Correction For Selection Bias Research Articles

Digital finance and MSME performance in India: evidence from World Bank Enterprise Survey data

PurposeThe impact of digitalisation on smaller firms remains sparsely studied across emerging economies. The paper aims to examine the relationship between digital adoption and multiple performance parameters of micro, small and medium enterprises (MSME) in a prominent emerging economy.Design/methodology/approachThe study employs data from the World Bank Enterprise Survey (WBES) 2022, capturing 9,024 Indian MSME firms spread across the country. Performance indicators are derived from growth in sales, employment and labour productivity (LPROD). Multiple regression estimates are derived that also correct for sample selection bias using Heckman’s two-step process.FindingsDigital proliferation is found to increase as firms mature up in terms of age, size and constitution. A significant difference could also be observed in business performance across digital and non-digital businesses, with sales growth (SG) and productivity higher for digital firms. Digital financial variables are found to have a significant impact on SG but not as much in the case of employment growth and LPROD. The results are robust to correction for sample selection bias in digital adoption using inverse mills ratio (IMR).Practical implicationsThe study highlights digital adoption gaps across various strata of MSMEs, highlighting lower adoption when firms are younger, smaller and lacking formal constitutional setup. Digital variables indicating positive association with SG highlight the need for concerted efforts at the public policy level for building appropriate skills and infrastructure for micro and small enterprises to boost their digital adoption to promote growth.Originality/valueThere is a lack of micro-level empirical evidence measuring the impact of advanced digital technology usage on multiple aspects of enterprise performance amongst micro and small firms. The study deploys unique digital variables including TReDS and use of online credit applications to assess the impact on business performance. The findings provide insights for practice and public policy, besides making the case for a higher focus on launching digital initiatives for smaller enterprises.

Migration and Remittances in Senegal: Effect on Households Members Left Behind

ABSTRACT This article investigates how migration and remittances affect labor market, and how remittances influence human capital in Senegal. Using parametric and nonparametric estimation procedures for the correction of selection bias, the results reveal that migration and remittances reduce labor market participation of household members left behind, and that remittances increase human capital development of the left-behind. The study suggests that both the migration status and the level of remittances are relevant in understanding labor market participation and human capital formation.

Inverse probability weighting for self-selection bias correction in the investigation of social inequality in mortality.

Empirical evaluation of inverse probability weighting (IPW) for self-selection bias correction is inaccessible without the full source population. We aimed to: (i) investigate how self-selection biases frequency and association measures and (ii) assess self-selection bias correction using IPW in a cohort with register linkage. The source population included 17 936 individuals invited to the Copenhagen Aging and Midlife Biobank during 2009-11 (ages 49-63 years). Participants counted 7185 (40.1%). Register data were obtained for every invited person from 7years before invitation to the end of 2020. The association between education and mortality was estimated using Cox regression models among participants, IPW participants and the source population. Participants had higher socioeconomic position and fewer hospital contacts before baseline than the source population. Frequency measures of participants approached those of the source population after IPW. Compared with primary/lower secondary education, upper secondary, short tertiary, bachelor and master/doctoral were associated with reduced risk of death among participants (adjusted hazard ratio [95% CI]: 0.60 [0.46; 0.77], 0.68 [0.42; 1.11], 0.37 [0.25; 0.54], 0.28 [0.18; 0.46], respectively). IPW changed the estimates marginally (0.59 [0.45; 0.77], 0.57 [0.34; 0.93], 0.34 [0.23; 0.50], 0.24 [0.15; 0.39]) but not only towards those of the source population (0.57 [0.51; 0.64], 0.43 [0.32; 0.60], 0.38 [0.32; 0.47], 0.22 [0.16; 0.29]). Frequency measures of study participants may not reflect the source population in the presence of self-selection, but the impact on association measures can be limited. IPW may be useful for (self-)selection bias correction, but the returned results can still reflect residual or other biases and random errors.

Labor market segmentation and the gender wage gap: Evidence from China.

