Exploring Accounting Doctoral Program Decline: Variation and the Search for Antecedents

by Issues in Accounting Education
INTRODUCTION

The declining number of doctorally qualified individuals entering the accounting faculty ranks constitutes one of the most important trends of the last two decades. This condition threatens the gains made by the discipline since the time when it was dominated by full-time faculty with lesser academic credentials and a large cadre of practitioners that were ''doing a little teaching on the side.'' Any group that cannot adequately replenish its ranks with dedicated full-time initiates becomes seriously threatened by aging and retirements.

The academic and practice literature has convincingly documented the shortage of terminally qualified accounting faculty (Plumlee et al. 2006; Leslie 2008, 2009). Any person with administrative responsibilities appreciates the difficulty of finding acceptable candidates to fill faculty positions. This ''seller's market'' becomes manifest in the aggregate each year when the American Accounting Association (AAA) provides clearinghouse placement services. In this tally, the list of schools with positions routinely outstrips the available candidates. Although those postings do not describe the entirety of the labor market, the continuing imbalance suggests a structural problem.

Fewer new faculty taking positions on the tenure track create numerous and considerable consequences. Ceteris paribus, fewer people with research training reduce contributions to the knowledge base of the accounting discipline. Without strong research coming from accounting, tenure-track faculty can be differentiated less readily from lower cost instructors. As a result, the accounting discipline could lose status and critical mass within business schools (Swanson et al. 2007). As current tenure-track faculty retire, they are less likely to be replaced in kind. Even in the shorter run, fewer new faculty mean less activity, less co-authors, and less opportunity to influence the direction of business schools. Excessive reliance on adjuncts may also erode student motivation and aspiration (Jaschik 2008). That practicing accountants have jumped into the fray, with support measured in the millions of dollars (Ruffet al. 2009), should certainly attest to the gravity of the situation for the accounting profession.

This paper explores the decline of Ph.D. programs in the U.S. by first providing better descriptive information about this trend. For these purposes, a specific set of questions about this trend is posed and addressed with the available data. This paper reports that the doctoral program decline has been far from uniform and might better be described as a shiftamong its sectors, and one that took place over a relatively brief part of the last two decades. Although the prestige of schools marks the pattern of change in doctoral student output, the relationship is highly nuanced and variable. Results also indicate that larger declines were more likely to occur at the schools that had been the largest producers and at schools that have had some success with their M.B.A. programs. Extrapolating from the current population of doctoral programs fails to support the prospects for a recovery over the near horizon.

The balance of this paper comprises three sections. The first develops a number of specific research questions. For these purposes, the existence of a decline acts as a maintained hypothesis. The second section details how the research questions were addressed, including the measurement choices that were made. The next section reviews the evidence that sheds light on the general research questions that have been posed. The final section provides a discussion of issues suggested by this inquiry.

BACKGROUND AND QUESTION DEVELOPMENT

Fogarty and Markarian (2007) document the rise and fall of academic accounting between 1982 and 2002. In the last decade of this period, these authors show a relatively steep loss of full- time and tenured-track positions in accounting. They suggest that some of the decline can be attributable to the insufficiency of ''freshly minted'' Ph.D. graduates in the discipline.

While these results correspond to the lived experience of many in academic accounting, they also open new questions worthy of consideration. Interested parties certainly would want to know where the decline has been concentrated and what sectors have been more resilient. To the extent possible, the category of ''doctoral school'' should be partitioned to determine if some subsets of schools have led or have followed the curve. The overall decline that has occurred may be the result of changes that occurred in some schools, but not at others. Even if accounting doctoral programs generate positive reputational capital, they have to be understood as part of a larger set of possible priorities that business schools possess.

No central authority governs doctoral training in accountancy. If one did, the decline in production and training might not have been allowed. Instead, autonomous decisions made by administrators of accounting departments and business schools all over the country contribute to market supply. Any trend in the data must be the result of people assessing shared market conditions and acting with some degree of similarity. A threshold research question requires an investigation of the uniformity of direction of these decisions. Since an overall decline might be the composite of a slide of some programs not completely offset by the gains of others, the answer to the composition question is not self-apparent. Therefore, the issue of similarity of direction and size of trend presents a general issue, which can be captured as:

RQ1: How uniform is the decline in accounting doctoral education?

