UNIVERSITY OF ILLINOIS UBRARY AT URBANACHAMPAJ6M ** STACKS Digitized by the Internet Archive in 2011 with funding from University of Illinois Urbana-Champaign http://www.archive.org/details/neighborhoodzoni626colw Faculty Working Papers College of Commerce and Business Administration University of Illinois at Urbana-Champaign FACULTY WORKING PAPERS College of Commerce and Business Administration University of Illinois at Urbana-Champaign November 12, 1979 NEIGHBORHOOD, ZONING, AND THE VALUE OF URBAN LAND Peter F. Colwell, Associate Professor, Depart- ment of Finance Paul Asabere, Doctoral Student, Department of Finance #626 Summary: This paper is an empirical study of land values in Champaign-Urbana. Land value depends on lot size relative to the typical lot size in the neigh- borhood. This result supports the notion that minimum lot area zoning can have externality effects. The supply effects of existing zoning appear to dominate any externality effects which might exist. This suggests that the zoning in Champaign-Urbana does more harm than good. Several location variables are introduced to deal with the fact that the value of land would vary across land use zones in the absence of governmental zoning. NEIGHBORHOOD, ZONING, AND THE VALUE OF URBAN LAND I. INTRODUCTION Various models have been used to explain urban land values. These models have exhibited differences in functional forms, levels of aggre- gation, and the explanatory variables selected. Most often linear func- tions and aggregate data are employed. In those few studies where dis- aggregation makes neighborhood variables possible, those which relate specifically to land characteristics are generally primitive and relate only to topographical features. Finally, the tendency to include only a single zoning classification or a small range of classifications has made it impossible to detect the impact of governmental zoning on land value within the empirical model. In contrast, this paper uses micro data, a transcendental function, a new neighborhood variable, and vari- ables to capture the effect of zoning at the two ends of the zoning hierarchy. Neighborhood variables have been utilized to explain urban property values in a number of studies. Some examples are average assessed value [3]; value of improvements [22]; degree of blight [9]; percent non-white [3, 25]; median income [1, 4, 22]; crowding index [3]; air pollution [2, 19]; and developed area [15]. Focusing on the aspect of land that differentiate neighborhood characteristics, this paper introduces a new variable — lot area relative to the typical or average neighborhood lot area. Zoning has also been used to explain urban property values in a number of studies [7, 8, 14, 17, 18, 21, 23], In order to make it pos- sible to detect the impact of governmental zoning on land values, the -2- sample in this paper includes all zoning classifications. A principal hypothesis of this paper is that governmental zoning has its impact on urban land values primarily through supply rather than externality effects. This hypothesis implies that governmental zoning is alloca- tively inefficient. In addition to the neighborhood and zoning variables, it is impor- tant to include location variables. The theory of urban land economics tells us that different land use zones would have different values in the absence of governmental zoning so the effect of governmental zoning can only be measured while holding location constant. Location variables have been used in several studies in various forms. Examples are dis- tance to CBD [1, 3, 4, 13, 22]; job access potential [3, 22]; distance to a regional shopping center [25]; and an urban function access index (uses time-distance) [9]. This paper utilizes five location variables: distance to a center of economic activity and dummy variables for cul-de-sac, growth path, corner lot, and busy street. II. HYPOTHESES Neighborhood : The primary importance of neighborhood variables (i.e., technological externalities or intra-neighborhood effects) in determining urban land values is undeniable. Those who believe that these effects are trivial have generally not considered the rapid rate at which the effects fall off with distance, and used definitions of neighborhood which are too large. Of course, there are substantial problems