.. United States General Accounting office ReDort to Congressional Reauesters Timber Offered for:’ SaJeNeedstoEk ‘: Iinmoved Resources, Community, and Economic Development Division IS-23891 I May 2,1990 The Honorable Sidney R. Yates, Chairman The Honorable Ralph Regula, Ranking Minority Membrl Subcommittee on Interior and Related Agencies Committee on Appropriations Ilouse of Representatives In response to your October 26, 1988, request, this is our report on the systems used to appraise timber offered for sale by the 1’5 Department of Agriculture’s Forest Service and the Department of the Interior’s Bureau 01’I,and Management. It evaluates the two methods currently in use and recommends certain :Il,tions to improve the appraisal process. As agreed with your offices, unless you rclcase its contents earlier, we plan no further distribution of this report until 30 days aftt,r the date of this letter. At t,hat time, we will send copies to the appropriate Senate and I louse Committees; interested Members of Congress; the Secretary of Agriculture; the>Chit,f of the Forest Service; and the Director, Office of Managcmcnt and Budget,. Copies will also be made available to other interested parties. This work was performed under the genctral direction of John W. Harman, Director, Food and Agriculture Issues (202) 275-5138. Other major contributors are listed in appendix III of the J. Dexter ieach Assistant Comptroller General Executive Summary The Department of ,4gricaulture’s Forest Service and the Department of Purpose the Interior’s Bureau of Land Management (131,hl) annually sell billions of board feet of timber from the nation’s forests. In fiscal year 1988, receipts from these sales totaled more than $1.4 billion. The Forest Service and I%M appraise timber to establish an advertised selling price. The Chauman and Ranking Minority Member, Subcommit- tee on Interior and Related Agencies, House Committee on Appropria- tions. asked GAO to examine whether the appraisal methods in current use (1) ensure that t lr~ government, receives fair market value for the timber and (2) result in minimum selling prices that, adequately protect the government’s inttarcst and enhance revenues. Various laws and regr(lations require that the Forest Service and N,M sell Background timber for its fair market value. Government-wide guidance also pro- vides that sound business management principles generally be used when selling federal resources. IIowever, neither agency is under a legal or regulatory requircmcnt to sell timber at a price that, will recover costs. The two appraisal m&hods used in government timber sales are called the “transaction evidcnccl” and “residual value” methods. The transac- tion evidence method establishes an appraisal prire based on an average for comparable timber sales, while the residual value method est,ablishes an appraisal price thai would enable a purchaser of average efficiency to harvest and procc>ss the timber at a “rtsasonable profit.” Three Forest Service regions USI’(he residual value met hod while the remaining six regions and H1.M IIW the transaction evidence method. After appraising t III’ I imber, the agencies determine a minimum sales price, advertise the ~tlc. accept bids, and award the sale to the highest, bidder. GAO analyzed fiscal year 1988 data for 3,3 1ti Forest Service tim- ber sales and 221 I<I.Ntimber sales. The Forest Service timber sales were from forests in all ninr, of its geographic: regions; H1.M sales were from its forest lands in wcstc%rn Oregon. GAO’S evaluation SI~IWS that using t.he transaction evidence appraisal Results in Brief method results in atl~~r~rtlsrd price-the lowest prices at which the gov- ernment will sell-uhic+r arc closer to fair market value than does the residual value rntxi hod. This o~urs bec.ausr the residual value method Page 2 GAO/RCED-90-135 Federal Timber Sales Appraisals has many problems in its implementation due to nonstatistical and out- dated data. While the transaction evidence method better estimates fair market value, it is being inconsistently applied by the Forest Service regions using it. Forest Service headquarters has provided only limited guidance and oversight to the regions to better ensure that their appraisals reflect fair market value and increase revenues to the government. Neither appraisal method is designed to establish a minimum sales price which recovers costs and would protect the government’s interest and enhance revenues. At the Forest Service, where a cost-accounting sys- tem has been in place since 1987,40 percent of the total fiscal year 1988 timber sales that GAO reviewed were advertised for less than just the costs of preparing and administering the timber sales. I3LM does not have such a cost-accounting system. However, on the basis of cost data pro- vided by BLM, about 1 percent of the IlLM sales that GAO reviewed were advertised for less than the costs of preparing and administering the sales. Principal Findings Transaction Evidence GAO'S analysis of fiscal year 1988 Forest Service timber sales data showed that the transaction evidence method, when consistently imple- Method Superior Predictor mented, resulted in advertised prices which were closer to fair market of Fair Market Value value than prices estimated by the residual value method. When aggres- sive competition exists, the appraisal method used makes little or no dif- ference because competition results in a selling price that equates to fair market value. However. when aggressive competition is lacking, the accuracy of the appraisal method in estimating fair market value is par- ticularly important to protect the government’s interests. For example, about 5 percent of the sales were sold at advertised prices in oral auc- tions with a single bidder. GAO'S analysis suggests that the transaction evidence method, on average, results in advertised prices that could range from 14 to 37 percent higher than the residual value method. Con- sequently, this suggrsts that the government may be able to enhance its revenues on noncompetitivr sales by using the transaction evidence method of appraisal The residual value method is being implemented with nonstatistical and outdated data. All bllt three Forest Servict regions have switched to the Page 3 GAO/RC,EI)-W-135 Federal Timber Sales Apprdsalfi Executive Summary transaction evidence, method, citing data problems with the residual value method as well as the fact that the transaction evidence method better estimates fair market value and is less costly to maintain. None- theless, the Forest Service’s two main timber-producing regions continue to use the residual \-alue method and cite limited staff resources and historic use as the primary reasons. Forest Service regions have received limited guidance or oversight from headquarters in deMoping or implementing the transaction evidence appraisal method. Accordingly, regions have developed differing approaches. One appreciable difference GAOidentified was in the “rollback,” or reduction the regions made to their appraisal estimates to stimulate competition and to compensate for any inaccuracies that they believed may have o\,erstat.ed the price developed by their appraisal estimates. One region reduced the appraised price developed by its transaction evidence> method by an average of 47 percent in 1988, whereas all other regions reduced the prices within a range of 5 to 25 percent. In fiscal ~‘c’irt. 1988. this region had 18 sales at advertised prices in single-bidder ()ri(I ;hllcTions. With a smaller percentage rollback applied in this rcgic 111,t 1~3government might enhance its revenue on such sales. GAO also found that I trtl Forest Service does not exercise adequate inter- nal control over thr) I imbcr appraisal process. For example, there is no routine hcadquarttlrs monitoring of how well regional appraisal systems are establishing bid Iiri(.(ss that approximate fair market value. IKM switched to tht> transaction evidence method during 1988 because of the residual value mchthod’s data problems and cost, and because the agency believed t 1~~t ransaction evidence method better estimates fair market value. G:\( I’S ;malysis of fiscal year 1988 ill,M competitive timber sales showed that sales appraised using the transaction evidence met,hod resulttsd in sfblling prices that better approximated fair market value than thosca ;~pl~r’aixcd using the residual value method. ..____ Costs Not Considered Neither appraisal mc\t hod ensures a minimum selling price that will ade- When Setting Prices quately protect t 11~government’s interest and enhance revenues because neither method t akos into account the costs of growing and selling the timber. In 1987, the, F’orc~+tService started using, on a test basis, a cost- accounting systc,m u trictl identifies all costs associated with it.s timber sale program. GA(1I iscd this system to compare the costs associated only Page 4 GAO/RCED-SO-135 Federal Timber Sales Appraisals Executive Summary with the sales preparation and administration functions to the adver- tised and sales prices. In fiscal year 1988, the Forest Service advertised about 40 percent of the sales GAO analyzed for prices that were less than these costs and actually sold 24 percent at prices where these costs exceeded the sales prices by over $22 million. These costs do not include the costs of growing the timber, overhead, or foregone interest on the government’s investmcbnt HLM does not have a cost-account,ing system for its timber sale program. However, RLM furnishcbd tiA0 with available data on sale preparation and administration costs which GAO did not verify. GAO’S comparison of these data with BLM'S fiscal year 1988 sales showed that about 1 percent of sales was advertised and that only one timber sale was actually sold for less than its sale preparation and administration costs. While there can be valid reasons for below-cost sales-e.g., diseased timber may adversely affect other forest resources-+x0 believes the reasons for such sales should be documented by the Forest Service. Information regarding the cost, and purpose of a sale is not currently documented or considered before a sale is advertised. However, accord- ing to Forest Service of‘fickls, the agency plans to adopt guidelines and procedures in late 1990 that would provide guidance on timber sales which do not recovc’r ,~sts. GAO recommends that (he Forest Service provide better guidance and Recommendations oversight to improve its timber appraisal process, including developing and using the transaction evidence method in Forest Service regions and discontinuing the use of the residual value appraisal method. GAO also recommends that the 1:orest Service complete actions to ensure that the government’s costs be considered before a sale is advertised and that the reasons for selling below cost be documented by the appropriate official. GM) discussed the information in this report with Forest Service and HLM Agency Comments officials. They generally agreed with the facts presented and with GAO’S conclusions and recommendations. GAO included their comments in the report where appropriate,. As requested, however, GAO did not obtain official comments on this report. Page 5 GAO/RCED90-135 Federal Timber Saks Appraisals Contents Executive Summary 2 Chapter 1 8 Introduction Timbt,r Is Sold Through Bids 9 Legal Requirements for Timber Appraisals 10 .\ppraisal Methods in Current I!se 11 !jbjectives, Scope. anti Methodology 13 Chapter 2 16 Improvements Needed Problems With Residual Value Method 16 Transaction Evidcnw Method Superior to Residual Value 18 in Forest Service ?Gthod in Estimating Fair Market Value Appraisal Process to BLM’s Results Consistent With Forest Service’s 20 Limitcad tiuidancc and Oversight in Establishing Appraisal 21 Better Reflect Fair Systems Market Value Conclusions 22 l~twmmendations 23 -- Chapter 3 24 Advertised Price for Costs Arc Not ITsed to Establish Advertised Prices 25 Costs Clan Now Bc Dc~tcrmined 25 Forest Service Timber Comparison of Costs ()f (‘onducting Sales With Advertised 27 Is Often Not Adequate and Actual Sales IQ-ices for the Forest Service Comparison of Costs of (‘onducting Sales With Advertised 29 to Recover Costs of and Actual Saks Prices for BLM Conducting Sales Sales Not Recovering ( ‘osts May Be Warranted in Some 29 Instances C:onclusions 30 Rwommendations 31 Appendixes Appendix 1: An Analysis of Appraisal Method Bias in 32 Advertised Priws of Forest Service Timber Sales Appendix 11:I+tcntial and Actual IJnrecovered Sales 45 I’reparation and ‘Idministration Costs on Fiscal Year 1988 Timber Saks, Forest Service Appendix Ill: Major (‘ontribut,ors t,o This Report 50 Tables Table 1.1: Exampk ot’ a Kcsidual Value Determination 11 Table 2.1: Percent agesof Overbids and Rollback Factors in 20 R@ons 1Ising Transaction Evidence Appraisals Page 6 GAO/RCED-90-135 Federal Timber Sales Appraisals Table 2.2: Comparison of Transaction Evidence and 21 Residual Value Appraisal Methods in 1988 Timber Sales, With Two or More Bidders at BLM Table 3.1: Potential IJnrecovcred Sales Preparation and 27 Administration Costs on Fiscal Year 1988 Timber Sales Table 3.2: Actual Unrecovered Sales Preparation and 2x Administration Costs on Fiscal Year 1988 Timber Sales Table I. 1: Description of Factors in Overbid Percentages 36 Model Table 1.2: Rollback Factor and Appraisal Method by 36 Region for Fiscal Year 1988 Table 1.3: Estimation Results When the Rollback Factors 39 Are Considered to Be Part of the Appraisal Process Table 1.4: Estimation Results When the Rollback Factors 42 Are Not Considered as Part of the Appraisal Process ~____~ Figure Figure 1.1: Forest Service Regions 9 Abbreviations 111M Bureau of Land Management GAO General Accounting Office Page 7 GAO/RCED-90.136 Federal Timber Sales Appraisals Chapter 1 Introdhion The federal government is a substantial supplier of timber available for harvest. In fiscal year 1988. it, sold more than 12 billion board feet of timber.’ Most federal timber lands are managed by two agencies-the Department of Agric~rilture’s Forest Service and the Department of Intc- rior’s Bureau of Land Management (IQAI). The Forest Service manages 191 million acres of national forest. system land, while I&M manages about 8 million acrts. In fiscal year 1988, the Forest Service sold about 11 billion board feet of timber for a total price of over $1.25 billion, and HI,M sold more than I billion board feet for about $153 million. The Forest Service is organized into nine regions (see fig. 1, I ), with the two regions on the I’acific Coast producing the bulk of the harvested timber in fiscal year 1988. In that year, these two regions, which include the states of Califoruia, Oregon, and Washingnm, accounted for nearly 6.9 billion board fret t)f timber (about 63 pcrcmt of the total timber vol- umc sold) and nearly %1.04 billion (83 pcrc*ent of the v;~luc received), according t,o the Fort+t Scrvicc’s 1988 annual report. ~r~.nz has 12 state offi<xx Most of ixx’s t,imber volume and value are obtained from salch III western Oregon. In fiscal year 1988, 92 percent of the vohime sold ant1 ! Iii Ixrcent of the valr~c received from HLM timber sales came from it 5 \\ ~~,tern Oregon land. Page R GAO/RCED-90-135 Federal Timber Sales Appraisals Figure 1.1: Forest Service Regions ._~ ~~ ~~~~~ In fiscal year 1988, t ho Forest Scrvic*tBand IILM offered timber in more Timber Is Sold than 2~i4,OOOindividual timber salts, ranging in value from less than Through Bids $300 to over $6 m illion Gonerally, only larger sales-those with selling pric,cs of more than SL!,(KN) or timber volumes of more than 2 m illion board t’cct-arr appr;~i~l. When the Forest Scrvlc,t, and HI.M advertise timber for sale, they desig- nate arcas to bc har\(~t rtd and solicit bids. The two agencies establish the advertised price>. \vlCch is the m inimum bid that will be accepted. by Page 9 GAO,:KCED-W-135 Federal Timber Saks Appraisals Chapter 1 Introduction appraising the timber before offering it for sale. Once the timber is appraised, it is advrrt,iscd, bids are accept,ed, and the sale is awarded to the highest bidder. Bidding is by either sealed bids or written bids fol- lowed by oral auctions. With sealed bidding, each potential purchaser submits a written bid. and t,he contract is awarded to the purchaser whose bid exceeded the advertised price by the greatest amount. With oral auction sales, written bids that equal or exceed the advertised value of the timber are required to qualify potential purchasers for further competition by oral bidding. As with sealed bidding, contracts are then awarded to the purchasers whose oral bids exceed the advertised prices by the greatest amount. provided that the purchaser is otherwise quali- fied and responsible Various laws and regulations require the Forest Service and BLM to sell Legal Requirements timber for fair market value. IIowever, neit,her agency is under a legal or for Timber Appraisals regulatory requirement to sell timber at a price that will recover costs. For the Forest Service>. thrl National Forest Management Act of 1978 (16 1T.S.C. lG00 et seq.) authorizes the Secretary of Agriculture to sell tim- ber and other forest products “at not less than appraised value.” Department regulations promulgated in accordance with this authority (36 C.F.R. 223.6) state‘ in part. “The objective of national forest timber appraisals is to est.imatc fair market value ,” The Forest Service Manual defines fair marktlt value as the “price accept,able t,o a willing buyer and seller both with knowledge of the relevant facts and not under pressure or compLllsion to deal.” In addition, GAO’S Office of the General Counsel has drtcrmincd that appraised value referred to in the law means fair markcst \rall~e.’ For lIl.M, two laws prmiarily dicta& the price to be received for timber. The act of 1937 concerning Oregon and California Railroad Grant Lands requires HLM to sell tmtber from former railroad grant lands “at reasona- ble prices on a normal market,.” IU%‘s main timberlands in western Ore- gon are included in thttse former railroad grant lands. BLM’S other timbe sales are guided by thtb Federal Land Policy and Management Act of 1976. This act sedtdtcs that the government is to receive fair market value of the use of public lands and their resourt.es “unless otherwise pro- vided for by statutts. ’ Additionally, Office, (bf Management and Budget Circular A-25 sets forth general policies for t,harging for government services and property. The Page 10 GAO/RCED-SO-135 Federal Timber Sales Appraisals Chapter 1 Introduction circular provides that.. “Where federally owned resources or property are leased or sold, a fair market value should be obtained. Charges are to be determined by application of sound business management prac- tices, and so far as practical and feasible in accordance with comparable commercial practices.” Currently, two methods are primarily used to appraise timber. These Appraisal Methods in two methods, explained b&)w, arc called the “residual value” and the Current Use “transaction evidence” methods. The residual value method is used by three Forest Service rtbgions while the transact.ion evidence method is used by the other six Forest Scrvic,r regions as well as KM’s Oregon State Office. Residual Value Method The premise behind the residual value appraisal method is that standing timber should be advertised at a price that enables a purchaser to (1) harvest the timber. (2 ) process it into finished products, and (3) sell those finished products at prices that recover all of the purchaser’s har- vesting and manufacturing costs and that also allow a margin for profit and risk. In calculating the price for standing timber, the agency starts with an average price for the‘finished pr0duct.s that can reasonably be expected to be produced from the timber in the sale. These products include pri- mary products such as lumber and plywood as well as byproducts such as wood chips used for pulp. From this price, the agency subtracts (1) the estimated costs of harvesting the timber and manufacturing it into finished products and (2) an allowance for profit, and risk. What remains is the residual value, or appraised price. Table 1.1 illustrates a calculation made under t Iris process in terms of a price per thousand board feet. Table 1.1: Example of a Residual Value Determination Sehng value of end produc:s $588 Less harvesting costs $293 manufactunng costs 222 Total costs 515 Price before proflt and r!sk 73 Less allowance for proflt alld risk 50 Appraised prrce using residual value $23 Page 11 GAO,‘RCEI)-YY-I35 Federal Timber S&s Appraisals ._ Chapter 1 Introduction A major consideration in estimating the selling values is the amount of lumber and other finished products that may be produced from a log of a given size and species. At the Forest Service, the amount of the main products-lumber and plywood-is commonly estimated using product recovery or “mill” stlldies. These studies are conducted at individual mills and may take up to a year to complete. They involve following selected logs from act,llal harvesting through the milling process to ascertain which products are produced. This procedure allows an esti- mate to be made of t,hc relative quantity of the various products and their quality. Product prices obtained from market indexes are then applied to the estimated volumes to produce an anticipated average sell- ing price for the principal products. The costs included m the appraisal equation are those that contribute to converting the standing timber to the finished product. They include the costs of logging, transportation, and manufacturing, and costs for such items as erosion control and road maintenance. The profit and risk mar- gin is a standard subtraction. At the Forest Service, this margin nor- mally has been in the: range of 9 to 13 percent of the selling value of the products, according t,o a 1987 report prepared by a national working group organized b> the Forest Service and Kational Forest Products Association. Three Forest Servir*ts regions-Region 5 (Pacific Southwest), Region 6 (Pacific Northwest), and Region 10 (Alaska)-use the residual value method as their primary appraisal system. In February 1988, HLM st.opped using the residual value method and started using the transac- tion evidence method as its primary appraisal system in its Oregon forests. Transaction Evidence The transaction evidtlnce method’s objective is to predict the fair market value of timber. 1TndtBrthis method, the Forest Service uses prior timber Method sales in each appraisal zone within the region to estimate the price that new timber sales can be expected to bring. According to Forest Service officials, transaction evidence method appraisals are based on the pre- mise that, if a compc~titivc market exists, the high bid received is a valid indicator of fair ma~~ket value. The transaction evidence method assumes that whilr, the timber will sell for a price that is close to the predicted price, half of the time timber will sell above that price and half of the time it will sell below that price. Page I2 GAO/RCEW90-135 Federal Timber Sales Appraisals To develop the appraisc.d price using the transaction evidence method, the appraisers must first calculate the average selling price for timber during a defined period of time. The time period used by BLM and the Forest Service varied in length from 1 to 3 years. The high-bid prices for all timber sales during this time in each area are then segregated by spe- cies, weighted by volume, and averaged. These prices then become the base period price upon which ot,her adjustments are made before arriv- ing at the final appraised prices. Adjustments are made to the base period price for a variety of reasons. For example, an adjustment, may be made to reflect market conditions that are different from those that existed in the base period. Adjust- ments are also made to reflect each sale’s individual characteristics. For example, adjustments UC made to recognize differences in factors such as the type of logging system that will be used, the distances that felled timber will be hauled for processing, the quality of the timber, and the amount of road maintenance required. In Region 3, these adjustments are made to individual sale appraisals if they exceed $3 per thousand board feet. For example. if a certain sale characteristic is favorable and exceeds this value whcln compared with the area average (for example, less than average harvesting costs), then the selling price on that sale will be adjusted upwards. If the specific characteristic is unfavorable, i he price will be lowcrc-d. In addition to the adjustments described above, a reduction, or “rollback,” is made to the predicted sale prices. Because the appraised price is based on average’s, which are expected to exceed the price at which the timber will sell for 50 percent of the time, a rollback factor is used to ensure that the advertised price is not set at a level which results in no bids or discourages competition. The value that remains after making adjustments and applying the rollback factor becomes the indicated advertised price Forest Service Regions 8 and 9 have used the transaction evidence appraisal method since the 1970s. Regions 1. 2, 3, 4, and BLM’s Oregon State Office have all switched from the residual value method to the transaction evidence method since 1986. .