0712.2820/ms.tex
1: \documentclass{emulateapj}
2: \usepackage{apjfonts}
3: 
4: \begin{document}
5: \slugcomment{Publications of the Astronomical Society of the Pacific, ACCEPTED}
6: 
7: \shorttitle{THE PRODUCTION AND EMPLOYMENT OF ASTRONOMERS}
8: \shortauthors{METCALFE}
9:  
10: \title{The Production Rate and Employment of Ph.D.\ Astronomers}
11: 
12: \author{Travis~S. Metcalfe}
13: 
14: \affil{High Altitude Observatory and Scientific Computing Division,\\ 
15: National Center for Atmospheric Research, P.O.\ Box 3000, Boulder, CO 
16: 80307 USA}
17: \email{travis@hao.ucar.edu}
18: 
19: \begin{abstract}
20: In an effort to encourage self-regulation of the astronomy job market, 
21: I examine the supply of, and demand for, astronomers over time. On the 
22: supply side, I document the production rate of Ph.D.\ astronomers from 
23: 1970 to 2006 using the UMI Dissertation Abstracts database, along with 
24: data from other independent sources. I compare the long-term trends in 
25: Ph.D.\ production with federal astronomy research funding over the same 
26: time period, and I demonstrate that additional funding is correlated with 
27: higher subsequent Ph.D.\ production. On the demand side, I monitor the 
28: changing patterns of employment using statistics about the number and 
29: types of jobs advertised in the AAS Job Register from 1984 to 2006. 
30: Finally, I assess the sustainability of the job market by normalizing this 
31: demand by the annual Ph.D.\ production. The most recent data suggest that 
32: there are now annual advertisements for about one postdoctoral job, half a 
33: faculty job, and half a research/support position for every new domestic 
34: Ph.D.\ recipient in astronomy and astrophysics. The average new astronomer 
35: might expect to hold up to 3 jobs before finding a steady position.
36: \end{abstract}
37: 
38: \keywords{sociology of astronomy}
39: 
40: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
41: \section{Background}
42: 
43: From 1982 to 1992, federal funding for astronomy research increased by 
44: more than 80\%, fueled primarily by NASA funding related to the Hubble 
45: Space Telescope. Trailing this trend by several years, the annual 
46: production of new Ph.D.\ astronomers more than doubled. Meanwhile, the 
47: total number of jobs advertised in the American Astronomical Society (AAS) 
48: Job Register remained fairly constant. Responsible undergraduate programs 
49: began to lecture incoming astronomy majors about this imbalance in the job 
50: market, comparing the statistical odds of a long-term career in astronomy 
51: to the chances of becoming a professional athlete. The faculty in some 
52: graduate programs started to debate the idea of limiting the number of 
53: incoming students (``birth control''), and they made efforts to track and 
54: publicize the long-term career progress of their Ph.D.\ recipients (e.g., 
55: see Dinerstein 1996).
56: 
57: It was in this atmosphere that Thronson (1991) devised a model to describe 
58: the surplus production of new astronomers. He observed that overproduction 
59: appears to be built into the system, making the mathematical formulation 
60: of the problem similar to that of industrial pollution---an unintended 
61: side effect of the process. He calculated the future astronomer surplus 
62: under a variety of possible circumstances, including expected retirement 
63: rates and increased research funding. The most likely of the scenarios 
64: predicted a short-lived reduction in the astronomer surplus over a decade 
65: or so, followed by continued growth in the overproduction rate.
66: 
67: A few years later, Harris (1994) identified some reliable sources of data 
68: to evaluate the astronomy job market over time. He looked at the number of 
69: dissertations published in the subject area of astronomy and astrophysics 
70: using the UMI Dissertation Abstracts database (UMI) to track new Ph.D.\ 
71: recipients. He noted that the American Institute of Physics (AIP) 
72: tabulates the number of degrees awarded by astronomy departments annually, 
73: but astronomers who receive their degrees from physics departments have 
74: not been tracked until the past few years. To quantify the number of 
75: long-term jobs available each year, Harris examined advertisements in the 
76: AAS Job Register. This has some inherent limitations, but it is still the 
77: best source available.
