assessing who gains and who loses from natural resource policy: distributional information and the...

Assessing who gains and who loses from natural resource policy

Recent legislation mandates an increased role for public participation in natural resource policy making. This paper presents an operational way to provide those impacted with relevant information and to enable public officials to assess public reaction to proposed policies. The methodology trans- forms results obtained from a multisector model of income formation and distribution into positive economic measures of distributional impacts. The usefulness of the methodology is illus- trated in the context of a policy to increase coal surface mining on public lands.

Adam Rose is with the Department of Mineral Economics, Pennsylvania State University, University Park, PA, USA; Brandt Stevens is with the California Ener- gy Commission, Sacramento, CA, USA: and Gregg Davis is with the Department of Economics, Marshall University, Hunting- ton, WV, USA.

‘This is not to deny the role of interest groups in natural resource policy before the mid-1970s; see eg P. Culhane, Public Lands Politics, Johns Hopkins University Press, Baltimore, MD, 1981. ‘There are other instances of the importance of non-normative distributional con- cerns in natural resource policy. For example, rent seeking behaviour may be en- couraged when the distribution of benefits

continued on page 283

Distributional information and the public participation process

Adam Rose, Brandt Stevens and Gregg Davis

The framework for public policy evaluation typically applied to natural resources is cost-benefit analysis. This well established approach is based on an aggregate measure of economic efficiency. The decision maker as custodian of resources, and of the public interest, will usually favour a policy if the community or nation as a whole are likely to reap more benefits than costs from its implementation. The associated distribution of the policy impacts is typically of secondary importance at best.

Since the mid-1970s however, an increasing number of natural resource policy decisions have been made in a different context. Rather than decisions being made by a single person or agency insulated from those impacted,’ this new setting mandates direct popular input for public participation in the policy evaluation process. It is accordingly necessary to develop a new analytical framework. The aggregate cost-benefit criterion is not adequate unless we impart an unwarranted sense of altruism to the individual citizen, who is at least as likely to be influenced by how he or she is affected personally. Thus, there is a need for disaggregated, or distributional, impact information and a policy analysis framework in which to place it.

The purpose of this paper is to specify three positive measures of the distribution of economic impacts, demonstrate an operational methodology by which they can be calculated and illustrate their role in the context of natural resource policy. We consider the measures to be in the domain of positive economics in contrast to the normative economics usually associated with distributional impacts. The need to dissemi- nate information on individual policy impacts and the ability to predict public reaction are important irrespective of value judgements pertain- ing to distributional justice.’ Our modelling approach is that of input-

282 0301-4207/89/040282-l 0$03.00 0 1989 Butterworth & Co (Publishers) Ltd

continued from page 282 or costs (or both) is highly unequal (see 0. Lee and D. Orr, ‘Two laws of survival for ascriptive government behavior’, in J. Buchanan, R. Tollison and G. Tullock, eds, Toward a Theory of the Rent-Seeking Eco- nomy, Texas A&M University Press, Col- lege Station, TX, 1980. Critics of US Forest Service policy on timber sales claim that the agency is pro-logging - too much timber is sold at prices below costs incurred by public owners and users of national forests. If the critics are correct, one potential explanation is that the benefits of US Forest Service timber sales are highly concentrated, while the costs are broadly dis- persed among users of national forests and general tax payers who thus have little incentive to pursue a share of the rents.

Still other examples include cases where public participation is instrumental to the success of policy implementation, as in the case of voluntary recycling program- mes (see eg R. Guttentag, T. Pytlar and N. Kutchukian, ‘Using public opinion surveys to design recycling programs’, Resource Recycling, September/October, 1987); and cases where public attitudes hold sway, as in the case of nuclear power (see eg W. Freudenberg and E. Rosa, eds Public Reaction to Nuclear Power: Are There Critical Masses?, Westview Press, Boulder, CO, 1984. ‘See P. Culhane and H.P. Friesema, ‘Pub- lic participation in RPA/NFMA’, in W. Shands, ed, A Citizen’s Guide to the Forest and Range Renewable Resource Planning Act, US Forest Service, Washington, DC, 1981; P. Mohai, ‘Public participation and natural resource decision-making’, Natural Resources Jour- nal, Winter, 1987. ‘See S. Dana and S. Fairfax, Forest and Range Policy: its Development in the US, Johns Hopkins University Press, Balti- more, MD, 1979; US Forest Service, Pub- lic Participation Handbook, Washington, DC, no date. %ee J. Krutilla, A. Fisher and R. Rice, Economical and Fiscal Impact of Coal De- velopment, Johns Hopkins University Press, Baltimore, MD, 1977; A. lsserman and J. Merrifield, ‘Quasi-experimental control group methods for regional analysis: the case of boomtowns’, Economic Geography, January 1987. %p tit, Ref 1.


output analysis extended to incorporate the multisector calculation of income distribution. We construct such an empirical model of the Monongahela National Forest economy. Our case study illustration involves simulating the distributional impacts of overturning the current moratorium on coal surface mining on the public land within the regional economy.

Public participation in natural resource policy decisions

In the USA the most notable examples of formal public participation procedures are those associated with the operation of the Forest Service, as required by the Forest and Rangelands Renewable Re- sources Planning Act of 1974 (RPA) and the Natural Forest Manage- ment Act of 1976 (NFMA), and of the Bureau of Land Management, as required by the Federal Land Policy and Management Act (FLPMA) of 1976. Both agencies have endeavoured to meet legislative requirements in encouraging citizen involvement in their decision making processes and are considered leaders among federal agencies in this area.”

