-delivery arrangements that span state boundaries, including telehealth, multistate health systems, and cross-border commuting.⁵ Occupational licensing more broadly has been shown to reduce labor mobility and create barriers to employment, particularly when workers must obtain new credentials after relocating across state lines.^6,7

In response, there is a growing trend among health care professions to advance multistate compacts. Specifically in nursing, the National Council of State Boards of Nursing developed the Nurse Licensure Compact (NLC), a mutual-recognition model that allows a nurse licensed in a participating home state to practice in other compact states without obtaining additional licenses.⁵

The enhanced compact (NLC 2.0), enacted in 2018, introduced uniform licensure requirements to address barriers that had stalled adoption of the original version.⁸ As of 2024, 41 states participate in the compact, reflecting widespread policy interest in licensure portability as a tool for addressing workforce shortages. While the compact was originally designed to facilitate temporary cross-border practice (eg, telehealth, travel nursing, border-community care), subsequent advocacy has increasingly emphasized permanent migration as an expected benefit—creating a tension between design and expectation that our analysis directly addresses.⁹

Prior research on the NLC has focused primarily on interstate nurse mobility, finding mixed results. DePasquale and Stange conducted the most rigorous study to date using a difference-in-differences design.¹⁰ They found limited effects on interstate migration using broad population-based measures, though their analysis focused on the first version of the compact (NLC 1.0), which predated the improvements contained in NLC 2.0.¹⁰ Using a difference-in-differences design, Ghani found an approximately 11% increase in interstate job flows between compact states during NLC 1.0.¹¹ Shakya, Ghosh, and Norris, using regression analysis of longitudinal National Provider Identifier records that track nurse practice locations, also reported positive mobility effects within compact states during the NLC 2.0 period, particularly for compact-to-compact migration.¹²

Although focused on the first version of the compact, Portlock examined an interesting subgroup of nurses using American Community Survey (ACS) data through 2015. They exploited the exogenous nature of military relocations to show that nurses whose spouses were reassigned between compact states experienced significantly fewer departures from the labor force than those relocated to non-compact states.¹³

Other work has examined the compact's influence on telehealth utilization, finding no significant effect on out-of-state telemedicine use,¹⁴ and has documented regulatory responses during COVID-19, including emergency licensure waivers.^15,16 Zhong, Martin, and Alexander examined potential safety concerns and found that multistate license holders exhibit lower discipline rates than single-state holders, suggesting the compact likely does not lower quality of care.¹⁷

Missing from the literature is research assessing whether adopting compacts could draw licensed nurses who are not currently employed in nursing back into nursing. Given the high proportion of nurses not working in nursing jobs, which, using the methods detailed below, we found to be almost 45% nationwide in 2023, this is a potentially important question. If the compact's adoption indeed helps bring licensed nurses back into the workforce, it could help address the nursing shortage. Whereas movement across states, to the extent that it occurs, creates winners and losers, the return of inactive nurses does not detract from another state's supply.

This study contributes to existing literature by examining both major state-level outcomes for first-time NLC 2.0 adopters. These outcomes are (1) interstate RN inflow, which captures the net change in state-level RN supply from interstate movement, and (2) the license-employment mismatch. We also describe state-level variation in the magnitude of these associations.¹⁸ The study also builds on the methodological strengths of prior difference-in-difference analyses by updating DePasquale and Stange for compact 2.0¹⁰ and by measuring actual residential moves rather than changes in provider registration records in the case of Shakya, Ghosh, and Norris for 2.0.^10,12

Methods

Conceptual framework

The mechanism by which NLC 2.0 adoption could affect either nurse supply outcome (increased interstate migration inflow or a reduction in the share of licensed nurses not employed in nursing jobs) is the same in both cases. Presumably, it reduces the administrative steps required to practice in a different state than the one in which a nurse obtained their original license.⁵

Most research has focused on interstate migration for NLC 1.0. Apart from differences with the enhanced 2.0 compact (changes intended to streamline the process further), it is important to recognize that many major forces driving migration could overshadow this effect. These include housing costs, family constraints, and local labor demand and wages.¹⁰

Similarly, many factors influence whether licensed nurses choose to work in nursing, including, importantly, working conditions, wages, and family situations. However, this second potential effect does not require geographic relocation and thus could be more sensitive to the compact. In particular, it may be attractive to nurses who moved to a new state for other reasons, eg, family-related reasons, and now must decide whether to take a nursing job in the new state.

