Souter, J.
These are consolidated appeals under RSA 151-C:9 (Supp. 1981) and RSA 541:6 brought by two unsuccessful applicants for a certificate of need to operate an intermediate care facility in North Conway. See RSA 151-C:2, XXIII (Supp. 1981), :4 (Supp. 1981) (amended by RSA 151-C:2, XXIII (Supp. 1983), :4 (Supp. 1983)). They raise a series of challenges to the application of statutory and regulatory criteria by the former Bureau of Institutional Health Services of the State Department of Health and Welfare (the agency) in awarding the contested certificate of need to Country Village Health Care, Inc. See RSA 151-C:2, XXIX (Supp. 1981). In the Appeal of Lemire-Courville Associates (No. 83-427) we affirm the denial of the certificate of need to Lemire-Courville, but in the Appeal of Gilmore-Holloway (No. 83-444) we vacate the agency’s order granting the certificate of need to Country Village and remand.
On April 16, 1982, Lemire-Courville sent the agency a letter of intent to construct a 120-bed nursing home, or intermediate care facility, in North Conway, although it later revised the proposed bed [23]*23capacity to 110. See RSA 151-C:7, IV (Supp. 1981). (Because this case arose in 1982, we will apply the 1979 version of RSA chapter 151-C which was then in effect.) The agency determined that a certificate of need review would be necessary under the statute and requested Lemire-Courville to apply for a certificate of need. See RSA 151-C:7, IV (Supp. 1981). The agency set a deadline for filing competing applications, see RSA 151-C:7, V (Supp. 1981); within the deadline Country Village filed such a competing proposal for an 80-bed facility, and Gilmore-Holloway filed a proposal for one of 72 beds.
The agency sent copies of the three competing proposals to the United Health Systems Agency (the HSA), which held a hearing on November 1, 1982, and on December 6, 1982, submitted a recommendation to the agency that Lemire-Courville’s application be granted. See RSA 151-C:7, XV (Supp. 1981). The agency held its own hearing, and on December 30, 1982, it announced its decision rejecting the recommendation of the HSA and granting Country Village’s application instead.
In its written decision, the agency mistakenly attributed to Country Village the lower construction and operating cost figures that Gilmore-Holloway had submitted. For that and other reasons, each of the losing applicants requested reconsideration, see RSA 151-C:8, V (Supp. 1981), and the agency granted the requests. Although the process of reconsideration was interrupted by a gubernatorial freeze on the issuance of certificates of need, the agency later held a five-day hearing for the reception of further testimony and exhibits. On September 19, 1983, the agency affirmed its original order, and issued a new decision in which it discussed some of the issues raised and responded to a total of more than two hundred requests for findings of fact submitted by Country Village and Gilmore-Holloway. After the agency had denied the losing parties’ motions for rehearing, see RSA 541:3, these appeals followed.
It is well to note that our consideration of the specific issues raised by the appellants is governed by RSA 541:13. The statute provides that “all findings . . . upon all questions of fact properly before [the agency] shall be deemed to be prima facie lawful and reasonable; and the order or decision appealed from shall not be set aside or vacated except for errors of law, unless the court is satisfied, by a clear preponderance of the evidence before it, that such order is unjust or unreasonable.” We note, too, that although RSA 151-C:9 (Supp. 1983) was enacted in its present form after the parties filed the applications in question, it sets essentially the same standard of review by providing that this court must “affirm the decision of the [24]*24[S]tate agency unless it finds it to be arbitrary or capricious or not made in compliance with applicable law.”
Mindful of these limitations on the scope of our review, we turn to several claims that the agen'cy committed errors of law, the first of which Lemire-Courville asserts in arguing that the agency failed to give proper consideration to the HSA’s recommendation in its favor. Although the agency’s response to the recommendation is open to fair criticism, its deficiencies fall short of reversible error.
