Heckman Approach Research Articles

Reconciling information gaps in organic farm borrowers’ dealings with farm lenders

Purpose – Organic outputs have been increasing at much lower rates than growth in consumer demand. Organic farmers’ debt aversion hinders them from obtaining business funds through borrowing. The purpose of this paper is to clarify that the farmers’ reluctance to use debt as a funding option can be more attributed to gaps in existing borrower-lender relationships, beyond sustainability principles. Design/methodology/approach – Empirical evidence collected from organic farmers and farm lenders establish differing expectations and perceptions that reinforce the organic farmers’ debt aversion. The farm lender survey data set was analyzed using the Heckman approach applied to two lenders’ decisions: their interest in lending to organic farm borrowers and loan amounts approved for successful loan applicants. The econometric results were reconciled with the compiled inputs provided by organic farmers interviewed. Findings – Results validate the farmers’ lower reliance on loans due to suspicions that lenders lack knowledge and consideration of organic farming conditions and principles. Farm lenders must depart from employing a uniform credit risk appraisal model and adopt borrower-specific versions of the model, but not necessarily delineating organic-conventional farming dichotomy that may not substantially affect credit risk measurement. Organic farms, on the other hand, need to better understand the credit risk appraisal principles and use their inherent business strengths to compete for loans with conventional farms without any special consideration. Practical implications – Borrower-lender relationships can improve if information gaps between lenders and borrowers can be minimized with more extensive outreach education efforts. Better relationships would increase organic farms’ credit access to effectively address an impending supply gap in an expanding industry. Originality/value – To the knowledge, a specific focus on organic farms in understanding farm borrower-lender relationships has never been explored in literature.

Addressing unobserved endogeneity bias in accounting studies: control and sensitivity methods by variable type

Together with their associated statistical routines, this paper describes the control and sensitivity methods that can be employed by accounting researchers to address the important issue of unobserved (omitted) variable bias in regression and matching models according to the types of variables employed. As with other social science disciplines, an important and pervasive issue in observational (non-experimental) accounting research is omitted variable bias (endogeneity). Causal inferences for endogenous explanatory variables are biased. This occurs in regression models where an unobserved (confounding) variable is correlated with both the dependent (outcome) variable in a regression model and the causal explanatory (often a selection) variable of interest. The Heckman treatment effect model has been widely employed to control for hidden bias for continuous outcomes and endogenous binary selection variables. However, in accounting studies, limited (categorical) dependent variables are a common feature and endogenous explanatory variables may be other than binary in nature. The purpose of this paper is to provide an overview of contemporary control methods, together with the statistical routines to implement them, which extend the Heckman approach to binary, multinomial, ordinal, count and percentile outcomes and to where endogenous variables take various forms. These contemporary methods aim to improve causal estimates by controlling for hidden bias, though at the price of increased complexity. A simpler approach is to conduct sensitivity analysis. This paper also presents a synopsis of a number of sensitivity techniques and their associated statistical routines which accounting researchers can employ routinely to appraise the vulnerability of causal effects to potential (simulated) unobserved bias when estimated with conventional regression and propensity score matching estimators.

Hurdles for sports consumption? The determining factors of household sports expenditures

The purpose of this article is to explore the determining factors of household expenditures on sports participation. Due to a relatively large amount of zero-expenditures, simple regression methods are not suited. Because of methodological reasons, the two-step Heckman approach is used over the Tobit approach and the Double Hurdle approach. The participation decision (spend money or not) is influenced by sports participation of the parents, family income, education, sports club membership, and sports frequency. Determining factors of the intensity decision (amount of money that is spent on sports participation) are family income, sports participation of parents during their youth, sports club membership, sports frequency, age of youngest child, and household size. Moreover, the results indicate that a two-stage approach is needed because it gives a more in-depth insight in the household spending behavior. For example, higher educated households more often spend money on sports participation. However, this research demonstrates that once higher educated households have decided to spend money on sports participation, the amount of money spent does not differ from lower educated households.