Although the Chinese government has implemented a variety of measures, the gender wage gap in 21st century China has not decreased. A significant body of literature has studied this phenomenon using sector segmentation theory, but these studies have overlooked the importance of the collective economy beyond the public and private sectors. Moreover, they have lacked assessment of the gender wage gap across different wage groups, hindering an accurate estimation of the gender wage gap in China, and the formulation of appropriate recommendations. Utilizing micro-level data from 2004, 2008, and 2013, this paper examines trends in the gender wage gap within the public sector, private sector, and collective economy. Employing a selection bias correction based on the multinomial logit model, this study finds that the gender wage gap is smallest and most stable within the public sector. Furthermore, the private sector surpasses the collective economy in this period, becoming the sector with the largest gender wage gap. Meanwhile, a recentered influence function regression reveals a substantial gender wage gap among the low-wage population in all three sectors, as well as among the high-wage population in the private sector. Additionally, employing Brown wage decomposition, this study concludes that inter-sector, rather than intra-sector, differences account for the largest share of the gender wage gap, with gender discrimination in certain sectors identified as the primary cause. Finally, this paper provides policy recommendations aimed at addressing the gender wage gap among low-wage groups and within the private sector.

Revising model for end-stage liver disease from calendar-time cross-sections with correction for selection bias

BackgroundEurotransplant liver transplant candidates are prioritized by Model for End-stage Liver Disease (MELD), a 90-day waitlist survival risk score based on the INR, creatinine and bilirubin. Several studies revised the original MELD score, UNOS-MELD, with transplant candidate data by modelling 90-day waitlist mortality from waitlist registration, censoring patients at delisting or transplantation. This approach ignores biomarkers reported after registration, and ignores informative censoring by transplantation and delisting.MethodsWe study how MELD revision is affected by revision from calendar-time cross-sections and correction for informative censoring with inverse probability censoring weighting (IPCW). For this, we revised UNOS-MELD on patients with chronic liver cirrhosis on the Eurotransplant waitlist between 2007 and 2019 (n = 13,274) with Cox models with as endpoints 90-day survival (a) from registration and (b) from weekly drawn calendar-time cross-sections. We refer to the revised score from cross-section with IPCW as DynReMELD, and compare DynReMELD to UNOS-MELD and ReMELD, a prior revision of UNOS-MELD for Eurotransplant, in geographical validation.ResultsRevising MELD from calendar-time cross-sections leads to significantly different MELD coefficients. IPCW increases estimates of absolute 90-day waitlist mortality risks by approximately 10 percentage points. DynReMELD has improved discrimination over UNOS-MELD (delta c-index: 0.0040, p < 0.001) and ReMELD (delta c-index: 0.0015, p < 0.01), with differences comparable in magnitude to the addition of an extra biomarker to MELD (delta c-index: ± 0.0030).ConclusionCorrecting for selection bias by transplantation/delisting does not improve discrimination of revised MELD scores, but substantially increases estimated absolute 90-day mortality risks. Revision from cross-section uses waitlist data more efficiently, and improves discrimination compared to revision of MELD exclusively based on information available at listing.

Assessment of impacts of adoption of improved sweetpotato varieties in Ghana: accounting for differences in male and female farmers

New agricultural technologies, such as high yielding varieties and improved agronomic practices could significantly increase agricultural productivity leading to improved welfare of those who participate in them. Feedback effects on adoption of agricultural technologies in low-income countries are important in prioritization of funding to the agricultural sector. Through a multistage randomized sample of 526 farmers and employing propensity score matching technique to correct selection bias, the study assessed the effect of adoption of improved sweetpotato varieties on farmers’ yields and incomes comparing separate effects on men and women farmers. Results revealed that income and yield increased with adoption. Men benefited more than women due to access to extension, improved varieties and productive inputs. Nonetheless, men and women had significant increases in yield and income. Results also showed that land and credit play a role in adoption of technologies. Supporting research extension systems to promote and disseminate improved varieties is recommended.

Male and female selection effects on gender wage gaps in three countries

A vast literature on gender wage gaps has examined the importance of selection into employment. However, most analyses have focused only on female labour force participation and gaps at the median. The Great Recession questions this approach because of the major shift in male employment that it implied. This paper uses the methodology proposed by Arellano and Bonhomme (2017) to estimate a quantile selection model over the period 2007–2018. Using a tax and benefit microsimulation model, I compute an instrument capturing both male and female decisions to participate in the labour market: the potential out-of-work income. Since my instrument is crucially determined by the welfare state, I consider three countries with notably different benefit systems – the UK, France and Finland. My results imply different selection patterns across countries and a sizeable male selection in France and the UK. Correction for selection bias lowers the gender wage gap and reveals a substantial glass ceiling with different magnitudes. Findings suggest that disparities between these countries are driven by occupational segregation and public spending on families.