At the outset it is clear that the agency is not bound by statute or regulation to adopt an HSA recommendation, but only to consider it. RSA 151-C:6, 1(a) (Supp. 1981); Dep’t of Health & Welfare Rules He-C 304.01(a). This it clearly did in the first decision, in which it referred to the specific HSA findings. Such cause for concern as there is rests on the requirement of regulation He-C 306.01(m), that “[i]f the . . . agency decision is inconsistent with the recommendation of the [HSA], the . . . agency shall submit to the [HSA] a written statement of the reasons for the inconsistency.” The statute likewise required a detailed statement of reasons for any inconsistency. RSA 151-C:8, IV (Supp. 1981).
The agency’s compliance with these mandates consists generally of references to evidence submitted, a reference to the HSA’s discussion of a given criterion, and the agency’s own conclusion about the application of that criterion. While it may therefore be possible to analyze the points of disagreement between the agency and the HSA, the agency’s approach certainly represents the minimum level of compliance with a regulation that calls for a statement of reasons.
Lemire-Courville next argues that the agency committed a series of legal errors in the application of criteria to evaluate the proposals before it. First among these is allegedly an error of omission consisting of the agency’s failure to adopt criteria for the evaluation of competing proposals as required by Appeal of Behavior Science Institute, 121 N.H. 928, 934-35, 436 A.2d 1329, 1332-33 (1981), which applied the rule of Ashbacker Radio Co. v. F.C.C., 326 U.S. 327, 333 (1945). In Behavior Science we adopted the Ashbacker doctrine that mutually exclusive applications must be considered together in order to insure the fair treatment, of each. Appeal of Behavior Science, 121 N.H. at 935, 436 A.2d at 1333. The agency, however, did just what these cases require.
It is implicit in both cases, of course, that each competing application will be evaluated under the same substantive criteria, and it is to the application of substantive criteria that we turn in considering Lemire-Courville’s next claim of legal error, that the agency made a mistake when it evaluated the proposals by reference, inter alia, to [25]*25the 1981 State health plan as prepared by the agency’s predecessor, the Office of Health Planning and Development. See RSA 151-C:2, XVIII (Supp. 1981). Lemire-Courville submits that the agency should have used the plan entitled “Health Choices 1982,” prepared by the HSA. This argument, however, ignores the requirement of RSA 151-C:6, I (Supp. 1981), that “the state agency . . . shall consider only the criteria which have been duly adopted and published in accordance with the requirements of this chapter as of the date of notification of the beginning of the review period.” Cf. RSA 151-C:8, I (Supp. 1983) (requiring the agency’s decision to be consistent with the State health plan in effect when the letter of intent is filed).
In this case, the notification of the beginning of the review cycle occurred on September 22, 1982, when the State agency caused such notification to be published in the Union Leader, a newspaper of general circulation in the State, and in the Irregular and Mt. Washington Valley News, a newspaper of general circulation in the proposed service area. See RSA 151-C:7, XI and XII (Supp. 1981). The State health plan in effect on this date was the 1981 version. The HSA plan on which Lemire-Courville relies, “Health Choices 1982,” was not “duly adopted” until it was approved by the U.S. Department of Health and Human Services on September 24, 1982. Therefore, it was not erroneous for the agency to evaluate Lemire-Courville’s application using the standards established in the 1981 State health plan, to the exclusion of “Health Choices 1982.”
Lemire-Courville’s next claim of legal error is that the agency applied inappropriate criteria when it evaluated the need for the proposed services, as required by RSA 151-C:6, 1(d) (Supp. 1981), by reference to the agency’s calculation of bed need in 1990. Lemire-Courville argues that the facility is not likely to be operational much before 1990, and that the agency ought therefore to look further into the future, at least to 1995, in evaluating the proposal. In support of this argument, Lemire-Courville cites a decision of the agency in 1982 that projections of need should extend at least through one-third of the anticipated debt service term of a facility that is proposed to meet that need.