-__ In a letter dated October 26, 1988, the Chairman and Ranking Minority Objectives, Scope, and Member, Subcommittee on Interior and Related Agencies, House Com- Methodology mittee on Appropriations. asked us to examine two issues with regard to the current appraisal rtwt hods: Page 13 GAO/RCED-90-135 Federal Timber Sales Appraisals Chapter 1 Introduction . Whether the current methods ensure that the government receives fair market value for the timber it sells. . Whether the current methods set minimum selling prices that are ade- quate to protect thr government’s interest and enhance revenues to the Treasury. To respond to the first ob.jective, we obtained data bases of Forest Ser- vice and HLM timber sales. For the Forest Service, we obtained a data base from the Forest Service’s Fort Collins computer center covering all nine regions. We conc*trntrated our review on those sold sales greater than $2,000 in valura or 2 million board feet in size. For fiscal year 1988, the Forest Service data base contained 3,316 sales which met these crite- ria. The data base) that we obtained for BLM timber sales showed that, sales comparable in size to those of the Forest Service were concentrated in western Oregon. The HLM data base contained 221 comparable sales, of which 60 were appraised using the residual value method and 161 were appraised using t ht transaction evidence method. In order to evaluate IIOW effective each appraisal method is in ensuring that the government receives fair market value, we developed an eco- nomic model to explain the relationships between fair market value, government advertiscbd prices, appraisal methods, and other factors. lrsing the Forest St>rvictb’s 1988 sales data, we estimated the parameters of the model with rogrcssion analysis. These estimates then served as our basis for comparmg the ability of the two appraisal methods to result in advertised ~~rict~sthat approach fair market value. For this regression analysis. WC used 2,80 1 of the 3,3 16 sales contained in the Forest Service data base. Of the 2.801 sales, 1,356 used the residual value appraisal method and 1,445 used the t.ransact,ion evidence appraisal method The details of our regression analysis are described in appendix I. We tested BLM'S cxpt,ricnce with how well its appraisal methods approx- imated fair market values. While we did not use regression analysis as a control for other factors, we computed the average advertised and high- bid prices to comI)arcJ the relat,ive difference in overbids for the two appraisal methods To respond to the scc,ond objective, we relied on the data bases referred to above. For our analysis, we excluded all sales that were coded “pend- ing” in the Forest Service data base. This resulted in a data base of 3,030 sales for our analysis. We performed a reliability assessment of selected data elcmc~nts from bot,h the Forest Service and IILM data bases. Page 14 GAO/KCED-90-135 Federal Timber Sales Appraisals -.-____ Chapter 1 Introduction We used random sampling techniques for the Forest Service data base and a loo-percent test of the BLM data base. We found an error rate of less than 1 percent for each data base, which we judged to be acceptable for our purposes. All discovered errors were corrected. In addition, we utilized the Forest Service’s Timber Sale Program and Information Reporting System to obtain the costs associated with the basic sales preparation and administration costs for each forest. These costs were then compared, on a per-thousand-board-foot basis, with the advertised and eventual selling prices of the timber to identify sales that were advertised and/or sold for prices that did not recover these costs. HIM does not have a cost-accounting system similar to that of the Forest Service. We relied on figures BI,M supplied us with t,o make a similar comparison. We did not attempt to verify the cost information supplied by either the Forest Service or BLM. We interviewed headquarters and regional or state officials for both agencies and reviewed pertinent documents. We discussed our findings with appropriate officials and incorporated their comments where appropriate. They generally agreed with the facts presented and with our conclusions and recommendations, Our review was performed between January and November 1989 in accordance with gencrally accepted government auditing standards. As requested, however. WV did not obtain official comment,s on this report. Page 15 GAO/RCED-90-135 Federal Timber Sales Appraisals Chapter 2 Improvements Needed in Forest Service Appraisal Process to Better Reflect Fair Maxket Value - - Our evaluation of the Forest Service data shows that the transaction evidence appraisal method results in advertised prices which were closer to fair market value than prices estimated by the residual value method. This occurs because the residual value method has many problems in its implementation. Although the transaction evidence method does a better job of estimating fair market value, it is being inconsistently applied. Forest Service headquarters has provided only limited guidance and oversight to the regions to better ensure that their appraisals reflect fair market value and enhance revenues to the gov- ernment. Our limited analysis of HLM data also showed that nt,&~‘sexpcri- ence with the appraisal methods was consistent with our finding for the Forest Service. - -___-~ The residual value aptnaisal method as explained in chapter 1 attempts Problems With to set prices that will allow the purchaser “of average efficiency” to Residual Value harvest the timber. nritntrfacture it into finished products, and make a Method profit. Most Forest Service regions and IDI’s Oregon Stat,e Office have moved away from using the residual value method. However, three Forest Ser- vice regions, inc~luding Regions 5 and 6. which accounted for 83 percent of all Forest Servict\ timber sold and 83 percent of all Forest Service timber receipts in fis(.al year 1988, continue to use the residual value method. Since 1%X;, four of the nine Forest Srrvjce regions have switched from the t-cwd~~;~l value method to the transaction evidence method, bringing to 51 r; the number of Forest Service regions primarily using the transactiorr r~vrtlencc method. IILM’S Oregon State Office switched from t,ht> r~cwtir~;tl value to the transaction evidence method in 1988. Forest Servicing ;Ind IL~I officials cited dissatisfaction with data accuracy, high maintcnancc costs. and inconsistent appraisal values as reasons for the change They also stated that the residual value method was outdated and st ai istically invalid, and that it did not result in a price reflective of’ ~‘;III market, value. Here arc some of the specific criti- cisms they voiced Forest Service off’ic~ials in Region 1 stat,ed that the product recovery studies used for t 11~residual value method were of little or no use and the information thcL\, prctvided could be obtained in less costly ways. Forest Service offic~rals In Region 3 stated that industry complained that the residual wrlw rltc’t ho(l set the price of timber too high in some Page I6 GAO/RCEIHWlBR Federal Timber Saks Appraisals Chapter 2 Improvements Need& in Forest Service Appraisal Prucess to Better Reflect Fair Market Value instances and in others, the advertised price was set so low that the tim- ber was eventually sold at a price two to three times greater than what was advertised. . A HLM issue paper cit,ed the failure of the residual value method to pre- dict the value of standing timber and the high maintenance costs as rea- sons for changing. The paper estimated the short-term savings of implementing the transaction evidence method to be 25 percent of total appraisal cost under the residual value method. RLM also reported that timber purchasers and its own appraisers had long been critical of the residual value method’s obsolescence. Our review of the residual value method showed that the averages used for harvesting and manufacturing costs lacked statistical validity. To calculate the average harvesting and manufacturing costs on which the method is based, in our opinion, it is first necessary to identify the log- ging and manufacturing companies located in each appraisal zone. If obtaining cost data from all of these companies is not feasible, a statisti- cally valid random sample can be selected. If the sample is statistically valid, an average can be computed that is representative of all compa- nies. According t,o Forest Service officials in Region 6, they do not use all logging and manufacturing companies located in each appraisal zone to calculate average costs, nor have they identified these companies so that a random sample can be selected. Instead, they rely on companies will- ing to supply cost data, and they use the same companies year after year whenever possible. As a result, the average cost data being used in the residual value appraisal process is not statistically valid and may not be representative of an “operator of average efficiency.” The problem of statistical validity also applies to the average selling val- ues used. For example, not all mills located in an appraisal zone are used for product recovery studies, and those mills which are used are not selected in a way that ensures statistical validity. As with cost data only willing mills are used for product recovery studies. These studies can take up to a year to complet,e and necessnate the positioning of up to 30 people, for a period of 1 week, at all stages of the manufacturing process to mark logs as they go through. This procedure is costly and greatly increases the mills’ normal processing times while each individ- ual log’s products are identified and measured. The selection of mills is based on their willingness to volunteer for this inconvenience and loss of productivity. An official in Region 5 told us that some of the product recovery studies currently used in the residual value appraisal process are over 30 years old. Page I7 GAO/RCED-9Q-135 Federal Timber Sales Appraisals Chapter 2 Improvements Needed in Forest Service Appraisal Process to Better Reflect Fair Market Value Despite the problems and limitations of the method, the residual value method continues as the primary appraisal method in the Forest Ser- vice’s two main timber-producing regions-Region 5 (Pacific Southwest) and Region 6 (Pacific Northwest). Region 10 [Alaska) also uses the residual value method as its primary appraisal method. In fiscal year 1988, Regions 5 and 6 accounted for 63 percent of all Forest Service timber sold and 83 percent of all Forest Service timber receipts. Officials in these two regions cited limited staff resources and historic use as two reasons for continued use of the residual value method. A Region 10 official cited a lark of comparable sales and a fear of industry collusion as the main reasons for keeping the residual value method. C&r analysis of Forest Service timber sales data for fiscal year 1988 Transaction Evidence indicates that the Forest Service’s advertised prices more closely reflect Method Superior to fair market value when they are determined using the transaction evi- Residual Value dence appraisal method rather than the residual value method. This suggests that the government could be losing revenue on those sales Method in Estimating which lacked aggressive competition (for example, oral auctions with a Fair Market Value single bidder) and for which the residual value method was used. Our results also show that inconsistent application of rollback factors can greatly affect the different appraisal methods’ apparent, ability to result in advertised prices which approach fair market value. Analysis of Timber Sales To conduct our evaluation of which appraisal method was better able to estimate fair market value, we developed an economic model to explain Data the relationship between final sale prices, or fair market value, and advertised prices as det,ermined by many factors, including the method of appraisal. IJsing t,he model, we were able to estimate the effect on advertised prices of selecting one appraisal method over the other, while simultaneously nccount.ing for the influence of many other factors on advertised prices. In arriving at our estimates of an appraisal method bias in advertised prices, however, we gave special consideration in the analysis to the role of “rollback” factors or the percentage by which appraisals are adjusted downward in arriving at advertised prices. Rollback factors are important because the actual factors used by the Forest Service are not consistent either across regions or appraisal methods. Further, it is unclear whether or not rollback factors should be considered as compo- nents of the appraisal process. Therefore, we conducted several versions of the analysis, including versions in which rollback factors are and are Page 18 GAO/RCED-90135 Federal Timber Sales Appraisals (‘haptrr 2 Improvements Needed in Forest Srnice Appraisal Procrss to Beltrr Reflect Fair Market Value not considered associated with the appraisal process, and a version which addresses t hc c’(Insistent use of the rollback factors across regions. The results of our analysis suggest that, either when rollback factors are reasonably consistent XI‘OSS regions or if they are considered distinct from the appraisal pro~~css, the transaction evidence method provides a more accurate ref1ec.t ion of’fair market value than does the residual value method. In gcnc~ral. VW found that the transaction evidence method may result in advertised prices that average anywhere from 14 to :I7 percent higher t ban those derived from the residual value method. (SW app. I for a more detailed discussion of our methodology and findings.) Varied Application of the Ah bough agency ot’f’ic~& believe the transaction evidence method pro- vides a more accuratll rcflcction of fair market value, we found that in Transaction Evidence one Forest Scrvict region. it was being applied in such a way as to have Method the opposite effect. The problem stems from the use of a rollback to set final appraisal raos. As chapter 1 explained, because the appraised price is based on av(bI.agcs, the predicted selling price of any timber sale may be set above Lvhat t hc market is willing to pay half of the time. To compensate for this possibility. a downward adjustment is made to the predicted bid price on all sales by using a rollback factor. There is no headquarters guidanc,c on the purpose of the rollback or parameters established as to its size. Our analysis .&wed that Region l’s applica- 1ion of the transact ion cvitlencc method resulted in the most significant dit’fercnce between ap~)raisals and high bids of any of the regions using the transact ion cvidcm~c method Table 2.1 shows the comparison bctwern the rollback f’;n?or usccl with the average percentage overbid for each Forest %,I.\ ICY’region using the transaction evidence method. Page 19 GAO/RCED-YO-135 Federal Timber Sales Appraisals .~~ Chapter 2 Improvements Needed in Forrst Srrvicr Appraisal Process to Better Rrflert Fair Market Value Table 2.1: Percentage of Overbids and Rollback Factors in Regions Using Average percentage of Transaction Evidence Appraisals Rollback factor overbid on competitive Region range sales= I 47" 205 2 5 10 70 3 5-10 40 4 5~10 69 8 .~ 10~25 39 9 15” 48 In fiscal year 1988. licq$on 1 had 18 sales at the advertised price that were oral auctions wirh only 1 bidder. Higher Advertised Prices When timber salts arf‘ competitive. the advertised price is only a start- ing point for the coml)catitivc bidding. Competition tends to raise the Could Enhance advertised price to fair market value. Accordingly, the accuracy of the Government Revenues on appraisal is of litt Ic unportance when competition exists. Some Sales For those sales wl11c91 lacked aggressive competition in the bidding, advertised prices ;issllm<’ greater importance regardless of which method is used. For c>xample, a single bidder at an oral auction need bid no more than thtb ad\ crtised price to be high bidder, so that the govern- ment receives no prcMum above t,he advertised price. In fiscal year 1988. about ,5 perching of the sales were sold at advertised prices in oral auctions with a singIt bidder. Therefore, for those sales which lack aggressive compel it ic~n. I he government may enhance revenues by using that appraisal mc*tttod which results in advertised prices that are closer to fair market viil\llt As previously stated, the transaction evidence method rt>sults in ;Icl\-clrtiscd prices that average from 14 to 37 percent hightlr than t hc ~~c~~~ti~~al value met hod. __~ We also tested IUI’S ~~xperiencc with how well its appraisal methods BLM’s Results approximated fair markrt values. Although we did not use regression Consistent W ith Forest analysis to control t‘c11’ c)ther factors, a simple comparison of averages Service’s suggests results (.olrxistc>nt with those of our analysis of the Forest Ser- \+cc,. The transac,i icII I t,vidence method more closely reflected fair mar- ket \,aluc on c~oml~c*t Iti\,caly bid sales in fiscal year 1988 than did the residual value mt’t IN id. The results are presented in table 2.2. The table Page 20 GAO/RCED-W-135 Federal Timber Sales Appraisals Chapter 2 Improvements Needed in Forest Service Appraisal Process to BetIer Reflect Fair Market Value shows that the average percentage of overbids using t,he residual value method was twice that of the transaction evidence method. Table 2.2: Comparison of Transaction Evidence and Residual Value Appraisal Difference/ Methods in 1988 Timber Sales, With Two Average advertised or More Bidders at ELM Number advertised Average Average price Method of sales price high bid difference (percent) Transaction evidence 147 $516,627 $731,315 $214,688 42 Residual value 60 $316,538 $584,809 $268,271 85 Although the transact ion evidence method is better than the residual Lim ited Guidance and value method in reflecting fair market value, we found that. the differ- Oversight in ences in the application of the rollback factor in Region 1 were lim iting Establishing Appraisal this method’s effect ivcncss wit,hin the Forest Service. The differences between regions rcflcct. in our view, lim ited guidance and oversight Systems from Forest Scrvicc, h(~adquarters. The Forest Service m itnual st,ates that, the development of appraisal data is a regional responsibility, Forest Service headquarters provides lim ited guidance to regions in establishing appraisal systems primarily through field trips to provitl(> t ethnical assistance in system design. Forest Ser- vice officials told IIS that while the regions have the responsibility to design their own appraisal systems, considerable influence is exerted throllgh the recomtn(,ndations of the assistance teams during the field visits. IIowever, t hc f’inal regional appraisal system does not need for- mal approval by Forest S(brvict> headquarters. The Forest Servic.tl alxo has little oversight of how the regions c.onduct the appraisal procc’ss WC found no ongoing Forest Service headquarters review of how wc4l c~rch region’s appraisal system is predicting the evcnt,ual selling prlc~~ of timber sales. Hids arc not being monitored nationally so that ;~l)l~raisal systems can be adjusted accordingly. An Agriculture Insptxc.tor General report on timber appraisals (Audit Report No. 08627.:%-SF). dated .January 1986, also criticized the Forest Scarvice for poor in(tLrnal controls over the appraisal process. In that report, the Inspec,tor Gcnc~ral c,rit,ic-ized the Forest Service fol Page 21 GAO/RCEI)-SO-135 Federal Timber Sales Appraisals Chapter 2 Improvements Needed in Forest Service Appraisal Process to Better Reflect Fair Markel Value identify and correct Kt~g~onal appraisal methods which do not result in advertised values which are reasonat~lc estimates of fair market value for Kational Forest timber Along with recommendations for better internal controls, the Inspector General’s report also recommended that Region 1 adopt a rollback factor that results in advert iscd rates being within 76-86 percent of the high bid and that all regions adopt standards that result in advertised rates t,hat are reasonable c,stimates of fair market value. In April 1990, Forest Service officials informed us that they were finalizing an action plan to address these recommendat ions. While we did not make a detailed review of the internal controls over appraisals at ISLM’SOregon State Office, our review of selected internal controls indicates that they were adequate. The appraisal system hand- book was reviewed hy hc>adquarters, and the appraisal system is main- tained at the state office. The state office supplies the district offices with the selling value and equations they are to use. Districts may mod- ify these equations to fit specific sale conditions. Bids are monitored by the state office, which prepares a “shadow appraisal” on every adver- tised sale. This appraisal is compared with the one the district prepares, and major differtm.css ;W explored. Our analysis shows that the transaction evidence appraisal method, Conclusions with consistent apt)lic ation of rollback factors. results in advertised prices which are closcbr to fair market value than does the residual value method. When coml)rstition exists, the appraisal method used makes lit- tle difference becatlc:(, c.ompetition encourages the receipt of fair market value. IIowever, when competition does not exist. the advertised price is very often the sellmg pri(e. Accordingly, where comparable sales data exist, it is impcratrx CLthat the Forest Service switch from the residual value to the transact ion c,vidence appraisal method to increase the ret,urn to the govc~t11r1~w1 by better approximating fair market value. We also found that : he Forest, Service does not exercise adequate inter- nal control OVL’I-the 1Imber appraisal process. There is no routine head- quarters monit,oring cIf how well regional appraisal systems are establishing bid pric.tts that approximate fair market value. As a result, the regions are inc~onsistently developing and applying appraisal sys- tems. In particular. Region l’s appraisal system establishes the advrr- tiscd price by using ;I rollback factor which is nearly twice that of any Page 22 GAO/RCED-90.135 Federal Timber Sates Appraisals Chapter 2 Improvements Needed in Forest Srrvice Appraisal Process to Brttrr Reflect Fair Market Value other region. We are concerned that in sales without competition, this could significantly reduce revenues to the government. .~- As a result of our analysis, we recommend that the Secretary of Agricul- Recommendations ture direct the Chief of the Forest Service to take the following actions to improve its timber appraisal process: . Improve the guidance i o regions on developing and maintaining timber appraisal systems. . Require routine headqltarters monitoring of how well regional appraisal systems are approximating fair market value. In addition, we recommend that in order to establish bid prices that bet- ter approximate fair nlarket~ value, the Chief of the Forest Service direct: . Regions 6 and 6 to hwttch to the transaction evidence appraisal method. l Region 1 to reductl its rollback factor to be more consistent with the other regions. Page 23 GAO/RCED-90-135 Federal Timber Sales Appraisals Chapter 3 Advertised Price for Forest Service Timber Is Often Not Adequate to Recover Costs of Conducting Sales The advertised timber prices established by the Forest Service do not ensure recovery of the costs of even preparing and administering the sales. As a result, man\. sales are advertised and eventually sold for prices which do not rc’c’ovcr these costs. We found that 40 percent of the sales in our data base for fiscal year 1988 were offered for less than the Forest Service’s estimated costs of preparing and administering them. Twenty-four percent of the sales were sold for about $22 million less than these cstimattttl (,osts. Moreover, these figures do not inchide the costs of growing the timber, overhead, or the foregone interest on the government’s investment in timber activiticx Before 1987, the Forclst Service did not have a cost-accounting system which could dev?l(Jp timber program costs. However, in 1987, the Forest. Service developed a u jst-accounting syst,em that determines costs associ- ated with growing thus timber, preparing and administering timber sales, and general overh<h;ttl. In this report. we chose to emphasize the fact that timber sales arc’ t’requcntly advertised and/or sold at prices that do not even recover thtGr preparation and administrative costs. If we had included all costs in our analysis, the percentage of sales which did not rt’covor costs would h;tvc~ been much higher. While the Forest Srr~ ICCis not required by law to recover or consider costs, we believe that For& Service policy should consider all vests when setting advcrtistd prices. We believe the cost-accounting system should be used to f’st ablish minimum advertised prices which will gcner- ally promote the re(‘o\ t’ry of the timber program costs. When the Forest Service chooses to ~6.11thr t,imber at less than these costs, it, should docu- ment the reasons 11h) the, sale is being advertised at a lower price. IKM does not have XII xcounting system that establishes timber sales costs. Cost data sulq)Iwd by KLM on fiscal year 1988 sales in Oregon ind- cate that three sales. I lr about 1 percent, kvere offered for salt at less than the costs of pr~~l)xing and administering the sales and that one was sold at, about $3.900 Itbss than these costs. We did not verify thcl accu- racy or completentlss of the cost data ISI.Mreported. Because HI.TVI lacks a cost-accounting syst~rn, we believt> that it is impossible to draw any c’on- elusions from our (comparison of NI.YI’Scosts of preparing and adminis- tering the sale with ;rtlvc>rtised or selling prices. Pa@? 24 GAO/RCEII-90-135 Frdwal Timber Sales Appraisals Chapter 3 Advertised Price for Forest Swvicr Timber Is Often Not Adequate to Rrrovrr Costs of Conducting Saks Aside from activities involved in growing the timber to harvest, the For- Costs Are Not Used to est Service timber salt, process can take over 10 years from the time Establish Advertised initial sale planning begins through the harvest. Typically, about 8 years Prices before the sale award. approximate sale boundaries are identified, the general conditions of the area are surveyed, and a brief logging and transportation plan is l)repared. Three years later, a sales description, including location and approximate timber volume, is published in each national forest’s listing of upcoming timber sales. During the next 5 years, a variety of other sale preparation activities are undertaken such as assessing the environmental impacts of harvesting the timber, esti- mating the timber volumc~ more accurately. and appraising the timber. The contract to harvest the timber is awarded under competitive bidding procedures to the highest bidder. The contract terms often call for the timber to be cut in 3 to 5 years, but cutting time can range from 1 or 2 months for small sales to 10 years for large sales. The Forest Service monit,ors the purchaser to ensure that access roads to the timber are built correctly. only designated trees are cut, the trees are cut according to contract specifications, damage to the soil or streams is minimized, and various other cant ract requirements are complied with. After the harvest, the area is rclforested by natural means or new trees planted. In addition to the costs of growing the timber, the Forest Service incurs costs to prepare timber for sale, supervise harvesting, and do subse- quent reforestation. IIowt~vrr, applicable legislation governing sales of Forest Service timber does not rcquirc the Forest Service to recover these costs on indi\,idual sales. As a result, these cost,s are not consid- ered by the Forest Servic,tb when appraising and setting the minimum advertised price for which timber is offert,d for sale. This, in turn, often results in the Forclst Service’s offering and eventually selling timber for less than even thcb c,osts of preparing and administering the sales. Before 1987. the Forest Sclrvicc’s accounting system could not provide Costs Can Now Be detailed cost informal ion associated with timber sales. In its fiscal year Determined 1985 appropriations, however. the Forest Service was requested to develop a cost-accounting system which would, among other things, pro- vide this information Tht Forest Service was instructed to work with us in developing the xysr tm This system was tested in fiscal years 1987 and 1988, and impl~mrntcd in fiscal year 1989. The Timber Sale l’rogram Informat,ion and Reporting System identifies timber program cysts on a forest-by-forest basis. It attempts to match Page 2.5 GAOjRCED-90-135 Frderal Timber Saks Appraisals Advertised Price for Forrsl Srr%ic~r Timber Is Often Not Adequate to Rrrovrr Costs of Conducting Sales costs with the revenues they produce by accumulating costs into multi- year pools. A certain dollar amount from the pools is then expensed annually on the basis of a formula which includes the amount of timber actually harvested. If no timber is harvested in a given year, then no costs are expensed from the pools. Two cost pools accumulate multiyear costs The first cost pool is called the “sale activity pool ” This pool includes costs which can be specifi- cally identified witlr individual sales on the forest. It contains the costs necessary to plan and prepare timber sales. For example, costs to iden- tify sale boundaries, prepare necessary environmental documents, and advertise sales are t>lirred in this pool. Each year. a certain amount of the accumulated costs is expensed. The amount to be expensed is deter- mined by dividing the total costs in the pool by the volume of sales under contract for that year and then multiplying the result by the vol- umc of timber harvesied during the year. The second cost pool is called the “growth activity pool” and includes those costs related to growing timber. F’or example, precommercial thin- ning, pest control. :III~ fertilization costs would be placed in this pool. This cost pool has ;I rnu~h longer life than that of the sale activity pool because the costs as~ckrted with it are those necessary to bring the tim- ber to the stage where> it is once again available for sale. These costs are generally described to be investments in future timber stands. Again, a certain amount of 1he costs in the pool is c~xpenscd annually on the basis of formulas dcvclopetl as part of the cost-accounting system. The annual amount expensed from this pool was not imludcd in our analysis. Other costs assock~tr30 with the timber sale program are not placed in pools because they ar11directly related to the revenue generated in the year in which they WP imurred. These costs are expensed annually as they OWL~I‘. These ;amrual charges include forest-level overhead and the costs associated with administering sales actually being harvested. Finally, the government incurs costs of foregone interest on its invest- ments in timber act.icrlies. These costs are not explicitly reflect,ed in the cost-accounting system, tbut would he relevant lo an economic analysis of timber program c osts and profitablility. The annual expensed amount from the sale activity pool and the costs of the annual harvest in:: expenses were included in our analysis of sales advertised and/or sold below their preparation and administrative c,osts. The exclusion I cIf’growth act,ivity pool costs and overhead co& Page 26 GAO/RCED-YO-13.5 Frdcral Timber Sales Appraisals Chapter 3 Advertised Price for Forrst Service Timber Is Often Not Adequate to Rrcwrr (‘osts of Conducting Sales ___-~ ~~ ___-- from this analysis does not imply that we believe they should not be recovered or considered when setting a timber sale’s advertised price. In this report. we chost~ to emphasize that timber sales are frequently advertised and/or sold at prices that do not even recover their prepara- tion and administrativc~ costs. If we had also included these other costs in our analysis, the pc~c’cntage of sales which did not recover cost,s would have been match higher. Our comparison of t hc Forest Service’s costs of preparing and adminis- Comparison of Costs tering timber sales wrtlt advertised sales prices showed that about 40 of Conducting Sales percent of the fiscal year 1988 sales were advertised at prices that were less than t,he Forest Scrvkc’s preparation ;md administration costs. With Advertised and Twenty-four JKT~YT~I 01’these sales wet-c sold at prices which did not Actual Sales Prices for recover about $22 ruilllorr of lhesr costs. the Forest Service Forest Service Advertised The Forest Service xl\ crtised about 40 percent of the 3,030 fiscal year 1988 sales we revic\vc~cl for prices that were less than their preparation Nearly 40 Percent of Its and administration co\ts Table 3.1 summarizes for each region the total Sales at Less Than Costs of nnmber of sales. t hcstlttmber advertised at less than their preparation Conducting Sales in Fiscal and administration costs. and the potential amoums of unrecovered Year 1988 preparation and atim~~list ration c,osts. (The results for the individual forests within cac.11rcS::iotr iIr(’ coniained in :iJIp. II. ) Table 3.1: Potential Unrecovered Sales Preparation and Administration Costs on Advertised at less than costs of Fiscal Year 1988 Timber Sales conducting sales Total number of Number of Potential Region sales sales amounta 304 254 519,015,266 2 96 57 2.944.998 3 72 28 1 135,952 4 118 59 1820740 5 426 208 19.535.714 6 759 110 8,761,883 8 721 216 3.207.848 9 528 282 4 660.160 IC tj 5 765,266 Total 3,030 1,219 $61,847,827 Page 27 GAO/KCW)-90.135 Federal Timber Saks Appraisals Chapter 3 Advertised Price for Forest Srrvice Timber Is Often Not Adequate tu Recover (:osts of Conducting Saks As table 3.1 shows, the%largest amount of potential unrecovered prepa- ration and administration costs occurred in Regions 1 (Northern) and 5 (Pacific Southwest). but all regions had sales which were advertised at less than these costs. If competition had not resulted in higher selling prices, the potential unrecovered preparation and administration costs would have totaled about $62 million. Not only did Regions 1 and 5 have the highest potential unrecovered preparation and administration costs, but they also had higher average costs per thousand board feet than the other regions. Without doing a detailed analysis, it is difficult to know why these two regions have higher timber sale preparation and administration costs and more poten- tial unrecovered costs than the others. Forest Service Actually The Forest Service actually sold 24 percent of the sales included in our analysis for $22 million less than their preparation and administration Sold 24 Percent of Its Sales costs in fiscal year 198X. The regions showing the largest amounts of for Less Than the Costs of unrecovered preparation and administration costs were also Regions 1 Conducting Sales and 5. These results are summarized in table 3.2. (The results for indi- vidual forests within each region are contained in app. 11.) Table 3.2: Actual Unrecovered Sales Preparation and Administration Costs on Sold at less than costs of conducting Fiscal Year 1998 Timber Sales sales Total number of Number of Region sales - sales Actual amount 1 304 120 $3,912,639 2 96 49 2.734,75% 3 72 ~21 923,448 4 -~ 116 -~ 44 1391 328 5 426 59 4,510 919 6 -~ 759 51 3,263,544 8 721 152 2,249,079 9 528 - 217 3,081,507 10 6 1 45,332 Total 3,030 714 $22.112,554 Paye 28 GAO/RCED-SO-135 Federal Timber Sales Appraisals Chapter 3 Advertised Price for Forest Sewicr Timber Is Often Not Adequate to Rerovrr Costs of Conducting Sales Our comparison showed that BLM advertised three of its fiscal year 1988 Comparison of Costs western Oregon timber sales for prices less than their preparation and of Conducting Sales administration costs, but, actually sold only one of their sales for less With Advertised and than these costs. The unrecovered costs on this sale amounted to $3,937. As stated earlier, ~I,Y does not have a cost-accounting system. In order Actual Sales Prices for to perform an analysis similar to that conducted at the Forest Service, it BLM was necessary to rely on agency-supplied cost figures. These cost figures represented sale preparation and administration costs on a district office basis and were not independently verified by us. Because HLM lacks a cost-accounting system, we believe that it is impossible to draw any conclusions from our comparison of HLM’S basic costs with adver- tised or selling prices. The multiple use object IVCSin national forest land use plans include gen- Sales Not Recovering erating a fair return to the government, as well as contributing to local Costs May Be and national economicIs and to nont,imber resources. In general, we Warranted in Some believe that advertising sales t.hat are not, expected to cover the costs of even preparing and st4ling the timber are not consistent with sound bus- Instances iness management practices. Office of Management and Budget govcrn- ment-wide guidance provides that sound business management principles generally bc used when selling federal resources. As private sellers would not gemlrally undertake such sales, they may also be inconsistent, with thtl notion of yielding fair market value. While there can be valid reasons for below-cost sales--c.g., diseased timber may affect other resourct’s--we believe that the reasons for such sales should be documented by the Forest Service. Forest Service Examines The Forest Service currently has three initiatives underway dealing with the issue of salts which do not recover costs. The first initiative Issues of Sales Which Do examines the feasibility. of establishing national guidelines and proce- Not Recover All Costs dures, the second initiative is the Below-Cost Commercial Timber Sale Pilot Test contained in I hts fiscal year 1991 budget proposal, and the third initiative is a stlttly of minimum bid rates. All three initiatives use data from the Timber Sale E’rogram Information and Reporting System. In August 1989, thr For& Service formed a task force to develop and test draft national guid+xlincs and procedures regarding timber sales which do not, recover all costs. The objectivcx of the guidelines is to pro- mote cost efficiency of Individual national forest timber sales programs. Page 29 GAO/RCED-SO-135 Federal Timber Sales Appraisals Chapter 3 Advertised Price for FOIYM Sm-vicrr Timber Is Often Not Adequate to Recowr (‘o&s of t‘onductiny Saks A draft of these guidelines states that each national forest should ana- lyze and utilize opportunities to reduce inefficiencies in its timber pro- gram by reducing costs and enhancing revenues. If a national forest’s timber program is not recovering all costs, the draft guidelines state that one action to be taken would be to increase the minimum acceptable bid price so that it COWI’s all current-year costs. The Forest Service is cur- rently testing the guidelines at four national forests and expects to report on the results later in fiscal year 1990. The President’s budgtt for fiscal year 199 1 included a proposed test to evaluate the implications of phasing out certain below-cost commercial timber sales and to determine whether the loss in local economic activity and revenues can bt, offstlt through the expansion of recreational pro- grams. The test will cXvahlatc the effects on 12 national forests. In March 1990, the Forest Service initiated a study of the minimum bid rates. The current. minimum rates were revised in 1979. The objectives of the study are to d~elop and evaluate alternative approaches for computing minimum hid rates for the timber being offered for sale. The study team will bc examining various cost -recovery alternatives using the cost information from the Timber Sale Program Information Keport- ing System. Forest Service officials believe that a 3- to 5-year phase-in period is needed 10 allow each national forest supervisor to thoroughly analyze the cost data that. are now available so that cost reductions and/ or program effic%~nc~~c~scan be identified and instituted. The Forest Scrvic,tB advertised 40 percent, of the fiscal year 1988 timber Conclusions sales we reviewed at prices which would not have recovered the costs of preparing and administering that sale. If competition had not resulted in higher selling pncc>s. the Forest Service would have experienced about a 662 million short t’all of these costs to revenues. With competition, this potential shortfall TV;ISreduced to around $22 million. As previously stated. we did not inc.lude all costs of growing and selling the timber in our analysis, but this does not imply that all costs should not be consid- cred when setting the’advertised price. If we had included all costs, the perccbntagc of salrs wl1ic.h was advertised below these costs as well as the percentage w1ric.h was sold that did not recover costs would have been much higher R.e believe that the main reason why even the prepa- ration and adminisl ration cost,s were not recovered on more sales was that the Forest SC~VICY~ does not, and is not required to, consider costs when setting the ;III\ t,rtised price. IIowcver, the Forest Service has Page 30 GAO/RCED-SO-135 Federal Timber Saks Appraisals Chapter 3 Advertised Price for Forest Service Timber Is Often Not Adequate to Rrcowr Costs of (imducting Saks started actions to consider all costs related to growing and selling the timber in establishing minimum acceptable bids. We believe that thr Forest Service should consider all of its costs related to timber sales when setting advertised prices for timber sales. In those instances where these costs exceed what the appraisal process predicts the timber is worth, a formal decision needs to be made to (1) raise the advertised price to covt’r these costs, (2) not, proceed wit,h the sale, or (3) sell the timber but document the reasons for doing so. -__ We recommend that thus %cretary of Agriculture direct the Chief of the Recommendations Forest Service to consider all t.imber sales costs in establishing adver- t,ised prices for timber salts. We also recommend t liar the Secretary of Agriculture direct the Chief of the Forest Service to adopt the formal decision-making process described above as an integral part of its forthcoming guidelines and procedures regarding timber salts which do not recover costs. Page 31 GAO/RCED-90-135 Federal Timber Sales Appraisals An Analysis of Appraisal Method Bias in Advertised prices of Forest Service Timber Sales This appendix presents our analysis of whether Forest Service advcr- tised prices for timber sales have been significantly higher or lower when the residual value appraisal method is used rather than the trans- action evidence appraisal met.hod. An appraisal method bias in Forest Service advertised prices can result, as discussed in chapter 2, from dif- ferences in the theory underlying the two appraisal methods and/or the manner in which the\. are applied across the country. Since the advcr- tised price is the minimum price that the Forest Service will accept on a sale, higher advertised prices generally imply that the Forest Service will obtain more rr’vcm~es for those sales where there is little or no com- petition among bidders. Ideally, we would cst mate any appraisal method bias in advertised prices by selecting ;I 1epresentative timber sale and then comparing the advertised price for that sale that would result from each of the two appraisal methods. Although available data did not permit this straight- forward method of a controlled experiment, we used regression analysis t.o approximate such a comparison. For our analysis, we developed a model to explain hoa, variations in timber sale overbid percentage, or sale prices as a perc~ontage of associated advertised prices. are deter- mined by a variety 01’administrative, market, and appraisal-related far- tors. Ey including t,hc appraisal method among t,he factors that ma> explain the overbid 1)erc*c.ntage, we could test statistically for the pres- ence of a bias in ad\ ( rt lscd prices that is associated with the selection and/or application 01 an appraisal method. We estimated the paramet,ers of the model with nonlinear regression analysis, facilitating a test for an appraisal method bitts while simultaneously controlling for the effects of many other factors cm I he overbid percentages. Our results suggest that advertised prices ar(’ significantly higher (anywhere from 14 to 37 per- cent) when determitllld by using the transact,ion evidence method rathttr than the residual value met hod. This appendix prcn’tcies / 1) the theoretical development of the model of Forest Service tinrbcxt. sales overbids, (2) a discussion of the estimation methodology and da1 a, and (3) a discussion of the estimation results and sensitivity analysis We developed a model of the overbid percentages from expressions for A Model of Timber each of the two c~ltnponc~~ts of an overbid; the sales price and the advcr- Sale Overbid tised price. Equation 1 describes the sales price as equal to timber ma).- Percentage kct value adjustcltl (tlthc3r up or down because of both administrative factors and oth(lr nl;lrkct factors. Page 32 GAO/RCED-90-135 Federal Timber Sales Appraisals Appendix I An Analysis of Appraisal Method Bias in Advertised Prices of Forest Service Tiimhrr Snlrs Equation 1: Sales priccb = [timber market value] * [l + f(administrative factors + market factors)] Timber market value represents the fair market value of the timber for sale under competitive market conditions. This value should, therefore, reflect supply and demand conditions for the timber of the sale and, in so doing, account for a number of considerations such as extraction costs and profit rates at levels considered to be industry norms under competitive conditions. Many administrative and market factors, how- ever, may deviate from industry norms on some sales and result, there- fore, in the deviation of the sales price from timber market value. Administrative fact,ors include the timber sale contract specifications, e.g., the type of cont,ract and whether the sales price is considered to be a flat rate or will be escalated according to future market prices. Market factors include the nature of the competition surrounding the sale, e.g., the number of biddtars or degree of competition and whether the bid method is an oral auc‘t,ion or sealed bid. As administrative or market factors deviate from industry norms on a given sale, so will each bid- der’s perspective on profit and risk, costs, or his/her own market (monopoly) power, in regards t.o that sale. Consequently, the resulting bids could result in a final sales price either more or less than the timber market value. Both the transaction cvidcnce and residual value methods can be viewed as attempts to approximate timber market value prior to a couple of adjustments which rc>sult in the final advertised price. Equation 2, then, describes the advertised price as being equal to the timber market value, adjusted by appraisal Pnctors. Equation 2: Advertised price = Itimber market value] * [I - rollback fac- tor] / [ 1 + g(salvage and appraisal method)] One adjustment, or apljraisal factor, applied to about 30 percent of all sales, and which results in lowering advertised price, is the classification of the t,imber sale as salvage, A sale can be classified as salvage if a large percentage (around 40 percent) of the timber is damaged. In such cases, the Forest Ser\ic,c will adjust advertised prices downward because it is anxious to dispose of the timber and minimize bug infesta- tion and other probkms associated with damaged or dead timber. Page 33 GAO/RCED-SO-135 Fedwal Timber Sales Appraisals Appendix I An Analysis of Appraisal Mrthcd Bias in Advertised Priers of Fcwvst Service Timber Salrs Another ad,justmc~nl ;lpplied to all sales is to rollback the timber market value approximated hy i he appraisal by some percentage to attract bid- drrs to the sale’ ‘1’h(>perc,entage of the rollback can differ across regions as does the application aspect of the different appraisal methods. It is difficult, therefore. to disassociate rollback factors from the application aspecat of the two mot I~ods of appraisal. Nonetheless, as discussed below, we deal with this issll(, of appraisal method and rollback factor associz tion by estimating the parameters of the model both with and without the rollback factors (~~nsidcred to be associated with the appraisal process. Finally. we inslutk t Ir11appraisal method among the factors that can affect advertised pt’lc’l’. This will permit a statistical test for the pres- ence of an appraisill n1(1thod bias in advertised prices. We obtain an expression in equal ion 3 for the overbid percentages (i.e.. the sales price divided by the advc*l.t tsed price) by dividing equation 1 by equation 2. Equation :3: Ovct’bid 1Ic~rc~t~ntage = [ 1 + f(administrative factors t mark(bt l’;~~iot~s)]* [ 1 t g(salvage and appraisal factors )I 11 -- rollback factor] The advantage of t~qll&on 3 is that timber market value, which appcarcd in hot h c’cl~i:lticnts 1 and 2, has been divided out, thereby elimi- nating the nr~tl to IIU11loI1hc>supply and demand for timber. Konethe- less, tbquation 3 still ;illo\ts us to test for tlie presence of an appraisal met hod bias in 120t.~~\1 St3rvic.e advertised 1jriccs. LVc obtaincLd data f 1.1ml t h(l Forest Service on all timber sales of more Estimation than $2.000 for fisc’;~I yc’ar 1988. The data set contained information on Methodology and Data a nurnbcr of :rdmin~s~ t.itt ivc, market, and appraisal factors. Those fat- tors we sclcctcd 10 [,I’#IVI~IV detail to equation 3 are described in table 1.1. pag:r 34 GAO/RCEI)-90-135 Federal Timhrr Sales Appraisals Appendix I An Analysis of Appraisal Method Bias in Advertised Prices of Fomst Sewire Timber Saks Table 1.1: Description of Factors in Overbid Percentages Model Factor Description Administrative factors NOSBA = 0 If the sale I; restricted to a Small E&&s AdmInIstratIon status firm and 1 If not SALEMETH = 1 If the method of sale (flnal sales prcce) IS a flat-rate, and 0 If the method IS quarterly escalation of sale price ROADS = 1 If road costs are Included in the advertrsed pnce and 0 if iltherwlse F&M6 = 1 If the contract form IS a 6 or 6A or 6T. and 0 If otherwise FORM9 = 1 If the contract form IS a 9 or ST,-and 0 If otherwrse FORtiT - = 1 If the contract form IS a 3T, 6T. 6A. or 9T, ImplyIng tree measured rather than scaled sale, and 0 If otherwrse The one remalnrng form not accounted for by any of the FORM factors is (ontract form 3. whrch serves as the base case Market factors BIDDERS = ihe number of bidders I” auction or closed bid sale SEALED = 1 If closed brd sale and 0 If open auction tiZE - = Estimated volume of sale in mllllon board feet ACRES = rhe acreage of the sale area SIZE/ACRES = A measure of density of the timber stand for sale HIGHBID = ’ If the SEA classrflcatlon of Ihe high bidder IS that of a large firm and 0 If otherwIse Appraisal factors NgSALVG = 1 If the sale IS not classrfred as a salvaae sale and 0 If ri IS a salvage sale TEA = 1 If transactjon-evidence method was used to arnve at advertcsed price and 0 11residual value method was used R@F = Rollback factor preset either to the mean of a range of rollback percentage used !n each region (different numbers for each rcgron), or the weighted average of means of ranges for all regions (one number), depending on the nature of the lest for apprarsal melhod bias The rollback factor can vary both within and across regions; however, variations within a region are constrained within a given range. Table 1.2 presents the rollback factor ranges for fiscal year 1988. As indicated in table 1.1, we considered two versions of the rollback factor in estimat- ing the parameters of the, model to achieve different perspectives on the test for an appraisal method bias. We believe these two perspectives were necessary, in part, hccaause,as can be seen in table 1.2, the average rollback factor for Region 1 was appreciably different from that of all other regions. Page 35 GAO/RCED90-135 Federal Timber Sales Appraisals Appendix I An Analysis of Appraisal Method Bias in Advertised Prices of Forest Service Timber Sales Table 1.2: Rollback Factor and Appraisal Method by Region for Fiscal Year 1988 Rollback factor range Region Appraisal method (percent) 1 Tra%actw- evidence 47” i! TransactIon evvjence -5 10 .~~ 3 Transactjon ewdence 5-10 4 Transaction ewdence 5~ 10 5 Residual value 11-15 6 Residual value 10 8 TransactIon ewdence lo-25 9 TransactIon evidence 1 5” “Represents the actual average rollback factor for 1988 Averages were not obtalned from all regions Substituting the specific factors described in table I. 1 for the general terms of equation 3, and taking the natural logarithm of both sides of equation 3, results in equation 4, which can be estimated. The reason for the logarithmic transformation is to convert what we assume to be a multiplicative error structure for equation 3 to an additive error struc- ture in equation 4 in order to facilitate estimation. Equation 4: Ln(ovcrbid percentage) = Ln[l + Al + A2*(1/BIDDERS) + A3*SEALED + A4*(SEALED/BIDDERS) + A5*SIZE + AG*(SIZE/ACRES) + ATHIGHBID + A8*NOSBA + AS*SALEMETH + AlO*ROADS + Al l*FORMG + Al2”IWRMS + AlS*FORMT] + Ln 11 + A14*NOSALVG +A15*TEA] - Ln[l - rollback factor], where Al through Al5 ar‘~ i he parameters to be estimated. The specific factors listed in table I. 1 are substituted into equation 3 in a straightforward manner with three exceptions. First, BIDDERS is inverted to impost t.he assumption that a change in the number of bid- ders has a relatively larger effect on sales price (or overbid percentage) if the number of bidders is small to begin with rather than large. In other words, we assume t,hat adding one more bidder will introduce rela- tively more competition if the number of initial bidders is only 2 or 3 rather than 10 or 11. Second, we introduce the interactive term (SEALED/BIDDERS) to account for the fact that the number of bidders is unknown to each bidder in sealed bid sales, whereas the number of bidders is known by all in oral auctions. This term will permit BIDDERS to have different effects on the overbid percentage, depending on the bid method. Third, we include a constant term Al among the administra- tive and market fac,tors to account for any factors which are not explic- itly represented in the expression. Page 3fi GAO/RCEDW135 Federal Timber Sales Appraisals Appendix I An Analysis of Appraisal Method Biasin Advertised Prices of Fwest Srrvice Timber Saks We estimated the parameters of equation 4 from Forest Service data on timber sales for fiscal yctar 1988 using nonlinear least squares regression analysis.’ The entire data set contains 3,3 16 sales but we used only 2,801 to estimate equation 4. We excluded the six sales from Region 10 (Alaska) from the sample because that region is different in many ways from all others. We also excluded 180 oral auction sales with only one bidder and 161 direct sales (sales made after an auction with no bidders) because they provide no information on how the various factors, includ- ing an appraisal method bias, contribute to the overbid percentage. For these sales, the lack of’ direct competition results in a zero overbid regardless of any other factors. IIowever, WC included sealed bid auc- tions with one bidder in the sample because, when the bidding is sealed: the one bidder will not know prior to submitting his/her bid that there are no other bidders, or competitors, for that sale. In addition, we excluded 168 sales from the sample because either some data fields were missing or we detected inconsistencies indicative of data entry errors. Since we consider the issue of whether rollback factors should be associ- Estimation Results ated with the appraisal process as unsettled, we present one set of results, in table 1.3, on the basis of the assumption that rollback factors are associated with t hcsappraisal process, and a second set of results, in table 1.4, on the basis 01 the assumption that there is no such associa- tion. When considered associated with the appraisal process, variations in rollback factors across regions reflect differences in the application, rather than theory. of t hc appraisal process. Therefore, we can apply an alternative perspect iv<, to t.he results in table I.3 as compared with those in table 1.4, in that the latter isolate t,he theoret,ical differences (apart from application diff‘t~r(~nces) between the two appraisal methods because rollback factor\ arc disassociated from the estimate of bias for table I.4 results. We present two sets of results from the estimation of equation 4 in table 1.3. The first, set is based on the sample of the 2,801 sales described Page 37 GAO/RCED-90-135 Federal Timber Sales Appraisals Appendix I An Analysis of Appraisal Method Bias in Advertised Prices of Forest Service Timber Saks above, while the scc.ond set is based on a sample that excludes Region 1 sales from the first set, sample, resulting in a sample of 2,571 sales. WC estimated equation 4 with and without Region 1 sales to examine the sensitivity of the estimation results to the fact that the rollback facto1 used in Region 1 is c,xtraordinarily large relat.ivc to that used in the other regions. We obtained the estimates presented in table 1.3 by imposing a constant rollback factor across all regions. Specifically, we adjusted the depcn- dent variable, overbid percentage. according to the weighted average of the rollback factors nationwide (the latter version of the rollback factor described in table I. 1 ).’ This assumption forces any effects on overbid percentages caused by actual differences in rollback factors to be reflected, at least in part, in the estimate of A1.5, the parameter for transaction cvidencc met hod that is used to test for an appraisal method bias. In other words, 1his assumption results in the treatment of rollback factors as associat4 with the appraisal process. The statistical signifi- cance of the paramc~tc~r Al5, then, is interpreted as a test for the pres- ence of the combination of theoretical and application components (including rollback factors) of any appraisal method bias. Page 38 GAO/RCED-90-135 Federal Timber S&s Appraisals Appendix I An Analysis of Appraisal Method Bias in Advertised Prices of Forest Service Timber Sales Table 1.3: Estimation Results When the Rollback Factors Are Considered to Be All regions Excluding Region I Part of the Appraisal Process Variate Parameter Estimate T-stat. Estimate T-stat. CONSTANT Al 1 53 17 56” 1 32 1579 1 /BIDDERS A2 -2 61 -15.83” -2 22 -14 22’ SEALED A3 - 56 -7 32’ - 25 ~ -3 T$+ SEALEDjBDDRS A4 1 75 10 43” 1 34 8.30’ SIZE A5 002 3 14” 002 343’ SIZE/ACRES A6 - 19 -I a7 - 14 -146 HIGHBID A7 10 2 47” 07 1 61 NOSBA A8 18 4 59” 15 4 09” SALEMETH A9 06 1 23 OS- 1 04 ROADS A10 - 11 -3 23” - .09 -2 63” FORM6 All 018 ?7 0003 01 FORM9 Al2 - 02 - 44 07 1 49 FORMT Al3 - 1i- -2 53” -.05 -1 02 NOSALVG Ai - 25 -1531a - 21 -12.28’ TEA Al5 - 018 - 83 - 15 -6 i+ R-SQUARED 30 33 ‘iStgnkxntly different from zex ,It the 95.percent canftdence level for a two-talled test In general, the two set5 of results in table I.3 are similar. One notable exception is the estimate of the parameter associated with the transac- tion evidence method variable, A15. When we include all regions, the estimate of Al 5 is not statistically significant, suggesting that there is no appraisal method blas in advertised prkes. However, when we exclude Region 1 sales from the sample, Al5 is statistically significant and negative, suggesting that overbid percentages are smaller, and con- sequently. adverti& prices are relatively higher, when the transaction evidence method is usc~l by the Forest Service to determine advertised prices.; The statistical significance of Al5 is sensitive to the inclusion of Region 1 in the sample for two reasons: (1) these estimates are made with the assumption that rollback factors are a component of the appraisal pro- cess so that differcnccs in the factors across appraisal methods are Page 39 GAO/RCED-90.135 Federal Timber Sales Appraisals Appendix I An Analysis of Appreisal Method Bias in Advertised Prices of Forest Service Timber Sales reflected in the estimate of Al5 and (2) Region 1 sales were all done with the transaction evidence method, and the average rollback factor applied to Region 1 sales was more than twice that of all other regions.” The inclusion of Region 1 in the sample, then, appreciably increases the average rollback factor for all transaction evidence method sales rela- tive to residual valur, method sales in the sample. Since larger rollback factors result in lowtar advertised prices, all else equal, it is not surpris- ing that the significance of Al5 is sensitive to the inclusion of Region 1. These results suggest that such a large difference or inconsistency in the application aspect (use of rollback factors) of the different appraisal methods is capable of’dominating the effect on advertised prices of any theoretical differences, or differences prior to rollback adjustments, that might exist between appraisal methods. Results in table I.3 xvhich are consistent with or without Region 1 sales include negative and significant parameter estimates both for l/BID- DERS, A2, suggesting that the overbid percentage rises with the addi- tion of bidders (i.e., more competition), and for NOSALVG, A14, suggesting that a salt. classified as salvage implies a lower advertised price and therefore a greater overbid percentage. Further, the parame- ter estimate for SEALED, A3, is negative and significant, suggesting that sealed bid auctions result in smaller overbid percentages. However, this result must be inter-pret,ed in conjunction with the interaction term SEALED/BIDDERS. which is intended to account for the fact that bid- ders in sealed auctions do not know the number of other bidders with whom they are competing. The estimate of its parameter, A4, is positive and significant, ant1 when evaluated at t.he mean number of bidders in sealed bid auctions suggests that sealed bids do not result in a smaller overbid percentagt, than oral auctions.’ Other consistent results include positive and significant parameter esti- mates for SIZE ant1 NOSHA, A5 and AS, suggesting that there are econo- mies of scale (lowtxr average costs) associated with large timber volumes and large firms. rts;l)ectively, that are not adequately captured in the Page 40 GAO/RCED-90-135 Federal Timber Sales Appraisals Appendix I An Analvsis of Aoaraisal Method Bias in Advert&d PI& bf Forest Service Timber Sales appraisal or advertised price, and that can result in larger overbid per- centages.x One last consistent result is the negative and significant parameter estimate for ROADS, AlO. We expected this result because A10 serves to mathematically correct for a distinction between the observable and the true advertised prices for sales with roads. Road costs are included in the observable advertised price and, therefore, also are reflected in the sales price (i.e., road costs are included in both the numerator and denominator of overbid percentage), but the government compensates the winning bidder for those road costs.” The negative esti- mate of AIO, then, accounts for a negative adjustment to overbid per- centage for sales with roads because the observable overbid percentage would have been greater had road costs been netted out of both the numerator and denominator of overbid percentage, as is effectively accomplished from the perspective of bidders through government com- pensation for road costs Table I.4 presents results of an alternative test for the significance of an appraisal method bias. For this estimation, we adjusted overbid percent- ages by the mean of the range of the rollback factor for each region to reflect advertised prices prior to the application of rollback factors.‘” This effectively precludes rollback factor differences from influencing the estimate of A15. The estimate of A15, then, should reflect all theo- retical and application components of a bias in appraisal methods apart from rollback factors, and thus will not be sensitive to rollback factor differences as are the results in table I.3 (where sensitivity to rollback factor differences across appraisal methods resulted in our presentation of results determined from samples with and without Region 1). Fur- ther, if we assume that any application components of a bias apart from the rollback factors are small, then any appraisal method bias evident in the estimate of Al5 would be interpreted as entirely due to theoretical differences in appraisal methods. Page 41 GAO/RCED-SO-135 Federal Timber Sales Appraisals Appendix I An Analysis of Appraisal Method Bias in Advertised Prices of Forest Service Timber Sales Table 1.4: Estimation Results When the Rollback Factors Are Not Considered as All regions Part of the Appraisal Process Variate Parameter Estimate- T-stat. CONSTANT Al 1 53 17 75” l/BIDDERS A2 -2 52 -15 64” SEALED A3 - 37 -4 82” SEALED/BDDRS A4 1 59 9 54” SIZE A5 002 3 29” SIZE/ACRES A6 - 16 -1% .