Any trend that is somewhat uniform may be the result of some programs taking the initiative and others following that direction. In any system of social stratification, more highly esteemed individuals or units tend to be emulated rather than to mimic others (Fallers 1954; Levitt and Dubner 2005). Fogarty and Markarian (2007) indicate that doctoral student production from 1982 to 2002 declined more rapidly within higher-prestige programs. This trend was somewhat offset by increased production at a group of new and lower- prestige programs. Extending this assessment requires more precise attention to the timing of change. High prestige programs may not have decreased production throughout the period, or even uniformly in parts of it. Low or middle prestige schools also may have changed their positional reaction. Several years have also elapsed since the first evaluation of the association between institutional prestige and doctoral student graduation rates. Whether or not doctoral student production after 2002 continues to reflect institutional prestige is unclear, especially through a period of system-wide instability. Therefore, the following question would be appropriate:

RQ2: Is the degree and the timing of decline of accounting doctoral student production related to the program's prestige?

The competition among business schools for favorable positioning in mass media rankings has changed much about business education (Pfeffer and Fong 2002). Focusing on M.B.A. programs, popular press renditions of who is better than who has created a recognition race that demands considerable resources from business schools (Zimmerman 2001). Reportedly, the rankings of M.B.A. programs by national media publications (e.g., Business Week, US News & World Report) have driven many business school deans to divert resources into arenas that would be noted with favor by these journalists. This expensive contest puts pressure on other less-visible business school programs, especially those that do not generate financial surpluses. A ranked M.B.A. program may be thought a more important reputational investment than doctoral education because the payback (stronger enrollments) can be measured in the short term. If accounting doctoral programs have fallen victim to this relatively recent business school focus, one would expect a more sudden and sharp decline by those schools more actively engaged in such efforts. On the other hand, since M.B.A. program rankings are not new, they might no longer exert much continuing opportunity cost influence on other business school programs. Therefore:

RQ3: Is the rate of the decline of accounting doctoral programs associated with M.B.A. program rankings?

Relatively few universities offer the doctoral degree in accounting. Other degrees, even within related subject matter such as economics, show a much wider distribution across the academy. Within the set of schools that do offer the terminal qualification in accounting, a further concentration exists. The two schools that have graduated the most students are responsible for 8.7 percent of all graduates. Adding the next six largest schools brings the cumulative total to 23.3 percent. The largest quartile of doctoral schools has generated 53.3 percent of all graduates (Hasselback 2010). Unlike many other disciplines, the production of accounting doctorates seems to be highly dependent upon a small set of decision makers. At the other end of the scale, 16 and 25 schools with this degree offering have each only produced no more than ten and twenty total graduates, respectively (Hasselback 2010). The disproportionality of size offers some obvious possibilities for the study of system-wide decline. Historical demographics of this sort suggest that total production may be highly influenced by the fortunes of a few large programs. However, declines experienced at larger schools might be compensated by increases in ''market share'' by schools that had been smaller producers. Thus, the relationship between program size and change in graduation rates should be considered with the following statement: RQ4: Has the decline in graduation at accounting doctoral programs been disproportionately experienced by the largest programs?

An under-appreciated, but highly consequential, trend in U.S. higher education is the decline of governmental funding for public schools. Whereas not too long ago, the state was paying nearly all the expenses of operation, this portion of funding has now slipped to unprecedented low points (Newfield 2008; Nichol 2008). State schools have come under increased pressure to develop alternative revenue streams. In such a difficult environment, expensive doctoral programs may be financially squeezed. In such an environment, the public interest rationales for doctoral program operations may be strained. These developments lead to the following inquiry:

RQ5: Has the decline in graduations at accounting doctoral programs been disproportionately experienced by public universities?

The last focus of this paper pertains to the projection of the future for accounting doctoral programs. In the last few years, the supply problem has been a sustained concern for more groups (e.g., the AAA, AICPA, the Big 4 firms) than ever before. This attention could be seen as an indication that the problem of inadequate doctoral students will not self-correct without intervention.

The fact that an inadequate supply of candidates has bid up the starting salaries of new accounting faculty should make this occupation more attractive in the future. Classic economic theory suggests that the prospect of higher compensation will bring forth more supply of future faculty. In other words, one could hope that a bottom has been reached and that a gradual rebuilding would progress. Alternatively, the continuation of the downward trend serves as a simpler expectation that requires no sudden turnaround.

