Saints Marching In, 1590-2012* Robert J. Barro and Rachel M. McCleary Harvard University January 2016 Abstract The Catholic Church has been making saints for centuries in the two-stage process of beatification and canonization. We analyze determinants of numbers beatified and canonized (non-martyrs) since 1590 across seven world regions. The number beatified is roughly proportional to a pope’s tenure and a region’s Catholic population, responds positively since the early 20th century to Catholic-Protestant competition and to secularization, and falls after the virtual ending of warfare between European Catholics and Protestants with the Peace of Westphalia in 1648. There is bias in favor of Italy, then Other Western Europe and Eastern Europe, and against Africa, Asia, Latin America, and North America. The number canonized rises with the stock of beatifieds not yet canonized, rises with Catholic-Protestant competition, and is lower after the Peace of Westphalia. Regional bias is minor for canonization, given stocks of beatifieds. The last two popes before Francis, John Paul II and Benedict XVI, are large positive outliers in numbers beatified, and John Paul II is also an outlier for numbers canonized. *This research was supported by the Paul M. Warburg Fund at Harvard University. We benefited from comments from Marios Angeletos, Davide Cantoni, Gary Chamberlain, Lawrence Cripe, Edward Glaeser, Caroline Hoxby, Emi Nakamura, Emily Oster, Ricardo PérezTruglia, Thomas J. Reese, S.J., Rodney Stark, José Ursúa, and Gina Zurlo. Alex McQuoid helped with data on blessed persons through 2005. Todd Johnson assisted with data on religious adherence as well as providing comments. We appreciate early input from Bradley Ruffle on the design of the data set. Some years ago, we went to Antigua, the colonial capital of Guatemala, and visited the tomb of Santo Hermano Pedro, Central America’s first saint. Crowds of people milled in the courtyard of San Francisco Church, where his sanctuary is located, purchasing candles, effigies, flowers, and crucifixes. A line of visitors curved down a cement ramp to the tomb’s entrance. Once inside, pilgrims approached the stately wood and iron casket on their knees, prayed, placed flowers, and tied onto the metal grill wax effigies of body parts for which they were seeking a miracle cure. Before leaving, supplicants knocked on Hermano Pedro’s tomb to ensure that he had heard their prayers. Hermano Pedro was born in 1626 and left his native Canary Islands to avoid entering into an arranged marriage. He made his way to Guatemala, where he became a member of the Third Order of Saint Francis and worked among the ill, lame, prisoners, and orphaned children. In Antigua, he established a hospital, homeless shelter, and school. Later, he founded a new order, the Bethlehemite Congregation, dedicated to overseeing the institutions he had founded for the indigent. Pope John Paul II beatified Hermano Pedro in 1980 and named him a saint in 2002. During our visit to Hermano Pedro’s tomb, we were struck by the devotion engendered by Central America’s first saint. In a country dominated by the growth of evangelical and Pentecostal denominations as well as neo-Pentecostal megachurches, the level of activity at San Francisco Church surrounding Hermano Pedro was remarkable. Guatemala has seen rapid growth in Protestantism, particularly Pentecostalism, at the expense of Catholicism. In canonizing Hermano Pedro, was the Catholic Church attempting to raise the enthusiasm of its followers and thereby compete against the growing Protestant threat? This question led us to ask more generally whether the Catholic Church has been using saint making as a way to compete against Protestantism in other countries. To answer this question, we began by constructing a data set on Catholic saints from 1588 to 2012. To preview our results, we found that saint making has played a role in CatholicProtestant competition particularly since the early 1900s and especially in traditionally Catholic Latin America. In addition, we found that the Church has been using sainthood to deter secularism, which we measure by adherence to no religion. This mechanism has been especially important in Western Europe, where adherence to no religion has soared. Part of the historical record on saint making is the unprecedented pace of beatification and canonization under Pope Francis’s immediate predecessors, Benedict XVI (2005-2013) and John Paul II (1978-2005). Another feature of the data is the shift since the early 1900s away from an almost exclusive focus on blessed persons from Italy and Other Western Europe and toward Eastern Europe, Latin America, North America, Asia, and Africa.1 To assess these geographic patterns and the roles of Catholic-Protestant competition and secularization, we undertake a long-term econometric analysis that seeks to explain the Catholic Church’s choices of blessed persons over time and across regions. Our research falls into the growing fields of religion and political economy and the sociology of religion. For surveys, see Iannaccone (1998), Sherkat and Ellison (1999), and McCleary and Barro (2006). Our application to saint making stresses the Catholic Church’s incentives to compete on one side with Protestantism and on another side with secularism (that is, no religion). These competitive mechanisms fall comfortably into the realm of economic analysis. The closest related research that we know of is Stark (2003, 2004), Ferrero (2002), and Pfaff (2013). 1 In describing John Paul II, Cunningham (2005, p. 122) says, “...the pope has a predilection for canonizing saints from regions other than the traditional geographical locations of Europe or the Middle East where saints have typically been found. He wants to show—by canonizing people from the Far East, India, Oceania, and so on—that the possibility of sanctity may be found in all places where the Catholic Church has been planted.” 2 I. A Brief Survey of Saint Making Our study relies on a long-term data set constructed by Barro, McCleary, and McQuoid (2011). These data include numbers and characteristics of blessed persons selected by the Catholic Church.2 The data apply to beatifications (stage one for becoming a saint) and canonizations (final approval as a saint). The main information in our data set covers the period since 1588, when official Vatican records began. We also have partial information on canonizations back to 1234. We focus on blessed persons known as confessors—individuals who lived a life of virtue but were not put to death for their faith. That is, as discussed later, our analysis excludes martyrs.3 We now provide a brief history of the selection procedures for blessed persons. For confessors, beatification requires the posthumous performance of a miracle—one since the 1983 Code of Canon Law reforms, two or more under the 1917 Code of Canon Law. Beatification is normally required for the second stage, canonization, which features another, post-beatification miracle (two additional miracles before the 1983 reforms). The data set has information on when beatification and canonization occurred and, hence, under which pope. Also included are characteristics of blessed persons, including dates and places of birth and death, gender, urban versus rural origin, whether ever married, status in the Catholic Church, whether a convert, and indicators of education and occupation. We gauge the location of a person chosen as blessed by residence at death—which typically corresponds to the place in which the person’s main religious work was performed. 2 The main data are from the Catholic Church, Congregatione pro Causis Sanctorum (1999), Burns (1995 and later years), and McBrien (1995, 2000). Information for recent years comes from the Vatican website. See www.vatican.va/news_services/liturgy/saints/index_saints-blesseds_en.html. 3 An earlier list of blessed persons, covering 993 to 1967, was compiled by Delooz (1962). This list includes martyrs, as well as persons described as blessed but who lacked formal approval by the pope. 3 The first recognized papal canonization occurred in 993 when Pope John XV canonized Ulric of Augsburg a mere 20 years after Ulric’s death. In contrast, the mean time between death and canonization since 1590 was 181 years. The process of canonization gradually became formalized up to the 12th century. In 1234, Pope Gregory IX declared the exclusive authority of the Holy See to bestow the title of “saint”. However, this decree did not deter bishops from conferring beatifications, thereby creating a clear distinction between “beatified” and “saint”. 4 Pope Sixtus V (1585-1590) formalized the saint-making process and created the Congregation of Rites to concentrate decision-making within the Holy See. The Congregation was assigned authority over canonization, particularly with regard to verifying miracles and virtues; for the first time the process included medical examiners (Harvey [2007, p. 1256]). Sixtus V also purged the Calendar of Saints of persons with questionable credentials.5 A long-overdue development in 1917 was the systematic organizing of the “vast and confusing collection of canonical materials into a single authoritative reference known as a code ...” [Peters (2001, p. xxiii)]. Begun in 1904 under Pope Pius X (1903-1914), this reorganizational process was promulgated in 1917 by Benedict XV (1914-1922). According to Peterson (1940, p. 240), “The outstanding event [after the Council of Trent of 1545-1563] was ... the promulgation of the new Codex Juris Canonici ... by ... Pope Benedict XV. This event in ecclesiastico-legal science was as a great step forward as had been the compilation of the [canon law in the 12th century].”6 For our purposes, the Code is important because it lays out the rules for beatification and canonization. Canon 1999 of the Code stipulates that only the pope has the authority to 4 For a discussion of the evolution of canonization, see Kemp (1945). The Calendar was first purged during the Council of Trent (1545-63) when nearly half of saints’ feast days were eliminated (Klauser [1979, pp. 117-152]). 6 Peters (2001) provides an English translation of the Pio-Benedictine Code. 5 4 canonize, whereas the Congregation of Rites is charged with overseeing the process, and that local ecclesiastical authorities must follow canon law. However, Canons 2038 and 2039 stipulate that the process of naming blessed persons should typically be initiated by local Catholic authorities.7 Given the significance of this codification of the saint-making process, we look at whether there was a break in the process of naming blessed persons in the early 20th century. In 1969, Pope Paul VI created the Congregation for the Causes of Saints (replacing the Congregation of Rites) and the Congregation for the Divine Worship. Pope John Paul II decentralized the process in 1983, while strengthening the Vatican’s ability to review cases by creating the College of Relators.8 A relator is a high-ranking member of the Roman Curia who supervises the preparation of materials submitted to the Congregation for the Causes of Saints. In the current structure, local ecclesiastical authority in the diocese where the blessed person died oversees the collection of evidentiary materials in the first phase of the beatification process. After a five-year waiting period following the death of the candidate (much shorter than the previously required 50 years), a formal petition can be submitted to Rome to open the case for beatification. Upon Rome’s approval, the local diocese appoints a postulator, who gathers evidentiary materials—writings by the candidate, testimonials from eyewitnesses, and secondhand accounts. The body of the candidate is exhumed and examined to ensure that the person existed. When the collection of evidence is complete, the report is sent to Rome to a relator, who reviews the report and appoints a medical expert to conduct an independent inquiry into the claimed miracles. The relator oversees the report, which is submitted to the Congregation for the Causes of Saints, consisting of 25 cardinals and bishops. If the candidate is deemed by the 7 Canon 2101 reaffirms that 50 years must elapse after a person’s death before a petition can be made. John Paul II also eliminated the “Devil’s Advocate” or “Promoter of the Faith,” who had for centuries assumed the role of posing objections to proposed blessed persons. See McBrien (2001, p. 45). 