P.S. Ruckman, Jr.
PSRuckman@aol.com

Abstract :

This study provides an empirical examination of the federal executive clemency policy in the United States. I briefly review current literature on clemency, provide a framework within which to develop and test hypotheses, and specify variables for testing in a multivariate model of clemency policy from 1934 to 1994. The dependent variable (the percentage of total clemency actions which are 'positive') is derived from a data set consisting of over 41,000 clemency applications and over 11,000 'positive' clemency decisions made in eleven administrations. ARIMA models of clemency policy suggest that while the personal background characteristics of presidents and administrative factors are important considerations in understanding overall trends in clemency policy, the actions of an administration are best understood by a focus on the political environment.

AN EARLIER VERSION OF THIS PAPER WAS PRESENTED AT THE ANNUAL MEETING OF THE SOUTHERN POLITICAL SCIENCE ASSOCIATION. ATLANTA, GA. 1996.

INTRODUCTION

The federal exercise of executive clemency has been directly related to (if not the critical feature of) some of the most salient political events in our nation's history. The clemency power has, furthermore, played a major role in the development of criminal law's recognition of an insanity defense, self-defense, compulsion, and more lenient treatment of juvenile offenders. Clemency decisions have also been the focus of numerous and, in some instances, classic Supreme Court decisions. Despite the fact that our nation's last seventeen presidents have averaged over two hundred acts of clemency per year (Ruckman, forthcoming), clemency policy has been generally ignored by political science scholarship. One might continue to agree with the conclusion of a 1939 survey by the U.S. Attorney General, "there has never been an adequate treatment of the subject of pardon. [It] has been a neglected orphan, allowed to grow without benefit of careful grooming which has been accorded other branches of law" (Survey, vii,ix).

A majority of the extant literature on executive clemency appears in the law review or editorial format. Examinations of clemency powers typically trace their interesting origins and early development, their transformation in common law, their brief consideration at the constitutional convention and subsequent developments in classic Supreme Court decisions (Adler 1989; Boudin 1976; Cowlishaw 1975; Duker 1977; Jorgensen 1993). Other scholars focus on the exercise of clemency in particular areas of law, such as the death penalty (Abramowitz and Paget 1964; Bedau 1990; Cobb 1989; Kobil 1993; Ledewitz and Staples 1993; Radelet and Zsembik 1993) or address more political concerns in highly publicized, controversial cases (Adler 1989; Boudin 1976; Clark 1984; Jorgensen 1993; Krajick 1979; Orman and Rudoni 1979; Rozell 1994; Shichor and Ranish 1980; Smith and Johnson 1989). Current examinations rarely involve the analysis, or even presentation of data.

In this study, I briefly review current literature as it relates to the development and exercise of clemency in the United States. I then extract from the literature several specific hypotheses with respect to the outputs of clemency policy. In a most significant departure from the extant literature, I then provide a multivariate test of each hypothesis in a data set of the 41,879 clemency applications received by the last eleven administrations from 1934- 1994. In the review of the literature and the process of hypotheses building, I have provided the first clear bridge between this topic and social science analysis. The empirical analysis is the first rigorous statistical examination of federal executive clemency policy.

Executive Clemency in the United States

The Article II, Section 2 power to "grant reprieves and pardons for offenses against the United States, except in cases of impeachment" is remarkably broad considering the experience of the Framers. At common law, the King possessed broad powers to pardon offenses, with or without condition, either before or after indictment, conviction and sentencing (Jorgensen 1993). The historical record shows Kings utilized the clemency power as a "tool of pecuniary and political aggrandizement" and a path to personal gain (Adler 1989, 213). The sale of pardons was a common abuse (Adler 1977; Cowlishaw 1975; Survey 1939) and Kings used them to refurbish diminishing armies (Duker 1977). Clemency was also employed to coax felons into assistance with 'settling' the colonies (Cowlishaw 1975) and to celebrate landmark years in an individual's reign (Duker 1977).

As clemency abuses resulted in various parliamentary upheavals, the standard practice in the royal colonies was to permit the governor to pardon in all cases except treason and wilful murder (Adler 1989; Duker 1977). In the remaining colonies, the chief executive exercised clemency powers with occasional assistance from other authorities (Abramowitz and Paget 1964). After the revolution, most state governments went even further by rejecting the legacy of complete executive control of clemency power. Eight of thirteen states vested the authority to remit punishment in an executive legislative council and governor jointly, or in the legislature alone (Kobil 1991). There were no federal clemency powers. Thus, it is not particularly surprising that neither the Virginia plan nor the New Jersey plan contained a provision for federal clemency. Nor did any such provision find its way into resolutions submitted by the Convention to the Committee on Detail. John Rutledge, serving as a member of the Committee, scrawled "the power of pardoning" (except in the case of "an impeachment") into the margin of the Virginia plan (Ruckman 1994).