Graduate education, long since recognized as an uncertain undertaking (Berelson 1960), might be even more problematic in accounting than in other fields. The attraction of alternative employments may increase the opportunity costs and mute the impact of possible future faculty compensation for doctoral candidates. Furthermore, schools might seek a lower number of graduate students, more or less ensuring that an inadequate number of students progress toward their terminal qualification. This general issue can be expressed as:

RQ6: Will the decline in graduations at accounting doctoral programs persist into the future, continuing to contribute to the shortage of U.S. accounting faculty?

In sum, the research questions that have been posed promise a more detailed exploration of the decline in the academic accounting discipline's training programs. The questions seek empirical evidence that helps understand what type of decline has occurred, how uniform it might be, and whether we can expect its continuation.

THE STUDY

Period

Since the purpose of the research was to illuminate the recent decline of accounting doctoral programs, a period coextensive with the decline was needed. The time frame spanning 1989 to 2008 was chosen as a 20-year period that began with the annual peak of Ph.D. student production (1989) to the last year for which full data are available (2008). Since such a lengthy period of time might not be homogeneous, total market size (graduations) was used to identify more uniform subperiods.

Measures

Most of the data from this study was obtained from the Accounting Faculty Directory 2009- 2010 (Hasselback 2010). This source contained the best available record of accounting doctoral degree completion from 1989 to 2008. This reference also contains information on graduate students currently in their program of study, data used for the last research question. The Hasselback (2010) data were supplemented by a previous edition of the directory (Hasselback 2005), which contained the best available reports of graduations between 1985 and 1989 needed to establish baseline production for the first year of the study period (1989).

Anyone familiar with doctoral education appreciates the uncertainty involved in completing a dissertation. Because the duration of a program of study is rather unpredictable, yearly graduation totals are unreliable as a metric of a school's commitment to the degree program. Multiple years summed together and averaged reduce the ''noise'' that one-year fluctuations tend to represent. Therefore, the major analysis uses moving averages over five-year periods. This interval (coincidentally approximating the current length of time needed to complete an accounting Ph.D. degree) was designed to be sufficiently long so as not to overstate the variation in the data and to focus instead on more sustained differences. Using the five-year average graduation rate (i.e., the graduations in the current year summed with the previous four years' graduations divided by five), percent change from the base year (usually 1989)1 was calculated. This served as the operationalized phenomenon of interest for the descriptive portion of this research. Percentage change, definitionally skewed by being bounded at ,100 percent but unbounded on the positive side, proved unusable. Therefore, the change in five-year average graduates from 2008 to 1989 was used for statistical tests.

Information in the Accounting Faculty Directory exists for 103 doctoral programs. However, one program (Bentley) had no graduates through 2008 and was deleted from further consideration. Of the remaining 102 programs, another four (American, Lehigh, Rensselaer, and Santa Clara) had become dormant before the first year under consideration (1989) and remained inactive for the entire subsequent 20 years. They were also eliminated. Of the remaining 98 programs, 22 had no graduates before 1989 but did have graduations before 2008 and, therefore, were included in the analysis, even though they would not have been considered doctoral programs at the interval's inception.

The prestige of accounting programs was taken from the composite index offered by Fogarty and Markarian (2007). This metric gives a prestige rank to each doctoral program based on its research productivity (quantity and quality), its positioning to indirectly influence scholarship, its ability to place its graduates, and other factors. Low ranks indicate higher prestige. Although any attempt to delineate social esteem precisely would be hard to justify, the general comparisons for which this scale is used in this research make the scale serviceable.

The possible impact of business school ranking required the consideration of whether or not a school appears in a national media ranking. This distinction was first treated as a binary variable. The Internet sites of Business Week, Financial Times, and US News & World Report were used to secure the most recent (2010) ranking information. A school that was ranked at least once by any of these sources was distinguished from the never-ranked schools. This rule assigned 41 schools a 1 (indicating ranked presence) and gave all others a 0 value. In a second measure, the number of times a school was ranked became the categorical variable. This created a scale that ranged from a 0 for the never-ranked schools (57 cases) to a maximum score of 3 for the schools ranked by all the publications (19 cases). No attempt was made to capture how highly a school was ranked.

Size of the doctoral program was measured in two ways. Using Hasselback (2005), total pre- 1989 accounting doctoral graduates by each school were captured as the primary measure. This offered a scale that ranged from a low of 0 to a high of 292. Fearing that a continuous measure with such wide ranges might inject too much variation into the analysis, this scale was reconfigured into a second measure that distinguished four size quartiles of approximately equal size.