78: 
79: Thanks to the Internet, most of these data sources are now much more 
80: accessible, and we can easily examine the trends in Ph.D.\ production and 
81: job availability over an extended time baseline. Further data on the 
82: trends in astronomy research funding are also available, allowing us to 
83: investigate the possible connection with Ph.D.\ production predicted by 
84: Thronson (1991). In section~\ref{DATA}, I document the sources of these 
85: data for future reference and I present tables that can be updated 
86: annually as new statistics become available. I examine the historical 
87: trends of both the supply and demand in section~\ref{RESULTS}, and I 
88: discuss the results in section~\ref{DISC}.
89: 
90: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
91: \section{Data Sources\label{DATA}}
92: 
93: The Statistical Research Center of the AIP has been surveying every 
94: astronomy Ph.D.\ program in the U.S.\ since 1972. An archive of their 
95: annual reports beginning in 1980 is available online\footnote{ 
96: http://aip.org/statistics/trends/archives/astrorost.htm}, and the most 
97: recent ``Roster of Astronomy Departments'' (Nicholson \& Mulvey 2007) 
98: appears on a separate 
99: page\footnote{http://aip.org/statistics/trends/gradtrends.html}. I 
100: received the complete set of data on astronomy Ph.D.\ production from AIP 
101: staff (P.~Mulvey, personal communication), which is listed under the 
102: heading ``AIP'' in Table~\ref{tab1}. These totals do not include degrees 
103: that were conferred by physics departments to Ph.D.\ recipients who 
104: complete a dissertation in a subject related to astrophysics, even though 
105: this is the most popular subfield among domestic first year physics 
106: graduate students (Mulvey \& Tesfaye 2006).
107: 
108: % TABLE 1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
109: \begin{table}[t]
110: \begin{center}
111: \caption{Ph.D.\ Production and Research Funding.\label{tab1}}
112: \begin{tabular}{lrrrrrr}
113: \hline\hline
114:  & \multicolumn{3}{c}{Ph.D.s} & 
115: \multicolumn{3}{c}{Research Funding (M\$)} \\
116: Year & AIP    & SED    & UMI & Total   & NASA    & NSF     \\
117: \hline
118: 1970 &$\cdots$& 111    & 141 & 136.455 & 100.875 &  19.922 \\
119: 1971 &$\cdots$& 113    & 119 & 126.852 &  90.688 &  23.236 \\
120: 1972 &  75    & 129    & 173 & 136.405 &  89.859 &  25.608 \\
121: 1973 &  80    & 131    & 147 & 123.085 &  82.020 &  27.080 \\
122: 1974 &  97    & 133    & 172 & 136.030 &  91.564 &  30.283 \\
123: 1975 &  76    & 131    & 162 & 168.319 & 122.002 &  29.093 \\
124: 1976 &  93    & 150    & 186 & 163.093 & 116.575 &  31.754 \\
125: 1977 &  96    & 120    & 164 & 197.305 & 140.702 &  41.585 \\
126: 1978 &  93    & 138    & 173 & 213.601 & 148.603 &  46.614 \\
127: 1979 &  90    & 115    & 152 & 284.429 & 213.272 &  51.064 \\
128: 1980 & 106    & 121    & 133 & 285.671 & 229.724 &  34.512 \\
129: 1981 &  82    & 109    & 144 & 281.027 & 198.674 &  59.146 \\
130: 1982 &  70    & 102    & 121 & 275.784 & 184.630 &  65.411 \\
131: 1983 &  81    & 115    & 129 & 357.416 & 265.687 &  64.638 \\
132: 1984 &  74    &  98    & 120 & 382.292 & 277.013 &  77.534 \\
133: 1985 &  66    & 100    & 150 & 414.743 & 310.668 &  76.554 \\
134: 1986 &  86    & 109    & 132 & 468.160 & 362.547 &  77.808 \\
135: 1987 &  72    & 100    & 166 & 522.187 & 410.651 &  76.789 \\
136: 1988 &  94    & 130    & 171 & 470.838 & 364.730 &  77.638 \\