The Supreme Court has been the ultimate interpreter of the clause which eventually granted the president the clemency power and commentators agree the scope of the power is quite broad. John Marshall set the tone for the Court's initial interpretation of the pardoning power by describing it as "exercised from time immemorial by the executive of that nation whose language is our language" and suggesting a look "into their books for the rules prescribing the manner in which it is to be used." The Court has since granted presidents the right to remit fines and forfeitures, pardon criminal contempt of court, award conditional pardons, and commute sentences (even against the wishes of the individual involved). The Court has also described the pardoning power as "unlimited" (with the exception of treason) and "not subject to legislative control."

It is actually quite difficult to locate a generation of Americans without prominent and/or controversial clemency decisions. Federalist 74 anticipated a "well timed" offer of pardon in seasons of "insurrection or rebellion" might "restore the tranquility of the commonwealth" and George Washington hoped for as much when, in 1795, he pardoned those convicted of treason in the Whiskey Rebellion. Four years later, John Adams pardoned individuals sentenced to be hanged for their participation in Fries Rebellion. Despite unanimous disapproval by the members of his Cabinet, "most Americans approved the pardon and applauded the President for granting it" (Smith 1962, 1034).

Clemency powers were also quickly utilized for advantages more personal in nature. Adams brazenly attempted to affect the outcome of the 1800 election by strategic use of pardons and Jefferson made good on campaign promises by pardoning those still imprisoned under the Sedition Act of 1789 (Ruckman 1994). Jefferson also employed the pardoning power in his elaborate attempt to manipulate evidence in the conspiracy trial of Aaron Burr. A sensational open-court rejection of a blank pardon may have been the critical factor which turned the trial in Burr's favor (Abernathy 1954, 233-4; Schachner 1964, 854).

Madison made extensive use of clemency in obtaining military assistance from Laffite's Barataria Pirates (Hickey 1989; Mahon 1972) and numerous deserters and polygamists received pardons in the early republic (Adler 1989; Ruckman 1994), but the clemency policy of Abraham Lincoln and Andrew Johnson provoked the first constitutional crisis (Adler 1989, 219; Boudin 1976, 14; Duker 1977, 509; Humbert 1941, 36-7; Sebba 1983, 59). Lincoln employed clemency to "encourage desertions from the enemy without fear of punishment" and so enraged the people of the north that many considered his assassination "a godsend to the country" (Dorris 1977, 94). The initial threats of Johnson's impeachment were sparked by his own clemency program (Dorris 1977, 305). Johnson's pardoning of the unsolicitous, unrepenting Jefferson Davis was more than many could tolerate.

No twentieth century clemency decision has prompted a crisis equal to that of the Reconstruction era, but some actions have received an unusual amount of attention. Warren Harding's 1921 commutation of the sentence of socialist Eugene Debs, for example, appeared to fly in the face of Justice Holmes, the Supreme Court and Woodrow Wilson. Richard Nixon's conditional commutation of the sentence of James R. Hoffa raised serious constitutional questions (Boudin 1976), but Gerald Ford's pardon of Nixon may be the most single most controversial clemency decision in our nation's history. The pardon followed Ford's explicit recognition that the American people "would not stand for a pardon" and statements, before Congressional confirmation committees, that he intended to avoid such action (Orman and Rudoni 1977). Ford's decision caused his standing in public opinion polls to drop considerably (Orman and Rudoni 1979).

Clemency issues were an important factor in the 1976 presidential campaign for another reason. In the first debate, the candidates were asked how they proposed to deal with Vietnam draft evaders and deserters. Ford described his clemency program as a "good opportunity" that need not go "any further." His Clemency Board had received 22,000 applications, estimated 113 to 300 persons "eligible," and 91,000 potential applicants remained (Browning 1977, 39). Carter suggested a blanket pardon in the primaries and throughout much of his campaign and, in his first executive order as President, he made good on campaign promises by granting a blanket pardon to all draft evaders who had not been involved in a violent act.

Perhaps the most recent example of a clemency decision resulting in considerable controversy is George Bush's decision to pardon former Secretary of Defense Casper Weinberger and five other former Reagan administration officials on Christmas Eve 1992 (see Jorgensen 1993, at note 136). Six years and thirty-five million dollars after the commencement of the independent counsel (Jost 1992), almost half of the respondents in a Gallup poll were willing to state their belief that Bush issued the pardons "in order to protect himself from legal difficulties or embarrassment resulting from his own role in Iran- Contra" (Hugick 1992).

In sum, the Article II provision of federal executive clemency powers is, from an historical viewpoint, surprisingly broad. The Supreme Court's interpretation of the provision has, furthermore, generally been one of wide deference to the decision making of the Chief Executive. Though relatively few clemency decisions have prompted national debate, the use of clemency has been the focus of both campaign rhetoric and voter concern. It has healed the wounds of a young nation in rebellion and exacerbated the divisions of a nation at civil war. Presidents have employed the pardoning power to make political statements and there is considerable evidence that they have also used it to solve conflicts of a more personal nature.

