Wage Dispersion and Firm Financial Performance: Evidence from

Wage Dispersion and Firm Financial Performance: Evidence
from Non-profit Hospitals
G. Nathan Dong*
Columbia University
March 20, 2013
ABSTRACT
Does wage dispersion provide incentives for competition or perceptions of unfairness among
workers? Prior literature suggests that pay distribution and pay equity are important factors
incentivizing employees as in the models of tournament and cohesive cooperation. If non-profit
organizations rely more on intrinsically motivated employees because they can not afford
paying higher salary than their for-profit counterparts, they will be more likely to exhibit wage
dispersion as part of their organizational strategies. The employee wage dispersion data from 48
non-profit hospitals in Maryland provides an ideal test bed to examine the relation between
worker wage dispersion and firm financial performance. Using this data set, along with hospital
financial statements, this paper finds a positive effect of wage dispersion among low-skill
workers on profitability and a negative effect of wage dispersion among high-skill workers on
profitability. Interestingly, the staff size in each skill category tends to offset the aforementioned
effects.
Keywords: Wage dispersion, pay inequality, worker incentive, non-profit organization, profit
JEL Codes: G30, J24, J31, L31, M12
__________________________
* Dept. of Health, Policy & Management, Columbia University. 600 W 168th Street, New York, NY 10032. Tel: 212342-0490. E-mail: [email protected]. No potential conflict of interest relevant to this article was reported.
“Disparities in workers' pay are more complex than a difference in gender…, a lay person could
see a disparity, but it could have nondiscriminatory reasons.”
--- John Sarno, the President of Employers Association of New Jersey
I. INTRODUCTION
It has been widely recognized that issues related to the wage inequality and firm performance
are of central policy and business importance, with the pay dispersion encouraging employees
exerting maximum efforts (tournament) on the one hand and the perceived unfairness offsetting
the positive incentive effects (cohesiveness) on the other. Given that prior research has mainly
relied on data of for-profit firms and the fact that most for-profit firms has various strict
workplace rules prohibiting disclosing salary information among co-workers, 1 it is indeed
paradoxical to understand how tournament or cohesiveness affects the behavior of
“uninformed” workers.2 Clearly, an understanding of how these two factors interact is crucial
to the creation of an efficient compensation policy in a firm; however, existing literature’s
dependence on for-profit firms’ wage information is not able to disentangle the relative
importance of these two theoretical effects on incentives of the “informed” employees.
If non-profit organizations rely more on intrinsically motivated employees because they
can not afford paying higher salary than their for-profit counterparts (Preston 1989),3 they will
be more likely to exhibit wage dispersion as part of their organizational strategies. 4 The
employee wage dispersion data from 48 non-profit hospitals in the state of Maryland provides
an ideal test bed to examine the relation between worker wage dispersion and firm financial
performance. All hospitals in Maryland are required to report their worker wage distribution
information to the Health Services Cost Review Commission (HSCRC), and we argue that this
The Paycheck Fairness Act introduced to the Congress in 2009 would allow employees to disclose salary
information with co-workers despite workplace rules prohibiting disclosure; however, this legislation which aims to
expand the scope of the Equal Pay Act of 1963 and the Fair Labor Standards Act has been twice rejected by the
Congress.
2 Nosenzo (2012) studies the disclosure of wage information and finds it detrimental for effort provision if employees
think they are treated unequally.
3 Wages might be low in non-profit organizations because these firms can not distribute net earnings (Hansmann
1980), or employees are willing to accept low compensation to perform socially desirable activities (Frank 1996), or
these non-profit organizations are in high competitive industries (Lakdawalla and Philipson 2006).
4 See Leete (2000) for more discussions on the worker motivation in the non-profit sector. The author uses the US
Census dataset, rather than a linked employer-employee dataset like the one used in this paper, to compare wage
equality between the non-profit sector and the for-profit sector.
1
2
regulation makes easier for general public to obtain salary information about hospital
employees.
In contrast to most other studies, which in most cases pool all workers of different job
functions in one regression, we are able to examine a specific labor market with detailed
classification of job functions and skills. We therefore also analyze the interrelation between the
dispersion of wage across different job functions based on skill levels next to the dispersion of
wage levels and firm performance.5 Our findings provide insights of the relative importance of
fairness considerations resulting from monetary inequality in firms compared to incentive
effects. The results can be considered to generate recommendations for firms’ wage policy.
It is worth noting that despite the advantages of this new data set, it suffers limitations
in that we end up studying a rather small group of employees, who could be selected into
various job categories, and not representing the worker compensation nationwide. Although
state economic situations in Maryland clearly influence the behavior of hospitals and hospital
employees, we believe that with focusing only on those having better information about
workers compensation should distinguish this research from prior literature that mainly reveal
the incentive effects in the private sector. Nonetheless, we advise readers to exercise caution
when interpreting such results.
The remainder of the paper is organized as follows. Section II reviews the relevant prior
research on the two contrasting effects of pay dispersion in general. Section III presents the
sample data and measurement choice. Section IV introduces the empirical method. Section V
evaluates the results. Section VI conducts the robustness tests. Section VII concludes.
II. RELATED LITERATURE
The publication of Lazear and Rosen (1981) and Akerlof and Yellen (1988, 1990) highlighted the
importance of worker wage distribution for the firms with one (tournament incentive) arguing
that a more differentiated wage structure encourages workers to exert maximum effort by
awarding large prizes to the most productive ones,6 and the other (income fairness) arguing that
wage compression or pay equity improves labor cohesiveness among workers.7 Since that time,
Groshen and Krueger (1990) examine the BLS data on 300 hospitals in 1985 and find a phenomenon of wageequality within hospitals that if a hospital paid relatively high wage to one occupation, it was likely to pay high
wages to workers in other occupations as well.
6 Also see Rosen (1986).
7 For other papers on fairness and equity considerations, see Fehr and Schmidt (1999), Levine (1991), and Milgrom
and Roberts (1990).
5
3
researchers have presented a variety of evidence how wage dispersion affects firm performance;
however, there is no consensus regarding the precise impact and which theory dominates the
other, in terms of explaining the observed patterns in the data: Lazear (2000) find that monetary
incentives matter for individuals and Abrevaya (2002), Becker and Huselid (1992), Bull, Schotter
and Weigelt (1987), Ehrenberg and Bognanno (1990), Main, O’Reilly and Wade (1993) report
that reward structure affects individuals’ effort. Camerer and Thaler (1995) and Cowherd and
Levine (1992) suggest that fairness considerations influence human behavior, and Charness and
Kuhn (2007) reveals no evidence in a laboratory experiment that coworkers’ wages actually
affect workers’ effort and productivity at all.
Although the argument that wage policies affect worker behavior is not new (Simon
1957), the literature examining the connection between wage dispersion and firm performance
is indeed recent. Lallemand, Plasman and Rycx (2004) and Heyman (2005) find a positive link
between the dispersion of wage levels and firm performance for Belgium and Sweden
respectively. Winter-Ebmer and Zweimüller (1999) argue that a high wage level in a firm
reflects high firm performance. They find an inversely U-shaped interrelation between wage
dispersion and the level of wages for white collar employees in Austrian firms. Similar evidence
is reported in Franck and Nüesch (2011) using German data and Yang and Klaas (2011) using
Korean data. Bloom (1999) shows that the level of wage dispersion among major league baseball
team members is negatively related to several measures of individual and team performance.