137: 1989 &  94    & 113    & 223 & 541.939 & 428.963 &  83.373 \\
138: 1990 &  89    & 128    & 240 & 597.154 & 462.320 & 104.613 \\
139: 1991 &  73    & 125    & 233 & 631.657 & 501.766 & 101.712 \\
140: 1992 &  93    & 134    & 254 & 739.057 & 584.504 & 116.126 \\
141: 1993 & 119    & 145    & 256 & 686.281 & 520.184 & 109.711 \\
142: 1994 & 117    & 144    & 271 & 747.590 & 562.962 & 122.065 \\
143: 1995 & 133    & 173    & 321 & 764.458 & 593.089 & 126.225 \\
144: 1996 & 126    & 192    & 303 & 728.958 & 570.450 & 119.946 \\
145: 1997 & 117    & 198    & 320 & 774.740 & 621.324 & 115.212 \\
146: 1998 & 116    & 206    & 310 & 732.150 & 575.093 & 113.248 \\
147: 1999 &  88    & 159    & 267 & 757.859 & 578.973 & 124.395 \\
148: 2000 & 139    & 185    & 291 & 880.267 & 710.006 & 130.702 \\
149: 2001 & 101    & 186    & 253 & 759.195 & 559.539 & 153.919 \\
150: 2002 & 102    & 141    & 251 & 751.527 & 540.243 & 167.921 \\
151: 2003 &  88    & 167    & 254 & 871.757 & 639.761 & 192.179 \\
152: 2004 & 116    & 165    & 297 & 920.763 & 678.304 & 202.092 \\
153: 2005 &  91    & 186    & 269 &$\cdots$ &$\cdots$ &$\cdots$ \\
154: 2006 & 119    &$\cdots$& 284 &$\cdots$ &$\cdots$ &$\cdots$ \\
155: \hline\hline
156: \end{tabular}
157: \tablecomments{Degree data for astronomy departments from the
158: American Institute of Physics (AIP), for astronomy and physics
159: departments from the Survey of Earned Doctorates (SED), and for
160: all astrophysics related dissertations from the UMI Dissertation
161: Abstracts database (UMI). Funding data for all R\&D (Total) with
162: the contributions from NASA and the NSF, reported as actual
163: dollar amounts (not adjusted for inflation).}
164: \end{center}
165: \end{table}
166: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
167: 
168: The NSF Division of Science Resources Statistics has documented the annual 
169: number of astronomy and astrophysics Ph.D.\ recipients through the Survey 
170: of Earned Doctorates (SED), a voluntary program administered by graduate 
171: deans at accredited universities. Over the past decade, more than 90\% of 
172: new doctorates have participated in this survey---at many institutions, it 
173: is now a requirement for graduation. Historical data from this survey were 
174: compiled in a recent NSF report (Thurgood et al.~2006), while data from 
175: the past decade appeared in a separate publication (Hill 2007). The data 
176: from 1970 to 2005, which include degrees conferred by both astronomy and 
177: physics departments, are listed under the heading ``SED'' in 
178: Table~\ref{tab1}.
179: 
180: % TABLE 2 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
181: \begin{table}[t]
182: \begin{center}
183: \caption{Number of AAS Job Register Ads.\label{tab2}}
184: \begin{tabular}{lrrrrrrrc}
185: \hline\hline
186: Year & All & PV     & TT     & NT     & R      & RS     & MO     & F      \\
187: \hline
188: 1984 & 275 &  77    &  66    &$\cdots$&  72    &$\cdots$&$\cdots$&$\cdots$\\ 
189: 1985 & 338 & 149    &  81    &$\cdots$& 118    &$\cdots$&$\cdots$&$\cdots$\\ 
190: 1986 & 257 &$\cdots$&$\cdots$&$\cdots$&$\cdots$&$\cdots$&$\cdots$&$\cdots$\\ 
191: 1987 & 240 &$\cdots$&$\cdots$&$\cdots$&$\cdots$&$\cdots$&$\cdots$&$\cdots$\\ 
192: 1988 & 321 & 157    &$\cdots$&$\cdots$&$\cdots$&$\cdots$&$\cdots$&$\cdots$\\ 
193: 1989 & 351 &$\cdots$&$\cdots$&$\cdots$&$\cdots$&$\cdots$&$\cdots$&$\cdots$\\ 
194: 1990 & 240 & 120    &  32    &$\cdots$&$\cdots$&$\cdots$&$\cdots$&$\cdots$\\ 
