OPINION

KEITH, Chief Justice.

This case, on certiorari from the Minnesota Tax Court, concerns the proper classification and valuation for real estate tax purposes of the Oak Ridge Conference Center (Oak Ridge) and adjacent land owned and operated by American Express Financial Advisors, Inc. (American Express). On appeal, American Express challenges the decision of the tax court classifying the facility as a special purpose property and estimating the market value of the property at $27,300,000 for 1994, 1995 and 1996 using only the cost approach to valuation. We hold that the tax court's finding that Oak Ridge is a special purpose property was clearly erroneous. Further, we hold that the court's exclusive reliance on the cost approach was an abuse of discretion in light of the fact that Oak Ridge is not a special purpose property and in light of the abundant data with which an income approach could have been performed. Additionally, in applying the cost approach, the tax court incorrectly estimated the value of the property according to its value-in-use rather than its market value. However, the tax court did not err in using the reproduction cost of the structure in its cost approach and it correctly allowed the inclusion of several disputed cost categories in this estimate. It also correctly used the market value of the site valued as though vacant in its cost approach. We affirm in part, reverse in part and remand for further findings with respect to the valuation of the subject property.

I.

In 1987, American Express began planning the construction of the Oak Ridge Conference Center in response to its difficulty finding a conference center which could accommodate its growing sales force of financial planners. The company acquired approximately 46 acres of land on the west side of Highway 41 in Chaska, Minnesota for the building site. American Express planned to use the facility primarily to house its financial planner training, renting excess space to third parties. To meet the particular needs of the financial planners, American Express constructed the facility with double occupancy rooms with dividers for privacy, larger conference rooms located near smaller "breakout" rooms, and a large banquet room. American Express also added design features to encourage third-party use including enlarged guest rooms, a lounge, a terrace, and additional parking spaces. In addition, American Express' architects designed Oak Ridge to be as flexible as possible. The guest and conference rooms were designed with non-load-bearing walls which could be easily moved and the guest room and conference room wings were structured to allow for expansion.

Oak Ridge was completed in February 1990 at a cost of approximately $23,000,000 for the facility and $1 for the land, deeded by the City of Chaska. Oak Ridge has 147 guest rooms with 240 beds, a 200-seat main dining room, a smaller executive dining room, and a 56-seat lounge. It contains 34 conference rooms of various sizes, including eight large conference rooms which can seat up to 140 persons theater-style. Its amenities include a half-court gymnasium, two racquetball courts, a weight room, a game room, walking trails, a softball diamond, two saunas, a volleyball court and a grand fireplace.

In scheduling the use of the facility, Oak Ridge gives first priority to American Express for its 10-day financial planner training program. It prioritizes other uses according to how much profit they will generate. Generally, Oak Ridge rents remaining space to other groups associated with American Express and outside corporations, although the facility has recently begun to seek business from transient business travelers as well. Oak Ridge charges American Express' financial planner group according to the estimated cost of their use of the facility, while it charges other American Express groups and third-party users market rates. The use of the facility by the financial planners, as expressed as a percentage of occupancy, has decreased in recent years, from approximately 21% in 1993 to 14% in 1995. Meanwhile, use of the facility for other purposes has increased from approximately 29% in 1993 to 35% in 1995. Oak Ridge competes for third-party business with Northland Inn and Executive Conference Center, the Radisson Hotel and Conference Center, the Riverwood Metro Business Resort, and other conference hotels in the southwest metropolitan area. The assessor estimated the fair market value of Oak Ridge to be $20,012,800 for 1994, 1995, and 1996. American Express appealed this valuation to the tax court and the matter proceeded to a week and one-half long trial concluding on August 7, 1996. At trial, American Express presented the testimony of two appraisers: Louis W. Frillman (Frillman), owner of Marquette Advisors, and Karen E. Rubin (Rubin), senior vice president of Hospitality Valuation Services. The appraisers agreed that the property's highest and best use as improved was as an income-producing conference center. Because income-producing property is typically purchased as an investment for which earning power is a critical element affecting property value, both appraisers placed primary emphasis on the income approach to value. Frillman utilized a discounted cash flow analysis for his cost approach, which entails forecasting the expected annual cash flows for a property over a specified holding period and then discounting them at a required rate of return to derive an indication of present value. Frillman used historical operating data, industry averages, and market considerations to forecast Oak Ridge's expected annual cash flows over an eleven-year period and arrived at an income valuation of $10,100,000 in 1994, $10,900,000 for 1995, and $12,400,000 for 1996. In contrast, Rubin utilized the direct capitalization method for her income approach, in which a single year's income is divided by an income rate or multiplied by an income factor to reach an indication of value. Rubin adjusted Oak Ridge's historical room rate and occupancy data according to data from local competitors and income and expense data according to industry averages to derive an income valuation of $8,800,000 for 1994, $10,000,000 for 1995, and $12,000,000 for 1996.