Whether a school was public or private was a needed measure for the fifth research question. This information was gathered from Peterson's Guide to Four Year Colleges (Peterson's College Division 2002), a widely used reference book on colleges and universities in the U.S. This resulted in the assignment of 0 for the 26 private schools and 1 for all others.

RESULTS

Table 1 provides the comprehensive data about accounting doctoral student graduations. For these purposes, listings under any given year represent the average of that year and the previous four years. This approach sought to mitigate ''lumpy'' production and, therefore, to better represent general graduation trends.

RQ1

Column totals depict the general trajectory of decline in the doctoral accounting sector. The largest year-to-year change in five- year averages occurred between 1998 and 1999, and again between 1999 and 2000 when the five-year average declined by 13 graduates. The rows represent the five-year averages at any particular doctoral program and cumulate into each school's percentage change since 1989 or near its inception (see footnote 1), whichever is later. For these purposes, a zero in a cell means the lack of even a single graduate over the five-year interval that ended in that year.

Fifty-four of the 77 programs (70 percent) that had active doctoral programs in 1989 had declining five-year averages of graduates over the 20-year interval. These schools ranged from the largest possible decline of 100 percent for several schools (e.g., Rice and St. Louis) to very small negative changes, such as the 7 percent and 10 percent experienced by The University of North Carolina and University of Colorado, respectively. The downward trajectory of the bulk of the schools exists as the most apparent interpretation of the change over time for accounting doctoral programs. Although this could be predicted from the total market direction, the extent of its pervasiveness among the individual schools adds to our appreciation of the trend. Not all schools share the pattern of decline, when such a decline is defined as a comparison between the five-year moving average of 1989 and the five- year moving average of 2008. Twenty schools that had non-zero 1989 production on this metric had more average graduates in 2008. Increases tended to be small, however, with only five cases of an increase of a single graduate in the five-year average. For all the increasing schools, the average increase was 0.63 graduates per year. Appendix A lists these schools.

RQ1 needs also to consider another type of exception to the general decline of Ph.D. student production. Schools that were not producing any graduates in 1989 (indicated by a five-year average of 0.0 in Table 1) may have begun production during the 20-year interval. In fact, an impressive 21 schools fit that description. However, four of these reverted to a zero average number of graduates by 2008, indicating a rather short-lived commitment to the accounting doctorate program. As with the continuing schools that have beginning-to-end increases, the new doctoral programs that are still producing graduates did not add large numbers of new faculty to the community. More than half of these 17 programs have a five- year average of no more than one graduate. Only two new programs (Rutgers, The State University of New Jersey and The University of Texas at Dallas) exceed two graduates per year. On average, the new doctoral programs have less than 1.2 graduates per year on a five- year moving average basis. Appendix B identifies the new programs since 1989, separating out those that are not producing graduates as of 2008.

An examination of total graduate production during the 20 years in five-year moving average terms allows the identification of three distinct periods. The first occurred between 1989 and 1995. During this time, production was mostly stable around what we now identify as the market peak. During most of these years, the total number of graduates (five-year average basis) started and ended at 195 and only ranged upward to 198 in 1992. The second period (1996 to 2003) saw rapid and steady declines. The moving averages of that period saw decreases of between 7 and 13 graduates per year. This brought the yearly production to its nadir in 2003 of 116 average graduates. The third period is one characterized by a modest rebound. During the last five years (2004 to 2008), doctoral schools collectively have added between three and six graduates per year to their moving average, bringing the total market to 140 in 2008. This restoration pushes the moving average to approximately the same point it had reached in 2000. The current position is 20 percent higher than its lowest point (2003), but still 29 percent down from its peak (1992), again in terms of five-year averages. Figure 1 illustrates these intervals.

On balance, the issue posed by RQ1 needs to be answered with care. Accounting doctoral student production has reduced overall over the last 20 years. However, the data suggest that the years of this decline were limited to the middle third of this interval. We also could assert that nearly 40 percent of the 98 schools did not contribute to the decline. Some of these nonparticipators actually increased their totals while others began programs that also contributed students to the academy. Thus, the decline of accounting doctoral programs production was quite limited, both institutionally and temporally. Nonetheless, the decline that did occur tended to be substantial.

RQ2

RQ2 pertained to the relationship between program prestige and decline. For these purposes, the Fogarty and Markarian (2007)2 index was used to discern patterns within the percent change in the five- year average data presented in Table 1. Since this prestige scale identifies high prestigious programs with lower numbers (''We're number 1''), a positive correlation indicates more decline in the more highly thought of schools.

