Different Ways of Losing: Public Defenders (and Private Counsel) at the Supreme Court of Argentina

Abstract

Though most countries have established public defense systems to represent indigent defendants, this is far from implying their offices are in good shape. Indeed, significant variation likely exists in the systems' effectiveness, across societies and at the subnational level. Defense agencies' performance likely depends on their configuration, including their funding, their internal arrangements, and their selection and retention mechanisms. Centered on public defense in Argentina, this article compares the performance of public and retained counsel at the country's Supreme Court. Public defenders' offices received a boost in the last two decades, and are institutionally well positioned to square off against prosecutors, putting them at least on par with the averaged retained counsel. Using a fresh dataset of around 3000 appeal decisions from 2008 to 2013, the study largely tests representational capabilities by looking at whether counsel meets briefs' formal requirements, a subset of decisions particularly valuable to reduce potential biases. It finds that formal dismissals are significantly less frequent when a public defender is named in an appeal, particularly when a federal defender is involved. It also discusses and tests alternative mechanisms. The article's findings illuminate discussions of support structures for litigation, criminal justice reform, and criminal defendants' rights.

1. Introduction

“[P]oliticians at every level of government remain almost completely silent about one of the biggest crises facing criminal justice: the utter collapse of indigent defense” (Pfaff 2016). Thus reads a recent article decrying the sorry state of public defense services in the United States. According to its author,

[i]f defendants had well-funded, effective representation, our adversarial system would do what it is intended to do. What we have right now, however, simply is not adversarial: relatively well-funded, well-staffed prosecutor offices square off against public defenders whose caseloads defy imagination. (Pfaff 2016)

The situation described in the United States may be to some degree generalizable. Rarely does the status of criminal defendants, and particularly indigent defendants, make the top list of societies' concerns. Most countries establish a public defense system for the latter, often coexisting with such other arrangements as a court-appointed counsel system. This does not mean, however, that public defense offices are in good shape, and reports highlighting the offices' dire straits probably do not come as a big shock. This represents but one more source of disadvantage for the worse off.

Still, this trend is likely not universal, and significant variation likely exists in the degree of effectiveness of public defenders' representation. In particular, the situation in Argentina, the subject of this article, seems to depart from descriptions of indigent representation in the United States, in part since, in the last decades, defenders' offices have received a boost. Especially at the federal level, but also in the case of some provinces, defenders are, on average, competent members of a bureaucracy that is lacking in some respects but is stable, autonomous, and relatively well staffed. They are also repeat players who have multiple iterations in court and can learn from past experiences. Also unlike in the United States, a defendant in Argentina can access a defender if he or she so chooses, regardless of the type and severity of the indictment or the appeal. Something similar may also be true of some other Latin American countries.

To start assessing the strength of public defenders' offices in Argentina, in this article I compare the performance of public and retained (i.e., private) counsel. I suggest that the offices' structure and organization would appear to make it possible for the average public defender (hereinafter, “defender”) to square off effectively against prosecutors, putting her on par with, or even on top of, the averaged retained counsel. Although they might manifest across the board, these advantages seem especially relevant concerning the focus of this article—appeals to the Supreme Court, the country's apex court (hereinafter, the “Court”). Appeals tend to involve technical expertise and are not as labor intensive as such other activities as sitting through a trial. I posit that the average defender is better positioned to master the technicalities involved in these appeals than the average retained attorney, or that, at the very least, she is not in a worse position. This stresses the relevance of an adequate institutional design (and proper funding) to ensure the representation of defendants, most of whom cannot find it in the private market.

Although I emphasize public defense, the previous intuitions say as much about private attorneys as they do about defenders. In Argentina, specialized criminal law firms are rare and the average defendant who can access private counsel likely cannot afford the best attorneys in town. Much as it has a bearing on the article's core ideas and findings, however, the structure of the private market for legal representation is partly a reflection of the centrality of public defense (Binder 2005). Demand for defenders has grown as a function of worsening socioeconomic indicators. Yet at least some of those who resort to a defender can afford a private attorney, so their choice may be guided by a perception of defenders' relative competence.

Given its institutional position and scarce resources, the Court summarily dismisses virtually all appeals it receives. There are, however, different ways of losing. Apart from a very infrequent win or defeat on the merits, defendant can receive either a dismissal due to formal deficiencies in the appeal brief or a dismissal at the Court's discretion. I largely test representational capabilities by looking at whether counsel (public or retained) meets appeal briefs' formal requirements, even if their case is dismissed on other grounds.

Three reasons explain that this way to test representational capabilities is adequate. Mastering the formal requirements of an appeal may yield only a Pyrrhic victory for counsel if the appeal will nonetheless be rejected on other grounds. However, losing due to procedural error more visibly shows the limitations of defendant's representation than receiving a discretionary dismissal, which may be a function of the Court's limited agenda. Meeting those requirements is a necessary condition for prevailing on the merits, and the least defendant can ask of counsel is that such errors are avoided. Secondly, irrespective of the value it has for defendants, avoidance of formal dismissals seems a good proxy for counsel competence, though one must bear in mind possible shortcomings regarding generalizability. Thirdly, this is an acceptable (though admittedly not perfect) strategy to minimize case-selection bias, since such requirements are independent of the potentially varying appeals' merits.

I find that the intervention of a defender is correlated with a lower proportion of dismissals due to formal deficiencies in appeal briefs (hereinafter “formal dismissals”). This effect is present in cases coming from both the federal and local jurisdictions, though it seems stronger in the former. Although these findings are consistent with the previous insights, I attempt to zero in on the mechanisms at play. I discuss an alternative set of theories that conceivably may explain outcomes, based either on the deference the Court may show defenders or on a less flexible stance toward retained attorneys. I thus attempt to disentangle three sets of mechanisms—the hypothesized one, labeled institutional capability, and the alternative ones, labeled deference and punishment, respectively. I believe it is the first that mostly explains outcomes.

The study combines quantitative and qualitative methodology. Among the very few to embark on a large-N exploration of the performance of public defense vis-à-vis retained counsel in Latin America, the study conducts regression analysis using a fresh data set of around 3000 decisions in criminal appeals issued by the Court in the period 2008–2013, a subset of these cases, and additional data from 2017. It also examines justices' coalitions for evidence of manipulation of standards, resorts to interviews of experts and officials, and explores subnational variation.

The questions I tackle are timely. As Binder, Cape, and Namoradze explain, “[i]n the last two decades, most Latin American countries have undergone, or are still undergoing, substantive changes in their criminal justice systems” (Binder et al. 2015: 3). These changes represent a shift from an inquisitorial approach to an adversarial approach and involve the strengthening of the rights of those subject to a criminal investigation (Binder et al. 2015). One key issue in such developments is defendants' right to an adequate defense, and, specifically, the consolidation of public defense offices. Although valuable comparative contributions have appeared on the right to defense in Latin America (Binder et al. 2015), to date there are only a small handful of studies directly assessing the performance of defenders in criminal cases. This article contributes to the literature on support structures for litigation, criminal justice reform, the rights of criminal defendants, and judicial politics.

2. Literature

The empirical scholarship on the effect of private and public counsel in criminal cases has mostly revolved around the United States, where defense services tend to be diagnosed as deficient, as the article's opening lines exemplify. I believe my emphasis on Argentina provides a good comparator for the United States, since public defense in the former seems to be better structured and financed than in the latter.