-. ,- - HIGHRIO A7 09 2 12 NOSBA A8 12 3 ooa SALEMETH A9 0% j4 ROADS A10 - 07 -2 02” FORM6 All - 004 - 10 F&l9 Al2 13 2 52” FORMT Al3 004 10 NOSALVG Al4 - 19 -12 4ld TEA Al5 - 19 -10 21” R-SQUARED .34 %gnlflcantly d!fferent fromi zero at the 95 percent confidence level for a ivmtakd test The results presented in table I.4 are similar to those presented in table I.3 for the sample which excludes Region 1 sales. Specifically, the csti- mate for parameter Al5 is negative and significant, suggesting that the overbid percentages are smaller, and therefore advertised prices are rel- atively higher, when the transaction evidence method is used by the Forest Service to determine advertised prices. In comparing the size and significance of the estimates for parameter A15, the results suggest that the appraisal method bias is stronger when rollback factors are considered apart from the application component of the appraisal met hods. On the basis of the sample means of the different factors of equation 1 and the estimated parameters in table I.4 and table I.3 (for the sampltl csxc*luding Region 1 sales), we evaluated the estimates of Al5 in terms of’tlow much higher advertised prices have been when determined from t he transaction evidence method rather than the residual value method. On the basis of the results presented in table I.3 for the sample t.hat excludes Region 1, or if rollback factors are consid- ered as part of thtb appraisal process but applied in a reasonably consis- tent manner, our estimate of Al5 suggests, with 95-percent confidence, that the transact,ioll evidence method will result in advertised prices from 14 to 29 perc.ent greater than the residual value method. On the basis of the results presented in table 1.4, or if rollback factors are not Page 42 GAO/RC,ED-SO-135 Federal Timber Sales Appraisals Appendix I An Analysis of Appraisal Method Bias in Advertised Prices of Forest Service Timber Sales considered a part of the appraisal process, our estimate of Al5 suggests, with 95percent confidence, that the transaction evidence method will result in advertised prices from 22 to 37 percent greater than the residual value method. Combining these two analyses, or regardless of the association between rollback factors (when consistently applied) and the appraisal process, our estimates of Al5 suggest that the trans- action evidence method results in advertised prices between 14 and 37 percent greater than under the residual value method. We conducted several other estimations to examine the sensitivity of the Sensitivity Analysis results presented above to alternative samples and specifications of the model. Specific alternative samples we tried included dropping all sal- vage sales, adjusting overbid percentages to exclude road costs from both the sales and advertised price components, and excluding some observations which were suspected of being influential (too highly weighted). The alternative specifications we examined included a log lin- ear. rather than nonlinear, structure, and different locations for the con- stant term in the nonlinear structure version. Finally, we also compared the mean overbid percentages across appraisal methods but without any attempt, to control for the effects of other factors. None of these alterna- tivc estimations or approaches produced results concerning the cxis- tence of an appraisal method bias which were different from those presented above. The results presented above suggest that the question of the existence of Conclusions an appraisal method bias is dependent on the role assigned to the rollback factors in the appraisal process and the extent to which the rollback factor is similarly applied. If rollback factors are considered as a component in the application of appraisal methods, then our results suggest that there was not an appraisal method bias for fiscal year 1988. Ilowever, the results also suggest that an appraisal method bias would have been found, indicating that transactions evidence appraisals result in advertised prices from 14 to 29 percent higher than residual value appraisals, if the rollback factor for Region 1 alone had been more in line with that of all other regions, or put differently, if rollback fac- tors were applied in a more consistent manner across regions. Further, this bias appears strongest, resulting in advertised prices, determined with the transaction evidence method, of from 22 to 37 percent greater than those determined with the residual value method, when rollback factors are considered not to be associated with the appraisal methods in any way. Paye 43 GAO/RCED-99-135 Federal Timber Sales Appraisals Appendix 1 An Analysis of Appraisal Method Bias in Advertised Prices of Forrst Srrvire Timber Sales Higher advertised prices achieved with the transaction evidence method rather than the residual value method imply that the government would likely increase revenue on sales for which there is little or no compcti- tion if the advertised prices were determined by the transaction tvi- dence method. In 1988, at least 167 sales involved no competition; specifically only one bidder was involved in an oral auction. There may have been other sales with less aggressive behavior on the part of the bidders, although WY‘can cite no evidence to document this possibility. Page 44 GAO/RCED-90-135 Federal Timber Saks Appraisals Appendix II Potential and Actual Unrecovered Sales Preparation and A dministration Costs on Fiscal Year 1988 Timber Sales, Forest Service Advertised sales Sold sales Total number of Number Potential Number Actual Forest sales below cost nonrecovery below cost nonrecovery Region 1 Beaverhead 5 4 $231,785 4 $116,853 Bitterroot 9 9 610,700 5 210,722 Panhandle 80 4,370,841 60 - 19 ~~ ~~ 356,682 ~~ Clearwater 35 25 1411 183 12 316.003 Custer 3 3 164417 - 3 ~- ~~84,377 ~~~~ Deerlodge 8 8 858971 6 221,374 Flathead 29 28 1,977.008 14 402,795 GFdlatln . . . . . Helena 9 9 476,428 8 225,432 Kootem 82 67 3 702,584 20 447,488 Lewis & Clark 6 ~6 278,770 6 217,774 Lolo 23 23 2,973 285 17 1,067,483 - Nezperce 15 12 1.959,294 6 245,656 Totals, Region 1 304 254 19,015,266 120 3,912,639 Region 2 iig Horn 1 41,850 . . - ~~- Black Hills 4 246,814 3 245.061 Grand Mesa-Uncompahgre- Gumson 16 530,724 16 529 232 Medlclne Bow 3 29,102 2 Y3.337 Nebraska 2 17,761 2 17,759 RIO Grande 5 20 818 1 2,839 Arapaho - Roosevelt 6 218,199 6 217,565 Rout7 4 379,700 4 369,168 - Pike & San Isabel 310,917 3 308,576 San Juan- 14,539 2 8,165 Shoshone 24,977 4 17,779 White River 1.109,597 6 995,277 Totals, Region 2 57 2,944,998 49 2,734,758 (contmued) Page 45 GAO/RCED-90-135 Federal Timber Sales Appraisals - Appendix II Potential and Actual Unrecovered Sales Preparation and Administration Costs cm Fiscal Year 1988 Timber Sales, Forest Service Advertised sales Sold sales Total number of Number Potential Number Actual Forest sales below cost nonrecovery below cost nonrecovery Region 3 Apache~Sltgreaves 16 2’ 6,774 . . CXSX 12 R 185,962 6 110,382 Clbola 2 . . . . Coconlno 8 1 33,282 I 1,962 Coronado . . . . . Glla 3 iI> 516.360 2 483.598 Kalbab 9 1 11 018 1 11.018 Lincoln 7 248,415 5 187 084 Prescott 2 Lr, 4,147 2 4,145 Santa Fe 11 Ia 129,994 4 125.259 Tonto 2 . . . . Totals, Region 3 72 28 1,135,952 21 923,448 Region 4 Ashley 17 Ii 442,538 11 317 110 Ease 20 L 53,800 1 77,630 Brldger-Teton 3 LI, 55,415 2 35,005 Caribou 2 a 505 . . Challis 1 30,943 1 12,529 Dwe 5 . . . . Fish Lake 1 34,178 1 25,070 Humboldt . . . . Manil-LaSal i 350 1 350 Payette 23 4 016 1 3.440 Salmon -~ 9 359 736 5 229 357 Sawtooth . . . . . Targhee 27 I F: 701 286 16 634,179 Tolyabe 2 L 33 838 2 32 905 Ulnta 1 . . . . Wasatch 4 I 96,135 3 83 753 Totals, Region 4 118 59 1,820,740 44 1,391,328 (continued) Page 46 GAO/RCED-96-135 Federal Timber Sales Appraisals Appendix II Potential and Actual Unrecovered Sales Preparation and Administration Costs on Fiscal Year 1988 Timber Sales, Forest Srrvic~ Advertised<* Sold sales ._._. ..-.. .--. _. Number PCktential Number Actual Forest sales below cost nonrecovery below cost nonrecovery Region 5 Angeles . . . . . Cleveland . . . . . Eldorado 26 8 75,351 2 10,774 lny0 ~5 4 185 267 . . Klamath 51 46 6,275,129 14 1504.398 Lassen 44 8 1,711,969 2 1,051,062 Los Padres . . . . . Mendoclno 17 12 984,927 6 778,992 Modoc 8 2 384,906 1 33,475 SIX hers 21 a 1 316.655 1 97,043 Plumas 85 26 763.748 9 421,431 San Bernadlno . . . . . Sequoia 13 10 765,352 5 322,498 Shasta lb I 307.605 . . Sierra 26 3 389,776 1 341 -I I Stanislaus 41 35 3,164,871 13 176,457 Tahoe 77 “l I? ,V 666.376 4 71,156 Trlnlty 36 26 2 543,782 1 40,222 Lk Tahoe Basin . . . . . Totals, Region 5 426 208 19,535,714 59 4,510,919 Region 6 Deschutes 21 8 1 193,485 7 Y66,YtV Fremont 28 2 169,639 2 169,639 Gifford Plnchot 75 6 403,580 2 76,391 Malheur 44 4 49,515 2 44,670 Mt Baker-Snoqualmle 57 3 53,141 1 7,164 Mt Hood 57 9 527,745 4 216,416 Ochoco 13 . . . . Okanogan 9 4 285376 3 229,182 Olympic 39 8 275,674 3 18,975 Rogue her 49 5 540,417 1 207,906 Sisklyou 41 18 3 070.609 2 65,495 Sluslaw 53 2 7,756 2 -7,756 Umatilla 15 6 664,442 4 590,124 (continued) Page 47 GAO/RCED-90.135 Federal Timber Sales Appraisals Appendix II Potential and Actual Unrecovered Saks Preparation and Administration Costs on Fiscal Year 1988 Timber Sales, Forest Service Advertised sales Sold sales Total number of Number Potential Number Actual below cost below cost nonrecovery Forest sales .~ nonrecovery ~.~ Umpqua 127 1 15,461 1 10,261 Wallowa-Whltman 31 0 674,742 6 380,869 Wenatchee 8 I 3,573 . . Wlllamette 27 2 9,230 1 7,658 Wlnema 34 13 586,292 6 230,442 Colvllle 31 9 231,206 4 33,614 Totals, Region 6 759 110 0,761,003 51 3,263,544 Region 6 Alabama 32 a 50,554 5 18,887 Daniel Boone 25 13 148,521 5 37.790 Chattachooche- Oconee 30 10 48,345 6 25.719 Cherokee-- 15 5 100,350 5 83,256 Florida 32 4 19,284 3 8,381 Kisatchle 85 6 8,856 2 1,423 MISSISSIPPI 130 ;‘5 358,490 24 323.399 Geo. WashIngton 32 :i 1 683,300 30 621,663 Ouachlta 109 ‘4 74 097 7 26,340 Ozark St FrancIs 47 I 105,561 5 47,289 North Carolina 39 33 829,026 24 426,033 Francls Marlon Sumter 53 ‘0 103,226 6 82,716 Texas 65 ‘> 222,941 6 180,342 Jefferson 27 25 455,297 24 365,841 Totals, Region 6 721 216 3,207,848 152 2,249,079 Region 9 Chequameqon 49 749 954 45 573 266 Chlppewa 41 450,391 - 22 179.229 Huron~Manlstee 32 62,489 9 27,805 Mark Twain 98 30.704 - 5 II ,540 Nlcolet 35 438,869 17 195,628 Ottawa 44 489.736 24 315,510 Shawnee a 452,492 7 445,515 Supenor 46 4C ~1,093,290 ~. 38 838,211 Hiawatha 38 ‘3: 335,651 25 183.590 Wayne-Hoosier 5 1,635 . . Allegheny 38 . . . (contjnued) Page 48 GAO/RCED-90135 Federal Timber Sales Appraisals Appendix II Potential and Actual Unrecovered Sales Preparation and Administration Costs on Fiscal Year 1988 Timber Saks, Forwt Service -____ Advertised sales Sold sales Total number of Number Potential Number Actual Forest sales below cost nonrecovery below cost nonrecovery .____- Green Mountain 18 16 385,577 16 257,696 Monongahela 28 s 43,444 4 35,049 White MountaIr 20 14 125,928 5 18,468 Mamstee 28 . . . . Totals, Region 9 528 282 4,660,160 217 3,oai,507 Region 10 Chugach . . . . . ~- - -_ Tongass 6 5 765.266~~-~~~. 1 45,332 Totals, Region 10 6 i 765,266 1 45,332 Grand Totals 3,030 1,219 561,847,827 714 $22,112,554 Page 49 GAO/RCED-W-135 Federal Timber Sales Appraisals Appendix III Major Contributors to This Report Gus Johanson, Assist ant Director Resources, John P. Murphy, Jr., Assignment Manager Community, and Gene Wichmann, Assignment Manager Scott Smith, Economist Economic Development Division, Washington, D.C. Leo H. Kenyon. Regional Management Representative Seattle Regional Office Jill J. Lund, Evaluator-in-Charge Hugo W. Wolter, I+,aluatoI Stan Stenersen, Evaluator (021025) Page 50 GAO/RCED-SO-135 Federal Timber Sales Appraisals sbouid be sent to: The @mt five copies of each report are free. Additional copies are $2.00 each There is a 26% discount on orders for 100 or more copies mailed to a s~ddress. 4Brders must be prepaid by cash or by check or money order made out to~t.he S@erintendent of Documents, ‘. -, : _.: -- .,,? ltleqwsa for copies of GAO reports should be sent to: UJ!& General A ccountingoffice Post t3Bice Box 6015 Gaithe~b~ Maryland 20877 Telephone 202-276-6241 The first five copies of each report are free. Additional copies are $2.00 each. There is a 25% discomnt on orders for 100 or more copies mailed to a singie address. Orders must be prepaid by cash or by check or money order made out tq the Superintendent of Documents.
Federal Timber Sales: Process for Appraising Timber Offered for Sale Needs to Be Improved
Published by the Government Accountability Office on 1990-05-02.
Below is a raw (and likely hideous) rendition of the original report. (PDF)