The answer to Lemire-Courville’s argument rests simply on the fact that we have no way of knowing whether it would be preferable in this instance to look to 1990 or to 1995. One may argue that sound planning requires a long view, or that the reliability of projections diminishes as the forecast period lengthens. No statute or regulation commands either the longer or the shorter view in this case, and we must therefore recognize that the choice is one for the [26]*26judgment of the agency. See Appeal of Concord Natural Gas Corp., 121 N.H. 685, 692-93, 443 A.2d 1291, 1296 (1981). That judgment may not be set aside without evidence that it is clearly unreasonable, and there is no such evidence before us.
The fourth and last of Lemire-Courville’s claims of legal error attacks the agency’s decision to split Carroll County into two sections and to evaluate the competing proposals by reference to the projected needs of the northern section alone. This challenge goes to the very basis on which the agency calculated the bed need in order to evaluate the competing proposals on their relative conformity with that need. The significance of the agency’s decision to split the county is apparent when we recall that Lemire-Courville proposed to build a 110-bed home, to serve a larger area and population than Country Village proposed to serve with 80 beds, or Gilmore-Holloway proposed to serve with 72. Therefore, the larger the area and population assumed in the agency’s calculation of bed need, the more desirable Lemire-Courville’s proposals would appear, at least initially, in contrast to those of its competitors. Thus the agency’s choice to determine bed need by reference only to a portion of the county had some tendency to favor the other two applicants over Lemire-Courville. Accordingly, Lemire-Courville argues in its brief that the agency “does not have the authority to create special areas for determination of need by arbitrarily splitting the county into sections.” This argument raises two related issues.
The first issue is whether the agency violated any statute, regulation or plan simply by deciding that the “population to be served” should be the population of only a part of a county. See RSA 151-C:6, 1(d) (Supp. 1981). This issue is easily resolved in the agency’s favor. The State plan set the statewide standard for the optimum number of nursing home beds in 1990 at fifty beds per thousand people aged sixty-five or over. The State plan did not, however, prescribe the size of nursing homes or the political boundaries of the populations they must serve. There is no mandate to follow county boundaries or any other particular geographical or political boundaries within the State. Rather, the agency must approach each case with only the basic object of promoting what the statute calls a “rational allocation of health care resources in the [SJtate.” RSA 151-C:1 (Supp. 1981). It follows that a proposal to build such a facility, like the agency’s decision to approve or deny it, necessarily must address two quite separate issues: the appropriate area to be served, as well as the ability of the applicant to serve the need of that area.
Therefore the agency’s decision to divide the county was not error per se, and we must turn to the second issue raised by Lemire-Courville’s challenge to this particular decision: whether it did promote [27]*27such a rational allocation or was, conversely, as arbitrary as Lemire-Courville charges. We can resolve this issue by tracing the agency’s own steps.
The agency began the process of identifying the population to be served by assuming that the home would be in North Conway, where each applicant proposed to build it. The agency then determined the accessibility of North Conway by reference to geographical features and consequent transportation routes. These considerations convinced the agency that “with a rural mountainous county such as Carroll. . . the . . . agency [would be] within its right to split the county based on geographical considerations.”
The agency therefore concluded that the population to be served by a home in North Conway should be the residents of the northern half of Carroll County. It went on to analyze corresponding population distribution and found that the southern portion of Carroll County was more densely populated than the northern part that included North Conway. Specifically, the 1980 census data showed that 40% of the county population aged sixty-five or older lived in the northern section. The agency therefore allocated the same percentage of the county-wide bed need to the northern section. It then identified both existing and proposed facilities, inside and outside the county, that would serve the population of the northern half. Based on the 1981 State plan’s standard of 50 beds per thousand aged sixty-five and above, the agency determined that by 1990 the northern section of the county would need 81 beds in addition to those that could otherwise be anticipated. This method for determining service area and need was obviously reasonable and must be upheld.