8 5 Congregation to have lived a virtuous life according to Catholic theology, the candidate receives the title “venerable,” the current status of Pope Pius XII (1939-1958). To be beatified, the candidate must be verified to have performed a miracle. This verification earns the candidate the title “blessed” as a beatified, the status accorded Pope John Paul II (1978-2005) in 2011. A second post-beatification miracle must be performed and verified before the candidate can be canonized as a saint. This process was completed for John Paul II in September 2013, with his canonization taking place in 2014 (along with that of Pope John XXIII). Pope Benedict XVI maintained the 1983 reforms, while making minor changes. The act of beatification can now take place anywhere in the world and, although a pontifical act, does not require the pope’s presence.9 The large expansion in beatifications by the last two popes—319 by John Paul II and 92 by Benedict XVI—stands out because the cumulative number beatified (non-martyrs) from 1590 to 2012 is only 674. At the beginning of Pope Francis’s term in 2013, the stock of beatifieds not yet canonized was 399, by far a record.10 Our study uses econometric methods to analyze determinants of numbers of persons beatified and canonized by pope and region since 1590. The main analysis uses a seven-way regional breakdown: Italy, Other Western Europe, Eastern Europe, Latin America, North America (defined as Canada and the United States), Asia, and Africa. We chose these regions to allow for cross-sectional variation while avoiding a proliferation of small geographical entities that had mostly zeroes for persons beatified and canonized. We are particularly interested in evaluating the hypothesis that, at least in recent decades, the naming of blessed persons is part of 9 For the official description of the process, see www.vatican.va/roman_curia/congregations/csaints/documents/rc_con_csaints_doc_20070517_sanctorummater_en.html. 10 The breakdown was 176 from Italy, 125 Other Western Europe, 32 Eastern Europe, 32 Latin America, 14 North America, 13 Asia, and 7 Africa. 6 the competition between the Catholic Church and Protestantism and between Catholicism and no religion. We should stress that martyrs are excluded from our definition of blessed persons, whom we limit to confessors. Since death in the cause of the Church is a prerequisite for martyrdom, the selection process for martyrs is intrinsically different from that for confessors. Moreover, no miracle is required to beatify a martyr—the act of martyrdom is understood as itself a miracle of grace. Aside from these differences, a significant practical complication is that martyrs are often beatified in large groups. For example, shortly after becoming pope in 2013, Francis canonized the 813 martyrs of Otranto, who were executed in 1480 following an Ottoman siege of their southern Italian city. These martyrs had been beatified in 1771. Our view is that the determinants of choices of martyrs would differ significantly from those of confessors, whom we are studying. It would be interesting to analyze choices of martyrs in detail, but this project would require a lot of additional data retrieval, going beyond the scope of the present study.11 II. The Data Set of Blessed Persons and other Variables A. Popes, Beatifieds, Saints Table 1 gives statistics on popes’ terms from Sixtus V (1585-1590) to Benedict XVI (2005-2013). The sample for our statistical analysis starts after Sixtus V’s major reforms; that is, with Urban VII (1590-1590). This sample comprises 38 popes from 1590 to 2012.12 Table 1 shows each pope’s ID number, name, start and end year of his term, and tenure in years (based on the specific day started and ended). For the columns under the heading 11 A possible concern for our analysis is that the existence of martyrs may influence the choices of confessors. For example, the beatification of the assassinated Archbishop Romero of El Salvador in 2015 might provide a substitute for a confessor from that country. 12 We attribute any blessed persons named from 2013 onward to Francis, rather than Benedict XVI. 7 “Beatified,” the stock is the cumulative number beatified but not yet canonized at the start of the pope’s term. The duration is the mean number of years from beatification to the start of the pope’s term for the stock of beatified. The flow is the number beatified during the pope’s term. For the columns headed “Canonized,” the stock is the cumulative number canonized at the start of the pope’s term, and the flow is the number canonized during the term. We computed beatification rates, defined as the ratio of number beatified to papal tenure. Figure 1 shows beatifications per year since 1590 for popes with four or more beatifications. For the 38 popes in our main sample, the mean beatification rate was 1.1, the median was 0.41, the minimum was 0 (for thirteen popes), and the maximum was 12.0 for John Paul II. As indicated in the figure, the beatification rate was between 0 and 2 until the last two popes—when the rate rose sharply to 12.0 for John Paul II (1978-2005) and 11.7 for Benedict XVI (2005-2013). This pattern suggests a marked diminution of standards for beatification, a conjecture that we support later through regression analysis. Figure 2 shows canonizations per year since 1590 for popes with four or more canonizations. For the 38 popes in our sample, the mean canonization rate was 0.72, the median was 0.18, the minimum was 0 (for thirteen popes), and the maximum was 5.3 for Benedict XVI. The pattern in the figure suggests that John Paul II (3.0 canonizations per year) was only a moderate outlier, whereas Benedict XVI (5.3 per year) was a clearer outlier. However, an important positive influence on Benedict XVI’s canonization rate is the large stock of beatified person’s left by John Paul II. Our regression analysis finds that John Paul II was an outlier in numbers canonized compared to earlier popes but that Benedict XVI was not. The sharp expansion of beatified stocks can explain Benedict XVI’s large numbers canonized within the structure applicable to popes prior to John Paul II. 8 One way to accelerate the pace of beatification temporarily is to shorten the time between death and beatification. Before the 1983 reforms, this interval was restricted to at least 50 years, although popes occasionally ignored this restriction.13 Over the full sample, 1590 to 2012, the mean time from death to beatification was 117 years, and the median was 88. Figure 3, for popes with four or more beatifications, suggests that this lag rose early on—from Paul V (1605-1621) to Clement XII (1730-1740). However, the lag fell back around the time of Pius XI (1922-1939) and has since been relatively stable. For John Paul II and Benedict XVI, the means were 109 and 98 years, respectively, and the medians were 86 and 84, respectively. These values accord roughly with those prevailing since Pius XI. For the lag between beatification and canonization, the mean from 1590 to 2012 was 49 years, and the median was 20. Figure 4 indicates that John Paul II was consistent with the overall experience—a mean of 48 and median of 18. However, Benedict XVI shortened the lag—to a mean of 19 and median of 14. We interpret Benedict XVI’s shortened lag for canonization as part of his response to the dramatic increase in number of persons beatified (relatively recently) by John Paul II. Figure 5 shows how characteristics of blessed persons changed from 1590 to 2012.14 The male share of beatifications and canonizations diminished from the 16th to the 19th century, from 70-80% to 50-60%. However, no clear trends apply since 1900. Shares with some formal schooling were high throughout, averaging around 80%, and have not changed substantially over time, despite the expansion of formal schooling in the overall population. Similarly, the share 13 From 1590 to 1982, 37 of the 285 beatifications (13%) took place less than 50 years after the blessed person’s death. From 1983 to 2012, 63 of the 385 beatifications (16%) featured a lag of less than 50 years from the blessed person’s death. Thus, there was no clear break after 1983. 14 Stark (2004, pp. 51-59) discusses analogous characteristics of saints chosen between 500 and 1500. His analysis— which includes the period before the formalization of procedures for canonization in 1234—uses a data set that Stark constructed by using a variety of criteria to filter out many commonly included “saints.” 9 originating from urban areas has averaged nearly 80% and shows no clear trend, despite the global trend toward urbanization. Over the full sample (1590-2012), the 670 beatifications broke down, when classified by residence at death, as 45% Italy, 33% Other Western Europe, 7% Eastern Europe, 8% Latin America, 4% North America, 3% Asia, and 1.3% Africa. The breakdown was similar for the 286 canonizations: 46% Italy, 34% Other Western Europe, 7% Eastern Europe, 7% Latin America, 4% North America, 2.4% Asia, and 0.7% Africa. These data reveal the Catholic Church’s striking “home bias” in choices of blessed persons. Our formal statistical analysis confirms that these raw patterns hold up after taking account of time-varying Catholic population by region and changes in Catholic-Protestant competition and secularization. For example, when considering determinants of numbers beatified over the full sample, a Catholic person in Other Western Europe counts only about 30% as one from Italy. The weights in other regions compared to Italy are even smaller—12% for Eastern Europe, 4.4% for Latin America, 5.8% for North America, 3.2% for Asia, and 2.6% for Africa. Figures 6 and 7 show changes over time in the regional composition of beatifications and canonizations. Through the 19th century, blessed persons were overwhelmingly from Italy and Other Western Europe. In contrast, there has been some globalization of the process in the 20 th century. From 1939 to 2012, the shares of beatifications were 8% Eastern Europe, 9% Latin America, 5% North America, 4% Asia, and 2% Africa. Our formal analysis explains part of these regional shifts from responses to changes in measures of Catholic-Protestant competition and secularization. 10 B. Population and Religious Adherence We constructed rough estimates of population by country and year from 1550 to 2012. Our starting point was McEvedy and Jones (1978), henceforth called MJ, who provide estimates since 1500 at 100-, 50-, or 25-year intervals for 68 individual countries and 16 broader groups typically centered on a principal country.15 The constructs account for border changes and provide estimates of historical population corresponding to borders around 1975. The 84 entities come close to covering the world. Since we focus our statistical analysis on blessed persons chosen in seven regions, the aggregation of countries into the 16 broader groupings is not a problem. We adjusted the MJ data for 1975 on population in each country or group to match estimates from the World Bank’s World Development Indicators (WDI). We then used the available annual data from WDI to cover 1960-2012. Then we interpolated backwards from 1960 to 1550, using as benchmarks the interval data from MJ (adjusted by the multiple applying to 1975) going back to 1500. In this manner, we estimated total population for each year for each country or group of countries from 1550 to 2012. We then combined data on countries or groups within each region to compute estimates of total population by year for our seven regions.16 These population data are informative for long-term trends even if not for variations over short intervals before 1900. 