involved in measuring the overall ambient quality of a neigh- borhood or the various components of quality. The obvious solution is to use dummy variables to characterize neighborhoods. But if dummy -3- variables are used to represent neighborhoods and neighborhoods are small, it is unlikely that a large enough number of sales will be found to provide sufficient degrees of freedom. Thus, for empirical as well as theoretical reasons, it is desirable to attempt to measure the abstract characteristics of neighborhoods. In order to move into this direction, this paper includes a variable that takes into account the impact of one neighborhood characteristic, the relationship of the lot's size or area to the average lot area in the neighborhood. It is hypothesized that the values of larger lots are pulled down while the values of smaller lots are raised according to their positions relative to a typical or average area. At least for residential property, this may be explained by the feeling of spaciousness that one experiences within neighborhoods of typically large lot sizes, and an oppressive-cramped feeling in neighor- hoods with small lot sizes. (1) SP ±j - Ica^/A/* where 1 > £1 > 0, 62 < 0»3i> 1 62 ! SP. . = the selling price of ith lot in the jth neighborhood, A.. = the area of lot i in neighborhood j, A. = the average lot size or area in jth neighborhood, and k = everything else. Equation (1) could be rewritten so that the arguments of the selling price function are lot are and average lot area rather than lot area and the ratio of lot area to average lot area. While this form is mathematically equivalent, it is not econometrically equivalent. This is because the ratio has lower colinearity than average lot area has with lot area. OO OO OO 00 -o -o -o ■o -» — » no ro — i r-j SELLING PRICE CD 33 7 i. TO -<• TO- -"• TO - -^ PO —J <-"- — ■ + + TO TO w (S3 rv> > > -" 1 ro i Ol TO TO ro l\J II * > > IN) -4- Figure 1 illustrates equation 1. Since lot 1 in neighborhood 1 is smaller than average (a,, < A-,) as shown in Figure 1, its selling price would be expected to be more than if it were in a neighborhood where it happened to be the average (SP, , > SP.. where a.. = A = a...). Similarly, suppose that lot 2 in neighborhood 1 is larger than average (a 2 - > A..) . As in Figure 1, its selling price would be expected to be less than if it were in a neighborhood where it happened to be the average (SP2, < SP2 where a = A z = a^J. Note that as average neighborhood area increases from A-^ to A£ the function which relates intra-neighborhood lot areas with selling price shifts upward. The selling price-lot area relationship in equation (1) and illustrated in Figure 1 has the same look as Deusenbery's well-known relative income hypothesis or Friedman's permanent income hypothesis without the propor- tionality assumption. Thus, this relationship will be referred to as the relative land area hypothesis. Zoning : What is variously known as hierarchical zoning, cumulative zoning, or progressively inclusive zoning operates by allocating a zone to a particular land use or any higher use in the government ally defined hierarchy. The rationale for hierarchical zoning suggests that it restricts the flow of negative externalities from lower to higher land uses in the hierarchy. If this were the only effect of governmental zoning, the value of the highest uses in the hierarchy would be raised as a result of the protection provided by the zoning ordinance (i.e., ceteris paribus). That is, those who desire to use land for residential purposes, usually the -5- highest use in the hierarchy, are able to choose from land in any zone, but would be willing to pay more for land in the protected residential zone, holding location and other factors constant. Thus the externality argument, which provides the rationale for the legal application of police powers to governmental zoning, implies that there should be a premium paid for residentially zoned land. On the other hand, governmental zoning may be put to other purposes. Special interests in and out of government may be able to shape govern- mental zoning to serve their own ends [11]. A local government may engage in fiscal zoning in order to directly protect its purse and indirectly beggar neighboring governments. Planners may have their