195: 1991 & 304 & 137    &  30    &$\cdots$&$\cdots$&$\cdots$&$\cdots$&$\cdots$\\ 
196: 1992 & 327 & 127    &  40    & 13     &  63    & 57     &  21    & 0.140  \\
197: 1993 & 315 &  66    &  59    & 12     &  42    & 60     &  23    & 0.160  \\ 
198: 1994 & 318 & 140    &  44    & 15     &  14    & 76     &  24    & 0.233  \\ 
199: 1995 & 425 & 200    &  54    & 23     &  73    & 52     &  17    & 0.268  \\ 
200: 1996 & 426 & 210    &  63    & 15     &  65    & 56     &  15    & 0.211  \\ 
201: 1997 & 469 & 208    &  74    &  9     &  55    & 49     &  65    & 0.230  \\ 
202: 1998 & 569 & 224    &  77    & 23     & 112    & 68     &  51    & 0.210  \\ 
203: 1999 & 623 & 206    &  94    & 43     & 130    & 38     &  67    & 0.230  \\ 
204: 2000 & 739 & 277    & 108    & 39     & 132    & 66     &  63    & 0.264  \\ 
205: 2001 & 715 & 310    &  93    & 59     &  93    & 56     &  78    & 0.288  \\ 
206: 2002 & 716 & 370    & 117    & 29     & 109    &  3     &  46    & 0.198  \\ 
207: 2003 & 722 & 389    & 119    & 22     &  78    & 29     &  85    & 0.072  \\ 
208: 2004 & 714 & 333    & 113    & 22     & 117    & 28     & 101    & 0.251  \\ 
209: 2005 & 673 & 334    & 111    & 14     &  88    & 19     & 102    & 0.276  \\ 
210: 2006 & 738 & 388    & 129    & 22     &  87    & 19     &  93    & 0.276  \\
211: \hline\hline
212: \end{tabular}
213: \tablecomments{The total number of AAS Job Register ads (All) along with
214: the number of Postdoctoral and Visitor (PV), Tenure-Track (TT), Non-Tenure
215: track (NT), Research (R), Research Support (RS), Management and Other (MO)
216: positions advertised, and the fraction of ads for Foreign (F) jobs.}
217: \end{center}
218: \end{table}
219: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
220: 
221: Although the SED data represent a substantially complete survey, they do 
222: not include dissertations that are relevant to astronomy and astrophysics 
223: but were completed in another primary field. Following Harris (1994), I 
224: used the online version of the UMI Dissertation Abstracts 
225: database\footnote{http://proquest.umi.com/login} to obtain an independent 
226: estimate of the number of dissertations related to astronomy and 
227: astrophysics that were published annually from 1970 to 2006. For each 
228: calendar year, I queried the ``Dissertations \& Theses'' database for 
229: doctoral dissertations with a subject code of astronomy or astrophysics 
230: that were written in English and awarded a Ph.D.\ by a domestic 
231: university. The results are shown in Table~\ref{tab1} under the heading 
232: ``UMI''. Since these queries included dissertations with primary subject 
233: codes in other fields (e.g., mathematics or chemistry) the net effect is 
234: to slightly inflate the supply of Ph.D.\ recipients, leading to a worst 
235: case scenario for job seekers. While the normalization of job market 
236: numbers to assess sustainability may be sensitive to the choice of degree 
237: data, the overall {\it trends} should be similar.
238: 
239: In addition to degree data, the NSF Division of Science Resources 
240: Statistics also documents the trends in federal funding for research and 
241: development. One of their recent reports provides a breakdown of federal 
242: funding from 1970 to 2003 by both agency and the detailed field of science 
243: and engineering (Meeks 2004). I obtained data for the most recent years 
244: from several additional NSF reports (Meeks 2005, 2006; Jankowski 2007). 
