Servicios
Los economistas expertos de Econ One tienen experiencia en una amplia variedad de servicios, como defensa de la competencia, certificación colectiva, daños y perjuicios, mercados financieros y valores, propiedad intelectual, arbitraje internacional, trabajo y empleo, y valoración y análisis financiero.
Industrias
Los economistas expertos de Econ One cuentan con una amplia experiencia en sectores específicos. Nuestra experiencia abarca numerosos sectores, como los mercados de la energía eléctrica, los mercados financieros, la sanidad, los seguros, el petróleo y el gas, la industria farmacéutica, etc.
Recursos
Los recursos de Econ One, que incluyen blogs, casos, noticias y mucho más, ofrecen una colección de materiales de los expertos de Econ One.
Blog
Conozca la economía desde dentro con los expertos.
Director General
- Los Angeles, CA
- bkriegler@econone.com
- (Principal) 213.624.9600
Educación
Doctorado en Estadística, Universidad de California, Los Ángeles
Máster en Estadística, Universidad de California, Los Ángeles
Licenciatura en Matemáticas/Economía, Claremont McKenna College
Experiencia laboral
Econ One, Agosto 2008 - Presente
Universidad de Pensilvania, 2007 - 2008
Universidad de California en Los Ángeles, 2007 - 2008
Consultor estadístico autónomo, 2004 - 2008
RAND Statistics Group, 2006
Lockheed Martin Misiles y Espacio, 2001 - 2003
Experiencia testimonial
Tribunal de distrito de EE.UU.
Tribunal del Estado
Arbitraje
Mediación privada
Industrias
Comparte este artículo
While this blog focuses on gender wage disparities between men and women, the methods described herein could be extended to non-binary, transgender, and other gender-diverse individuals.
Índice
Statistical modeling is a commonly used technique as a means for understanding wage disparities. In this article, we will explore how various forms of regression analysis can be used to identify and quantify gender-based pay gaps. Particularly in California under the Equal Pay Act, discrepancies across genders in terms of earnings have become a hot button topic.
Key Takeaways
- Statistical modeling, including linear and data adaptive non-linear regression, plays a crucial role in identifying and quantifying gender wage gaps.
- California’s Equal Pay Act mandates comparable pay for employees performing substantially similar work irrespective of gender.
- Engaging an experienced statistician or economist is vital for conducting a robust analysis of potential wage gaps.
Introducción
Statistical modeling is a powerful tool in evaluating Equal Pay Act claims. When analyzing potential wage disparities, the practitioner’s goals typically include (i) modeling the data reasonably well, (ii) isolating the association between earnings and gender, and (iii) assessing the magnitude of said association.
Over the decades, statistical modeling has evolved significantly. While economists and statisticians historically have utilized traditional techniques such as linear regression, the field now includes more advanced and flexible models as well. These newer techniques can provide additional insights and greater precision.
In this exploration, we will discuss these various statistical techniques and how they can be applied to evaluate pay gaps.
The Equal Pay Act and its Significance
California’s Equal Pay Act, established under Labor Code 1197.5, is a foundational legal framework that mandates equal pay for employees performing substantially similar work under similar conditions. This act prohibits pay discrepancies between men and women who perform jobs that require substantially equal skill, effort, responsibility, and under comparable working conditions. If an employee believes there may be a difference between earnings among men versus women, then this can lead to litigation and/or a governmental audit.
As compensation practices receive greater visibility and heightened awareness, so too do the statistical techniques used to evaluate employees’ earnings. Analyzing historical employee data helps organizations gain insights into the factors contributing to wage differences.
The Role of Statistical Modeling in Gender Wage Gap Analysis
A gender pay gap analysis typically entails modeling earnings as a function of years of experience, years at the company, job title, other potentially relevant attributes, along with gender. This allows the practitioner to assess the magnitude of wages for men versus women, holding all other variables constant.
Key Statistical Techniques Used in Gender Wage Gap Analysis
Regression modeling can take on a number of forms when evaluating earnings. Practitioners must make a series of decisions regarding the type of model to use, the data to consider, and the functional form. Two broad categories of techniques are described below.