Table 2 illustrates the data that underlie the test of this research question and contains the result of the statistical tests. For the first purpose, schools are grouped into prestige quartiles. Change in the five-year average of doctoral student production was then aggregated within these tiers. The important conclusion drawn from these numbers is that prestige differences do not systematically correlate with the change in graduates. The most prestigious schools (top quartile) have smaller average losses in graduates than the middle two tiers. The most apparent prestige patterning of the data is the distinction between the lowest prestige quartile and the other tiers. These low-prestige schools were not only collectively less likely to have declines in graduates, but averaged an increase of 0.33 students. The change in direction from the loss of students in the third tier to the increase of students in the fourth tier can also be attested to by the large standard deviation for the former. An F-test statistic of 104.72 suggests that prestige is strongly significant (p , 0.01). Thus, prestige does pattern the doctoral student change data, but not in the simple linear way previously suspected by Fogarty and Markarian (2007).

The above evidence does not consider the timing of the diminished doctoral student production. For these purposes, the three sub- periods identified with the five-year average data from Table 1 were used to reconsider the relationship between prestige and change in accounting doctoral student production. Table 3 summarizes this approach that again uses prestige quartiles to aggregate the graduation change data under analysis.

Across the three periods, the data show shifting but systematic variation in the basic relationship between institutional prestige and accounting doctoral student production. During the early period of growth (1989-1995), losses in the higher prestige tiers were offset by large gains in the lower tiers. Thus, the total growth of the period was a result of the shifting of graduation volume among prestige sectors. The period of rapid decline (1996-2003) exhibited a disruption of that reallocation. Although the production of all prestige tiers was down, the first and third tiers showed more losses in graduates. However, the least prestigious group continued to experience the smallest losses. The higher prestige schools recovered more lost ground than any other sector during the 2004- 2008 period. However, continuing strength in the low prestige group prevented the change in average graduates during the last period from being continuous. All F-statistics for sub-periods reported in Table 3 are significant at p , 0.01, indicating the viability of prestige grouping.

In sum, the evidence on RQ2 indicates a highly nuanced relationship between the extent of doctoral student production and institutional prestige. The belief that high prestige programs were more likely to reduce their Ph.D. programs in accounting may have been true in earlier time periods, but does not really describe the last 20 years. Nonetheless, prestige as an important patterning factor remains apparent, albeit in a non-linear way. For these purposes, the emergence of the lower prestige strata as a counter- balance against more general decline has been the most important development.

RQ3

This research question sought to identify whether schools that had gained some degree of success in the M.B.A. program rankings experienced greater declines in their accounting Ph.D. programs. The two measures used to capture the former allowed a comparison of schools that had achieved some degree of M.B.A. recognition with those that had not, and then to consider relative degrees of M.B.A. program success.

The statistical evidence for both tests of this research question is detailed in Table 4. Panel A shows that the t-value produced by the test of the relationship between the change in doctoral production (using data from Table 1) and the binary M.B.A. ranking score is 0.36, which is not statistically significant (p . 0.05). The Panel A results can be clarified when the alternative measure of M.B.A. ranking success is used. Schools that are regularly recognized for their M.B.A. programs show lower accounting Ph.D. program reductions than schools that are either not ranked or infrequently ranked. This categorical M.B.A. ranking profile variable has an F-statistic of 39.67, significant at p , 0.01. Thus, the Table 4, Panel B results suggest that the most frequently ranked schools are more dissimilar to all other tiers in terms of changes in their accounting doctoral graduation rates. These schools show net increases while all other tiers show losses. Interestingly, schools that are never ranked appear to be reducing their accounting doctoral programs more slowly than schools that are sometimes (but not always) ranked for their M.B.A. programs. The downturn in accounting doctoral programs is most extreme among schools that are sometimes, but not always, ranked for their M.B.A. programs by the popular media. In this more subtle way, the decline of Ph.D. programs in accounting could be said to be intertwined with the struggle for M.B.A. program recognition by institutions of higher education in the U.S.

The RQ3 results were also analyzed on a sub-interval basis. In results not shown, the categorical M.B.A. recognition variable produced an F-statistic that is significant at the p , 0.01 level in all sub-periods. Again, the never-ranked and the always-ranked schools show results similar to each other and distinct from the two groups of sometimes-ranked schools. This result proves particularly strong in the earliest (1989-1996) period.