Gender wage gap in the formal and informal sector in the COVID-19 crisis period, of the national labor market and the state of Coahuila

One of the Sustainable Development Goals of the 2015-2030 agenda establishes guaranteeing equal remuneration between women and men for work of equal value. This objective is taken up in the 2018-2024 nation project for Mexico. In the project, the Federal Government aims to eliminate inequality gaps between women and men, as well as the obstacles that women face in all areas. The objective of this research is to estimate the Gender Wage Gap in the formal and informal sector of the labor market of the country and the state of Coahuila in the period of the COVID-19 crisis, to observe the trend of wage differences in the period 2019 – 2022. Mincerian wage equations (1974) were estimated for each gender with data from the National Occupation and Employment Survey (ENOE), for subordinate workers. The variables considered the sector in which the employee works and the condition of the working day. The Gender Wage Gap was estimated using the Blinder and Oaxaca (1973) methodology, with selection bias correction (Heckman, 1979). It was observed that the Gender Wage Gap in the context of crisis generally tended to close, this was because in the formal and informal sector the salary difference for women tended to decrease in 2020, while in 2022, during the economic recovery the Gender Wage Gap returned to the parameters of 2019 both nationally and in the state of Coahuila.

Tobacco use and employment status: A cross-sectional analysis of the CONSTANCES cohort

This study aims to estimate the prevalence of tobacco use in 2017 and 2019 in the French population covered by the Régime Général d'Assurance Maladie according to employment status. From the French national CONSTANCES cohort, 18,008 randomly recruited volunteers aged between 18 and 69 years, affiliated to the Régime Général d'Assurance Maladie and enrolled in 2017, were included in the analysis. The prevalence of tobacco use according to employment status was estimated. Estimates of these prevalence data were calculated in 2017 and 2019 after correction for selection bias at inclusion and non-response at follow-up. In 2019, smoking prevalence was higher among unemployed people (29.2% among men and 20.7% among women) than among employed people (16.5% among men and 13.8% among women). Smoking prevalence was highest among those not in work for health reasons (38.5% among men and 35.8% among women). Smokers were more likely to be unemployed than non-smokers (OR 2.63 [95% confidence interval (CI): 1.79; 3.85] in men and OR 1.55 [95% CI: 1.08; 2.22] in women). Among men, the prevalence of employed people among light smokers (&lt;10 cigarettes/day) significantly decreased between 2017 (87.1%) and 2019 (74.8%). These results underline the importance of reinforcing smoking prevention campaigns among the unemployed, particularly for health reasons.

PhD holders propensity to work in research-intensive sectors: evidence from Italy

PhD holders play an important role in exploring new research areas relevant to the development of knowledge-based economies, and the investment in PhD education is seen as part of a strategy for tomorrows’ society, with doctorates historically being absorbed by an academic career dedicated to teaching and research. However, these days are gone, and the number of PhD holders is far beyond the academic vacancies available. The present study aims at estimating the probability of PhD holders to work in research intensive sectors in Italy. We use data available from the National Institute of Statistics to estimate a probit model with the Heckman correction for the sample selection bias. The study has political consequences related to PhD holders career orientation.

Correction to: Correcting Selection Bias in Big Data by Pseudo-Weighting

Correction to: Correcting Selection Bias in Big Data by Pseudo-Weighting

Migration, Selection and Income Differentials in Brazilian Labor Market.

Migration is understood in the economic literature as a decision based on prospects of better working conditions and the search for higher remuneration at the destination, compared to the origin of the workforce. In this sense, this article aims to test the hypothesis of favorable migratory selectivity in Brazil based on the model with sample selection bias correction proposed by Heckman (1979). Based on census information for the years 2000 and 2010, the aim is to analyze the population aged between 15 and 60 years who declared themselves to be employed in the Brazilian labor market. Suppose it is confirmed that migrants are a positively selected population group. In that case, they have unobservable characteristics that affect the decision to migrate and, consequently, labor earnings. The second step is to decompose the characteristics that affect the earnings differentials between migrants and non-migrants, considering those of an observable and non-observable nature. The results show that Brazilian intercity migrants make up a positively selected group. Concerning the breakdown of income differentials, labor income is more remarkable in favor of migrants, and the largest share of income differences between migrants and non-migrants is due to unobservable factors. Therefore, the implications of this study show that the migration of qualified human capital in Brazil, that is, those with the best professional performance in the labor market (positively selected), may end up deepening regional socioeconomic inequalities since migrants always seek opportunities in more economically dynamic regions. This suggests that policies be developed to reduce regional inequalities that aim, above all, to boost the growth of less developed regions so that their human capital has the opportunity to develop internally in their regions, contributing to the growth and development of the original region.

