Sign up

The rise and impact of nurse practitioners and physician assistants on their own and cross-occupation incomes.
Abstract:
There has been a dramatic increase in the authority granted to nurse practitioners (NP) and physician assistants (PA). This "expanded" authority has changed who can provide health-care services and has weakened the control physicians have traditionally held over the provision of medical services. These changes in regulation have varied by occupation, state, and year and provide variation that can be exploited to empirically measure the individual and collective impacts of changes in NP authority and PA authority on practitioner incomes. It is found that changes in NP and PA regulatory authority do impact the labor markets of all three practitioner categories. NPs having greater practice authority brings physician incomes down, has differential impacts on PA incomes, and improves their own earnings, other factors held constant. PAs having increased authority has a downward effect on NP earnings, a positive impact on physician income, and little impact on their own incomes. (JEL I18, J18, J44, H75)

Article Type:
Report
Subject:
Nurse practitioners (Compensation and benefits)
Nurse practitioners (Practice)
Nurse practitioners (Influence)
Physicians' assistants (Practice)
Physicians' assistants (Compensation and benefits)
Physicians' assistants (Influence)
Caregivers (Practice)
Author:
Perry, John J.
Pub Date:
10/01/2009
Publication:
Name: Contemporary Economic Policy Publisher: Western Economic Association International Audience: Academic; Trade Format: Magazine/Journal Subject: Business; Economics Copyright: COPYRIGHT 2009 Western Economic Association International ISSN: 1074-3529
Issue:
Date: Oct, 2009 Source Volume: 27 Source Issue: 4
Topic:
Event Code: 280 Personnel administration; 200 Management dynamics
Geographic:
Geographic Scope: United States Geographic Code: 1USA United States
Accession Number:
211959183
Full Text:
I. INTRODUCTION

A. Motivation

Health care is a large, dynamic, and growing segment of the economy. Of its many interesting facets is the fact that its occupations are nearly universally regulated. A mandatory condition to work in a regulated occupation is to hold the appropriate license or certification. Typically, these licenses not only authorize a practitioner to practice but also give parameters of that practice: what the occupation can and cannot do. Any change in these parameters has the potential of dramatically altering the occupational landscape as there are direct and indirect impacts on the occupation experiencing the change as well as on other related occupations. These impacts are of interest generally, but especially for an industry as large and prominent in public policy concern as health care.

There has been a dramatic change in occupational regulation in the health-care industry leading to nurse practitioners (NPs) and physician assistants (PAs) rising as primary health-care providers. Both NPs and PAs have experienced authority expansions at different times in different states that has resulted in their practice authority resembling that of traditional general care physicians. This change has gone largely uninvestigated in the literature. This lack of attention coupled with the importance of the health-care industry in which these changes have taken place motivates this research. It is found that the changes NPs and PAs have experienced have had economically significant impacts on their own and physician incomes.

B. Literature

There has been a significant amount of attention in the larger academic literature concerning NPs and PAs. Nearly all the developed literature can be grouped into three broad categories. The first traces the rise of NPs and PAs as primary caregivers, chronicling their histories and the role each plays in the medical care system today. (1)

A second category, and of particular importance to this research, is a literature on the quality of care provided by NPs and PAs as compared to that of physicians. The research consistently finds that care given by NPs, PAs, and physicians is generally indistinguishable, a result that is robust to how quality of care is measured. (2) In addition to objective quality of the care measures, the research investigating patient satisfaction finds that NP and PA care scores at least as high as that provided by a physician. (3)

A third branch of the literature has examined the impact of NP and PA regulatory changes on the populations of NPs and PAs. In 1994, Sekscenski et al. (1994) examined the relationship of state practice environments and the number of NPs, PAs, and nurse-midwives in a state. The authors constructed an "index" of state practice environments and found that higher index scores (more favorable practice environment) were positively correlated with the relative number of NPs, PAs, and nurse-midwives.

Wing et al. (2004) expanded on Sekscenski et al. (1994) and looked at the relationship of state practice environments and provider populations for years 1992 and 2000. They used the basic framework developed by Sekscenski et al. (1994) but made some modifications to the practice index value. They found that the differences between states' index values narrowed over the time period, implying that practice environments became similar between states over time. They also found that provider populations were positively correlated with state practice environments. In related research using a more sophisticated, regression approach, Kalist and Spurr (2004) found that in states where NPs had greater practice authority, enrollments in masters nursing programs were higher. (4)

A fourth, undeveloped category of the literature, and where this research seeks to expand the body of knowledge, concerns the effect of NPs and PAs gaining greater practice authority on their own incomes as well as that of physicians. Dueker et al. (2005) made a first contribution to this question. They investigated the impact of advanced practice nurses (APNs) having expanded practice authority on the incomes of APNs, PAs, and physicians. (5) Using the Current Population Survey (CPS) from 1988 to 2002, the authors estimated that APNs having greater practice authority lead to lower incomes for APNs and higher incomes for PAs and found some evidence that it lowers physician incomes.

These findings, however, seem counterintuitive. That APNs having greater practice authority in a state leads to their own incomes falling and that of a substitute caregiver occupation's (PAs) increasing, even after addressing potential endogeneity issues, probably does not meet the prior expectation. The hypothesis provided by the authors is that as APNs have gained greater authority, the established physician population has tended to favor hiring PAs instead of APNs when their own medical practices make a nonphysician hire. This leads to the demand for APNs to presumably fall and that of PAs to increase. This explanation is given some plausibility when it is understood that in all states, PAs, as a condition to practice, must be employed by a supervising physician. There is no such general requirement that APNs must practice under a physician.

Dueker et al.'s (2005) findings are also consistent with some parts of the limited literature investigating the impact of changes in occupational regulation on a profession. Sass and Nichols (1996) find that as physical therapists gained greater professional independence from physicians they experienced a fall in incomes. However, the literature is not settled and in research by Goldsmith (1989), it is found that with respect to dental hygienists, the less autonomy and authority they have from dentists, the lower their incomes.

C. Contribution

The work of Dueker et al. (2005) is a first venture into understanding the impact of the dramatic change of allowing nonphysician caregivers to obtain authorities traditionally held by physicians. This research seeks to expand and refine the literature. Instead of examining the APN group, which is composed of very distinct occupations such as midwives and nurse anesthetists that face significantly different labor market conditions, its largest and most general care focused component, NPs, is considered. This is also important because the changes in practice authority examined pertain most directly to NPs.

In addition, instead of focusing exclusively on the changes in NP authority, changes in PA authority are specifically incorporated. If NP authority is important to NP, PA, and physician incomes, it would be expected that similar changes in PA authority would also be important for each of the practitioner groups. Thus, accounting for changes in PA authority is at least as important as accounting for NP authority.

A third advancement is employing richer and more detailed data with respect to NPs and PAs. The CPS, the data used by Dueker et al. (2005), does not specifically identify NPs. In addition, the CPS has a relatively limited sample of PAs (only 51 are reported in the 2005 CPS March Supplement). While a standard workhorse in economics research, these unavoidable limits of the CPS constrain the analysis for NPs and PAs.

A final contribution of this work is to control for an additional avenue of authority obtained by NPs and PAs. While the measure of authority incorporated into the Dueker et al. (2005) investigation (prescriptive authority) is the primary authority considered in other related literature, it is not the only measure of regulation that has changed or that has been studied in the literature. Expanding and controlling for the scope of what occupational changes have occurred provide the opportunity to understand the changes in authority more fully.

II. REGULATORY CHANGE OVERVIEW

No single source provides an accounting of the changes in NP and PA authority. Rather, the literature provides bits and pieces. With this, states can be widely different how they approach occupational regulation. Accounting for every change to both NPs and PAs is not feasible. Thus, the first task was to determine what changes in regulation were material and the second was to compile those changes in a usable format.

The established literature informs on the relative importance of the changes in regulation that have occurred. In Sekscenski et al. (1994), the two largest categories of practice regulation that comprise their "practice environment score" are prescriptive authority and reimbursement authority. Wing et al. (2004), while altering the weights of each authority in their revised practice score, still employ both reimbursement and prescriptive authority as primary components. Dueker et al. (2005) focus solely on prescriptive authority. Following the larger literature, this research focuses on two regulatory areas of authority: prescriptive authority and reimbursement authority.

Prescriptive authority is the area where the most change has occurred and has probably been the most visible to the outside world. Prescriptive authority is defined in this research as the authority to write and administer prescriptions for both noncontrolled and controlled pharmaceutical substances. (6) Twenty years ago most states did not allow NPs and PAs this authority. By 2006, however, nearly all states had authorized NPs and PAs prescriptive authority, though there has been variation in the timing and extent of this authority by state and profession.

NPs and PAs have long been able to see and counsel patients, take medical records, and provide diagnosis and treatment suggestions. However, without full prescriptive authority, they are limited in the total level of care they can provide. For example, if an NP sees a patient with a broken arm, without full prescriptive authority, the pain medications that would typically be prescribed to the patient could not be given. Rather, a physician who can prescribe the required medication(s) is needed so that the patient care can be "completed." As NPs and PAs have obtained full prescriptive authority, they gained the ability to see a larger proportion of patients in totality without the involvement of a physician.

The second area of regulatory change is reimbursement authority. Reimbursement authority concerns a legal mandate that services provided by NPs or PAs be compensated by third-party payers and insurers. NP and PA services have not always been recognized by or been independently billable to third-party payers and insurers. Rather, reimbursement was facilitated through an employing physician practice or medical care facility. By granting specific reimbursement authority, a state requires payers to cover services provided by these nonphysician practitioners.