Antiparkinson Drug Use and Adherence in Medicare Part D Beneficiaries With Parkinson’s Disease

BackgroundAntiparkinson drugs (APDs) are the mainstay of managing Parkinson’s disease (PD). However, there is paucity of evidence documenting patterns of APD use and examining factors associated with adherence to APDs. ObjectivesOur goal was to provide updated, comprehensive population-based data on APD use and adherence and to examine characteristics associated with adherence behaviors. MethodsWe analyzed data from 7583 beneficiaries with PD who had ≥2 APD prescription fills and were continuously enrolled in Medicare Parts A, B, and D for up to 19 months (from June 1, 2006, through December 31, 2007) or until death in 2007. We assessed 5 patterns of APD use: (1) concurrent use of ≥2 APD classes for ≥30 days; (2) switching of APDs from 1 to another; (3) augmentation of the existing regimen with a new APD; (4) duration of therapy, defined as days of APD treatment; and (5) adherence measured by using the medication possession ratio (MPR). We corrected for sample selection bias inherent in patients’ self-selection into either a Part D plan or a Medicare Advantage Prescription Plan by using Heckman’s 2-stage procedures. ResultsAPD users were pre-dominantly aged ≥65 years (93.6%), female (59.9%), and white (89.3%). Almost one half (43.2%) of APD users concurrently used ≥2 APD classes. One in 4 APD users experienced changes in their APD regimen, with 16.4% switching medications and 21.1% augmenting their current regimen. Three quarters of APD users had therapy lasting ≥436 days (75.3%) and an MPR ≥0.8 (72.7%). Multivariate analyses revealed that patients aged ≥65 years, of non-white race, non–low-income subsidy recipients, late Part D enrollees, cognitively impaired, highly comorbid, and who experienced multiple changes in APD therapy were less likely to adhere to APD therapy. We were able to generalize our findings to all Part D enrollees by correcting for sample selection bias using the Heckman approach. These population-level, generalizable observations provide better understanding of APD use and adherence and assist in the design of interventions for poor adherence. Limitations include cross-sectional study design and constraints in administrative data that preclude measurement of other potential factors related to adherence. ConclusionsA substantial proportion of these Medicare beneficiaries with PD used multiple APDs concurrently, experienced switching and/or augmentation of APDs, and had poor adherence to APDs. Patient characteristics and clinical and drug-related factors were important predictors of APD adherence.

An explanation of the relationship between adults’ work trip mode and children’s school trip mode through the Heckman approach

Most studies on children’s travel hypothesize that the characteristics of children, households, schools, and neighborhoods exogenously affect their travel mode to school. This study makes an additional assumption that children’s mode to school and household adults’ travel mode to work are interrelated because of a lifestyle choice made by parents and caregivers. With this assumption, Heckman probit models were used with data from the 2009 US National Household Travel Survey to predict household adults’ travel mode to work and children’s travel mode to school jointly. It found strong evidence that household adults’ decision to drive to work significantly increases the probability of children being dropped off at school and decreases their likelihood of walking and bicycling, but not vice versa. As adults’ mode choice is more fundamental in the household decision-making process, the study suggests that children’s mode choice studies should not ignore how parents or caregivers travel to work.

Diferenciais salariais para o setor cultural e ocupações artísticas no Brasil

This paper analyses the wage differentials among workers from the cultural sector and the artists in Brazil using data from PNAD - a household survey conducted by the Brazilian Statistics Bureau (IBGE) - from 2002 to 2007. In order to identify the features that contribute to the differences we use the fixed effects estimator in a cohort pseudo panel model, paying attention to the correction of the selection bias using the Heckman's approach. In general, workers from the cultural sector and artists have higher wages than other workers. After correcting the selection bias and fixed effects, these differences are even higher. By the Oaxaca decomposition analysis, the cultural sector characteristics are the most important factor that generates these wage differences.

Self‐reported side effects in buprenorphine and methadone patients receiving antiretroviral therapy: results from the MANIF 2000 cohort study

The aim of the study was to investigate the relationship between methadone and buprenorphine treatment and self-reported symptoms in HIV-infected opioid dependent individuals receiving antiretroviral therapy (ART). Longitudinal study. The French MANIF2000 cohort was used to compare self-reported symptoms in buprenorphine and methadone patients also receiving ART. We selected individuals receiving ART and OAT (342 visits among 106 patients). Symptoms were self-reported using a list of 14 symptoms (e.g. nausea, fatigue, fever) perceived during the previous 4 weeks, including three painful symptoms (abdominal or muscular pain, headaches). A two-step Heckman approach enabled us to account for the non-random assignment of OAT: a probit model identified predictors of starting either buprenorphine or methadone. A Poisson regression based on generalized estimating equations (GEE) was then used to identify predictors of the number of symptoms while adjusting for the non-random assignment of OAT. The median (interquartile range) number of symptoms was 4 (1-6) and 2 (1-6) among buprenorphine and methadone patients, respectively. After adjustment for non-random assignment of OAT type, depressive and opioid withdrawal symptoms, anxiolytics consumption and daily cannabis use, methadone patients were more likely to report a lower number of symptoms than those receiving buprenorphine. Methadone patients on ART report fewer symptoms than buprenorphine patients on ART under current treatment conditions in France. Further experimental research is still needed to identify an OAT-ART strategy which would minimize the burden of self-reported symptoms and potential interactions, while assuring sustainability and response to both treatments.