245: Table~\ref{tab1} includes the total research funding, along with the 
246: amounts contributed by NASA and the NSF, reported as the actual dollar 
247: values (in millions) for each year\footnote{The specific programs included 
248: in these totals were determined by each agency in their response to the 
249: annual Survey of Federal Funds for Research and Development}. To correct 
250: these values for inflation, I used the composite outlay deflators for 
251: ``other grants'' from the FY\,2006 U.S.\ 
252: budget\footnote{http://gpoaccess.gov/usbudget/fy06/sheets/hist10z1.xls}, 
253: which normalizes all values to constant FY\,2000 dollars. Unfortunately, 
254: even the historical values of the inflation indices adopted by the 
255: government appear to change slightly from one budget to the next. So, any 
256: future updates to the research funding data in Table~\ref{tab1} should 
257: re-adjust the historical numbers using deflators from the most recent 
258: federal budget.
259: 
260: % FIGURE 1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 
261: \begin{figure}[t]
262: \includegraphics[height=\linewidth, angle=-90]{f1.eps}
263: \caption{Left axis: Total federal funding for astronomy research and 
264: development in millions of FY\,2000 dollars (upper solid line), including 
265: the annual totals from NASA (upper dashed line) and the NSF (lower dotted 
266: line). Right axis: Historical Ph.D.\ production from AIP surveys of 
267: astronomy departments (upper dotted line), from the NSF Survey of Earned 
268: Doctorates (SED; lower dashed line) and from all dissertations included in 
269: the UMI Dissertation Abstracts database and categorized in the subject 
270: area of astronomy and astrophysics (lower solid line).\label{fig1}}
271: \end{figure}
272: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
273: 
274: The total number of jobs advertised in the Job Register from 1984 to 2006 
275: is available from the AAS Annual Reports, published each year in the BAAS 
276: (e.g., Marvel 2007). A detailed breakdown by advertiser-specified job type 
277: has been compiled consistently since 1992, including the following 
278: categories: postdoctoral and visitor (PV), tenure-track (TT), 
279: non-tenure-track (NT), research (R), research support (RS), management and 
280: other (MO). The available data for each of these categories are listed in 
281: Table~\ref{tab2}. Note that the total number of jobs advertised in each 
282: year (shown under the heading ``All'') is not exactly equal to the sum of 
283: the six categories listed. The total also includes jobs categorized as 
284: predoctoral, which are omitted from Table~\ref{tab2} because they are not 
285: relevant to this study. Statistics about the type of institution offering 
286: the jobs are also compiled, including whether it is in a foreign country. 
287: To evaluate this contribution to the demand for astronomers, the final 
288: column of Table~\ref{tab2} lists the fraction of jobs that were classified 
289: as foreign (F). The primary limitation of the Job Register data is that 
290: they provide no way of evaluating which advertised positions are {\it 
291: new}, i.e. which jobs arise from retirement or death as opposed to faculty 
292: shuffle. This limitation is generally not relevant to shorter-term 
293: postdoctoral jobs, and assuming a reasonably constant number of faculty 
294: shuffles over time it will affect the normalization but not the trends in 
295: longer-term positions. This may partially offset the impact of adopting 
296: the slightly inflated degree data from UMI.
297: 
298: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
299: 
300: \section{Results\label{RESULTS}}
301: 
302: Using these data, drawn from reliable and accessible sources that are
303: likely to be updated annually, we can evaluate the current state of the 
304: astronomy job market and how it has changed over time. 
305: 
306: % FIGURE 2 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
307: \begin{figure}[t]
308: \includegraphics[height=\linewidth, angle=-90]{f2.eps}
309: \caption{The correlation between inflation-adjusted federal research 
310: funding and Ph.D.\ production related to astronomy and astrophysics 4 
311: years later. The best linear fit (solid line) suggests that an increase of 
312: \$100 million in federal funding ultimately leads to the production of 
313: 40-50 additional Ph.D.\ recipients.\label{fig2}}
314: \end{figure}
315: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
316: 
317: \subsection{Supply of Astronomers}
318: 
319: The data from Table~\ref{tab1} on the supply of astronomers, and the 
320: inflation-corrected federal research funding that might be one of its 
321: primary drivers, are shown in Figure~\ref{fig1}. The left axis shows 
322: funding levels in millions of FY\,2000 dollars for the total federal 
323: investment (upper solid line), and for the contributions from NASA (upper 
324: dashed line) and the NSF (lower dotted line). The most striking thing 
325: about these data is how NASA dominates the total federal funding for 
326: astronomy research, and how it is responsible for nearly all of the 
327: volatility. By contrast, astronomy funding from the NSF has shown fairly 
328: steady growth, nearly doubling its real value over the last 35 years. 