While both NPs and PAs have experienced changes in reimbursement authority, there is an important distinction in the authority between the two practitioners. For NPs, reimbursement authority implies the ability to bill for services independent of a physician. Having this authority allows an NP to claim payment from insurers in the absence of a physician, altering the more traditional hierarchical relationship.

For PAs, reimbursement authority has a different meaning, which stems from the core practice relationship they have with physicians. From the inception of the PA occupation, PAs have been required to practice under a "supervising" physician. While the meaning of supervising is different from state to state, all states require a PA to be employed by a supervising physician. Thus, being granted reimbursement authority mandates that a PA's services be recognized and paid by third-party payers and insurers, but it does not imply independence from physicians. This, while not as strong as the authority it implies NPs have from physicians, is important in that it alters the previous state of the labor market.

To compile a by state, by year record of the two areas of practitioner authority, a variety of sources were used. For NPs, the initial source employed was the annual "Legislative Update" published annually in the journal Nurse Practitioner (1992-2006). This publication provides a brief overview of state laws concerning NP practice in all 50 states and the District of Columbia. The Nurse Practitioner's Legislative Updates were then supplemented (and corrected) with individual state statue and regulation research to yield a complete panel of state regulations concerning prescriptive and reimbursement authority for NPs for the period 1992-2005.

For PA authority, the American Academy of Physician Assistants' (AAPAAC) Physician Assistants: State Laws and Regulations series of publications was the primary source. This publication has been produced sporadically over the past decade (American Academy of Physician Assistants, 2002, 2000, 1998, 1992) and contains systematic information, by state and year, on PA regulation. This publication was used as a foundation and then supplemented by individual state statute and regulation research that resulted in a complete panel of regulations for the period 1992-2005.

With data on both NPs and PAs, a global picture of how each practitioner group's authorities changed can be seen. Table 1 provides a summary overview of the authority of these practitioners and how their authorities have changed. It highlights the dramatic increase in the percent of states authorizing NPs and PAs both types of authority. In particular, the increase in prescriptive authority has been particularly active with less than half of the states allowing either NPs or PAs to write controlled substance prescriptions in 1992 to all but a handful not allowing this authority by 2005. While the proportion of states granting reimbursement authority is not as great, the level of change between 1992 and 2005 is dramatic, especially for PAs. The variation by state and year is precisely what is needed to empirically estimate the effects of these policy changes. Table A.1 in the appendix provides a year-by-year breakdown of the number of states granting the different authorities.

III. DATA AND EMPIRICAL METHODOLOGY

The data employed in this investigation come from a variety of sources. The variables of interest are the regulatory changes experienced by NPs and PAs, whether they had prescriptive authority and reimbursement authority. The compilation of regulations provided a database indicating by state and year whether NPs and PAs had full prescriptive authority and whether each occupation had reimbursement authority. From this database, four dichotomous variables were created.

The first pair of variables concerns the ability of NPs and PAs to prescribe. NP_Rx takes on a value of 1 if NPs in a state have full authority to write prescriptions and 0 if not. Similarly, PA_Rx takes on the value of 1 if the state authorizes PAs full prescriptive authority and 0 if it does not.

Concerning the reimbursement status of NPs and PAs, NP_ReimbAuth and PA_Reim-bAuth were created. NP_ReimbAuth takes on a value of 1 if NPs are granted this authority and 0 otherwise. PA_ReimbAuth takes on a value of 1 if PAs have this authority extended by the state and 0 if not. Note that NP_ReimbAuth and PA_ReimbAuth have different meanings, as noted previously, though both concern the authority to have reimbursement.

A number of data sets are used in conjunction with this regulatory data. For NPs, the data source for earnings is the National Sample Survey of Registered Nurses (NSSRN). The NSSRN is a national survey of registered nurses (RNs) conducted every 4 yr by the U.S. Department of Health and Human Services. The sample is a probability sample of the universe of RNs, as obtained from the state boards of nursing. While the focus of the survey is RNs, NPs are identifiable since NPs are also RNs. The NSSRN data observational level is the individual and it contains a host of demographic information (age, sex, race, state, marital status, etc.) as well as professional information (income, area of practice, type of employers, typical hours worked, etc). The NSSRN provided a sample of just more than 1,500 NPs over the 4 year (1992, 1996, 2000, and 2004).

The primary data for PAs were the American Academy of Physician Assistants Annual Census. The AAPAAC is a proprietary national survey administered by the American Academy of Physician Assistants (AAPA) annually to all PAs in the United States with an active license to practice. The data were obtained from the AAPA under a special agreement for this research. Its observation level is the individual and contains both demographic (age, sex, race, state, etc.) and professional information (income, area of practice, types of employers, typical hours worked, etc.). The AAPAAC was available for the period 1996-2004 and provided a sample of about 95,500.

The CPS Annual Social and Economic Supplement was the data employed for physician earnings. (7) The CPS is a public data survey, conducted and maintained by the U.S. Bureau of Labor Statistics and the U.S. Census, commonly used in economic research. It contains a host of demographic and labor market information at the individual level and identifies physicians as an occupation though the data does not identify the specialty of physician. The years 1996-2005 were employed in this analysis resulting in a total sample of just more than 3,700 physicians. (8)

To empirically estimate the impact of NP and PA prescriptive and reimbursement authority on the incomes of NPs, PAs, and physicians, a standard wage regression model is employed, with the regulatory variables of interest included. The following augmented Mincer general equation is the first equation estimated.

(1) In [wage.sub.[practitionertype,t,i]] = [[beta].sub.1] + [delta][Rx.sub.[t,i]] + [lambda][Reimbursement.sub.[t,i]] + [pi][X.sub.[t,i]] + [psi][S.sub.[t,i]] + [e.sub.[t,i]],

where Rx is a matrix of state prescriptive authority for NPs and PAs (NP_Rx and PA_Rx) and Reimbursement is matrix of state-level reimbursement authority for NPs and PAs (NP_ReimbAuth and PA_Reim-bAuth). X is a vector of personal characteristics and interactions and S is state real per capita income. In addition, year and state fixed effects are included in the estimations.

The dependent variable being explained in the example is In wage, the natural log of the real earnings (year 2000 dollars) of individual i in time t. The same base model is employed for NPs, PAs, and physicians but is modified to suit the available data for each practitioner category.

While it is possible that some policy or regulatory change could have an immediate effect on a labor market, often the impact of a policy change develops over time. To take this "maturing" into account, the basic model represented by Equation (1) is augmented. To do this, four additional variables were constructed: NP_Rx_TimeSince, PA_Rx_Time-Since, NP_ReimbAuth_TimeSince, and PA_ ReimbAuth_TimeSince. These measure the number of years that NPs and PAs have had the specific authority in a state since 1992. For example, if a state granted NPs the authority to write controlled substance prescriptions in 1996, NP_Rx_TimeSince takes on a value of 4 for year 1999. If a state has never granted authority, the time variable is recorded as 0, as is the contemporaneous variable (NP_Rx). This results in the enhanced model:

(2) In [wage.sub.[practitionertype,t,i]] = [[beta].sub.1] + [delta][Rx.sub.[t,i]] + [delta][Rx_TimeSince.sub.[t,i]] + [lambda][Re imbursement.sub.[t,i]] + [empty set][Reimbursement_TimeSince.sub.[t,i]] + [pi][X.sub.[t,i]] + [psi][S.sub.[t,i]] + [e.sub.[t,i]],

where Rx_TimeSince is a matrix containing the two duration variables and Remibursement_TimeSince is a matrix containing the two duration variables with respect to reimbursement authority.

IV. ESTIMATION RESULTS

A. NP Estimation Results

Equations (1) and (2) are estimated employing the NSSRN data for years 1992, 1996, 2000, and 2004 for full-time NPs between the ages of 25 and 60. Real per capita income as well as state and year fixed effects are included to account for any state or year-specific characteristics that could influence the earnings of NPs. Huber-White standard errors are also employed. Table 2 provides a summary of the estimated coefficients of interest. The full results can be found in Table A3 in the Appendix. Summary statistics and a description of the variables can be found in Table A2 in the Appendix.

The estimation results of Equation (1) indicate that no regulatory variable coefficient of interest is significantly different than zero. When the "time since" variables are included (column 2), the results change. By controlling for the effect of these authorities over time, there is some evidence that both NP and PA expanded authority affect NP incomes. The coefficient estimate on the number of years NPs have had the authority to prescribe is positive and significant at the 5% level, implying that having this authority tends to increase NP earnings over time, all else equal. This conforms to the expectation that as a practitioner receives greater authority, their incomes will also improve. The interpretation is that every year NPs have this authority, their earnings increase on average from the year before by roughly 1.6%, all else equal. No other NP authority was estimated to be statistically significant.

With respect to the impact from PAs having greater authority, the effect is found in the duration measure of PAs having prescriptive authority. The coefficient estimate is negative and significant at nearly the 5% level. This implies that the longer PAs have the authority to write controlled substance prescriptions, the bigger the effect on NP incomes. This finding also fits with the prior expectation that a competitor to NPs having a greater level of practice authority would tend to damage NP incomes. The economic effect of this is also material, with NP earnings falling on average just more than 1% for each year PAs have this authority. No other PA authority was found significant.

B. PA Estimation Results

Equations (1) and (2) are also estimated for PAs, employing the AAPAAC data. The AAPAAC data, however, deserves additional attention. While the AAPAAC data have a large sample size and a number of individual control variables, earnings are recorded as "income bands," generally in $5,000 increments. For example, if a PA has annual earnings of $67,000, this is reported in the AAPAAC as having earnings of between $65,000 and $69,999. The data, being banded, also has left and right censoring (at $30,000 to the left and $150,000 to the right).