The Politics of Clemency Policy

As the discussion above suggests, there are many factors which might influence both specific clemency decisions and the overall policy of an administration. W.H. Humbert's 1941 study, The Pardoning Power of the President, remains the sole systematic, empirical investigation of factors affecting executive clemency policy. Humbert examines clemency warrants issued by presidents from 1860 to 1936, but follows two distinct paths of discussion and analysis. The first focuses on the reasoning behind clemency decisions (utilizing justifications which are specifically stated in clemency warrants). A second path of analysis is more anecdotal and focuses on political events and processes which affect trends in clemency policy. The latter discussion relies partially upon frequency analysis and aggregate data. Ruckman (forthcoming) notes Humbert's study and other analyses direct us to a distinct set of private-personal, political and administrative factors which influence clemency policy. Humbert, in fact, assumes the difficulty of determining the "actual" reasons why positive clemency decisions are made since "officials favoring the granting of clemency probably give as their reasons only those which will evoke least opposition to the use of the pardoning power" (1941, 124). Most commentators would agree "there are motivating forces which lead pardon officials to grant pardons or commutations [which] are not given as reasons for acting" (Newman 1968; see also Scott 1952).

Ruckman thus notes the background, experiences, and attitudes of the president have clearly been reflected in clemency policy (forthcoming). Presidents Lincoln and Johnson, for example, are commonly supposed to have been more favorable toward the requests of petitioners from their native states, Kentucky and North Carolina. While there is considerable evidence indicating such favoritism (Dorris 1977; Ruckman and Kincaid 1994), similar patterns have not been systematically explored in other administrations.

Some presidents take office with a well-advertised 'hard-line' stance on crime, the sincerity of which they may feel obligated to demonstrate in areas like clemency policy (Ruckman, forthcoming). Humbert agrees "the most effective influence in checking the growth in acts of clemency" has been the "determination of officials, who administered the pardoning power [to] avoid any flagrant misuse" and the desire for "more rigid enforcement of the laws" by individual administrations (1941, 120-1). Writing over forty years later, Clark (1984) attributes part of the decrease in pardons during the Reagan administration (the least generous administration in a half-century) to its "hard line on crime" (2878). Thus, when explaining trends in clemency policy, personal characteristics of the president, ideological positions and views of the criminal justice system should be considered.

Ruckman notes the clemency policy of an administration is also affected by a broad set of "political" influences (e.g., public opinion, foreign policy concerns, war). Clark (1984) agrees the pardon process "is never totally immune from political winds" (2879). "Political" influences can, of course, be quite complex and interrelated. Public opinion, as noted above, can both encourage and discourage presidential action, as well as offer after- the-fact condemnation and/or praise. Warren Harding supported amnesty for political prisoners in his campaign for the presidency, but there is little doubt three hundred thousand signatures and seven hundred organizational endorsements encouraged his simultaneous release of Debs and twenty-three others (Adler 1989). Lyndon Johnson, on the other hand, received so much criticism for his clemency actions, he reduced pardons and commutations significantly in 1968 (Clark 1984; Krajick 1979).

Narrow and specific foreign policy considerations, as well as the out-break of war, can have a tremendous effect on an administration's clemency policy (Ruckman, forthcoming). George Washington signed the first pardon issued to a non-US citizen, and gave no specific reason for the action other than "sentiments of respect" for the Republic of Geneva. Jimmy Carter, likewise, viewed the opinion of the "international community" a factor even more critical than lack of remorse in an individual who attempted to assassinate President Truman, and in three others who sprayed bullets from the gallery of the U. S. House of Representatives. Carter, in explaining the commutations, cited pressure from Puerto Ricans in and outside the United States and pressure from various Third World countries (Clark 1984). Humbert, interestingly, suggests wars "have perhaps occasioned the most distinctive increases and decreases in the use of the pardoning power" (1941, 122).

Ruckman categorizes a third broad set of influences on clemency policy as "administrative factors" (forthcoming). The Office of the Pardon Attorney is a central actor in the clemency process, but its decision making is affected by structural developments within the legal system and idiosyncratic characteristics of each administration. Throughout out history, major reforms in the federal system have provided alternative release mechanisms which have transformed the role of the Office. In 1910, for example, the Federal Parole Act was enacted into law. The first Federal Probation Act followed fifteen years later in 1925. Scholars routinely note acts of clemency were more necessary before the federal system's institutionalization of mechanisms which provide alternative forms of release (see, for example, Goldfarb and Singer 1973, 319; Moore 1989, 61; Stoke 1927; Survey 1939; 296-300).

More narrowly, the work of the Pardon Attorney is affected by factors idiosyncratic to particular administrations including: changes in personnel, the specific policies of an administration, and perceptions of the administration held by potential petitioners. The general tendency for clemency activity to increase throughout an administration is undoubtedly due (in part) to an "adjustment period" for new members of the incoming Justice Department (Humbert 1941, 123; Ruckman, forthcoming). Clark (1984) notes, however, that the Reagan Justice Department "tightened" the rules in May of 1983, "by increasing the time that pardon applicants must wait to become eligible after completing a prison term, or after a conviction in cases where the sentence did not involve prison." Applicants were required to wait five years for lesser crimes and seven years for serious crimes ("the longest period ever") and the department "expanded list of crimes considered serious enough to require the longer waiting period" (2878). Reagan granted fewer pardons and commutations than any president in twentieth century president before him.