A comprehensive review of the US literature, which began in the 1960s, is beyond the scope of this article (see Emmelman n.d.). The scholarship varies in terms of the dependent variable studies employ—conviction rates, length of incarceration, probability of a guilty plea or a bail decision—as well as in the data they resort to and the way (if any) researchers deal with possible case-selection biases. As Williams points out, the literature is divided between studies that show “little to no difference” between counsel types regarding conviction rates and other outcomes and studies that do find a difference, mostly showing that retained counsel comes out ahead in some way (Williams 2013: 205–206). The former includes Wolf-Harlow (2000), Hanson and Ostrom (1998), Williams (2002), and, largely, Hartley et al. (2010), while the latter includes Nagel (1973), Champion (1989), and Williams (2013). Notably, no research seems to show that public defenders have clear advantages over retained counsel. Some studies have compared public defenders with court-appointed counsel, finding that the former achieve better outcomes (Anderson and Heaton 2012; Iyengar 2007). While this very roughly summarizes the literature, I next highlight selected arguments and findings.

In 1997, Stuntz suggested that the coexistence of overworked prosecutors and defenders and low-paid appointed counsel under a criminal procedure meant to protect defendants generated a bias against the poor (Hoffman et al. 2005; Stuntz 1997). Prosecutors do not have resources to litigate the time-consuming issues that may be raised by better-paid private counsel, so they choose to disproportionately prosecute the poor, whose counsel faces severe constraints (Stuntz 1997). Hoffman et al. (2005) found that defenders in Denver achieved poorer outcomes than private counsel in terms of the sentences defendants received. Part of the explanation lay in overburdened defenders, but they suggested the existence of self-selection by “marginally indigent” defendants, capable of hiring private counsel if charges were sufficiently serious but also, and perhaps especially, if they were innocent or thought they had a strong case; defenders probably received weaker cases (Hoffman et al. 2005).

Anderson and Heaton (2012) compared the performance of court-appointed attorneys and defenders in murder cases in Philadelphia, where one in five indigent murder defendants are randomly assigned a defender, and the remaining are assigned an attorney by the court. The presence of a defender was correlated with a decrease in conviction rates by almost 20 percent and a decrease in the probability of a life sentence by around 60 percent. They cited, among contributing factors, the low compensation for appointed counsel and the “relative isolation” of such counsel compared to the better-organized public defender units (Anderson and Heaton 2012). Before this, Iyengar (2007) had achieved comparable results.

In a rare study of a non-US jurisdiction, Huang et al. (2010) show that, in Taiwan, “defendants represented by public defenders tend to have higher conviction rates, but shorter sentences” if convicted than those represented by government-contracted attorneys. What about defenders in Latin America? In one of the few empirical studies I am aware of testing counsel performance, Quintana-Navarrete and Fondevila found that, in a populated Mexican state following a criminal justice reform, having a public defender “increased the likelihood of conviction,” what the authors credited to the “limited personal and institutional resources” of defenders (Quintana-Navarrete and Fondevila n.d.). Smulovitz (2019) focused on subnational variation of local offices in Argentina to analyze the determinants of defenders' availability and effectiveness. The dependent variable in the study, provision of defense, was operationalized in two ways, supply of defenders and a set of institutional features said to correlate with (i.e., arguably promoting) effectiveness. The author coded provincial offices according to four variables indicating levels of institutional strength—a coding I take advantage of. She found that supply of defenders and institutional design were uncorrelated, and that the former was mainly determined by the size of both the province's population and territory and the size of the lawyers' local market. Levels of local poverty and litigiousness did not correlate with defenders' supply (Smulovitz 2019). Another valuable contribution along these lines is Madeira's (2014), analyzing the gap between formal achievements of independence of Brazilian states' defenders' offices and “what actually occurs.”

In this article, I aim to expand the still embryonic exploration of public defense's effectiveness in Latin America.

3. Argentina

3.1 Public Defense

Criminal defendants without means in Argentina are by default assisted by defenders (López-Puleio 2007; La Rosa, 2001). This diverges from the situation in many nations (e.g., the United States, France, and Germany) where defendants are often represented by an attorney from either an association or the bar, even if the cost is shouldered by the state. Another difference with many states is that defenders in each of Argentina's jurisdictions (the country being a federal system) act within a single bureaucracy (Binder et al. 2015). A defendant can access the services of a defender if he or she so chooses. Unlike in many nations, like in the United States, this is a right the exercise of which depends neither on a means test nor on the type and severity of the indictment or the appeal. The Supreme Court has derived the right from the country's Constitution's right to defense in court (Section 18). The Court has long said that, barring exceptional circumstances, a criminal defendant cannot go unrepresented.¹ While a statute establishes a process of recoupment of fees incurred by a defender if defendants are able to afford them, the amounts set by courts are low and seldom enforced (Ley 24.946, Sections 63 and 64; Pérez-Otero 2015). Defenders' salary does not depend on recoupment.

Since the early 2000s, the country's Supreme Court has underscored the right to counsel (Asociación por los Derechos Civiles [ADC] 2005, 2008). The Court has said that the right should not be interpreted merely as one to access counsel, but that counsel must be minimally effective (ADC 2005). In doing so, the Court was acknowledging deficiencies in (public and private) representation, particularly though not exclusively at the local level.

3.1.1 Use

Partly because of availability, demand for public counsel in criminal cases is high, with a likely impact on the quality of services (Pérez-Otero 2015). The defense in Argentina has been dealing with a notable upsurge in demand, with the 2001 economic crisis counting as one of the factors behind it.² Defenders intervene in as much as 80–90 percent of federal criminal cases and in up to 80 percent of local criminal cases (Smulovitz 2019: 136). Demand is still high at the upper rungs of the judiciary, though the number of cases decreases, and the type of activity demanded at the appeal level is less time-consuming.

Most defendants who are socioeconomically worse off resort to the public defense, and, since most defendants are relatively worse off,³ most defendants do. But not only economically disadvantaged individuals use it. For example, hundreds of defendants in my initial data set were being prosecuted for their alleged involvement in crimes against humanity during the 1976–1983 dictatorship. Although these defendants ranged from top-ranked military officers to operatives, they clearly were not destitute on average. Yet, 84 percent of these defendants' appeals featured a defender. While this figure may partly be due to other factors, defendants likely would not access defenders' services if they perceived them as utterly incompetent.

3.1.2 Organization

Both each of the country's 24 local jurisdictions and the federal government have their own code of criminal procedure and their own regulations concerning the public defense, although crimes are listed in a uniform penal code for the entire country. Defenders at both the federal and the so-called national⁴ levels are grouped into the Federal Public Defender's Office (nowadays the Ministerio Público de la Defensa). While I mostly focus on this office, offices in several provinces have a roughly similar structure.

The federal defense received a significant boost in the last 25 years.⁵ Before 1994, the office had neither functional nor financial autonomy. Both prosecutors and defenders were part of the Ministerio Público, an institution presided over by the prosecutors' head. The institution itself was dependent upon the executive and did not feature in the Constitution. In 1994, a constitutional amendment introduced a new section (120) regarding the Ministerio Público. The Ministerio received recognition as an “independent organ with functional and financial autonomy [autarquía]” (Constitución de la Nación Argentina [1994], Section 120), and the functions of the prosecution and the public defense were divided, placing both on equal footing.