These characteristics of the data, particularly the censoring, make standard ordinary least squares (OLS) techniques inappropriate for estimation purposes. A Tobit model, based on Tobin (1958), is more appropriate when censored data are involved. However, another estimation approach that preserves the interval characteristics of the banded income data while accommodating the censoring issue exists. This estimation model, referred to as "interval regression," is based on work by Amenmiya (1973). Interval regression is essentially a generalization of the Tobit model. (9)

The estimation results, using the interval regression method, of Equations (1) and (2) can be found in Table 3. (10) The full results can be found in Table A5 in the Appendix. Summary statistics and a description of the variables can be found in Table A4 in the Appendix.

For Equation (1), the coefficient estimate for NPs having the authority to prescribe controlled substances is found to be negative and significant. This implies that in states where NPs have this authority, PA incomes are lower on average, all else equal, by around 1.4%. (11)

The estimation results for Equation (2) also show a significant and negative impact from NPs having controlled substance authority in the contemporaneous measure. Interestingly, PAs having expanded authority is not found to impact their earnings, either contemporaneously or through time.

It is also found in the estimation of Equation (2) that NPs having payment authority impacts PA earnings. This effect is found in the estimated coefficient on the duration measure. Of particular interest is that the coefficient is positive, which implies that as NPs have reimbursement authority, PA incomes increase over time.

C. Physician Estimation Results

The final practitioner category considered is physicians. Both Equations (1) and (2) were estimated using the CPS ASES data, as described previously. Standard OLS, with state and year fixed effects and Huber-White standard errors, was employed for full-time physicians between the ages of 30 and 60 with a variety of controls. While it would be ideal to control for the specialty of the physician since the data have no information on specialty, this was not possible. The estimation results for the variables of primary interest are provided in Table 4 and full estimation results can be found in Table A7 in the Appendix. Summary statistics and a description of the variables can be found in Table A6 in the Appendix.

The estimation results of Equation (1) for physicians, shown in column 1, indicate that no contemporaneous measure of NP or PA practice authority is significantly different than zero at conventional levels. However, when the length of time NPs and PAs have had prescriptive and reimbursement authorities is included, there is some evidence that both NP and PA authority levels matter. Column 2 shows the coefficient estimates for Equation (2). The coefficient estimate for the length of time NPs have had prescriptive authority is negatively signed and significant at the 5% level. The interpretation for this result is that physician incomes are on average lower by roughly 7.6% for each year NPs have authority to prescribe, all else equal. No other NP authority measure is statistically significant.

Interestingly, the effect on physician earnings is the opposite if PAs have controlled substance prescriptive authority. The coefficient estimated for how long PAs have had prescriptive authority is positive and statistically significant. The interpretation of this coefficient is that each year PAs have the authority to write controlled substance prescriptions, physician incomes increase on average, all else equal.

D. Endogeneity and Selection Bias

The variation being exploited in the current investigation derives from state-level changes in occupational regulation. Even though the investigation examines the impact on individuals, there is some concern that regulatory changes may not be exogenous with respect to provider incomes. If the regulations are not exogenous then OLS estimates are not credible. Before a full discussion of the estimation results, some attention should be paid to this issue.

To assess this concern, the endogeneity of the regulatory variables was tested. To do this, the Durbin-Wu-Hausman lest was performed for the four potentially endogenous regulatory variables: NP prescriptive authority, NP reimbursement authority, PA reimbursement authority, and PA payment authority for each practitioner sample. This test essentially compares the coefficient estimates obtained by estimating via OLS to two-stage least squares estimates. (12) If the potential endogenous variables are in actuality exogenous then both OLS and two-stage least squares are consistent, but the OLS estimates are more efficient. Hausman (1978) proposed a straightforward method to investigate this, and Wooldridge (2006) expands the test to the case where there are multiple suspected endogenous variables.

To implement this test, appropriate instruments must be selected for the potentially endogenous variables. Finding appropriate instruments is always a challenge, and for the current investigation, instruments were taken from a public choice framework that employs political variables as instruments for regulatory and policy changes. Specifically, the pool of political variables chosen as instruments included population per state senator, population per state house representative, political party of the governor, percent of Senate that are Republicans, percent of House that are Republicans, whether the legislature has annual sessions, whether the legislature can call itself into session, whether there is a time limit on legislative sessions, and whether the legislative body can determine its agenda in special sessions. Intuitively, these political variables make reasonable instruments as changes in regulation are by their nature political. In evaluating the chosen instruments for each practitioner group, they met the conditions of being correlated with the suspect regulatory variables as well as not being correlated with the error term of the original estimation, Equation (1). (13) In addition, F statistics testing the joint significance of the instruments in the first stage are provided in Table A8 show that the instruments have explanatory power.

There was no evidence that the regulatory variables were endogenous. This implies that the regulatory variables can be treated as exogenous and OLS is efficient. This finding mirrors that of Dueker et al. (2005) who found that APN prescriptive authority was exogenous to provider incomes.

Another empirical concern, especially in labor market investigations, is selection bias. While believed to be mitigated in this research since it investigates highly educated and specialized professionals (especially in the case of physicians), it can always be a concern. To help further alleviate the concern, the presence of selection bias was tested following the procedure developed by Heckman (1976). Essentially, the participation decision is modeled on the entire sample and the inverse Mill's ratio is calculated. The inverse Mill's ratio is then included in the log wage regression. For both NPs and physicians, the coefficient on the inverse Mill's ratio was insignificant, which implies that sample bias is not a concern.

One requirement for the Heckman procedure is that the first-stage selection equation have at least one variable not in the second stage (Wool-dridge 2006). Unfortunately, the PA data are very limited in the demographic controls it contains. Given this, the sample selection issue could not be credibly investigated. However, the finding that selection bias was not a concern for the NP and the physician's estimations should help mitigate the concern for PAs.

V. DISCUSSION

A. Results

A number of interesting results can be taken from this research. The first is that there is some evidence that NPs having greater practice authority, as measured by prescriptive authority, is found to increase NP earnings. This result is counter to the negative impact on earnings found by Dueker et al. (2005). In a larger sense, the current finding that expanded practice authority increases NP earnings is intuitive. It is likely that the different finding is the result of using data that more cleanly identifies NPs from traditional RNs as well as from other types of APNs, such as mid-wives and nurse anesthetists.

NPs having prescriptive authority is also found to impact PA and physician incomes. For PAs, the impact is found in the contemporaneous measure of the NP prescriptive authority while in the duration measure for physicians. However, in both instances, the result is negative, implying greater NP authority leads to a decrease in PA and physician incomes. The impacts are statistically and economically significant. When NPs have prescriptive authority, PA incomes fall on average about 1.4%. For physicians, the impact is larger. The strict interpretation of the coefficient on NPs having prescriptive authority is that physician incomes, all else equal, fall by roughly 7.6% for each year NP authority. This result could imply that physician incomes are sensitive and when they face significant competition from another, substitute caregiver, their incomes, and potentially the cost of care tend to fall materially. Taken with the finding that this authority tends to increase NP incomes by about 1.6% a year, a convergence of incomes between NPs and physicians is also implied. Thus, as NPs and physicians become more substitute caregivers (which NPs having prescriptive authority facilitates), there is a shrinking difference in their average incomes. While this research was not able to specifically identify the specialty of the physician, it is intuitive that the downward effect of NP authority on physician incomes would be greater for (and perhaps is driven by) general practice physicians as the overlap in the typical practice of NPs and general physicians is closer than that of NPs and specialist physicians. Of note is that the finding that NP authority tends to harm physician incomes is in line with the findings of Dueker et al. (2005).

One impact of NP authority that does not seem intuitive is the positive impact NPs having payment authority has on PA incomes. The effect was found in the duration measure of NP payment authority and had a magnitude of about 0.3% per year. While a relatively small economic effect, it is one that grows over time. Though at first, this seems peculiar, it becomes less so as it is considered in context of what this authority implies. NPs with this authority have the ability to receive payment for services they provide independent of a physician. This decouples the traditional need for NPs to work for a supervising physician to be compensated by insurers and third-party payers. To the extent that physician-led medical practices still demand the services of nonphysician caregivers, a role filled by both NPs and PAs, and to the extent that NPs remove themselves from such positions by seeking other, more independent opportunities, the indirect effect is an increase in the demand for PAs who are always constrained to work under a supervising physician. This relative shift in demand for PAs, in turn, leads to PAs experiencing an increase in earnings at the margin. It is not clear that this is the source of the positive effect, but given the somewhat contentious relationship physicians and NPs have had over the appropriate levels of authority NPs should possess in some states, it is a reasonable hypothesis that could be tested with data on medical practice staffing.

The effect of PAs having greater authority meets the expectation that it would have a downward effect on NP earnings. The impact on NPs is found in the duration measure of PAs having prescriptive authority. When a state grants PAs controlled substance prescriptive authority, the results indicate that NP incomes fall on average roughly 0.8% a year. This, as with the effect found of NP prescriptive authority on PAs, meets the general expectation that when one occupation receives a greater authority to practice, other substitute occupations are damaged.

A curious result of the estimation of physician earnings is the positive impact of PAs having prescriptive authority. Not only is the effect estimated to be positive and significant, the economic significance is also strong. The strict interpretation of the coefficient estimate is that PAs having prescriptive authority tends to increase physician earnings by about 8% for each year of authority. While at first striking in sign and magnitude, the result is not so odd upon further consideration of the relationship between physicians and PAs and the physician data employed.