The Office of the Pardon Attorney has also noted prison populations react to the varying practices of particular administrations (Goldfarb and Singer 1973, 337). George Bush issued even fewer pardons and commutations than Reagan, and the number of yearly requests continued to generally decline throughout his administration. The election of Bill Clinton appears to have raised expectations however. Yearly requests increased significantly in 1993. The 808 requests received by the Clinton administration in fiscal year 1994 were the highest number received by the Office since 1967.

Despite the literature's suggestion that federal executive clemency policy is considerably influenced by a host of political and administrative factors, and factors idiosyncratic to particular presidents and administrations, there has been no formal testing of multivariate models of clemency policy. Fortunately, data on clemency policy are available and, as the review above clearly suggests, the literature provides numerous hypotheses which are amenable to empirical testing.

Overview of the Data - Dependent Variable

In order to investigate correlates of clemency policy, I have obtained data from the Office of the Pardon Attorney, U.S. Department of Justice. The data consist of 41,879 clemency requests in the administrations of eleven presidents, from 1934 to 1994. The data include the number of clemency requests per fiscal year as well as the number of pardons, commutations and remissions granted, the number of requests denied, and the number of requests pending. The data suggest presidents exercise clemency with much greater frequency than might be commonly assumed. Few might guess, for example, that the pardon of Richard Nixon was but one of 409 clemency actions by the Ford administration (including 382 pardons and 27 commutations). Similarly, it is probable that few are aware the Reagan administration issued 406 clemency warrants (including 393 pardons and 13 commutations). The clemency policies of the eleven presidents in the data have, in fact, resulted in over eleven thousand acts of clemency (including 9,419 pardons, 1,122 commutations and 493 remissions of fines). Presidents have averaged over one hundred and eighty acts of clemency per year, in the sixty-one year period.

The data are particularly interesting from the standpoint of the percentage of total requests that result in 'positive' action (a grant of some form of clemency), 'negative' action (denial or closing of the application), or no action whatsoever. On average, the eleven administrations granted some form of clemency in 12 percent of the total number of applications per fiscal year unit. While the average 'negative' action rate per fiscal year unit is 34 percent, Franklin D. Roosevelt typically rejected almost 50 percent of the requests received during his administration. More recent presidents have exhibited a distinct tendency to simply take no action on clemency requests. In a typical fiscal year unit, John F. Kennedy, Lyndon Johnson, Jimmy Carter, Ronald Reagan, George Bush and Bill Clinton took no action on at least 55 percent of all clemency requests.

Multiple and varied explanations for lack of action on clemency requests and a general desire to judge administrations on the basis of action (contra inaction) encourage the selection of a more informative viewpoint utilized in previous discussions. As did Clark (1984) and Ruckman (1995; forthcoming), I have re-examined the data in terms of the population of requests for which some action was taken. Actions were thus categorized as being either 'positive' (those granting some form of clemency) or 'negative' (those involving explicit denial or 'closing' of an application) and the percentage of the total actions per

Figure One - About Here

fiscal year unit that were 'positive' was calculated. Figure 1 presents these data from 1934- 1994. Breaks occur in the time-series when multiple presidents served in the same fiscal year unit. A break does not appear in fiscal year 1969, however, as the last six months of the administration of Lyndon Johnson (1969) and the first six months of Richard Nixon's administration (1969) resulted in no grants of clemency. It should also be noted that President Clinton granted no requests for clemency in fiscal years 1993 and 1994.

The data reveal the eleven presidents, on average, granted clemency in 25 percent of actions taken per fiscal year unit. The highest 'positive' clemency rate (59 percent) appears in the last six months of Eisenhower administration (1961). The highest 'positive' clemency rates for complete fiscal years appear in the administration of Harry S. Truman (55 percent in 1950, 48 percent in 1952, and 47 percent in 1951). The lowest 'positive' clemency rate for a fiscal year unit is '0' and the mark is shared by four presidents. As noted, Johnson and Nixon granted no clemency in fiscal year 1969. The final months of Nixon's presidency (fiscal year 1975) were also marked by a complete halt in clemency activity. George Bush and Bill Clinton, however, are the only presidents in the data to issue no clemency warrants in a complete fiscal year unit. Bush granted no requests in either fiscal year 1990 or 1992. President Clinton granted no requests in fiscal year 1994.

The administrations of Harry Truman (1946-1953), John F. Kennedy (1961-1964) and Gerald Ford (1975-1977) are notable for having an average 'positive' clemency rate of at least 35 percent. The administrations of Ronald Reagan (1981-1989), George Bush (1989- 1993) and Bill Clinton (1993-1994) are notable as the only three with an average 'positive' decision rate of less than 15 percent. It quite evident from Figure 1 that there are short- term trends in 'positive' clemency rates, but there is also an apparent decline in 'positive' clemency rates over time.

