A class of improved heteroskedasticity-consistent covariance matrix estimators

The heteroskedasticity-consistent covariance matrix estimator proposed by White (1980), also known as HC0, is commonly used in practical applications and is implemented into a number of statistical software. Cribari–Neto, Ferrari & Cordeiro (2000) have developed a bias-adjustment scheme that delivers bias-corrected White estimators. There are several variants of the original White estimator that also commonly used by practitioners. These include the HC1, HC2 and HC3 estimators, which have proven to have superior small-sample behavior relative to White’s estimator. This paper defines a general bias-correction mechamism that can be applied not only to White’s estimator, but to variants of this estimator as well, such as HC1, HC2 and HC3. Numerical evidence on the usefulness of the proposed corrections is also presented. Overall, the results favor the sequence of improved HC2 estimators.

Conditional treatment and its effect on recidivism

The objective of this paper is to evaluate the effect of the 1985 ”Employment Services for Ex-Offenders” (ESEO) program on recidivism. Initially, the sample has been split randomly in a control group and a treatment group. However, the actual treatment (mainly being job related counseling) only takes place conditional on finding a job, and not having been arrested, for those selected in the treatment group. We use a multiple proportional hazard model with unobserved heterogeneity for job seach and recidivism time which incorporates the conditional treatment effect. We find that the program helps to reduce criminal activity, contrary to the result of the previous analysis of this data set. This finding is important for crime prevention policy.

Conditional cash transfers, schooling and child labor : micro-simulating bolsa escola

Cash transfers targeted to poor people, but conditional on some behavior on their part, such as school attendance or regular visits to health care facilities, are being adopted in a growing number of developing countries. Even where ex-post impact evaluations have been conducted, a number of policy-relevant counterfactual questions have remained unanswered. These are questions about the potential impact of changes in program design, such as benefit levels or the choice of the means-test, on both the current welfare and the behavioral response of household members. This paper proposes a method to simulate the effects of those alternative program designs on welfare and behavior, based on microeconometrically estimated models of household behavior. In an application to Brazil’s recently introduced federal Bolsa Escola program, we find a surprisingly strong effect of the conditionality on school attendance, but a muted impact of the transfers on the reduction of current poverty and inequality levels.

A conditional likelihood ratio test for structural models

This paper develops a general method for constructing similar tests based on the conditional distribution of nonpivotal statistics in a simultaneous equations model with normal errors and known reducedform covariance matrix. The test based on the likelihood ratio statistic is particularly simple and has good power properties. When identification is strong, the power curve of this conditional likelihood ratio test is essentially equal to the power envelope for similar tests. Monte Carlo simulations also suggest that this test dominates the Anderson- Rubin test and the score test. Dropping the restrictive assumption of disturbances normally distributed with known covariance matrix, approximate conditional tests are found that behave well in small samples even when identification is weak.

The long-term effects of conditional cash transfers on child labor and school enrollment

This paper investigates the long-term e ects of conditional cash transfers on school attainment and child labor. To this end, we construct a dynamic heterogeneous agent model, calibrate it with Brazilian data, and introduce a policy similar to the Brazilian Bolsa Fam lia. Our results suggest that this type of policy has a very strong impact on educational outcomes, sharply increasing primary school completion. The conditional transfer is also able to reduce the share of working children from 22% to 17%. We then compute the transition to the new steady state and show that the program actually increases child labor over the short run, because the transfer is not enough to completely cover the schooling costs, so children have to work to be able to comply with the program's schooling eligibility requirement. We also evaluate the impacts on poverty, inequality, and welfare.

Speculative bubbles and contagion: analysis of volatility’s clusters during the DotCom bubble based on the dynamic conditional correlation model

Revendo a definição e determinação de bolhas especulativas no contexto de contágio, este estudo analisa a bolha do DotCom nos mercados acionistas americanos e europeus usando o modelo de correlação condicional dinâmica (DCC) proposto por Engle e Sheppard (2001) como uma explicação econométrica e, por outro lado, as finanças comportamentais como uma explicação psicológica. Contágio é definido, neste contexto, como a quebra estatística nos DCC’s estimados, medidos através das alterações das suas médias e medianas. Surpreendentemente, o contágio é menor durante bolhas de preços, sendo que o resultado principal indica a presença de contágio entre os diferentes índices dos dois continentes e demonstra a presença de alterações estruturais durante a crise financeira.