Since PAs must be an employee, they obtain their authority to practice through their employing physician as well as through state regulation. That they are always an economic agent of a physician, it is reasonable that expanding their authority would positively reflect on physicians who employ PAs. If an employee becomes more productive and is constrained to always remain an employee, it is reasonable that the fruits of this productivity will show up for the employer. When PAs can write the full spectrum of medications, they are essentially allowed to be complete caregivers and are therefore more productive. This would benefit the supervising physician in at least two ways. First, it would allow the physician to continue seeing and treating her own patients without the interruption of seeing patients referred by an employee PA who could not complete the patient's visit. Second, it would increase the total number of patients the employee PA could see, which presumably would increase the medical practice's revenues.

It is an open question, however, as to whether all physicians are made better off from PAs having this expanded authority. The results inform that on average physicians in a state are made better off. The CPS data does not, however, record physician practice area or give information about whether a physician employs a PA. So, if general practice physicians are impacted differently than other specialist physicians, or if employing physicians are impacted differently than nonemploying physicians, any differential impacts are masked. In particular, it is plausible that specialist physicians gain more from PAs having expanded authority than would general practice physicians. As PAs obtain greater practice authority, they become more like general practice physicians, able to fill most of the roles a general physician would in a typical general medical practice. This would tend to harm general practice physicians from a purely competitive aspect while benefiting specialist physicians. Along with this, specialists outnumber general practice physicians by roughly two to one (Smart, 2007).

It could also be the case that medical practices, particularly general care practices, would fill open positions with PAs instead of hiring another physician into the practice. If the PA can perform all or most of the activities a new general care physician would for the practice but would be constrained to be an employee rather than a potential partner, it could enhance the incomes of the existing general care physicians in the practice while also having a downward effect of the hiring of new general practice physicians. This could also be why the effect shows up in the duration, implying that this filling of positions with PAs occurs over time as practices expand.

These questions could be explored with more detailed staffing data on medical practices and information about physician specialty. However, without this type of data immediately available, all that can be said from the current results it that physicians, as a group, are benefited from PAs being granted greater authority.

VI. CONCLUSIONS

The rise of NPs and PAs as caregivers has been dramatic and the impacts of this rise have been largely ignored in the research literature. This article provides the most comprehensive examination to date of the NP and PA impact by first explicitly accounting for both NP and PA authority and then by using data better suited for the analysis. In so doing, it provides a clearer understanding of the impacts of the dramatic shift in the health-care provider market and signals where some future research may be aimed.

A primary contribution of this article is that incorporating both NP and PA authorities is important. This is intuitive as NPs and PAs are generally considered substitutes for one another. The previous research literature has recognized the importance of NP authority, but finding significant results from both NP and PA authority demonstrates that incorporating both in any investigation is important.

This research also illustrates the importance of controlling for different authorities specifically. In the current case, both prescriptive and payment authority were explicitly controlled. The current results also indicate that just controlling for the contemporaneous measure of authority can miss the ultimate impact of a regulatory change. Rather, since labor market responses can occur over time, incorporating a measure of the length of time the regulatory changes have been in place is important.

A number of policy implications can be drawn from this research concerning the impact of NPs and PAs being granted greater practice authority. A first is that there have been impacts from changing NP and PA authority on all three categories of practitioners. NPs having greater prescriptive authority was found to help their own earnings while causing a fall in physician and PA incomes. Given that NPs and general care physicians are often seen as substitute caregivers, one potentially important policy result from this, especially with the literature finding that the quality of NP and physician care is clinically similar, is that there are potential cost savings to allowing greater competition between these practitioners.

PAs having authority was not found to specifically impact their own earnings, but it was found to lower NP earnings. Interestingly, PA expanded authority was found to improve physician incomes, all else equal. This result likely stems from the employer-employee relationship between physicians and PAs and that physicians of all specialties were included in the physician sample. Neither issue could be addressed given the limits of the physician data in the current research, but the question does raise other questions concerning the impact on medical practice productivity stemming from PA authority as well as the impact on provider population levels.

The larger research literature is split on the ultimate effect of an occupation being granted greater freedom to practice. The current research provides evidence that own authority can increase own earnings. No evidence that own authority hurts own earnings was found. It also shows that other occupations' authority can both help and hurt related occupations. This dual finding implies that there are likely inherent differences in specific, though related, labor markets. This underscores the importance of understanding such differences in policy making. While currently unavailable, a contribution could be made in disentangling these differences by having detailed data on medical practices and staffing trends for a number of years.

With dramatic changes in authority being seen across the medical occupation landscape, this makes it increasingly important to carefully examine these types of impacts, especially since the literature provides no generalizable results. As one example of another major development and authority change, in 2002. New Mexico granted psychologists the authority to write some prescriptions for clients in therapy, a practice traditionally the sole domain of psychiatrists. In another example, the Federal Drug Administration recently announced that it was considering allowing pharmacists to prescribe some medications without physician involvement (Fox News, 2007). The ultimate impact of these types of changes has potentially wide-ranging effects and is of substantial policy interest.

Health care will continue to receive a substantial amount of policy interest. A part of that attention should be aimed at understanding the impact of changes in the occupational regulation of its occupations. This research provides a foundation for examining these types of changes in other areas of the health industry while specifically educating on the impact NP and PA authority changes have had on the earnings of NPs, PAs, and physicians. It also provides a foundation for further study that will provide a deeper understanding of the impacts of changing NP and PA authority on their own and physician labor markets.

REFERENCES

Amenmiya, T. "Regression Analysis when the Dependent Variable is Truncated Normal." Econometrica, 41, 1973, 997-1016.

American Academy of Physician Assistants. Physician Assistants: State Laws and Regulations. 6th ed. Alexandria, VA: Author. 1992.

--. Physician Assistants: State Laws and Regulations. 7th ed. Alexandria, VA, 1998.

--. Physician Assistants: State Laws and Regulations. 8th ed. Alexandria, VA, 2000.

--. Physician Assistants: State Laws and Regulations. 9th ed. Alexandria, VA, 2002.

Brown. S. A., and D. E. Grimes. "A Meta-analysis of Nurse Practitioners and Nurse Midwives in Primary Care." Nursing Research, 44(6), 1995, 332-39.

Cawley, J. "Physician Assistants in the Health Care Workforce." in The Roles of Physician Assistants and Nurse Practitioners in Primary Care, edited by D. K. Clawson and M. Osterweis. Washington, DC: Association of Academic Health Centers, 1993, 21-39.

Cooper. R. A. "Health Care Workforce for the Twenty-First Century: The Impact of Nonphysician Clinicians." Annual Review of Medicine, 52, 2001, 51-61.

Cooper, R. A., and L. H. Aiken. "Human Inputs: The Health Care Workforce and Medical Markets." Journal of Health Politics. Policy and Law, 26(5), 2001, 925-38.

Cooper, R. A., T. Henderson, and C. L. Dietrich. "Roles of Nonphysician Clinicians as Autonomous Providers of Patient Care." Journal of the American Medical Association, 280, 1998, 796-802.

Druss, B. G., S. C. Marcus, M. Olfson, T. Tanielian, and H. A. Pincus. "Trends in Care by Nonphysician Clinicians in the United States." New England Journal of Medicine, 348, 2003, 130-37.

Dueker, M. J., A. K. Jacox, D. E. Kalist, and S. J. Spurr. "The Practice Boundaries and Advanced Practice Nurses: An Economic and Legal Analysis." Journal of Regulatory Economics, 27, 2005, 309-29.

Fox News. "FDA May Allow Pharmacists to Give Drugs without Prescriptions." 2007. Accessed October 3, 2007. http://www.foxnews.com/printer_friendly_story/0,3566(299162),00.html.

Goldsmith, A. "Dental Hygienists' Wages: The Role of Human Capital and Institutional Factors." Quarterly Review of Economics and Business. 29. 1989. 56-67.

Hausman, J. A. "Specification Tests in Econometrics." Econometrica, 46, 1978, 1251-71.

Heckman, J. J. "The Common Structure of Statistical Models of Truncation, Sample Selection, and Limited Dependent Variables and a Simple Estimator for Such Models." Annals of Economic and Social Measurement. 5, 1976. 475-92.

Hooker, R. S., and L. McCaig. "Use of Physician Assistants and Nurse Practitioners in Primary Care, 1995-1999." Health Affairs, 20(4), 2001, 231-39.

Horricks, S., E. Anderson, and C. Salisbury. "Systematic Review of Whether Nurse Practitioners Working in Primary Care Can Provide Equivalent Care to Doctors." British Medical Journal. 324, 2002. 819-23.

Kalist, D. E., and S. J. Spurr. "The Effect of State Laws on the Supply of Advanced Practice Nurses." International Journal of Health Care Finance and Economics, 4(4). 2004, 271-81.

Kennedy, P. "Estimation with Correctly Interpreted Dummy Variables in Semilogarithmic Equations." American Economic Review. 71, 1981. 801.

Kinnersley. P., E. Anderson, K. Perry, J. Clement, L. Archard. P. Thurton, A. Stainthorpe, A. Fraser, C. C. Butler, and C. Rogers. "Randomized Controlled Trial of Nurse Practitioner Versus General Practitioner Care for Patients Requesting 'Same Day' Consultations in Primary Care." British Medical Journal, 320, 2000, 1043-48.

Lenz. E. R., M. O. Mundinger, R. L. Kane, S. C. Hopkins, and S. X. Lin. "Primary Care Outcomes in Patients Treated by Nurse Practitioners or Physicians." Medical Care Research and Review, 61. 2004, 332-51.