biases. "The almost universal preference, as expressed in zoning statutes, for single- family dwellings probably inspires planners to ... overallocate land for single-family use." [14] • If planners are ideologically at odds with the expansion of business activity locally, they will have little trouble finding political allies. "The owners of land currently zoned for (commercial and industrial) use prefer to limit its supply. They may be joined in their efforts to restrict supply by owners of resi- dential land who fear the effects of negative externalities" [14] . Thus, zoning may not only increase efficiency by separating incom- patible land uses and reducing the flow of negative externalities. It may also create inefficiency by distorting the supply of land to the var- ious uses. The nature of hierarchical zoning causes such distortions to be asymmetric. It can only overallocate land to the highest uses and underallocate land to the lowest uses. The reverse of underallocating land to the highest and overallocating land to the lowest uses is -6- impossible. Thus where there are supply effects from governmental zoning, there would be a tendency for residential land values to be depressed and commercial land values to be raised by the zoning. Recall that the exter- nality argument suggests that there would be a premium paid for residential land. Thus, any net effect of residential zoning on land value indicates whether zoning operates primarily to improve the allocation of land or to misallocate land. If the partial effect of commercial zoning is to increase land value, this would be evidence of the misallocation at the low end of the zoning hierarchy. Location ; Five location variables are employed in this study. The first of these variables is distance to the center of activity. For Champ aign-Urb ana, a typical campus town and the subject of this study, the north end of the University of Illinois 'quad' is the center of ac- tivity. The university serves as the principal regional employer, the main night-life rendezvous, and campus town at the north end of the quad serves some commercial functions. The downtowns (CBDs) for Champaign and Urbana are not explicitly used as proxies for the centers of activity because of their relative decline in importance in recent years along with the development of peripheral shopping centers. However, it should be noted that the north end of the 'quad' is approximately on a line halfway between the two CBDs and thus may act as the centroid of the activity which remains. The second location variable measures the impact that cul-de-sac location has on land value. The inclusion of this variable is based on our belief that the cul-de-sac plays 3 main roles. First, it lends it- self to flexibility in arrangement and orientation of houses and, thus, -7- provides for more variety in spatial arrangements. Second, the cul-de-sac reduces pedestrian, bicycle, and automobile traffic and, thus, reduces noise and dirt and increases security. Finally, neighbors around a cul- de-sac may be more socially integrated than those on traditional grid- iron patterns, because the cul-de-sac neighborhood is well defined and small. These factors promote club formation and cohesion as well as the resulting public goods production (e.g., manicured lawns, freshly painted facades, and help when needed). Based on such attributes, being on a cul-de-sac should have a positive impact on selling price. The third location variable is intended to pick up the impact of being in the path of rapid growth. Most developments south of Kirby/ Florida Avenue appear to be post 1960, and most post 1960 developments appear to be south of Kirby/Florida Avenue. Thus the growth path var- iable is a dummy indicating whether the lot is north or south of this street. The fourth location variable is included to capture the effect of corner location on land value. It is expected that corner location increases land value. This is especially so for commercial properties. Corner location enhances the visibility of the property. It provides more access and more exposure due to the double frontage. Corner loca- tion provides desireable separation for residential property. Thus, corner location is probably preferred for both residential and commercial land users. The corner location