Data sources and study design

We constructed a state-year analytic panel from multiple data sources. For the interstate mobility outcome, we used the ACS 1-year Public Use Microdata Sample.¹⁹ For the license-employment mismatch outcome, we combined active RN license counts from the National Council of State Boards of Nursing (NCSBN) National Nursing Database with employment counts from the Bureau of Labor Statistics (BLS) Occupational Employment and Wage Statistics (OEWS).^20,21 We drew state-year covariates from the Bureau of Economic Analysis Regional Price Parities (cost of living),²² the American Hospital Association Annual Survey (RN vacancy rates, lagged 1 year),²³ BLS OEWS (RN wages),²¹ state legislative records (right-to-work status), US Census Bureau population estimates (population change),²⁴ and CDC hospitalization data (COVID-19 inpatient burden).²⁵ We coded NLC 2.0 adoption timing from NCSBN compact implementation documents using the first full implementation year for each state.²⁰

We identified registered nurses in the ACS using Census occupational codes and restricted the sample to individuals in the labor force. We excluded respondents with unknown migration status and those who moved from abroad and applied person weights adjusted for pooled annual data. The analytic window spans 2010 through 2023. The ACS is a sample survey, and the number of unweighted RN respondents per state-year is small in less populous states. For example, Wyoming and Vermont have fewer than 100 respondents per year, compared with more than 2000 in Florida and New York (see Table SA1).

The regression models use 364 state-year observations in the baseline specification and 310 in the controlled specification.

For Outcome 2, we excluded traditional NLC 1.0 states and used data from 2015 through 2023. The baseline model includes 234 state-year observations, and the controlled model includes 206 observations across 26 states (12 treated, 14 comparison).

We defined treated states as those that adopted NLC 2.0 but were never members of the original compact. The treatment indicator equals one in years at or after a state's first full implementation year. We excluded 25 legacy NLC states. Figure 1 displays the adoption timeline, while Table SA2 documents the coding rules, sample counts, and event-time support.

Figure 1.

NLC 2.0 Adoption by State and Year. Source: Authors' analysis of National Council of State Boards of Nursing compact implementation documents and study analytic files. Note: Panel A classifies states as first-time NLC 2.0 adopters (treated), non-adopters (comparison), or legacy NLC states (excluded). Panel B reports the cumulative number of NLC 2.0 participating states by first full implementation year, disaggregated by legacy NLC states that transitioned and first-time adopters.

Outcome measures

The first outcome is interstate RN inflow share, defined as

$$InflowShare=\frac{\sum w\cdot 1[movedin]}{\sum w\cdot 1[employedinstates]}$$

where w is the ACS person weight adjusted for pooled annual data, the numerator counts RNs employed in state s in year t who resided in a different state 1 year earlier, and the denominator counts all RNs employed in state s in year t. This captures the share of a state's nursing workforce that moved from another state in the prior year.

The second outcome is the share of active RN licenses not matched to observed payroll nursing employment, defined as

$$\mathrm{NEN}\text{-}\mathrm{RN}=1-\frac{\mathrm{Employment}}{\mathrm{Active}\mathrm{Licenses}}$$

where Employment is the BLS OEWS count of registered nurses, nurse practitioners, nurse anesthetists, and nurse midwives employed in states in year t, and active licenses is the NCSBN count of active RN licenses in that state-year.^20,21 We included these advanced practice roles because they require an active RN license. The denominator counts licenses rather than unique individuals, and the numerator counts wage-and-salary workers in nonfarm establishments. This measure captures a state-level gap between licensure and employment. It is not a person-level indicator of unemployment or labor-force participation. Table SA3 provides the full construction methodology.