Pursuant to this amendment, in 1998, the Congress passed a statute (Ley 24,946) dividing the Ministerio Público into the Ministerio Público Fiscal (the prosecution) and the Ministerio Público de la Defensa (the defense).⁶ The heads of both the prosecution (the General Prosecutor) and defense (the General Defender) are appointed by the President with the Senate's consent via a two-third vote (Ley 24,946, Section 5). All other prosecutors and defenders are selected through competitive procedures. To be appointed, or to compete for another position, candidates sit for an exam.⁷ With variation, this is also true of defenders in some provinces which underwent similar processes, though, on average, the quality of local agencies is probably lower.

Defenders are officials serving for life, with their own office and staff. They are part of a large bureaucratic structure with different levels of management. The federal defender's office has a loosely hierarchical structure. According to a recent statute, which likely recognized what was already the case, defenders have “autonomy and technical independence” in dealing with cases and must “channel the instructions of [the represented individual] to search for the solution that most favors him or her, acting according to their technical expertise” (Ley 27,149 [2015], Section 17). Yet, they are subject to both instructions and general guidelines from superiors, the latter consisting of strategies in classes of cases. Performance-based evaluations do not have a direct impact upon the defender's career. Defenders, however, can be disciplined, either by the federal office's head or, in serious cases that can lead to removal, by a special panel (Ley 27,149 [2015], Sections 55–62).⁸

Defenders are largely organized mirroring the organization of the judiciary—trial stage, appellate stage, and so on. This has been criticized, since individual defenders are probably less acquainted with each case than they might be if they remained involved throughout; according to some observers, it also generates a scarcely efficient work distribution (CELS 2004). At the same time, these potential costs might be offset by the better specialization of defenders involved in each stage. While a small number of officials deal with most appeals to the Court at the federal level, others are in principle able to file as well. To illustrate, appeals from four defenders make up two-thirds of all federal appeals by defenders in the dataset. Yet, as many as 34 federal defenders intervened in appeals during the period (plus around 50 local ones).

3.1.3 Budget

Given the scarcity of data and “the differences in the scope and eligibility for legal aid and the legal service products the states make provision for” (CEPEJ 2014: 74), it is difficult to compare funding of public defense services across nations, though levels in Argentina do not seem considerably inadequate. Taking per capita spending in criminal and noncriminal defense services as a share of per capita GDP, Argentina's approximate figures for the federal and national levels only (around 0.05 percent in 2014)⁹ place the country above the mean in strongly varying Europe (0.023),¹⁰ with a relative level like that of Ireland (CEPEJ 2014).¹¹ Systematized figures for Latin America are unavailable, but the few I was able to process present strong variation as well, ranging from around 0.11–0.14 (in Costa Rica and Paraguay) to around 0.02–0.04 (in Chile, Ecuador, Guatemala, Colombia) to 0.005 (Bolivia).¹² Argentina is the only federal system of these, and, since the previous figure only covers the federal defense, a comparison is hard to establish. In the country, around 90 percent of the federal public defense's budget is devoted to salaries (Ministerio Público de la Defensa [MPD] 2018: 242).

3.2 Appeals at the Supreme Court

The Court deals with appeals coming from either the federal judiciary or local judiciaries insofar as they present a constitutional or federal issue that affects petitioner and is central to the case's outcome (Constitution of Argentina [1853], Sections 97 and 98; Ley 48 [1863]). Appellants file in the relevant lower court, which analyzes whether the relevant conditions have been complied with. This first appeal is called recurso extraordinario. If the appeal survives this supervision, the lower court sends the entire case file to the Court.¹³ If the lower court denies leave, which happens most of the time, appellant can still file a cert petition in the Court (queja) (Código de Procedimiento Civil, Sections 282, 285). The task in this instance is to show that the lower erred in dismissing the appeal. If the Court agrees, it asks the lower to send it the case file; if it rejects the appeal, the lower decision stays.

3.2.1 Requirements

Appellants file within an extremely short period of receiving notice of the decision (10 or 5 days); their appeal must challenge a “definitive” lower decision¹⁴ that comes from a so-called “superior court.” These form a subset of all formal legal requirements, as opposed to substantive legal issues. In 2007, the Supreme Court enacted a rule to further standardize the minutiae of appeals (Supreme Court, Acordada 4/2007).¹⁵ Table 1 summarizes these requirements; while some are entirely formal and unconnected to the case's or the brief's substantive merits, others are less clearly so. The Court embedded an escape valve in this rule by saying that it can determine, at its own discretion, to overlook an appeal brief's formal deficiencies if they “are not an obstacle to granting [the appeal]” (Supreme Court, Acordada 4/2007, Section 11). Decisions concerning this rule (“Acordada 4”) will be central to the analysis.

Table 1. Supreme Court Appeal Briefs' Requirements by Acordada 4 Section

Appellants deposit a fee to file a queja (Código de Procedimiento Civil, Section 286). Only if the Court overrules the lower court is the deposit returned (Section 287). During the period of this study, the fee amounted to US $800–1600. Appellants can avoid paying the fee if they resort to a proceeding in the lower court to show they cannot afford it. Defendants may defer payment until after the decision (Ley 21,859). Court regulations (Acordadas 13/90 and 35/90) establish that, if defendants choose to defer, they must provide their full name, address, and ID for the Court to demand payment if they lose. If this information is missing, the Court demands it once and, if there is still no compliance, it rejects the appeal. I study decisions regarding these regulations as well (I refer to them as “Acordada 13” decisions).

A recent study testing representational advantage at the Court looked at the operation of Acordada 4 (Muro et al. 2018). It codified the identity of litigants as government, corporations, prosecutors, and individuals. The authors found that the government (excluding prosecutors) received significantly fewer formal dismissals than nonformal (mainly discretionary) dismissals, and that this did not happen in the case of corporations. They also observed a lower proportion of formal dismissals in criminal appeals. But these included those filed by prosecutors and any type of counsel; they did not test the relevance of counsel type in criminal appeals.

3.2.2 Discretionary Dismissals

Only if appeals are free from formal errors can they be decided on the merits, though the Court is not forced to so decide. Confronting a rapidly growing docket, a statute in 1990 assigned the Court the power to dismiss any appeal that, at its discretion, “does not feature a sufficient federal issue” or revolves around “insubstantial or inconsequential issues” (Código de Procedimiento Civil, Sections 280, 285).¹⁶ I take advantage of the difference between formal dismissals (including Acordada 4 dismissals) and discretionary dismissals. Notice that, while these two types of decisions have a sequential logic, they really do not comprise two independent stages of decisionmaking.

3.2.3 Secretaría Penal

The Court is a large institution with career staff. Each justice is assisted by a staff of clerks and employees. Moreover, the Court features specialized offices receiving appeals by subject matter (criminal law among them). The relevant office undertakes a preliminary supervision of formal requirements and drafts a first memo to be distributed among the justices (Muro et al. 2018). Justices study the file and can freely alter the memo.

4. Theory and Mechanisms

I believe that, because of the way it is arranged, the public defense is either better equipped than retained counsel to deal with appeals or at least as well equipped as the latter. While one can regard this as applying in principle to both decisions on the merits and those supervising compliance with formal requirements, I restrict attention to the latter for reasons I explore below concerning selection bias.

