dissenting:
I agree with the majority that the Secretary did not act arbitrarily and capriciously when he switched from using the number of budgeted beds to using the number of physical beds for calculating the Medicare IME adjustment. I also agree that the Secretary’s interpretation of “available beds” as presumptively meaning physical beds is entitled to deference from this court and is reasonable as a general matter. However, because I believe that County/USC met its burden of rebutting the presumption and showing that certain beds should have been excluded from Med Center’s available bed count in the fiscal year ending June 30, 1994 (“FY1994”)— namely the 123 beds that made up the difference between the number of physical beds and the number of budgeted beds — I would hold that the Provider Reimbursement Review Board’s (“Board”) decision was arbitrary and capricious.
As the majority explains, the relevant interpretive regulation directs that “beds available at any time during the cost reporting period are presumed to be available during the entire cost reporting period.” PRM-1 § 2405.3; Maj. Op. at 1081. However the majority relegates to a parenthetical a crucial caveat: hospitals may *1083provide “evidence to the contrary” in order to “exclude beds from the count.” Id. In other words, hospitals have the opportunity to prove that certain beds should not be considered available and thus should be excluded from the count. At a hearing before the Board, County/USC presented evidence that the 123 physical beds that were not included in the budget should be excluded from the IME calculation because those beds were not actually available for patient use during the year. Disregarding the substantial evidence presented by County/USC, the Board concluded that all 1,320 physical beds should be considered available beds and denied County/USC’s request to adjust the figure. Thus, the issue in this case is not whether the Secretary’s interpretation of the term “available beds” as presumptively meaning physical beds instead of budgeted beds is arbitrary and capricious, but rather whether the Board’s ruling that County/USC did not meet its burden of showing that the physical beds that were not budgeted for were not actually available and thus should be excluded from the count.
I
At the hearing before the Board, County/USC presented substantial evidence that in FY1994, Med Center did not actually use any of the extra physical beds that were not included in the budgeted beds figure. According to the Chief Financial Officer of Med Center, there was not a single day that the hospital had the resources to staff all of the budgeted beds to say nothing of the physical beds. He testified that the average number of beds that were actually available in the relevant year was 1,056, and that that number could vary on any given day by 50. Therefore, on the busiest days of the year, Med Center had only 1,106 beds available for patient care. If anything, the 1,197 budgeted beds that County/USC seeks to use as the available bed figure is too high, not too low. If the budgeted beds figure is too high, then the 1,320 physical bed figure that the Secretary proposes is certainly too high. With this testimony alone, County/USC more than met its burden of proving that the 1,197 budgeted beds was a more accurate count of available beds in FY1994 than the 1,320 physical beds.
Even counsel for the Secretary admitted during the hearing that in the relevant year “there’s a fair cushion between the amount that the hospital calls budgeted beds and what its census was.” He stated that the question of whether the hospital could actually use the physical beds that were not included in the budgeted bed count was “hypothetical” and “abstract” because actual data indicates that the hospital did not even use all of the budgeted beds that year.
The Board’s decision did not discuss or acknowledge the strong testimony County/USC offered. Instead, it found that “[t]he provider had 1,320 beds available, and on any given day its budgeted cap of 1,197 may have been exceeded.” In light of the CFO’s undisputed testimony and the admission of the Secretary’s own lawyer that the budgeted cap was never actually exceeded, it is apparent that the “ ‘explanation for [the Board’s] decision runs counter to the evidence before the agency.’ ” Ranchers Cattlemen Action Legal Fund United Stockgrowers of America v. United States Dept. of Agric., 499 F.3d 1108, 1115 (9th Cir.2007) (quoting Motor Vehicle Mfrs. Ass’n v. State Farm Mutual Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983)). On that basis, I would hold that the Board’s decision was arbitrary and capricious under that APA.
II
In addition to holding that County/USC met its burden of showing that the budget*1084ed beds figure more accurately reflected the actual number of available beds at Med Center in FY1994 than the physical beds figure, I would also hold that County/USC offered sufficient proof that budgeted beds is the appropriate figure to use to measure available beds at Med Center as a general rule.
The Board’s findings and decision to use the physical bed count to calculate Med Center’s IME adjustment ran counter to the evidence presented by County/USC at the hearing. First, the Board found that County/USC stipulated that Med Center had 1,320 beds “available for patient care” and that this was “the strongest argument that this number is appropriate for the IME calculation.” The Stipulation does not support the Board’s finding. The Stipulation states: “For the fiscal year ended June 30,1994 ... the Intermediary agrees to revise the Provider’s indirect medical education (TME’) payment amount using 1320 beds in the ratio of interns and residents-to-beds.” (emphasis added). The reason that it was the Intermediary (Blue Cross) that stipulated to this number— instead of County/USC, as the board claims — is that Blue Cross originally sought to use the full count of licensed beds at Med Center, which was 2,045, but later agreed to lower that number to 1320 after County/USC demonstrated that some of those licensed beds could not be considered for the purposes of the IME calculation. The stipulation binds Blue Cross to a lower bed count number than it originally proposed. It does not represent, as the Board suggests, a concession on the part of County/USC that 1,320 is the appropriate number of beds to use for the IME calculation. In fact, the Stipulation clarifies that it “only partially settles the IME bed count issue that is still pending at the PRRB” — that is, the discrepancy between physical beds and actual beds that is now before us. Additionally, there is nothing in the Stipulation that discusses whether the beds were “available for patient care” and County/USC certainly did not stipulate that they were, as the Board found.