Pfeffer and Langton (1993) find decreasing research productivity and collaboration among
college and university faculty with increasing wage dispersion. There are some papers only
investigating wage distribution among senior managers, for example, Eriksson (1999) finds a
positive relationship between the pay spread among executives and firm profitability in
Denmark, whereas Leonard (1990) and O’Reilly, Main and Crystal (1988) do not find such
results among managers in U.S. firms. Beaumont and Harris (2003) use the ratio of non-manual
and manual labor costs per employee as a proxy for wage dispersion to overcome the lack of
individual wage information, and find a positive association between worker wage dispersion
and value-added per employee in UK manufacturers. To some extent, our measure of wage
dispersion using the average value of hourly wage ranges at the job function level is technically
better than the rough measure used in Beaumont and Harris (2003) because there are 65 job
functions within each hospital in our sample, and we believe the distribution mean of this
sample size for each hospital is statistically meaningful. In addition to the cross-sectional
4
evidence, there are other papers studying aggregate time-series data. Hibbs and Locking (2000)
find more positive than negative effects of wage dispersion on firms’ real value added in
Swedish. Bingley and Eriksson (2001) examine the skewness of intra-firm wage distribution,
and find a U-shaped relation with firm productivity in Denmark. Interestingly, Grund and
Westergaard-Nielsen (2008) show a negative association between the dispersion of wage
growth and firm performance in Denmark, but is mainly driven by white-collar rather than
blue-collar workers. The authors argue that the dispersion of wage growth is more revealing
than the dispersion of wage levels for purposes of comparing wage dispersion’s positive
incentive effects with its adverse morale effects.
The brief review of prior literature shows that empirical evidence, however, is mixed in
regards to whether wage dispersion improves firm productivity. Given that the majority of
prior research only studies the economic consequences of wage structure using data of forprofit companies in various industries and wage differentials between blue and white collar
workers, this article seeks to present a contribution by focusing on workers with different
skills,8 and on firms that usually do not offer employees other incentive schemes such as bonus,
profit sharing or stock options: not-for-profit hospitals. Such an update in the literature is
critical to understanding how pay dispersions at different skill levels affect the overall
productivity of firms that disproportionally rely on intrinsically motivated employees.
The two strands of the literature obviously predict contradicting results for the
interrelation between wage dispersion and firm performance. Tournament theory predicts
increasing effort levels with increasing wage premiums for winners of rank-order tournaments.
Hence, incentives are induced by monetary rewards to the winners. The effect of monetary
incentives on firm productivity depends on the kind of relevant production function
transforming effort to production and the group of people who are more sensitive to monetary
incentives. The simplest one is to aggregate the effort of individual workers to the productivity
of the firm, as suggested by Adams (2006). However, different teams within the firm might
response to monetary incentives differently due to the nature of their job functions. For low-pay,
blue-collar or low-skill workers, monetary rewards might be their primary, if not only,
Li (2013) develops a theoretical model to study job mobility and wage dispersion under asymmetric information
and suggests that the wage distribution becomes more spread out (greater pay inequality) if technological change is
skill-biased and favors general skills over firm-specific skills. One of the empirical implications of this model is a
positive association between firm productivity and wage dispersion among workers with general (non-firm-specific)
skills. If firm financial performance represents skill-biased technological change and the group of low-skill employees
represents general-skill workers, our results provide empirical evidence supporting this theory; however, the
causality is in the opposite direction by assumption.
8
5
motivation; this type of production function implies that the observed productivity effect is
expected to be increasing in the overall effort. However, for high-pay, white-collar, or high-skill
workers in a not-for-profit organization, their motivation might well be based on a high degree
of work-group cohesion or the welfare of their customers rather than monetary incentives; as a
result, they might exert effort that can be equally detrimental to productivity, such as
complaining to co-workers, ingratiation, duplicity, verbal aggression, bad-mouthing, or even
antagonistic exit. This argument suggests a negative link between the wage dispersion and firm
performance.
Until now, there is no definite empirical evidence about this issue whether high- and
low-skill workers response to wage dispersion differently due to their different perceptions to
tournament incentives and pay unfairness. It is an empirical question which effect dominates in
practice. We will study this interrelation in the following empirical examination. Regressing the
dispersion of wage on firm performance controlling for worker skills, we can examine the
empirical shape of the relationship. Then we can check, whether there is a positive or negative
relationship between wage dispersion and firm performance for different group of employees
in these non-profit organizations.
Using the data set of hospital employee wage dispersions, along with hospital financial
statements, we provide new evidence on quantitatively understanding how different aspects of
pay dispersions, worker skills, and staff sizes influence firm performance. Specifically, we find a
positive effect of wage dispersion among low-skill workers on firm financial outcomes and a
negative effect of wage dispersion among high-skill workers on outcomes. More interestingly,
there is an offsetting effect from the staff size of each skill type that mitigates the associations
between pay distribution and firm performance.
III. DATA
Our primary data source is the Wage and Salary Survey Results, a self-reported administrative
data set that contains information on maximum, minimum and average wages for 64 categories
of hospital workers. This unique panel data originates from the Maryland Health Services Cost
Review Commission (HSCRC) and is used for disclosure and served as the basis for the Labor
Market Adjustor to account for differences in area wage levels. The important feature of this
data set is that it is possible to associate workers’ wages with their employers. The wage
dispersion data cover all employees who are not medical doctors and managing executives in 48
6
hospitals during the 2010 fiscal year, and include a detailed breakdown of 64 job functions.9
Employers are defined by their hospital identification number, and job categories are defined by
their job function code.
To study the long-term effects of the wage distribution on firm performance, we obtain
the hospital financial statements for the fiscal year of 2012 from HSCRC. We believe using the
firm performance data in the same year (2010) or the following year (2011) to study the longterm economic consequence of worker pay dispersion does not provide enough basis to draw
adequate conclusions simply because it takes time for people to learn about the wage
differential between coworkers to be incentivized or discouraged. Although 2-year can be
hardly considered as long-term, it does provides a reasonable compromise between providing
the lagged performance measure and avoiding any confounding effects from changing
economic and market conditions. Nevertheless, we can examine the financial performance of
these hospitals in the following three or even five years, so long as the financial statements
become available in the future.
Our dependent variable is the financial profit of the hospitals. We use two different
ratios to measure firm performance: profitability. The first one is the operating margin, the ratio
of the operating income divided by the total revenue, and the second one is the net profit
margin, defined as the net income divided by the total revenue:
OperatingMargini 
OperatingIncomei
Revenuei
NetParofitMargini 
(1)
NetIncomei
Revenuei
(2)
It is often argued that net profit margin is not a reliable profitability measure because net
income (numerator in net profit margin) is more likely to be manipulated by company
executives than operating income (numerator in operating margin); therefore, we use both
measures as dependent variables to examine the financial outcomes of hospitals. The formal
definitions and descriptions of these firm performance measures are reported in Table I.
[Insert Table I Here]
Though some doctors are hospital employees, most doctors are not. Non-employee doctors are independent
contractors. This data set does not include either employee or non-employee doctors.
9
7
The summary statistics in Table II illustrates that the average net profit margin is 3.1% with a
standard deviation of 5.9%, whereas the average operating profit margin is 1.1% with a
standard deviation of 5.5%. It appears that hospitals in the state of Maryland are more
profitable than the rest of hospitals in the U.S. For all U.S. hospitals that reported their 2010
financial statements to the Centers for Medicare & Medicaid Services (CMS), the average net
profit margin of is 1.4% and the operating margin is -1.5%. 10 Additionally, the number of
employees can be used as a measure of firm size. The total assets can be considered as the size
of a firm’s financial and fixed capital, whereas the number of employees can be treated as the
size of a firm’s human capital, and it is commonly used in the economics literature when
alternative measures of firm size such as the total assets is not available (Cabral and Mata 2003).
