Fixed effects estimation of structural parameters and marginal effects in panel probit models

Fernández‐Val, Iván

doi:10.1016/j.jeconom.2009.02.007

Cited by 193 publications

(187 citation statements)

References 43 publications

Supporting

Mentioning

180

Contrasting

Unclassified

Order By: Relevance

“…Firstly, a fixed effects probit model is theoretically not possible (Cameron and Trivedi, 2005). Additional discrete-choice models (logit or tobit) allow us to adjust firm-specific effects but the coefficients could be severely biased with small Tperiods and a high number of individuals (Nickell, 1981;Greene, 2004;and Fernández-Val, 2009) or, as in our case, it might not be possible to estimate the model because of the excessive number of firm dummies. These problems are especially relevant in samples like ours which cover a very short time period and have a large number of individuals.…”

Section: _________________________mentioning

confidence: 94%

Exit from Exporting: Does Being a Two-way Trader Matter?

2015

View full text Add to dashboard Cite

The aim of this paper is to investigate whether the probability of ceasing to export is lower for firms that simultaneously import intermediate inputs and export (vertically specialized firms à la Hummels et al., The nature and growth of vertical specialization in world trade, 2001), once other firm characteristics are controlled for. On the basis of the estimation of a random-effects probit model with panel data, the authors find that the superior characteristics of these types of two-way trading firms (in terms of size, productivity, foreign ownership and skilled labor) explain their greater resistance to losing their status as exporters. However, even when these distinctive traits are controlled for, the effect of sourcing inputs from abroad on export exit is significantly different for large and small firms, playing a role in continuing to export only for the latter. Thus, it seems that small firms which are both importers of intermediates and exporters have an added advantage which enables them to confront the uncertainty of foreign markets in better conditions and translates to a lower likelihood that they will stop exporting.

show abstract

Section: _________________________mentioning

confidence: 94%

Exit from Exporting: Does Being a Two-way Trader Matter?

2015

View full text Add to dashboard Cite

show abstract

“…The coefficient estimates on FEEDIFF are 20% to 30% lower than in the original. These coefficients imply that a $434 fee increase raises cesarean rates by 2-2.5 percentage points.For this calculation we use the average marginal effect, the theoretically appropriate measure, increasingly recognized as preferable to the marginal effect evaluated at the mean of the independent variables, the measure used by GKM (Fernandez-Val, 2007). The two differ substantially here, as Table 2 shows, because cesarean probabilities are largely bifurcated across mothers: they are high for mothers with clinical indications for a cesarean, and low for those without such indications.…”

mentioning

confidence: 99%

Physician financial incentives and cesarean delivery: New conclusions from the healthcare cost and utilization project

Grant

2009

Journal of Health Economics

101

View full text Add to dashboard Cite

Abstract:This paper replicates Gruber, Kim, and Mayzlin's (1999) analysis of the effect of physician financial incentives on cesarean delivery rates, using their data, sample selection criteria, and specification. Coincident trends explain much of their estimated positive relation between fees and cesarean utilization, which also falls somewhat upon the inclusion of several childbirth observations that had been inadvertently excluded from their estimation sample. The data ultimately indicate that a $1000 increase, in current dollars, in the reimbursement for a cesarean section increases cesarean delivery rates by about one percentage point, one-quarter of the effect estimated originally. JEL Codes: I11; I18Keywords: cesarean section; financial incentives; HCUPThe cesarean section is one of the most common surgical procedures, and one of those most studied by economists, for good reasons: it absorbs billions of dollars of health care resources annually, is used with widely varying frequency across regions and across providers, and is potentially responsive to a variety of economic forces, including source of payment, malpractice liability, and financial incentives.Yet there are few studies of the last of these, financial incentives. One of the most recent and most expansive, by Gruber, Kim, and Mayzlin (1999, hereafter GKM) The accessible data, simple design, and striking conclusions of this study make it a suitable candidate for replication. Accordingly, we obtained the same data used by GKM originally and attempted to reproduce their results, generously assisted by the original authors, who shared their data assembly programs with us. In addition, as standard empirical practice in panel studies of this type has evolved somewhat since the original study was published, we also explored the effect of some of these changes on the findings-particularly the inclusion of state trends, which are clearly 2observable in the data, in the regression specification. These trends and, to a lesser degree, data assembly issues, are both important. We ultimately conclude that financial incentives may influence cesarean utilization in these data, but, if so, the effect is much smaller than originally advertised. This analysis related the probability of a cesarean delivery to the Medicaid physician cesarean fee premium (cesarean fee -vaginal fee) in that state in that year, FEEDIFF; state and year fixed effects, : and J; and controls for maternal demographics, hospital characteristics, and clinical indicators for cesarean section, X. Estimated using a logit model, the specification is as follows: where i, s, and t index individuals, states, and time; C is a dummy that equals one if individual i's delivery was performed by cesarean section and zero otherwise; 7 is the logistic distribution 3 function; and $ and ( are regression coefficients. We are mostly interested in the value of (. The effect of fee changes was predicted using the marginal effect of a one unit change in FEEDIFF at the mean of the independent variables. The model...

show abstract

“…First, the incidental parameters problem is most pronounced in large N, small T data sets (Heckman 1981), while it is less of a concern in datasets with long timeseries, such as ours (from 1781-2011). Second, we report marginal effects, and a recent study shows that even where the estimated coefficients of the fixed effects probit model might be biased, the bias for the marginal effects tends to be negligible (Fernández-Val 2009). However, to make sure that our choice of the fixed effects probit model is not driving our results, we re-estimate the baseline specification with a fixed effects conditional logit model, which does not suffer from the incidental parameters problem.…”

Section: Resultsmentioning

confidence: 99%