I discuss first the article's main theoretical insights to account for a potentially observed empirical regularity—a higher probability of surviving a formal dismissal in the case of public defenders' appeals. I then turn to alternative explanations of this regularity. A crucial difference between the two sets of explanations is that only in the case of the former are the Court's outcomes a function of counsel's legal skills as revealed on appeal briefs. Only if the main theory holds can we say that being assisted by a defender is beneficial to or at least not detrimental for, a defendant, and, further, that a potentially observed lower rate of formal dismissals in the case of defenders can be taken as an indicator of their competence.

4.1 Main Theory: Institutional Capability

Taking a case to the Supreme Court demands knowledge of the Court's internal proceedings, case law, and substantive law. Considering that appeals are filed within a very short time span, the task can be taxing even if discrete requirements (e.g., attaching copies) are easy to comply with. Several organizational traits may help defenders navigate this labyrinthine process, relative to the average retained attorney. Three sets of traits are next presented that, separately or jointly, may give the average defender an advantage.

4.1.1 Competence, Stability, and Specialization

The competitive recruitment and permanent tenure of defenders likely promote the selection of competent professionals who are attracted by the promise of a life career and are able to develop expertise.¹⁷ While one expects these traits to benefit defenders in general, they likely give them an edge at the level of appeals, where case overload is not as serious an issue and cases tend to reflect the attorney's skills at legal argument in writing instead of more labor-consuming activities.

The more competitive selection processes in place since defense agencies were overhauled—particularly at the federal level—likely improved the average competence of appointed defenders. Defenders are also full-time specialized attorneys with a subject matter to focus on, whereas most private attorneys in Argentina often represent cases in manifold matters. Specialized criminal law firms are not so common, and few defendants can afford to retain the best criminal law attorneys. On average, this may lower the competence of retained attorneys appearing at the Court compared to defenders in general and those specializing on appeals to it in particular.

Public defense in Argentina seems to feature most of the elements Weber posed as defining characteristics of a bureaucracy, particularly recruitment based on specialization, expertise, and office holding as a career (Weber 1978). Similarly, in Down's account, a bureaucracy's members are full-time workers who depend on their job for most of their incomes and are hired, promoted, and retained based upon performance assessment (Downs 1965; Jaffee 2008). Now, defenders' performance is not directly relevant for retention, so a valid question is: Why would they be diligent? A first possibility is that at least some of the office's members are, following Weber, professionals with a career vocation. A second possibility concerns incentives. Defenders still face informal pressure or reprimands in case of poor performance, outside of the boundaries of what is considered “normal.” And, although remote, there is a chance that they face disciplinary action in case of notably bad performance. A big mistake might also hurt future chances at promotion.

4.1.2 Directions, Communication, and Support

Compared to the likely isolation of the average private counsel, both top-down directions from the organization's head and peer-to-peer formal and informal communication may keep defenders updated on regulations and promote consistency.¹⁸ From the bottom-up, defenders also benefit from their reliance on staff, which may help them lighten their workload. These traits facilitate the transmission of knowledge among defenders and staff and the pooling of resources and create a support structure for defenders.

4.1.3 Repeated Interactions

Many more individual defenders than individual attorneys are repeat players in that the former appeal more often to the Court (Galanter 1974). Compared to the fewer than 100 defenders intervening in two-thirds of all cases in my dataset, over 650 attorneys are named in appeals in the remaining third (some may belong to the same firm). Since appeals involve a degree of intellectual gymnastic, counsel's frequent interactions with the Court can be an indicator of their focus on the Court's regulations and case law. Interactions may also facilitate the gradual accrual of experience, allowing counsel to learn from past errors. In the case of defenders, this can be the silver lining to their high demand, though frequent interactions likely yield diminishing returns at some level.

Now, the hypothesized capability of defenders likely results from a whole set of traits that include but reach beyond repeated iterations. Also, the little evidence there exists seems to indicate that experience by itself may not be enough. In a study of the South African Supreme Court of Appeal, Haynie and Sill (2007) found that counsel's capabilities (i.e., previous litigation success) were a significant predictor of success while litigation experience (i.e., previous appearances) was not.

4.1.4 Federal versus Local Defenders

At the subnational level, there seem to be clashing theoretical insights regarding the article's comparison of representational capabilities. On the one hand, the previous mechanisms would seem to be particularly at play at the federal level. Federal defenders appear more often at the Supreme Court than local defenders do, and, on average, they seem better prepared than their counterparts if all the country's local jurisdictions are bundled together.¹⁹ On the other hand, there are still fewer specialized criminal law firms—or firms that litigate at the Court—at the subnational level. In this case, even a relatively lacking local public defense may compare well to local retained counsel.

4.2 Alternative Explanations: Deference and Punishment

The explanation of defenders' hypothetically observed advantage, particularly at the Court's gatekeeping stage of supervision of briefs' formal requirements, may lie elsewhere. A crucial assumption so far has been that, by and large, the Court (and the specialized office) is unbiased when undertaking such supervision. I posited that formal dismissals and discretionary dismissals followed a lexicographic order, according to which the Court and staff only resort to the latter to dismiss if the appeal cannot be dismissed on formal grounds. But maybe justices and staff are more willing to turn a blind eye to defenders' briefs' formal errors—and to avoid pointing fingers at them. Why may this be the case?

A first answer (that I label “deference”) combines four complementary explanations. Firstly, there may be the sense on the part of Court justices and staff that defenders—officials, like them, of the justice system writ large—deserve special institutional respect. Relatedly, the difference may be due to the informal relations the defenders' interactions may develop (Galanter 1974: 99). Thirdly, the shared notion may be that defenders have a high workload. What would be gained from conveying the message that they are doing their job poorly? Fourthly, justices and staff may sense that, since defendants assisted by defenders are relatively worse off, additional sources of disadvantage must be kept at a minimum.

According to these explanations, the Court may opt to disguise a technical error made by a defender by issuing a discretionary dismissal instead of a formal dismissal, thus being deferential (or forgiving) toward them. This type of manipulation would hardly extend to decisions on the merits since it is implausible to conceive of justices as willing to overturn the lower court just because a defender is named. To the extent that deference does exist in decisions regarding briefs' requirements, it might emerge more strongly at the federal level. The Court's justices, federal defenders, and their respective staff spend their workdays at the relatively close-knit institutions structuring the federal judiciary, with a degree of interaction among them. On the other hand, provincial defenders may face a high workload as well, and justices might also be forgiving toward them because of this.

Much connected to "deference", to the point that it may be indistinguishable from the latter, another explanation is centered on retained attorneys. I label it “punishment.” In a limited number of appeals (recursos extraordinarios), upon the attorney's petition, the Court establishes a fee that counsel collects from defendant (in addition to retainers). If a case is dismissed due to an Acordada 4 error, counsel forfeits the right to collect (Section 11). Perhaps the Court wishes to penalize retained attorneys when they do not properly discharge their duties by applying the requirements without exception. Since defenders virtually never collect this fee, it would not make much of a difference which motive of dismissal the Court invokes when they intervene. Yet, the Court very rarely sets any fees in criminal appeals, so the explanation can be partial at best.