Empirical strategy

For both outcomes, the primary specification is a TWFE difference-in-differences model:

$$Y_{st}={\alpha }_s+{\lambda }_t+\beta \cdot \mathrm{NLC}2\mathrm{pos}{\mathrm{t}}_{st}+{\epsilon }_{st}$$

where ${\alpha }_s$ represents state fixed effects, ${\lambda }_t$ represents year fixed effects, and $\mathrm{NLC}2\mathrm{pos}{\mathrm{t}}_{st}$ equals 1 if the state has implemented NLC 2.0 by year t. Standard errors are clustered at the state level.²⁶

The controlled model adds time-varying state covariates:

$$Y_{st}={\alpha }_s+{\lambda }_t+\beta \cdot \mathrm{NLC}2\mathrm{pos}{\mathrm{t}}_{st}+X_{st}^{\mathrm{\prime }}\theta +{\epsilon }_{st}$$

For Outcome 1, ${\mathit{X}}_{\mathit{s}\mathit{t}}^{\mathrm{\prime }}$ includes RN wages, cost of living, RN vacancy rates, COVID-19 inpatient burden, right-to-work status, and population change. For Outcome 2, ${\mathit{X}}_{\mathit{s}\mathit{t}}^{\mathrm{\prime }}$ includes the cost of living, RN vacancy rate, COVID-19 inpatient burden, and right-to-work status. The covariate sets differ due to differences in data availability between the two outcome panels. Because Outcome 1 is a share bounded between zero and one, we also estimate a fractional-logit model as a robustness check and report average marginal effects.²⁷

To assess the parallel trends assumption and examine dynamic effects, we estimate event-study specifications:

$$Y_{st}={\alpha }_s+{\lambda }_t+\sum _k{\delta }_k\cdot D_{st}^k+X_{st}^{\mathrm{\prime }}\theta +{\epsilon }_{st}$$

where $D_{st}^k$ equals one if state s is k years relative to its adoption year. The omitted reference period is t = −1. The event window spans −5 to +5 years. The main text presents controlled specifications (Figure 2); raw specifications are reported in Table SA4.

Figure 2.

Dynamic Effects of NLC 2.0 Adoption. Source: Authors' analysis of ACS RN-only analytic files, NCSBN licensure data, BLS OEWS data, AHA Annual Survey nurse vacancy data, and linked state-year policy and market covariates. Note: Panel A shows controlled event-study estimates for interstate RN inflow share. Panel B shows controlled event-study estimates for license-employment mismatch. Coefficients are reported in percentage points. The omitted reference period is t = −1. Points show coefficient estimates, and whiskers show 95% confidence intervals.

NLC 2.0 adoption is staggered across six cohorts (2018-2023), raising concerns about potential bias in conventional TWFE estimators when treatment effects are heterogeneous across cohorts or grow over time.^28-30 As a robustness check, we also implement the Callaway and Sant’Anna estimator, which uses only never-treated states as the comparison group.²⁹

NLC 2.0 adoption is not randomly assigned and may correlate with state political or labor-market characteristics. State fixed effects do not absorb the effects of concurrent policies, such as Medicaid expansion.

To examine state-level variation in the treatment effect, we replace the single treatment indicator with 12 state-specific post-adoption indicators. This is a descriptive decomposition of the pooled estimate, allowing us to identify which states drive the overall result. Because this specification estimates 12 separate coefficients from 12 treated-state clusters, these state-specific estimates should be interpreted with caution, given limited post-adoption observations per state.

