Omitted Variables Bias Research Articles

Which freedoms benefit the poor? A two‐horse race between economic and political freedoms on health‐adjusted life expectancy and child mortality, 1990–2020

AbstractWe examine theoretically and empirically whether democratic rights might be superior to economic rights in a two‐horse race, utilizing indicators of poverty reduction, such as health‐adjusted life expectancy (HALE) and child mortality as outcomes. The results show robustly that economic freedoms associate positively with HALE and negatively with child mortality, while the effect of democracy is more mixed. Studies reporting a negative effect of political freedoms on child mortality without accounting for economic freedoms, thus, are potentially mis‐specified. These results are robust to a barrage of tests, alternative data, estimating method and formal tests of omitted variables bias.

Estimando el efecto del clima en la agricultura con el valor productivo de la tierra agrícola

Objective: the Ricardian model, in which market land values are modeled as a function of climate, has been estimated extensively in the context of developing countries where its core assumptions are likely to fail due to missing or incomplete markets. This article proposes a new measure of land valuation that better reflects agricultural productivity in such contexts: the Productive Value of Agricultural Land (PVAL). Methodology: a three-step empirical strategy is applied to data from a survey of Mexican rural households. The first step is the estimation of an agricultural production function. In the second step, parameter estimates are used to calculate PVAL. In the third step, PVAL is used as the dependent variable in a Ricardian regression. Results: suggests that PVAL increases with more precipitation and decreases with by extreme heat. When the Ricardian regression is estimated using market land values, the positive effect of precipitation is underestimated and the effect of extreme heat on land productivity is null. Limitations: omitted variables bias could still influence the Ricardian estimates obtained using PVAL. Originality: a novel version of the Ricardian model is estimated, one that does not rely on market values of land. Conclusions: failing to account for the market setting of agricultural producers, particularly in developing countries, may lead to an underestimation of the effects of climate change in agricultural productivity.

Assessing omitted variables bias in intention-behavior linkages: Mitigation strategies and research implications

Omitting relevant variables in research models is a significant challenge in academic research involving non-experimental research methods. Omitted variables may bias the empirical findings and lead to erroneous conclusions about relationships between factors underlying information systems phenomena. Using data coded from 128 samples reported in 105 prior studies, this study applies meta-regression methods to quantify the extent to which omitted variables bias the reported effect sizes between behavioral intention and system use. Since a direct examination of omitted variables is not possible, this study quantifies four measures: independent variables common to both behavioral intention and system use, moderators for the relationship between behavioral intention and system use, moderators for relationships between other independent variables and system use, and control variables for system use. Meta-regression results show that the effect size for the relationship between behavioral intention and system use decreases when independent variables common to behavioral intention and system use or moderators for the relationship between behavioral intention and system use are included in research models. This implies that the non-inclusion of relevant variables distorts the effect size for the relationship between behavioral intention and system use resulting in a biased understanding of the relationships. It is crucial for research models to include relevant variables such that the omitted variables bias can be effectively handled. Several mitigation strategies and research implications are described.

A Factor‐Augmented New Keynesian Phillips Curve for the European Union Countries

AbstractIn this paper, a factor‐augmented version of the New Keynesian Phillips curve (NKPC) is assessed using a data set comprised of a large panel of European Union (EU) member countries. The factor‐augmentation is natural given that country‐level inflation rates are highly co‐moving. The presence of unattended common factors is important because it raises the issue of omitted variables bias, as the real marginal cost, which is a regressor of the NKPC, is likely to load on the same factors as inflation. One possibility here is to employ the regular instrumental variables approach. However, if the external instruments load on the same factors as the error term of the NKPC, the instruments would be invalid and the results would therefore likely be misleading. Motivated by this last observation, the present paper proposes a new estimator of the NKPC that allows for very general forms of factor dependencies and endogeneity. Our results provide evidence in support of the NKPC, but only after the presence of common factors has been appropriately accounted for.

Deworming and micronutrient status by community open defecation prevalence: An observational study using nationally representative data from India, 2016-2018.