15 These groupings are Botswana (including Namibia), Caribbean (5 countries), Central America (7 countries), Czechoslovakia (Czech Republic and Slovak Republic), East Africa (6 sub-Saharan African countries), Guyana (including Suriname), Indian sub-continent (comprising Bangladesh, India, and Pakistan), Malaysia (including Singapore), Nigeria (including 11 other sub-Saharan African countries), Palestine (including Israel), Russia (including Belarus, Lithuania, and Ukraine), Sahara (4 African countries), Saudi Arabia (including Bahrain, Kuwait, Qatar, UAE, and Yemen), South-Central Africa (Malawi, Zambia, and Zimbabwe), South-West Africa (6 subSaharan African countries), and Yugoslavia (Bosnia, Croatia, Macedonia, Serbia, and Slovenia). 16 Among possibly ambiguous cases, we defined Western Europe to include Greece; Eastern Europe to include Cyprus and Russia; Asia to include Australia and other Pacific-area countries; and Latin America to include Caribbean, Cuba, and Puerto Rico. 11 The blue graph in Figure 8 shows world population for selected years since 1550. This population reached 6.6 billion in 2009. (For comparison, the number for world population in 2009 from the World Bank’s World Development Indicators is 6.8 billion.) In 2009, 4.0 billion of the world’s total population was in Asia, 992 million in Africa, 576 million in Latin America, 402 million in Western Europe (including Italy), and 339 million in North America. Johnson (2010) estimates adherence by country for the major world religions in 1900, 1950, 1970, and 2000. We use the data for Catholic, a broad concept of Protestant (including Anglicans, independent Christian churches, and “marginal Christians” such as Mormons and Seventh Day Adventists), and Orthodox. Note that these adherence numbers exclude unaffiliated Christians. The sum of the three categories gives an estimate of overall affiliated Christians.17 For later purposes, we also use from Johnson (2010) information back to 1900 on adherence numbers to no religion. Johnson (2010) estimates Evangelical adherence by country for 1900, 1950, 1970, and 2000. The underlying concept begins with the number of persons affiliated with Evangelical churches, categorized as Evangelical based on belief structure. Specifically, Evangelicals believe in the inerrancy of the Bible, one God (in trinity), Jesus as deity, personal salvation through belief in Christ, the presence of the Holy Spirit in believing Christians, and resurrection into heaven and damnation for sinners.18 The estimated number of Evangelicals then adds in persons who self-identify as Evangelical. 17 Double-counting arises because some persons have affiliations to more than one Christian religion. However, the data in Barrett, Kurian, and Johnson (2001, Table 1-1) for 2000 indicate that doubly-counted religionists are less than 1% of affiliated Christians (if we assume that all the doubly-counted are Christians). It is possible, however, that Barrett, Kurian, and Johnson understate dual affiliations, particularly by neglecting continuing affiliations of “Christians” with indigenous faiths, particularly in Africa and Latin America. 18 See the statement of faith of the National Association of Evangelicals, http://www.nae.net/about-us/statement-offaith, accessed January 25, 2011. 12 Similarly, Johnson (2010) estimates numbers of Pentecostals by country in 1900, 1950, 1970, and 2000 by including persons affiliated with Pentecostal churches and then adding persons affiliated with non-Pentecostal churches who self-identify as Pentecostal. Pentecostalism is a form of evangelicalism that came into existence in the United States in 1901 as a schism from the holiness movement (Wacker [2003, p. 6]). This movement emphasizes personal conversion and a second Holy Spirit experience. Pentecostals (also described as Renewalists) are a subset of evangelicals in the sense of accepting the evangelical statement of faith. The belief structure of Pentecostals emphasizes gifts of the Holy Spirit such as speaking in tongues and divine miracles.19 For years before 1900, we used data from Barrett and Johnson (2001, Table 7-2) to estimate adherence numbers for Catholic, Protestant (broadly defined), and Orthodox for six regions: Western Europe (including Italy), Eastern Europe, Latin America, North America, Asia, and Africa. These data are for 1500, 1650, 1750, 1800, 1850, and 1900. We assumed that adherence ratios for each country or group after 2000 equaled those in 2000. We estimated adherence ratios from 1900 to 1950, 1950 to 1970, and 1970 to 2000 by interpolating linearly between the end points. We multiplied the computed adherence ratios by the estimates of total population for each country or group to estimate adherents to each religion annually since 1900 for each country or group. Finally, we added up across countries or groups in each region to estimate adherents to each religion in our seven regions: Italy, Other Western Europe, Eastern Europe, Latin America, North America, Asia, and Africa. Before 1900, we interpolated the available adherence ratios for six regions (Western Europe, Eastern Europe, Latin America, North America, Asia, and Africa) between benchmark 19 See the statements of faith of the Pentecostal Assemblies of God of America, http://www.paga.org/statemen.htm, accessed January 25, 2011 and The Church of God Full Gospel (Cleveland, TN), http://www.churchofgod.org/beliefs/bylaws-of-the-church-of-god, accessed January 22, 2016. 13 dates: 1500, 1650, 1750, 1800, 1850, and 1900.20 We then multiplied the adherence ratios by the population of each region to estimate adherents to Catholic, Protestant, and Orthodox religions for 1550-1900. For Evangelical, we assumed that the ratio of Evangelicals to Protestants in each region in 1900 applied to all earlier years.21 For Pentecostals, we accepted the treatment by Johnson (2010), which regards this movement as starting around 1900. Therefore, the Pentecostal numbers equal zero before 1900. Figure 9 shows the evolution since 1550 of the world’s affiliated Christian population by major type. In 2009, the total of 2.16 billion persons broke down into 1.09 billion Catholics, 0.83 billion Protestants (broadly defined), and 0.24 billion Orthodox. The estimated share of Catholics in the overall Christian population was 56% in 1550 and 50% in 2009. The figure also shows the number Evangelical, which equaled 238 million in 2009, compared to 88 million in 1950 and 68 million in 1900. Most of these fall into the broad Protestant category. The number Pentecostal in 2009 was 405 million, having grown dramatically from 9 million in 1950 and 0 in 1900. Figure 10 shows the breakdown of the world’s Catholic population into regions since 1550. The graph shows the shift over time of Catholic numbers from Western Europe toward Latin America, Africa, and Asia. In 2009, among an estimated 1.09 billion Catholics worldwide, the distribution was 43.5% Latin America, 17.5% Western Europe (including Italy), 14.2% Africa, 12.6% Asia, 6.7% North America, and 5.5% Eastern Europe. 20 For Italy, included with Western Europe in these data, we assumed that the adherence ratios for 1900--nearly 100% Catholic—applied to earlier years. A minor discrepancy is that the compositions of Western and Eastern Europe used by Barrett and Johnson (2001, Table 7-2) do not correspond precisely to those we used. 21 Evangelicalism might be viewed as having originated with George Whitefield and John Wesley in England and Jonathan Edwards in the United States; see Noll (2003, Introduction). However, since there were previous aspects of “evangelicalism” in England, elsewhere in Western Europe, and the United States, we did not set the world Evangelical number to zero at a date such as 1730. Our procedure generates a world Evangelical population of 17 million in 1800, 94% of which was in Western Europe and North America. In contrast, Barrett and Johnson (2001, Table 10-2) estimate the world population of Evangelicals in 1800 to be 25 million. 14 Figure 11 shows a comparable diagram for Protestant population. Western Europe was the main home for Protestants up to the early 1800s, but North America became increasingly important during the 19th century. In the later part of the 20th century, the dominant places for Protestants became Asia and Africa, with Latin America also rising in importance. In 2009, among an estimated 835 million Protestants worldwide, the distribution was 32.7% Africa, 24.7% Asia, 17.6% North America, 12.2% Western Europe (including Italy), 11.4% Latin America, and 1.5% Eastern Europe. An analogous figure for Orthodox population would show that Eastern Europe (including Russia) was dominant throughout the period since 1550. C. Catholic Religious-Competition Measures We suppose that the Catholic Church is most interested in competing with Protestants (or possibly other religions) in areas that have substantial representation of both Catholics and Protestants. Our approach emphasizes contacts between affiliates of the two religions. To gauge the likelihood of contacts, we constructed for each country the product of the adherence rates for Catholics and Protestants (multiplied by two). This variable gives the probability that a randomly selected pair of persons in a country will consist of one Catholic and one Protestant.22 This measure is small when a country’s religion market is nearly monopolized by either Catholics or Protestants and reaches a peak when the market is evenly divided (for a given total of Catholics and Protestants). The Catholic-Protestant competition measure is also small when the religion market is dominated by groups that are neither Catholic nor Protestant (including persons with no religious affiliation). We constructed analogous measures for likelihood of 22 One perspective is that the encounters associate with potential religious conversions. We imagine that the Catholic Church undertakes investments—including saint-making—intended to influence conversions toward Catholicism and away from Protestantism. We assume that, once a Catholic-Protestant encounter occurs, the probabilities of the three possible outcomes (conversion to Catholic, conversion to Protestant, and no change) do not depend on adherence shares in the overall population but that more Catholic investment makes Catholic conversion more likely and Protestant conversion less likely. We also assume that encounters between Catholics and other religions involve negligible probabilities of conversion. The optimal level of Catholic investment (for given investment by the Protestant leadership) is then increasing in the competition variable as we specified it. 15 contact and, hence, the extent of competition between Catholics and other Christian groupings: Evangelical, Pentecostal, and Orthodox. Since 1900, we construct the competition measures by country and then compute the regional variables by averaging over countries within each region while weighting each country’s value by its share of the region’s Catholic population. (Given our subsequent model of choices of blessed persons, this weighting scheme approximates the correct variable for the regional analysis.) Before 1900, we lack benchmark numbers for adherence numbers by country.23 Our computed competition variables rely on the assumption that adherence shares prior to 1900 equaled their values in 1900. Therefore, the calculations will be increasingly inaccurate as we move further back in time. Since the subsequent analysis suggests that Catholic-Protestant competition is important for saint making mainly since the early 20 th century, the deficiencies in the pre-1900 data may not matter a lot. Figures 12 and 13 show the changes across regions in Catholic-Protestant competition. The most important shifts since 1900 in Figure 13 are in Latin America and Africa. For Latin America, the value starts at 0.03 in 1900, then rises sharply due to the rise of Protestantism, reaching 0.26 in 2010. Similarly, in Africa, the competition variable is 0.01 in 1900 and 0.23 in 2010. According to our model, these patterns would motivate heightened saint making over time in Latin America and Africa. In contrast, in Asia and North America, the competition variable is high throughout, equaling 0.21 and 0.20, respectively, in 1900 and 0.28 and 0.20, respectively, in 2010. Although the average levels of these competition variables are high, the absence of substantial changes suggests that competitive pressure would not account for major shifts over time in saint making in Asia and North America. 