variable used in this study is a dummy indicating whether the lot is a corner lot or not. The fifth and final location variable is a dummy for high traffic volume streets. It is hypothesized that location on a busy street has a positive impact on land value. Commercial activity favors location on busy streets because of the visibility and high potential for attract- ing customers because of the sheer numbers who pass by the property. Time of Sale ; It is hypothesized that during the sample period, 1977 and 1978, land appreciated in value at a rate which was relatively constant and that the sale price of lot i depends on its time of sale in the following manner. Slot, (2) SP i = he where Bio = rate of appreciation, t^ = time of sale of i lot, and h = everything else. III. THE MODEL All the hypotheses developed above were brought together into the following equation: (3) SP i = B aJUa i ./A ) e2 exp[e 3 COMM i + Bi+SRES^^ + B 5 QUAD ;L + B6CdeS ± + B 7 GRTH i + & 8 CORN,. + BgHTP^ + BioMC^ where SP. = selling price of lot i, a. . = area of lot i in neighborhood j in thousands of square 2 feet, A.: = average area in the jth neighborhood (i.e., block) in thousands of square feet, COMM. = a dummy variable assigning 1 if lot i is located in a commercial zone and for all other zones, SRES. = a dummy variable assigning 1 if lot i is in a single- family residential zone and for all other zones, •9- QUAD. = distance in miles of lot i from the north end of the University of Illinois 'quad' CdeS. = a dummy variable assigning 1 if lot i is on a cul-de-sac and if it is not located on a cul-de-sac, GRTH = a dummy variable assigning 1 if lot i is located in the growth path — south of Kirby/Florida Avenue and if it is located north of it, CORN. = a dummy variable assigning 1 if lot i is a corner lot and if it is not. KTRF. = a dummy variable assigning 1 if lot i is located on a street with an average-daily traffic volume of 5000 or more and for less than 5000. MOS. = the month cf sale of lot i. i The sample data consist of all recorded sales in the cities of Champaign and Urbana during the years 1977 and 1978. The Sale Price data was taken from transfer tax and deed records while the lot size data was taken from platbooks . Zoning information for the city of Urbana was taken from the Champaign County Regional Planning Commission while that of Champaign was taken from the Champaign City (Planning) Office. The model was estimated by taking natural logarithms of both sides of equation (3) and utilizing Ordinary Least Squares. The results of the estimation are as follows: (4) In SP. = 1.934 + 0.416 In a - 0.211 ln(a. ./A ) + 0.405 COMM 1 (5.226) (3.923) J (-1.923) 1J J (1.461) - 0.702 SRES - 0.150 QUAD. + 0.266 CdeS. (-3.871) (-1.643) 1 (1.737) 1 + 0.304 GRTH. + 0.207 CORN. + 0.428 HTRF + 0.013 MOS. (1.866) 1 (1.391) X (1.535) (1.087) 1 (t ratios in parentheses; d.f. = 114) -10- The adjusted coefficient of determination is 0.35. (A correlation matrix for the explanatory variables is shown in Table 1.) The coeffi- cients on the In a.., In a. ./A., SRES . , CdeS . , GRTH., are significantly ij ij J i i i different from zero at the 90% level of confidence. The coefficient on the QUAD, variable is significantly negative (one-tail) at the 90% level of confidence. The coefficients on the COMM , CORN and HTRF dummy var- iables are significantly positive (one-tail) at the 90% level of confi- dence. The magnitude of the annual rate of appreciation is 15.6% this is the same rate estimated by Colwell and Sirmans [5] for the period 1969- 1975. However the coefficient here does not differ significantly from zero, whereas it does in the Colwell and Sirmans paper. The main reason for this remarkable difference is that as the urban bid-rent function shifts upward over time, the price of peripheral land in transition from agricultural to urban uses is determined by the agricultural land price and not by the height of the bid-rent curve. Most sales in the 1977-78 period were peripheral. Thus the coefficient on the month of sale variable is more indicative of the experience of agricultural land prices than urban land prices. There is independent evidence which suggests that agricultural land prices were relatively stable over the study period while they increased dramatically over the earlier period. The