Explaining the Gender Wage Gap in Georgia

This paper evaluates gender wage differentials in Georgia between 2000 and 2004. Using ordinary least squares, we find that the gender wage gap in Georgia is substantially higher than in other transition countries. Correcting for sample selection bias using the Heckman approach further increases the gender wage gap. The Blinder Oaxaca decomposition results suggest that most of the wage gap remains unexplained. The explained portion of the gap is almost entirely attributed to industrial variables. We find that the gender wage gap in Georgia diminished between 2000 and 2004.

Predicted earnings and the propensity for self‐employment

PurposeThe purpose of this paper is to investigate the influence of the predicted earnings differential between self‐employment and wage‐employment on self‐employment propensities in Sweden using a large data set from the year 2003.Design/methodology/approachThe analysis in the paper is based on the presumption that the individual chooses to work in either the self‐employed or the wage‐employed sector. The separate earnings functions for the self‐employed and the wage‐employed are estimated in order to predict an individual's earnings in each sector. In order to overcome selectivity problems a Heckman approach is used at this stage. Finally, a structural probit model, where the difference in predicted earnings from the two sectors is included as an independent variable, is estimated.FindingsThe main result is that the predicted differential between self‐employment and wage‐employment earnings plays an important role for the self‐employment decision and that an increase in this earnings differential will lead to a higher self‐employment rate and to an increase in total employment in Sweden.Originality/valueThe policy relevance of this question is evident since previous research has shown that self‐employed individuals do not only create jobs for themselves but also for others. Thus, an increase in the earnings from self‐employment relative to the earnings from wage‐employment will increase the self‐employment rate as well as total employment.

Estimating differences in gasoline consumption among socio‐economic groups in Greece

AbstractThis study presents insights into the determinants of gasoline consumption for transportation based on cross section data for Greece. Given that car ownership decisions and gasoline demand decisions are taken jointly, in the empirical analysis, we employ the two‐step Heckman approach. Quantile regression analysis is used to investigate disparities in behaviour among different levels of gasoline consumption. The empirical results report that demographic and economic characteristics specify the consumer's choices with respect to gasoline consumption. An energy conservation framework can improve upon the situation by facing consumers as socio‐economic subgroups with different transportation needs and different lifestyle aspects.

Évaluation du coût associé à l’obésité en France

Summary Objective To estimate the costs associated with obesity in France. Methods We used data from the 2002 survey on health and social protection as well as health care consumption data from the EPAS sample of the national health insurance funds. After excluding those younger than 18 years, pregnant women and people diagnosed with cancer, we classified subjects as underweight, normal weight, or obese and then studied two subpopulations: (P1) obese patients (body mass index (BMI) ≥ 30 kg/m2) and (P2) patients who were obese or overweight (BMI ≥ 27kg/m2) and had additional cardiovascular risk factors: hypertension, dyslipidemia, type 2 diabetes, myocardial infarction (present or history), transient ischemic attack, intermittent claudication or sleep apnea. The Heckman approach was used to adjust for several factors. Results The study included 4651 individuals, with a mean age was 47 years. The obesity prevalence rate was estimated at 10.7% for P1; the P2 prevalence was 16.0%. Average healthcare costs for the entire study group were 1534€, 1105€ for ambulatory care (72%) and 429€ (28%) for inpatient care. Costs for obese subjects (P1 and P2) averaged 2500€, twice the costs for normal-weight people (1263€). After taking into account age, sex, socioeconomic status, alcohol consumption and smoking, the 100% reimbursement rate for chronic diseases (all else being equal), the extra cost of obesity, compared with normal weight, was estimated between 506€ (P1) and 648€ (P2). Conclusion The annual total cost of obesity was estimated to range from 2.1 to 6.2 billion euros in 2002 and account for 1.5 to 4.6% of total health expenditures in France.