Both appraisers also performed the market comparison approach, in which an estimate of market value is derived from the sales prices of comparable properties. Frillman compared the sales prices of two local properties, whereas Rubin performed a nationwide sales comparison to arrive at a range of values for each valuation year. However, both appraisers largely discounted this approach due to the lack of a sufficient quantity of data from comparable properties. Rubin did not prepare an estimate of value under the cost approach because she determined that the approach would not be used by a knowledgeable buyer. Frillman performed a cost-based valuation using the Marshall Valuation Service index to estimate replacement cost, but ascribed it little weight because he found that the construction was not consistent with market demands. In his valuation, Frillman found no functional obsolescence, but attributed external obsolescence to the property ranging from approximately 42 to 57% of the total replacement cost on the basis of the property's net operating losses and arrived at a cost valuation of $9,500,000 for 1994, $10,600,000 for 1995, and $12,800,000 for 1996.

Frillman's final estimate of market value for Oak Ridge was identical to his estimate under the income approach. Rubin's final market value estimate varied only slightly from her estimate under the income approach; she arrived at a final valuation of $9,000,000 for 1994, $10,300,000 for 1995, and $12,200,000 for 1996.

Carver County presented the testimony of Michael Buchalski, a partner in the firm Buchalski, Reynolds and Brodowski, and Joseph Novelli, a consultant to the firm, regarding their appraisal of Oak Ridge. Buchalski and Novelli performed all three approaches to value, but gave no weight to the sales comparison approach because they found the range of sales prices was too broad to give any indication of value. Buchalski found that, based on his analysis of the cost savings to American Express, the highest and best use of the property as improved was as a corporate conference center. However, Buchalski also performed an estimate of Oak Ridge's market value as a hybrid corporate/conference center under the income approach. Using the income capitalization method, Buchalski adjusted the historical occupancy and room rates of Oak Ridge to meet market expectations using data from local competitors and arrived at values of $25,840,000 for 1994, $26,670,000 for 1995, and $27,110,000 for 1996.

Finally, Buchalski analyzed the value of Oak Ridge under the cost approach. He used the Means Assemblies and Mean Construct Cost Data Manuals to estimate the reproduction cost, which is the estimated cost to construct an exact duplicate of the building at current prices. Buchalski's reproduction cost estimate included a 5% construction contingency and no allowance for functional or economic obsolescence. In his cost approach, Buchalski used the land valuation performed by Paul Bakken, principal in the real estate valuation firm of Bakken Liedl, Inc. Bakken found that the highest and best use of the land valued as vacant was for light industrial or corporate campus use. Under this use, he deemed sixteen acres of the subject property excess and valued this land separately and at a higher rate than the main parcel. Buchalski used these land values in his cost approach to value Oak Ridge at $30,090,000 for 1994, $29,980,000 for 1995, and $30,130,000 for 1996.

Buchalski relied primarily on the cost approach to derive his final valuation figures of $30,500,000 for each of the three valuation years. He premised this reliance heavily on his opinion that Oak Ridge was a special purpose property. This opinion was contrary to the views of Frillman and Rubin, who both testified that Oak Ridge was not a special purpose property.

The tax court found that Oak Ridge was a special purpose property because it was specifically designed to meet the needs of American Express' financial planner training program and is marketed only to uses consistent with that program. While the court determined that this finding supported the use of the cost approach to value Oak Ridge, it also considered the market data and income approaches before deciding to rely solely on the cost approach. The tax court rejected both the income and market approaches on the basis that there was insufficient data with which to make a meaningful valuation. In utilizing the cost approach, the tax court relied on Buchalski's reproduction cost analysis, rather than Frillman's replacement cost analysis. The court agreed with Buchalski that there was no external obsolescence, but discounted his estimate of entrepreneurial profit and increased the deduction for physical depreciation to arrive at an estimated market value of $27,300,000 for each assessment year.

II.

American Express raises a number of issues on appeal. The first issue before us is whether the tax court's finding that Oak Ridge was a special purpose property was clearly erroneous. We hold that it was.

This court has carefully defined the "special purpose property" classification:

Special purpose property is property that is treated in the market as adapted to or designed and built for a special purpose. This definition combines both functional and structural aspects: a special purpose property becomes such either by its use for unique functions or by its distinctive, specially-designed structural details.