[Insert Table II Here]
Data on employee wage rate are aggregated to 64 different job functions by HSCRC. For
each hospital we obtain the maximum, minimum, and average rate of hourly wage and
calculate the pay dispersion or range for each job function:
Dispersioni  MaximumRatei  MinimumRatei
(3)
This measure of pay dispersion or range is sometimes called the spread of pay as in Gupta et al.
(2012) and used widely in the prior literature including Becker and Huselid (1992), Kepes et al.
(2009), Shaw et al. (2002), and Leonard (1990). The mean of the pay dispersion or range as
reported in Table II is 0.42 with a standard deviation of 0.15. It should be cautioned that using
range to measure dispersion is very sensitive to outliers because range is determined by the
furthest outliers at either end of the distribution, hence reflecting information about extreme
values but not necessarily about typical values. However, when the range is narrow, meaning
that there are no outliers, it does reflect information about typical values in the data. Given the
minimum pay dispersion of 0.16 and the maximum dispersion of 0.69, we argue that range
provides similar measure of dispersion in this sample of 48 hospitals.
By matching the worker compensation data to the hospitals’ financial statements, we
construct a set of variables to control for hospital characteristics. The standard measure of firm
size in this article and in most of the relevant literature is the total assets, which is reported in
the balance sheet of a hospital’ financial statements. To avoid problems of skewed distribution
10
This is based on the data at the Healthcare Cost Report System (HCRIS).
8
of firm size and potential outliers that may bias the regression results, we use a natural
logarithm transformation of the total assets to normalize its distribution:
(4)
AssetSizei  log(Total Assetsi )
As mentioned earlier, the staff size is an alternative measure of firm size, and we include the
number of employees with a unit of measure of one thousand as a control variable:
StaffSizei 
Numof Employeesi
1, 000
(5)
Financial leverage is the degree to which a firm (or shareholders) is utilizing borrowed
money, mainly in the form of debt financing. Leverage magnifies increases in earnings during
periods of rising operating income due to the tax benefits of debt, but adds significant risks for
stockholders and creditors because of added interest obligations. Because companies that are
highly leveraged may be providing better return to investments and having higher probability
of bankruptcy at the same time, leverage has been commonly used in the corporate literature as
a measure of a firm’s risk-taking behavior. For hospital i, we define the financial leverage as its
debt-to-asset ratio:
Leveragei 
Debti
TotalAsseti
(6)
For this sample of 48 hospitals, the mean leverage is 0.58 with a standard deviation of 0.24.
The current ratio is a financial ratio that measures whether or not a firm has enough
resources to pay its debts over the next year. It compares a firm's current assets to its current
liabilities:
Liquidityi  CurrentRatioi 
CurrentAsseti
CurrentLiabilitiesi
(7)
The current ratio is an indication of a firm's market liquidity and ability to meet creditor's
demands.
The Pearson’s correlations are reported in Table II.
[Insert Table II Here]
An examination of the correlation matrix indicates that correlations between independent
variables are generally smaller than 0.7 except one case of the logarithm of total assets and
logarithm of number of employees (0.61) due to the fact that both of these two variables
measure the size of hospitals. The low correlation among the covariates helps prevent the
9
problem of multicollinearity that causes high standard errors and low significance levels when
both variables are included in the same regression. Further diagnostics indicate no obvious
evidence of serious multicollinearity among the covariates.
IV. METHODOLOGY
We estimate the pay dispersion’s effect on the firm performance by fitting hospital-level
regression equations that take the form:
Profitability i     1  Sizei   2  Leveragei   3  Liquidity i   4  StaffSizei   5  Dispersioni   i (8)
Here the profitability of the hospital is measured by two different variables: net profit margin
and operating margin. Pay dispersion, which is the ratio of pay range and average wage rate, is
the variable of interest in this regression analysis. To control for hospital i-specific effects, we
employ control variables to account for firm size, financial leverage, and operating liquidity.
However, there might be differences across hospitals that are not captured by the control
variables and that affect firm performance and the wage dispersion simultaneously. This may
lead to biased and inconsistent parameter estimates; therefore, we add firm (hospital) fixedeffects to the models, and examine the sensitivity of these regression estimates to the inclusion
and exclusion of the firm fixed-effects.
Still, even with the firm fixed-effects, the regression estimates have two key limitations.
First, to interpret the coefficient (λ) as the causal effects of the pay dispersion, or at least to argue
the coefficient estimate is unbiased, one must assume that the pay dispersion is uncorrelated
with unobserved determinants of firm performance as measured by the hospital’ financial
profitability (εi). Second, fitting this regression model using hospital-level aggregate data is
prone to the influence of outliers because essentially pay dispersion is the range between the
maximum and minimum wage rates scaled by the average wage rate. Remember, there are 48
hospitals in the state of Maryland, meaning a very large maximum wage rate or a very small
minimum wage rate in one hospital could affect the coefficient estimates in a dramatic way. To
circumvent the problem of small sample bias, we aggregate the data by the job functions. Table
IV illustrates the breakdowns of the job functions based on their titles and functions. We rank
the importance and complexity of these job functions, and categorize them to two skill levels
based on the ranking: high and low. After obtaining the number of employees and wage
dispersion for both high-skill and low-skill categories, we conduct the following regression
analysis with and without the firm fixed-effects:
10
Profitability i     1  Dispersioni  Sizei   2  Leveragei   3  Liquidity i   4  LowSkillStaffSizei
  5LowSkillPayDispersioni   5  HighSkillStaffSizei   6 HighSkillPayDispersioni   i
(9)
The model specification is very similar to the previous equation (8) but with the breakdowns of
low- and high-skill workers based on the job functions as defined in the Table IV to study in
more detail how wage distribution affect firm performance.
[Insert Table III Here]
To identify the underlying forces of this effect, we further break down the wage dispersion into
low-, mid- and high-skill worker groups, and run the following regressions with and without
hospital fixed-effects:
Profitability i     1  Dispersioni  Sizei   2  Leveragei   3  Liquidity i   4  LowSkillStaffSizei
  5LowSkillPayDispersioni   5  MidSkillStaffSizei   6 MidSkillPayDispersioni
  7  HighSkillStaffSizei   8 HighSkillPayDispersioni   i
(10)
As noted in the data section, this is an aggregate dataset that provides the distribution
information of employee wage at the job function level for each hospital in the state of
Maryland. If we choose to conduct the analysis using the data as is, there will be an important
concern confounding the results obtained in the regressions: there is no separate performance
measure for each job function within the organization. We can proxy it with the firm
performance variable as in equation (9); however, it will be troublesome to interpret the
coefficients because the impact of wage dispersion from each of the 64 job functions on firm
performance might be too small to be considered as a significant economic effect. Therefore, we
created the aforementioned samples that category workers to two and three types (low, mid,
and high) and aggregate the job function-level data to firm-skill-level data before running
regressions of equations (9) and (10).
V. RESULTS
We first discuss the results using firm-level aggregate data. In these specifications, we estimate
the effects of the wage distribution of all employees within the firm on the future financial
performance of the firm. Table IV reports coefficient estimates for the covariates of interest and
11
other control variables, by which we account for firm size, financial leverage, and operating
liquidity.