In terms of the implications for defendants, are not all theories identical? Ultimately, no matter what theory accounts for it, defendants lose most of the time. Yet, there are different ways of losing. Only if the main theory is at play are defenders' briefs not crippled by relatively simple, foreseeable errors as often as retained attorneys' briefs are. Save for exceptional circumstances, complying with formal requirements is necessary to get a favorable decision, and the least one can expect from counsel is that such errors are avoided. Defenders' appeals may be discretionarily dismissed, but this does not necessarily point to a problem in their capabilities. A good way to conceive of this is that, ex-post, it may not matter for defendant how a case is rejected. Ex ante, however, it does. If defendant had the chance to choose counsel, he or she would select one who promised to avoid formal errors. Under the alternative explanations, defenders often avoid a formal dismissal even if they have committed an error, but there really is no attached benefit to defendants from accessing a defender.

4.3 Expectations

I offer two testable expectations. One has to do with the regularity itself. Appeals filed by defenders will face a lower proportion of formal dismissals than those filed by retained counsel, or at most the same proportion. The second regards the mechanisms at play. I argue that it is the main theory instead of the alternative explanations that accounts for the outcomes.

Exclusively looking at decisions on the merits may have been considered an intuitive way to go about comparing counsel performance, but case-selection bias becomes a very important hurdle. An apparent feature of these decisions is that they can deal with much varied substantive matters. Not enough data are available to match all cases by crime, and no data are available to match cases raising similar issues. Since there is no random assignment of cases to either public or retained counsel, it can be that each counsel type deals with cases of a different nature, and, thus, that differences in outcomes are explained by unobserved peculiarities of the cases each type receives (Anderson and Heaton 2012).

Several factors may bias the estimation of differences. Conceivably, moral hazard might underplay the performance of defenders. Defendants assisted by defenders do not fully internalize the costs of appeals, so, on average, they may tend to reach the Court with weaker cases. On the other hand, it could be that the pool of appeals featuring a retained attorney disproportionately excludes cases of defendants who—either because of the merits of their case, the quality of counsel received, or both—avoided an indictment or negotiated a plea before reaching the Court. In addition, a defendant with retained counsel at the lower courts may have turned to a defender at the Supreme Court or the other way around. Thus, the substantive merits of the appeal may be hard to disentangle from the activity performed throughout the case, which may or may not have been undertaken by counsel of the same type.

Focusing on briefs' requirements strongly minimizes these problems, though, admittedly, they do not entirely disappear. The issue of compliance with requirements is (or should be) by and large independent from the appeals' substantive merits and the substantive issues involved. This is particularly so in the case of most formal Acordada 4 (and Acordada 13) regulations, which are entirely unconnected to the merits. Further, if the Court decided to dismiss, there would not be much reason for it to apply formal requirements in a biased way because of the case's varying merits. In the case of other appeals requirements, particularly that the decision is “definitive,” a link cannot be discarded between the Court's decision and the lower decision, but this is expected to disadvantage defenders. Defenders often are instructed by the agency's head to appeal decisions even if they are not considered “definitive” by the Court, likely to promote a change of precedent. Overall, decisions regarding formal requirements lend themselves as particularly strong candidates for assessing counsel performance.

A second possible source of bias is the Court's manipulation of formal requirements to benefit or disadvantage counsel types, and the alternative explanations entirely rely on this notion. Some considerations may suggest that this bias is unlikely present across the board. Several of the requirements are crystal clear, and full-scale manipulation may be hard to sustain over time without potential observers likely noticing it. Also, appeals are first filed in the lower court, which undertakes a first analysis of compliance with requirements. This is an indisputably limited, but perhaps nonnegligible, instance of control.

5. Data and Empirical Strategy

5.1 Data

I use a fresh data set of around 3000 Supreme Court decisions in criminal defendants' appeals from the period 2008–2013. (I also collected very basic data from decisions issued in 2017.) I consider the time range appropriate to measure the application of the new regulation enacted in 2007. I searched decisions via the Court's webpage by entering two keywords which were likely to be present in all appeals (causa and penal). I then discarded appeals from the prosecution, misclassified cases, and a very small handful of cases with unidentified counsel or where a legal aid organization was intervening. I also dropped all cases I could identify as being related to the prosecution of crimes against humanity during the country's last dictatorship. In 2005, the Court struck down statutes limiting such prosecution, and new cases cascaded. An argument can be made that these cases are to some degree bundled together and settled by macro definitions of judicial politics instead of at the individual level. I was careful to ensure that this decision did not affect results.

Although it is difficult to determine exactly what percentage of total criminal appeals were captured by entering those keywords—the percentage is likely very high—I am certain that the strategy did not introduce any bias. Taking the Supreme Court's official statistics for 2012–2013, my observations for those years amount to half of criminal appeals reported, but the official numbers include cases I discarded, including appeals by the prosecution and the revenue service and hundreds concerning the dictatorship.

Using the scarce information displayed in each decision, I coded several variables including decision date, lower court intervening, type of appeal, name of counsel, type of decision, and, if available (around 30 percent of the time), the crime being prosecuted; the motive of appeal was largely unavailable. I created dummies for case dispositions, discretionary dismissals, and formal dismissals, discriminating in the latter case between Acordada 4 dismissals, Acordada 13 dismissals, and other formal dismissals. I sorted counsel into public and retained, sometimes resorting to external sources. When a case name appeared more than once, only one decision was preserved at random to the extent that: (1) the case file was the same and (2) the appellant was the same. Appeals with repeated names were kept otherwise—it could be that a person was also being indicted for other crimes or that a case file involved several defendants. Thirty-five cases were dropped. The main results largely do not change if only keeping one appeal at random in cases involving the same file, resulting in the exclusion of around 120 cases.

5.2 Empirical Strategy

Supervision of compliance with formal requirements should be undertaken independently of the cases’ merits. Yet, I tried to minimize the possibility of selection bias concerning the cases each type of counsel receives by creating a subset of comparable cases. To make up for the paucity of information available in the Supreme Court decisions, I secured access to around 50,000 decisions coming from the federal appeals court immediately below the Court.²⁰ I searched among them for every federal case file in my Court dataset to get data that were missing from the Court's decision. I then identified appeals to a conviction involving actual jail time (i.e., a sentence of more than 3 years), arguably among the most serious appeals in the dataset, resulting in a total of 207 cases after the exclusion of a handful of cases following the strategy identified above for repeated files. I believe this subset identifies sufficiently comparable appeals, not in the arguments raised, but at least in what was at stake. I report results of tests employing both the general dataset and this subset.

On a small number of occasions (17), two appeals within the context of the same file featured a different counsel type. I report results of tests using this matching, since it represents another way to control for cases' merits. Finally, to test whether it is defenders' repeated interactions that explains differences, I codified the number of interventions of each defender and calculated the proportion of each defender's formal dismissals (and Acordada 4 dismissals) taking into account all decisions involving her. The mean intervention for a defender was around 24 cases (sd = 69), and 53 (sd = 101) for federal ones. I assume that defenders' work is stable over time. I did not attempt the same for retained counsel, since 75 percent of attorneys intervened in just one case and 90 percent in at most two (mean = 1.46, sd = 1.15).

I modeled decisions in favor or against appellant as a function of counsel type, case origin, and the identity of repeated defenders, plus some controls and an interaction. Given the nested nature of the data, one initially intuitive option was to create a counsel-level data set by collapsing decisions by individual counsel and generating rates of “wins” or “loses” at the Court to use as dependent variables. A very serious obstacle for this was retained attorneys' low-intervention rates. I estimated logistic regression models for both the whole dataset and the subset of convictions involving jail, each corresponding to different ways a counsel type comparison can be conceived. The dependent variable always codifies a binary outcome, the Court's disposition, but considers different types of decisions. To deal with the nested nature of the data, I clustered standard errors by counsel, and, in case that more than one counsel was present, by the first to be named.