relative lot area hypothesis was borne out by the estimation. Figure 2 shows the estimated relations between value and area relative to average neighborhood area over the range of 250-18,000 square feet. In constructing Figure 2, it is assumed that the lot is north of Kirby, 1 mile from the 'quad,' zoned for single-family residential, not located on a cul-de-sac, not on a corner lot, not located on a busy street, and sold just at the end of 1978. vO o CM rH id • CM vO o • co ^-> O CM CM ^"\ • rH »x> n v— / x^ ii r-» P». Q. o\ o\ IvO rH , rH | • | • 1 (0 I « I £> I -C loo 1 <*> \CM 1 ^ 1 • 1 • ICO 1 "> lii 1 II la lc/> CO in . i— cr «/> o o o a; CM O cr> UD CM o CO CO VD CM (000.)30IHd 9NIT13: NVIQ3W -11- As shown in Figure 2, 3 plots were made: one shows the relation- ship between value and lot area given that lot area just equals average neighborhood lot area. The other 2 plots assumed that average neighbor- hood lot equals 6000 square feet and 12,000 square feet, respectively. The results as shown in Figure 2 are consistent with the hypothesis as presented in Figure 1. Land values of larger lots are pulled down while values of smaller lots are raised according to their positions relative to a typical or average area in the neighborhood. Note that as the aver- age increased from 6000 to 12,000 square feet, the function, which re- lated intra-neighborhood lot areas with selling price, shifted upward. The coefficient on a^. is significantly greater than and less than 1 at the 99% level of confidence, while the coefficient on (a.. /A.) is significantly negative at the 95% level of confidence. The coefficient on a., is significantly greater than the absolute value of the coeffi- cient on (a. ./A.) at the 95% level of confidence. The estimated coefficients on the zoning variables strongly suggest that governmental zoning has done more harm than good. The dummy variable C0MM. (Commercial) proved to have a substantial positive impact on land values. Commercial zoning adds 50% to value. The dummy variable SRES. (single- family residential), on the other hand, proved to have a substantial negative impact on land value. It appears to cause a 51% decline in value. The commercial dummy variable is significantly positive at 90% level of confidence while the single-family residential dummy variable is significantly negative at 99% level of confidence. The negative impact on land values of single-family residential zoning means that the supply effect has swamped any externality effects which might exist. -12- An interpretation of these results is that land in Chaiapaign-Urbana has been overallocated to residential uses and underallocated to commercial uses. The location variables all worked as expected. The land value gra- dient turned out to be .150. The magnitude of the coefficient on the 'cul-de-sac' dummy variable establishes that, all things being equal, a lot would be expected to gain 30% in value if it were located on a 'cul-de-sac' The GRTH^ (growth path) dummy variable proved to give positive impacts on land values. Location south of Kirby/Florida Avenue, the growth path, would be expected to lead to a 36% gain over values than north of it. The CORN, (corner lot) dummy variable establishes that, all things being equal, a lot would be expected to gain 23% in value if it were located on a corner. The HTRF. (high traffic) dummy variable proved that location on a busy street would be expected to lead to a 53% gain in value over other locations. IV. SUMMARY AND CONCLUSIONS This paper has offered some empirical evidence of the hypothesized relationship between urban land values and neighborhood, zoning, loca- tion and time. The model in this paper specifies that selling price of urban land is a function of both lot area and lot area relative to typical neigh- borhood area. This relative lot-area hypothesis is at least visually akin to Deusenbery's relative income hypothesis. The relative lot-area hypothesis provides empirical support for minimum lot-area zoning as an externality type of zoning. The empirical development of this concept is probably one of the most important contributions of this paper. -13- Zoning appears to do more harm than good. The coefficient on the commercial zoning dummy variable is significantly positive while the coefficient on the