[Insert Table IV Here]
We find negative and highly significant effects of employee wage dispersion on
operating margin in specification (1). In addition, larger firms as measured by the number of
employees, tend to have better financial outcomes, which is consistent with the findings in the
previous studies including Abowd, Kramarz and Margolis (1999), Brown and Medoff (1989),
Ferrer and Lluis (2008), and Oi and Idson (1999). The fixed-effects estimation in specification (2)
confirms such effect of wage dispersion even with hospital specific factors. Taking this result at
face value, a natural interpretation is that wage dispersion within a firm might have fostered a
negative perception of unfair employment practices; hence reduced the workers’ willingness to
cooperate with other workers; as a result, wage dispersion increased transaction costs and
decreases efficiency. When we change the firm performance measure to net profit margin in
specifications (3) and (4), the coefficient estimates become insignificant for both specifications
with and without the fixed-effects. It is not surprising to see the vanishing significance of the
influence of wage dispersion on firm performance. As mentioned earlier, there is evidence in
the existing literature supporting both theories of the positive effect as shown in Beaumont and
Harris (2003), Eriksson (1999), Heyman (2005), Lallemand, Plasman and Rycx (2004), and
Winter-Ebmer and Zweimüller (1999) and the negative effect as reported in Bloom (1999),
Charness and Kuhn (2007), Grund and Westergaard-Nielsen (2008), Leonard (1990), O'Reilly,
Main and Crystal (1988), and Pfeffer and Langton (1993).
To test the hypothesis that high-skill workers are more concerned about the pay equity
and welfare and low-skill workers are more likely to be incentivized by monetary rewards, we
break down the workers in each firm to low- and high-skill categories as discussed in the
previous methodology section, and report the parameter estimates in Table V.
[Insert Table V Here]
Notice that the sign on the regressor of wage dispersion of the low-skill workers is statistically
positive across all four specifications, suggesting that this group is very sensitive to the
12
incentive value of the monetary reward and that the wage gap among the workers is large
enough to be of considerable interest. The fact that the interaction term between staff size and
wage dispersion has a negative coefficient implies a crowding-out behavior; that is, as the
number of workers increase in this group, monetary incentives are being crowded out and
fairness considerations are becoming stronger and stronger for these low-skill workers. For
high-skill workers the situation is reversed. The coefficient of the wage dispersion of this group
of mostly highly paid white-collar workers is statistically significantly positive. This result is
somewhat surprising, given that bonus, stock and options have been widely used in
compensation contracts of high-tech workers, middle managers and senior executives. We must
remember, however, that these hospitals in our samples are not publically-trade companies (not
even for-profit companies) but, instead are non-profit organizations. Thus, the most plausible
interpretation is that these high-skill workers in hospitals, mostly technicians, senior nurses,
and mid-managers, primarily care about maximizing the patients’ welfares rather than the
hospitals’ financial outcomes. More interestingly, the interaction term between staff size and
wage dispersion is negative, suggesting that as the size of this group increases the negative
effect of wage inequality die away.
Table VI gives results for the model specifications that are similar to that reported in
Table V, but with 3-level skill breakdowns: low, mid, and high.
[Insert Table VI Here]
With this more detailed data on worker categorization for each firm, we can not only reduce the
risk of omitted variable bias, but also assess the robustness of the previously obtained results.
The trade-off is that we will lose two degrees of freedom for the additional estimators: staff size
and wage dispersion of mid-skill workers. For this sample of 144 observations (3 years data of
48 hospitals), the concern is whether the improvement in fit can outweigh the loss of the degree
of freedom. As expected, the adjusted R-squares in all four specifications are smaller than the
ones reported in Table V; however, the estimated coefficients and statistical significances of the
independent variables remain almost unchanged. Interesting, the staff size and wage dispersion
of the mid-skill worker group have no impact on firm performance, confirming the size and
dispersion effects from the high- and low-skill worker groups. This result holds for both
13
measures of financial outcomes (operating margin in specifications 1 and 2, and net profit
margin in specifications 3 and 4) and with the fixed effects estimation in specification (2) and (4).
It is worth mentioning that the wage dispersion over years in firms is also affected by
retirees who recently left the company and newcomers who just joined the company. We are
aware of the fact that our study is primarily focused on incumbents, because employee turnover
in non-profit organizations is relatively low: the average job turnover rate among non-profit
employers is 21 percent, less than half the average of their for-profit counterparts (Philanthropy
Journal 2008).
VI. ROBUSTNESS
It should be noted that the economic interpretation of statistical significance in correlations
between household financing choices and personal characteristics deserves caution because the
empirical results reported in the previous section could be driven by endogeneity concerns.
Specifically, there might be significant omitted variable(s) correlated with both wage dispersion
and firm performance driving our results spuriously. One possible omitted variable is the
personal characteristics such as age, gender and wealth. Younger, female, or poorer workers are
more likely to have lower wage than their older, male or richer counterparts in the same
organizations. To specifically address this endogeneity, we need to separate the worker groups
that have “exogenous” wage dispersion from those that have “endogenous” wage dispersion.
Specifically, we run the regression as specified in equation (11) using the wage data of practical
nurses and nurse managers of type A, both of whom are most likely to be female.
Profitability i     1  Sizei   2  Leveragei   3  Liquidityi   4  StaffSizei   5  Dispersioni
  6  NurseWageDispersioni   7  ManagerWageDispersioni   i
(11)
The results reported in Table VII show a negative effect of wage dispersion among nurse
managers on profitability, providing supporting evidence for the hypothesis that high-skill
workers are more concerned about the pay equity and welfare. Unfortunately, due to the
statistically insignificant coefficient estimate for the wage dispersion variable of practical nurses,
we are not able to make a meaningful inference about the second half of the hypothesis that
low-skill workers are more likely to be incentivized by monetary rewards.
[Insert Table VII Here]
14
Another endogeneity concern could be that wage dispersion is the outcome of a firm’s
financial performance. One argument can be that firms with higher earnings are likely to offer
tournament incentives to their employees, and the other argument can be that firms with lower
earnings are likely to compress wage dispersion. To address this issue, we identify the 2008
financial crisis as an exogenous shock, and assume that this national economic crisis increased
the unemployment rate and increased the wage distribution temporarily.11 In fact, we find that
the pay dispersion of hourly wage among all employees increased from $15.5 in 2008 to $16.1 in
2010. 12 For low-skill workers, the dispersion increased from $9.8 to $10.5, whereas for high-skill
workers, it increased from $19.3 to $20.2. During the same period, the median profit margin of
all Maryland hospitals increased from 0.92% in 2008 to 5.04% in 2010, suggesting that the
expansion of wage gap might have improved productivity in terms of financial profitability
among the sample hospitals. The question which remains, however, is whether the opposite
causal relationship holds true, i.e. when the wage gap shrinks, do workers change their
behavior. To address this question, we will need to identify a negative shock. If and when such
events occur, future research will be valuable to understanding workers’ responses to wage
dispersion shrinkage, particularly if a particular employee group, of low- or high-skill, that
reacts strongly that eventually affects the financial outcomes.