Micronutrient deficiencies are widespread in India. Soil-transmitted helminth (STH) infections are acquired by interaction with soil and water contaminated by human feces and lead to blood loss and poor micronutrient absorption. The current recommendation for control of STH-related morbidity is targeted deworming, yet little is known about the effectiveness of deworming on micronutrient status in varying sanitation contexts. Ranging between 1% and 40% prevalence across Indian states, open defecation (OD) remains high despite India's investments at elimination by promoting community-wide sanitation. This variation provides an opportunity to study the relationship between deworming, micronutrient status, and OD at-scale. Cross-sectional datasets that were representative for India were obtained the Comprehensive National Nutrition Survey in 2016 to 2018 (n = 105,060 individuals aged 1 to 19 years). Consumption of deworming medication was described by age and community OD level. Logistic regression models were used to examine the relationship between deworming, cluster OD, and their interactions, with anemia and micronutrient deficiencies (iron, zinc, vitamin A, folate, and vitamin B12), controlling for age, sex, wealth, diet, and seasonality. These regression models further allowed us to identify a minimum OD rate after which deworming becomes ineffective. In sensitivity analyses, the association between deworming and deficiencies were tested in subsamples of communities classified into 3 OD levels based on statistical tertiles: OD free (0% of households in the community practicing OD), moderate OD (>0% and <30%), or high OD (at least 30%). Average deworming coverage and OD prevalence in the sample were 43.4% [IQR 26.0, 59.0] and 19.1% [IQR 0, 28.5], respectively. Controlling for other determinants of nutritional status, adolescents living in communities with higher OD levels had lower coverage of deworming and higher prevalence of anemia, zinc, vitamin A, and B12 deficiencies. Compared to those who were not dewormed, dewormed children and adolescents had lower odds of anemia (adjusted odds ratio 0.72, (95% CI [0.67, 0.78], p < 0.001) and deficiencies of iron 0.78, (95% CI [0.74, 0.82], p < 0.001) and folate 0.69, (95% CI [0.64,0.74], p<0.001)) in OD free communities. These protective effects remained significant for anemia but diminished for other micronutrient deficiencies in communities with moderate or high OD. Analysis of community OD indicated a threshold range of 30% to 60%, above which targeted deworming was no longer significantly associated with lower anemia, iron, and folate deficiency. The primary limitations of the study included potential for omitted variables bias and inability to capture longitudinal effects. Moderate to high rates of OD significantly modify the association between deworming and micronutrient status in India. Public health policy could involve sequencing interventions, with focus on improving deworming coverage in communities that have achieved minimum thresholds of OD and re- triggering sanitation interventions in high OD communities prior to deworming days, ensuring high coverage for both. The efficacy of micronutrient supplementation as a complementary strategy to improve nutritional outcomes alongside deworming and OD elimination in this age group needs further study.

From aid to resilience: Assessing the impact of climate finance on energy vulnerability in developing countries

Climate change is a major threat, and tackling it requires massive investment in sustainable practices and technologies. This paper explores the impact of the Global Climate Finance Fund (CF), including the Adaptation Finance Fund (AF) and Mitigation Finance Fund (MF), on energy vulnerability from 2000 to 2019 in a panel of 74 developing countries. Estimation using the Ordinary Least Squares (OLS) method showed that the CF significantly reduces energy vulnerability in developing countries. Furthermore, these results show that the MF has a greater reduction effect on energy vulnerability than the AF. To deal with the omitted variables bias and reverse causality, we applied the Oster (2019) stability test, and the Instrumental Variable Generalized Method of Moments (IV-GMM). The results confirm previous findings that climate finance reduces energy vulnerability, with mitigation finance having a larger effect. These results remained robust to the introduction of additional controls, to the exclusion of outliers, and to the use of alternative climate finance measures. Instrumenting climate finance by the share of important votes at the UN General Assembly and using three different instrumental estimation techniques, such as the Instrumental Variable Two-Stage Least Squares (IV-2SLS), Lewbel (2012), and Kiviet (2020), does not change our results. Finally, an asymmetry analysis through the application of the Canay (2011) quantile regression demonstrated that climate finance reduces energy vulnerability only for countries with a medium and high level of energy vulnerability. Based on these results, policy recommendations have been formulated.

Gender composition of children and sanitation behavior in India

Open defecation has been linked to various environmental and public health issues and has gained significant policy attention. Investing in better sanitation has also been advocated to provide women with privacy and protection from harassment. Nonetheless, previous research has shown that because of son-biased preferences, households in India underinvest in outcomes for their female children. Linking the gender composition of children to the sanitation behavior in Indian households, I find that households increase toilet ownership in the presence of female children by 5% in urban and 2.5% in rural areas, respectively. These results are robust to various empirical specifications, falsification tests, and bounding analysis to test for omitted-variable bias. Various heterogeneity analyses suggest that the decline in open defecation in response to having a female child is driven by households living in regions with a higher prevalence gender-based harassment and weaker law and order. These findings provide new evidence that differences in the gender composition of households create differential incentives to adopt toilets, which can be harnessed to reduce open defecation in India.

