Publications

Working papers

2021

Prison Rehabilitation Programs: Efficiency and Targeting

William Arbour, Guy Lacroix, Steeve Marchand

Increasing evidence suggests that incarceration, under certain circumstances, can improve inmates' social reintegration upon release. Yet, the mechanisms through which incarceration can lead to successful rehabilitation remain largely unknown. This paper finds that participation in social rehabilitation programs while incarcerated can significantly reduce recidivism. This result is entirely driven by inmates whose risk and needs were evaluated by a widely used assessment tool identifying their criminogenic needs. For this group, we estimate that participation in these programs reduces recidivism by about 9 percentage points within three years following release. Our results suggest targeting criminogenic needs is crucial for successful rehabilitation. We also find considerable heterogeneous program treatment effects:  inmates with a high overall risk score, or who exhibit procriminal attitudes, benefit little if at all from program participation. We investigate the stability of the treatment effect coefficients and conclude they unlikely suffer from an omitted variable bias.

Download document wp_eepp 2021-001

Immigrants' Economic Performance and Selective Outmigration: Diverging Predictions from Survey and Administrative Data

Charles Bellemare, Natalia Kyui, Guy Lacroix

We show that survey and administrative data-based estimates of a panel data model of earnings, employment, and outmigration yield very different qualitative and quantitative predictions. Survey-based estimates substantially overpredict outmigration, in particular for lower performing immigrants. Consequently, employment and earnings of immigrants who remain in the country are overpredicted relative to model predictions from administrative data. Importantly, estimates from both data sources find opposite self-selection mechanisms into outmigration. Differences hold despite using the same cohort, survey period, and observable characteristics. Differences in predictions are driven by difficulties of properly distinguishing non-random sample attrition from selective outmigration in survey data.

Download document eepp 2021-002

Infections, Accidents and Nursing Overtime in a Neonatal Intensive Care Unit (published, European Journal of Health Economics)

Marc Beltempo, Georges Bresson, Jean-Michel Etienne, Guy Lacroix

The paper investigates the effects of  nursing overtime on  nosocomial infections and medical accidents in a neonatal intensive care unit (NICU).  The literature lacks clear evidence on this issue and we conjecture that this may be due to empirical and methodological factors. We model the occurrences of both events using a sample of  3,979 neonates who represents over 84,846 observations (infant/days).  We  exploit an important change in workforce arrangement that was implemented in June 2012, and which aimed at reducing overtime hours to identify a causal impact between the latter and the two outcomes of interest.  We contrast the results using a standard mixed-effects logit model with those of a semiparametric mixed-effects logit model.  Contrary to the mixed-effects logit model, the semiparametric model  unequivocally shows that both adverse events  are impacted by nursing overtime as well as being highly sensitive to infant and NICU-related characteristics.  Furthermore, the mixed-effects logit model is rejected in favour of the semiparametric one.

Download document  wp eepp 2021-003

Can Recidivism Be Prevented From Behind Bars? Evidence From a Behavioral Program

William Arbour

Incarcerated offenders are offered a wide range of programs to encourage their chances of successful reintegration into society. Little is known, however, about the degree to which such programs improve prisoners’ reentry. In this paper, I study the effects of a cognitive-behavioral program implemented in Quebec, Canada, with a rich micro-level dataset. To manage the econometric issue of inmates’ self-selection into the program, I exploit inmates’ random assignment to probation officers who exhibit varying propensities to recommend the rehabilitation measure. I find large, negative, and significant effects of the program on recidivism, as measured by an inmate’s probability of serving a subsequent sentence: within one year following release, the program reduces recidivism by up to 18 percentage points. Moreover, the program is shown to decrease the number of future offenses. Further analyses indicate that the most plausible mechanism can be attributed to the program’s success in altering offenders’ preferences towards crime.