VII. DISCUSSIONS AND CONCLUSION
Despite a large body of theoretical and empirical papers on how wage dispersion and pay
fairness might be part of organizational strategies to incentivize or discourage employees to
maximize their efforts, prior literature of examining the relation between wage dispersions at
different skill levels and firm financial outcomes is scarce, and little research has been done into
aspects of the staff size effects among workers of different skill levels on firm performance. This
paper is to take a step further to answer these two interesting questions using a data set of
employee wage dispersion from 48 non-profit hospitals in Maryland. It has been argued that
non-profit organizations rely more on intrinsically motivated employees because they can not
afford paying higher salary than their for-profit counterparts. If this is indeed the case, these
The U.S. unemployment rate increased from 5.8% in 2008 to 9.6% in 2010, according to Bureau of Labor Statistics
data.
12 One potential argument against the assumption of the financial crisis causing the wage dispersion to rise is that
managers might avoid wage reductions in slumps, a phenomenon commonly known as wage rigidity (Campbell and
Kamlani 1997, Bewley 1999, Zeoga and Karlsson 2006, Radowski and Bonin 2010); however, our sample indicates a
widening wage gap after the crisis.
11
15
not-for-profit firms will be more likely to exhibit wage dispersion as part of their organizational
strategies; hence this unique data set of Maryland hospitals provides an ideal test bed to
examine the relation between wage dispersion and firm performance.
Wage dispersion is an issue that has attracted considerable debate amongst practitioners,
academics, and politicians in recent years. In part, this has reflected a desire to understand the
factors behind the worker incentive design and pay equity in both for-profit firms and nonprofit organizations. However, there is also a growing recognition that the relationship between
wage distribution and firm performance depend on the skills of the workforce and the
uncertainty of the firm economic environment, which could potentially dilute the results and
conclusions obtained from prior research.
Given that the majority of prior research only studies the economic consequences of
wage structure using data of for-profit companies in various industries and wage differentials
between blue and white collar workers, this article seeks to present a contribution by focusing
on workers with different skills, and on firms that usually do not offer employees other
incentive schemes such as bonus, profit sharing or stock options: not-for-profit hospitals. Such
an update in the literature is critical to understanding how pay dispersions at different skill
levels affect the overall productivity of firms that disproportionally rely on intrinsically
motivated employees.
Using the data set of hospital employee wage dispersions, along with hospital financial
statements, we provide new evidence on quantitatively understanding how different aspects of
pay dispersions, worker skills, and staff sizes influence firm performance. Specifically, we find a
positive effect of wage dispersion among low-skill workers on firm financial outcomes and a
negative effect of wage dispersion among high-skill workers on outcomes. More interestingly,
there is an offsetting effect from the staff size of each skill type that mitigates the associations
between pay distribution and firm performance: The positive effect of wage dispersion among
low-skill workers on firm outcomes is reduced when the number of low-skill workers increases.
Similarly, the negative effect of wage dispersion among high-skill workers on outcomes is
reduced when the number of high-skill worker increases. One explanation of this negative
offsetting effect from the staff size among low-skill workers could be the consequence of “public
wages” as in Nosenzo (2012). The likelihood of wage distribution becoming public information
to the low-skill workers may increase with the size of the workforce. When workers learn that
they are underpaid relative to co-workers, they tend to exert less effort; hence, pay fairness
16
becomes an important concern in this case. One response to this phenomenon for the managers
is to reduce the wage dispersion within this worker group. For the high-skill workers, this
positive offsetting effect from the staff size can be due to “monitoring difficulty” as in Garen
(1985). It becomes more difficult to monitor the high-skill workers, mainly senior assistants,
technicians and mid-managers, when the size of this workforce increases in the hospitals. As a
result, senior managers are less likely to detect the subtler aspects of worker quality and their
exerted efforts; therefore, they have to rely more on tournament incentives to motivate highskill employees.
The correlations between wage dispersion, staff size, skill level, and firm performance
do not necessarily imply causality. However, the evidence reported in this paper using lagged
financial outcomes in hospitals do suggest that pay distribution polices, such as high dispersion
for a small group of low-skill workers and low dispersion for a large group of high-skill
workers, should be preferred to policies that only concern about tournament incentive or pay
fairness on a universal basis. Furthermore, to address the endogeneity concern, we compare the
wage dispersions between practical nurses and nurse managers, both of whom are most likely
to be female and identify the 2008 financial crisis as an exogenous shock, and we find similar
evidence. Finally, it is worth noting that we are studying a rather small and special group of
employers: non-profit hospitals. It is arguable to what extent the compensation practice in nonprofit hospitals represents the wage policy nationwide. This suggests that future research might
be directed towards analyzing whether our findings are robust for other institutional
environments as well. It may well be the case that either tournament incentives or pay fairness
considerations are less or even more important in hospitals than other for-profit and not-forprofit organizations.
17
REFERENCE
Abowd, John, Francis Kramarz, and David Margolis, 1999, High Wage Workers and High Wage
Firms, Econometrica 67, 251–333.
Abrevaya, Jason, 2002, Ladder Tournaments and Underdogs: Lessons from Professional
Bowling, Journal of Economic Behavior and Organization 47, 87–101.
Adams, Christopher, 2006, Optimal Team Incentives with CES Production, Economics Letters 92,
143–148.
Akerlof, George and Janet Yellen, 1990, The Fair Wage-Effort Hypothesis and Unemployment,
Quarterly Journal of Economics 105, 255–283.
Becker, Brian and Mark Huselid, 1992, The Incentive Effects of Tournament Compensation
Systems, Administrative Science Quarterly 37, 336–350.
Bewley, Truman, 1999, Why Wages Don't Fall During a Recession, Harvard University Press.
Bloom, Matt, 1999, The Performance Effects of Pay Dispersion on Individuals and Organizations,
Academy of Management Journal 42, 25–40.
Brown, Charles and James Medoff, 1989, The Employer Size-Wage Effect, Journal of Political
Economy 97, 1027–1059.
Bull, Clive, Andrew Schotter and Keith Weigelt, 1987, Tournaments and Piece Rates: An
Experimental Study, Journal of Political Economy 95, 1–33.
Campbell, Carl and Kunal Kamlani, 1997, The Reasons for Wage Rigidity: Evidence From a
Survey of Firms, Quarterly Journal of Economics 112, 759–789.
Camerer, Colin and Richard Thaler, 1995, Anomalies: Ultimatums, Dictators and Manners,
Journal of Economic Perspectives 9, 209–219.
Charness, Gary and Peter Kuhn, 2007, Does Pay Inequality Affect Worker Effort? Experimental
Evidence, Journal of Labor Economics 25, 693–723.
Cowherd, Douglas and David Levine, 1992, Product Quality and Pay Equity between LowerLevel Employees and Top Management: An Investigation of Distributive Justice Theory,
Administrative Science Quarterly 37, 302–320.
Ehrenberg, Ronald and Michael Bognanno, 1990, Do Tournaments have Incentive Effects?,
Journal of Political Economy 98, 1307–1324.
Eriksson, Tor, 1999, Executive Compensation and Tournament Theory: Empirical Tests on
Danish Data, Journal of Labor Economics 17, 262–280.
18
Fehr, Ernst and Klaus Schmidt, 1999, A Theory of Fairness, Competition and Cooperation,
Quarterly Journal of Economics 114, 817–868.
Ferrer, Ana and Stéphanie Lluis, 2008, Should Workers Care about Firm Size?, Industrial and
Labor Relations Review 62, 104–125.
Franck, Egon and Stephan Nüesch, 2011, The Effect of Wage Dispersion on Team Outcome and
the Way Team Outcome is Produced, Applied Economics 43, 3037–3049.
Frank, Robert, 1996, What Price the Moral High Ground?, Southern Economic Journal 63, 1-17.