329: Between 1982 and 1992, NASA funding for astronomy more than doubled. This 
330: expansion was driven primarily by preparations for the Hubble Space 
331: Telescope, but also by the Compton Gamma Ray Observatory and other 
332: missions (E.~Suckow, personal communication).
333: 
334: The right axis of Figure~\ref{fig1} shows the total number of Ph.D.\ 
335: recipients each year in astronomy and astrophysics from UMI (lower solid 
336: line) and SED (lower dashed line), as well as the subset of those coming 
337: from pure astronomy departments (upper dotted line). The correlation 
338: between the trend in Ph.D.\ recipients and the trend in federal funding is 
339: apparent, though there may be a delay of several years between the two 
340: curves. Degrees from astronomy departments account for about 25\% of the 
341: observed increase in Ph.D.\ recipients between the mid-1980's and 
342: mid-1990's, while most of the new degrees came from other departments.
343: 
344: To examine the relationship between federal funding and Ph.D.\ production 
345: more quantitatively, Figure~\ref{fig2} shows the correlation between 
346: inflation-adjusted research funding in a given year and the number of new 
347: degrees related to astronomy and astrophysics 4 years later (from UMI). 
348: Similar correlations were produced for time delays ranging from 0 to 5 
349: years, but a 4 year delay minimized the fractional errors on the fit 
350: parameters (formally, all of the fits were statistically equivalent). The 
351: best linear fit is shown as a solid line, with the $\pm1\sigma$ limits on 
352: the fit parameters indicated with dotted lines. The correlation suggests 
353: that an incremental increase of \$100 million in federal funding may lead 
354: to the production of 40-50 additional Ph.D.\ recipients in subsequent 
355: years (4 years is the median time to Ph.D.\ after the initial 2 years of 
356: coursework).
357: 
358: % FIGURE 3 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
359: \begin{figure}[t]
360: \includegraphics[height=\linewidth, angle=-90]{f3.eps}
361: \caption{The number of jobs advertised in the AAS Job Register from 1984 
362: to 2006 (upper solid line). A detailed breakdown by advertiser-specified 
363: job type is available beginning in the early 1990's, and is shown for 
364: positions categorized as postdoctoral and visitor (dotted line), research 
365: and research support (dashed line), tenure-track (lower solid line), 
366: non-tenure-track (long dashed line), along with management and other 
367: positions (dot-dashed line).\label{fig3}}
368: \end{figure}
369: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
370: 
371: \subsection{Demand for Astronomers}
372: 
373: The data from Table~\ref{tab2} on the demand for astronomers from 1984 to 
374: 2006 are shown in Figure~\ref{fig3}. The total number of jobs advertised 
375: (upper solid line) is shown along with the numbers of postdoctoral and 
376: visitor positions (dotted line), research and research support jobs 
377: (dashed line), as well as tenure-track (lower solid line), 
378: non-tenure-track (long dashed line), management and other (dash-dot line) 
379: advertisements. The growth in the total number of jobs advertised 
380: throughout the 1990's is remarkable, and is dominated by an increase in 
381: the number of postdoctoral positions, though all job categories 
382: experienced some growth during this period. The most recent data show some 
383: anti-correlated trends after the year 2000, suggesting a possible shift 
384: from research and support positions to postdocs, and from non-tenure-track 
385: to tenure-track jobs.
386: 
387: To assess the relative sustainability of the job market, we can normalize 
388: these job advertisements by the number of new Ph.D.\ recipients in each 
389: year (from UMI). These data are shown in Figure~\ref{fig4} for all 
390: advertised jobs (upper solid line), postdoctoral and visitor positions 
391: (dotted line), research and research support jobs (dashed line), 
392: tenure-track and non-tenure-track positions (lower solid line), as well as 
393: management and other advertisements (dash-dot line). By normalizing in 
394: this way, the vertical axis can be thought of as a relative probability of 
395: obtaining a given type of position for each class of Ph.D.\ recipients. It 
396: is not exactly a probability, since the number of foreign Ph.D.\ 
397: recipients competing for some jobs may exceed the number of foreign jobs 
398: advertised, and because some Ph.D.\ recipients from previous years will be 
399: competing for some jobs while others will eventually choose to work 
400: outside the field or in a foreign country.