Mundinner, M. O., R. L. Kane, E. R. Lenz, A. M. Totten, W.Y. Tai, P. D. Cleary, W. T. Friedewald, A. L. Siu, and M. L. Shelanski. "Primary Care Outcomes in Patients Treated by Nurse Practitioners or Physicians." Journal of the American Medical Association. 283, 2000. 59-68.

Oliver. D. R., J. E. Conboy. W. J. Donahue. M. A. Daniels, and P. A. McKelvey. "Patient Satisfaction with Physician Assistant Services." Physician Assistant, 10(7), 1986, 51-60.

--"1991-92 Update: How Each State Stands on Legislative Issues Affecting Advanced Nursing Practice." Nurse Practitioner, 17(1), 1992. 14-23.

--" 1992-93 Update: How Each State Stands on Legislative Issues Affecting Advanced Nursing Practice." Nurse Practitioner, 18(1), 1993, 23-38.

--"Annual Update of How Each State Stands on Legislative Issues Affecting Advanced Nursing Practice." Nurse Practitioner, 19(1), 1994, 11-53.

--"Annual Update of How Each State Stands on Legislative Issues Affecting Advanced Nursing Practice." Nurse Practitioner, 20(1), 1995, 13-51.

--"Annual Update of How Each State Stands on Legislative Issues Affecting Advanced Nursing Practice." Nurse Practitioner, 21(1), 1996, 10-70.

--"Annual Update of How Each State Stands on Legislative Issues Affecting Advanced Nursing Practice." Nurse Practitioner. 22(1), 1997, 18-86.

--"Annual Update of How Each State Stands on Legislative Issues Affecting Advanced Nursing Practice." Nurse Practitioner, 23(1), 1998, 14-66.

--"Annual Update of How Each State Stands on Legislative Issues Affecting Advanced Nursing Practice." Nurse Practitioner, 24(1), 1999, 16-83.

--"Twelfth Annual Legislative Update." Nurse Practitioner, 25(1), 2000, 16-68.

--"Thirteenth Annual Legislative Update." Nurse Practitioner, 26(1), 2001, 7-57.

--"Fourteenth Annual Legislative Update." Nurse Practitioner, 27(1), 2002, 10-52.

--"Fifteenth Annual Legislative Update." Nurse Practitioner, 28(1), 2003, 26-58.

--"Sixteenth Annual Legislative Update." Nurse Practitioner, 29(1), 2004, 26-51.

Perry, K. "Patient Survey: Physician Extenders. Why Patients Love Physician Extenders." Medical Economics, 72(16), 1995, 58, 63, 67.

Phillips, S. J. "Seventeenth Annual Legislative Update: A Comprehensive Look at the Legislative Issues Affecting Advanced Nursing Practice." Nurse Practitioner, 30(1), 2005, 14-47.

--"Eighteenth Annual Legislative Update: A Comprehensive Look at the Legislative Issues Affecting Advanced Nursing Practice." Nurse Practitioner, 31(1), 2006, 6-38.

Roblin, D. W., E. R. Becker. K. Adams, D. H. Howard, and M. H. Roberts. "Patient Satisfaction with Primary Care: Does Type of Practitioner Matter?" Medical Care, 42, 2004", 579-90.

Rudy, E. B., L. J. Davidson, B. Daly, J. M. Clochesy, S. Sereika, M. Baldisseri, M. Hravnak, T. Ross, and C. Ryan. "Care Activities and Outcomes of Patients Cared for by Acute Care Nurse Practitioners, Physician Assistants, and Resident Physicians: A Comparison." American Journal of Critical Care, 7, 1998, 267-81.

Ryan, S. "Nurse Practitioners: Educational Issues, Practice Styles, and Service Barriers," in The Roles of Physician Assistants and Nurse Practitioners in Primary Care, edited by. D. K. Clawson and M. Osterweis. Washington, DC: Association of Academic Health Centers, 1993, 41-49.

Sass, T. R., and M. W. Nichols. "Scope-of-Practice Regulation: Physician Control and the Wages of Non-Physician Health Care Professionals." Journal of Regulatory Economics, 9, 1996, 61-81.

Sekscenski, E. S., S. Samson, C. Bazell, M. E. Salmon, and F. Mullan. "State Practice Environments and the Supply of Physician Assistants, Nurse Practitioners, and Certified Nurse-Midwives." New England Journal of Medicine, 331, 1994, 1266-71.

Smart, D. Physician Characteristics and the Distribution in the US. 2007 ed. Chicago: American Medical Association, 2007.

Stata. STATA Base Reference Manual, Vol. 1, A-J, Release 9. College Station, Texas: Stata Press Publication, 2005.

Tobin, J. "Estimation of Relationships for Limited Dependent Variables." Econometrica, 26, 1958, 24-36.

United States Congress, Office of Technology Assessment. Nurse Practitioners, Physician Assistants, and Certified Nurse-Midwives: A Policy Analysis. Health Technology Case Study 37. OTA-HCS-37. Washington, DC: Government Printing Office, 1986.

Wilson, I. B., B. E. Landon, L. R. Hirschlorn, K. McInnes, L. Ding, P. V. Marsden, and P. D. Cleary. "Quality of HIV Care Provided by Nurse Practitioners, Physician Assistants, and Physicians." Annals of Internal Medicine, 143, 2005. 729-36.

Wing, P., M. H. Langelier, E. S. Salsberg, and R. S. Hooker. "The Changing Professional Practice of Physician Assistants, Journal of the American Academy of Physician Assistants, 17, 2004, 3-49.

Wooldridge, J. M. Introductory Econometrics: A Modern Approach. Mason, OH: Thomson/South-Western, United States, 2006.

ABBREVIATIONS

AAPA: American Academy of Physician Assistants

AAPAAC: American Academy of Physician Assistants Annual Census

APN: Advanced Practice Nurse

CPS: Current Population Survey

NP: Nurse Practitioner

NSSRN: National Sample Survey of Registered Nurses

MSA: Metropolitan Statistical Area

OLS: Ordinary Least Squares

PA: Physician Assistant

RN: Registered Nurse

Online Early publication May 15, 2009

doi: 10.1111/j.1465-7287.2009.00162.x

JOHN J. PERRY *

* I would like to thank Glenn Blomquist, John Garen. Kenneth Troske, Joseph Fink, Dave Ziebart, the late Mark Berger, Chris Clark, Mike Clark, and Barry Boardman for guidance and thoughts on this article's development. I would also like to thank the AAPA and Kevin Kraditor for the use of the Academy's data on PAs.

Perry: Assistant Professor of Economics, Economics and Financial Economics Program, Centre College, 600 West Walnut Street. Danville. KY 40422. Phone (859) 913-7933, Fax (859) 238-5256. E-mail john.perry@eentre.edu

(1.) See Cooper, Henderson, and Dietrich (1998), Cooper (2001), Cooper and Aiken (2001), Cawley (1993), Druss et al. (2003), Hooker and McCaig (2001), Ryan (1993), and United States Congress, Office of Technology Assessment (1986).

(2.) See Brown and Grimes (1995), Horricks, Anderson, and Salisbury (2002), Lenz et al. (2004), Mundinger et al. (2000), Rudy et al. (1998), and Wilson et al. (2005).

(3.) See Kinnersley et al. (2000), Oliver et al. (1986), Perry (1995), and Roblin et al. (2004).

(4.) Note that masters-level programs in nursing are not exclusively NP producing.

(5.) APN is a term used to describe nurses in advanced practice. NPs, nurse-midwives, nurse anesthetists, and certified nurse specialists are all classified as APNs.

(6.) Prescription pharmaceutical substances are classified into two broad categories by the Controlled Substances Act passed by Congress in 1970: noncontrolled substances and controlled substances. Noncontrolled substances are medications generally thought to have a low probability of addiction or abuse and include medieations such as antibiotics and acid reflux medications. Controlled substances, on the other hand, are pharmaceuticals identified to have a high risk of addiction or abuse and are thus regulated more heavily. There are five schedules of controlled substances. Schedule I is considered the most addictive and classified as having no medical use. Schedule V is the lowest schedule and is composed of drugs that, while potentially addictive, are not considered "highly" addictive.

(7.) It should be noted that total income was used for physicians (as well as NPs and PAs) relying on the accuracy of the reported income data from the CPS as well as NSSRN and AAPAC.

(8.) While including earlier years of data, before 1996, the CPS top-coded earnings at $100,000 were explored. While this is not likely a critical truncation of earnings for most occupations, it is for physicians. In 1995, more than half of the sampled physicians had top-coded earnings.

(9.) See STATA Manual Reference A-J Release 9 (Stata, 2005), p. 513.

(10.) The estimations were also performed using OLS and Tobit techniques using the midpoint of each income band as the income. Nicely, the results were not sensitive to the estimation technique. The PA income data bands were also adjusted to real terms using the CPI (2000 real dollars).

(11.) Note that in a semilog model, the coefficient estimate on a dummy variable is approximately the percent change in the dependent variable from the variable being tuned "on." See Kennedy (1981).

(12.) It should be noted the estimation results for PAs in this article are not OLS estimates. However, the results from using the interval regression model and OLS are materially the same. This provides some assurance that this endogeneity test is appropriate for the PA estimates as well as for NPs and physicians.