RQ4

The next research question considered the impact of the program size. The existence of a few disproportionately large accounting doctoral programs necessitated the consideration of the shifting distribution of ''market-share'' among programs. When measured as the number of total pre-1989 graduates from the program, size proved negatively correlated (p , 0.01) with the five-year average change in the number of graduates.

In order to test the robustness of the size effect, Table 5, Panel A groups doctoral programs into size quartiles. Again, size is strongly related to change in doctoral program (F 1/4487.45, p < 0.01). Both the average graduate change and the median graduate change show monotonic variation from the smallest quartile to larger ones. Losses in the largest tier are disproportionate, exceeding an average of two graduates per program. The smallest school size tier shows graduate increases.

Panel B of Table 5 conducts a similar analysis for the sub- intervals that have been identified by the Table 1 moving average totals. Generally, the patterning of changes in graduations by institutional size holds true in the first two of the shorter periods. During these intervals, the largest schools experienced the largest declines, and the smallest schools grew or resisted declines. However, stronger rebound growth existed at the largest and second largest tiers of schools than the second-smallest tier schools in 2004-2008. F-statistics from all three time intervals are significant at p , 0.01.

RQ5

The fifth research question pertained to the public/private classification of the doctoral schools. For these purposes, reasons exist to believe that doctoral student production in accounting relates to how those programs are financed. The result of a simple correlation between the binary distinction between public and private schools and the 1989-2008 change in doctoral student production produced a t-statistic of 15.21, significant at p , 0.01. Public schools reduced the total amount of their accounting doctoral production much more substantially than private schools. In fact, private schools show very small increases. Table 6, Panel A provides more statistical information about this test.

Table 6, Panel B considers the public-private school distinction across the three sub-periods. Public schools did not have larger losses of accounting doctoral graduations in each of the three sub- intervals. Such a conclusion proved true only in the middle period, as indicated by the t-statistic of 14.72 (significant at p , 0.01) for the 1996-2003 time interval. In other intervals, private schools grew slightly (but not significantly) more than public schools. The t-statistic for the time periods other than that of great decline (1996-2003) proved insignificant.

RQ6

The last research question attempts to shed light on the future of doctoral student production in accounting. The continuation of declines was cast into doubt by the aggregate data that show that the 2004-2008 period represents an improvement in production over the previous sub-sections of the 20-year interval (see Table 1). This empirical fact adds some credence to the theoretical logic that markets tend to be self-correcting. However, even if strong downward trends do not inevitably continue, the magnitude of their reversal remains unclear.

For projection purposes, current doctoral program students reported by Hasselback (2010) were assumed to represent an adequate measure of the doctoral cohort proceeding toward graduation. The length of the typical accounting doctoral program makes it unlikely that anyone not currently in a doctoral program would be in a graduate cohort over the next few years. Thus, the number of current graduate students enables a ''best case scenario'' to be confidently projected. The 828 existing doctoral students could be expected to yield 166 new faculty if one-fifth of them graduate each year and 207 new faculty if one-quarter of them finish annually.3 Either of these results would elevate the moving average totals reported on Table 1.

Table 7 provides an effort to derive a more realistic forecast of new terminally qualified Ph.D.s in accounting. After grossing up the total to estimate a small element of missing data,4 the graduate student population was discounted by an estimate of attrition. Doctoral students abandon, or are asked to leave, their pursuit of a degree for a variety of reasons. Compounding the losses due to inadequate fit and quality control are those attributable to post- graduation departure from the U.S. U.S. doctoral programs in accounting have become very attractive to foreign students, many of whom take positions in their home country following their studies. These departees may contribute significantly to the scholarship of the discipline, but make themselves unavailable to fulfill other aspects of the academic role within the U.S. and, therefore, cannot be considered part of the doctoral shortage solution. Their separate consideration responds to that part of RQ6 that pertains to the faculty consequences of doctoral student shortages.

Estimates of future attrition were derived by making estimates of past attrition. For these purposes, comparisons between the rosters of the AAA Doctoral Student Composition and the Accounting Faculty Directory were conducted. The years 2000-2005 of the former were traced into the 2010 Directory in the belief that such an interval would be sufficient to enable degrees to be completed and faculty positions assumed. In that the Doctoral Consortium representatives should be an assembly of the ''best and brightest'' of a year's cohort, and of students that are well into their doctoral studies, attrition from such a group creates a conservative estimation of overall attrition. The average loss of over 13 percent, as shown in Table 7, is higher than many would have expected. Table 7 also estimates the emigration of U.S. doctoral graduates at over 14 percent. This leaves 622 people likely to join the U.S. academy in accounting. If these students all graduate and took positions within the U.S. academy over the next four years, 156 graduates per year would be produced, a rate sufficient to modestly raise the five- year moving average. However, if one were to assume a five-year training period, only 124 graduates a year can be expected. This total would return the discipline closer to its low point of doctoral student production, reversing most of the gains of the 2004- 2008 interval and changing the slope of the last segment of the line shown in Figure 1.