401: 
402: % FIGURE 4 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
403: \begin{figure}[t]
404: \includegraphics[height=\linewidth, angle=-90]{f4.eps}
405: \caption{Job advertisements in several categories, normalized by the UMI 
406: measure of astronomy and astrophysics Ph.D.\ production from 1990 to 2006. 
407: There is now more than one postdoctoral or visitor position (dotted line), 
408: about half a tenure-track or non-tenure-track job (lower solid line) and 
409: half a research or research support position (dashed line) advertised for 
410: each new Ph.D.\ recipient. If all advertised jobs (upper solid line) are 
411: filled each year, then the average new Ph.D.\ might expect to hold between 
412: 2 and 3 of them before finding a stable position.\label{fig4}}
413: \end{figure}
414: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
415: 
416: There now appears to be more than one postdoctoral position advertised for 
417: every new domestic Ph.D.\ recipient related to astronomy and astrophysics. 
418: This surge in the relative probability of obtaining a postdoc position can 
419: be traced to both a slightly declining population of new Ph.D.\ recipients 
420: since a peak in the mid-1990's (Figure~\ref{fig1}), as well as the larger 
421: number of postdoc positions advertised---possibly at the expense of 
422: longer-term research jobs (Figure~\ref{fig3}). At best, half of all recent 
423: Ph.D.\ recipients might expect to obtain a faculty position, while the 
424: other half will likely end up in a research or support position. While 
425: this situation may not be ideal for job seekers who prefer to remain in 
426: academia, it was much more competitive a decade ago with nearly 6 new 
427: Ph.D.\ astronomers in 1995 for every new tenure-track job. If we assume 
428: that all advertised jobs are eventually filled by someone in a related 
429: field, then the upper solid line in Figure~\ref{fig4} suggests that the 
430: average new Ph.D.\ might expect to hold between 2 and 3 jobs before 
431: finding a more permanent position.
432: 
433: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
434: 
435: \section{Discussion\label{DISC}}
436: 
437: We are now in a position to evaluate the predictions of Thronson (1991), 
438: who made 20 year projections of several models for the production of 
439: astronomers over time. His definition of the ``astronomer surplus'' was 
440: the annual ratio of new astronomy Ph.D.\ recipients tabulated by the AIP 
441: relative to new tenure-track faculty positions advertised in the AAS Job 
442: Register, which he calculated to be around 2.5 in 1991. Regardless of the 
443: absolute level of overproduction, his most realistic models included the 
444: competing effects of increased research funding (which leads to growth in 
445: the surplus), and retirement (which briefly absorbs some of the surplus). 
446: These models predicted a gradual decline in the astronomer surplus over 
447: about 10 years, followed by continued overproduction:{\it ``A momentary 
448: abundance of jobs will be subsequently compensated for by training of new 
449: graduate students by a new cohort of young professors''}. Indeed, the 
450: anticipated impact of increased research funding on Ph.D.\ production has 
451: now been quantified in Figure~\ref{fig2}. Furthermore, the slowly evolving 
452: ratio of faculty positions to new Ph.D.\ recipients in Figure~\ref{fig4} 
453: shows exactly the predicted behavior---peaking in 2001, and gradually 
454: leveling off or declining in more recent years.
455: 
456: Despite the accuracy of Thronson's predictions, a broader definition of 
457: sustainability may now be necessary, since the character of the astronomy 
458: job market appears to have shifted in the meantime. Astronomy research is 
459: increasingly moving towards automation and large collaborations where 
460: service positions are becoming more important (Sage 2001). This cultural 
461: shift might be reflected in the surge of new research and support 
462: positions between 1997 and 2002, which may have absorbed some of the 
463: astronomer surplus generated during the more competitive conditions of the 
464: early 1990's. More recently, as the new cadre of young faculty have 
465: attempted to recruit a limited supply of graduate students, some may have 
466: turned to postdocs to maintain their research productivity. This could be 
467: responsible for the unprecedented expansion of postdoctoral jobs after 
468: 1999---but as startup funds are depleted and external research funding 
469: stays relatively constant, these ``holding pattern'' positions could 
470: evaporate. The most recent data show the first signs of a possible 
471: slowdown.