(13.) One important note is that instruments for all years of data for each state were not available. Thus, the endogeneity test was performed on a smaller sample than the total sample for each of the practitioners.
TABLE 1

Percent of States Granting Specific Regulatory Authority for Controlled
Substances Prescriptions or Reimbursement to NPs and PAs, Selected
Years

                                                     Year

Regulatory Authority                        1992  1996  2000  2005

NPs have controlled substance prescriptive   37    61    78    90
authority (%)

NPs have reimbursement authority (%)         43    53    63    63

PAs have controlled substance prescriptive   47    63    76    86
authority (%)

PAs have reimbursement authority (%)          2     4    16    20

TABLE A1

Number and Percent of States Granting NPs and PAs Specified Regulatory
Authority

        NPs have Controlled         NPs have
        Substance Authority      Reimbursement
                                    Authority

Year    States    Percent      States    Percent

2005      46        90           32        63
2004      46        90           32        63
2003      45        88           32        63
2002      43        84           32        63
2001      41        80           32        63
2000      40        78           32        63
1999      37        73           29        57
1998      34        67           27        53
1997      33        65           28        55
1996      31        61           27        53
1995      29        57           27        53
1994      24        47           25        49
1993      22        43           24        47
1992      19        37           22        43

        PAs have Controlled         PAs have
        Substance Authority       Reimbursement
                                    Authority

Year    States    Percent      States    Percent

2005      44        86           10        20
2004      42        82           10        20
2003      41        80           10        20
2002      40        78            9        18
2001      40        78            8        16
2000      39        76            8        16
1999      39        76            6        12
1998      34        67            6        12
1997      33        65            3         6
1996      32        63            2         4
1995      32        63            2         4
1994      28        55            2         4
1993      25        49            1         2
1992      24        47            1         2


TABLE 2

NP Earnings Estimation, Selected Results (Natural Log of Earnings)

                                 Coefficient        Coefficient
                                 Estimate (t        Estimate (t
                                   Value)             Value)

                                  Column 1           Column 2

NPs have controlled substance   -0.038 (0.95)      -0.025 (0.63)
authority (NP_Rx)

Years since 1992 NPs have had                       0.016 (2.24)
controlled substance authority
(NPRx_TimeSince)

NPs have reimbursement           0.004 (0.07)       0.036 (0.44)
authority (NP_ReimbAuth)

Years since 1992 NPs have had                      -0.002 (0.53)
reimbursement authority
(NPReimbAuth_TimeSince)

PAs have controlled substance   -0.032 (0.72)      -0.063 (1.18)
authority (PA_Rx)

Years since 1992 PAs have had                      -0.011 (2.54)
controlled substance authority
(PARx_TimeSince)

PAs have reimbursement          -0.021 (0.27)      -0.151 (0.92)
authority (PA_ReimbAuth)

Years since 1992 PAs have had                       0.033 (1.39)
reimbursement authority
(PAReimbAuth_TimeSince)

                                [R.sup.2] = .1975  [R.sup.2] = .2063

                                            n = 1,559

                                   Years: 1992, 1996, 2000, 2004

Note: Values in bold indicate significant at the 5% level and t scores
are in parentheses. State and year fixed effects are included with
robust standard errors. Earnings are adjusted to year 2000 dollars.
Estimation incorporates sample NSSRN weights and includes full-time NPs
between the ages of 25 and 60. Full regression results can be found in
Table A2.

TABLE A3

NP Earnings Regression Results

            Variable                Coefficient        Coefficient
                                    Estimate (t)       Estimate (t)

NPs have controlled substance      -0.0384 (0.95)     -0.0245 (0.63)
authority (NP_Rx)

Years since 1992 NPs have had                          0.0159 (2.24)
controlled substance authority
(NPRx_TimeSince)

NPs have reimbursement authority    0.0044 (0.07)      0.0363 (0.44)
(NP_ReimbAuth)

Years since 1992 NPs have had                         -0.0018 (0.53)
reimbursement authority (N
PReimbAuth_TimeSince)

PAs have controlled substance      -0.0315 (0.72)     -0.0628 (1.18)
authority (PA_Rx)

Years since 1992 PAs have had                         -0.0111 (2.54)
controlled substance authority
(PARx_TimeSince)

PAs have reimbursement authority   -0.0211 (0.27)      0.1510(0.92)
(PA_ReimbAuth)

Years since 1992 PAs have had                          0.0336 (1.39)
reimbursement authority
(PAReimbAuth_TimeSince)

Self-employed                       0.0083 (0.12)      0.0101 (0.14)

Part-time student                  -0.0148 (0.52)     -0.0124 (0.44)

Experience                          0.0114 (3.75)      0.0116 (3.80)

Experience squared                 -0.0002 (1.97)     -0.0002 (2.07)

Married                             0.0004 (0.00)      0.0064 (0.07)

Female                              0.0011 (0.02)      0.0045 (0.07)

Female x married                    0.0131 (0.15)     -0.0196 (0.21)

Female x children                  -0.0540 (0.65)     -0.0526 (0.64)

White                              -0.0310 (1.16)     -0.0302 (1.12)

Black                               -0.0065 (0.17)    -0.0081 (0.21)

Other race                            Dropped             Dropped

Children                            0.0825 (0.95)      0.0843 (0.97)

Work in hospital setting            0.1028 (1.42)      0.1167 (1.49)

Work in nursing home                0.0849 (1.10)      0.0933 (1.10)

Work in an educational setting      0.0790 (0.88)     (0.0943 (0.95)
Work in an public health           -0.0133 (0.18)     -0.0038 (0.05)
setting

Work in an student health          -0.0413 (0.53)     -0.0294 (0.35)
setting

Work in an occupational health        Dropped             Dropped
setting

Work in a private health-care       0.1002 (1.38)      0.1119 (1.39)
office setting

Work in any other setting           0.0874 (1.04)      0.0980 (1.10)

Highest education: high school      0.0137 (0.34)      0.0156 (0.38)
diploma

Highest education: associates         Dropped             Dropped

Highest education: bachelors        0.0274 (0.78)      0.0281 (0.80)

Highest education: MS/PhD           0.1105 (3.42)      0.1100 (3.35)

State real per capita income       -0.0229 (2.04)     -0.0319 (2.41)
(000s)

Have more than one job             -0.0017 (0.09)     -0.0058 (0.31)

Live in an MSA                      0.0495 (2.99)      0.0491 (2.95)

Constant                           11.5217 (23.87)    11.7202 (22.43)

                                  [R.sup.2] = .1975  [R.sup.2] = .2063

                                               n = 1,559

                                    Years: 1992, 1996, 2000, and 2004

Notes: Slate and year fixed effects are included with robust standard
errors. NSSRN sample weights employed.

TABLE A2

NP Earnings Data Summary Statistics

Variable                                              Mean   Standard
                                                             Deviation

Real income (year 2000 dollars)                      59,195    16,628

NPs have controlled substance authority (NP_Rx)       0.64      0.48

Years since 1992 NPs have had controlled substance    4.40      4.62
authority (NPRx_TimeSince)

NPs have reimbursement authority (NP_ReimbAuth)       0.46      0.50

Years since 1992 NPs have had reimbursement           3.26      4.52
authority (NPReimbAuth_TimeSince)

PAs have controlled substance authority (PA_Rx)       0.67      0.47

Years since 1992 PAs have had controlled substance    4.92      4.84
authority (PARx_TimeSince)

PAs have reimbursement authority (PA_ReimbAuth)       0.14      0.34

Years since 1992 PAs have had reimbursement           0.66      1.94
authority (PAReimbAuth_TimeSince)

Self-employed                                         0.03      0.17

Part-time student                                     0.06      0.24

Experience                                           18.10      8.77

Experience squared                                  404.51    341.06

Married                                               0.69      0.46

Female                                                0.93      0.26

Female x married                                      0.63      0.48

Female x children                                     0.33      0.47

While                                                 0.87      0.33

Black                                                 0.05      0.21

Other race                                            0.08      0.27

Children                                              0.37      0.48

Work in hospital setting                              0.37      0.48

Work in nursing home                                  0.04      0.19

Work in an educational setting                        0.03      0.17

Work in an public health setting                      0.17      0.38

Work in an student health setting                     0.03      0.18

Work in an occupational health setting                0.02      0.13

Work in a private health-care office setting          0.30      0.46

Work in any other setting                             0.04      0.20

Highest education: high school diploma                0.06      0.23

Highest education: associates                         0.05      0.22

Highest education: bachelors                          0.15      0.35

Highest education: MS/PhD                             0.74      0.44

Have more than one job                                0.20      0.40

Weeks                                                51.68      0.72

Hours                                                42.12      7.02

y 1992                                                0.15      0.36

y 1996                                                0.21      0.41

y 2000                                                0.29      0.46

y 2004                                                0.34      0.47

Live in an MSA                                        0.75      0.43

State real per capita income                         28,247     4,941

Alabama                                               0.02      0.14

Alaska                                                0.01      0.08

Arizona                                               0.01      0.11

Arkansas                                              0.01      0.11

California                                            0.08      0.27

Colorado                                              0.02      0.15

Connecticut                                           0.01      0.11

Delaware                                              0.00      0.05

District of Columbia                                  0.01      0.10

Florida                                               0.06      0.24

Georgia                                               0.03      0.18

Hawaii                                                0.01      0.07

Idaho                                                 0.00      0.06

Illinois                                              0.03      0.18

Indiana                                               0.02      0.12

Iowa                                                  0.01      0.08

Kansas                                                0.01      0.12

Kentucky                                              0.01      0.11

Louisiana                                             0.01      0.08

Maine                                                 0.01      0.09

Maryland                                              0.02      0.15

Massachusetts                                         0.03      0.17

Michigan                                              0.02      0.14

Minnesota                                             0.01      0.11

Mississippi                                           0.01      0.12

Missouri                                              0.02      0.12

Montana                                               0.00      0.07

Nebraska                                              0.00      0.06

Nevada                                                0.01      0.07

New Hampshire                                         0.01      0.08

New Jersey                                            0.02      0.15

New Mexico                                            0.01      0.09

New York                                              0.11      0.32

North Carolina                                        0.03      0.18

North Dakota                                          0.00      0.06

Ohio                                                  0.03      0.17

Oklahoma                                              0.01      0.08

Oregon                                                0.02      0.13

Pennsylvania                                          0.06      0.24

Rhode Island                                          0.00      0.07

South Carolina                                        0.01      0.12

South Dakota                                          0.00      0.06

Tennessee                                             0.03      0.17

Texas                                                 0.05      0.22

Utah                                                  0.01      0.08

Vermont                                               0.00      0.04

Virginia                                              0.04      0.18

Washington                                            0.02      0.15

West Virginia                                         0.00      0.05

Wisconsin                                             0.01      0.12

Wyoming                                               0.00      0.04

                                                         n = 1,559

Note: Summary statistics are weighted with NSSRN weights.