With the resolution of these legal issues about appropriate criteria raised by Lemire-Courville, we turn to the claims of both Lemire-Courville and Gilmore-Holloway, that the agency’s decision is clearly unreasonable when judged on the basis of the evidentiary record before us. On this point, Lemire-Courville’s argument rests on generalities that do not warrant extended analysis. Gilmore-Holloway’s challenge, however, is of a different order. It is best understood by first analyzing the agency’s responsibilities to investigate and compare capital and operating costs entailed by proposals before it, and then examining the agency’s September, 1983 findings about the costs of the proposed projects.
In the enactment of RSA chapter 151-C, the legislature declared that new institutional health services ought to be developed in a “manner which .. . contains or reduces increases in the cost of delivering services.” RSA 151-C:1, I (Supp. 1981). This object was, and [28]*28remains, reflected in the statutory and regulatory criteria for reviewing applications for certificates of need, which require the agency to consider, inter alia:
“[the] probable impact [of the proposal] on the costs of and charges for providing health services by the person proposing the new institutional health service,” RSA 151-C:6, 1(f) (Supp. 1981); see RSA 151-C:6,1(e) (Supp. 1983);
“in the case of a construction project . . . the costs and methods of the proposed construction . . . and . . . the probable impact of the . . . project... on the [applicant’s] cost of providing . . . services”; see RSA 151-C:6, I(n) (Supp. 1981); cf. RSA 151-C:6,1(m) (Supp. 1983); and the capital and operating costs of the project proposed and the likely effect of those costs on patient charges.
See Dep’t of Health and Welfare Rules He-C 304.01(v). To underscore the significance of these considerations, the statute requires the agency to make “specific written findings ... as to [t]he capital and operating costs [and] the potential impact of those costs on patient charges . . . .” RSA 151-C:6, 111(b)(2) (Supps. 1981 and 1983).
The State health plan reflected these same concerns by providing that “[decisions regarding new institutional health services should take price competition into account.” State of New Hampshire, State Health Plan 1981 § IV. F.(2), par. 158. And it provided even more specifically that “[w]here there is no demonstrable difference in quality among competitors, the provider whose services are priced at or below the median for providers of the same services should be encouraged to expand, and the higher priced provider should not.” Id., par. 160.
At the reconsideration proceeding, Country Village and Gilmore-Holloway addressed the foregoing cost criteria and mandates for economy by submitting requests for specific findings and rulings on the costs entailed by their proposals. Like the applications to which they related, the applicants’ requests for findings of fact were mutually exclusive, since each applicant sought findings that its costs for providing the proposed services would be the lowest of the three.
In responding to these requests, however, the agency granted requests for apparently contradictory findings, some favorable to Country Village and some favorable to Gilmore-Holloway. We say “apparently,” because the language of the findings is not so consistent as to allow us to be absolutely certain of what the agency found. The most that we can say is that nothing that the agency has written [29]*29either clarifies the terminology or resolves the apparent conflicts among findings.
Specifically, the agency found that “[o]n a construction cost per bed, Country Village is less costly to build than either of the competing applicants,” and, further, that “[t]he Country Village application proposes the lowest overall costs and charges of the three competing applications.” Consistently with these findings, the agency denied Gilmore-Holloway’s request to find that Gilmore-Holloway proposed “the most overall economical project.”
Simultaneously, however, the agency found that Gilmore-Holloway had the least costly “per bed project cost” and went on to grant Gilmore-Holloway’s following requested finding in its entirety:
“The 1981 State Health Plan states at page 227:
“‘160. “Quality of care” should be the primary consideration for approval of new institutional health services where there is need for expansion of services. Where there is no demonstrable difference in quality among competitors, the provider whose services are priced at and below the median for providers of the same services should be encouraged to expand, and the higher priced provider should not.’
“Gilmore-Holloway’s total project, [sic] costs, operating costs, and bed rates are all the most competitive amongst the three applicants, and therefore Gilmore-Holloway should be the applicant encouraged to expand.”