Moreover, I studied voting coalitions at the Court (more on this below) and conducted 10 semistructured interviews with specialists and officials. A challenge was to decide on a sample universe for potential participation (Robinson 2014). I privileged the view of officials at the Court by interviewing two officials (relatores) currently in charge of drafting memos within the specialized office in criminal matters and five justices' clerks or relatores (a permanent position). Four of the latter were present during all or part of the period 2008–2013, three of them no longer at the Court. The remaining subject presently works at the Court but did not do so during the period of the study. I also interviewed one leading criminal law scholar and two public defense officials, one at the federal level and one at the provincial level. All were guaranteed anonymity in the hope that this would incentivize candid answers to capture the mechanisms.

5.2.1 Telling the Mechanisms Apart

If the predicted regularity is observed, one can only be relatively confident that the main theory accounts for it if the influence of the alternative explanations can be either rejected or minimized, and no further biases are suspected. At the core of the alternative hypotheses is manipulation of formal standards. I first search for the presence of manipulation indirectly by resorting to data coming from different sources—the general dataset as well as fresh cases from 2017—and looking at different regulations—Acordada 4 decisions, Acordada 13 decisions, and decisions on all formal regulations combined. There is perhaps less reason to suspect manipulation at the Court if the regularity is observed at two different points in time (2008–2013 and 2017, the latter featuring a change of personnel, with four justices leaving and two joining a now five-member Court) and across regulations.

I more directly address the possibility of manipulation of formal requirements both by studying justices' voting coalition profiles and, importantly, by using interviews. Concerning the former, concurring opinions are particularly useful. If the main plurality opinion—that is, the Court's opinion—decides to invoke lack of compliance with requirements, a concurring opinion may choose instead to advocate for a discretionary dismissal, and vice versa. Thus, concurrencies may be a sign or signal of manipulation. Firstly, I study the Court's decisions to issue a formal dismissal. A concurrent opinion suggesting dismissing the appeal but through a discretionary dismissal might be a sign of its author's willingness to overlook the requirements. If these concurring opinions are more widespread in the presence of defenders than in the presence of retained counsel, this may suggest that at least some justices at the Court (though not a majority) are more willing to do so in the former case.

Afterward, I focus on the opposite—the Court's discretionary dismissals. A higher number of concurring opinions agreeing that the appeal must be dismissed but arguing that requirements were not complied with in cases with defenders may point to manipulation of formal standards by the plurality when defenders are present. This would suggest bias on the part of a majority of justices. The sign or signal would be missing if all justices were involved in such manipulation, but this is unlikely since even a perfunctory look at cases reveals that the Court does often issue formal dismissals when defenders intervene.

One could entertain the possibility that manipulation originates not with the justices but with the specialized office that prepares a first draft reacting to the case. If so, even if the justices themselves were not in favor of it, manipulation might persist because the justices are disinclined to spend time modifying the original draft, keen as they may be on optimizing their resources. However, the office serves the justices, who likely would not tolerate such manipulation over the long run in case they did not condone it. To the extent that manipulation exists, it is likely to be reflected in justices' opinions.

6. Results and Discussion

6.1 Descriptive Statistics

Both discretionary dismissals and formal dismissals make up over 90 percent of all decisions (59 and 33 percent, respectively). Acordada 4 dismissals represent slightly less than half of formal dismissals, and almost 90 percent of the former involve the clearly formal requirements identified in Table 1.

Table 2 breaks down appeals by defender type and the origin of appeals. Of all cases, 83 percent are federal (or “national”) appeals and two-thirds feature a defender. Defenders are considerably more prevalent at the federal level. Table 3 displays different case outcomes by type of counsel. The proportions of formal dismissals and the subset of Acordada 4 dismissals are smaller for defenders than for retained attorneys, while the opposite is true in the case of discretionary dismissals. Only in 5 percent of federal appeals with defenders do defenders receive a dismissal that cites the Acordada 4 requirements, compared to around 38 percent in the case of retained counsel; the difference is slightly smaller in local appeals. A difference still exists but it is smaller in formal dismissals other than those citing the Acordada 4 regulations, two-thirds of which concern the lack of a “definitive” decision.

Table 2. Private and Public Counsel at the Supreme Court by Appeal Origin

Federal cases include 24 local appeals in which defender was federal.

Table 3. Supreme Court Outcomes by Type of Decision, Counsel, and Appeal Origin

The table also shows similar proportions of granted appeals—overruling the lower—for both counsel types taking all cases into consideration. Notice in Table 4 that, once formal dismissals are excluded, the proportion of cases prevailing on the merits is eight points larger in the case of retained counsel in federal appeals. Apart from the instruction to appeal defenders often receive from the office's head, the discrepancy might conceivably be explained by moral hazard, as noted above. Perhaps there is some indication of this. Once discretionary dismissals are also excluded (bottom row of Table 4) and attention is restricted to the cases the Court decides on the merits—an admittedly imperfect measure of case strength—the proportion of “wins” becomes almost identical for both counsel types at the federal level. This likely reinforces the inadvisability of looking at merits decisions.

Table 4. Supreme Court Outcomes by Counsel and Appeal Origin Excluding Formal Dismissals

6.2 Main Outcomes

Table 5 displays the outcomes of three estimations. For the first two (1–2), I employ all appeals, while model 3 excludes formal dismissals and only considers both discretionary dismissals and decisions on the merits. The two models in Table 6 (4–5) replicate models 1–2 but by employing the subset of appeals to a conviction involving jail time. The dependent variable in all models is binary. In models 1 and 4, it is whether appeals survive a formal dismissal. In models 2 and 5, it is whether they survive the subset of Acordada 4 requirements. These four models are keys to the analysis. In model 3, the dependent variable is the disposition of the case once formal dismissals are excluded. While the previous discussion suggested that these cases are not the best-suited for a comparison of representational capabilities, I included a model considering them to provide a more complete view. Since there are only 12 overturned appeals in the subset of convictions involving jail, I do not offer the equivalent of model 3 using these cases.

Table 5. Regression Outputs, All Appeals

Counsel-clustered se values (by first counsel named in appeal) in parentheses. In 24 local cases, in which defender was federal, federal takes value 1.

Table 6. Regression Outputs, Appeals to Convictions Involving Jail

Counsel-clustered se values (by first counsel named in appeal) in parentheses.

The presence of a defender is always correlated with fewer formal (models 1 and 4) and Acordada 4 (models 2 and 5) dismissals, with a p value always lower than 0.05. In terms of average marginal effects, the probability of surviving a formal dismissal is predicted to be around 38–46 percentage points higher in the case of a public defender than in the case of a retained attorney, and the probability of surviving an Acordada 4 dismissal is predicted to be around 22–33 points higher in the case of the former. (Considering confidence intervals instead of single point estimates, and depending on the models, these differences range from a 13-point increase in the presence of a defender to a 66-point increase.)