Results

Our analytic sample includes 26 states: 12 that adopted NLC 2.0 for the first time between 2018 and 2023 and 14 that did not adopt during the study period. Twenty-five legacy NLC states were excluded. Five states adopted in 2018, two in 2019, two in 2020, and one each in 2021, 2022, and 2023 (see Figure 1). Washington was adopted in 2024, after the end of the analytic period, and is classified as a comparison state.

Treated and comparison states differ on several dimensions (see Table 1; Table SA5 provides expanded statistics). Treated states have lower mean hourly RN wages ($32.21 vs $40.40; P < 0.001), lower cost of living (95.7 vs 103.9; P < 0.001), and a higher prevalence of right-to-work laws (65% vs 14%; P = 0.016). RN vacancy rates and COVID-19 inpatient burden do not differ significantly between groups. While treated states have somewhat lower wages and a lower cost of living than comparison states, these differences are controlled through state fixed effects and time-varying covariates. They are unlikely to bias the estimated treatment effects as they are time-invariant. The mean rate of licensed nurses not employed in nursing is 45.0% (SD = 8.81), indicating that nearly half of active RN licenses are not matched to payroll nursing employment. Several factors inflate this ratio: (1) nurses may hold active licenses in multiple states simultaneously, inflating the denominator; (2) retired nurses often maintain licenses for insurance or contingency purposes; (3) some licensed nurses work in occupations not captured by SOC 29-1141 (eg, nursing education, health administration); and (4) the OEWS establishment survey excludes self-employed nurses and those in nontraditional settings. The relevant finding is the change in this ratio following adoption, which is unaffected by these level differences as they are time-invariant within states.

Table 1.

Descriptive characteristics of the analytic sample.

Panel A. Interstate RN inflow
Variable	Full sample	Treated states	Comparison states
State-year observations	310	155	155
States	26	12	14
RN inflow share (%)	3.08 (3.03)	2.71 (1.93)	3.44 (3.80)
RN employment (weighted N)	4626.6 (4325.7)	3977.1 (3286.0)	5276.0 (5089.0)
RN hourly wage ($)	36.31 (6.69)	32.21 (4.46)	40.40 (6.01)
Cost of living (RPP)	99.80 (7.34)	95.74 (6.35)	103.9 (5.87)
RN vacancy (AHA)	45.35 (35.71)	42.41 (29.87)	48.29 (40.61)
Population change (%)	0.52 (0.91)	0.55 (0.81)	0.49 (1.00)
COVID inpatient rate (%)	2.31 (4.30)	2.34 (4.37)	2.29 (4.24)
Right-to-work (indicator)	0.40 (0.49)	0.65 (0.48)	0.14 (0.35)
NLC 2.0 adopted	0.13 (0.34)	0.26 (0.44)	0.00 (0.00)

Panel B. License-employment mismatch
Variable	Full sample	Treated states	Comparison states
State-year observations	206	103	103
States	26	12	14
Licensed RNs not employed in nursing (%)	45.00 (8.81)	44.40 (8.11)	45.61 (9.46)
Licensed RNs employed in nursing (%)	55.00 (8.81)	55.60 (8.11)	54.39 (9.46)
RN employment (N)	4871.3 (4518.0)	4220.6 (3489.6)	5522.0 (5290.9)
RN annual wage ($)	78 542.4 (14 372.6)	69 525.2 (9381.8)	87 559.6 (12 764.6)
Cost of living (RPP)	99.78 (7.60)	95.66 (6.74)	103.9 (6.03)
RN vacancy (AHA)	55.37 (37.91)	52.98 (30.76)	57.75 (43.94)
COVID inpatient rate (%)	3.48 (4.87)	3.52 (4.96)	3.44 (4.81)
Right-to-work (indicator)	0.42 (0.49)	0.68 (0.47)	0.16 (0.36)
NLC 2.0 adopted	0.20 (0.40)	0.40 (0.49)	0.00 (0.00)

Authors' analysis of ACS registered nurse (RN) only analytic files, NCSBN licensure data, BLS OEWS data, AHA Annual Survey nurse vacancy data, and linked state-year policy and market covariates. Panel A reports descriptive statistics for the interstate RN inflow sample. Panel B reports descriptive statistics for the license-employment mismatch sample. Treated states are first-time Nurse Licensure Compact (NLC) 2.0 adopters. Comparison states are nonlegacy states that did not adopt during the analytic period. Continuous variables are reported as mean (SD). Right-to-work and enhanced compact (NLC 2.0) adoption are indicators.