Federal Reserve Bank of Minneapolis v. State of Minnesota, 313 N.W.2d 619, 621-22 (Minn. 1981) (emphasis added). A structure does not qualify as a special purpose property simply because it was built for a particular purpose. Rather, a special purpose property is one that, due to its unique function or design, is not likely to be sold on the market and cannot readily be converted to other uses without a large capital investment or a substantial loss in the investment value of the property's special features. Id. at 622-23. For example, in Federal Reserve, we upheld the tax court's determination that the Federal Reserve Building was a special purpose property because 15% of the building's area was devoted to a bank vault which could never be converted to other uses and at least 60% of the building was "super constructed" with high-cost materials and features such as a firing range, heliport, and an underground level for coin storage and pallet trucks which would lose substantial investment value if converted to general office uses. Id. Similarly, in McCannel v. County of Hennepin, we upheld the tax court's determination that Northwest's airport facility was a special purpose property because of the substantial loss in the investment value of features particular to its use as an airport facility which would occur if the facility were used for any other purpose. 301 N.W.2d 910 (Minn. 1980).

In finding that Oak Ridge was a special purpose property, the tax court relied heavily on the fact that it was designed and operated to meet the specific needs of American Express' financial planner training program, rather than to maximize use by third parties. While the tax court is correct that certain features were installed based on the particular needs of the financial planners, such as double-occupancy rooms and small, "breakout" conference rooms, these features do not inhibit the use of the property for other general hospitality uses. Oak Ridge's adaptability to other uses is evidenced by its active competition with other local conference centers for third-party business. Indeed, Oak Ridge rents more of its space for non-training uses than for its designed use. Additionally, the fact that the conference and guest rooms were constructed with non-load-bearing walls designed to be easy to move indicates that the facility could be altered with minimal capital expenditure.

The Oak Ridge facility is inapposite to the facilities we deemed special purpose properties in Federal Reserve and McCannel because there is no loss in investment value in renting the facility to third parties for conferences. Third parties utilize the conference rooms, guest rooms, and other facilities in the same manner as the financial planners. Furthermore, Oak Ridge's decision to market to third parties which are compatible with American Express' training program is not the type of use restriction which would render the facility a special purpose property. Because neither the design nor the current use of the conference center renders Oak Ridge incapable of being adapted to other uses, we conclude that the tax court clearly erred in finding that the facility was a special purpose property.

III.

The next issue before us is whether the tax court's reliance on the cost approach as the sole method of valuing Oak Ridge was an abuse of discretion. This court recognizes the three traditional approaches to determining the market value of real estate: 1) the market comparison, based on the prices paid for comparable properties in market transactions; 2) the cost approach, founded on the proposition that an informed buyer would pay no more for the property than the cost of building a new property with the same utility as the subject property; and 3) the income approach, predicated on the capitalization of the income the property is expected to generate. Equitable Life Assur. Soc. of U.S. v. County of Ramsey, 530 N.W.2d 544, 552 (Minn. 1995). We have stated that, whenever possible, the court should apply at least two approaches to market value because the alternative value indications derived can serve as useful checks on each other. Equitable Life, 530 N.W.2d at 553. Furthermore, because Minnesota law requires that every assessor "consider and give due weight to every element and factor affecting the market value" of real property for purposes of taxation, Minn.Stat. § 273.12 (1996), we have stated that circumstances will rarely warrant giving weight to only one approach to value. Equitable Life, 530 N.W.2d at 554. The cost approach is an imprecise measure which tends to inflate the value of a property because of the difficulty in making appropriate deductions for functional and economic obsolescence, especially for older buildings. See Montgomery Ward Co., Inc. v. County of Hennepin, 450 N.W.2d 299, 303 (Minn. 1990). See also The Appraisal Institute, The Appraisal of Real Estate 339 (11th ed. 1996) ("Because the estimation of depreciation and entrepreneurial [profit] is difficult, the cost approach may be of limited usefulness in valuing older improved properties."). For this reason, we have stated that it is a useful method for putting a ceiling on the value of the property being appraised. Montgomery Ward, 450 N.W.2d at 303. In the narrow context of special purpose properties, we have generally permitted the use of a cost approach alone because, "[t]he very nature of special purpose property is such that market value cannot readily be determined by the existence of an actual market, and therefore other methods of valuation, such as reproduction cost, must be resorted to." McCannel, 301 N.W.2d at 924. However, as we have held that the subject property is not a special purpose property, reliance upon the cost method as the sole measure of value would be justified only in the absence of valuable income and market comparison data.