The interaction of defenders and federal appeals is positive in models 1–2—the presence of a federal defender is correlated with fewer formal dismissals—with the coefficient's p value lower than 0.05. In line with what Table 4 showed, the presence of a defender is negatively correlated with case dispositions once formal dismissals are excluded (model 3), with a p value close to, but larger than, 0.05. The probability of a favorable case disposition is predicted to be around 7 points lower in the case of a defender, and a difference ranging from a 3 percentage-point increase in the presence of a defender to a 17-point decrease is also compatible with the data.

Of interest among the controls in models 1–2 and 4–5 are defenders who are frequent appellants, operationalized in the former as those who filed at least 100 appeals²¹ and in the latter as those who filed at least 20. The picture concerning them is unclear. Only in the case of two defenders (#36 and #62) do coefficients with the same sign have a p value lower than 0.05 in more than one model, but the sign is different across defenders. Defenders handling appeals formerly without counsel (in forma pauperis) receive fewer Acordada 4 dismissals in model 2 (p value <0.05), a result compatible with the notion that the Court is more lenient toward these defendants. The Buenos Aires province public defense stands out in models 1–2 (and 3 as well); its presence is always correlated with a higher probability of a positive outcome.

To complement these findings, Figure 1 presents histograms of the proportions of appeals surviving an Acordada 4 dismissal by individual defender, federal, or local. (The pattern of overall formal dismissals is similar.) Performance is more uniform at the federal level. The proportions of formal and Acordada 4 dismissals for each defender are also modeled as a function of number of appearances. Figure 2 depicts scatterplots of the correlations, with regression lines. While the slopes point upward, the effect is small and p values are larger than 0.05. It is not evident that repeated interventions directly affect performance. A nonlinear model might better fit the data, suggesting an optimal number of interventions, though this is not explored since observations above that conjectural level are too few.²²

Figure 1. Histograms of Federal and Local Defenders' Proportions of Appeals Surviving an Acordada 4 Dismissal (with Number of Interventions). [Color figure can be viewed at wileyonlinelibrary.com]

Figure 2. Correlation of Individual Defenders' Proportions of Appeals Surviving a Formal and Acordada Dismissal, Respectively, and Number of Interventions. [Color figure can be viewed at wileyonlinelibrary.com]

Since many local defense agencies have not received the overhaul operated at the federal level, the good outcomes that the subset of local defenders' appeals show relative to retained attorneys are intriguing. The well-reputed defense from the Buenos Aires province accounts for part of this effect, but other local defenders do well too. To explore subnational variation, I constructed a conceivable proxy of institutional strength resorting to Smulovitz (2019). In it, local offices got a minimum score of 2 and a maximum of 9, with the latter indicating greater institutional strength. The author ranked local defense offices according to institutional measures only partly coinciding with those I discussed—the separation of prosecution and defense offices, the defense's policy and budgetary decision-making power, the permanence of defenders throughout a case, and defenders' specialization.

This measure of public defense strength is uncorrelated with a variable codifying each province's proportion of defenders' local appeals surviving a formal or Acordada 4 dismissal.²³ Interestingly, however, my dataset only features 4 provinces with at least 10 appeals naming a defender,²⁴ and 7 provinces (out of 24) were outright unrepresented in the pool of defenders' appeals. While the mean “Smulovitz score” of the 17 provinces with at least 1 defender appeal was 5.5, the mean of those that did not have any was 3.3, and the mean of those with at least 10 appeals was 6.5. Herein may lie the explanation of the finding that local defense seems not to lag much behind its federal counterpart. A biased sample of local offices is observed at the Court, and those getting to it are better on average.²⁵ This unanticipated outcome points to a serious access-to-justice problem in some provinces.

I complement the previous tests. Firstly, I consider dismissals invoking the Acordada 13 regulations (for lack of inclusion of appellant's information) among the cases in the data set. A regression is not needed—only in 1 out of 40 dismissals invoking the Acordada 13 did the appeal feature a defender. Next, I look at Acordada 4 dismissals by using fresh data from 2017. As Table 7 shows, around 70 percent of appeals featured a defender, a result I obtained by taking a sample of the year's appeals. Yet, of all dismissals I found based on lack of compliance (86 cases), around 90 percent were in appeals with a retained attorney. Finally, I consider the 17 occasions in which two appeals were submitted in the context of the same file with different counsel type. Of the 34 appeals, 10 were dismissed for lack of compliance with Acordada 4. Two featured a defender.

Table 7. Acordada 4 Decisions by Type of Counsel in 2017

The previous findings seem to suggest that defenders do well on appeal comparatively when decisions on briefs' formal requirements are considered. The results show across categories of cases and regulations and at different points in time, perhaps lending some support to the main theory instead of the set of rival explanations.

6.3 Alternative Explanations: Voting Coalitions

I argued that, if justices are in some way biased in favor of defenders, this may show in their voting profile. I studied decisions to dismiss both out of discretionary motives and out of formal deficiencies, specifically those citing the Acordada 4 regulations. Table 8 summarizes the findings.

Table 8. Voting Coalitions at the Supreme Court

Firstly, I looked at dismissals in federal appeals based on lack of compliance with Acordada 4 regulations. A concurrent opinion advocating instead a discretionary dismissal might be an invitation on the part of the concurring justice to turn a blind eye to formal errors. Looking at all cases where the Court invoked the Acordada 4, concurrent opinions suggesting using instead the discretionary dismissal powers were not more frequent when defenders were named. Two justices' votes account for 90 percent of all votes advocating this criterion—Justice Zaffaroni, a liberal criminal law expert, signed 53 percent of them, followed by Justice Highton (36 percent). These justices likely had qualms about the application of the new regulations across the board to defendants.

I did observe in these dismissals more dissenting opinions in the presence of defenders, particularly in the direction of granting the appeal. This seemed in part a function of the legal issues involved (justices limited themselves to citing a precedent). Justice Zaffaroni, again, signed most of these dissenting votes. Moreover, in a handful of cases, concurring or dissenting justices explicitly said that they would exempt appellant from meeting formal requirements, though this was not more prevalent in cases with defenders.²⁶

Next, I focused on the opposite—decisions to issue a discretionary dismissal. I analyzed a sample of slightly over 20 percent of federal appeals thus decided with each counsel type. Concurrent opinions invoking lack of compliance with briefs' requirements—a potential sign that the plurality is engaged in manipulation—were not more widespread in the case of defenders. Indeed, they were more prevalent in the case of retained attorneys, so it might be thought that individual justices (though not a majority) were inclined to show less leniency toward the latter. But this is unlikely, since all these concurring votes in the sample were authored by either Justice Petracchi or Justice Argibay (or both), and some did pertain to appeals with public defenders. If these justices had a less deferential position toward retained attorneys, they likely would not point to Acordada deficiencies in public defenders' appeals in cases where the majority dismissed on discretionary grounds.

In sum, the notion that justices generally or disproportionately engaged in manipulation to cater to defenders (or to disadvantage retained attorneys) does not seem to be supported by the data thus far.

6.4 Interviews and Discussion

If the previous outcomes pointed in the direction of the main explanation, the interviews support that explanation as well, albeit introducing some nuance. What follows from virtually all the interviews is that public defense comes out ahead, at least, but perhaps not only, at the level of appeal briefs' requirements.