Independent Variables

I have employed eight substantive variables (and one control variable) in an attempt to understand the short and long term trends apparent in Figure 1. While each of the eight variables was chosen on the basis of considerations addressed in the literature, the model represents the first multivariate test of individual hypotheses. The categories of independent variables are roughly based upon Ruckman (forthcoming): Characteristics of the President, Administrative Factors, and "Political" factors.

Characteristics of the President

The model will first address the relationship between the partisan identification of the president and clemency policy. The recent rhetorical emphasis of the Republican party on 'law and order' and getting 'tough' on crime, in combination with the Democratic party's emphasis on due process and civil rights, encourage the hypothesis that Republican presidents should be less generous in clemency decision making and, therefore, should be associated with a reduction in 'positive' clemency rates. In his frequency analysis of clemency from 1900-1993, Ruckman (forthcoming) notes an average difference of ten percentage points between the 'positive' clemency rates of Republican and Democratic administrations per fiscal year unit.

Marked recruitment patterns leave little to distinguish between the social backgrounds of individuals who have served as president. Each has been white and male. Each has typically been born in a relatively prosperous family, and can trace his ancestry to the British Isles. Presidents have also generally been members of high-status Protestant denominations and have received their education at prestigious private institutions (Watson and Thomas 1988, 133). In terms of qualifications, or previous experience, most presidents served a long apprenticeship in public office before reaching the presidency. Approximately two-thirds have served in either the House or the Senate, or both (Watson and Downing, 119). In most instances where social backgrounds and/or experiences do vary, there is little theoretical cause to expect an accompanying variation in clemency policy. There are, however, two types of background experiences which seem particularly relevant to clemency decision making: gubernatorial experience and experience as an attorney.

Governors play a central and, in most instances, critical role in state clemency decisions. Unlike the president, a governor is not "remote" from the scene of crimes and more often must make clemency decisions "amid intense feelings" (Humbert 1941, 5; see also Abramowitz and Paget 1964, 172-5; Kobil 1991, 606-11; Scott 1952, 98). Clemency decisions are, moreover, a "significant factor" in the political lives of governors (Kobil 1991, 607) and any decision(s) in the 'wrong' direction, can mean "political suicide" (Bedau 1990, 268; see also Kobil 1991, 607-8; Krajick 1979). My model contains a variable controlling for the effects of gubernatorial experience and hypothesizes individuals with such experience will approach clemency decisions with a reticence reflected by a decrease in the 'positive' clemency rate. On the other hand, I will test the hypothesis that experience as an attorney is associated with an increase in a president's clemency activity. Individuals with legal experience are less likely to be baffled, or intimidated, by the technical nature of the clemency process, or the host of legal considerations present in individual cases.

"Political" Factors

The current literature lends itself only to a few hypotheses related to "political" factors which can be tested at the aggregate level. The first variable in this category will control for those years in which the nation is at war. The following years are coded '1' to denote the presence of this factor: 1939 to 1945; 1950 to 1953; 1965 to 1973. Consistent with Humbert's analysis, I hypothesize war will contribute toward an increase in 'positive' clemency rates (1941, 122).

A second variable will control for fiscal years in which a presidential election occurs. As the literature gives little information that would guide one as to the expected direction of an hypothesized relationship, I will simply test to see whether clemency practices are sensitive to the occurrence of presidential elections. A third variable, however, will distinguish between elections before and after the "law and order" campaigns of Richard Nixon. Issues related to crime have been a prominent feature of elections since 1968, and the model will hypothesize 'positive' clemency rates decreased as a result of this significant change in the landscape of campaign issues.

A fourth and final variable will simply control for the crisis in the presidency which eventually resulted in the resignation of Richard Nixon. Although 207 pardon and 49 commutation petitions were "pending" at the beginning of fiscal year 1975, the Justice Department did not receive new petitions or issue grants of clemency in the short period of time Nixon remained in office. Complete inaction by the Justice Department on both fronts appears in no other fiscal year unit in the data. A variable will thus control for the observed negative effects of the "crisis."

Administrative Considerations

Three variables will control for factors which are 'administrative' in character, two represent major policy interventions. Humbert notes "the periodic shifting of administrative personnel also contributes to the increases and decreases in the use of the pardoning power" (1941, 123). Indeed, there is little reason to believe administrations will enter office with clemency policy as a top priority and, again, it is reasonable to assume incoming members of the Department of Justice will require a period of adjustment. Noting 'positive' clemency action presupposes familiarity with both the process and individual cases, Ruckman finds 'positive' clemency decision making typically increases throughout a president's term (forthcoming). I will thus hypothesize an increase in tenure of office will lead to an increase in 'positive' clemency decisions.

Two variables will control for major policy changes which occurred in the time period covered by the data. The first variable will test the intervention effect of a policy change which took place in the Office of the Pardon Attorney during the Kennedy administration. Prior to 1962, the Office sent only those petitions to the president which, in the opinion of the Pardon Attorney, deserved positive action and those petitions involving the death penalty. Beginning in 1963, however, the Office began also sending petitions to the president for which it recommended denial (28 C.F.R.  1.3). A second variable will test the effect of the Reagan administration's 1983 increase in the waiting period for pardon applicants (see discussion above). A step function will be utilized to test the hypothesis that each of these major change in policy resulted in a decrease in 'positive' clemency rates.