Garen, John, 1985, Worker Heterogeneity, Job Screening, and Firm Size, Journal of Political
Economy 93, 715–739.
Groshen, Erica and Alan Krueger, 1990, The Structure of Supervision and Pay in Hospitals,
Industrial and Labor Relations Review 43, 134S–146S.
Grund, Christian and Niels Westergaard-Nielsen, 2008, The Dispersion of Employees’ Wage
Increases and Firm Performance, Industrial and Labor Relations Review 61, 485–501.
Gupta, Nina, Samantha Conroy, and John Delery, 2012, The Many Faces of Pay Variation,
Human Resource Management Review 22, 100–115.
Hansmann, Henry, 1980, The Role of Nonprofit Enterprise, Yale Law Journal 89, 835–901.
Heyman, Fredrik, 2005, Pay Inequality and Firm Performance: Evidence from Matched
Employer-Employee Data, Applied Economics 37, 1313–1327.
Hibbs, Douglas and Hakan Locking, 2000, Wage Dispersion and Productive Efficiency:
Evidence for Sweden, Journal of Labor Economics 18, 755–782.
Kepes, Sven, John Delery, and Nina Gupta, 2009, Contingencies in the Effects of Pay Range on
Organizational Effectiveness, Personnel Psychology 62, 497–531.
Lakdawalla, Darius, and Thomas Philipson (2006), Non-Profit Production and Industry
Performance, Journal of Public Economics 90, 1681–1698.
Lallemand, Thierry, Robert Plasman, and François Rycx, 2004, Intra-Firm Wage Dispersion and
Firm Performance: Evidence from Linked Employer-Employee Data. Kyklos 57, 533–558.
Lazear, Edward, 2000, Performance Pay and Productivity, American Economic Review 90, 1346–
1361.
Lazear, Edward and Sherwin Rosen, 1981, Rank-Order Tournaments as Optimum Labor
Contracts, Journal of Political Economy 89, 841–864.
Leete, Laura, 2000, Wage Equity and Employee Motivation in Nonprofit and For-Profit
Organization, Journal of Economic Behavior and Organization 43, 423–446.
19
Levine, David, 1991, Cohesiveness, Productivity, and Wage Dispersion, Journal of Economic
Behavior and Organization 15, 237–255.
Leonard, Jonathan, 1990, Executive Pay and Firm Performance, Industrial and Labor Relations
Review 43, S13–29.
Li, Jun, 2013, Job Mobility, Wage Dispersion, and Technological Change: An Asymmetric
Information Perspective, European Economic Review 60, 105–126.
Main, Brian, Charles O’Reilly, and James Wade, 1993, Top Executive Pay: Tournament or
Teamwork?, Journal of Labor Economics 11, 606–628.
Milgrom, Paul and John Roberts, 1990, The Efficiency of Equity in Organizational Decision
Processes, American Economic Review 80, 154–159.
Nosenzo, Daniele, 2012, Pay Secrecy and Effort Provision, Economic Inquiry, forthcoming.
Oi, Walter and Todd Idson, 1999, Firm Size and Wages, Handbook of Labor Economics 3B,
Ashenfelter, Orley and David Card (Eds.), 2165–2214.
O'Reilly, Charles, Brian Main, and Graef Crystal, 1988, CEO Compensation as a Tournament
and Social Comparison: A Tale of Two Theories, Administrative Science Quarterly 33, 257–274.
Pfeffer, Jeffrey and Nancy Langton, 1993, The Effect of Wage Dispersion on Satisfaction,
Productivity, and Work Collaboratively: Evidence from College and University Faculty,
Administrative Science Quarterly 38, 382–407.
Philanthropy Journal, 2008, Nonprofit Turnover Lower than For-profit, March 5th, 2008, the
Institute for Nonprofits at North Carolina State University.
Preston, Anne, 1989, The Nonprofit Worker in a For-Profit World, Journal of Labor Economics 7,
438–463.
Radowski, Daniel and Holger Bonin, 2010, Downward Nominal Wage Rigidity in Services,
Direct Evidence from a Firm Survey, Economics Letters 106, 227–229.
Rosen, Sherwin, 1986, Prizes and Incentives in Elimination Tournaments, American Economic
Review 76, 701–715.
Shaw, Jason, Nina Gupta, and John Delery, 2002, Pay dispersion and workforce performance:
moderating effects of incentives and interdependence, Strategic Management Journal 23, 491–512.
Simon, Herbert, 1957, The Compensation of Executives, Sociometry 20, 32–35.
Winter-Ebmer, Rudolf and Josef Zweimüller, 1999, Intra-firm Wage Dispersion and Firm
Performance, Kyklos 52, 555–572.
20
Yang, Hyuckseung and Brian Klaas, 2011, Pay Dispersion and the Financial Performance of the
Firm: Evidence from Korea, International Journal of Human Resource Management 22, 2147–2166.
Zoega, Gylfi and Thorlakur Karlsson, 2006, Does Wage Compression Explain Rigid Money
Wages?, Economics Letters 93, 111–115.
21
Table I. Variable definitions and summary statistics
Mean
Standard
Deviation
Net income divided by gross revenue on the
income statement
2.46
5.94
Operating margin (%)
Operating income divided by gross revenue
2.16
5.19
log(Total assets)
Natural log of total assets on the balance sheet
12.1
1.06
Leverage
Total debt divided by total assets
0.635
0.342
Liquidity
Current assets divided by current liabilities
1.85
0.906
Num of employees (,000)
Total number of employees in thousands
1.57
1.25
Pay dispersion ($)
Weighted average range of hourly wage among all
job functions within the hospital. The weight is the
number of employees in each job function.
17.3
7.01
High/Low Skill Breakdowns
(See Table III)
Num of low-skill workers (,000)
Total number of employees with low-skill in
thousands.
0.683
0.496
Num of high-skill workers (,000)
Total number of employees with high-skill in
thousands.
0.878
0.76
Pay dispersion of low-skill
workers ($)
Weighted average range of hourly wage among
low-skill workers. The weight is the number of
employees in this category.
11.3
4.69
Pay dispersion of high-skill
workers ($)
Weighted average range of hourly wage among
high-skill workers. The weight is the number of
employees in this category.
22.1
9.11
High/Mid/Low Skill Breakdowns
(See Table III)
Num of low-skill workers (,000)
Total number of employees with low-skill in
thousands.
0.595
0.425
Num of mid-skill workers (,000)
Total number of employees with mid-skill in
thousands
0.243
0.223
Num of high-skill workers (,000)
Total number of employees with high-skill in
thousands.
0.723
0.618
Pay dispersion of low-skill
workers ($)
Weighted average range of hourly wage among
low-skill workers. The weight is the number of
employees in this category.
10.8
4.65
Pay dispersion of mid-skill
workers ($)
Weighted average range of hourly wage among
mid-skill workers. The weight is the number of
employees in this category.
16.3
6.17
Pay dispersion of high-skill
workers ($)
Weighted average range of hourly wage among
high-skill workers. The weight is the number of
employees in this category.