472: 
473: The most important aspect of the ongoing cultural shift in the astronomy 
474: job market is the persistent gap between expectations and reality. When 
475: first year graduate students are surveyed by the AIP, fully 87\% of those 
476: in astronomy departments say they would like to end up in an academic 
477: position, while only 8\% express a desire to work in a national lab or 
478: research position (Mulvey \& Tesfaye 2006). By contrast, Figure~\ref{fig4} 
479: suggests that less than 50\% will ultimately obtain academic 
480: positions---and probably fewer, since the turnover at universities appears 
481: to be episodic. Graduate programs in astronomy should prepare their 
482: students for this reality.
483: 
484: Having documented the historical trends in the astronomy job market and 
485: the forces that help shape it, it should be straightforward to update the 
486: analysis periodically as new data emerge. In the future, I hope to make 
487: such updates available through the Career Services section of the AAS 
488: website\footnote{http://aas.org/career/}, and disseminate them through 
489: {\tt astro-ph}. With such information readily available, incoming graduate 
490: students can make informed decisions about their long-term career 
491: prospects, and faculty can nurture appropriate skills in the next 
492: generation of astronomers.
493: 
494: 
495: \acknowledgements 
496: This work was motivated by my involvement with the AAS Employment 
497: Committee, which is charged with facilitating the professional development 
498: and employment of astronomers at all career stages and on all career 
499: paths, and promoting balance and fairness in the job market. The National 
500: Center for Atmospheric Research is a federally funded research and 
501: development center sponsored by the National Science Foundation.
502: 
503: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
504: 
505: \begin{thebibliography}{}
506: 
507: \bibitem[Dinerstein(1996)]{din96} Dinerstein, H.\ 1996, BAAS, 28, 1277 
508: 
509: \bibitem[Harris(1994)]{har94} Harris, H.~C.\ 1994, ASP Conf., 57, 12 
510: 
511: \bibitem[Hill(2007)]{hill07} Hill, S.~T.\ 2007, NSF Report 07-305,
512: (www.nsf.gov/statistics/nsf07305/)
513: 
514: \bibitem[Jankowski(2007)]{jan07} Jankowski, J.~E.\ 2007, NSF Report 07-323, 
515: (www.nsf.gov/statistics/ nsf07323/)
516: 
517: \bibitem[Marvel(2007)]{mar07} Marvel, K.\ 2007, BAAS, 39, 283
518: 
519: \bibitem[Meeks(2004)]{meeks04} Meeks, R.~L.\ 2004, NSF Report 04-335, 
520: (www.nsf.gov/statistics/nsf04335/)
521: 
522: \bibitem[Meeks(2005)]{meeks05} Meeks, R.~L.\ 2005, NSF Report 05-307,
523: (www.nsf.gov/statistics/nsf05307/)
524: 
525: \bibitem[Meeks(2006)]{meeks06} Meeks, R.~L.\ 2006, NSF Report 06-313, 
526: (www.nsf.gov/statistics/nsf06313/)
527: 
528: \bibitem[Mulvey \& Tesfaye(2006)]{mt06} Mulvey, P.~J. \& Tesfaye, C.~L.\ 
529: 2006, AIP Report R-207.35
530: 
531: \bibitem[Mulvey(2007)]{mul07} Mulvey, P.\ 2007, personal communication
532: 
533: \bibitem[Sage(2001)]{sag01} Sage, L.\ 2001, \nat, 414, 4 
534: 
535: \bibitem[Suckow(2007)]{suc07} Suckow, E.\ 2007, personal communication
536: 
537: \bibitem[Thronson(1991)]{thr91} Thronson, H.~A., Jr.\ 1991, \pasp, 103, 90 
538: 
539: \bibitem[Thurgood et al.(2006)]{tgh06} Thurgood, L., Golladay, M.~J. \&
540: Hill, S.~T.\ 2006, NSF Report 06-319, (www.nsf.gov/statistics/nsf06319/)
541: 
542: \end{thebibliography}
543: 
544: \end{document}
545: