2 New Mexico Business / August 2007
NM Population Estimates...
(cont. from page 1)(continued on page 6)
New Mexico1,819,0461,852,7401,876,2871,899,8461,929,7131,968,3522,010,570Bernalillo556,678571,440581,118590,153600,439614,508628,188Catron3,5433,5893,5953,5953,6433,7123,824Chaves61,38261,34361,14661,08561,72262,20363,166Cibola25,59525,86526,14026,49828,27828,50628,683Colfax14,18914,30414,32614,35114,35114,37514,540Curry45,04445,26745,39545,60945,67046,28946,666De Baca2,2402,2582,2552,2632,2742,2562,271Dona Ana174,682177,981180,162182,497185,872192,474198,625Eddy51,65851,72851,81351,87751,90952,16752,449Grant31,00231,19131,23231,30031,33731,51131,733Guadalupe4,6804,7224,7304,7304,7254,7434,821Harding810809811805790778814Hidalgo5,9325,9195,9135,9035,9185,9665,960Lea55,51155,58755,64455,78356,65757,00658,175Lincoln19,41119,78319,87120,33320,90421,89822,523Los Alamos18,34318,52418,55118,88819,19019,86419,906Luna25,01625,42525,55425,76626,35026,39427,844McKinley74,79875,63876,24776,79277,12578,01379,781Mora5,1805,2115,2335,2485,3805,4405,472Otero62,29862,59862,60962,64363,19063,99466,027Quay10,15510,22010,21710,17410,10910,10610,208Rio Arriba41,19041,65241,76941,91142,11843,02443,530Roosevelt18,01818,25218,07618,29318,42918,77118,858Sandoval89,90892,99096,16199,231102,862106,165111,493San Juan113,801116,075119,690122,019124,808126,008127,618San Miguel30,12630,33730,49730,56130,60630,71930,817Santa Fe129,292132,462134,645137,358139,760143,306147,409Sierra13,27013,58513,63813,65213,64713,65713,726Socorro18,07818,27618,32118,36218,38918,51318,656Taos29,97930,38230,87031,30131,55631,93132,127Torrance16,91117,30717,36217,47117,69218,28218,302Union4,1744,1794,1754,2404,2104,3154,470Valencia66,15267,83868,52369,15269,80471,45971,888
Source: UNM, Bureau of Business and Economic Research; Census 2000 data fromU.S. Census Bureau, Census 2000.
CountyCensus2000Table 2. Bureau of Business and Economic Research Population EstimatesJuly 1, 2001 to July 1, 2006BBER Estimates as of July 1…
method may explain the large gap between their estimates and thoseproduced by BBER-PEP, but it also raises questions about the validityof their method for estimating the New Mexico population.Table 1 (on page 1) presents an alternative set of estimates basedon a conservative housing unit-based procedure and the CensusBureau’s own published data on New Mexico housing stock. It focusesupon state-level differences, but the reader should be aware that county-by-county differences are also present between CB-PEP and BBER-PEP estimates (see Table 2 above, which presents the recently-released 2006 BBER-PEP county and state estimates for July 1,2006). The first row of estimates in Table 1 represent a hypotheticalhousing unit-based estimate using 2000-2006 housing stock estimatesfrom CB-PEP and Census 2000 values for housing occupancy rates(OR) and persons-per-household (PPH). This method is conservativebecause it holds the OR and PPH values constant, but CB-PEPdoes exactly that when producing the sub-county estimates. Thesecond row presents the current CB-PEP state population estimatesproduced using the administrative records approach and the third rowpresents the BBER-PEP 2001-2006 state population estimates basedon its own estimates of the New Mexico housing stock and Census2000 values for OR and PPH. The remaining rows contrast each ofthe estimates in terms of absolute and percent differences. Figure 1on page 7
reviews the results graphically. It is clear that the CB-PEPestimates would be substantially higher if they were using a housingunit-based approach. Their estimates would be quite similar acrossthe 2001 to 2006 period to those produced by BBER-PEP (again,the only difference between the estimates is the housing-unit stock)and substantially higher than the current estimates (CB-AR in thetable).The results of this thought-experiment suggest that the CB-PEPestimates might be improved by considering the housing unit-basedmethod for producing county and state estimates. CB-PEP, however,might defend the use of the administrative records approach basedupon the insensitivity of the housing unit approach indicated here tointercensal changes in OR and PPH. The method is insensitive,again, because 2000 values of OR and PPH are held constant.