23 We have estimates at the regional level, but these do not provide good estimates of country-weighted regional averages of the competition variable. 16 In Figure 12, the average levels of the competition variable for Italy, Other Western Europe, and Eastern Europe are much smaller than those for the other regions. Most importantly, the lack of large (absolute) changes in Figure 12 suggests that changing competition would not have a major impact on saint making in these regions. D. Secularization (no religion) As a general matter, secularization refers to a decline in various dimensions of individual religiosity and in the power of formal religion. Our application of this concept focuses on individual religiosity and uses a variable—the adherence rate to no religion—that is available for most countries back to 1900.24 A rise over time in the no-religion share means that a greater fraction of the population has strayed away from religion. (The reported no-religion shares for 1900 are typically close to zero.) If we multiply the current no-religion fraction of the population by the Catholic share of the population in 1900, we get an indication of the fraction of the wayward population that could be (or could have been) swayed to be Catholic. As an example, in Italy, the no-religion fraction of the population rose from 0.2% in 1900 to 16.5% in 2010. If we multiply the 16.5% by the Catholic adherence fraction in 1900, 99.7%, we get that roughly 16.5% is the potential share of the population in 2010 that could be targeted through a campaign to counter secularization. For other countries in Western Europe, the advance in the no-religion fraction is even stronger—for example, this fraction rose in France from 0.3% in 1900 to 19.8% in 2010 and in Germany from 0.3% in 1900 to 23.2% in 2010. However, in many Western European countries, the Catholic adherence share in 1900 is much smaller than that in Italy; for example, the fraction for Germany is 35.6% (while France is 98.1%). Thus, for Germany, the multiple of the no-religion share in 2010 (19.8%) and the 24 Johnson (2010) has estimates for 1900, 1950, 1970, 2000, 2005, and 2010. We lack good long-term data on alternative measures, such as attendance at formal religious services or the extent of religious beliefs. 17 Catholic adherence share in 1900 (35.6%) gives 7.0% as the fraction of the population that the Catholic Church might fruitfully target in 2010 through saint making. We constructed the secularization variable for each country for each year since 1900 as the multiple of the no-religion fraction and the Catholic adherence fraction in 1900. Then we constructed the regional variables, as above for competition, by weighting by the Catholic population share of each country within a region. Figures 14 and 15 show the time patterns in this secularization variable across regions since 1900. A prominent pattern in Figure 14 is the sharp rise over time for Italy and Other Western Europe. We predict that these patterns would induce increased saint making over time in these regions (or, at least, less of reduction than would otherwise have occurred). For Eastern Europe, the secularization variable rises strongly through 1970 and then levels off. In Figure 15, the absolute changes over time in the other regions are minor compared to those shown in Figure 14. Comparing Figures 12/13 with Figures 14/15 we see that the competition and secularization variables behaved very differently across time and regions. The competition variable should be most important for explaining increased saint making in recent decades in Latin America and Africa (Figure 13). The secularization variable should be most important for explaining heightened saint making in recent decades in Italy and the rest of Europe (Figure 12). An interesting case that is not captured by our overall approach is the United Kingdom. Because the Catholic share of the U.K. population has not been high (6.3% in 1900 and 9.0% in 2010), our Catholic-Protestant-competition and secularization variables have also not been high (even though the no-religion share rose from 1.9% in 1900 to 14.2% in 2010). The dominant adherence in the United Kingdom is to the Anglican Church (Church of England), comprising 64.3% in 1900 and 42.2% in 2010. The Anglican Church derives from King Henry VIII’s ouster 18 of the Catholic Church in 1534, and, in many theological respects, the Anglican Church remains similar to the Catholic Church. This closeness may explain the beatification of the English Cardinal John Henry Newman (1801-1890) by Benedict XVI in 2010. A year before Newman’s beatification, Benedict XVI issued an invitation to discontented Church of England members and clergy to join the Catholic Church. The Vatican set up a structure whereby Anglicans, including married priests, could practice Catholicism while maintaining much of their own identity and liturgy. Moreover, Newman’s beatification occurred at a time of a potentially serious schism in the Anglican Church over the ordination of female and homosexual priests. Thus, this particular beatification may reflect a special form of competition—the Catholic Church seeking converts from Anglicanism. E. The Peace of Westphalia Our regression sample for blessed persons begins in 1590 and, thereby, post-dates the Reformation in 1517 in Germany by Martin Luther, who was followed in Switzerland by Ulrich Zwingli and John Calvin. Our sample also post-dates the start of the Catholic CounterReformation, which began in the mid 1500s and featured the Council of Trent of 1545-1563. The Council reaffirmed the validity of sainthood and embraced the saint-making process as part of the effort to intensify the enthusiasm of the faithful: “The Council of Trent represents the attitude of the Catholic Church in its juridical understanding. The Catholic Reformation also understood that the recognition of saints by canonization served as a strong instrument of evangelization by emphasizing that within the church there existed the means by which people reached the heights of holiness.” (Cunningham [2005, p. 63].) Our post-1590 sample begins in this atmosphere, featuring intense Catholic-Protestant competition that frequently showed up as religious wars, culminating in the Thirty Years’ War 19 from 1618 to 1648. The end of this war with the Peace of Westphalia in 1648 was a watershed for religious warfare in Europe: “The Peace of Westphalia (1648) ended a generation of war and a century of strife. It was the end, not only of the Thirty Years War, but of religious wars in general. During the next century and a half the peace of Europe was often broken … but religion seldom provided the pretext.” (Cragg [1960, p.9].) A key point is that the Peace of Westphalia likely lessened the extent of CatholicProtestant competition, notably in Europe. We anticipate that this lessened competition implies diminished pressure for namings of blessed persons. That is, other things equal, the incentives for beatifications and canonizations will be stronger in the pre-Westphalia part of our sample (1590-1647) than subsequently.25 In our regression analysis, we look for this effect from a dummy variable for popes whose terms concluded before the Peace of Westphalia. This variable takes on the value one for the eight popes who started office between 1590 and 1623, with the last (Urban VIII) concluding his term in 1644 (Table 1). III. Framework for the Regression Analysis We use count-model regressions to assess the determinants of numbers of persons beatified and canonized over time by pope and region. This framework is appropriate because the dependent variable—numbers beatified or canonized—must take on non-negative, integer values. 25 Similarly, we predict that saint-making pressure would be weaker in the pre-Reformation period than afterwards. However, our sample excludes this period. Pfaff (2013, Table 3) considers effects from saint making on the probability that cities would shift after the Reformation away from Catholicism (measured by the abolition of the Catholic mass). He finds in a sample of 145 German-speaking towns that greater saint-making—gauged by the number of Catholic shrines erected up to 1530—was inversely related to the probability of shifting away from Catholicism in the period 1523-1545. This finding makes sense because the shift away from Catholicism was based on local decisions, as were the choices of beatifieds and creations of shrines. 20 The sample comprises 38 popes’ terms, from Urban VII (1590-1590) to Benedict XVI (2005-2013), as described in Table 1. The main analysis uses seven regions, as noted before: Italy, Other Western Europe, Eastern Europe, Latin America, North America, Asia, and Africa. The beatification variable is the number beatified during a pope’s term in each region.26 The canonization variable is defined analogously. Overall, there are 266 observations each for numbers beatified and canonized. Table 2, organized by pope, and Table 3, organized by pope/region, provide statistics on variables used in the regressions. Let Bij be the number beatified and Cij the number canonized by pope i (=1, …, 38) in region j (j=1, …, 7). We express the determinants of Bij and Cij as effects from pope-related variables, Xi, and pope/region-related variables, Zij. The Xi include dummy variables for John Paul II and Benedict XVI (who are also the only non-Italian popes in our sample), the pope’s overall years in office, Ti, whether the pope’s term preceded the Peace of Westphalia, and the pope’s age at the start of his term. The Zij include for the start of pope i’s term the Catholic population, the stock of persons beatified but not yet canonized, the mean durations in years for these beatified stocks, the Catholic-Protestant competition variable, and the secularization variable. For Bij, our key measure of the pool of candidates for beatification is the regional population of Catholics. This measure also gauges the primary set of customers for blessed persons chosen in the region. For Cij, we measure the pool of candidates for canonization by the regional stock of persons previously beatified but not yet canonized.27 26 We could look at each year within a pope’s term, but the inclusion of time variation for a given pope is unlikely to add much information. 27 We include in this stock four persons who were canonized since 1588 but reported as having been beatified well before 1588 (all from Italy). The problem is that the records are often unclear about what constitutes a beatification in the pre-1588 period. 