For the 96 schools for which data exist, 76 report more current doctoral students than they have had graduates in the last five years. This suggests also that only 22 schools will see their populations of future graduates necessarily dwindle below what now could be defined as that school's new normal level. However, as attested to by Table 7, the conversion of doctoral students into faculty members remains problematic even with adequate numbers of people engaged in the process.

DISCUSSION

Although to suggest that a decline has occurred in the doctoral segment of accounting education might belabor the obvious, this research undertook the task of describing the phenomenon with greater precision. The current research has also sought explanations for the patterns formed by the trends of five-year moving averages. This metric smoothes out what would otherwise be a discontinuous production outcome in individual years by individual schools and reveals a more systematic way to describe commitment to the disciplinary area.

This project shows that the accounting doctoral program decline, although pervasive and sharp, is hardly comprehensive or uniform. At one extreme, some programs have declined precipitously over a 20- year period. At the other end, the years were marked by many schools that began doctoral programs in the discipline. Although some decline is shared by the majority of schools, the absolute importance of the magnitude of the decline experienced by a few schools cannot be denied. These changes were sufficiently consequential to overwhelm the small increases in accounting doctoral student production at many continuing and new programs.

The use of five-year moving averages enables the identification of three distinct sub-intervals of the 20-year period that would not have been as apparent if the annual totals of graduates had been used. These shorter periods enable greater precision in the consideration of the nature of the decline. The data suggest three distinct sub-periods that often varied the importance of the influences that were considered.

Following suggestions in the literature, the behavior of doctoral programs was parsed according to their institutional prestige. High prestige accounting programs had compressed the number of their graduates in the 1980s prior to the period under study (Fogarty and Markarian 2007). The current research found that a different sort of prestige-based change continued to occur over the next two decades. Larger declines in the middle prestige tiers were more characteristic of the period under study (1989-2008), as if these schools emulated the ''quality not quantity'' strategy seen in earlier times at the top tier of schools. At the same time, lower prestige programs pursued a much different strategy. In the low prestige sector, actual increases (or at least smaller reductions) in graduates can be seen over the time period.

The logic of this research's prestige finding may rely upon the fact that the reputations of some schools are more secure than others. The school that believes itself to be highly esteemed can impose severe restrictions on the number of future graduates without fear of reputational injury. The schools that are still looking to establish their reputation appear more likely to pursue higher prestige by increasing their numerical presence in the doctoral sector. The results mitigate our belief in the importance of institutional prestige as an unvarying force in academic accounting. This source of variation cannot be considered apart from the general tenor of the times. When the market is in a freefall state (as it was between 1997 and 2003), prestige does not create the usual differentiation that is does in more bountiful times. During these years, declines in graduation occurred in all prestige tiers, even if they were still less extreme in the lower one. The basic prestige rationale also did not anticipate the stronger rebuilding that has occurred since 2004 in the highest prestige tier. In short, prestige differentiation offers systematic but not monotonic variation in accounting doctoral programs.

This research showed that the decline of accounting doctoral education occurred with greater magnitude at the large cohort doctoral programs. When large producing schools decide not to continue in that capacity, smaller producing schools collectively have the theoretic capacity to compensate. However, over the 20 years considered in this study, the small producers added to the problem by also becoming just a bit smaller in the aggregate. In retrospect, the doctoral student shortage has been a result of the excessively concentrated supply side that leftthe discipline vulnerable to a relatively few decision makers.

That some programs exist at public universities and others are contained within private institutions grounds explanations in available resources. This dichotomy puts the recent diminishment of public funding into the picture. This trend suggests that public schools may have faced consequential financial pressure over a component of the time period studied. Such exigency may have translated into lesser resources for doctoral programs that have not increased in more recent times. On the other hand, public schools may be more obligated to respond to the educational needs of the state's population, a constraint that may continue the production of accounting academics at previous levels. The results bear out the more prominent declines at public schools. In addition to the much steeper declines of the 1997-2003 period, public schools were unable to outpace private schools in the rebound era of 2004-2008. Over the longer time frame, economic pressures appear to have trumped the need to respond to the system needs for more terminally qualified faculty in the accounting area.