Solar PV adoption in rural Burkina Faso

Energy is a rare commodity in Africa. Many households use unclean fuels for domestic activities. While more than 90% of rural households use fuelwood and kerosene as a source of energy in Sub-Saharan Africa, this study examines the determinants of energy diversity through solar PV adoption by rural household. Employing primary data on 105 villages from Burkina Faso, a sample of 6300 households is investigated. Performing the probit and using a sample selection bias correction technique, the findings show that rural households engaged in economic activities are more likely to adopt a solar PV system. The entrepreneurial spirit has a positive impact on the adoption of solar PV. Moreover, the use of solar PV technology is strongly associated with household head age, gender, family size, and education. The findings suggest that financial support and micro-credit must target more rural households.

Role of immune checkpoint inhibitors in metastatic uveal melanoma: a single-center retrospective cohort study

Background Uveal melanoma is an orphan malignancy with very limited data on treatment options in metastatic setting. Methods In this single-center retrospective study, we describe real-world epidemiological and survival data on 121 metastatic uveal melanoma (MUM) patients registered in our institution. As a large tertiary referral center, almost 30% of all diagnoses in the Flemish region of Belgium were covered. Primarily, we determined whether introduction of immune checkpoint inhibitors (ICI) led to improved overall survival (OS) in MUM patients. Secondarily, response rates to ICI were assessed and we evaluated whether first-line ICI could be a valid alternative to liver-directed therapy (LDT) in liver-only disease. Results The initially perceived 10.8 months survival benefit from treatment with ICI disappeared after correction for immortality bias. By analyzing treatment type as time-varying covariate on OS, no significant benefit of ICI over other systemic therapies (HR = 0.771) or best supportive care (BSC) (HR = 0.780) was found. Also comparison of the pre-ICI versus ICI era showed no OS improvement after introduction of ICI in our center (p = 0.7994). Only liver-directed and local oligometastatic approaches were associated with a lower chance of mortality when compared to ICI (p = 0.0025), other systemic therapies (p = 0.0001) and BSC (p = 0.0003), yet without correction for selection bias. We reported overall response rates on ICI ranging from 8–15% and we found some support for neoadjuvant strategies with ICI resulting in remission or downsizing, allowing oligometastatic approaches later on. In first-line liver-only disease, median real-world progression-free survival and OS did not significantly differ between patients treated with LDT or ICI upfront (p = 0.2930 and p = 0.5461 respectively). Conclusion Although we documented responses to ICI, our analyses do not demonstrate an OS benefit of ICI over alternative treatment strategies for MUM. However, local treatment options, whether liver-directed or for oligometastatic disease, may be beneficial and should be considered.

The Effect of Parental Occupation and Family Arragements on Occupational Choices among Brazilian Youth

This paper estimates the added worker effect and discouraged worker effect for young people in metropolitan Brazil. Its main contribution is to reveal how the transition to unemployment by the household head impacts labor supply among youth, considering the demographic changes – especially the size and composition of family arrangements – that Brazil has undergone in recent decades. We designed multinomial probit models (with and without correction of sample selection bias) covering the main metropolitan regions of the country using data from Brazil’s Monthly Employment Survey (PME) for the years 2002 to 2015, which compiles information on monthly transitions in the Brazilian labor market. Our findings point to evidence of a greater transition of young men and women to activation in the labor market, specifically from inactivity to unemployment, in periods when the household head became unemployed. On the other hand, there is no evidence of a statistically significant discouraged worker effect impacting these individuals. Furthermore, the parents’ occupational situation seems to have little effect on the transition to activation of young males in low-income families, unlike those belonging to wealthier families.