21 We assume that the expected number beatified by pope i in region j, E(Bij), follows a standard form for count models: (1) E(Bij) = exp[c1∙log(Ti) + c2∙log(CATHPOPij) + c3∙(competition variable) + c4∙(secularization variable) + regional dummies + dummies for John Paul II and Benedict XVI + other variables]. The exponential form on the right side guarantees that the expectation of Bij is positive. If Bij is proportional to pope’s tenure, Ti, then c1=1, which turns out to accord roughly with the data. Note from Table 1 that Ti varies substantially across popes, from less than 0.1 year for Urban VII (1590), Leo XI (1605), and John Paul I (1978) to 31.7 for Pius IX (1846-1878) and 26.5 for John Paul II (1978-2005). If Bij is proportional to the number of candidates and customers in a region, then c2=1, again roughly in line with the data. If the choice of blessed persons were region neutral, the right-hand side would include a single constant, and the regional dummies would be unimportant—our results strongly reject this conjecture. If, given the other right-hand-side variables, John Paul II and Benedict XVI operated like previous popes, the dummies for these two popes would be zero—not what we find. We anticipate that higher values of the Catholic-Protestant competition variable and the secularization variable in a region will raise the number beatified. Hence, we predict c3>0 and c4>0 in equation (1). In addition, we investigate other forms of Catholic-Christian competition— involving Evangelicals, Pentecostals, and Orthodox. 22 One of the other explanatory variables on the right-hand side of equation (1) is the pope’s age at the start of his term. If older popes try hard to reach a target number of persons beatified during their terms, then this variable would have a positive coefficient. Our main analysis uses maximum likelihood in a negative-binominal form to estimate the coefficients in equation (1). The conclusions for the count-model regressions are similar with a simpler Poisson specification.28 The form for the expected number canonized by pope i in region j, E(Cij), is analogous: (2) E(Cij) = exp[c1∙log(Ti) + c2∙log(BEASTOCKij) + c3∙(competition variable) + c4∙(secularization variable) + regional dummies + dummies for John Paul II and Benedict XVI + other variables]. The main difference from equation (1) is that log(BEASTOCKij) replaces log(CATHPOPij), where BEASTOCKij is the stock of persons previously beatified but not yet canonized in region j at the start of pope i’s term. Note that regional variations in the beatified stock will already reflect regional biases given by the regional dummies in equation (1). Therefore, the regional dummies in equation (2) now reflect only biases conditional on the stock of beatifieds by region. Not surprisingly, we find much less indication of regional bias in this conditional sense. Similarly, the dummies for John Paul II and Benedict XVI in equation (2) are conditional on stocks of beatifieds. We find that, once we control for the sharp expansion of beatifieds under John Paul II, the dummy variable for Benedict XVI is unimportant for canonization. The list of other explanatory variables in equation (2) includes the mean duration in years of the stock of beatifieds in region j at the start of pope i’s term. If popes dislike leaving blessed 28 For discussions of count-model regression, see Cameron and Trivedi (1986) and Wooldridge (2002, Ch. 19). 23 persons waiting too long, then the coefficient on this variable would be positive. However, if beatified persons have varying and unobserved “quality,” then the estimated coefficient of this duration variable tends to be biased downward (for usual reasons related to unobserved heterogeneity). IV. Regression Results Statistics for the variables used in the regressions are in Tables 2 and 3. The results from count-model regressions, using a negative-binomial specification, are in Table 4. The regressions follow the forms of equations (1) and (2). A. Beatifications Consider the results for numbers beatified by pope and region, Bij, in Table 4, column 1. 1. Pope’s Tenure. The estimated coefficient on the log of pope’s tenure is 1.37 (s.e.=0.17), which is significantly positive and just significantly different from one at the 5% level (p-value=0.033). Thus, the results are roughly consistent with the property that twice as much time in office leads to twice as many persons beatified, other things equal. 2. Catholic Population. The estimated coefficient on the scale variable for candidates—the log of Catholic population in a region—is 0.63 (s.e.=0.17), which is significantly positive and just significantly different from one at the 5% level (p-value=0.033). Thus, the results are roughly consistent with constant returns to scale with respect to a region’s Catholic population. The shortfall of the estimated coefficient from one may result from attenuation bias due to measurement error in Catholic population by region, particularly in the pre-1900 sample. 24 3. Regional Dummy Variables. The regional dummy variables show substantial bias in beatification first toward Italy and second toward Europe. Each coefficient should be interpreted relative to Italy. For example, for Other Western Europe, the point estimate, -1.19 (s.e.=0.20), means that the expected number beatified compared to Italy is multiplied by 0.30, the exponential of -1.19. Hence, Western European Catholics count about 30% as much as Italian Catholics as influences on the expected number beatified. Each of the estimated regionaldummy coefficients is negative and significantly different from zero. The point estimates imply multiples relative to Italy of 12% for Eastern Europe, 4.4% for Latin America, 5.8% for North America, 3.2% for Asia, and 2.6% for Africa. Thus, the home bias in the Catholic Church’s saint making is remarkable. 4. John Paul II and Benedict XVI. The estimated coefficients on the dummy variables for John Paul II (1978-2005) and Benedict XVI (2005-2013) are each significantly positive— 1.78 (s.e.=0.27) and 1.18 (0.31), respectively. These estimates imply that, compared to earlier popes, the propensities to beatify are multiplied by 5.9 for John Paul II and 3.3 for Benedict XVI. The estimated coefficient for John Paul II is just significantly different from that for Benedict XVI at the 5% level (p-value=0.046). Thus, with respect to beatification, John Paul II is a positive outlier29 and Benedict XVI is also a positive outlier but by less than John Paul II. 5. Catholic-Protestant Competition. In Table 4, the Catholic-Protestant competition variable applies only to popes that started in 1900 or later, thereby comprising the last nine popes (Table 1). The break point of 1900 corresponds roughly to our prior assessment of when Catholic-Protestant competition became important for saint making (as discussed before) and 29 Our findings on John Paul II are consistent with observations by Cunningham (2005, pp. 121-122): “The prodigal use of this process [making saints] has been the subject of some wonder and criticism, both in Rome and in other parts of the church. The criticism comes mainly from those (including some in the Roman curia) who think both that the process is too hasty and that the multiplication of new saints cheapens the whole notion of those who are in the canon of the saints.” 25 also turns out to maximize the likelihood for the beatification regression. We discuss later alternative starting dates for the competition effect. The Catholic-Protestant competition variable has a significantly positive coefficient, 5.2 (s.e.=1.9). This estimated value implies that a rise in the competition variable by one standard deviation (0.083 in Table 3) raises the expected number beatified by 54% (from exponentiation of the product of 5.2 and 0.083). Therefore, the estimated response to competition with Protestants is quantitatively important. 6. Secularization. The secularization variable also applies only to popes that started in 1900 or later. However, in practice, the value of this variable is close to zero for earlier years if we assume that the adherence rate for no religion remained close to the small values reported for 1900. Therefore, the results are virtually unchanged if we allow the secularization variable to apply also to pre-1900 popes. The estimated coefficient on the secularization variable is significantly positive, 4.6 (s.e.=2.2). This estimated value implies that a rise in the variable by one standard deviation (0.020 in Table 3) raises the expected number beatified by 10%. 7. Peace of Westphalia. The regressions include the pre-Westphalia dummy variable, which takes on the value one for popes who ended office up to 1644 (Table 1). In Table 4, column 1, the estimated coefficient is positive and significantly different from zero, 0.66 (s.e.=0.28). This estimate implies that the numbers beatified before the Peace of Westphalia were multiplied by 1.9 compared with the later period. Hence, as conjectured, the end of religious warfare in Europe appears to have diminished rates of beatification. 8. Pope’s Age. The estimated coefficient on pope’s age at the start of his term, 0.034 (s.e.=0.015), is positive as expected and statistically significantly different from zero at the 5% level. A one-standard-deviation increase in a pope’s starting age (by 7.8 years, as shown in Table 2) is estimated to raise the number beatified by 30%—but for a given papal tenure. Hence, 26 there is evidence that older popes—typically with lower life expectancy—tend to accelerate the process of beatification. This result makes sense if popes have target numbers for blessed persons chosen during their terms. B. Canonizations Consider now the results for numbers canonized by pope and region, Cij, in Table 4, column 2. We highlight the results that differ from those for beatification. 1. Pope’s Tenure. The estimated coefficient on the log of pope’s tenure is 0.76 (s.e.=0.15), which is significantly positive but insignificantly different from one at the 5% level (p-value=0.12). Thus, as with beatifications, the results are roughly consistent with the property that twice as much time in office leads to twice as many persons canonized, other things equal. 2. Stock of Beatifieds. An important difference from the specification for beatification is that the canonization propensity is conditioned on the number of persons already beatified but not yet canonized. Hence, the beatified stock at the start of each pope’s term now appears instead of Catholic population as the key scale variable in the regression.30 In Table 4, column 2, the estimated coefficient on the log of a region’s beatified stock is 0.96 (s.e.=0.19), which is significantly positive but insignificantly different from one at the 5% level (p-value=0.82). Hence, the results are consistent with constant returns in the response of number canonized to the number of candidates as represented by the beatified stock. Note that, in contrast to numbers of Catholics, the numbers of persons already beatified by region would not be subject to important 30 A minor difficulty concerns the measurement of beatified stocks at the start of the sample, 1590. Our numbers include only 4 pre-1588 beatifications, corresponding to 4 persons subsequently canonized (all from Italy). There are also 11 cases of post-1588 canonizations for which no prior beatification was noted in the records (4 Italy, 3 Other Western Europe, and 4 Eastern Europe). We treat these 11 as lacking prior beatification. Thus, in our data, there is a small probability of being canonized without having previously been beatified. To allow for this possibility, we adjusted the beatified stocks in 1590 upward slightly (by 0.5 for each region) from 4 for Italy and 0 for the other regions. The value 0.5 was chosen to maximize the likelihood of the canonization regression. 27 measurement error. This difference likely explains why the estimated scale effect is closer to unity in the regression for numbers canonized as compared to that for numbers beatified. The regression for canonization also includes the mean duration of the stock of beatifieds (those previously beatified but not yet canonized) at the start of a pope’s term. 31 The estimated coefficient, 0.0052 (s.e.