We found no statistically significant effect of NLC 2.0 adoption on the interstate RN inflow share (see Table 2), although the trend was positive. The estimated effect is +0.18 percentage points in the baseline model (SE = 0.56; P = 0.76; N = 364) and +0.27 percentage points in the controlled model (SE = 0.61; P = 0.67; N = 310). Fractional-logit average marginal effects are comparable, ranging from +0.20 to +0.22 percentage points (Table SA6).

Table 2.

Main difference-in-differences estimates.

	Outcome 1. Interstate RN inflow share		Outcome 2. License-employment mismatch
Variable	(1) Baseline	(2) Controlled	(3) Baseline	(4) Controlled
NLC 2.0 post	0.18	0.27	−5.32b	−3.30a
	(0.56)	(0.61)	(2.04)	(1.63)
State fixed effects	✓	✓	✓	✓
Year fixed effects	✓	✓	✓	✓
Covariates		✓		✓
Observations	364	310	234	206
States	26	26	26	26

Authors' analysis of ACS RN-only analytic files, NCSBN licensure data, BLS OEWS data, AHA Annual Survey nurse vacancy data, and linked state-year policy and market covariates. Columns 1 and 2 report estimates for Outcome 1, interstate RN inflow share. Columns 3 and 4 report estimates for Outcome 2, license-employment mismatch. All coefficients are reported in percentage points. Standard errors clustered by state are shown in parentheses. All models include state and year fixed effects. Controlled models add the listed state-year covariates.

^aP < 0.10.

^bP < 0.05.

The event-study estimates do not show a significant shift toward higher inflow shares after adoption (see Figure 2, Panel A). Pre-period coefficients range from −0.10 to +1.08 percentage points, with confidence intervals spanning zero. Post-adoption estimates are similarly imprecise and do not trace a sustained upward trajectory.

In contrast, we found that NLC 2.0 adoption is associated with a statistically significant reduction in the share of licensed nurses not employed in nursing (see Table 2). In the baseline model, the estimate is −5.32 percentage points (SE = 2.04; P = 0.015; N = 234). In the controlled model, the estimate is attenuated to −3.30 percentage points but remains significant (SE = 1.63; P = 0.054; N = 206).

The event-study estimates show flat pre-trends and progressively larger post-adoption reductions (see Figure 2, Panel B). Pre-adoption coefficients range from −2.37 to −1.13 percentage points and are statistically indistinguishable from zero (all P > 0.20). Post-adoption estimates turn more negative over time: −3.48 percentage points at t = +1 (P = 0.017), −6.13 at t = +2 (P = 0.017), −10.03 at t = +3 (P = 0.004), and −12.83 at t = +4 (P < 0.001). This gradual pattern is consistent with a sustained change rather than a transient shock at the time of adoption. However, limited post-periods for later-adopting cohorts warrant caution in interpreting this dynamic pattern; the staggered adoption design means that longer-horizon estimates are identified primarily from earlier adopters.

As a robustness check for staggered treatment timing, the Callaway and Sant’Anna estimator yields an overall estimate of −8.43 percentage points (SE = 1.98; P < 0.001) with a consistent event-study pattern (Table SA7). The larger magnitude is consistent with the attenuation expected under conventional TWFE when treatment effects grow over time.²⁸

At a baseline rate of 45.0%, the controlled estimate of −3.30 percentage points corresponds to a 7.3% relative reduction and an increase in the share of licensed nurses employed in nursing from 55.0% to 58.3%.