Inference in differences-in-differences with few treated groups and heteroskedasticity

Differences-in-Differences (DID) is one of the most widely used identification strategies in applied economics. However, how to draw inferences in DID models when there are few treated groups remains an open question. We show that the usual inference methods used in DID models might not perform well when there are few treated groups and errors are heteroskedastic. In particular, we show that when there is variation in the number of observations per group, inference methods designed to work when there are few treated groups tend to (under-) over-reject the null hypothesis when the treated groups are (large) small relative to the control groups. This happens because larger groups tend to have lower variance, generating heteroskedasticity in the group x time aggregate DID model. We provide evidence from Monte Carlo simulations and from placebo DID regressions with the American Community Survey (ACS) and the Current Population Survey (CPS) datasets to show that this problem is relevant even in datasets with large numbers of observations per group. We then derive an alternative inference method that provides accurate hypothesis testing in situations where there are few treated groups (or even just one) and many control groups in the presence of heteroskedasticity. Our method assumes that we can model the heteroskedasticity of a linear combination of the errors. We show that this assumption can be satisfied without imposing strong assumptions on the errors in common DID applications. With many pre-treatment periods, we show that this assumption can be relaxed. Instead, we provide an alternative inference method that relies on strict stationarity and ergodicity of the time series. Finally, we consider two recent alternatives to DID when there are many pre-treatment periods. We extend our inference methods to linear factor models when there are few treated groups. We also derive conditions under which a permutation test for the synthetic control estimator proposed by Abadie et al. (2010) is robust to heteroskedasticity and propose a modification on the test statistic that provided a better heteroskedasticity correction in our simulations.

Inference in differences-in-differences with few treated groups and heteroskedasticity

Differences-in-Differences (DID) is one of the most widely used identification strategies in applied economics. However, how to draw inferences in DID models when there are few treated groups remains an open question. We show that the usual inference methods used in DID models might not perform well when there are few treated groups and errors are heteroskedastic. In particular, we show that when there is variation in the number of observations per group, inference methods designed to work when there are few treated groups tend to (under-) over-reject the null hypothesis when the treated groups are (large) small relative to the control groups. This happens because larger groups tend to have lower variance, generating heteroskedasticity in the group x time aggregate DID model. We provide evidence from Monte Carlo simulations and from placebo DID regressions with the American Community Survey (ACS) and the Current Population Survey (CPS) datasets to show that this problem is relevant even in datasets with large numbers of observations per group. We then derive an alternative inference method that provides accurate hypothesis testing in situations where there are few treated groups (or even just one) and many control groups in the presence of heteroskedasticity. Our method assumes that we know how the heteroskedasticity is generated, which is the case when it is generated by variation in the number of observations per group. With many pre-treatment periods, we show that this assumption can be relaxed. Instead, we provide an alternative application of our method that relies on assumptions about stationarity and convergence of the moments of the time series. Finally, we consider two recent alternatives to DID when there are many pre-treatment groups. We extend our inference method to linear factor models when there are few treated groups. We also propose a permutation test for the synthetic control estimator that provided a better heteroskedasticity correction in our simulations than the test suggested by Abadie et al. (2010).

Speculative bubbles and contagion: analysis of volatility's clusters during the DotCom bubble based on the dynamic conditional correlation model

Reviewing the de nition and measurement of speculative bubbles in context of contagion, this paper analyses the DotCom bubble in American and European equity markets using the dynamic conditional correlation (DCC) model proposed by (Engle and Sheppard 2001) as on one hand as an econometrics explanation and on the other hand the behavioral nance as an psychological explanation. Contagion is de ned in this context as the statistical break in the computed DCCs as measured by the shifts in their means and medians. Even it is astonishing, that the contagion is lower during price bubbles, the main nding indicates the presence of contagion in the di¤erent indices among those two continents and proves the presence of structural changes during nancial crisis