TABLE 3

PA Earnings Interval Regression Estimation, Selected Results (Natural
Log of Earnings)

                                   Coefficient         Coefficient
                                Estimate (z Scor)   Estimate (z Score)

                                    Column 1            Column 2

NPs have controlled substance     -0.014 (4.34)       -0.015 (4.45)
authority (NP_Rx)

Years since 1992 NPs have had                         -0.001 (0.97)
controlled substance authority
(NPRx_TimeSincc)

NPs have reimbursement             0.005 (0.99)       -0.002 (0.36)
authority (NP_ReimbAuth)

Years since 1992 NPs have had                          0.003 (3.95)
reimbursement authority
(NPReimbAuth_TirneSince)

PAs have controlled substance      0.004 (1.07)        0.006 (1.41)
authority (PA_Rx)

Years since 1992 PAs have had                         -0.000 (0.15)
controlled substance authority
(PARx_TimeSince)

PAs have reimbursement             0.004 (1.01)        0.005 (1.11)
authority (PA_ReimbAuth)

Years since 1992 PAs have had                         -0.002 (1.40)
reimbursement authority
(PAReimbAuth_TimeSince)

                                [[chi].sup.2]   [[chi].sup.2] = 16.317
                                 = 16,307
                                             n = 95, 570

                                           Years: 1996-2004

Notes: Values in bold indicate significant at the 5% level and z scores
are in parentheses. State and year fixed effects are included with
robust standard errors. Earnings are adjusted to year 2000 dollars.
Estimation includes full-time PAs between the ages of 23 and 60.
Clustered on individuals. Full regression results can be found in Table
A5.

TABLE A5

PA Earnings Regression Results (Interval Regression)

       Variable        Coefficient Estimate    Coefficient Estimate
                                (z)                     (z)

NPs have controlled       -0.0141 (4.34)          -0.0146 (4.45)
substance authority
(NP_Rx)

Years since 1992 NPs                              -0.0012 (0.97)
have had controlled
substance authority
(NPRx_TimeSince)

NPs have                   0.0046 (0.99)          -0.0020 (0.36)
reimbursement
authority
(NP_ReimbAuth)

Years since 1992 NPs                               0.0027 (3.95)
have had
reimbursement
authority
(NPReimbAuth_
TimeSince)

PAs have controlled        0.0043 (1.07)           0.0057(1.41)
substance authority
(PA_Rx)

Years since 1992 PAs                              -0.0002(0.15)
have had controlled
substance authority
(PARx_TimeSince)

PAs have                   0.0041 (1.01)           0.0052 (1.11)
reimbursement
authority
(PA_ReimbAuth)

Years since 1992 PAs                              -0.0015 (1.40)
have had
reimbursement
authority
(PAReimbAuth_
TimeSince)

White                      0.0010 (0.33)           0.0010 (0.33)

Black                     -0.0177 (2.53)          -0.0177 (2.54)

Other race                   Dropped                  Dropped

Female                    -0.0821 (34.40)         -0.0821 (34.40)

Experience                 0.0233 (49.94)          0.0233 (49.93)

Experience squared        -0.0006 (29.19)         -0.0006 (29.18)

Federal government        -0.0995 (23.94)         -0.0994 (23.92
employee

State or local            -0.0773 (14.27)         -0.0772 (14.23)
government employee

Private firm                 Dropped                  Dropped
employee

Work in hospital           0.1789 (17.75)          0.1789 (17.73)
setting

Work in a private          0.1480(14.68)           0.148 (14.66)
health-care office
setting

Work in an                   Dropped                  Dropped
educational setting

Work in any other          0.1328 (13.14)          0.1326 (13.11)
setting

General practice          -0.0088 (1.43)          -0.0087 (1.42)
specialty

Surgical specialty         0.1058 (16.87)          0.1059 (16.89)

Pediatric specialty          Dropped                  Dropped

Internal medicine          0.0057 (0.93)           0.0058 (0.93)
specialty

Other medical              0.0645 (10.47)          0.0646 (10.48)
specialty

State real per             0.0020 (1.45)           0.0012 (0.78)
capita income
(000's)

Constant                  10.6855 (240.48         10.7555 (220.95)

                      [[chi].sup.2] = 16,307  [[chi].sup.2] = 16,317
                                        n = 95,570

                                     Years: 1996-2004

Note: State and year fixed effects are included with robust standard
errors.

TABLE A4
PA Earnings Data Summary Statistics

                     Variable                      Mean    Standard
                                                          Deviation

Real income (year 2000 dollars, middle of hand)   69,878    18,310

NPs have controlled substance authority (NP_Rx)    0.73      0.44

Years since 1992 NPs have had controlled           5.19      4.30
substance authority (NPRx_TimeSince)

NPs have reimbursement authority (NP_ReimbAuth)    0.52      0.50

Years since 1992 NPs have had reimbursement        4.05      4.53
authority (NPReimbAuth_TimeSince)

PAs have controlled substance authority (PA_Rx)    0.78      0.42

Years since 1992 PAs have had controlled           6.11      4.34
substance authority (PARx_TimeSince)

PAs have reimbursement authority (PA_ReimbAuth)    0.16      0.37

Years since 1992 PAs have had reimbursement        0.67      1.84
authority (PAReimbAuth_TimeSince)

White                                              0.82      0.38

Black                                              0.02      0.15

Other race                                         0.15      0.36

Female                                             0.50      0.50

Experience                                         9.11      7.90

Experience squared                               145.49    197.10

Federal government employee                        0.05      0.21

State or local government employee                 0.04      0.19

Private firm employee                              0.92      0.28

Work in hospital setting                           0.38      0.49

Work in a private health-care office setting       0.41      0.49

Work in an educational setting                     0.01      0.10

Work in any other setting                          0.20      0.40

General practice specialty                         0.33      0.47

Surgical specialty                                 0.23      0.42

Pediatric specialty                                0.03      0.18

Internal medicine specialty                        0.17      0.37

Other medical specialty                            0.24      0.43

State real per capita income                      29.160     3,880

y 1996                                             0.08      0.27

y 1997                                             0.09      0.28

y 1998                                             0.09      0.28

y 1999                                             0.12      0.32

y 2000                                             0.12      0.33

y 2001                                             0.13      0.34

y 2002                                             0.13      0.33

y 2003                                             0.14      0.34

y 2004                                             0.11      0.31

Alabama                                            0.01      0.08

Alaska                                             0.01      0.08

Arizona                                            0.02      0.14

Arkansas                                           0.00      0.04

California                                         0.07      0.25

Colorado                                           0.02      0.15

Connecticut                                        0.02      0.15

Delaware                                           0.00      0.05

District of Columbia                               0.00      0.07

Florida                                            0.05      0.22

Georgia                                            0.03      0.18

Hawaii                                             0.00      0.05

Idaho                                              0.01      0.08

Illinois                                           0.02      0.15

Indiana                                            0.01      0.09

Iowa                                               0.02      0.13

Kansas                                             0.01      0.12

Kentucky                                           0.01      0.11

Louisiana                                          0.01      0.08

Maine                                              0.01      0.11

Maryland                                           0.02      0.15

Massachusetts                                      0.02      0.15

Michigan                                           0.05      0.21

Minnesota                                          0.02      0.13

Mississippi                                        0.00      0.03

Missouri                                           0.01      0.09

Montana                                            0.01      0.07

Nebraska                                           0.01      0.12

Nevada                                             0.01      0.08

New Hampshire                                      0.01      0.08

New Jersey                                         0.01      0.09

New Mexico                                         0.01      0.09

New York                                           0.10      0.30

North Carolina                                     0.06      0.24

North Dakota                                       0.01      0.07

Ohio                                               0.02      0.15

Oklahoma                                           0.02      0.13

Oregon                                             0.01      0.10

Pennsylvania                                       0.06      0.24

Rhode Island                                       0.00      0.06

South Carolina                                     0.01      0.09

South Dakota                                       0.01      0.09

Tennessee                                          0.01      0.12

Texas                                              0.07      0.26

Utah                                               0.01      0.10

Vermont                                            0.00      0.06

Virginia                                           0.02      0.13

Washington                                         0.03      0.17

West Virginia                                      0.01      0.10

Wisconsin                                          0.03      0.17

Wyoming                                            0.00      0.06

                                                     N = 95,570


TABLE 4

Physician Earnings Estimation, Selected Results (Natural Log of
Earnings)

                                    Coefficient         Coefficient
                                    Estimate (t         Estimate (t
                                       Score)              Score)

                                      Column 1            Column 2

NPs have controlled substance       -0.012 (0.16)      -0.031 (0.39)
authority (NP_Rx)

Years since 1992 NPs have had                          -0.076 (2.47)
controlled substance authority
(NPRx_TimeSince)

NPs have reimbursement authority    -0.116 (0.91)      -0.084 (0.58)
(NP_ReimbAuth)

Years since 1992 NPs have had                          -0.015 (0.97)
reimbursement authority
(NPReimbAuth_TimeSince)

PAs have controlled substance        0.047 (0.45)       0.050 (0.46)
authority (PA_Rx)

Years since 1992 PAs have had                           0.081 (2.78)
controlled substance authority
(PARx_TimeSince)

PAs have reimbursement authority    -0.102 (0.99)      -0.118 (0.73)
(PA_ReimbAuth)

Years since 1992 PAs have had                          -0.003 (0.10)
reimbursement authority
(PAReimbAuth_TimeSince)

                                  [R.sup.2] = .1969  [R.sup.2] = .2004

                                                n = 3,243

                                            Years: 1996-2005

Notes: Values in bold indicate significant at the 5% level and t scores
are in parentheses. State and year fixed effects are included with
robust standard errors. Earnings are adjusted to year 2000 dollars.
Estimation incorporates CPS weights and includes full-time physicians
between the ages of 30 and 60. Full regression results can be found in
Table A7.

