900 Walnut Owner LLC v. Michael Dauphin, Assessor, City of St. Louis

STATE TAX COMMISSION OF MISSOURI

900 WALNUT OWNER LLC,	)
Complainant,	)	Appeal No.19-20163
	)	Parcel No. 0195-00-0021-0
	)
v.	)
	)
MICHAEL DAUPHIN, ASSESSOR,	)
CITY OF ST. LOUIS, MISSOURI	)
Respondent.	)

 

DECISION AND ORDER

            900 Walnut Owner LLC, (Complainant) appeals the City of St. Louis Board of Equalization’s (BOE) decision finding the true value in money (TVM) of the subject commercial property on January 1, 2019, was $22,410,900. Complainant produced substantial and persuasive evidence showing the TVM of the subject property on January 1, 2019, was $17,000,000. The BOE decision is SET ASIDE.^[1]

Complainant is represented by attorney Peter Corsale. Respondent is represented by attorney Abby Duncan. Both parties filed a post-hearing brief and a reply brief.

 

FINDINGS OF FACT

1. The Subject Property. The subject property is commercial real property located at 900 Walnut Street in the City of St. Louis, Missouri. The subject property consists of a 0.61 acre site improved with seven-story building with approximately 127,000 square feet of gross building area. (Ex. 2 at vi; Ex. A at 70) The building was last renovated in 2018, when the owner completed approximately $4,000,000 of electrical and mechanical work and other building upgrades. (Ex. 2 at 35; Ex. A at 104)

As of January 1, 2019, the building was used as a carrier hotel data center. A carrier hotel data center serves as a connection point for internet service providers to facilitate data exchange. (Ex. 2 at 35; Ex. A at 16) The subject property is configured to provide suites from which tenants operate private data centers. Tenants may connect to a “Meet Me Room” (MMR), which is a dedicated space where fibers are brought into the data center and made available to tenants for additional connectivity. An MMR is considered a business service separate from the real property. (Ex. 2 at 38; Tr. 62:23-25)

Data centers require a substantial, uninterrupted power supply to maximize the time data services and connections are online, referred to as “uptime.” The amount of uptime is a significant valuation factor in the data center market. (Ex. 2 at 45-47; Ex. B at 26)   The specialized services and equipment associated with a data center, in conjunction with the importance of uptime, requires a physical configuration unique from other commercial properties. Much of the equipment in a data center must be placed on raised floors to permit air flow and the installation of climate control systems necessary to maintain constant temperature. Raised floors are also necessary to enable wiring and connections to the servers. (Ex. A at 13) The subject property has 53,635 square feet of raised floor space. (Ex. A at 70)

Most data centers also utilize redundant systems to maximize uptime in the event of interruptions in the primary power, cooling or connectivity systems. (Ex. 2 at 45-47; Ex. A at 12-13) These redundant systems can have limited life spans relative to the systems in typical office or industrial buildings. Additionally, technological changes may result in accelerated obsolescence and depreciation. (Ex. 2 at vii; Ex. B at 26)

The importance of uptime is also reflected in the classification of data centers into four tiers according to their capabilities, with particular emphasis on power supply and system redundancy. (Ex. 2 at 45-47; Ex. A at 15) A Tier 1 data center has the least capability in terms of power supply and system redundancy, while a Tier 4 data center has the most capability. For instance, a Tier 1 facility has a single, non-redundant path for power and cooling distribution. By contrast, a Tier 4 facility has multiple redundant power and cooling distribution paths. The net result is that the annual uptime for a Tier 4 facility is typically 99.995% (26.3 minutes annual downtime) while the annual uptime for a Tier 1 facility is typically 99.671% (28.8 hours annual downtime). (Ex. 2 at 47; Ex. A at 15)

Given the market emphasis on uptime, the tier levels are an accepted basis for comparing properties. (Ex. A at 15) The subject property is a Tier II facility. (Ex. A at 102)   Critically, the main power source for the subject property has a single point of failure because the power lines servicing the data center come from a single electrical substation. (Tr. at 74:7-11)

In addition to the capability-focused tier level, data centers are also compared by market location. The different markets are referred to as primary markets, secondary markets, tertiary markets, and emerging markets. (Ex. A at 24-25) St. Louis is a tertiary market. The consensus among market participants is that the subject property’s location in the tertiary St. Louis data center market is a limiting factor. (Ex. A at 121; Ex. 2 at vii)

2. Assessment and Valuation. The BOE determined the TVM of the subject property on January 1, 2019, was $22,410,900.
3. Complainant’s Evidence. Complainant introduced Exhibits A through E.   All of these exhibits were admitted into evidence. The exhibits are as follows:

Exhibit A	Appraisal report prepared Kevin O’Bryan
Exhibit B	O’Bryan’s written direct testimony
Exhibit C	O’Bryan’s written rebuttal testimony
Exhibit D	Appraisal report for the subject property prepared in June 2020 by Respondent’s appraiser, Christopher Fudacz, on behalf of another client (Bank Appraisal).
Exhibit E	Appraisal report prepared by Fudacz’s predecessor, John Parsons, as part of the appeal from the 2017 assessment of the subject property (2017 City Appraisal)

 

Exhibit A is an appraisal report prepared by Kevin O’Bryan. O’Bryan is a licensed Missouri real estate appraiser and has testified as an expert in Missouri and Kansas. O’Bryan received the MAI designation from the Appraisal Institute in 2003. (Ex. A at 193-94)

Fudacz prepared the Bank Appraisal in June 2020 for another client. The Bank Appraisal valued the leased fee interest in the subject property as of December 31, 2019. In the Bank Appraisal, Fudacz noted the actual occupancy rate in 2019 was 70.9% and estimated a stabilized occupancy rate of 72% as of December 31, 2019. Fudacz’s Bank Appraisal did not apply a lease-up discount. (Ex. D at 72, 100)

The 2017 City Appraisal was prepared by Fudacz’s predecessor, John Parsons, as part of the appeal from the 2017 assessment of the subject property. The 2017 City Appraisal concluded the subject property’s actual occupancy rate on January 1, 2017, was 84.2%, with a stabilized occupancy of 90%. (Ex. E at 85)

Complainant introduced the Bank Appraisal and the 2017 City Appraisal “for the limited purpose of challenging the methodology resident in the City Appraisal.” (Compl. Br. at 7) Specifically, Complainant notes Fudacz deducted deferred maintenance in the Bank Appraisal but did not do so in Exhibit 2. Complainant further notes that in Exhibit 2, Fudacz utilized many of the same comparable sales as used by Mr. Parsons in the 2017 City Appraisal. Complainant asserts these issues render Fudacz’s appraisal report and testimony unreliable and inadmissible. (Compl. Br. at 12, 15)

O’Bryan Appraisal

O’Bryan determined the highest and best use of the subject property as improved is continued use as a data center. (Ex. A at 108) O’Bryan’s value estimate is based on the income approach and sales comparison approach, with emphasis on the sales comparison approach. (Ex. A at 8, 189) O’Bryan concluded the indicated value from the income approach was $18,200,000 and the indicated value from sales comparison approach was $16,000,000. O’Bryan reconciled these indicated values and estimated the fair market value of the fee simple interest in the subject property was $17,000,000 as of January 1, 2019. (Ex. A at 189; WDT at 34)

Income Approach

O’Bryan utilized the income approach to support his sales comparison analysis. (Ex. A at 189) To determine the potential gross income (PGI) for the subject property, O’Bryan estimated market rent based on triple net and power-based leases. Triple net leases require the tenant to pay their share of all operating expenses. A power-based lease is typically on a modified gross basis requiring the owner to pay some operating expenses while the tenant pays for electricity. (Ex. A at 117) O’Bryan concluded the best unit of measure for the subject property’s real estate is a triple net lease based on net rentable area. (Ex. A at 123)

O’Bryan’s market rent estimate relied on market participant interviews, surveys, and data from his comparable sales analysis. The market participants indicated triple net lease rates generally range between $10 and $30 per square foot. (Ex. A at 117, 121-22) The net operating income (NOI) indications from data center sales “ranged from $5.41 to $29.89 per square foot with quartile indications of $12.26 to $19.43 per square foot and an overall average of about $16.29 per square foot annually.” (Ex. A at 118) O’Bryan noted that market lease rates for the subject property would likely be slightly lower because the average effective age of the sales considered was 26 years while the subject property’s effective age is approximately 34 years. (Ex. A at 118) O’Bryan estimated a “rate from near the middle of the overall range would be most appropriate under this type of lease structure, or about $13.00 to $18.50 per square foot.” (Ex. A at 119) These lease rates equate to approximately $1,570,000 to $2,230,000 in annual rent on a triple net basis. (Ex. A at 119)

O’Bryan considered power-based lease rates as a check on his triple net lease analysis. (Ex. A at 124) Surveys indicated power-based leases range from $80 to $175 per kilowatt per month. (Ex. A 119-23) Similarly, market participants indicated power-based lease rates ranged from $80 to $125 per kilowatt. O’Bryan estimated the average power-based lease was $135 per kilowatt for Tier 3, “turnkey” facilities and $110 per kilowatt for Tier 2 facilities. (Ex. A at 123) O’Bryan qualified these averages by noting the subject property operates as a carrier hotel data center offering considerably less infrastructure than turnkey facilities, so the lease rates for the subject “would obviously be much lower.” (Ex. A at 123) O’Bryan concluded the market rate for a power-based lease of the subject property would be $58 per kilowatt per month for 5 megawatts of power, which equates to a PGI of $3,480,000.[2] (Ex. A at 124)

O’Bryan calculated the effective gross income (EGI) by deducting market vacancy and collection loss. O’Bryan’s vacancy estimate was based on national surveys indicating a trend toward increasing data center vacancies, including a CBRE survey indicating vacancy rates of 15% in the fourth quarter of 2018. (Ex. A at 135-26) Based on this data, and the subject’s market location and characteristics, O’Bryan estimated a stabilized vacancy and collection loss rate of 15%. (Ex. A at 126) Deducting the vacancy and collection loss from the PGI yields an EGI of $2,958,000. (Ex. A at 133)

To estimate market expenses, O’Bryan relied on national surveys and market participants to calculate a range of expenses based on a percentage of the subject property’s EGI. Public records from a publicly traded Real Estate Investment Trust (REIT) with a portfolio of approximately 225 data centers indicated expenses ranged from about 24.4% to 32.8% of total income, exclusive of reimbursements. (Ex. A at 127) Market participants indicated expenses generally equate to $25 to $40 per kilowatt per month, or approximately 15% to 30% of the power-based, modified gross lease rate. (Ex. A at 128) O’Bryan estimated market-based expenses for the subject property would total $6.75 per rentable square foot, or 27.5% of the EGI. O’Bryan also estimated replacement reserves of 4.50% of EGI and noted a special assessment expense of $126,171, which accounted for 4.27% of the EGI.[3] The total annual expense estimate was $1,070,000, which equates to 36.27% of the EGI of $2,958,000. (Ex. A at 133)

Deducting the estimated 15% vacancy rate and estimated $1,070,000 in annual expenses results in an estimated NOI of $1,890,000, or an estimated triple net equivalent lease rate of $15.67 per square foot ($1,890,000/120,586 sq. ft. = $15.67/sf). (Ex. A at 124, 132) O’Bryan also noted “the $1,890,000 net income generated from the power based rental analysis is at or near the average of the $1,570,000 to $2,230,000 range of the facility based triple net lease rate projections, suggesting limited risk in attaining that projection in the market.” (Ex. A at 124)

O’Bryan estimated the capitalization rate by consulting market participants, surveys, and overall rate indications from recent data center sales. (Ex. A at 134-145) Market participants familiar with the subject property indicated the “overall rate indications for the subject ranged from the middle 6% to 9% range with a median indication in the lower 7% to lower 8% range.” (Ex. A at 145) O’Bryan concluded that “[g]iven the historical operation of the subject and its current and anticipated occupancy, and, further considering the data cited and given our broader experiences in the market, an overall rate of 7.75% is supported (prior to consideration of any tax load).” (Ex. A at 146) O’Bryan added an effective tax rate of 2.613% to estimate an overall, loaded capitalization rate of 10.363%. (Ex. A at 147) Dividing the unrounded NOI ($1,885,270) by the overall capitalization rate (.10363) yielded an estimated direct capitalization value of $18,192,982, which O’Bryan rounded to $18,200,000. (Ex. A at 147)

Sales Comparison Approach

            O’Bryan emphasized the sales comparison approach to support his final value estimate of $17,000,000. O’Bryan’s initial step utilized a “trending analysis considering all data center sales that have taken place from 2016 to 2020.” (Ex. A at 150) O’Bryan utilized the trending analysis “to assist in identifying any significant functional limitations of the subject and to assist in providing support for the adjustments made for a variety of property features and physical attributes.” (Ex. A at 150) The trending analysis data set included 191 arm’s-length transactions. O’Bryan utilized multiple indicators, such as price per square foot, and determined “the minimum, maximum, and average indicators, as well as the 1st and 3rd quartile indications.” (Ex. A at 151) Arranging the data in quartiles “demonstrates central tendencies and mitigates some of the risk of a few outliers impacting the indicators.” (Ex. A at 151)

Next, O’Bryan filtered the 191 sales utilized in the trending analysis by multiple characteristics, such as building size, raised floor and critical power space, power per square foot, and tier level. This filtering pared the initial data set of 191 sales down to 61 sales that “better reflect the current market and those properties that are most comparable to the subject.” (Ex. A at 159)

To perform a sales comparison analysis, O’Bryan filtered the 61 sales to identify sales involving properties with characteristics similar to the subject property and which O’Bryan deemed most relevant to value, specifically: (1) Price per square foot (61 sales); (2) Age (35 sales); (3) Tier Rating (13 sales); (4) Use/Utilization (8 sales); (5) Critical/Raised Floor Area (61 sales); (6) Available Power (17 sales); and (7) Power Density (21 sales). (Ex. A at 167-86) O’Bryan testified these characteristics “are critical features that are commonly recognized in the market” as important to market participants.   (Tr. at 79:22-25; 80:1-3) O’Bryan developed a value indication based on each of these seven characteristics by determining the average, lower quartile (25^th percentile), and upper quartile (75^th percentile) value indications for each of these seven characteristics. O’Bryan then performed a sales comparison analysis of the value indications for each of the seven characteristics by comparing them to the same characteristics of the subject property. (Ex. A at 167-186) As part of his comparative analysis, O’Bryan adjusted each value indicator along seven parameters to account for differences between the subject property and the comparable properties. (Ex. A at 167-186)

For instance, to develop a value indication based on price per square foot, O’Bryan determined the average price per square foot was $178 per square foot. The lower quartile cutoff was $99 per square foot and the upper quartile cutoff was $258 per square foot. O’Bryan then determined adjustments for: (1) time of sale (-1.5%); (2) location (no adjustment); (3) size (no adjustment); (4) tier rating (-14% adjustment); (5) critical floor space ratio (-1%); (6) power per square foot of critical floor space (-5%); and (7) the age and condition of the subject relative to the average comparable (-8%).  Accounting for all adjustments, and assigning 50% weight to the average and 25% to the lower and upper quartiles, O’Bryan estimated the adjusted price per square foot of the subject property is $125.80 per square foot, which equates to a value estimate of approximately $15,170,000. (Ex. A at 170, 187).[4]

O’Bryan performed the same analysis for each of the remaining six value indicators; i.e., age, tier rating, use, raised floor area, available power, and power density. (Ex. A at 171-187) The final reconciliation is as follows:

Value Indication          Reconciled Value       Weighting            Contributory Value

Price/SF overall	$15,170,000	25%	$3,792,500
Price/SF similar age	$17,440,000	20%	$3,488,000
Price/SF similar tier	$13,510,000	15%	$2,026,500
Price/SF similar use	$13,030,000	15%	$1,954,500
Price/SF raised floor	$18,060,000	10%	$1,806,000
Price/SF MW power	$20,950,000	7.50%	$1,571,250
Price/SF power density	$17,990,000	7.50%	$1,349,250
Overall Range	$16,592,857	100%	$15,988,000

 

O’Bryan noted the “value indications exhibit a modest range” due to the nature of the data center market and the lack of transparency. (Ex. A at 188) O’Bryan concluded the sales comparison approach supported “a final value of $16,000,000 … or about $133 per square foot.” (Ex. A at 188) Based on the income approach, and with emphasis on the sales comparison approach, O’Bryan estimated the TVM of the subject property as of January 1, 2019, was $17,000,000.

4. Respondent’s Evidence. Respondent introduced Exhibits 1, 2, 4, 6 and 8.

All of these exhibits are admitted into evidence. The exhibits are as follows:

Exhibit 1	Written direct testimony of Christopher Fudacz
Exhibit 2	Appraisal report prepared by Christopher Fudacz
Exhibit 4	Fudacz’s written rebuttal testimony
Exhibit 6	Signature page showing the date Respondent engaged Fudacz to appraise the subject property. The entire agreement is reproduced in Addendum D to Exhibit 2.
Exhibit 8	Fudacz’s written sur-rebuttal testimony.

 

Fudacz Appraisal

Fudacz is CBRE Advisory Services’ National Data Center Practice Leader for valuations and a licensed Missouri real estate appraiser.  In Exhibit 2, Fudacz concluded the highest and best use of the property, as improved, is consistent with the existing use as a data center. (Ex. 2 at 79) Fudacz utilized the sales comparison approach and the income approach, with emphasis on the income approach. Fudacz concluded the indicated value was $23,400,000 based on the sales comparison approach and $23,300,000 based on the income approach. Fudacz reconciled these indicated values and estimated the fair market value of the fee simple interest in the subject property was $23,300,000 as of January 1, 2019. (Ex. 2 at 106)

Sales Comparison Approach

Fudacz utilized eight comparable sales to estimate the TVM of the subject property as of January 1, 2019, was $23,400,000. (Ex. 1 at 85-86) The eight sales are as follows:

Sale         Location             Square Feet       Sale Price/Price per SF     NOI per SF

1	Charlotte, NC	18,068	$8,260,000  / $457.16	$34.32
2	Austin, TX	62,237	$20,200,000 / $269.30	$15.75
3	Miami, FL	284,986	$84,000,000 / $294.75	$17.69
4	San Francisco, CA	105,800	$71,000,000 / $671.08	$36.24
5	Hopkins, MN	135,240	$19,900,000 / $147.15	$11.08
6	Boston, MA	150,967	$77,000,000 / $510.05	$28.15
7	Charlotte, NC	26,036	$10,000,000 / $384.08	$33.61
8	Las Vegas, NV	160,700	$37,000,000 / $230.24	$19.80

 

The eight sales ranged from $147.15 to $671.08 per square foot. Fudacz concluded the wide range of values made it “impossible to properly adjust these sales for differences in location and income characteristics due to the difficulty in accounting for the divergence in land values and economic profiles, as well as the influence of the demographic on rent.” (Ex. 1 at 84) On cross-examination, Fudacz confirmed he made no adjustments in his sales comparison approach. (Tr. 185: 1-5)

Rather than adjusting the sales to account for differences, Fudacz utilized the net income multiplier (NIM) method. A NIM is calculated by dividing the sale price per square foot by the NOI per square foot.   (Ex. 2 at 85) To estimate value, the NIM is multiplied by the NOI per square foot of the subject property. Fudacz testified the NIM accounts for different property characteristics because the rent a property generates is a function of those characteristics. (Tr. 119:10-16)

Fudacz concluded comparables 1, 2, 3, 5 and 8 best represent the subject property. Fudacz calculated the NIM for each of these sales and multiplied each NIM by the subject property’s estimated NOI of $17.33 per square foot. These calculations yielded indicated values ranging “from $201.52 to $296.32 per square foot.” (Ex. 2 at 85) “Based on the subject’s age and secondary St Louis location,” Fudacz concluded the subject property’s value would fall in “the mid-to-lower end of the range of the most comparable sales between $210.00 and $240.00 per square foot.”[5] (Ex. 2 at 85) Based the conclusion the subject property has 113,653 square feet of net rentable area (NRA), multiplying the NRA by the lower and upper NIM ranges results in an estimated value range of $23,867,130 (113,653 x $210.00) and $27,276,720 (113,653 x $240.00). Fudacz estimated an indicated value of $26,000,000 with a $2,560,000 lease up discount, yielding a rounded, indicated value of $23,400,000, or $205.89 per square foot. (Ex. at 86)

Sales 1, 2, 3, 5, and 8 include properties with different attributes than the subject property. The subject property is a Tier II facility in a tertiary market. (Ex. A at vii) Sales 1, 2, and 5 involved data centers in Charlotte, Austin, and Minneapolis. (Ex. 1 at 58)[6] The tertiary St. Louis data market is substantially smaller than these secondary markets, registering at one-fifth the size of the Minneapolis market, which is the “smallest secondary market tracked by CBRE’s Data Center Solutions group[.]” (Ex. 1 at 72) (Minneapolis). Fudacz accounted for market location by selecting “the mid-to-lower end of the range of the most comparable sales between $210.00 and $240.00 per square foot.”

Unlike the subject property, which has no parking (Ex. 1 at 35), Sale 3 included an adjacent 396-space parking garage with quoted rates at $85 to $120 per month, per parking space. (Ex. 1 at 82) Unlike the subject property, Sale 3 also included two retail spaces within the “Class B building” located “in an ‘A’ location with good frontage along Biscayne Boulevard in the heart of Downtown Miami.” (Ex. 1 Addenda, Sale No. 3) Sale 8 involved the first LEED-certified building in Nevada and included an adjacent five-story parking garage. (Ex. 1 at 84) Fudacz made no adjustments for these property differences and income streams not shared by the subject property.[7]

Income Approach

Fudacz relied primarily on the income approach. To estimate market rent, Fudacz considered the leases of eight carrier hotel data centers in the central and eastern United States. Fudacz testified he chose these comparables because of “their overall similarities as highly connected data center[s], possessing highly connected data center characteristics.” (Tr. at 107:1-4) The comparables are located in “primary, secondary and tertiary markets, with a focus on secondary and tertiary markets.” (Tr. at 107:5-7) “The comparables reflect a range from $30.50 to $74.50 per square foot gross for space ranging from 397 to 17,000 square feet; and a NNN range from $19.00 to $27.00 per square foot for space ranging from 6,000 to 135,590 square feet.” (Ex. 1 at 87-88)

Fudacz gave most weight to comparables 1, 3, 6, and 7, located in Baltimore, Houston, Miami, and Boston, respectively. The buildings on these properties were built between 1925 and 1985, and had lease rates from $27.00 to $52.14 per square foot. (Ex. 1 at 88) Due in part to “the secondary St. Louis location,” Fudacz concluded “it is reasonable to expect the subject to fall into the mid-to-lower end of the range of the gross deals.” (Ex. 1 at 88) Based on the “comparable data, market participants and the in-place leases at the subject[,]” Fudacz concluded the market rent for the subject property was “$35.00 to $45.00 per square foot [full service gross], say $43.00.” (Ex. 1 at 89) Multiplying $43.00 by the Fudacz’s concluded net rentable area of 113,583 square feet yields a PGI of $4,887,079. (Ex. 1 at 91)

Fudacz testified that estimating market rent based on different types of leases requires adjustments for the reimbursement structure. (Tr. 196:1-17) Fudacz conceded no such adjustments are reflected in his appraisal report. (Tr. 196:22-25)

Fudacz estimated the EGI by subtracting an estimated 10% vacancy loss and adding $25,000 in “other income.” (Ex. 1 at 95) The 10% vacancy loss was based on an estimated 90% market occupancy rate based on the general “office market vacancy and historical occupancy at the subject.” (Ex. 2 at 93). Fudacz also considered “underwriting criteria of investors in the marketplace” indicating the long-term vacancy “should total 10% and be inclusive of a credit loss amount.” (Ex. 2 at 94) Although acknowledging the subject property was only 66.6% leased as of January 1, 2019, Fudacz concluded it would take 15 months to reach the stabilized occupancy rate of 90%. (Ex. 2 at 94, 104) Fudacz subtracted market expenses of $1,544,362 to estimate the annual NOI for the subject property was $2,876,509, or $25.31 per square foot. (Ex. 2 at 100-01)

Fudacz estimated a capitalization rate based on sales 1, 5, 7, and 8 utilized in the sales comparison approach. These sales produced capitalization rates ranging from 7.53% to 8.75%. Further, an investor survey indicated capitalization rates ranging from “5.50% to 9.00% for research and development properties.” (Ex. 1 at 103) Fudacz estimated the capitalization rate for the subject property would fall in “the upper end” of both ranges due to the subject’s age and secondary St Louis location. (Ex. 1 at 102, 103) The final, unloaded capitalization rate estimate was 8.00%, with a loaded rate of 11.314% after adding the effective tax rate of 3.14%.   (Ex. 1 at 103)

Dividing the NOI ($2,876,509) by the loaded capitalization rate (0.114) yields an estimated value of approximately $25,800,000. (Ex. 2 at 105) To determine the final value estimate, Fudacz deducted lease-up costs on the assumption a 15-month lease-up period would result in 90% occupancy. Fudacz calculated the lease-up costs by adding lost rent, tenant improvement costs, leasing commission, and lost profit, for a total lease-up discount of $2,560,000. (Ex. 2 at 104) Subtracting the lease-up costs results in a final indicated value of $23,300,000 (rounded). (Ex 2 at 105)

Carson Walbridge Report

            Fudacz attached as an addendum to Exhibit 2 a “Conditions Assessment For Data Center Use” report prepared by Carson Walbridge, a consulting firm. The 18-page report prepared in April 2016 details the condition of the data center building and mechanical systems. The Carson Walbridge report identified roof curb and dunnage repairs as a “Priority 2 potentially critical” item that should be corrected “expeditiously” within a year to avoid service interruptions or building damage. The report further recommended a new chilled water system as a “Priority 3 Necessary – Not Yet Critical” item that should be addressed within two to five years. Finally, the report identified seven items as “Priority 4” and recommended improvements “to reduce long term maintenance and/or improve the quality level of the property.” (Ex. 2, Addendum at 18) Unlike the Bank Appraisal prepared in June 2020, Fudacz did not deduct deferred maintenance from his value estimate from his October 2020 appraisal report in Exhibit 2.

5. Value. The TVM of the subject property on January 1, 2019, was $17,000,000.

CONCLUSIONS OF LAW

1. Assessment and Valuation. Commercial real property is assessed at 32% of its TVM as of January 1 of each odd-numbered year. Section 137.115.5(1)(c). “True value in money is the fair market value of the property on the valuation date, and is a function of its highest and best use, which is the use of the property which will produce the greatest return in the reasonably near future.” Snider v. Casino Aztar/Aztar Mo. Gaming Corp., 156 S.W.3d 341, 346 (Mo. banc 2005) (internal quotation omitted). The fair market value is “the price which the property would bring from a willing buyer when offered for sale by a willing seller.” Mo. Baptist Children’s Home v. State Tax Comm’n, 867 S.W.2d 510, 512 (Mo. banc 1993). “True value in money is defined in terms of value in exchange not value in use.” Tibbs v. Poplar Bluff Assocs. I, L.P., 599 S.W.3d 1, 7 (Mo. App. S.D. 2020) (internal quotation omitted).  “Determining the true value in money is an issue of fact for the STC.” Cohen v. Bushmeyer, 251 S.W.3d 345, 348 (Mo. App. E.D. 2008).

“For purposes of levying property taxes, the value of real property is typically determined using one or more of three generally accepted approaches.” Snider, 156 S.W.3d at 346. The three generally accepted approaches are the cost approach, the income approach, and the comparable sales approach. Id. at 346-48; see also St. Louis Cty. v. Sec. Bonhomme, Inc., 558 S.W.2d 655, 659 (Mo. banc 1977). The STC has wide discretion in selecting the appropriate valuation method but “cannot base its decision on opinion evidence that fails to consider information that should have been considered under a particular valuation approach.” Snider, 156 S.W.3d at 348.

2. Evidence. The hearing officer is the finder of fact and determines the credibility and weight of the evidence. Kelly v. Mo. Dep’t of Soc. Servs., Family Support Div., 456 S.W.3d 107, 111 (Mo. App. W.D. 2015). “Although technical rules of evidence are not controlling in administrative hearings, fundamental rules of evidence are applicable.” Mo. Church of Scientology v. State Tax Comm’n, 560 S.W.2d 837, 839 (Mo. banc 1977).

Complainant asserts Fudacz’s appraisal report and testimony should be stricken from the record. (Comp’l Br. at 8-15, 23) Complainant’s argument to strike Fudacz’s testimony is based on three assertions.

First, Complainant asserts Exhibit 2 is tainted by confidential information Fudacz obtained from Complainant while conducting the Bank Appraisal. Specifically, Complainant notes Fudacz deducted $3.2 million in deferred maintenance in the June 2020 Bank Appraisal but did not do so in the October 2020 appraisal report prepared for Respondent. Complainant asserts the fact Fudacz did not deduct deferred maintenance in Exhibit 2 – when he deducted it in the Bank Appraisal less than five months earlier – amounts to a tacit admission the deferred maintenance information was confidential, thus creating a conflict of interest.   (Comp’l Br. at 9 n.1)[8]

On cross-examination, Fudacz conceded that if deferred maintenance should have been deducted from the Bank Appraisal, it should have been deducted from the appraisal report in Exhibit 2. (Tr. 151:16-23) Complainant’s argument assumes the deduction of deferred maintenance is an established fact. Complainant’s argument is undermined by the fact that, like Fudacz, O’Bryan also did not deduct for deferred maintenance. Complainant’s argument that Fudacz’s failure to deduct for deferred maintenance shows a conflict of interest is undermined by the conclusions of Complainant’s own expert.

Second, Complainant asserts Fudacz’s failure to deduct for known deferred maintenance shows he ignored vital data, thus rendering his report unreliable, in violation of the Uniform Standards of Professional Appraisal Practice, and inadmissible as expert testimony pursuant to Section 490.065. (Compl. Br at 8-12) Complainant asserts O’Bryan “did not have access to or even know the existence of the CBRE property condition reports seen in Mr. Fudacz’s Bank Appraisal.” (Compl. Br. at 10 n.2) This argument is unpersuasive because Complainant, as the owner of the subject property, has both access to the subject property and incentive to provide its appraiser with all necessary information regarding deferred maintenance. The fact O’Bryan also did not deduct deferred maintenance undermines Complainant’s assertion that Fudacz’s lack of a deduction renders his opinion inadmissible. Under these circumstances, the differences between Fudacz’s Bank Appraisal and Exhibit 2 go to the weight of Fudacz’s testimony and report, not their admissibility.

Finally, Complainant argues Fudacz’s testimony and report are unreliable and inadmissible due to similarities between the data utilized in the Bank Appraisal and the 2017 City Appraisal. For instance, six of the eight data center sales Fudacz used for his comparable sales analysis in Exhibit 2 are the same sales Parsons used in the 2017 City appraisal. However, Fudacz used only one of these eight sales of office and data centers in the Bank Appraisal. Complainant asserts this selective “copy and paste” approach renders Fudacz’s report unreliable and inadmissible. (Comp’l Br. at 15)

The similarity in data between the 2017 City Appraisal and Exhibit 2 does not mean Exhibit 2 is so unreliable that it is inadmissible. Both Fudacz and O’Bryan testified obtaining reliable market information on data center transactions is difficult because the market lacks transparency. The data overlap between the 2017 City Appraisal and Exhibit 2 is not unreasonable given the relative lack of transparency and paucity of information regarding the specifics of the data center market. Even if Fudacz failed to analyze transactions that were more pertinent to January 1, 2019, valuation date, any such oversight would go to the weight of the evidence, not its admissibility. This conclusion is consistent with the general rule that questions regarding the “the sources and bases of the expert’s opinion affect the weight, rather than the admissibility, of the opinion, and are properly left to [the trier of fact].” Doe v. McFarlane, 207 S.W.3d 52, 62 (Mo. App. W.D. 2006).

Complainant’s argument that Fudacz’s report is inadmissible and should be stricken is unpersuasive. Fudacz’s appraisal report and testimony are admissible.[9]

3. Complainant’s Burden of Proof. The taxpayer bears the burden of proof and must show by a preponderance of the evidence that the property was misclassified or overvalued.  Westwood P’ship v. Gogarty, 103 S.W.3d 152, 161 (Mo. App. E.D. 2003).  The BOE’s valuation is presumptively correct. Tibbs, 599 S.W.3d at 7. The “taxpayer may rebut this presumption by presenting substantial and persuasive evidence that the valuation is erroneous.” Id. (internal quotation omitted). The taxpayer also must prove “the value that should have been placed on the property.” Id. “Substantial evidence is that evidence which, if true, has probative force upon the issues, and from which the trier of fact can reasonably decide the case on the fact issues.” Savage v. State Tax Comm’n, 722 S.W.2d 72, 77 (Mo. banc 1986) (internal quotation omitted). Evidence is persuasive when it has “sufficient weight and probative value to convince the trier of fact.” Daly v. P.D. George Co., 77 S.W.3d 645, 651 (Mo. App. E.D. 2002); see also White v. Dir. of Revenue, 321 S.W.3d 298, 305 (Mo. banc 2010) (noting the burden of persuasion is the “party’s duty to convince the fact-finder to view the facts in a way that favors that party”).
4. Complainant Produced Substantial and Persuasive Evidence of Overvaluation.

 

The evidence shows the data center market presents unique valuation challenges. The nature of a data center requires specialized mechanical systems and physical adaptations. Further complicating matters, the market lacks transparency, making it difficult to obtain specific data, thus enhancing the challenge of separating real estate value from business value. O’Bryan and Fudacz accounted for these valuation challenges in different ways. O’Bryan emphasized the sales comparison approach. Fudacz emphasized the income approach. However, only O’Bryan adjusted for market-based differences among comparable properties to infer a persuasive value estimate. Fudacz did not adjust for property differences, relied on an unpersuasive occupancy estimate, and did not perform a true sales comparison analysis as a check on the validity of his income approach analysis. O’Bryan’s TVM estimate is substantial and persuasive evidence showing the TVM of the subject property was $17,000,000 as of January 1, 2019.

Adjustments

Fudacz made no adjustments in his comparable sales approach. By contrast, O’Bryan premised his sales comparison approach on identifying seven market-driven value indicators and making adjustments along seven parameters within each of those seven value indicators. (Ex. A at 167-186) By making adjustments for different property characteristics and the time of sale, O’Bryan reduced the data center sales information to common units of comparison with subject property, thus enabling a more persuasive inference of value.[10]

The comparable sales approach is premised on adjustments accounting for property differences. Snider, 156 S.W.3d at 348. Unless the comparable properties are identical to the subject property, adjustments are required. Appraisal Institute, The Appraisal of Real Estate (14th ed. 2013) 388. The evidence established that no comparables in either expert’s appraisal report were identical to the subject property. O’Bryan made adjustments, Fudacz did not. Because the STC “cannot base its decision on opinion evidence that fails to consider” the adjustments that “should have been considered” in applying the comparable sales approach, Fudacz’s sales comparison approach is unpersuasive. Snider, 156 S.W.3d at 348.

In addition to utilizing a more persuasive methodology, O’Bryan’s sales comparison approach was well-supported by the sales data. After filtering the initial 191 sales by various property characteristics, O’Bryan identified 61 sales that “better reflect the current market and those properties that are most comparable to the subject.” (Ex. A at 159) O’Bryan filtered those 61 sales by seven characteristics to identify factors relevant to value. (Ex. A at 167-86) O’Bryan testified these characteristics “are critical features that are commonly recognized in the market” as important to market participants.   (Tr. at 79:22-25; 80:1-3) O’Bryan utilized the sales data to develop a value indication for the subject property based on each of these seven characteristics. He then adjusted these value indicators along seven parameters reflecting market-based concerns. (Ex. A at 167-186) This multi-layered analysis enabled a market-based comparison between the characteristics driving the value of both the comparable sales and the subject property. (Ex. A at 167; Tr. 79:18 – 84:16) O’Bryan’s sales comparison approach is substantial and persuasive evidence of value showing the TVM of the subject property on January 1, 2019, was $17,000,000.

Fudacz relied primarily on the income approach but, like his sales comparison approach, his income approach is unpersuasive because he made no adjustments to the comparable leases used to estimate market rent.

To estimate market rent, “data for comparable space in the market is assembled so that equivalent market rents can be estimated and reduced to a unit of comparison.” The Appraisal of Real Estate 466. Estimating “equivalent” rents so they can be “reduced to a unit of comparison” requires the “[adjustment] of comparable rental data.” Id. Thus, “[t]he rents of comparable properties can provide a basis for estimating market rent for a subject property once they have been reduced to the same unit basis applied to the subject property.” Id. Adjustments to comparable leases are made “just as the transaction prices of comparable properties are adjusted in the sales comparison approach.” Id. Thus, inferring market rent from comparable leases presupposes that differences between the properties and the lease terms are accounted for and adjusted.

While Fudacz testified that deriving market rent from different types of leases requires adjustments, (Tr. 196:10-17), his appraisal report and testimony confirm no adjustments were made. (Ex. 2 at 84, 88-89; Tr. 185:1-5; 196:22-25) The difficulty in quantifying adjustments does not obviate their necessity. To the contrary, the specialized nature of the data center market – with its emphasis on the unique physical characteristics necessary to maximize “uptime” – implies there are identifiable, market-based value indicators that can serve as units of comparison to enable persuasive inferences of value from the available market data. O’Bryan performed that analysis, Fudacz did not.   O’Bryan’s analysis is more persuasive.

The closest Fudacz came to making market-based adjustments is his observation the subject property’s rent would fall into the “mid-to-lower end of the range of the gross deals.”   (Ex. 2 at 88)[11] This qualitative bracketing is unpersuasive.

Fudacz gave the most weight to leases 1, (excluding a $74.50 per square foot lease), 3, 6, and 7 because each involved an older carrier hotel with fiber connectivity similar to the subject property. (Ex. 2 at 88) Only leases 1, 6, and 7 are “gross deals.” (Ex. 2 at 87) The unadjusted gross leases ranged from $36.00 to $52.14 per square foot, with a mid-point of $44.07.[12] The $44.07 mid-point is consistent with Fudacz’s assertion the subject would fall in the “mid to lower range of the gross deals” and with his estimate that the market rent of the subject was $43 per square foot.^[13]

Although Fudacz’s market rent estimate is consistent with these bracketed gross lease rates, no adjustments were made to account for different expense bases in the three “gross deals.”[14] There was no adjustment for the fact that, unlike the subject property, the leases in Miami (comparable 6) and Boston (comparable 7) are located in two of the largest data center markets in North America.[15] The net result is that Fudacz’s market rent estimate of $43.00 per square foot is based in large part on a small sample size of three unadjusted leases representing different market dynamics than those of the subject property. Fudacz’s estimated market rent is unpersuasive.

Occupancy and Lease-Up

Fudacz’s value estimate is also premised on the conclusion the stabilized occupancy for the subject property is 90%, thus requiring only a 10% deduction for vacancy and collection losses.   The 90% was based on the general “office market vacancy and historical occupancy at the subject.” (Ex. 2 at 93). Neither metric persuasively supports a 90% stabilized occupancy rate.

Both experts testified data centers are a unique asset subject to unique market forces. The vacancy rates for office space, therefore, do not necessarily translate to the data center market. Likewise, the historical occupancy of the subject provides no precedent for a 90% occupancy. The record shows the occupancy as of January 1, 2017, was approximately 84% and that it declined to approximately 66% as of January 1, 2019. By December 31, 2019, the subject was only 70% occupied. The historical trend shows declining occupancy followed by stagnation. Fudacz’s conclusion that the subject’s stabilized occupancy is 90% is unpersuasive given the vacancy trend established by the evidence. The subject’s historical occupancy does not support the conclusion that the subject’s stabilized occupancy is 90%.

Fudacz’s stabilized occupancy conclusion is also undermined by his June 2020 Bank Appraisal, in which he concluded the subject property’s stabilized occupancy rate was 72% as of December 31, 2019, and that no lease up discount was warranted. (Ex. D at 99) Although the Bank Appraisal focused on the leased fee interest of the subject property, Fudacz testified on cross-examination the market data would be the same and the occupancy analysis “would be similar” in both the leased fee and fee simple appraisals. (Tr. 149:15-25; 150:1-11) Moreover, Fudacz testified that supply and demand in the St. Louis data center market were balanced. (Tr. 151:24-25; 152:1-3) Yet, in the October 2020 appraisal completed for purposes of this appeal, Fudacz concluded stabilized occupancy rate for the subject property was 90% as of January 1, 2019, and that a $2,560,000 lease up discount was warranted. (Ex. 2 at 94, 104) Given his testimony that market supply and demand were balanced, Fudacz’s competing occupancy estimates cannot both be true. Fudacz’s Bank Appraisal and testimony further undermines his conclusion that the subject’s stabilized, market occupancy was 90% as of January 1, 2019.

The unpersuasive occupancy analysis undermines Fudacz’s lease-up analysis. Fudacz applied a $2,560,000 lease-up discount on the assumption it would take 15 months to reach 90% occupancy. (Ex. 2 at 104) Considered in conjunction with Fudacz’s testimony that supply and demand were balanced, the fact the subject property’s vacancy remained below 72% for all of 2019 undermines the conclusion that a 15 month lease-up period is sufficient to achieve 90% occupancy. Fudacz essentially conceded this point by testifying that given the occupancy conclusion in the Bank Appraisal and the subject’s actual performance, “the lease-up did look on the aggressive side.” (Tr. 155:7-8) If, as Fudacz conceded, the 15 month lease-up period is “aggressive” in light of balanced supply and demand, then achieving 90% occupancy would take longer and require a greater discount, thus reducing the final value estimate. Alternatively, if the 15 month lease-up period is unrealistic due to balanced supply and demand, then Fudacz’s 90% occupancy estimate is overstated and the final value estimate must be reduced. Either way, the evidence regarding occupancy indicates Fudacz’s stabilized occupancy estimate overstates the subject property’s income potential and undermines his value estimate based on the income approach.

Sales Comparison

Fudacz’s sales comparison approach utilized a net income multiplier analysis. As the name implies, the NIM analysis is premised on comparing properties by their NOI. Similarly, estimating value with an income direct capitalization analysis requires dividing the NOI by a capitalization rate. Consequently, “[a] value indication developed using net operating income per square foot as a unit of comparison is not independent of a value indication developed using direct capitalization, which negates the checks and balances providing by using more than one approach to value.” The Appraisal of Real Estate 387. It follows that if a NIM analysis is used as a check on value estimate based on direct capitalization, “the results suffer from circular logic.” Id.

By analyzing sales in terms of a NIM rather than in terms of market-based adjustments, Fudacz’s sales comparison approach was essentially the inverse of his income approach. Consequently, Fudacz’s income approach, based as it is on a small sample of unadjusted leases with an unpersuasive stabilized occupancy, is not backed up by an independent sales comparison analysis. Fudacz’s value estimate therefore lacks the “checks and balances” provided by consideration of a pure sales comparison analysis.

CONCLUSION AND ORDER

Complainant produced substantial and persuasive evidence rebutting the BOE presumption and “showing the value that should have been placed on the property.” Tibbs, 599 S.W.3d at 7. The BOE’s decision is set aside.  The TVM of the subject property on January 1, 2019, was $17,000,000.

Application for Review

A party may file an application for review of this decision within 30 days of the mailing date set forth in the certificate of service for this decision.  The application “shall contain specific detailed grounds upon which it is claimed the decision is erroneous.”  Section 138.432.  The application must be in writing, and may be mailed to the State Tax Commission of Missouri, P.O. Box 146, Jefferson City, MO 65102-0146, or emailed to Legal@stc.mo.gov.  A copy of the application must be sent to each person listed below in the certificate of service.

Failure to state specific facts or law upon which the application for review is based will result in summary denial.  Section 138.432.

Disputed Taxes

The Collector of the City of St. Louis, and the collectors of all affected political subdivisions therein, shall continue to hold the disputed taxes pending the possible filing of an application for review, unless the disputed taxes have been disbursed pursuant to a court order under the provisions of section 139.031.

SO ORDERED July 16, 2021.

Eric S. Peterson

Senior Hearing Officer
State Tax Commission

 

 

 

Certificate of Service

I hereby certify that a copy of the foregoing has been electronically mailed and/or sent by U.S. Mail on July 16, 2021, to:  Complainant(s) and/or Counsel for Complainant(s), the County Assessor and/or Counsel for Respondent and County Collector.

 

Elaina Mejia
Legal Coordinator

 

Contact Information for State Tax Commission:
Missouri State Tax Commission
421 East Dunklin Street
P.O. Box 146
Jefferson City, MO 65102-0146
573-751-2414
Fax 573-751-1341

 

 

 

[1] Complainant timely filed a complaint for review of assessment. The State Tax Commission (STC) has authority to hear and decide Complainant’s appeal.  Mo. Const. art. X, sec. 14; Section 138.430.1, RSMo 2000. All statutory citations are to RSMo 2000, as amended.

[2] A “megawatt” is a unit of power equivalent to 1,000,000 watts. Webster’s Ninth Collegiate Dictionary (1989) 739. A five megawatt lease at $58 per kilowatt per month equates to: (5,000,000/1,000) x $58 x 12 months = $3,480,000 annual rent.

[3] Typically, taxes are not included as an expense because the point of an appraisal in an ad valorem tax appeal is to determine the value subject to a tax assessment. For this reason, taxes are typically accounted for by adding or “loading” the effective tax rate to the capitalization rate, which results in a lower value estimate accounting for the negative impact of taxes on value. The special assessment is not a proper expense.

[4] The calculation would be as follows: [$8,380,000 x 0.25] + [$15,110,000 x 0.50] + [$22,050,000 x 0.25] = $15,170,000 (rounded).

[5] Fudacz’s discussion of the subject property’s “secondary St. Louis location” is inconsistent with his earlier conclusion that St. Louis is a tertiary market. (Ex. 2 at vii)

[6] Hopkins, Minnesota is Minneapolis suburb.

[7] Sale 2 included 11.28 acres of adjacent land, but Fudacz excluded the estimated land value of $3,439,498 to calculate an effective price of $16,760,502 ($269.30 per square foot) for the improvements. (Ex. 1 at 82)

[8] Complainant asserted a similar argument in a pre-hearing motion to disqualify Fudacz and strike his testimony and appraisal report. The motion was overruled by an order dated December 16, 2020.

[9] Section 138.060 applies to the City of St. Louis and provides “the assessor shall not advocate nor present evidence advocating a valuation higher than that finally determined by the assessor or the value determined by the board of equalization, whichever is higher, for that assessment period.” Respondent’s Exhibits 1 and 2 conclude the TVM of the subject property exceeds the value set by the BOE.  Respondent does not advocate a value higher than that determined by the BOE.  (Resp. Br. at 41) Exhibits 1 and 2 are admissible as evidence for sustaining the value assigned by the BOE.  12 CSR 30-3.075(1).

 

[10] Respondent asserts O’Bryan failed to conduct various statistical analyses to ensure outliers did not skew the results. (Resp. Br. at 31-36) O’Bryan accounted for this issue by calculating both an average and quartile ranges. O’Bryan testified this approach “demonstrates central tendencies and mitigates some of the risk of a few outliers impacting the indicators.” (Ex. A at 151) Respondent’s argument does not undermine O’Bryan’s analysis.

[11] Fudacz utilized a similar analysis in his sales comparison approach, concluding the subject property “falls within the lower end of the unadjusted range at $205.89 per square foot and is considered reasonable.” (Ex. 2 at 86) (Emphasis added).

[12] The mid-point calculation is as follows: (52.14-36.00) / 2 + 36.00 = 44.07).

[13] Fudacz stated his bracketing was based on the “gross deals.” Complainant asserts Fudacz’s bracketing is unpersuasive because the four leases to which he gave the most weight ranged from $27 to $52.14 per square foot, which yields a mid-point of $39.57, somewhat below Fudacz’s estimated market rent of $43 per square foot. Complainant’s critique overlooks the fact that the $27 per square foot lease (lease 3) is triple net, not gross, and therefore is not among the “gross deals” Fudacz referenced.

[14] The expense basis of leases 1 and 6 were “Gross + E” while the expense basis of lease 7 was “MG + E.” (Ex. 2 at 87)

[15] Along with New York, San Francisco, and Los Angeles, the Miami and Boston markets are home to the largest carrier hotels in North America. (Ex. A at 16)