Download document wp eepp 2021-004

Évaluation de l'impact de la générosité des prestations sur la demande de congés parentaux au Québec

Nathalie Havet, Guy Lacroix, Morgane Plantier

Dans cette étude, nous nous intéressons à l'impact de la générosité des prestations parentales sur la demande de congés suite à une naissance au Québec. À partir des données administratives du ministère du Travail, de l'Emploi et de la Solidarité sociale (MTESS), couplées aux données fiscales de Revenu Québec (RQ), pour la période de 2006 à 2017, nous proposons pour la première fois d'évaluer l'effet causal du niveau des prestations familiales accordées sur la demande de congés des mères à bas revenus. En utilisant l'approche par Regression Kink Design (RKD), nos résultats concluent à un impact positif de la générosité des prestations sur la durée des congés parentaux demandés par les mères québécoises les plus modestes. Ils montrent également que cet effet varie selon la situation familiale de la mère, et en particulier qu'il tend à diminuer avec l'implication croissante du père. Notre étude permet ainsi de valider la pertinence de mesures basées sur un mécanisme d'incitation financière à destination des mères les plus modestes, mais également de quantifier l'effet attendu de ce type de dispositif.

Download document wp eepp 2021-005

Intrahousehold Resource Allocation and Individual Poverty: Assessing Collective Model Predictions against Direct Evidence on Sharing (published in The Economic Journal)

Olivier Bargain, Guy Lacroix, Luca Tiberti

Welfare analyses conducted by policy practitioners around the world usually rely on equivalized or per-capita expenditures and ignore the extent of within-household inequality. Recent advances in the estimation of collective models suggest ways to retrieve the complete sharing process within families using homogeneity assumptions (typically preferences stability upon exclusive goods across individuals or household types) and the observation of exclusive goods. So far, the prediction of these models has not been validated, essentially because intrahousehold allocation is seldom observed. We provide such a validation by leveraging a unique dataset from Bangladesh, which contains information on the fully individualized expenditures of each family member. We also test the core assumption (efficiency) and homogeneity assumptions used for identification. It turns out that the collective model predicts individual resources reasonably well when using clothing, i.e., one of the rare goods commonly assignable to male, female and children in standard expenditure surveys. It also allows identifying poor individuals in non-poor households while the traditional approach understates poverty among the poorest individuals.

Download document wp eepp 2021-006

The impact of tobacco tax reforms on poverty in México (published in Springer Nature: Business & Economics)

Luis Huesca, Abdelkrim Araar, Linda Llamas, Guy Lacroix

This paper investigates the impact of increasing the tobacco taxes on the poverty rate in Mexico.  Unlike most LMIC countries, the prevalence of smoking in Mexico is higher among the well-off than among the poor. Yet, tobacco tax rates in Mexico are lower than those in most LMIC countries. There is room, thus, to implement tax reforms and compensating policies to mitigate their impact on the poor.  Our analysis is based upon the  stochastic dominance approach.  More precisely, several tax reforms are analyzed through the so-called Consumption Dominance curves.  In addition, the reforms are assumed to be revenue neutral and to give rise to compensating subsidies on specific goods.  Our results show that if the Mexican government were to implement a WHO-type reform, poverty among households with at least one smoking member would increase by 2.6 percentage points.  Yet, the deleterious effects are entirely mitigated by price subsidies on staple foods.

Download document wp eepp 2021-007

Robust dynamic panel data using epsilon-contamination (forthcoming, Advances in Econometrics)

Badi H Baltagi, Georges Bresson, Anoop Chaturvedi, Guy Lacroix

This paper extends the work of Baltagi et al (2018) to the popular dynamic panel data model.  We investigate the robustness of Bayesian panel data models to possible misspecification of the prior distribution. The proposed robust Bayesian approach departs from the standard Bayesian framework in two ways. First, we consider the epsilon-contamination class of prior distributions for the model parameters as well as for the individual effects. Second, both the base elicited priors and the epsilon-contamination priors use Zellner’s (1986) g-priors for the variance-covariance matrices. We propose a general ``toolbox'' for a wide range of specifications which includes the dynamic panel model with random effects, with cross-correlated effects à la Chamberlain, for the Hausman-Taylor world and  for dynamic panel data models with homogeneous/heterogeneous slopes and cross-sectional dependence. Using a Monte Carlo simulation study, we compare the finite sample properties of our proposed estimator to those of standard classical estimators.  The paper contributes to the dynamic panel data literature by proposing a general robust Bayesian framework which encompasses the conventional frequentist specifications and their associated estimation methods as special cases.