Of the 12 treated states, nine show reductions after adoption. The largest reductions occur in Wyoming (−19.8 pp; P < 0.001), Oklahoma (−15.8 pp; P < 0.001), Vermont (−10.7 pp; P < 0.001), and Kansas (−9.6 pp; P < 0.001). Moderate reductions are observed in Georgia (−6.1 pp), Ohio (−4.7 pp), Florida (−3.8 pp), and West Virginia (−2.1 pp). After Benjamini–Hochberg correction, 9 of 12 estimates remain significant at q < 0.05 (Table SA8).

Three states show increases: Louisiana (+1.0 pp; not significant), New Jersey (+2.1 pp; P = 0.047), and Alabama (+5.9 pp; P < 0.001). Excluding Wyoming and West Virginia as a sensitivity check for small, unweighted ACS respondent counts does not materially change the pooled estimate (−4.74 vs −4.59 pp; Table SA9).

Limitations

There are several limitations to this study. First, the ACS is a sample survey, and the number of unweighted RN respondents per state-year is small in less populous states, fewer than 100 in Wyoming and Vermont, compared with several thousand in larger states. This could affect the precision of the interstate inflow estimates in those states. The absence of a statistically significant migration effect should not be interpreted as definitive evidence of no impact on interstate mobility; rather, any such effect may be too small or too diffuse across origin-destination pairs to detect with our aggregate state-level measure. Table SA1 reports unweighted respondent counts by state and year.

Second, the license-employment mismatch measure is an aggregate measure. The denominator counts active licenses rather than unique individuals, and the employment data capture payroll positions in nonfarm establishments. Nurses working as independent contractors, in self-employment, or in community-based settings outside standard payroll systems are not counted.

Third, the study cannot observe individual nurse behavior within this black box. We cannot be certain whether the reduction in the mismatch between licensed and employed nurses reflects nurses reentering clinical practice, consolidating licenses, or taking cross-state positions to provide telehealth enabled by the compact. Results are simply state-level associations.

Fourth, adoption of NLC 2.0 is not randomly assigned and may correlate with unobserved state characteristics. Concurrent time-varying policies, including Medicaid expansion, are not absorbed by state fixed effects. Table SA9 documents the overlap between NLC 2.0 adoption timing and Medicaid expansion across treated states.

Fifth, conventional TWFE estimators may exhibit bias when treatment timing is staggered and treatment effects are heterogeneous.^28,29 As a robustness check, we implement the Callaway and Sant’Anna estimator, which uses only never-treated states as the comparison group. The resulting overall ATT estimate is −8.43 percentage points (SE = 1.98, P < 0.001), larger in magnitude than the conventional TWFE estimate, with a similar event-study pattern (Table SA7). The attenuation of the TWFE estimate relative to the staggered estimator is consistent with the direction of bias predicted by Goodman-Bacon when treatment effects grow over time.²⁸

Discussion

Our results suggest that NLC 2.0 adoption is associated with a reduction in the share of licensed nurses not employed in nursing, but not with increases in interstate RN inflows. The controlled estimate for license-employment mismatch (−3.30 percentage points) is consistent across specifications, and the event-study pattern shows flat pre-trends followed by progressively larger post-adoption reductions. While the interstate inflow estimates are in a positive direction (consistent with the expectation that reducing licensing barriers would make a state more attractive to nurses in other states), the estimates are not statistically significant and do not trace a sustained post-adoption shift.