23.2
9.96
Variable
Definition
Net profit margin (%)
22
Operating margin
0.846
log(Total assets)
0.088
0.135
Leverage
-0.278
-0.232
-0.163
Liquidity
0.251
0.151
0.000
-0.568
Num of employees
0.203
0.289
0.759
-0.126
-0.004
Pay dispersion
-0.018
-0.042
0.391
0.080
-0.046
0.454
0.218
0.314
0.783
-0.138
0.015
0.985
0.443
0.190
0.267
0.732
-0.119
-0.015
0.994
0.455
0.961
0.058
0.119
0.370
0.106
-0.099
0.441
0.887
0.442
0.434
-0.021
-0.061
0.379
0.091
-0.015
0.423
0.981
0.424
0.417
0.810
0.222
0.317
0.786
-0.140
0.024
0.977
0.441
0.998
0.950
0.438
0.425
0.190
0.282
0.701
-0.092
-0.030
0.967
0.407
0.930
0.975
0.410
0.368
0.914
0.188
0.262
0.735
-0.128
-0.012
0.993
0.466
0.963
0.997
0.440
0.428
0.953
0.957
0.044
0.089
0.369
0.110
-0.081
0.422
0.895
0.427
0.414
0.992
0.824
0.426
0.389
0.419
0.057
0.130
0.393
0.096
-0.166
0.501
0.757
0.489
0.501
0.832
0.692
0.484
0.475
0.505
0.794
-0.014
-0.058
0.360
0.094
0.009
0.397
0.963
0.401
0.389
0.781
0.992
0.403
0.346
0.399
0.800
Num of low-skill workers
(High/Low)
Num of high-skill workers
(High/Low)
Pay dispersion of low-skill
workers (High/Low)
Pay dispersion of highskill workers (High/Low)
Num low-skill workers
(H/M/L)
Num of mid-skill workers
(H/M/L)
Num of high-skill workers
H/M/L)
Pay dispersion of low-skill
workers (H/M/L)
Pay dispersion of mid-skill
workers (H/M/L)
Pay dispersion of highskill workers (H/M/L)
Pay dispersion of
mid-skill workers
Pay dispersion of
low-skill workers
Num of high-skill
workers H/M/L)
Num of mid-skill
workers (H/M/L)
Num low-skill
workers (H/M/L)
Pay dispersion of
high-skill workers
Pay dispersion of
low-skill workers
Num of high-skill
workers(High/Low)
Num of low-skill
workers(High/Low)
Pay Dispersion
Num of employees
Liquidity
Leverage
Log(Total Assets)
Operating Margin
Net Profit Margin
Table II. Correlation matrix
0.619
Table III. Classification of job skill levels and job functions
2-level Job Skill
3-level Job Skill
(0=low, 1=high)
(1=low, 2=mid, 3=high)
Job function
Code
0
1
1
IP/OP ADM & REGISTRATION CLERK
0
1
2
BILLING CLERK
0
1
3
CASHIER - BUSINESS OFFICE
0
1
4
GENERAL OFFICE CLERK
0
1
10
DIETARY AIDE I
0
1
12
ENVIRONMENTAL SERVICE WORKER
0
1
14
LABORATORY ASSISTANT
0
1
19
MAIL CLERK
Job Function Definition
0
1
20
MAINTENANCE MECHANIC I
0
1
22
MEDICAL RECORDS AND CODING CLERK
0
1
28
MEDICAL TRANSCRIPTIONIST
0
1
32
PATIENT CARE/NURSING AIDE I
0
1
33
PATIENT CARE/NURSING AIDE II
0
1
34
PHARMACY TECHNICIAN
0
1
40
ADMINISTRATIVE ASSISTANT
0
1
41
SECURITY OFFICER
0
1
45
MATERIALS HANDLER, OPERATIONS LEVEL
0
1
48
TELEPHONE OPERATOR
0
1
50
UNIT CLERK/SECRETARY
0
1
55
COLLECTION CLERK
0
1
56
COOK
0
1
64
PHLEBOTOMIST
0
2
6
DIETITIAN
0
2
7
EKG CARDIOLOGY TECHNICIAN
0
2
8
EXECUTIVE ASSISTANT
0
2
8
EXECUTIVE SECRETARY
0
2
15
LABORATORY TECHNICIAN I
0
2
16
LABORATORY TECHNICIAN II
0
2
23
MED RCRDS & CODING TECH, ART OR CMRT
0
2
46
SURGICAL TECHNICIAN
0
2
47
MATERIALS HANDLER, PROF. LEVEL
0
2
54
ACCOUNTING/ACCTS PAYABLE/PAYROLL CLERK
0
2
65
PHYSICAL & OCCUPATIONAL THERAPIST ASSIST.
1
2
9
INFORMATION SYSTEMS/STAFF LEVEL
1
2
18
LICENSED PRACTICAL NURSE
1
2
21
MAINTENANCE MECHANIC II
1
2
25
MEDICAL SOCIAL WORKER M.S.W.
1
2
26
SOCIAL WORKER (B.S.)
1
2
27
MDCL. TECHNOLOGIST (A.S.C.P)
1
2
36
RADIOLOGIC TECHNOLOGIST (RRT REG.)
1
2
38
RESPIRATORY THERAPHY TECH.(CERTIFIED)
1
2
44
STATIONARY ENGINEER LICENSED
1
2
53
ACCOUNTANT/FINANCIAL ANALYST
1
2
68
HUMAN RESOURCES, ASSOCIATE LEVEL
1
3
11
GENERAL DUTY NURSE
1
3
13
INFORMATION SYSTEMS/PROFESSIONAL LEVEL
1
3
17
PHYSICIAN'S ASSISTANT
1
3
29
NUCLEAR MEDICINE TECHNOLOGIST (REG.)
1
3
31
NURSE PRACTITIONER/CLIN. NURSE SPEC
1
3
35
PHYSICAL THERAPIST
1
3
37
MRI TECHNOLOGIST
1
3
39
RESP. THERAPIST (NBRT REGISTERED)
1
3
42
SPECIAL PROCEDURES TECHNOLOGIST
1
3
43
STAFF AND CLINICAL PHARMACIST
1
3
49
ULTRASOUND TECHNOLOGIST
1
3
57
C.T. TECHNOLOGIST
1
3
58
NURSE EDUCATORS
1
3
59
NURSE MANAGER A
1
3
60
NURSE MANAGER B
1
3
61
NURSE MANAGER C
1
3
62
NURSE MANAGER D
1
3
63
OCCUPATIONAL THERAPIST
1
3
66
UTILIZATION REVIEW/QUAL ASSUR SPEC/CASE MGR
1
3
67
SPEECH LANGUAGE PATHOLOGIST OR AUDIOLOGIST
1
3
69
HUMAN RESOURCES, MGMT. LEVEL
25
Table IV. Regressions of pay dispersion and profit margins
The dependent variable is operating margin in percentage terms in specifications (1) and (2) and net profit margin in
percentage terms in specifications (3) and (4). The independent variables include log total assets, financial leverage,
liquidity, number of employees (in thousands), and pay dispersion. All four specifications use OLS regression, and zstatistics are shown in the parentheses with ***, ** and * indicating its statistical significant level of 1%, 5% and 10%
respectively.