The majority does not discuss the Board’s finding on this issue despite the fact that the Board considered it the “strongest argument” for using the number of physical beds instead of the number of budgeted beds. Because the Board’s finding clearly “runs counter to the evidence before [it],” I would hold that it was arbitrary and capricious in violation of the APA. See Ranchers Cattlemen Action Legal Fund United Stockgrowers of America, 499 F.3d at 1115.
Next, the Board found that the budgeted beds count was an appropriate figure to use for planning staffing numbers and other administrative purposes, but that it did not constitute “an absolute cap on bed utility.” However, the evidence before the Board proved that the budgeted beds figure did serve as a limit to the number of beds available for use at Med Center.
According to testimony presented by County/USC, the term “budgeted beds” connotes how much “staff, materials and supplies, pharmaceuticals, support is available to provide care and ... sets a ceiling for the operational capacity of the organization.” Plainly stated, the budgeted beds figure “sets a cap on the capability of the organization to deliver safe, appropriate patient care.” Because the “budgeted beds” figure is “a proxy for the nursing care and the dietary and linen contracts,” the number of budgeted beds represents “the number of inpatient days” Med Center can provide. In fact, County/USC monitors the bed count “every single day” and any deviations are immediately corrected.
County/USC also presented testimony that, unlike a private hospital in which the budget functions as a “floor” for planning purposes and can be expanded to accom*1085modate increased demand or emergencies, the budget for Med Center “is really a ceiling.” Med Center’s budget is determined by the Los Angeles County Board of Supervisors and cannot be altered during the year. The section head of the Controller’s Division responsible for the hospital’s budget team testified that in his ten to fifteen years of experience, he did not recall a hospital ever getting the number of budged beds adjusted during the year.
Ignoring the extensive testimony to the contrary, the majority nevertheless argues that the physical beds that were not budgeted were available because they were “plugged in and ready to go.” Maj. Op. at 1082. This is not so. They’re not even “fired up and ready to go.” Med Center’s CFO testified that a bed is only available when “there’s a physical bed that meets the needs of the patient and there is staff to operate them.” A bed that does not have staff, linens, or food is no more available for use by a patient than the painting on the wall. If there is an empty bed, “plugged in” or not, but no nurses to staff it, a “patient would have to wait in the Emergency Room until such time as the bed became available to transfer him or her.” Once the Emergency Room is saturated, patients are turned away rather than put in the empty beds that fall outside of the budgeted number. In other words, the physical beds are not necessarily available. Only those beds that are budgeted for can reasonably be considered available. Thus, the budget serves as a fixed cap on bed utility.
Although one hospital official testified that from a budgetary perspective, it was “marginally” possible to provide inpatient hospital care in more beds than the budget provides, such a scenario would be almost impossible. If the hospital could save money in another area, or if every patient who came to the hospital required only a low cost service, or if there were a sudden increase in paying patients, it might be possible to “bump up” the number of beds by the end of the year to “operate a few more.” There are so many fixed variables in the “budgeted beds” figure, however, that “it would be very difficult to increase the number of beds that [Med Center] operate[s] on the basis of those changes.” Additionally, on any given day the number of available beds could theoretically exceed the number of budgeted beds, but that increase would have to be balanced out by a proportional decrease later in the year. As discussed above, this did not actually happen on any day in FY1994 — and the figure for budgeted beds has never been altered in the ten to fifteen years that the head of the responsible Controller’s Division has been at County/USC.
The majority characterizes the parties’ disagreement as being whether the IME calculation should be based on the staffing level of the hospital or the size of the hospital. Maj. Op. at 1079-80. According to the majority, the Secretary reasonably chose to base the calculation on the size of the hospital, not, as County/USC advocated, on “a budgeted bed count that turns on staffing.” Maj. Op at 1080. This analysis misses the point. The budgeted bed count does not “turn[ ] on staffing.” Maj. Op. at 1080. Just the reverse — the level of staffing turns on the number of budgeted beds. Similarly, the number of beds available for patient care (i.e. the size or capacity of the hospital) is at most the number of budgeted beds.