Évaluation du coût associé à l'obésité en France

To estimate the costs associated with obesity in France. We used data from the 2002 survey on health and social protection as well as health care consumption data from the EPAS sample of the national health insurance funds. After excluding those younger than 18 years, pregnant women and people diagnosed with cancer, we classified subjects as underweight, normal weight, or obese and then studied two subpopulations: (P1) obese patients (body mass index (BMI) >or= 30 kg/m2) and (P2) patients who were obese or overweight (BMI >or= 27 kg/m2) and had additional cardiovascular risk factors: hypertension, dyslipidemia, type 2 diabetes, myocardial infarction (present or history), transient ischemic attack, intermittent claudication or sleep apnea. The Heckman approach was used to adjust for several factors. The study included 4651 individuals, with a mean age was 47 years. The obesity prevalence rate was estimated at 10.7% for P1; the P2 prevalence was 16.0%. Average healthcare costs for the entire study group were 1534 euro, 1105 euro for ambulatory care (72%) and 429 euro (28%) for inpatient care. Costs for obese subjects (P1 and P2) averaged 2500 euro, twice the costs for normal-weight people (1263 euro). After taking into account age, sex, socioeconomic status, alcohol consumption and smoking, the 100% reimbursement rate for chronic diseases (all else being equal), the extra cost of obesity, compared with normal weight, was estimated between 506 euro (P1) and 648 euro (P2). The annual total cost of obesity was estimated to range from 2.1 to 6.2 billion euros in 2002 and account for 1.5 to 4.6% of total health expenditures in France.

The Influence of Consumer Characteristics and Attitudes on the Demand for Organic Olive Oil

The aim of the present study is to examine consumer factors that affect the demand of organic olive oil (socioeconomic characteristics as well as attitudes towards organic products, food safety and the environment). Heckman approach for single equation procedure was applied in order to (1) identify the profile of regular buyers of organic olive oil based on the aforementioned factors and (2) to estimate income elasticity for the same product. Results indicate that the demand for or ganic olive oil is strongly affected by socioeconomic characteristics such as income size and occupation status, and to a lesser extent by attitudes towards organic products, food safety and the environment.

US physician board certification and labor market returns

PurposeThis paper aims to determine the characteristics of board certification among US physicians and to test whether accounting for the expected gains to certification alters the pattern of the determinants of board certification.Design/methodology/approachSplitting the sample into sub‐samples by characteristics associated with certification/non‐certification identified in a probit, the incremental gain to certification from log‐earnings equations is identified. Realizing that these methods are susceptible to sample selection, correction is made for it using the Heckman approach. Using the sample selection corrected equations, the expected gain to certification among those who certify is then predicted and those who do not certify is then predicted and this difference is included as a proxy for the expected gain in the original probit to ascertain whether including the expected gain alters the determinants of certification.FindingsAccounting for the expected gain alters the pattern of the determinants of certification. Although some groups such as blacks appear less likely to certify, after accounting for their expected return to certification, they are not as less likely. This is explained in terms of the expected marginal return to certification, market structure and practice setting.Research limitations/implicationsThe data used in the analysis apply only to young physicians in the USA. Also, these results may be applicable only to the particular cohort used in this analysis.Practical implicationsThe findings help to explain the absence of minority board certified physicians within the USA.Originality/valueThis paper is the first to simultaneously estimate the returns to physician board certification and the decision to obtain certification.

Deriving conditional and unconditional marginal effects in log earnings equations estimated by Heckman's procedure

Although the Heckman approach has often been used in empirical analysis, the marginal effects, necessary to interpret the effect of the regressors on the dependent variable, appeared to be overlooked. Using the Heckman approach, general expressions are derived for calculating the conditional and unconditional marginal effects. Based on a sample of Brazilian women, the conditional and unconditional return to education are calculated for the logarithm of earnings equation estimated by Heckman's procedure, comparing them to the marginal effect of education obtained without correcting for selectivity bias. The same analysis is carried out for a discrete variable ‘black’.

Income from Work Among Immigrants in Sweden

This paper uses a two‐step Heckman approach to investigate to what extent there are differences in income from work between immigrants and natives in Sweden. Contrary to previous studies this study takes the selection effect, i.e. the probability of having an income from work, into account when calculating the effect on income from work of a change in any of the explanatory variables. Our study shows that when the selection effect is taken into account, the differences in income from work between immigrants and natives are smaller than when the selection effect is not considered. We find that immigrants have a lower income from work than the native population when we control for variables such as schooling, experience, civil status and region of residence. Furthermore, immigrant cohorts with a recent year of immigration have a considerably lower income from work than earlier immigrant cohorts.