The paper has asserted that doctoral programs in accountancy should be placed within the context of the objectives of the modern business school. In such an environment, the prioritizing of some programs and the career aspirations of some students might result in the relative neglect of others. The media attention given to M.B.A. programs in recent years suggested the prospect that some schools may have diverted resources in this area, and this may have occurred to the detriment of accounting Ph.D. programs. The results suggest that such a prioritization has been more likely at schools less certain about their M.B.A. program rankings. This suggests that schools that are ''in the hunt'' for broader M.B.A. program recognition act differently than those who have less hope for such and those who can take such nomination for granted.

The future of doctoral education in accounting cannot be known with perfect clarity. The projection of the future production of accounting doctoral students was made more difficult by the rebound of total graduates during the last few years (2004-2008) of the entire period under study (1989-2008). On balance, strong optimism about the continuation of this turnaround does not seem warranted. This research sought to use data about the population of doctoral students to estimate future faculty candidates that are likely to be part of the U.S. academy. This effort needs to recognize the rather lengthy production cycle, whereby graduate students take many years to supplement the total of faculty members. The forecast offered in this paper suggests that, after losses to attrition and emigration are considered, a material increase in the ranks of ''freshly minted'' terminally qualified U.S. professors is unlikely. The fact that a sizable contingent of doctoral graduates now is employed outside the U.S. suggests that future research needs to separate the doctoral student production issue from the U.S. faculty shortage problem.

Several of the research questions of this paper are grounded in the economics of accounting doctoral student production. Unlike most other programs of business schools, doctoral education increases in cost with the size of the programs. This, combined with the absence of revenues, allows us to suspect that the doctoral student shortage is an institutional reaction to program costs. This was approached in several ways, such as the suspicion that public schools were under greater financial pressure (RQ5) and that larger programs had overcommitted in this area and would, therefore, retrench production (RQ 4). However, the focus on costs does not provide sufficient answers for the U.S. doctoral market as a whole. As RQ3 indicates, school priorities tend to make costs into valuable investments that schools are willing to incur. Furthermore, the RQ2 results suggest that lower prestige schools see doctoral programs as viable ways to enhance their relative position.

Limitations

The paper needed to use imperfect measures, especially as it pertained to the rationales that exist for schools to diminish the size of their accounting doctoral program. When attempting to ascribe motives based on activity levels, researchers would prefer access to the cognitive processes of decision makers. However plausible archival measures might be, they only provide inferential substitutes.

This paper charts the course of accounting doctoral student production from 1989 until 2008. Because this particular 20-year period possesses no natural uniqueness, a temporal limitation exists. The paper could have used a longer period (beginning before 1989) or a shorter one (beginning after 1989). The former would have appended an era of impressive ascendency that has now been compromised. The latter would have added several early years during which the discipline's doctoral student production was strong and steady and arguably closer to the market demand. In either event, the downward trajectory of the discipline's new scholar volume that was concentrated around the millennium would have remained the primary focus of our attention.

No attempt has been made in this paper to partition doctoral student production by the subfields of the accounting discipline. Previous studies of the shortage commissioned by the American Accounting Association have shown that ample graduate students that specialize in financial accounting have been available to U.S. schools over the last few years (Leslie 2009), while critical shortages plague areas such as audit and tax (AAA/APLG/FSA 2008). The extreme shortage areas merit more focused study in the future.

The efforts of this research to predict the future necessarily involve more limitations than that of the descriptions of the past. That no extraordinary exogenous shocks to the system occur that would materially alter the level of production serves as a critical assumption. Changes to the basic model of accounting doctoral education could also alter the number of years that are needed. This research only endeavored to estimate the next few years, making changes of this nature less likely.5

The data of this paper limited its ambition to the academic market in the U.S. Data on the number of doctorate degree graduates outside the U.S. were not available. Moreover, institutional prestige tends to be constrained by cultural parameters. The U.S. restriction might be more problematic if U.S. schools routinely recruited faculty from non-U.S. doctoral schools. Nonetheless, the U.S.-centric nature of this research question should be taken as a study limitation.

Copyright American Accounting Association May 2012

(c) 2012 Issues in Accounting Education. Provided by ProQuest LLC. All rights Reserved.

Related Articles