Dependent Variable:
Operating Margin (%)
Net Profit Margin (%)
(1)
(2)
(3)
(4)
log(Total Assets)
-1.006*
(-1.691)
-0.986
(-1.650)
-0.975
(-1.409)
-0.878
(-1.336)
Leverage
-2.364
(-1.592)
-2.235
(-1.492)
-2.913*
(-1.687)
-2.360
(-1.433)
Liquidity
0.327
(0.599)
0.348
(0.634)
1.005
(1.585)
1.090*
(1.807)
Num of Employee
2.107***
(4.079)
2.108***
(4.067)
1.697***
(2.826)
1.701***
(2.985)
Pay Dispersion
-0.131**
(-1.987)
-0.140**
(-2.087)
-0.0779
(-1.015)
-0.116
(-1.584)
Constant
14.23**
(2.023)
13.48*
(1.898)
12.94
(1.583)
9.511
(1.218)
Year Fixed-Effects
No
Yes
No
Yes
N
143
143
143
143
0.141
0.135
0.113
0.201
5.65***
4.16***
4.61***
6.09***
Adj. R-square
F-Test
26
Table V. Regressions of pay dispersions with high/low skills and profit margins
The dependent variable is operating margin in percentage terms in specifications (1) and (2) and net profit margin in
percentage terms in specifications (3) and (4). The independent variables include log total assets, financial leverage,
liquidity, number of employees (in thousands) of high- and low-skills, and pay dispersion of employees of high- and
low-skills. All four specifications use OLS regression, and z-statistics are shown in the parentheses with ***, ** and *
indicating its statistical significant level of 1%, 5% and 10% respectively.
Dependent Variable:
Operating Margin (%)
Net Profit Margin (%)
(1)
(2)
(3)
(4)
log(Total Assets)
-2.176***
(-3.567)
-2.165***
(-3.547)
-2.019***
(-2.667)
-1.982***
(-2.804)
Leverage
-2.479*
(-1.858)
-2.338*
(-1.744)
-3.054*
(-1.844)
-2.466
(-1.588)
Liquidity
0.494
(0.997)
0.503
(1.014)
1.108*
(1.802)
1.154**
(2.010)
Num of low-skill workers
23.30***
(5.371)
23.93***
(5.481)
21.44***
(3.981)
23.84***
(4.716)
Num of high-skill workers
-8.501**
(-2.606)
-8.999***
(-2.739)
-8.232**
(-2.033)
-10.13***
(-2.662)
Pay dispersion among low-skill
workers
1.195***
(5.348)
1.198***
(5.355)
1.069***
(3.855)
1.076***
(4.154)
Pay dispersion among high-skill
workers
-0.503***
(-4.676)
-0.515***
(-4.765)
-0.413***
(-3.098)
-0.459***
(-3.665)
Num of low-skill workers × Pay
dispersion among low-skill workers
-1.088***
(-4.035)
-1.111***
(-4.108)
-1.212***
(-3.622)
-1.297***
(-4.142)
Num of high-skill workers × Pay
dispersion among high-skill workers
0.272***
(2.614)
0.284***
(2.713)
0.322**
(2.490)
0.365***
(3.013)
Constant
21.80***
(3.264)
21.07***
(3.143)
19.88**
(2.398)
17.08**
(2.200)
Year Fixed-effects
No
Yes
No
Yes
N
143
143
143
143
0.321
0.319
0.198
0.302
8.40***
7.04***
4.91***
6.59***
Adj. R-square
F-Test
27
Table VI. Regressions of pay dispersions with high/mid/low skills and profit margins
The dependent variable is operating margin in percentage terms in specifications (1) and (2) and net profit margin in
percentage terms in specifications (3) and (4). The independent variables include log total assets, financial leverage,
liquidity, number of employees (in thousands) of high-, mid- and low-skills, and pay dispersion of employees of
high-, mid- and low-skills. All four specifications use OLS regression, and z-statistics are shown in the parentheses
with ***, ** and * indicating its statistical significant level of 1%, 5% and 10% respectively.
Dependent Variable:
Operating Margin (%)
Net Profit Margin (%)
(1)
(2)
(3)
(4)
log(Total Assets)
-2.454***
(-3.803)
-2.430***
(-3.766)
-2.347***
(-2.977)
-2.261***
(-3.074)
Leverage
-2.686*
(-1.901)
-2.545*
(-1.795)
-3.008*
(-1.742)
-2.500
(-1.547)
Liquidity
0.403
(0.774)
0.428
(0.820)
1.021
(1.604)
1.108*
(1.864)
Num of low-skill workers
22.74***
(3.709)
23.55***
(3.819)
19.51**
(2.604)
22.45***
(3.192)
Num of mid-skill workers
16.21
(1.184)
16.80
(1.225)
18.81
(1.125)
20.93
(1.338)
Num of high-skill workers
-10.99***
(-2.655)
-12.01***
(-2.864)
-10.02**
(-1.981)
-13.70***
(-2.865)
Pay dispersion among low-skill
workers
0.905***
(2.845)
0.885***
(2.775)
0.800**
(2.059)
0.730**
(2.006)
Pay dispersion among mid-skill
workers
0.0915
(0.520)
0.103
(0.584)
0.128
(0.594)
0.169
(0.841)
Pay dispersion among high-skill
workers
-0.367***
(-3.585)
-0.373***
(-3.637)
-0.331***
(-2.644)
-0.351***
(-3.007)
Num of low-skill workers × Pay
dispersion among low-skill workers
-0.954**
(-2.428)
-0.977**
(-2.476)
-1.019**
(-2.122)
-1.101**
(-2.447)
Num of mid-skill workers × Pay
dispersion among mid-skill workers
-0.409
(-0.698)
-0.384
(-0.654)
-0.711
(-0.993)
-0.621
(-0.928)
Num of high-skill workers × Pay
dispersion among high-skill workers
0.246**
(2.086)
0.254**
(2.157)
0.332**
(2.309)
0.363***
(2.700)
Constant
24.72***
(3.490)
23.71***
(3.332)
23.24***
(2.685)
19.61**
(2.418)
Year Fixed-effects
No
Yes
No
Yes
N
143
143
143
143
0.273
0.274
0.171
0.279
5.45***
4.83***
3.44***
4.93***
Adj. R-square
F-Test
28
Table VII. Regressions of pay dispersions among nurses and net profit margins
The dependent variable is net profit margin in percentage terms in all specifications. The independent variables
include log total assets, financial leverage, liquidity, number of employees in thousands, pay dispersion of all
employees, and pay dispersion of practical nurses and nurse managers. All four specifications use OLS regression,
and z-statistics are shown in the parentheses with ***, ** and * indicating its statistical significant level of 1%, 5% and
10% respectively.
(1)
(2)
(3)
(4)
log(Total Assets)
-0.114
(-0.202)
-0.156
(-0.308)
-0.0896
(-0.157)
-0.107
(-0.209)
Leverage
-3.312**
(-2.308)
-2.698**
(-2.076)
-3.241**
(-2.239)
-2.537*
(-1.944)
Liquidity
0.912*
(1.956)
1.053**
(2.502)
0.925*
(1.974)
1.083**
(2.572)
Num of Employee
0.513
(1.117)
0.500
(1.209)
0.558
(1.189)
0.593
(1.410)
-0.0313
(-0.496)
-0.0648
(-1.141)
Dependent Variable: Net Profit Margin
Pay Dispersion of All Employees
Pay Dispersion of Practical Nurses
-0.0453
(-0.969)
-0.0452
(-1.069)
-0.0345
(-0.668)
-0.0229
(-0.493)
Pay Dispersion of Nurse Managers
-0.0850**
(-1.997)
-0.0876**
(-2.286)
-0.0838**
(-1.961)
-0.0853**
(-2.225)
6.227
(0.927)
4.145
(0.684)
6.181
(0.917)
3.994
(0.659)
Year Fixed-Effects
No
Yes
No
Yes
N
143
143
143
143
0.129
0.294
0.123
0.296
3.88***
7.08***
3.34***
6.46***
Constant
Adj. R-square
F-Test
29