=0.0032), is positive but not significantly different from zero at the 5% level (p-value = 0.11). The point estimate implies that an extra 44 years of mean duration in the stock of beatifieds at the start of a pope’s term (the standard deviation of this variable in Table 3) raises the number canonized by 26%. We anticipated a positive effect because popes would likely find it undesirable to keep candidates for sainthood—who have already been judged worthy of beatification—waiting too long. The effect of the duration of the beatified stock on the canonization propensity is analogous to effects of duration on hazard rates (Kiefer [1988]). Unobserved heterogeneity in characteristics, in our case the “quality” of candidates for sainthood, can obscure a rising hazard rate. That is, since low quality candidates likely remain longer in the pool, the estimated coefficient on duration tends to be biased downward compared to the true effect. Since we do not observe multiple spells for a given candidate (because successful persons for sainthood never return to the pool of candidates), it is unclear how to proceed statistically to eliminate the effect from unobserved quality differences. However, since we do estimate a positive effect from duration of the beatified stock on canonization propensity, it may be that unobserved quality differences among beatifieds are not important in the saint-making process. In any event, the estimated coefficient, 0.0052, of the mean duration variable likely understates the true effect of duration. 31 The durations were set to zero for cases with zero stocks of beatifieds. 28 3. Regional Dummy Variables. The estimates of the coefficients on the regional dummy variables in Table 4, column 2, show preferences for Italy (the left-out region) and Europe, but the effects are quantitatively much smaller than those for beatification. Moreover, none of the estimated coefficients on regional dummies in the canonization regression are significantly different from zero at the 5% level, and the hypothesis that the six coefficients are jointly zero is accepted with a p-value of 0.95. These findings make sense because the regional biases in the beatification process would already be transmitted to canonization through effects from beatified stocks, even with no additional biases at the canonization stage. 4. John Paul II and Benedict XVI. In Table 4, column 2, the estimated coefficient on the dummy variable for John Paul II (1978-2005) is significantly positive, 1.93 (s.e.=0.40), implying a multiple of 6.9 on numbers canonized compared to earlier popes. However, the estimated coefficient for Benedict XVI (2005-2013), -0.18 (0.49), differs insignificantly from zero. The hypothesis of equality for the coefficients on the dummy variables for these two popes is decisively rejected (p-value=0.000). Hence, there is evidence that John Paul II was a sharp positive outlier in numbers canonized, given the values of the explanatory variables. However, there is no indication that Benedict XVI behaved abnormally with regard to canonization propensity when compared to popes prior to John Paul II, given the values of the explanatory variables. The key element in the explanatory variables for Benedict XVI is that John Paul II dramatically raised the stock of beatifieds left behind. This stock expanded from the 105 available to John Paul II in 1978 to the 344 available to Benedict XVI in 2005 (Table 1). In the absence of this expansion, the estimated coefficient on the beatified stock in Table 4, column 2, 29 implies that the number canonized by Benedict XVI would have been around 14, rather than the observed number of 42.32 Although Benedict XVI was not an outlier in numbers canonized, given the large stock of beatifieds when he entered office, the canonizations tended to be concentrated among persons beatified relatively recently, notably by John Paul II. This mechanism explains why the average time between beatification and canonization fell sharply under Benedict XVI, as shown in Figure 4. In contrast, because John Paul II was able to work off the backlog of persons beatified long ago by previous popes, the average time between beatification and canonization did not fall sharply during his tenure.33 5. Other variables. Results for other variables for canonization in Table 4, column 2, are analogous to those for beatification. Significantly positive effects are found for the CatholicProtestant competition variable, the pre-Westphalia dummy, and the pope’s age at the start of his term. One difference is that the estimated coefficient on the secularization variable differs insignificantly from zero. C. Extensions of the Model 1. Catholic-Protestant competition over time. The regressions for numbers beatified and canonized in Table 4 use the variables for Catholic-Protestant competition and secularization effective only for the nine popes starting since 1900; that is, those entering office in 1903, 1914, 1922, 1939, 1958, 1963, 1978, 1978, and 2005 (Table 1). We can extend the analysis by treating the starting date as a parameter to be estimated along with the other coefficients in the regression. Using the property that -2∙log(likelihood ratio) is distributed asymptotically as a chi- 32 Pope Francis had a beatified stock of 399 at the start of his term in 2013, and he already canonized 25 persons through 2015. This beatification rate—over 8 per year—dwarfs those of John Paul II (3.0 per year) and Benedict XVI (5.3 per year). 33 At the end of 2015, there were 57 candidates remaining who had been beatified before 1978. 30 squared distribution with 1 degree of freedom, the 5% confidence interval for the starting date in the beatification regression turns out to be between 1878 (sample of 10 popes beginning with Leo XIII) and 1922 (7 popes beginning with Pius XI). The likelihood is maximized with a starting date of 1903 (consistent with Table 4, column 1), but the likelihood is very flat for starting dates between 1903 and 1922. In practice, the main influence on the likelihood is the Catholic-Protestant competition variable—the secularization variable plays a negligible role because it remains small in each region throughout the relevant time frame. If we take an analogous approach for the canonization regression (Table 4, column 2), we find that the likelihood is maximized at a starting date of 1922 (sample of 7 popes). In this case, the 5% confidence interval for the starting date is between 1903 (sample of 9 popes) and 1930 (6 popes). Therefore, the assumed starting date of 1900 in Table 4 lies within the 5% confidence interval for both the beatification and canonization regressions. The bottom line is that Catholic-Protestant competition was an important force on choices of blessed persons starting early in the 20th century. The most likely date of onset for significant effects was in the range of 1903 (Pius X), 1914 (Benedict XV), and1922 (Pius XI). A possible explanation for the timing is the promulgation of the Pio-Benedictine Code of Canon Law in 1917, discussed earlier, which laid out the rules for saint making. Another possible element in the timing is the loss of the Papal States and the pope’s temporal powers between 1860 and 1870, leading up to the full unification of Italy in 1871. The pope then lived in a political limbo until the Lateran Pacts of 1929 established the Vatican as an independent city-state and restored the political sovereignty of the pope. 2. Other forms of Catholic-Christian competition. The results in Table 4 consider religious competition only in the form of Catholic-Protestant. We also examined analogously 31 constructed competition variables for Catholic-Evangelical, Catholic-Pentecostal, and CatholicOrthodox. We entered these variables one at a time to the equations estimated in columns 1 and 2 of Table 4. In most cases, the estimated coefficient of the new competition variable differed insignificantly from zero, and the estimated coefficient for Catholic-Protestant competition did not change substantially from that reported in Table 4, columns 1 and 2. As an example, for numbers beatified, the estimated coefficient of Catholic-Orthodox competition is -1.2 (s.e.=7.3), while that for Catholic-Protestant competition is 5.2 (2.0). A possible interpretation of the weak results for Catholic-Orthodox competition is that the Catholic and Orthodox Churches—which separated in the Great Schism of 1054—came to an understanding long ago not to interfere with the relationship of each church to its own members. From this perspective, it is reasonable that the Catholic Church would not be competing much for members through saint making with the Orthodox Church. Another point is that the sample features little within-country contact between Catholic and Orthodox populations—only in Eastern Europe to a noticeable extent. For Catholic-Evangelical and Catholic-Pentecostal competition, the only statistically significant result was the negative estimated coefficient, -7.7 (s.e. = 3.7), for CatholicPentecostal in the equation for beatification. In this case, the coefficient for Catholic-Protestant competition became 7.1 (2.7), higher than that in Table 4, column 1. These results suggest that the overall population of Protestants may be better than the sub-sets of Evangelicals or Pentecostals as gauges of competition with Catholics. A likely explanation is that the data on Evangelicals and Pentecostals are more prone than those on total Protestants to measurement error. 32 3. Poisson count model. We redid the count-model regressions from Table 4 using a Poisson specification instead of the negative binomial. The Poisson form is the special case of the negative binomial when the shape or variance parameter shown in Table 4 equals zero, so that the log of this parameter equals minus infinity. We can test the Poisson constraint by comparing the log likelihood of the fitted Poisson model with that of the negative binomial. Asymptotically, twice the log of the likelihood ratio is distributed as a chi-squared variable with one degree of freedom. For numbers beatified, the Poisson constraint is accepted with a p-value of 0.86. Given this high p-value, it is unsurprising that the estimated coefficients of the explanatory variables in the Poisson specification are close to those shown for the negative-binomial model in Table 4, column 1. For numbers canonized, the Poisson constraint is accepted with a lower p-value, 0.097. In this case, the changes in some of the estimated coefficients are more noticeable; for example, the coefficient of the dummy variable for John Paul II becomes 1.61 (s.e.=0.27), compared with 1.91 (0.40) in Table 4, column 2. However, the main results in the Poisson specification do not differ much from those in the negative-binomial model. 4. Individual effects for France, Germany, and Spain. As discussed before, our main specification considers the regional breakdown of Italy, Other Western Europe, and 5 other regions. We can go further to break out individual countries but going too far results in a proliferation of zeroes. We redid the estimation while breaking out the three Western European countries with (aside from Italy) the largest Catholic populations—France, Germany, and Spain. Aside from the coefficients for regions/countries, the results for the beatification regression are close to those reported in Table 4, column 1.34 The principal change is that the standard errors of 34 For the canonization regression, the regional dummies were less important, and the breaking out of France, Germany, and Spain has a smaller effect on the results. 33 the coefficients are higher. The main new information is the estimates on coefficients of dummy variables for the separated countries. The previous estimate for Other Western Europe in Table 4, column 1, was -1.19 (s.e.