The complete model thus employed is as follows:

       Yt    = % of 'positive' clemency actions per fiscal year unit, 1900-1993
       þ1    = party identification of president (1=Democrat, 0=Republican)
       þ2    = president with previous experience as lawyer (1=yes, 0=no)
       þ3    = president with previous experience as governor (1=yes, 0=no)
       þ4    = election year (1=yes, 0=no)
       þ5    = war year (1=yes, 0=no)
       þ1    = intervention, 1968 campaign (0=before 1968, 1=1968 and after)
       þ6    = control for Nixon crises (1=1975, 0=other)
       þ2    = intervention, 1962 policy (0=before 1962, 1=1962 and after)
       þ3    = intervention, 1983 policy (0=before 1984, 1=1984 and after)
       þ7    = counter, number of fiscal year units covered by administration

Analysis

In order to test each of the hypotheses in a multivariate context, I have employed Box-Jenkins, or ARIMA (AutoRegressive Integrated Moving Average) techniques. ARIMA models are particularly appropriate for this type of analysis since they control for components or "causes" of the variation in a time series that depend on time (including drifts, trend, and seasonal change). The removal of these stochastic processes allows one to appropriately assess the hypothesized impact of the independent variables on the series. The Box-Jenkins model is typically denoted as :

                                 Yt  = f (It) = Nt

where the dependent time series (Yt) is the result of the deterministic component or intervention(s) (It) and the stochastic noise component (Nt). The initial "identification" of a time series model includes specification of the order of differencing necessary to produce stationarity, estimation of the order of autoregressive effects, and estimation of the order of moving average effects that influence the series.

A casual glance at Figure 1 indicates the series is nonstationary. Statistical confirmation of its nonstationarity was obtained in an analysis which found that the autocorrelation function failed to 'die out' rapidly. A 'dampening' effect in the autocorrelations and 'truncation' of the partial autocorrelations was also noted and the probabilities associated with Box-Ljung Q (or modified Box-Pierce) statistics indicated a significant level of autocorrelation in the series. The series was then differenced. After differencing, the autocorrelation functions and partial autocorrelation functions were, again, characterized by a truncation effect, suggesting the appropriateness of an Integrated Moving Average, or (0,1,1) model. Finally, I noted the residuals of the tentative model were independent at the first and second lag and Box-Ljung Q statistics were non-significant. The results of the (0,1,1,) transformation thus indicated the residual distribution might then be properly characterized as "white noise."

The model in Table 1 clearly suggests clemency policy is affected by personal, "political" and administrative factors. The coefficients for previous experience as a lawyer (þ2) and governor (þ3) are in the expected direction and are statistically significant. Experience as a lawyer is associated with a 15 percent increase in the 'positive' clemency rate (significant at .001) while experience as a governor is associated with a 11 percent decrease in the 'positive' clemency rate (significant at .01).

Table 1 - about here

Interestingly, the impact of variables representing characteristics of the president are somewhat dwarfed by the impact of "political" and administrative variables. The coefficient for partisan identification of the president (þ1) actually fails to attain statistical significance. The coefficient for the step function representing fiscal years after 1967 (þ1) may shed substantive light on why there is no significant difference between the 'positive' clemency rates of Democratic and Republican administrations. The "law and order" campaign of 1968 and subsequent salience of issues related to crime in presidential campaigns have had a dramatic effect on clemency policy. The period from 1968 to 1994 is associated with a 34 percent drop in 'positive' rates (significant at .001). The rhetoric of presidential campaigns and concerns of the electorate have, evidently, placed parameters on the clemency policy of administrations, regardless of the party identification of the president.

Results for the two remaining substantive "political" variables are mixed. The pulse function þ4 indicates fiscal years in which an election takes place are associated with a 3 percent increase in 'positive' clemency rates. The relationship fails to attain statistical significance however. The pulse function þ5 indicates (unexpectedly) fiscal years in which the nation is at war are associated with an 11 percent decrease in 'positive' clemency rates. It is possible the unexpected negative effect of war may reflect the president's preoccupation with other matters, or suggest an important separation between clemency policy which is domestic in focus and impact and clemency decisions which are related to more broad, international policy goals. The historical use of amnesty might, furthermore, lead one to expect clemency activity to increase in the aftermath of war.

Both of the step functions representing major policy changes (þ2 and þ3) are significant at the .05 level. The 1962 policy (again unexpectedly) is associated with an 18 percent increase in 'positive' clemency rates while the 1983 policy coefficient (-16.9399) is in the expected direction.

The estimate for 'counter' variable representing the fiscal year unit of the administration (þ7) is in the expected direction and its value indicates increases in 'positive' clemency ratios are, in fact, associated with an increase of tenure in office. The average 'positive' clemency rate for administrations in the first fiscal year unit is 17 percent. In second and third fiscal year units, administrations have average 'positive' rates of 20 and 27 percent. The average for the last fiscal year unit of each administration (excluding the Clinton administration) is 32 percent. When controlling for the effects of other variables in the model, an increase of one fiscal year unit is associated with a 3 percentage point increase in the 'positive' clemency rate (statistically significant at .001).