Summarizing all of this together, the agency found that Country Village “proposes the lowest overall costs and charges,” but that Gilmore-Holloway’s “total project, [sic] costs, operating costs and bed rates are all the most competitive” and that Gilmore-Holloway “should be . . . encouraged to expand.” These findings appear to be squarely contradictory, even making allowances for the inconsistent terminology.
Country Village has addressed the problem of these contradictions by several arguments aimed at sustaining the agency’s action in its favor. In our judgment, however, nothing that either Country Village or the agency has said is sufficient to reconcile the contradictions or to provide a reason to sustain the decision in spite of such contradictions.
First, Country Village’s brief has attempted to blunt the inconsistency by referring to the Gilmore-Holloway cost figures as merely “projected,” while pointing to an evidentiary basis in support of its own figures. Country Village thus argues implicitly that the conflict [30]*30is illusory, as between Gilmore-Holloway’s proposed but unsupported figures and Country Village’s substantiated numbers as accepted by the agency. There are, however, two impediments to accepting this interpretation.
A close reading of the findings in question lends no support for the view that the agency accepted Country Village’s numbers and rejected Gilmore-Holloway’s. One of the findings quoted above states that Country Village “proposes” the lowest overall costs, while another states flatly that “Gilmore-Holloway’s total project [sic] costs ... are .. . the most competitive ....” (Emphasis added.)
What is far more important than these verbal distinctions, however, is that the decision on reconsideration includes no cost analysis by the agency itself that would support the argument that the agency found the Country Village figures more reliable than Gilmore-Holloway’s. As a part of its first decision in 1982, the agency prepared such an analysis, in an attempt to meet its statutory obligation to “make specific written findings as to . . . [t]he capital and operating costs ... of the proposed new institutional health service.” RSA 151-C:6,111(b)(2) (Supps. 1981 and 1983). It did this in the form of a comparative analysis of costs, attached to the agency’s decision on each of the competing applications. That analysis projected Lemire-Courville’s costs and charges as the highest among the three. As between the other two applicants, the agency found that Gilmore-Holloway would build for the lowest total- cost per bed and the lowest cost per square foot. But, it also found that Country Village would build for the lowest construction cost per bed, would have the lowest fixed costs, the lowest operating cost per bed and the lowest proposed first year charges per bed.
It was in this 1982 analysis, however, that the agency mistakenly attributed Gilmore-Holloway’s lower interest costs to Country Village, and the latter’s higher interest costs to Gilmore-Holloway. So far as we are able to tell from the record, this mistake led the agency to commit error in finding that Country Village would have lower fixed costs, lower operating costs per bed and, perhaps, a lower proposed charge per bed. In any case, it is stating the matter most conservatively to say that the agency’s 1982 mistake tainted every calculation that took interest cost into account.
Although in the 1983 reconsideration decision the agency acknowledged this error, it never prepared corrected findings of probable costs and charges. As we have said above, on the record before us it apears that if other factors remained equal, such corrected findings would have favored Gilmore-Holloway on most, if not all, points on which Country Village was given the advantage in the 1982 evaluation. In any case, it is again speaking conservatively to say that the [31]*31present record provides no basis to suggest that the agency found the Gilmore-Holloway cost and charge projections less worthy of acceptance than the comparable figures for Country Village. The contradiction among the findings still stands.
As a second argument to sustain the decision, Country Village emphasizes the agency’s finding that Gilmore-Holloway underestimated the number of potential medicaid patients who would be served by the facility and planned to set aside too few beds for them. In contrast, it found that Country Village had a more realistic appreciation of medicaid need, would probably obtain more patients in the first year of operation because it would actively seek medicaid patients, and had more realistic revenue projections as a result.
Accepting all of these findings, however, they are not enough to explain why the agency found in favor of Country Village. For it would still appear that Gilmore-Holloway should be “encouraged to expand” by adjusting its sights to a more realistic medicaid level, since, as the agency found, its costs and rates “are all the most competitive.”