Ethnicity and homophily effects in US M&amp;As

AbstractWe show that ethnic diversity of CEOs of merging firms has been increasing and report evidence of ethnic homophily effect in M&amp;As transactions. Specifically, M&amp;As perform better when the CEOs of the merging firms share a common ethnic background. In a sample of 444 US mergers completed between 2000 and 2018, we find that ethnic homophily improves the probability of deal completion. Furthermore, we report mild variation of performance across ethnic groups. White CEOs exhibit better postmerger performance while Asian CEOs instigate a better market reaction (announcement return). Subsequent analyses show that White CEOs tend to have longer tenure and longer experience while Asian CEOs tend to be more transparent. We discuss our findings in light of a few extant theories and show that our findings are robust to several additional tests including instrumental variables, Heckman's selection bias correction, and several variations in the model specifications and definitions of key variables.

Full-Information Selection Bias Correction for Discrete Choice Models with Observation-Conditional Regressors

We examine self-selection in polychotomous choice models that construct attribute values for each alternative conditioned on observed choices. Using observations made only when the alternative was chosen ignores private information which was a basis for the decision, biasing resulting estimates. We suggest a full-information maximum likelihood procedure that performs well at the extremes of the choice set in our sample, and use an “identification at infinity” weighting to identify levels. We apply the model to understanding fishing location choice in the economically significant Bering Sea pollock fishery, where expected catches at each location are constructed from harvests observed when that location is chosen.

Correcting Selection Bias in Big Data by Pseudo-Weighting

Abstract Nonprobability samples, for example observational studies, online opt-in surveys, or register data, do not come from a sampling design and therefore may suffer from selection bias. To correct for selection bias, Elliott and Valliant (EV) proposed a pseudo-weight estimation method that applies a two-sample setup for a probability sample and a nonprobability sample drawn from the same population, sharing some common auxiliary variables. By estimating the propensities of inclusion in the nonprobability sample given the two samples, we may correct the selection bias by (pseudo) design-based approaches. This paper expands the original method, allowing for large sampling fractions in either sample or for high expected overlap between selected units in each sample, conditions often present in administrative data sets and more frequently occurring with Big Data.

The Closure Relations in Optical Afterglow of Gamma-Ray Bursts

Gamma-ray bursts (GRBs) are extremely high-energy events that can be observed at very high redshift. In addition to γ rays, they can emit in X-ray, optical, and sometimes radio wavelengths. Here, following the approach in Srinivasaragavan et al.; Dainotti et al.; and Dainotti et al., we consider 82 GRBs from Dainotti et al. that have been observed in optical wavelengths and fitted with a broken power law. We consider the relations between the spectral and temporal indices (closure relations; CRs) according to the synchrotron forward-shock model evolving in the constant-density interstellar medium (ISM; k = 0) and the stellar wind environment (k = 2) in both slow- and fast-cooling regimes, where the density profile is defined as n ∝ r −k . We find the regime is most favored, where ν c and ν m are the cooling and characteristic frequencies, respectively. Finally, we test the 2D Dainotti correlation between the rest-frame end time of the plateau and the luminosity at that time on GRBs that fulfill the most-favored CRs. When we compare the intrinsic scatter σ int of those 2D correlations to the scatter presented in Dainotti et al., we see the scatters of our correlations generally agree with the previous values within 1σ, both before and after correction for selection bias. This new information has helped us to pinpoint subsamples of GRBs with features that could drive the GRB emission mechanism, and eventually allow for GRBs to be used as standard candles.

Phase Estimation for Distributed Scatterers by Alternating Projection

Decorrelation is a major obstacle to the application of multitemporal interferometric synthetic aperture radar (InSAR) in areas with low coherence. Distributed scatterers (DSs) with similar backscattering in the spatial neighborhood are the key to improving the observational density and accuracy of deformation estimation in fast decorrelation regions. Phase series estimation from all possible interferograms is expected to improve the signal-to-noise ratio (SNR) and further enhance the sensitivity of deformation measurement. The coherence bias and the efficiency of the phase estimation raise concerns. In this article, we propose a computationally attractive algorithm for the interferometric phase estimation, namely alternating projection (AP), which is a combination of the alternating maximization and the projection matrix decomposition methods. The homogenous pixel selection and coherence estimation bias correction are conducted by the FaSHPS algorithm and DSIpro software toolbox. Results of simulations and real SAR data show that the proposed AP method could reconstruct credible phase series comparable to the quasi-Newton optimization algorithm [Broyden–Fletcher–Goldfarb–Shanno (BFGS)] while having three times the efficiency gain.