Further estimations of the model included separate tests of dummy variables representing each of the eleven presidents and the tenure of each of ten pardon attorneys. None of these tests resulted in findings which would led one to believe the model above has been seriously mispecified. Indeed, the failure of variables representing individual presidents and pardon attorneys bolsters the conclusion that while clemency policy may be affected by the personal background characteristics of presidents, and even more so by a host of administrative factors, it is perhaps best understood in the context of the political environment.

Discussion and Conclusion

While there is a sizeable literature which addresses the topic of clemency, there is precious little systematic analysis of the clemency policies of our nation's presidents. The rare discussions of clemency policy typically rely upon broad, untested generalizations about the exercise of clemency, or note moderate relationships while utilizing uncontrolled frequency analysis. This study, however, provides a context for empirically tests of hypotheses regarding clemency policy. Each of the hypotheses in the model above is, furthermore, derived from the extant literature. In sum, I have provided the first multivariate examination of federal executive clemency policy.

The findings in this study reaffirm the current literature's suggestion that clemency is affected by a wide range of factors, but also provides powerful evidence that the literature's current emphasis on the partisan identification of the president should be reconsidered. Future multivariate analyses might instead benefit from further exploration of the impact of more specific characteristics and/or life experiences of the individuals who serve as President. "Politics" have clearly affected the clemency policies of our nation's presidents, but the models (and variables) provided here certainly do not exhaust the avenues of research. The experience of the Lyndon Johnson (see discussion above) suggests, for example, the need to consider the possible relationship between presidential approval ratings and clemency policy. Likewise, the significant effect of specific policy changes in the models here suggests the need to further explore the impact of such changes on the outputs of the Office of the Pardon Attorney. In sum, the models provide a solid base upon which future models of clemency policy can be built.

While I believe the findings above are a significant contribution to the clemency literature (and, more broadly, the literature which addresses the office of the presidency), I do not claim the multivariate model tested here provides significant insight into the decision making process in specific acts of clemency. Individual level data on clemency petitions (and requests) are clearly needed in order to provide a rich analysis which could result in such insights. I do claim, however, to have provided significant evidence of important, broad influences on the overall trends in clemency policy outputs. For this reason, I believe this study will encourage further empirical research on the generally ignored clemency power and perhaps lead toward a more general theory of clemency policy.

Clemency is, of course, a critical part of the system of checks and balances in our governmental framework and clemency decisions have been directly related to (if not the central feature of) some of the most controversial political events in our nation's history. The existence of the clemency power in national legislature and in the fifty states makes it even more remarkable that political scientists have provided little systematic research in this area. My review of the literature for this study uncovered a mere four articles concerning clemency in social science journals. Three of the four articles dealt with the Ford-Nixon pardon while the fourth dealt with the rare proclamations of formal amnesties. The inattention of social scientists is all the more remarkable given the wealth of research on the various failures, abuses and malfunctions of the criminal justice system - problems which the clemency power can potentially counteract or minimize.