Does cycle commuting reduce the risk of mental ill-health? An instrumental variable analysis using distance to nearest cycle path.

Previous studies have linked cycling with improved mental wellbeing but these studies tend to use cross-sectional survey data that have small sample sizes and self-reported health measures, and are potentially susceptible to omitted-variable bias and reverse causation. We use an instrumental variable approach and an objective measure of mental ill-health taken from linked administrative data to ask: 'Does cycle commuting reduce the risk of mental ill-health?' Our study links data on commuting in Edinburgh and Glasgow from the Scottish population census with mental health prescriptions from the National Health Service Prescribing Information System records. We use road distance from home to nearest cycle path as an instrumental variable for cycle commuting. In total, 378 253 people aged 16-74years living and working in the City of Edinburgh and Glasgow City council areas at the 2011 census were included in our study; 1.85% of commuters in Glasgow and 4.8% of commuters in Edinburgh cycled to work. Amongst cyclists, 9% had a prescription for mental health compared with 14% amongst non-cyclists. Using a bivariate probit model, we estimate a mean average reduction in prescriptions for antidepressants and/or anxiolytics in the 5years following the census of -15.1% (95% CI: -15.3% to -15.0%) amongst cycle commuters compared with those who use any other mode to commute. This work suggests that cycle commuting is causally related to reduced mental ill-health and provides further evidence in support of the promotion of active travel to encourage commuters travelling shorter distances to shift to cycle commutes.

How spousal cognitive functioning affects the level of depression in middle-aged and older adults: An instrumental variable study based on CHARLS in China.

A better understanding of the causal relationship between spousal cognitive functioning and depression levels among middle-aged and older adults is vital for effective health policymaking under the globally severe aging challenge. However, the related evidence is often limited by potential omitted-variable bias and reverse causation. This study uses an instrumental variables approach, namely the two-stage least squares (2SLS) method, to examine the impact of spousal cognitive functioning on depression levels among middle-aged and older adults in China. The data were sourced from the China Health and Retirement Longitudinal Study (CHARLS) of 2020, including a total of 3,710 couples aged 45 years and above. Depression levels were measured using the Center for Epidemiologic Studies Depression Scale (CES-D-10), while cognitive functioning was assessed using the Mini-Mental State Examination (MMSE). Spousal social participation was employed as the instrumental variable to address omitted-variable bias and reverse causation. Additionally, an interaction effect test between gender and spousal cognitive functioning was conducted. The results show that for each one-point increase in the spouse's MMSE score, the CES-D-10 score of middle-aged and older adults decreased by 17.1% to 68.2%. The OLS results indicated that women, rural residents, and middle-aged individuals were more sensitive to these changes. The interaction effect test results confirmed that women were more affected by changes in spousal cognitive functioning. However, after a more reliable 2SLS analysis, the results for age groups shifted, showing that middle-aged individuals were more sensitive to these changes, with a decrease in depression levels reaching 70.0%, compared to 60.2% for the elderly group. Nonetheless, given the prevalence of depression among the elderly, the impact of spousal cognitive decline on depression in this group should not be overlooked. Our findings highlight the importance of spousal cognitive health in managing depression among both middle-aged and older adults, with particular attention to women and rural populations.

CEO succession and auditor going concern decisions: An analysis of outsider CEOs and generalist skills

This study shows how external CEO succession can mitigate external auditors’ propensity to issue a going concern opinion (GCO) in financially distressed firms, particularly when the outsider CEO possesses higher generalist skills relative to the outgoing CEO. We show that our findings are unlikely to be driven by factors such as pre-succession restructuring, institutional ownership, and board quality that could trigger strategic changes and the appointment of an outsider CEO. Further, we provide evidence to rule out the possibility of our results being influenced by self-selection or omitted variables biases. We also find that the ability of outsider CEOs to mitigate going concern opinions in their appointment year significantly contributes to the subsequent improvements in the financial performance of distressed firms. Further tests on some mechanisms through which new external CEOs can reduce the likelihood of going concern opinions show that firms with such CEO appointees are more likely to reduce dividends and labor costs, issue equity, and reduce product similarity of prospector firms relative to their peers.