Calculating marginal effects in dichotomous - continuous models

In many economic settings, individual decisions can be viewed as a sequential process where a dichotomous choice is followed by a continuous choice. These processes are frequently encountered in consumption demand studies, where the decision of whether or not to consume a particular commodity is followed by the choice of how much to consume. The Heckman two-step approach has been extensively used in estimating these models. Expressions are derived for calculating marginal effects of regressors in dichotomous-continuous models. It is proposed that the marginal effect expressions are incomplete in almost all consumption demand studies that use the Heckman approach. In dichotomous-continuous models, a change in an explanatory variable that is common to both stages of the decision process has two effects: (1) it affects the likelihood of whether the commodity will be consumed; and (2) if the commodity is consumed, it affects the expenditure on that commodity. The first effect has so far been omitted from applied demand studies. The correct marginal effect expressions are derived for single-commodity and multiple-commodity demand models. An application to consumption survey data on 12 food commodities shows that erroneous marginal effect expressions can introduce substantial bias in demand elasticity estimates.

Exposing Our Ignorance: The Only “Solution” to Selection Bias

In his presidential address to the National Council on Measurement in Education (NCME) in 1982, Robert Linn (1983) argued that the problems of making inferences from biased (or selected) samples form an area sorely needing further research because they arise in so large a number of important educational studies. Wainer (1989) is to be congratulated for bringing together a wealth of ideas from several fields and for discussing both the problems of selection bias (or nonignorable nonresponse) and modern attempts to describe its effects in the context of educational research. His edited volume (Wainer, 1986) also deserves serious attention. In our opinion the key idea is analysis. There is no solution to the problem of selection bias in the sense of some automatic data analytic procedure that, once performed, removes the need for analysts to think further about selection bias. However, analysts can evaluate the sensitivity of their conclusions to various plausible models for Wainer discusses three approaches to model-based adjustment for nonignorable nonresponse: Rubin's mixture, Heckman's selection, and Tukey's selection models. In comparing Rubin's and Heckman's approaches, he contrasts Rubin's use of sensitivity analysis with Heckman's take-it-or-leave-it approach. Even though Heckman may not have intended his model to be used in this way, it is easy for a user of Heckman's approach to be fooled into thinking that once the model is estimated, the time for giving consideration to selection bias is over. This is because Heckman presents his model as something that may or may not apply to a given problem. The decision that the model applies is not based on data but on intuition or on prior knowledge. In Rubin's approach, the sensitivity of conclusions to assumptions is in the foreground and cannot be ignored. Tukey's selection model was suggested as a vehicle for making sensitivity analyses easier within Heckman's approach-hence, simplified selection. Because this may not be clear from Wainer's discussion, we shall compare the three approaches in terms of sensitivity analyses. For more details on Rubin's and Heckman's approaches, we commend the chapters on nonignorable nonresponse in Little and Rubin (1987) and in Rubin

Determinants of the Intrahousehold Allocation of Food in the Rural Philippines

AbstractThis study shows that the opportunity cost of time for husbands and wives can affect the intrahousehold distribution of food in a developing country. Equations which explain the relative allocation of calories within households were estimated for husbands, wives, and children with data for the rural Philippines. The estimating equation is derivable from both a joint household utility function and a bargaining model. Because the same households were included in the four survey rounds, the various observations for a given individual were averaged for each variable in the estimation procedure. Wages were estimated using the Heckman approach.

Allocation of goods through non-price mechanisms: Evidence on distribution by willingness to wait

This paper discusses a model of market clearing when prices are rigid. The estimated model is one in which the waiting time necessary to obtain the scarce goods brings the excess demand in line with current supply in a manner of a flexible price. The mechanism of such a market has been discussed by Barzel and by Nichols, Smolensky and Tideman, but little empirical evidence has been reported. Estimation of consumer demand in such a market is presented here. Heckman's approach to truncated sample estimation has been employed. The approach allows modelling of time costs as a two-part tariff. The results confirm that consumer's response to waiting time is measurable and in the same order of magnitude as expected from response to changes in prices. The study also indicates that rationing by willingness to wait does not necessarily distribute goods to the poor.