This is not true of all hospitals. The staffing levels of private hospitals are not limited by the number of budgeted beds. When a private hospital experiences an unexpectedly busy year, its revenues increase along with its patient census because private hospitals treat many more paying and insured patients. The hospital *1086then uses those unanticipated revenues to hire additional staff and provide the services necessary to transform a physical bed into an available bed, thereby increasing the capacity of the hospital. Thus, the number of budgeted beds does not provide a limit on the staffing levels or the number of available beds because both can be increased during the year. The physical bed count is the only accurate measure of a private hospital’s size. At Med Center, however, the budget limits the staffing level and the size of the hospital (the number of available beds). Consequently, at Med Center, there is no meaningful distinction between the size or capacity and the level of staffing because both are fixed by the same figure' — the number of budgeted beds. Distinguishing between the size and the level of staffing or the number of budgeted beds at Med Center was arbitrary and capricious.1
Furthermore, the Board found that “there was no evidence that the Provider closed off various floors or areas of the hospital.” The majority asserts the same. Maj. Op. at 1082. This is true as a factual matter, but it should not have any effect on the outcome of the case. The interpretive regulation refers to closed wings as an example of one situation in which a bed would not be considered “available for use and housed in patient rooms or wards.” PRM, § 2405.3G. The regulation never directs that a bed will be considered available unless it is kept in a closed wing. Although some hospitals may well have to place unused beds in a closed ward in order to prove that they are not available, Med Center presented substantial evidence that the physical beds that were not included in the budget were not available for patient care. Requiring the hospital to move those beds to a closed ward would be unreasonable as there are 123 physical beds that will necessarily remain unused for the entire year, it is immaterial whether the area in which they are placed is closed off physically. The interpretive regulation does not mandate that the only way to exclude physical beds from the count is to move them to a closed ward and the Board’s determination that such a requirement applied in all cases was arbitrary and capricious.
The Board and the majority also argue that the budgeted beds figure did not indicate which beds were available and which beds were not. Again, I fail to see how identifying which particular beds are available and which are not makes any difference in determining the size of the hospital for the purpose of the IME adjustment. Neither the majority nor the Board provide an explanation.
Finally, the Board found that because Med Center “was reimbursed capital costs” for the 1,320 physical beds, “it would be inconsistent” for Med Center *1087“not to include those beds in computing the Provider’s IME adjustment.” The Board does not specify which capital costs would be reimbursed based on the number of physical beds, nor does the Board explain how capital cost reimbursement relates to bed availability.
The Chairman of the Board questioned the CFO about the issue of capital costs. According to the CFO, the hospital files its cost reports based on the number of budgeted beds not the number of physical beds. Even the Chairman acknowledged that “the building cost is the same whether there’s [sic] 1300 or 1500” beds. The only other capital cost that the Board could possibly be referring to that was discussed at the hearing was depreciation, but, as the CFO testified, “[depreciation is not based on the number of beds that’s [sic] in use.” There is no evidence in the record that the hospital was receiving extra reimbursement money for capital costs because it had a number of physical beds that were not available for use. The Board’s finding on this issue is not supported by the evidence.
Even if the unavailable physical beds did qualify Med Center for additional capital costs, this fact would be irrelevant to the available bed count for the purpose of the IME adjustment. There are many beds that are not included in the available bed figure that presumably the hospital is reimbursed for, including the bassinets and psychiatric beds that the regulations expressly exclude from the IME count. The purpose of the IME adjustment is to reimburse hospitals for the costs involved in treating patients while also teaching medical students. The number of available beds is relevant because it reflects the workload of actual doctors working with actual patients. Whether or not a physical bed entitles the hospital to receive additional reimbursement for capital costs is irrelevant to that calculation. I would hold that the Board’s finding on this issue was arbitrary and capricious because it was not supported by the evidence and it was irrelevant to the question before it.
Conclusion
Although the Secretary’s adopted presumption about the appropriate figure to use in calculating “available beds” is entitled to deference from this court, the Board is not permitted to impose that presumption categorically without regard to the evidence before it. Because I believe that County/USC more than met its burden of proving that the physical beds at Med Center that were not in the budget appropriation were not available for the any portion of FY1994 and thus should be excluded from the bed count used to determine the IME adjustment, I would hold that the Board acted arbitrarily and capriciously when it held the opposite. Therefore I respectfully dissent.
. Indeed, this crucial distinction between the function of Med Center’s budget and that of a private hospital also explains why Altoona Hosp. v. Thompson, 131 Fed.Appx. 355 (3rd Cir.2005), is not relevant to this case. Altoona involved a private hospital that argued that some of its physical beds should not be included in the IME adjustment calculation because they were not able to be staffed within 24-48 hours. Id. at 356. However, the hospital admitted that it kept the beds because there was "a potential for an increase in patient census” and it "hoped to gain market share.” Id. Thus, staffing levels at Altoona Hospital did not determine the size of the hospital. At Med Center, however, the size of the hospital and the staffing level are coterminous. Clark Reg’l Med. Ctr. v. Dept. of Health and Human Servs., 314 F.3d 241 (6th Cir. 2002), is also inapposite. That case involved the question of whether beds that were used for two purposes should be counted when the regulation excluded one of the purposes from the calculation. Unlike the unbudgeted physical beds in this case, the beds in question at Clark Regional Medical Center were "always staffed and available for acute case inpatients.” Id. at 248.