single-family residential dummy is significantly negative. These price effects are indirect evidence of distortions in the allocation of land caused by governmental zoning. It is suggested that land is being overallocated to the highest uses to the extent that any positive impact due to reductions of negative externalities must be swamped by this supply effect. The opposite side of the coin is that land is overallocated to the lowest uses. Our zoning results are inconsistent with the results of Maser, Ricker and Rosett for Rochester, New York [10] . But their comparisons were only within residential types and x^ithin commercial and industrial not across the entire spectrum. So they conclude that zoning did not alter market allocations whereas we not only find that zoning alters market alloca- tions but that is does so in a counter-productive way. A problem with their interpretation is that no effect would have the same look as off- setting effects. That is, the public good effects of externality type of zoning may be offset by supply effects. An alternative rationale for zoning may provide a weak defense of the supply effects which have been found in this study. If there are distortions in other markets (e.g., credit) which have differential im- pacts on demand for various land uses, then supply effects of zoning might be used to overcome the defects in these other markets. Whether those who zone have the information, analytical capability, and the legal authority necessary to do this is another question. Certainly, the use of the supply effect zoning to overcome the defects in other -14- markets is a circuitous approach. One wonders why it would not be superior to attack the problem directly. There may be yet another rationale for the supply effects of zoning. A stated objective of zoners in the study area is to "conserve the tax- able value of land." [26] Suppose that this is taken to mean that gov- ernmental zoning ought to produce higher aggregate land values than the market. This can be achieved via the supply effects of zoning if the constrained land use has a higher price elasticity of demand for land [17]. However, this method of increasing aggregate land values is un- certain in its effects, because the relative demand elasticities are far from obvious. When this method works, the government is utilizing the same techniques as a price discriminating monopolist would. Thus, even when the method successfully increases aggregate land value, it reduces the efficiency with which land is allocated. The problem with the commercial zoning variable is that it may be a proxy for the location factors which attract commercial activity. If governmental zoning merely follows the market, commercial activity would be found on higher priced land having the desirable location factors. But it can never be certain that the right location factors have been identified and included in the regression. Thus, it is possible to complain if commercial zoning shows positive effect that the impact of governmental zoning has not been detected. Rather, commercial is cap- turing the effect of some omitted locational factors. Therefore, loca- tion variables must be selected with care. The location factors which are commonly thought to attract commercial activity ought to be included. Centrality, location in the path of most urban growth, corner location, -15- and location on a busy street were used in this study. Land values were shown to be a negative exponential function from the University of Illinois quad. Locations in the path of most urban growth, corner location, and location on a busy street proved to give positive impacts on land values. Probably distance from the quad, cul-de-sac, and corner location, and location on a busy street are important for residential uses. -16- REFERENCES 1. Alonso, William. Location and Land Use . Cambridge, Harvard University Press, 1964. 2. Anderson, R. J., Jr., and T. D. Crocker. "Air Pollution and Residential Property Values," Urban Studies 8 (October 1971), 171-180. 3. Brigham, Eugene F. A Model of Residential Land Values . Santa Monica: The Rand Corporation, 1964. 4. Brodsky, Harold. "Residential Land and Improvement Values in a Central City," Land Economics 46 (August 1970), 229-47. 