Firm life cycle and cost of debt

Theory provides several channels linking the corporate life cycle and lending risks. Using a sample of 20,307 firm-loan observations spanning 5,076 publicly traded U.S. firms, we find an economically significant relationship between firm life cycle stage and lending spreads. Based on the Owen and Yawson (2010) life cycle stage classification, young firms are expected to pay at least 15 bps more than mature firms, whereas mature firms pay at least 11 bps more than old firms. According to the Dickinson (2011) cash flow classification, firms in the introduction and decline phases pay lending spreads that are 6 and 12 percent greater than firms in the mature phase. We explore omitted variables bias and instrumental variable estimation in robustness testing to alleviate endogeneity concerns. A mechanism analysis shows that credit risk, systematic risk, and idiosyncratic risk follow the corporate life pattern in accordance with loan spreads, suggesting that banks charge a premium to compensate for risk. Our results support the theoretical prediction that structural changes occur as firms evolve across the corporate life cycle.

Omitted Variable Bias of Lasso-Based Inference Methods: A Finite Sample Analysis

Abstract We study the finite sample behavior of Lasso-based inference methods such as post–double Lasso and debiased Lasso. We show that these methods can exhibit substantial omitted variable biases (OVBs) due to Lasso's not selecting relevant controls. This phenomenon can occur even when the coefficients are sparse and the sample size is large and larger than the number of controls. Therefore, relying on the existing asymptotic inference theory can be problematic in empirical applications. We compare the Lasso-based inference methods to modern high-dimensional OLS-based methods and provide practical guidance.

The returns to education and wage penalty from overeducation: New evidence from Vietnam

AbstractThis study intends to estimate the rate of returns to education in Vietnam, the distributive effects of education on wages, and the wage penalty from the incidence of overeducation in the Vietnamese labor market during 2004–2016. This study employs a pseudo‐panel approach to address omitted variables bias and the unconditional quantile regression to identify the heterogeneity of returns to education across the income distribution. Our main finding indicates that the estimated rate of returns to education in Vietnam is approximately 6.5%, showing a downward bias from previous estimates. The returns vary across wage distributions, where a lower rate of return is observed in lower quantiles and a higher rate among those individuals at the higher quantiles. The returns to education have declined since 2008, confirming the oversupply of highly educated workers in the Vietnamese labor market with an estimated wage penalty of 17%. Government assistance measures are needed to reduce the overeducation and the wage penalty issues in the Vietnamese labor market.

State ownership and credit rationing: Evidence From China

We investigate the impact of a firm’s ownership on credit rationing. We find that state-owned enterprises (SOEs) are less likely to be credit rationed than non-state-owned enterprises (non-SOEs). After controlling for a large set of control variables, we find that SOEs are 12.22 percent less likely to experience credit rationing in comparison to non-SOEs. This finding is robust to different definitions of SOEs and other regression methods accounting for omitted variables bias and endogeneity problems. Furthermore, the effects of state ownership on credit rationing are heterogeneous among different cities. Specifically, SOEs are less likely to be credit rationed in cities with low financial development and high government intervention. This may reflect the fact that implicit government guarantees of SOEs are more effective in these cities.

Revisiting the Twin Deficit Hypothesis in presence of exchange rate nonlinearities: evidence from India using nonlinear ARDL model

PurposeThis paper is an attempt to re-examine the validity of the Twin Deficit Hypothesis in the Indian economy, which is characterised by mounting inequality and liquidity constraints. The authors augment the econometric analysis with two important mediating variables, exchange rate and trade openness, to analyse their impact on current account deficit.Design/methodology/approachThe authors have used a ground-breaking asymmetric cointegration technique proposed by Shin et al. (2014) to investigate the short-run and long-run asymmetric nexus between gross fiscal deficit and current account deficit. In addition, the study has used asymmetric dynamic multipliers to see the dynamics of nonlinear adjustment from disequilibrium in the short run to equilibrium in the long run. The study has also used generalised impulse response functions to check the robustness of our cointegration results.FindingsUsing annual time series data from 1970 to 2018, the empirical exercise validates the presence of asymmetries in the Twin Deficit Hypothesis for the Indian economy. This study's robust findings demonstrate that the two deficits are asymmetrically related in the long run. The authors also found that exchange rate asymmetrically affects current account deficit thus validating the asymmetric J-curve phenomenon. From the causality analysis, the authors infer that there is a weak unidirectional causality running from fiscal deficit to current account deficit.Research limitations/implicationsFiscal deficit may cause current account deficit via changes in other macroeconomic variables that were not taken care of in this study. Therefore, the estimation techniques used in the present study might suffer from the issue of omitted-variable bias. Further research should include other macroeconomic variables where the twin deficit nexus is also influenced by other relevant variables. This will help in disentangling the indirect transmissions by which fiscal deficit translates into current account deficit.Practical implicationsThe results from our econometric exercise strongly suggest that the twin deficits are asymmetrically related. From a policy perspective, the asymmetric twin deficit nexus offers strong policy implications for the development of policies that are flexible enough to respond to shifts in internal and external sector dynamics. While framing the mechanism of fiscal prudence, policymakers in emerging countries like India must take into account the regime-changing behaviour of twin deficits.Originality/valueThe present paper is a significant contribution to the existing body of literature by being the first study in India which has analysed the Twin Deficits phenomenon in a nonlinear framework with the incorporation of asymmetric exchange rate dynamics in the model.