Download document wp eepp 2021-008

2022

Using Machine Learning to Predict Nosocomial Infections and Medical Accidents in a NUCI

Marc Beltempo, Georges Bresson, Guy Lacroix

Frail newborns admitted to a NICU are at high risk of contracting a nosocomial infection and are vulnerable to medical incidents due to the intensity of care they require. In this paper, we use a gen-eralized mixed-effects regression tree model with random effects (GMERT-RI) to elaborate predictions trees for the two outcomes. GMERT-RI is an extension of the standard tree-based methods which is suit-able for binary unbalanced panel data with random effects. These account for neonate-specific unobserved variables that may impact their outcomes. The sample includes 4,356 neonates admitted in the CHU de Québec Level-III NICU between 10 April 2008 and 28 March 2013. The algorithm is fed numerous neonates- and NICU-specific daily predictors and manages to unearth non-linear relationships between them as well as identify critical predictor variables thresholds. Interestingly, in addition to usual clinical predictors, important ones are within management’s realm. Hence, machine learning algorithms such as GMERT-RI may prove to be a useful decision tool for management and pediatricians alike.

Download document wp eepp 2022-01

Home Ownership, Labour Market Transitions and Earnings

Carole Brunet, Thierry Kamionka, Guy Lacroix

The paper investigates the links between home ownership, employment and earnings. The motivation stems from the lack of consensus in the literature linking these outcomes. Our analysis is cast within a dynamic setting and the endogeneity of each outcome is assessed through the estimation of a flexible panel multivariate model with random effects. The data we use are drawn from the French sample of the EU Survey on Income and Living Conditions for the years 2004–2013. Our results show that while homeowners have longer employment and unemployment spells, they must contend with lower earnings than tenants upon reemployment. Importantly, our results highlight the importance of distinguishing between outright and indebted homeowners. Indeed, the latter are found to behave more or less like tenants on the labour market. At the aggregate level, thus, the positive relationship between home ownership and unemployment rate, known as Oswald’s conjecture, might thus depend on the share of leveraged homeowners in the population.

Download document wp eepp 2022-02

The Impact of Parental Benefits on Disadvantaged Households

Nathalie Havet, Guy Lacroix, Morgane Plantier

Over the past 25 years, the Government of Quebec (Canada) has introduced a number of relatively novel policies aimed at fighting poverty such as the Universal Child Care Program (UCCP) in 1997 and the Quebec Parental Insurance Program (QPIP) in 2006. The Canadian government has recently proposed to introduce the Early Learning Child Care Plan which more or less mimics the UCCP.  It is also considering introducing a similar parental leave program.  Since its inception, the QPIP has provided a means-tested supplementary benefits scheme for disadvantaged households.  The scheme yields a well-defined kink in the benefits schedule with respect to two entirely exogenous criteria. Using the QPIP administrative data files from 2006 to 2017, we estimate the causal impact of the supplemental benefits on leave duration and participation of poor households within a sharp Regression Kink Design (RKD) approach. Our results indicate that single mothers are relatively responsive to additional benefits.  Conversely, partnered mothers are not found to respond to the supplemental benefits, irrespective of fathers’ own participation in the parental leave. Our results may prove useful for the design of the Canadian parental leave program.

 Download document wp eepp 2022-03

Parole, Recidivism, and the Role of Supervised Transition

William Arbour, Steeve Marchand

We estimate the causal effect of parole on recidivism by exploiting the random assignment of parole board members to hearings in Quebec prisons. Board members vary in their propensity to grant parole and to place parolees to supervised halfway houses. We find that parole decreases the likelihood of recidivism by 8 percentage points within 5 years. Parolees at the margin of remaining incarcerated spend on average 4 fewer months incarcerated during the course of the next 5 years. This effect is largely driven by the direct release of parolees, but also by reduced incarceration time in future sentences. We further investigate the role of halfway houses in the reintegration process by estimating their effect on different groups of compliers. Our analysis shows that a stay in a halfway house is especially effective for convicts at the margin of remaining incarcerated.

Download document wp eepp 2022-04