5. Clonts, Howard A. "Influence of Urbanization on Land Values at the Urban Periphery," Land Economics 46 (November 1970), 489-97. 6. Colwell, Peter F. and C. F. Sirmans. "Area, Time, Centrality and the Value of Urban Land," Land Economics 54 (November 1978), 514- 19. 7. Courant, Paul N., "On the Effects of Fiscal Zoning on Land and Housing Values," Journal of Urban Economics 3 (1976), 88-94. 8. Crecine, John P., Otto A. Davis and John E. Jackson, "Urban Property Markets: Some Empirical Results and Their Implications for Municipal Zoning," Journal of Law and Economics 79 (1967), 111-132. 9. Czamanski, Stanislaw. "Effects of Public Investment on Urban Land Values," Journal of American Institute of Planners 37 (July 1971), 289-300. 10. Davis, Otto A. and Andrew B. Whinston. "The Economics of Complex Systems: The Case of Municipal Zoning," Kyklos 17 (1964), 491. 11. Davis, Otto A. "The Economic Elements of Municipal Zoning Decisions," Land Economics 39 (1963), 375. 12. Downing, Paul B. "Factors Affecting Commercial Land Values: An Empirical Study of Milwaukee, Wisconsin," Land Economics 49 (February 1973), 44-56. 13. Knos, Duane S. "The Distribution of Land Values in Topeka, Kansas," In Spatial Analysis , pp. 269-89. Edited by B. J. L. Berry and D. F. Marble. Englewood Cliffs: Prentice Hall, Inc., 1968. 14. Maser, Steven M., William H. Ricker and Richard N. Rosett. "The Effects of Zoning and Externalities on the Price of Land: An Empirical Analysis of Monroe County, New York," The Journal of Law and Economics 20 (1977), 111-132. -17- 15. Mohring, Hebert. "Land Values and Measurement of Highway Benefits," Journal of Political Economy 69 (June 1961), 236-49. 16. Oates, Wallace E. "The Effects of Property Taxes and Local Public Spending on Property Values: An Empirical Study of Tax Capitali- zation and the Tiebout Hypothesis," Journal of Political Economy 77 (November-December 1969), 957-71. 17. Ohls, James C, Chadbourn R. Weisberg, and Michele J. White. "The Effect of Zoning on Land Value," Journal of Urban Economics 1 (1974) 428-444. 18. , "Welfare Effects in Alternative Models of Zoning," Journal of Urban Economics 3 (1976), 95-96. 19. Ridker, R. G. and J. A. Henning. "The Determinants of Residential Property Values with Special Reference to Air Pollution," Review of Economics and Statistics 49 (May 1967), 246-256. 20. Rueter, Frederick H., "Externalities in Urban Property Markets: An Empirical Test of Zoning Ordinance of Pittsburgh," Journal of Law and Economics 16 (1973), 313-349. 21. Stull, William J., "Community Environment, Zoning and the Market Value of Single Family Homes," Journal of Law and Economics 18 (1975), 535-557. 22. Wendt, Paul F. and William Goldner. "Land Values and the Dynamics of Residential Location," In Essays in Urban Land Economics , pp. 189-213, Los Angeles, University of California, 1966. 23. White, J. Michelle. "The Effect of Zoning on the Size of Metro- politan Areas ," _j£urnaJ L _of_JJr_baji_Jcan£mj£ p s 2 (1975), 279-290. 24. Wings, Lowdon, Jr. Transportation and Urban Land . Washington, D.C.: Resources for the Future, Inc. 1961. 25. Yeates, Maurice H. "Some Factors Affecting the Spatial Distribution of Chicago Land Values, 1910-1960," Economic Geography 41 (January 1965), 57-76. 26. Zoning Ordinance: City of Urbana, Illinois (1977). M/E/167 EC n to M M JO ss co O E3 o CO 3 03 p o co SB a u H- H- s— ' H- to H- O o O 1 o © 1 O o 1 O o • • • • • • • • • n o o O M o to o o M o IO oo to I- 1 .fc- vo o ^J M pd CO CO VO os to ^J VD VO ^J h? t M JN M o .fc- M ^J ■> to H- O to O Cn as OS Co Cn o X> o o o 1 O I o 1 O O 1 O n • • • • • • • • O- o ro o o o o H 1 M n> IO o o O0 to M N5 JN CO IO Cn to M as »J H to H> ^J •vl Cn O OO Os *> o oo 1 to 1 ho 1 o H 1 VO to vj CO < o o o o o O O 6 H M o o o M Co (-■ O N5 (-• ^J Cn to Cn 0) .> M VD oo Cn J> >-J k: cr >J ^J tO to Os to O H- H 1 ■~J to 1 »-J 1 CO OS (0 O o n H P9 o O o o O O H cr • • • • • • s ro H M o t-> o o o o H (B "■J 03 Cn Cn to OS co DJ vo JI- CO to CO o M- rr H H" VD 00 as Os I> H- Co ^ o to Cn o O 3 1 o 1 o o o O * • • • • • JO rt u> M M o (-• a K i- 1 ^J Cn OS to 5 H- cn VO O o .> K IO CO OS oo «-J H- ^J -E> M cn M 1 O 1 o O o • • • • H O o o Cn 3 as lo o 00 o tO Cn as 05 ca Cn H 1 •^1 IO O Os as 1 1 1 H o o o 3 • • • /— \ o o o 0) «J «~J o •»«• Cn vo *- > O O to s-^ to ■> M 1 O 1 o • • CO o CO Cn CO s 00 Co CO CO M M- VD CO M so CO Cn O O