The tax court found that it had insufficient information to estimate value under either a market or an income approach. In examining the market data, the court rejected the two local properties used in Frillman's market sales comparison: the Radisson, because it was a sale under distress, and the Airport Hilton, because it was not a comparable property. It also rejected the range of market sales prices presented by Rubin based on her own statement that the differences in the properties prevented her from using them to derive a precise indication of Oak Ridge's value.

In considering the income approach, the tax court recognized that a potential purchaser could view Oak Ridge as an investment property, in which case it would be appropriate to determine value by capitalizing the net operating income. The tax court noted that Oak Ridge's income and expenses would have to be adjusted to reflect market conditions because the facility is not operated to maximize income or minimize expenses. However, the tax court then rejected Oak Ridge's local competitors as sources of adjustment data due to the fact that they were operated and marketed differently than Oak Ridge. The tax court also rejected calculations based on national averages for conference centers and other national average data as insufficiently specific to provide adjustments. American Express contends that the tax court erred both in its rejection of the market comparison approach and the income approach. American Express argues that, as it presented the testimony of two appraisers with personal knowledge of the Scanticon sale, Frillman and Robert Strachota, who stated that the sale was not under distress, it should have been considered a comparable property under the market approach. American Express also argues that the tax court improperly ignored valuable data from Oak Ridge's competitors which could be used to perform an income approach simply because there would be some difficulty in making the adjustments.

While we might agree that there was information with which the tax court could have performed a market comparison approach of some limited value, we hold that the tax court's determination that the market data was insufficient was not clearly erroneous. Rubin, American Express' own appraiser, discounted the validity of the Scanticon sale as a comparable and the value of the market data as a whole to provide a reliable indication of Oak Ridge's value. In light of this conflicting testimony, we cannot state that the tax court's determination was against the weight of the evidence.

However, we hold that the tax court's finding that the income data was insufficient to perform a useful income approach was against the clear weight of the evidence. Each of the three appraisers who testified at trial performed and relied on an income approach to valuation in assessing Oak Ridge's value. Furthermore, two of the appraisers, Rubin and Buchalski, used an income approach which adjusted Oak Ridge's occupancy and room rates based on data from local competitors to account for the fact that the facility was not operated to maximize profits and minimize expenses. The tax court was not bound to accept the valuation of either appraiser. However, the reasoning the court presented for rejecting the data from local competitors which it could have used to perform its own income approach was illogical. The tax court rejected rate and occupancy data from local competitors because the competitors were operated as for-profit executive conference centers without restrictions on marketing. However, this difference in operation is precisely the reason that the competitors' data is valuable in adjusting Oak Ridge's historical performance to reflect market conditions. Differences in amenities at each conference center may be accounted for and are not so great as to render the data from the competitors inapplicable.

We also disagree with the tax court's determination that the data concerning national averages of income and expenses for conference centers was insufficiently specific to make adjustments to Oak Ridge's income and expenses. In performing her income approach, Rubin utilized a detailed composite operating statement based on industry averages. She also testified at length as to her consideration of industry averages in adjusting Oak Ridge's income and expense ratios under the income capitalization approach. In addition, the national average data contained in Buchalski's valuation report is broken down into income and expense categories which are almost identical to those used on Oak Ridge's spreadsheets. In light of this wealth of local and national income data, the tax court's determination that there was insufficient evidence to perform an income approach was clearly against the weight of the evidence. Moreover, considering that there was ample income data available and the fact that Oak Ridge is not a special purpose property, the tax court's decision to rely solely on the cost approach was an abuse of discretion.

IV.

The next issue raised by American Express is whether the tax court determined the value of the subject property according to value-in-use rather than market value. Minnesota law requires that real property be assessed at its market value. Minn.Stat. § 273.11, subd. 1 (1996). Fair market value for property assessment purposes is the compensation which a willing purchaser not required to buy the property would pay to an owner willing but not required to sell it, taking into consideration the highest and best use of the property. Ferche Acquisitions, Inc. v. County of Benton, 550 N.W.2d 631, 634 (Minn. 1996); see also Minn.Stat. § 272.03, subd. 8 (1996). Value-in-use, in contrast, is the value of the property to the present owner and may be considered by the assessor only to the extent that such present use reflects the property's value to buyers in the marketplace. Space Center, Inc. v. County of Hennepin, 302 N.W.2d 17, 21 (Minn. 1981).

American Express argues that the tax court wrongly determined the value of Oak Ridge according to its value to the parent company, rather than its value to a potential market purchaser. In support of this claim, American Express points to the tax court's statement in support of its decision to rely on the cost approach that "The specific design and operational choices made by Petitioner result in a facility which is more valuable to Petitioner than to an investor." American Express also argues that the tax court's focus on value-in-use is also demonstrated in its failure to analyze whether the features of the facility which were designed particularly for American Express' training program added or detracted from the value of the property in the general marketplace.