=0.20). Now we get -1.42 (0.22) for France, -2.84 (0.60) for Germany, -0.88 (0.23) for Spain, and -2.13 (0.30) for the rest of Western Europe. (The hypothesis that these four coefficients are equal is strongly rejected.) Thus, within Western Europe, the bias compared to Italy is especially strong for Germany, weaker for France, and weaker still for Spain. The remainder of Western Europe lies between France and Germany. D. Diminished Standards for Naming Blessed Persons Our interpretation of the coefficients on the dummy variables for John Paul II and Benedict XVI in Table 4, column 1, is that the sharp upward movements in beatifications under these two popes represent diminished standards for declaring persons to be blessed (see n. 29). This pattern of reduced standards also applies in column 2 to canonizations under John Paul II but not to those under Benedict XVI. Another possible indicator of lowered standards is the tendency to beatify previous popes. The data since 1590 up to the end of Benedict XVI’s term show only two popes canonized35—Pius V in 1712 (beatified in 1672) and Pius X in 1954 (beatified in 1951). There were four more beatified36 but not canonized through 2012—Innocent XI in 1956, Pius IX in 2000, John XXIII in 2000, and John Paul II in 2011. Hence, the three recent beatifications of popes under John Paul II and Benedict XVI is a large number when placed in historical perspective. Moreover, the rapid elevations of John XXIII and John Paul II to sainthood in 2014 35 The data since 1234 contain one additional canonization of a pope—Celestine V (pope in the one year 1294), who was canonized in 1313 by Clement V (1305-1314). 36 Our data, based on Catholic Church, Congregatio pro Causis Sanctorum (1999), do not include as beatified two popes that are commonly described as blessed: Benedict XI (1303-1304) and Urban V (1362-1370). 34 and the ongoing consideration of several other recent popes for beatification—Pius XII, Paul VI, and John Paul I—are unprecedented. By its recent canonizations of popes, the Church is reaffirming the universal and unique role of the pope in Christendom. This approach resembles Pope Pius IX’s plan when he declared the dogma of the infallibility of the pope in 1870. This dogma was declared the same year that the Church lost its Papal territories and the pope his temporal power. In asserting the dogma of papal infallibility, Pius IX successfully appealed to the religious allegiance of Catholics living in countries with governments opposed to the papacy. Naming popes as saints has a similar effect of drawing on and strengthening the allegiance of the faithful worldwide, particularly since popes are already superstars of the Catholic Church. The only drawback to this strategy, if done excessively, is that the special effects may weaken. More broadly, the observation that the Catholic Church has diminished its standards for naming blessed persons does not imply that this policy change has been a mistake. In fact, our conjecture is that Pope John Paul II and his successors have found an innovative strategy for intensifying the devotion of Catholics and, thereby, competing more effectively against Protestantism and no religion. However, despite the anecdote about Guatemala’s saint Hermano Pedro that began this paper, we have not isolated the effects of saint-making policies on religiosity. Such an analysis would be important but challenging. V. Concluding Observations Our research uses a data set that encompasses numbers and characteristics of blessed persons (non-martyrs) selected by the 38 popes aside from Francis that started office since 1590. Our results assess determinants of numbers beatified and canonized over time for seven regions. 35 We found several pieces of evidence consistent with saint making as a competitive strategy. First, the numbers beatified and canonized since the early 20th century responded positively to heightened Catholic-Protestant competition. Second, beatification responded positively to secularization, amounting to competition with no religion. Finally, the numbers beatified and canonized were significantly higher, other things equal, under popes whose terms ended before the Peace of Westphalia of 1648. Our interpretation is that the ending of religious warfare in Europe diminished the incentives of the Catholic Church to compete with Protestantism via saint making. We verify that the last two popes are outliers in terms of increased beatification. John Paul II is also an outlier in canonization, but Benedict XVI’s large numbers canonized can be explained as a normal response to the dramatic rise in stocks of beatifieds left behind by John Paul II. Benedict XVI’s working off of these stocks entailed canonizations of many persons beatified by his predecessor and, therefore, required a sharp reduction in the interval between beatification and canonization. Our immediate plan for future research is to apply the framework to the Catholic Church’s selection of Cardinals. For this purpose, we can use the remarkable data set on Cardinals constructed over many years by Miranda (2010). This data set gives names and characteristics, including residence at birth, of the 4260 Cardinals chosen from 492 to 2010. Thus, these data cover a much longer period and comprise many more designees compared to the sample of non-martyr blessed persons. We anticipate giving particular attention to the changing geographical composition of the Cardinals. Our results suggest that choices of numbers and locations of blessed persons since the early 20th century reflect the Catholic Church’s desire to invigorate the Catholic faithful and 36 avoid conversions to Protestantism and no religion. We view these findings as constituting a first-stage analysis in a model where the second stage comprises the responses of religious intensity and conversion to the Church’s policies. We would particularly like to know whether the Church has succeeded in using saint making as a strategic device to inspire more intense religiosity, less conversion to Protestantism, and a reduced propensity to drop religion entirely (secularization). This analysis will be challenging because the Church’s choices on saints are endogenous to Protestant inroads and secularization and because religiosity and patterns of adherence depend on many variables other than saint making. 37 References Barrett, D.B. and T.M. Johnson (2001). World Christian Trends, AD 30-AD 2200, Pasadena CA, William Carey Library. Barrett, D.B., G.T. Kurian, and T.M. Johnson (2001). World Christian Encyclopedia, v.1, Oxford, Oxford University Press. Barro, R.J., R.M. McCleary, and A .McQuoid (2011). “The Economics of Sainthood (a Preliminary Investigation),” Chapter 10 of R.M. McCleary, ed., The Oxford Handbook of the Economics of Religion, Oxford, Oxford University Press. Burns, P., ed. (1995 and later years). Butler’s Lives of the Saints, Collegeville MN, Burns & Oates, The Liturgical Press. Cameron, A.C. and P.K. Trivedi (1986). “Econometric Models Based on Count Data: Comparisons and Applications of Some Estimators and Tests,” Journal of Applied Econometrics, 1, January, 29-53. Catholic Church, Congregatione pro Causis Sanctorum (1999). Index ac Status Causarum, 2nd edition, Vatican City. Cragg, G.R. (1960). The Age of Reason, 1648-1789, New York, Pelican Books. Cunningham, L.S. (2005). A Brief History of Saints, Oxford, Blackwell. Delooz, P. (1962). “Pour une etude sociologique di Canonizzazione della Controriforma,” Archives de Sociologie des Religions, 13, 17-43. Ferrero, M. (2002). “Competition for Sainthood and the Millennial Church,” Kyklos, 55, 335-360. Harvey, J.C. (2007). “The Role of the Physician in Certifying Miracles in the Canonization Process of the Catholic Church III,” Southern Medical Journal, 100, 12, December, 1255-1258. Iannaccone, L.R. (1998). “Introduction to the Economics of Religion,” Journal of Economic Literature, 36, September, 1465-1496. Johnson, T.M., general editor (2010). World Christian Data Base, available on the Internet at www.worldchristiandatabase.org through [email protected]. Kemp, E.W. (1945). “Pope Alexander III and the Canonization of Saints: The Alexander Prize Essay,” Transactions of the Royal Historical Society, fourth series, 27, 13-28. 38 Kiefer, N.M. (1988). “Economic Duration Data and Hazard Functions,” Journal of Economic Literature, 26, June, 646-679. Klauser, T. (1979). A Short History of the Western Liturgy, 2nd edition, Oxford, Oxford University Press. McBrien, R.P., ed. (1995). The Harper Collins Encyclopedia of Catholicism, New York, Harper Collins. McBrien, R.P. (2000). Lives of the Popes: The Pontiffs from St. Peter to John Paul II, San Francisco CA, Harper. McCleary, R.M. and R.J. Barro (2006). “Religion and Economy,” Journal of Economic Perspectives, 20, Spring, 49-72. McEvedy, C. and R. Jones (1978). Atlas of World Population History, New York, Penguin Books. Miranda, S. (2010). “The Cardinals of the Holy Roman Church,” available on the Internet at www2.fiu.edu/~mirandas/cardinals.htm. Noll, M.A. (2003). The Rise of Evangelicalism, Downers Grove IL, InterVarsity Press. Peters, E.N. (2001). The 1917 or Pio-Benedictine Code of Canon Law, San Francisco CA, Ignatius Press. Peterson, V.A. (1940). “The Development of the Canon Law since 1500 A.D.,” Church History, 9, September, 235-252. Pfaff, S. (2013). “The True Citizen of the City of God: the Cult of Saints, the Catholic Social Order, and Urban Reformation in Germany,” Theory and Society, 42, March, 189-218. Sherkat, D. and C. Ellison (1999). “Recent Developments and Current Controversies in the Sociology of Religion,” Annual Review of Sociology, 25, 363-394. Stark, R. (2003). “Upper Class Asceticism: Social Origins of Ascetic Movements and Medieval Saints,” Review of Religious Research, 45, September, 5-19. Stark, R. (2004). Exploring the Religious Life, Baltimore, Johns Hopkins University Press. Wacker, G. (2003). Heaven Below: Early Pentecostals and American Culture, Cambridge MA, Harvard University Press. Wooldridge, J.M. (2002). Econometric Analysis of Cross Section and Panel Data, Cambridge MA, MIT Press. 39 Name ID Sixtus V Urban VII Gregory XIV Innocent IX Clement VIII Leo XI Paul V Gregory XV Urban VIII Innocent X Alexander VII Clement IX Clement X Innocent XI Alexander VIII Innocent XII Clement XI Innocent XIII Benedict XIII Clement XII Benedict XIV Clement XIII Clement XIV Pius VI Pius VII Leo XII Pius VIII Gregory XVI Pius IX Leo XIII Pius X Benedict XV Pius XI Pius XII John XXIII Paul VI John Paul I John Paul II Benedict XVI Francis 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 Table 1 Popes’ Terms from 1585 to 2013 Start End Tenure Beatified Canonized (years) Stock Duration Flow Stock Flow 1585 1590 5.35 4 98 0 34 1 1590 1590 0.04 4 103 0 35 0 1590 1591 0.87 4 103 0 35 0 1591 1591 0.17 4 104 0 35 0 1592 1605 13.09 4 104 3 35 2 1605 1605 0.07 7 70 0 37 0 1605 1621 15.72 7 70 13 37 2 1621 1623 2.41 20 36 2 39 5 1623 1644 20.99 17 43 8 44 2 1644 1655 10.32 24 40 3 46 0 1655 1667 12.13 27 46 1 46 2 1667 1669 2.47 26 58 1 48 2 1670 1676 6.24 25 59 5 50 5 1676 1689 12.90 25 58 3 55 0 1689 1691 1.32 28 64 1 55 5 1691 1700 9.22 24 60 0 60 0 1700 1721 20.31 24 69 2 60 4 1721 1724 2.83 22 82 0 64 0 1724 1730 5.74 22 85 3 64 9 1730 1740 9.58 16 66 4 73 4 1740 1758 17.69 16 66 8 77 4 1758 1769 10.59 20 66 4 81 6 1769 1774 5.32 18 67 2 87 0 1775 1799 24.53 20 66 19 87 0 1800 1823 23.45 39 52 7 87 4 1823 1829 5.38 41 62 5 92 0 1829 1830 1.67 46 61 0 92 0 1831 1846 15.34 46 62 4 92 5 1846 1878 31.67 45 76 16 97 5 1878 1903 25.42 55 88 23 102 11 1903 1914 11.05 68 80 11 113 4 1914 1922 7.39 75 80 4 117 3 1922 1939 17.02 76 86 30 120 24 1939 1958 19.62 83 82 38 143 30 1958 1963 4.60 92 80 6 173 10 1963 1978 15.14 88 81 33 183 16 1978 1978 0.09 105 75 0 199 0 1978 2005 26.52 105 75 319 199 80 2005 2013 7.87 344 28 92 279 42 2013 --399 30 -321 -- 40 Notes to Table 1 Tenure is in years, based on number of days as pope. For Beatified: Stock is the cumulative number previously beatified, but not yet canonized, at the start of a pope’s term; Duration is the mean years from beatification to the start of the pope’s term for the stock of beatified; and Flow is the number beatified during the pope’s term. For Canonized: Stock is the cumulative number at the start of a pope’s term, and Flow is the number canonized during the pope’s term. The number 35 for the canonized stock at the end of the term of Sixtus V (in 1590) is the number canonized between 1234 and 1589. The year 1234 corresponds to the declaration by Pope Gregory IX that papal approval was required for canonization. However, the requirements for beatification remained unclear at this time. In 1588, Pope Sixtus V enacted detailed reforms of procedures for canonization and beatification. However, the Papacy did not gain complete control of the process until the regime of Urban VIII, who was pope from 1623 to 1644. Of the 286 persons canonized since 1590, the reports on canonizations indicate that 4 were beatified before 1590, and 11 were not noted as previously beatified. The number 4 for the stock of beatified for Sixtus V reflects the 4 pre-1590 beatifications among persons canonized since 1590. This treatment assumes that no other persons were “officially” beatified before 1590. Hence, the cumulative number beatified at the start of 2013 in our data is 399 (stock of beatified as of 2013) + 321 (stock of canonized as of 2013) – 11 (canonized since 1590 without prior beatification) –35 (stock of canonized in 1590) = 674. Of these, 670 were beatified since 1590. 