TABLE A6
Physician Earnings Data Summary Statistics

Variable                                           Mean    Standard
                                                          Deviation

Real income (year 2000 dollars)                  154,120   122,457

NPs have controlled substance authority (NP_Rx)    0.68      0.47

Years since 1992 NPs have had controlled           4.66      4.46
substance authority (NPRx_TimeSince)

NPs have reimbursement authority (NP_RcimbAuth)    0.49      0.50

Years since 1992 NPs have had reimbursement        3.77      4.57
authority (NPReimbAuth_TimeSince)

PAs have controlled substance authority (PA_Rx)    0.69      0.46

Years since 1992 PAs have had controlled           5.38      4.73
substance authority (PARx_TimeSince)

PAs have reimbursement authority (PA_ReimbAuth)    0.16      0.36

Years since 1992 PAs have had reimbursement        0.71      1.91
authority (PAReimbAuth_TimcSincc)

Female                                             0.26      0.44

Female x married                                   0.16      0.37

Female x children                                  0.12      0.32

Children                                           0.55      0.50

White                                              0.80      0.40

Black                                              0.05      0.22

Other race                                         0.15      0.36

Age                                               42.40      8.59

Age squared                                       1,872      738

Married                                            0.78      0.42

Self-employed                                      0.30      0.46

Federal government employee                        0.02      0.14

State or local government employee                 0.10      0.30

Private firm employee                              0.58      0.49

Work in hospital setting                           0.37      0.48

Work in an educational setting                     0.04      0.19

Work in a private health-care office setting       0.52      0.50

Work in any other health-care setting              0.04      0.20

Work in any other setting                          0.03      0.17

State real per capita income                      29,494     4.035

yl996                                              0.09      0.29

y1997                                              0.10      0.30

yl998                                              0.10      0.30

y1999                                              0.09      0.28

y2000                                              0.10      0.30

y200l                                              0.10      0.30

y2002                                              0.11      0.31

y2003                                              0.11      0.31

y2004                                              0.11      0.31

y2005                                              0.10      0.30

Alabama                                            0.01      0.12

Alaska                                             0.00      0.03

Arizona                                            0.01      0.11

Arkansas                                           0.01      0.09

California                                         0.09      0.29

Colorado                                           0.01      0.12

Connecticut                                        0.02      0.13

Delaware                                           0.00      0.05

District of Columbia                               0.00      0.05

Florida                                            0.06      0.23

Georgia                                            0.02      0.13

Hawaii                                             0.00      0.05

Idaho                                              0.00      0.05

Illinois                                           0.05      0.22

Indiana                                            0.02      0.13

Iowa                                               0.01      0.10

Kansas                                             0.01      0.10

Kentucky                                           0.02      0.13

Louisiana                                          0.01      0.11

Maine                                              0.01      0.07

Maryland                                           0.04      0.19

Massachusetts                                      0.03      0.18

Michigan                                           0.04      0.20

Minnesota                                          0.02      0.12

Mississippi                                        0.01      0.09

Missouri                                           0.02      0.14

Montana                                            0.00      0.04

Nebraska                                           0.00      0.07

Nevada                                             0.01      0.07

New Hampshire                                      0.00      0.06

New Jersey                                         0.03      0.16

New Mexico                                         0.00      0.07

New York                                           0.08      0.28

North Carolina                                     0.02      0.15

North Dakota                                       0.00      0.04

Ohio                                               0.04      0.19

Oklahoma                                           0.01      0.11

Oregon                                             0.01      0.09

Pennsylvania                                       0.07      0.25

Rhode Island                                       0.01      0.08

South Carolina                                     0.01      0.09

South Dakota                                       0.00      0.05

Tennessee                                          0.02      0.15

Texas                                              0.08      0.27

Utah                                               0.01      0.07

Vermont                                            0.00      0.05

Virginia                                           0.02      0.15

Washington                                         0.02      0.15

West Virginia                                      0.01      0.08

Wisconsin                                          0.02      0.14

Wyoming                                            0.00      0.03

Not in an MSA                                      0.12      0.33

Located in an MSA (100k-250k)                      0.05      0.21

Located in an MSA (250k-500k)                      0.11      0.31

Located in an MSA (500k-1,000k)                    0.15      0.35

Located in an MSA (l,000k-2,500k)                  0.29      0.45

Located in an MSA (2,500k-5,000k)                  0.15      0.36

Located in an MSA (more than 5,000k)               0.14      0.35

                                                      N = 3,243

Notes: Summary statistics weights with CPS weights.
TABLE A7
Physician Earnings Regression Results

            Variable              Coefficient         Coefficient
                                  Estimate (t)        Estimate (t)

NPs have controlled substance     -0.0124(0.16)      -0.0315 (0.39)
authority (NP_Rx)

Years since 1992 NPs have had                        -0.0764 (2.47)
controlled substance authority
(NPRx_TimeSince)

NPs have reimbursement            -0.1165 (0.91)     -0.0840 (0.58)
authority (NP_ReimbAuth)

Years since 1992 NPs have had                        -0.0157 (0.97)
reimbursement authority
(NPReimbAuth_TimeSince)

PAs have controlled substance      0.0474 (0.45)      0.0495 (0.46)
authority (PA_Rx)

Years since 1992 PAs have had                         0.0812 (2.78)
controlled substance authority
(PARx_TimcSince)

PAs have reimbursement            -0.1022 (0.99)     -0.1178 (0.73)
authority (PA_ReimbAuth)

Years since 1992 PAs have had                        -0.0033 (0.10)
reimbursement authority
(PAPayReimb_TimeSince)

Female                            -0.2459 (2.42)     -0.2339 (2.36)

Female x married                  -0.0588 (0.45)     -0.0707 (0.55)

Female x children                 -0.0510 (0.43)     -0.0511 (0.43)

Children                           0.2518 (4.82)      0.2509 (4.75)

White                              0.0205 (0.37)      0.0136(0.24)

Black                             -0.0076 (0.07)     -0.0247 (0.24)

Other race                           Dropped            Dropped

Age                                0.1572 (6.54)      0.1568 (6.45)

Age squared                       -0.0015 (5.18)     -0.0014 (5.09)

Married                            0.0618 (0.96)      0.0663 (1.03)

Self-employed                     -0.1346 (2.09)     -0.1418 (2.19)

Federal government employee       -0.1347 (1.34)     -0.1407 (1.38)

State or local government         -0.1852 (3.22)     -0.1774 (3.09)
employee

Private firm employee                Dropped            Dropped

Work in hospital setting           0.0482 (0.52)      0.0546 (0.59)

Work in an educational setting       Dropped            Dropped

Work in a private health-care      0.3022 (3.22)      0.3101 (3.29)
office setting

Work in any other health-care      0.1496 (1.18)      0.1496(1.18)
setting

Work in any other setting          0.0517 (0.40)      0.0608 (0.47)

Not in an MSA                     -0.0168 (0.16)     -0.0204(0.19)

MSA 100                           -0.0133 (0.12)     -0.0168 (0.16)

MSA 250                            0.0073 (0.06)      0.0097 (0.08)

MSA 500                           -0.1049 (0.82)     -0.1083(0.84)

MSA 1,000                          0.0413 (0.46)      0.0349 (0.39)

MSA 2,500                         -0.0466 (0.46)     -0.0433 (0.43)

MSA 5,000                            Dropped            Dropped

Stale real per capita income       0.1261 (2.74)      0.1504(2.80)
(000's)

Constant                           4.8772 (4.27)      4.4163 (3.66)

                                [R.sup.2] = .1969  [R.sup.2] = .2004

                                              n = 3,243

                                           Years: 1996-2005

Notes: State and year fixed effects are included with robust standard
errors. CPS sample weights employed.


TABLE A8
F Statistics of Testing Joint Significance in First Stage of
Instruments of State-level Practitioner Authority

                             Variable

Practitioner      NP            NP           PA           PA
   Sample     Controlled  Reimbursement  Controlled  Reimbursement
              Substance     Authority     Substance    Authority
              Authority                  Authority

NPs             38.85         68.82         38.04         12.89
PAs            714.82       1491.55       1887.58       1122.81
Physicians      46.76        760.39        125.01        234.36
Gale Copyright:
Copyright 2009 Gale, Cengage Learning. All rights reserved.