In the view of interviewee A (legal scholar), unlike “battleground” counsel at the investigatory or trial stage, most defenders filing appeals, particularly at the federal level, are specialized “desktop” defenders with guiding criteria and competence relative to the average attorney reaching the Court. According to interviewee B (specialized office), public defenders at the federal level but also in some provinces are generally very competent professionals. While on occasion defenders overlook arguments, they are technically superior to retained attorneys; “compared to retained defenders,” said B, “public defenders are excellent.” B added that, “in general, public defenders are very neat,” and mentioned that the “public defense adjusted itself well to the Acordada [4].” Interviewee C (specialized office) maintained that defenders' appeals (particularly in cases coming from federal courts and the Buenos Aires province) seldom present technical errors, while interviewees D, H, I, and J (current or former clerks) underscored public defense competence as well, either in absolute terms or relative to retained attorneys (or both). Interviewee G (official in a well-reputed local office) described the training those filing appeals receive, including on Acordada 4 requirements, and that officials frequently meet to discuss the Court's new case law and regulations. At the office, they have “more than one check of whether the [briefs' formal] requirements […] were complied with.” In contrast, added G, already at the level of the local Supreme Court, “one observes that retained attorneys are rather lacking regarding technical issues.”

The question remains whether the Court enforces formal requirements in a biased way based upon counsel type. Interviewees C, D, H, I, and J said that, to the extent of their knowledge, such a difference did not exist. C and D added that a minority of justices (especially former Justice Zaffaroni) was more disinclined to invoke the Acordada regulations, and H said that, when the Acordada was adopted, some justices considered that it was not to be applied in criminal cases, a view that did not prevail. G expressed a similar view, and said that, in cases of minor errors, the Court tends not to be so rigid in the supervision of Acordada requirements. C argued that in cases where Acordada deficiencies are not too serious, the office often suggests two dispositions, applying the Acordada and issuing a discretionary dismissal, for justices to decide on one. In turn, J mentioned that, if the appeal presents weighty arguments, the Court may be inclined to overlook minor formal errors, but that this was irrespective of counsel type.

F (former clerk) said that it could be that cases that should have been formally dismissed received a discretionary dismissal, and added that this partly depended on the “knowledge and rigorousness” of the official writing the first memo or studying the case. In line with this, D said that, in some cases, justices decided on a discretionary dismissal because they could not agree on an Acordada dismissal. F also argued that justices on some occasions dismissed cases on formal or discretionary grounds when the correct choice was to decide them on the merits. Like C, H, I, and J, however, both F and D denied, according to their experience, that type of counsel mattered, though added that officials at the specialized office may know more about the issue.

Asked about whether manipulation based on counsel type could take place, interviewee A entertained the possibility, “perhaps at the level of the specialized office,” but said that it was not “a determining factor.” Interviewee E (federal defense) said something similar regarding the federal defense, and added that, right after the adoption of the new regulations (Acordada), it could be that the Court felt somewhat “shy” about applying it to public defenders, but that over time the application became more even. (As Figure 3 shows, dismissals grew somewhat after the first year, but in a more sustained way in the case of retained counsel.)

Figure 3. Proportion of Acordada 4 Dismissals by Year and Counsel Type. [Color figure can be viewed at wileyonlinelibrary.com]

In B's view, the Court is “more demanding with retained attorneys” when applying the formal requirements of Acordada 4. This is because it is understood that public defenders represent worse-off defendants—people who “need a public defender and are exercising their right to appeal in an extreme form,” regardless of whether they filed in forma pauperis. When retained attorneys fail to comply with requirements, the Court often reacts more strictly. By taking this stance, the Court conveys something like: “You are charging money for this.” B's statements (diverging from the rest) initially lend support to the “deference” and “punishment” explanations.

In B's words, if the Court is disinclined to hear a public defender's appeal since it does not present a weighty federal issue, and the appeal contains a minor formal error, the Court may decide to overlook the error and choose a discretionary dismissal instead. However, B implies that this more flexible approach toward defenders takes place somewhat at the margins, since, “when there is lack of compliance by a public defender, the problem usually is very small.” B said that “one rarely finds a defenders' brief that does not comply with the formal Acordada 4 requirements.” And, when the error is “very slight,” the Court tries to avoid dismissing out of lack of compliance, largely irrespective of counsel type.

To summarize, the predicted regularity was extensively found in the data. In adjudicating between the rival sets of explanations accounting for it, I believe I found strong support for the main theory. Following interviewee B's responses (and perhaps A's and E's), I cannot dismiss the possibility that the “deference” and “punishment” explanations have pull as well, but there does not seem to be reason to suspect that these mechanisms are widespread. While this is admittedly conjectural, B's answers are conceivably compatible with a different explanation—that what appears at first sight as partiality (whether justified or not) on the part of at least some officials at the specialized office and arguably at the Court really is the fact that defenders commit fewer and less serious errors. As the interviewed subjects emphasized, defenders seem better equipped to deal with appeals and to do better on average at least at the gatekeeping stage. In sum, I cannot reject the possibility that the alternative mechanisms have some explanatory force. Yet it still appears that the main theory received the most empirical support.

7. Conclusions

The article's main question was whether public defense in criminal appeals at the Supreme Court did well compared to retained lawyers. The intuition was that this should be the case, particularly in terms of surviving the supervision of briefs' requirements. The main result I found is that defenders received considerably fewer formal dismissals than retained attorneys did. The observed advantage survived controls and was present both at the federal and local level, though it was bigger in the former. I tested alternative explanations under the twin notions that justices may cater to defenders (“deference”) and/or reprimand retained attorneys who fail to discharge their duties (“punishment”). Although there may be hints of a certain level of manipulation in the supervision of formal standards to benefit public defenders, there are no indications of those mechanisms being in place across the board. In short, the public defense comes off quite well in the comparison, and its strength seems largely the result of its stable and relatively well-structured bureaucracy.

These outcomes are far from inviting the conclusion that the public defense cannot be improved, since some of its shortfalls are likely serious. The intense use of services and a reportedly lacking distribution of work count among them, and deficiencies in some local jurisdictions may be so radical that cases are just not observed at the Court. All in all, though, the results show that, at the very least, criminal defendants assisted by a defender are not carrying an extra burden at the appeal level. Are these findings generalizable? Within Argentina, it could be that defenders lose their advantage at the more labor-intensive trial or investigative stages, although some interviewees seemed to think otherwise. More broadly, it may be that some institutional traits are more important than others. A standardized mapping of traits can be a good starting point for a wider comparison of counsel competence both at the subnational level and across countries.

The findings enrich the literature on the impact of institutions, in the direction of such studies as Anderson and Heaton (2012). But, are they good news for defendants? Defendants (and particularly indigent ones) face countless disadvantages, both during the criminal procedure and after sentencing. In Latin America, these include profiling, police violence, the abuses of pretrial detention, and the substandard condition of the prison system. Good counsel is necessary though not nearly sufficient to alleviate those conditions. Perhaps as important, this study shows defendants virtually always losing. My data only allow to conclude that defenders are (at least) as prepared to face off prosecutors as are retained lawyers, at least regarding appeals' technicalities. Having said this, the conclusion seems entirely remarkable given the low priority societies generally assign the public defense.

In short, not only is defenders' intervention not taxing upon criminal defendants at the appeal level, but it is likely advantageous. This lends credence to a piece of advice I once heard: “If you really have a lot to spend, go find yourself the best attorney—otherwise you'll be better off with a defender.”