References

Abernathy, Thomas Perkins. 1954. The Burr Conspiracy. New York: Oxford University Press.
Abramowitz, Elkan and David Paget. 1964. "Executive Clemency in Capital Cases," New York University Law Review, 39: 136-192.
Adler, David Gray. 1989. "The President's Pardon Power," in Inventing the American Presidency, Thomas E. Cronin (editor), University Press of Kansas.
Barber, James David. 1992. The Presidential Character. Predicting Performance in the White House, Englewood Cliffs : New Jersey.
Barton, William E. 1925. The Life of Abraham Lincoln. (Volume Two). Bobbs Merrill Company : Indianapolis.
Bedau, Hugo Adam. 1990. "The Decline of Executive Clemency in Capital Cases," Review of Law and Social Change, 18: 255-272.
Berry, William D. and Michael S. Lewis Beck. 1985. New Tools for Social Sciences. SAGE Publications : Beverly Hills.
Boudin, Leonard B. 1976. "The Presidential Pardon of James R. Hoffa and Richard M. Nixon: Have the Limitations on the Pardon Power been Exceeded?" University of Colorado Law Review, 48: 1-39.
Browning, Peter. 1977. "Amnesty for All: To Begin a New Life," The Nation, 224: 39-42.
Clark, Charles S. 1984. "Reagan Parsimonious Use of Pardon Power," Congressional Quarterly Digest, November 3, 1984.
Cobb, Paul Whitlock, Jr. 1989. "Reviving Mercy in the Structure of Capital Punishment," Yale Law Journal, 99: 389-409.
Cowlishaw, Patrick R. 1975. "The Conditional Presidential Pardon," Stanford Law Review, 28: 149-177.
Cross, Mercer. 1977. "Carter's Inaugural Hope: 'Unity, Trust,'" Congressional Quarterly Weekly Report, 35: 103-105.
Dorris, Jonathan Truman. 1977. Pardon and Amnesty under Lincoln and Johnson, Chapel Hill : The University of North Carolina Press.
Duker, William F. 1977. "The Presidential Power to Pardon: A Constitutional History," William and Mary Law Review, 18: 475-538.
Garber, Ronald L. and Linda R. Singer. 1973. After Conviction. Simon and Schuster: New York.
Hargrove, Erwin C. 1966. Presidential Leadership. Personality and Political Style. MacMillan: New York.
Hickey, Donald R. 1989. The War of 1812, Chicago : University of Illinois Press.
Hugick, Larry. 1992. "Iran Contra Pardons Opposed," The Gallup Poll Monthly, December: 21-22.
Humbert, W. H. 1941. The Pardoning Power of the President. Washington, D.C. : American Council on Public Affairs.
Jorgensen, James N. 1993. "Clemency and Pardons Note," University of Richmond Law Review, 27: 345-370.
Jost, Kenneth. 1993. "Walsh, Bush Tarnished By Pardons," American Bar Association Journal, February.
Kelly, Alfred H., Winfred A. Harbison, and Herman Belz. 1991. The American Constitution: Its Origins and Development, New York : W.W. Norton and Company.
Kobil, Daniel T. 1991. "The Quality of Mercy Strained: Wresting the Pardoning Power from the King." Texas Law Review, 69: 569-641.
Kobil, Daniel. 1993. "Due Process in Death Penalty Commutations: Life, Liberty, and the Pursuit of Clemency" University of Richmond Law Review, 27: 201-226.
Krajick, Kevin. 1979. "The Quality of Mercy," Corrections Magazine, 5: 46-53.
Ledewitz, Bruce and Scott Staples. 1993. "The Role of Executive Clemency in Modern Death Penalty Cases," University of Richmond Law Review, 27: 227-239.
Mahon, John K. 1972. The War of 1812, Gainesville, Florida : University of Florida Press.
McLeary, Richard and Richard A. Hay, Jr. 1980. Applied Time Series Analysis. SAGE Publications : Beverly Hills.
Moore, Kathleen Dean. 1989. Pardons : Justice, Mercy, and the Public Interest. Oxford University Press : New York.
Moore, Kathleen Dean. 1993. "Pardon for Good and Sufficient Reasons," University of Richmond Law Review, 27: 281-288.
Newman, Charles L. 1968. Sourcebook on Probation, Parole and Pardons. Charles C. Thomas: Springfield, Il.
Orman, John M. and Dorothy Rudoni. 1979 "Exercise of the President's Discretionary Power in Criminal Justice Policy," Presidential Studies Quarterly, 9:415-427.
Ostrom, Charles W. 1990. Time Series Analysis. 2nd Edition. SAGE Publications: New York.
Radelet, Michael L. and Barbara A. Zsembik. 1993. "Executive Clemency in Post Furman Capital Cases," University of Richmond Law Review, 27: 289-314.
Rozell, Mark. 1994. "President Ford's Pardon of Richard M. Nixon: Constitutional and Political Considerations," Presidential Studies Quarterly, 24: 121-137.
Ruckman, P.S., Jr., "Executive Clemency in the United States: Origins, Development, and Analysis (1900-1993)," Presidential Studies Quarterly, forthcoming.
Ruckman, P.S., Jr. 1994. "Policy as an Indicator of 'Original Understanding:' Executive Clemency in the Early Republic (1789-1817)." Paper presented at the annual meeting of the Southern Political Science Association, Atlanta, GA.
Ruckman, P.S., Jr. 1995. "Presidential Personality and Executive Clemency: A Reexamination." Social Science Quarterly, 76: 213-221.
Ruckman, P.S., Jr. and David Kincaid. 1994. "The Forgotten Side of Lincoln's Clemency Policy." Paper presented at the annual meeting of the Illinois Political Science Association, Chicago, IL.
Sebba, Leslie. 1983. "Amnesty and Pardon," in Encyclopedia of American Justice. Sanford H. Kadish (editor in chief), New York: The Free Press.
Schachner, Nathan. 1964. Thomas Jefferson: A Biography. New York: Thomas Yoseloff.
Scott, Austin W., Jr. 1952. "The Pardoning Power," Annals of the American Academy of Political Science, 284: 95-100.
Shichor, David and Donald R. Ranish. 1980. "President Carter's Vietnam Amnesty: An Analysis of a Public Policy Decision," Presidential Studies Quarterly, 10: 443-50.
Smith, Christopher E. and Scott P. Johnson. 1989. "Presidential Pardons and Accountability in the Executive Branch," The Wayne Law Review, 35: 1113-1131.
Smith, Page. 1962. John Adams. Garden City, New York: Doubleday and Company, Inc.
Stoke, Harold W. 1927. "A Review of the Pardoning Power," Kentucky Law Journal 16: 34- 42.
U.S. Attorney General. 1939. Survey of Release Procedures, v. III, U.S. Government Printing Office: Washington D.C.

Return