41 Table 2 Statistics for Popes (N=38) Variable Mean Median Number of beatifications 17.6 4.0 Number of canonizations 7.5 3.5 Beatifications per year 1.13 0.41 Canonizations per year 0.72 0.18 Stock of beatifieds (start of term) 45.3 25.0 Mean duration of stock of beatifieds 69.8 68.0 (years, start of term) Stock of canonized (start of term) 87.8 75.0 Tenure of pope (years) 11.0 9.9 Age of pope (years, start of term) 65.5 66.0 s.d. 52.9 14.7 2.61 1.18 57.8 18.3 Max 317 78 12.0 5.33 344 104 Min 0 0 0 0 4 28 56.0 8.7 7.8 279 31.7 80 35 0.04 51 Table 3 Statistics for Regression Sample, Popes/Regions (N=266) Variable Mean Median s.d. Max Number of beatifications 2.51 0 10.7 125 Number of canonizations 1.07 0 3.34 34 Catholic population (millions) 27.2 10.5 48.5 452.8 log(Catholic population) 1.59 2.35 2.66 6.12 Stock of beatifieds (start of term) 6.47 1.00 15.07 150 log(stock of beatifieds + 0.4) 0.54 0.34 1.62 5.01 Mean duration of stock of beatifieds 32.7 2.5 43.9 189 (years, start of term) log(tenure of pope) 1.69 2.30 1.72 3.46 Competition: Catholic/Protestant 0.090 0.079 0.083 0.274 Secularization variable 0.0070 0.0018 0.0204 0.1644 Min 0 0 0.0 -5.22 0 -0.92 0 -3.34 0.004 0.0000 Note: See notes to Table 1 for further discussion. Data are for 38 popes’ terms from Urban VII (start year 1590) to Benedict XVI (start year 2005). Beatifications (canonizations) per year equal the number of beatifications (canonizations) during a pope’s term divided by the pope’s tenure in years. Stock of beatifieds is the cumulative number beatified, but not yet canonized, at the start of a pope’s term. Duration, applying to the stock of beatifieds at the start of a pope’s term, is the mean years from beatification to the start of the pope’s term. Stock of canonized is the cumulative number at the start of a pope’s term. Data in Table 3 are for the regression sample used in Table 4. The competition and secularization variables are described in Sections II.C and II.D. 42 Table 4 Negative-Binomial Count Regressions for Numbers Beatified and Canonized Beatified Canonized Independent variable (1) (2) -6.28** -8.77** Constant (1.32) (1.59) 1.368** 0.761** log(tenure of pope (0.171) (0.152) 0.633** -log(Catholic population) (0.171) -0.957** log(beatified stock + 0.5) (0.190) -0.0052 Duration of beatified stock (0.0032) (years) -1.19** [.30] -0.20 [.82] Other Western Europe (0.20) (0.25) -2.09** [.12] -0.52 [.59] Eastern Europe (0.25) (0.50) -3.13** [.044] -0.42 [.66] Latin America (0.33) (0.62) -2.85** [.058] -0.55 [.58] North America (0.43) (0.80) -3.43** [.032] -0.82 [.44] Asia (0.49) (0.99) -3.65** [.026] -1.08 [.34] Africa (0.48) (1.09) 1.78** [5.9] 1.93** [6.9] John Paul II 1978-2005 (0.27) (0.40) 1.18** [3.3] -0.18 [.84] Benedict XVI 2005-2009 (0.31) (0.49) 5.2** 5.4* Competition Catholic(1.9) (2.3) Protestant 4.6* -3.7 Secularization variable (2.2) (3.5) 0.66* [1.9] 1.24** [3.5] pre-Westphalia dummy (0.28) (0.45) 0.034* 0.069** Age of pope, start term (0.015) (0.018) (years) 0.94 0.83 R-squared -3.33* -1.97** log(variance or shape parameter) (1.48) (0.69) 43 Notes to Table 4 Observations are by pope for the 38 popes from Urban VII (1590-1590) to Benedict XVI (2005-2013); see Table 1. The dependent variable in Table 4 is the number beatified in column 1 and the number canonized in column 2. The numbers by pope are for seven regions (based on a blessed person’s residence at death): Italy, Other Western Europe, Eastern Europe, Latin America, North America, Asia, and Africa. The count regressions are estimated by maximum likelihood in a negative-binomial form. The fitted values from this equation give the log of the expected number of persons beatified or canonized by each pope in each region. Standard errors of the estimated coefficients are in parentheses. For the six regional dummies (Italy excluded), the numbers shown in brackets, computed by exponentiation of the point estimates, correspond to proportionate effects compared to Italy. For example, in column 1, the number of persons beatified in Other Western Europe was estimated to be 0.30 that of Italy, for given values of the other explanatory variables. For the dummies for John Paul II and Benedict XVI, the numbers in brackets correspond to proportionate effects compared to earlier popes. For the pre-Westphalia dummy (which takes on the value one for the eight popes who started office between 1590 and 1623), the number in brackets gives the proportionate effect pre-Westphalia compared with post-Westphalia. The population, religious-competition, and secularization variables are described in sections II.B, II.C, and II.D of the text. The competition and secularization variables apply to the 9 popes (aside from Francis) starting since 1900 (Table 1). The stock of beatified persons (not yet canonized) is discussed in Table 1. The duration of this stock is the mean number of years from beatification to the start of a pope’s term for persons beatified but not yet canonized in each region. *Significant at 0.05 level. **Significant at 0.01 level. 44 Figure 1 Beatifications per Year (popes with 4 or more beatifications) 14 JPII Ben XVI 12 10 8 6 Pius XII 4 Pius XI 2 Paul VI John XXIII 0 233235 239246247248 250251 252254255256257258 259260261262264 265 Pope ID Number Note: Data are described in Section II.A. 45 Figure 2 Canonizations per Year (popes with 4 or more canonizations) 6 Ben XVI 5 4 JPII 3 John XXIII Pius XII 2 Pius XI Paul VI 1 0 234 239 241 243 245 246 247248 251 254 255 256 257 259 260 261 262 264 265 Pope ID Number Note: Data are described in Section II.A. 46 Figure 3 Years from Death to Beatification (popes with 4 or more beatifications) 240 200 160 120 Pius XI John XXIII JPII 80 Paul VI 40 Pius XII 0 233 235 239 246 247248 250 251 252 254 255 256 257 258 259 260 261 262 264 265 Pope ID Number yearsdeath to beatification (mean) yearsdeath to beatification (median) Note: Data are described in Section II.A. 47 Ben XVI Figure 4 Years from Beatification to Canonization (popes with 4 or more canonizations, among previously beatified) 120 100 John XXIII 80 Pius XI 60 Pius XII Paul VI JPII 40 Ben XVI 20 0 234 239 241 243 245 246 247 248 251 254 255 256 257 259 260 261 262 264 265 Pope ID Number yearsbeatification to canonization (mean) yearsbeatification to canonization (median) Note: Data are described in Section II.A. 48 Figure 5: Beatifications and Canonizations: Fractions Male, Schooled, and Urban since 1590 Beatifications 1.0 0.8 0.6 0.4 0.2 0.0 1590-1699 1700-99 1800-99 1900-49 1950-79 1980-89 1990-99 2000-12 Canonizations 1.0 0.8 0.6 0.4 0.2 0.0 1590-1699 1700-99 1800-99 1900-49 1950-79 1980-89 1990-99 2000-12 male fraction formal schooling fraction urban fraction Note: Data are described in Section II.A. 49 Figure 6 Beatifications by Region over Time 200 Total Beatified Italy Other Western Europe Eastern Europe Latin America North America Asia Africa 160 120 80 40 0 1590 1700 1800 1900 1950 1980 Starting Year for Period Note: Data are described in Section II.A. 50 1990 2000 Figure 7 Canonizations by Region over Time 90 80 70 60 50 Total Canonized Italy Western Europe Eastern Europe Latin America North America Asia Africa 40 30 20 10 0 1590 1700 1800 1900 1950 1980 Starting Year for Period Note: Data are described in Section II.A. 51 1990 2000 Figure 8 Population by Region 7,000 6,000 World Population Italy Other Western Europe Eastern Europe Latin America North America Asia Africa millions 5,000 4,000 3,000 2,000 1,000 0 1550 1650 1750 1850 1950 2009 Note: Data, described in section II.B, are based on McEvedy and Jones (1978) and World Development Indicators. 52 Figure 9 World Christian Population by Type 2,400 2,000 Total Christian affiliated population Catholic Protestant (broadly defined) Evangelical Pentecostal (by beliefs) Orthodox millions 1,600 1,200 800 400 0 1550 1650 1750 1850 1950 2009 Note: Total Christian is the sum of affiliated persons in the categories of Catholic, Protestant (broadly defined to include Anglicans, independent Christian churches, and marginal Christians such as Mormons and 7th-Day Adventists), and Orthodox. Most Evangelicals (affiliated with Evangelical churches) fall into the Protestant category. Pentecostals are defined by belief structure, rather than affiliation and are mostly a subset of Evangelicals. Data, described in section II.B, apply religious-adherence shares from Johnson (2010) and Barrett and Johnson (2001) to population data described in Figure 8. Before 1900, the ratio of Evangelical to Protestant population in each region was assumed to equal the ratio for 1900. Pentecostal numbers equal zero before 1900. 53 Figure 10 Catholic Population by Region 1,200 1,000 World Catholic Population Italy Other Western Europe Eastern Europe Latin America North America Asia Africa millions 800 600 400 200 0 1550 1650 1750 1850 1950 2009 Note: Data, described in section II.B, apply religious-adherence shares from Johnson (2010) and Barrett and Johnson (2001) to population data described in Figure 8. 54 Figure 11 Protestant Population by Region 900 800 World Protestant Population Italy Other Western Europe Eastern Europe Latin America North America Asia Africa 700 millions 600 500 400 300 200 100 0 1550 1650 1750 1850 1950 2009 Note: Data, described in section II.B, apply religious-adherence shares from Johnson (2010) and Barrett and Johnson (2001) to population data described in Figure 8. 55 Figure 12 Catholic-Protestant Competition Variable since 1900: Italy, Other Western Europe, Eastern Europe .30 .25 .20 .15 .10 .05 .00 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Italy Other Western Europe Eastern Europe Note: See Section II.C. of the text for definitions. 56 Figure 13 Catholic-Protestant Competition Variable since 1900: Latin America, North America, Asia, Africa .30 .25 .20 .15 .10 .05 .00 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Latin America Asia Note: See Section II.C. of the text for definitions. 57 North America Africa Figure 14 Secularization Variable since 1900: Italy, Other Western Europe, Eastern Europe .20 .16 .12 .08 .04 .00 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Italy Other Western Europe Eastern Europe Note: See Section II.D. of the text for definitions. 58 Figure 15 Secularization Variable since 1900: Latin America, North America, Asia, Africa .20 .16 .12 .08 .04 .00 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Latin America Asia Note: See Section II.D. of the text for definitions. 59 North America Africa
© Copyright 2026 Paperzz