This second finding is consistent with DePasquale and Stange's null migration finding, which concluded that there were limited effects of the original compact (NLC 1.0) on interstate mobility using broad population-based measures.¹⁰ It may reflect the persistent reality that housing costs, family ties, and local labor demand primarily shape interstate relocation.³¹ The positive mobility effects reported by Shakya, Ghosh, and Norris may reflect differences in the movement measure and study period, rather than a contradiction, since their analysis focused specifically on compact-to-compact migration among nurses who relocated.¹²

The return of inactive nurses to the workplace has not been examined before. Our results are consistent with the tied-mover interpretation advanced by Johnson and Kleiner (2020): if many nurses are trailing spouses whose relocation decisions are determined by household-level factors, then reducing licensure barriers would have a greater effect on labor market participation at the destination than on the migration decision itself.

One plausible explanation for our positive finding is that some licensed nurses who moved to a new state for personal or family reasons had not obtained a new state license and were therefore not practicing. Under the compact, these nurses can work using their existing multistate license without additional applications or fees.¹⁸

The gradual deepening of the effect over the post-adoption period is consistent with this interpretation: nurses may not reenter the workforce immediately upon adoption but may do so over several years as they become aware of the new option and as employers post new jobs. Our findings could also be partially explained by Portlock's findings, which suggested that nurses whose military spouses were relocated to compact states were more likely to remain in the workforce. This may mean that the compact is especially effective for nurses who relocate for family reasons.

It is important to acknowledge, again, that our aggregate measure of the share of licensed nurses not employed in nursing cannot isolate supply-side mobility responses from contemporaneous demand-side hiring adjustments that may also follow NLC adoption. Nor does our aggregate measure of interstate RN inflows differentiate telehealth arrangements, which enable cross-border practice without relocation. Lastly, neither measure detects compositional changes in the licensed workforce.¹³

The state-level heterogeneity results suggest that the largest reductions occurred in states with the highest baseline rates of licensed nurses not working in nursing, such as Wyoming, Oklahoma, Vermont, and Kansas. After false discovery rate correction, 9 of 12 state-specific estimates remain significant. States where mismatch increased after adoption (Alabama, New Jersey, Louisiana) may have experienced concurrent labor-market disruptions that offset any compact benefit.

From an external validity perspective, treated states (new NLC 2.0 adopters) tend to be lower-cost, right-to-work states with relatively lower baseline nursing wages. Thus, this selection may limit generalizability to higher-cost states or those with stronger union presence, where the effects of compact adoption on nurse labor supply may differ. Our estimates should be interpreted as local average treatment effects for the specific set of early adopters. Excluding Wyoming and West Virginia, two states with few unweighted ACS respondents per year, does not materially change the pooled estimate (−4.74 vs −4.59 pp; Table SA9).

Policy implications

These findings may have practical implications for states considering compact adoption. While statements by governors suggest that compact adoption would attract nurses to their states,^32,33 a more important mechanism may be activating licensed nurses already residing within their borders but not currently working in nursing. States with large pools of licensed but non-practicing nurses may therefore stand to benefit most from adoption.

These findings in no way suggest that licensure reform alone can resolve staffing shortages. Shortages are more likely a result of inadequate compensation, poor working conditions, or insufficient training capacity.^5,34 However, policies that pair licensure portability with reentry support, such as refresher programs, flexible scheduling, or return-to-practice initiatives, could amplify the compact's effects. States may also differ in the degree to which this effect would benefit them, given that the baseline share of nurses who are licensed but inactive varies dramatically.

Conclusion

This study finds that NLC 2.0 adoption is associated with reductions in the share of licensed nurses not employed in nursing but has almost no effect on interstate RN migration. The compact appears to function as a policy that lowers licensing-related barriers for nurses already residing in adopting states, rather than as a mechanism for geographic redistribution of the nursing workforce. These findings position licensure reform as one component of a broader workforce strategy. That strategy must also address the major drivers of shortages, such as the ongoing challenges relating to compensation, working conditions, and training pipelines.

Disclaimer

The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Supplementary material

Supplementary material is available at Health Affairs Scholar online.

Funding

This work was funded through a Cooperative Agreement for a Regional Center for Health Workforce Studies (U81HP26493) Health Resource and Services Administration.