The third argument that might be thought to reconcile the findings with the decision rests on the agency’s attempt to discount the significance of Gilmore-Holloway’s lower projected construction costs. Gilmore-Holloway proposed lower construction costs, in part because it would not depend on financing guaranteed by the Department of Housing and Urban Development and thus would not be bound by H.U.D. construction standards. The agency responded to this by observing that “[t]he difference in construction cost would indicate that there would have to be a difference in the materials and standards of construction. In fact, Gilmore-Holloway would build a wood-protected building versus Class A construction.” The agency suggests, in other words, that Gilmore-Holloway’s construction will be cheaper only because of lesser quality; quality will therefore not be equal as among the applicants, and there is justification for favoring Country Village because of its preferable construction.
This explanation is unavailing, however, because even if the reasoning had been logical, the agency apparently had the facts wrong. According to the applications, it is Country Village that proposes a wholly “wood-protected building,” while Gilmore-Holloway proposes to use some exterior brick construction. Thus, Gilmore-Holloway’s lower project cost per bed cannot be explained on the basis of all wood as against better quality construction.
Moreover, even if the agency were correct that Gilmore-Holloway’s construction would somehow be less desirable than Country Village’s, it is difficult to see how this fact would justify the present [32]*32decision for Country Village. As we have noted, the relevant State plan provided that if anticipated quality of care is equal as between two applicants, the applicant with anticipated lower “priced” services should “be encouraged to expand.” In this case the agency found originally that there would be no differences in the quality of care provided by the applicants, and it is clear from the specific finding quoted from the reconsideration decision that the agency continues to find such equality among them.
Two things follow: the differences in proposed construction would not affect quality of care, and the applicant with lowest cost should be “encouraged.” On the present record, therefore, Gilmore-Holloway has at least as good, if not a better, claim than Country Village.
The fourth and last point urged for sustaining the present decision is that Country Village’s proposal for 80 beds is closer to the projected need of 81 than Gilmore-Holloway’s proposal for 72. This is true but not sufficient to justify the decision, for it simply raises anew the question why Gilmore-Holloway with its “most competitive” position on costs was not “encouraged” to amend its proposal by increasing bed capacity to 80 or 81.
Thus, despite all of these attempts at explanation and reconciliation we are back where we started, with a set of contradictory findings of fact about the applications of Country Village and Gilmore-Holloway. Such contradictory findings on material issues, made on the basis of a given evidentiary record, are necessarily capricious and insufficient to support a judgment. Malstrom v. Consolidated Theatres, 4 Utah 2d 181, 183, 290 P.2d 689, 690-91 (1955); Wenzel v. Wenzel, 283 S.W.2d 882, 887 (Mo. Ct. App. 1955). Indeed, any judgment so grounded is unreasonable and must be vacated under RSA 541:13.
Accordingly, we remand the case to the agency, which is now reconstituted and known as the New Hampshire Health Services Planning and Review Board. Laws 1985, 378:6-8.
To avoid wasteful proceedings, the reconsideration should be limited to the issues on which the most recent decision is deficient: the issues of capital and operating costs and their effects on charges to patients. We have no basis to say whether the agency should receive further evidence on these subjects, and leave that to the agency’s reasonable discretion. At the conclusion of the proceedings, the agency’s decision addressing the issues of costs and charges must satisfy the following requirements:
A. The decision must define its own terminology.
B. The decision must contain specific findings as to each of the two competing applicants on the “capital and operating [33]*33costs, [and] the potential impact of those costs on patient charges.” RSA 151-C:6,111(b)(2) (Supps. 1981 and 1983).
C. The findings must be free of internal contradiction.
D. The agency must discuss the application of the provision of the 1981 State plan, quoted above, that as among applicants who will provide care of equivalent quality, the applicant with lower “priced” services should be “encouraged to expand.”
No. 83-J+27 affirmed; No. 83-UUU vacated and remanded.
All concurred.