Breaking barriers and building sustainability: The importance of board feminization in ESG disclosures

This research examines the nexus between board feminization and environmental, social, and governance (ESG) disclosures in listed Kenyan firms. The positivist paradigm informed the study's research design. The dynamic generalized method of moments (GMM) was applied to eliminate endogeneity bias, omitted variables bias, measurement error, and firm-specific heterogeneity. Panel data regressions were applied on 467 firm‐year observations from 36 firms listed on the Nairobi Securities Exchange (NSE) from 2006–2019. The results showed that board feminization positively affects firms’ overall ESG disclosure measures. However, a component-level analysis revealed substantial heterogeneity in the findings among the different ESG pillars. This research attempts to address a gap in the existing knowledge by examining the relationship between female board presence and ESG reporting in emerging economies. The research supports regulators' efforts to enforce women's participation on boards and sustainable performance. This paper is among the first to provide in-depth empirical evidence on the link between board feminization on overall ESG disclosure and the different ESG pillars in Kenya. This research also adds to the expanding literature on ESG disclosure and board feminization in developing markets.

Currency Risk Premiums Redux

Abstract We study a large currency cross-section using asset pricing methods that account for omitted-variable and measurement-error biases. First, we show that the pricing kernel includes at least three latent factors that resemble (but are not identical to) a strong U.S. “dollar” factor and two weak high Sharpe ratio “carry” and “momentum” slope factors. Evidence for an additional “value” factor is weaker. Second, using this pricing kernel, we find that only a small fraction of the over 100 nontradable candidate factors considered have a statistically significant risk premium, mostly relating to volatility, uncertainty, and liquidity conditions, rather than macro variables. Authors have furnished an Internet Appendix, which is available on the Oxford University Press Web site next to the link to the final published paper online.

Do migrant remittances have state de-legitimizing tendencies? A micro-survey based evidence from Africa

This paper examines the micro-level link between migrant remittances and state legitimacy. The paper argues that there are two theoretical channels through which remittances may ultimately erode state legitimacy. First, remittance income earners may use remittance income to bribe state institutions, and thus may feel that they do not need to abide by the laws those state institutions enact. Second, remittances provide funds for political mobilization and thus may enhance uprisings against autocratic states. I test these claims using individual-level pooled data from 3 rounds of the Afrobarometer surveys. For the quantitative analysis, I use multiple linear regression, propensity score matching, and an informal method of testing for omitted variables bias. Overall, the findings of the study show that remittance-receiving Africans hold de-legitimizing behavior towards the state.

Does the economic growth target overweight induce more polluting activities? Evidence from China.

In China, official promotion evaluation based on economic performance motivates local governments to develop high economic growth targets, which has played an active role in boosting China's economic growth in the past decades, whereas its environmental consequences have not been fully exploited. This paper finds that the economic growth target overweight has a stronger positive impact on the output of high-polluting industries than on the output of low-polluting industries, thus inducing more polluting activities. To deal with the issues of reverse causality and omitted variables bias, we take an instrumental variable approach. Examining mechanisms, we show that economic growth target overweight promotes polluting activities through the deregulation of the polluting activities in high-polluting industries. We also find an increase in the impact of the economic growth target overweight after the 2008 global economic crisis. Our study provides new evidence for explaining the dual presence of rapid economic growth and heavy environmental pollution in China.

An Alternative Approach to Estimating Foreclosure and Short Sale Discounts

Current research documents astonishingly large price discounts for foreclosures and short sales. However, such outsized estimates may largely be due to omitted variables bias. We propose an innovative methodology relying on appraisers’ ability to match properties along both observable and unobservable attributes when performing appraisals. Our empirical approach, which relies on the use of appraisal fixed effects, produces foreclosure and short sale discounts of approximately 5% after controlling for a rich set of characteristics, including quality and condition, attributable mostly to the stigma associated with distress itself. We show that these lower estimates are not due to appraisers selecting high-price distressed properties as comps and are robust across a wide variety of subsamples and under alternative estimation methods.