Linear Regression
One of the strengths of linear regression lies in its simplicity and interpretability. Consider the graph below, which shows a hypothetical company’s earnings conditional on years of experience and gender:
The corresponding linear regression model for this company is as follows:
Average Earnings = $70,000 + $2,000 x Years of Experience + $10,000 x Male
In the above model, the interpretation of the model is as follows:
- The average starting salary for a woman with no experience is $70,000
- Holding other variables constant, earnings increase by an average of $2,000 for each additional year of experience
- Holding other variables constant, men receive an average of $10,000 more than women
Data-Adaptive Nonlinear Regression
An alternative technique is data-adaptive nonlinear regression. This is a flexible technique that adjusts its form based on the specific characteristics of the dataset. Unlike linear regression, this method does not assume any predefined relationships between the outcome variable and predictor variables. This approach leaves open the possibility of capturing many complex non-parametric interactions.1
Before modeling the data, it may be unknown as to whether the relationship between the outcome variable and each predictor variable is linear, stepwise, piecewise, exponential, or some combination of these. Additionally, there may well be meaningful and nonlinear interactions between predictor variables. Data-adaptive nonlinear regression generally is capable of accurately identifying these associations.
Data-adaptive nonlinear regression can be useful in gender wage gap analysis in the event that there are intricate, nonlinear relationships between variables. As a result, it offers a more robust and reliable approach to analyzing and addressing gender wage disparities.
By way of a simple example, consider an employer who aspires to quantify the potential wage gap between men and women. The available data attributes include years of experience and gender. With this approach, the “best” data fit may include (i) a linear relationship between earnings and experience for the first 10 years, (ii) an exponential jump in earnings after 10 years, and (iii) an interaction between experience and gender. These data are depicted in the graph below.
Key takeaways from the above graph include the following:
- The patterns in the data among women are distinct from those among men.
- These data show a nonlinear relationship between earnings and years of experience. In this instance, a nonlinear model provides an accurate representation of trends in the data.
- There appears to be slightly more variation in earnings among men than women. For a given number of years of experience, the red data points signifying women tend to be relatively close to the corresponding trend line in black. Conversely, the blue data points signifying men appear to vary more widely around the corresponding trend line in dark green.
- Pay gaps between men and women appear to be negligible for people with less than 12 years of work experience. Thereafter, there are noticeable gaps between men and women.
One minor limitation of data-adaptive nonlinear model is that typically there is no simple equation to observe. As a result, graphical representations of the data become important assets to describe how each of the predictor variables is associated with the outcome variable. These types of graphs and diagnostics will be covered in a future blog post.
The Importance of Working with an Experienced Statistician or Economist
Working with an experienced statistician or economist is crucial in conducting accurate and reliable wage gap analyses. They generally will have access to advanced computational programs and techniques, allowing them to build and evaluate various models. This expertise increases the likelihood that the analysis will be thorough and aid in providing actionable insights into wage disparities.
In addition to technical skills, experienced statisticians bring a deep understanding of the data points and the context of the analysis. They can assess which approach makes the most sense based on the specific characteristics of the dataset, ensuring that the findings are accurate and relevant. Moreover, they can apply the proper interpretation to the data and explain the results in ways that the audience can comprehend, making the analysis accessible and impactful.
Resumen
The analysis of gender wage gaps through statistical modeling is a powerful and commonly accepted approach to evaluating Equal Pay Act claims. Practitioners are tasked with deciding which regression technique is appropriate to use, as well as which data attributes to consider.
Both linear and data adaptive nonlinear regression have their strengths and can be used to uncover meaningful insights into wage disparities. Linear regression provides a relatively straightforward and interpretable approach. Data adaptive non-linear regression offers substantial flexibility and accuracy in capturing complex relationships.
Preguntas frecuentes
What role does statistical modeling play in analyzing gender wage gaps?
Statistical modeling, especially regression analysis, plays a crucial role in identifying and quantifying gender wage gaps by analyzing employee data while controlling for factors such as experience, job department and/or education. This approach provides a clearer understanding of the disparities and informs policy decisions.
How does regression help in a gender wage gap analysis?
Regression models are essential for analyzing the gender wage gap as they quantify the impact of various factors on earnings, facilitating the identification of gender-based pay disparities. By isolating these effects, these models offer insights into potential inequalities in compensation.
What are the advantages of using linear regression to quantify wage gaps across genders?
Linear regression is particularly advantageous when there are linear associations between earnings and a given predictor variable. These circumstances allow the practitioner to provide a straightforward mathematical equation for estimated earnings.
What are the advantages of using data adaptive non-linear regression to quantify wage gaps across genders?
Data adaptive non-linear regression is particularly advantageous when the relationship between earnings and a set of predictor variables includes numerous complex patterns. The data may well suggest that compensation is a by-product of linear, stepwise, piecewise, exponential, and/or interactive predictor variables.
Why is it important to work with an experienced statistician for a wage gap analysis?
Experienced statisticians and economists can construct multiple regression models, test the robustness of these models, and interpret the results. For instance, they will assess whether the data show linear associations between earnings and the predictor variables, or alternatively, if the patterns are more complex and nuanced.
Referencias
1 One commonly accepted data-adaptive technique is known as “boosting.” This approach will be described in greater detail in a future post.