We agree with American Express that the tax court improperly determined the value-in-use of Oak Ridge, rather than its market value, in making its cost valuation of the property. The tax court relied on its determination that Oak Ridge was designed, built, and operated as a corporate conference center to find that the facility experienced no external obsolescence. Yet, the only evidence offered to show that Oak Ridge would be more valuable to a purchaser as an owner-occupied property than as an investment property was Buchalski's testimony that the highest and best use of the facility was as an owner-occupied corporate conference center, which was based on the cost savings to American Express. Indeed, the tax court itself recognized that a typical purchaser in the marketplace might view Oak Ridge as either an in-house training facility or an investment property. Therefore, to properly analyze the value of the property on the market, the tax court should have considered whether there was any functional or economic obsolescence based on the property's use as an executive conference center which generates income or, as it is currently used, as a dual-use property. Given that the tax court itself pointed out many features of the facility which render it less desirable than other conference centers for third-party business, including the facility's room sizes, amenities, and location, the tax court should have considered these features in determining whether there was any functional or economic obsolescence in using the facility as an investment or dual-use property. Additionally, the tax court should have considered the evidence of economic obsolescence presented by American Express based on the financial losses of the facility. Because the tax court failed to analyze the data from the perspective of a typical purchaser and instead, focused on the value of the facility to American Express, we hold that the court erred as a matter of law in determining the value of Oak Ridge according to its value-in-use.

V.

American Express also attacks the tax court's use of reproduction cost, rather than replacement cost, in making its cost approach valuation. Replacement cost is the estimated cost to construct a building with an equivalent utility to the building being appraised, at current prices, using modern materials, standards, design and layout. Reproduction cost, on the other hand, is the estimated cost to construct an exact replica of the subject property using the same materials, standards, design and layout and embodying all the deficiencies and obsolescence of the subject building. The Appraisal of Real Estate at 345. Replacement cost is generally lower and may provide a better indication of current value than reproduction cost because it does not embody obsolescent features which would not be constructed in a new building. However, reproduction cost can be adjusted to account for functional obsolescence through appropriate deductions. Id. at 370-71. Both methods have been used to build cost approaches to value in our courts. See, e.g., Federal Reserve, 313 N.W.2d at 623-25 (Minn. 1981) (using reproduction cost); Northwest Racquet Swim Health Clubs, Inc. v. County of Dakota, 557 N.W.2d 582, 585 (Minn. 1997) (using replacement cost). Provided that proper deductions for functional obsolescence are made, it is not error to use reproduction cost in estimating value under the cost approach.

VI.

Finally, American Express argues that the tax court improperly included several cost categories in its estimate of reproduction cost and incorrectly used the market value of the land valued as vacant in its cost approach. Specifically, American Express contends that the tax court wrongly included a construction contingency, construction period interest on 100% of the hard costs, and items claimed to be personal property in the amount of $292,137 in its estimate of reproduction cost. Additionally, American Express claims that it was improper to use the estimated market value of the land valued as vacant in performing the cost approach when the land had a different highest and best when valued as vacant than valued as improved. We reject these challenges and hold that the inclusion of each of these items was within accepted appraisal practice.

Reproduction cost requires the consideration of all costs, including capital expenditures for direct costs such as labor and materials, and indirect costs such as financing and marketing costs. The Appraisal of Real Estate, at 345-47. Construction contingencies and construction period interest are indirect costs of construction which would be encountered in the construction of a new building and must be accounted for in the reproduction cost. Fixtures are considered part of the property and must also be included in the reproduction cost. Id. at 9. Although American Express claims that certain kitchen equipment, a walk-in refrigerator, and a walk-in freezer were improperly included in the reproduction cost, it presents no evidence that these items were personal property rather than fixtures. Without such evidence, we will not disturb the tax court's inclusion of the items in the cost estimate. We also hold that the use of Bakken's land appraisal based on the highest and best use of the land vacant, which valued a portion of the land surrounding the conference center as excess, was proper. A cost analysis requires that the value of the site be added to the total depreciated cost of all improvements to arrive at the value of the property. Id. at 340. Site value is determined according to the value of the land based on its highest and best use as though vacant. Id. at 323. Therefore, the tax court did not err in using Bakken's land appraisal based on the highest and best use of the land as vacant in its cost valuation.

Affirmed in part, reversed in part and remanded for further findings consistent with this opinion.

BLATZ, J., took no part in the consideration or decision of this case.