The public hearing is the most prominent formal institution for promoting public participation. Other formal mechanisms for public input include, for example, working groups, invited comment to public documents and ombudsmen. Otherwise, interpersonal contacts serve as an important informal vehicle. In addition to these conventional forms of public involvement, citizen groups have often been successful at providing input to policy decisions through the courts.4

For the public participation process to be successful, however, some important practical conditions must be met. Of greatest importance is the dissemination of sufficient impact information to create an enlight- ened citizenry. In addition, there is a need for a summary measure by which a public agency can predict and evaluate the overall public reaction to its policies if the public participation process were to work effectively. Finally, it would be useful to be able to gauge public reaction in the face of impediments to the orderly functioning of the participation process. The new set of operational concepts that are needed is the major concern of this paper.

Most policy impact models to date have been oriented toward the determination of aggregate measures such as net benefits, employment and fiscal balance.’ But while public participation would best be facilitated by the dissemination of estimates of the impact on individuals, data and modelling limitations make it impossible to provide information below the group level. This information will be useful in the public participation process if the groupings are defined so that an individual can make the necessary association and if the variation of impacts within groups is low relative to that of the population as a whole. Otherwise the distribution of impacts within the group must be known so that an expected value of impacts can be calculated.

Choosing the most appropriate grouping is thus very important. Culhane’s study of Forest Service decision making delineates constituent groups in terms of major interests: economic, environmental, recreational and other (schools, government, professional etc).6 While this categorization is convenient because it pertains to formal visible organizations, it is not clear that it yields the greatest information per group (or per dollar expended on information gathering) for the purpose of the participants or resource managers themselves.

RESOURCES POLICY December 1989


7See eg F. Golladay and R. Haveman, The Economic Impact of Tax-Transfer Policy, Academic Press, New York, 1977; and A. Rose. B. Nakavama and B. Stevens, ‘Mod- ern energy region development and income distribution’, Journal of Environmen- tal Economics and Management, June 1982. %ee eg H. Scarf and J. Shoven, Applied Genera/ Equilibrium Analysis, Cambridge University Press, New York, 1984; M. Hazilla and Ft. Kopp, An introduction to Numerical General. Equilibrium Models Used for Applied Welfare Analysis, Re- sources for’the Future, Washington, DC, 1984. Note that the microsimulation models are not distributional models perse, but they do provide detailed information on how individuals are affected.

The analysis in this paper focuses on income or utility (I/U) levels as the group categories. The strengths of this choice are:

I/U levels are unique for each individual (ie alternative constituent groupings involve overlaps, since an individual may simultaneously be an employee of a recreation company, an environmentalist, and a member of a professional group). I/U levels represent a common denominator for comparing impact types (eg wages, property value increases, disamenities). I/U levels facilitate the calculation of second order impacts (eg the direct impacts on owners of lodging places may be small in comparison to the indirect impacts of increased economic activity). I/U levels facilitate the calculation of normative indices of the distribution of impacts, or equity, for related analyses. I/U levels facilitate the incorporation of important principles of the theory of consumer choice (eg attitudes toward risk and altruistic behaviour).

Another reason for choosing I/U levels is that they are the units of analysis for the majority of a new generation of economic models that focus on distributional impacts. These range from input-output approaches’ to microsimulation models.* An important feature of these new models is that they have been able to overcome the long standing obstacle of lack of data, and are therefore operational.

The case study setting

The Monongahela National Forest (MNF), consisting of all or part of 10 counties in eastern West Virginia, is a major recreational area for the Middle Atlantic States. At the same time it is also a major potential mining and logging site. A transition from the former role to one that places a greater emphasis on extractive industries would involve both gains and losses to the MNF-area economy. In the case of mining, gains would stem from both direct and higher order increases in employment, income and output. Losses would stem from negative externalities, such as aesthetic blight, water pollution, increased flooding and property damage.

Within the MNF’s outer boundaries there are a total of 1 647 000 acres, of which 852 000 acres, or 52%) is National Forest land. Among the resources found there, coal is the most abundant of the fossil fuels, with approximately 150 million tons of coal in the MNF under federal government ownership rights. In 1982 production in the lo-county area totalled 6 156 000 and 12 078 000 tons of surface- and deep-mined coal respectively, or approximately 25% of West Virginia’s surface coal production and 12% of its underground coal production.

The Forest Service owns the coal outright on over 60% of the MNF but on some remaining acreage the Forest Service has more limited control of mining activity, since private interests own the mineral rights. On still other acreage within the lo-county area not classified as National Forest the Forest Service has no control over mining activity, since ownership of surface and mineral rights is private. There is currently a national moratorium on coal leasing of federal lands, as well as a prohibition on strip mining of eastern National Forest lands according to PL 95-87. The strip mining in the MNF economy is currently taking place only where private surface ownership prevails.

284 RESOURCES POLICY December 1989


Opening up the public lands portion to strip mining would therefore necessitate a major policy change.

gFor a detailed description of the model, the reader is referred to A. Rose, B. Stevens and G. Davis, Nafural Resource Policy and home distribution, Johns Hopkins University Press, Baltimore, MD, 1988. “‘See US Forest Service, impact Analysis S&em for Plannina IIMPLAN). Rockv iountain Experimeni Station, it. Collins, CO, 1983. “The estimates were adapted from a study of south-eastern Kentucky by A. Randall ef al, Estimating Environmental Damages from Surface Mining of Coal in Appalkhia: Final Report to Us, EPA, Uni- versitv of Kentuckv. Lexinqter. KY, 1978. Major- adjustment6 ‘were made for differ- ences in population and economic activity between the two regions. The reader is referred to op tit, Ref 9 for a more detailed discussion. ‘*A reduced form of the basic input-output balance equation was used of the form AY = VBAF. where AY is a vector of the change in income payments by income bracket, V is the multisector income distribution matrix, B is the Leontief inverse, and AF is the simulated change in final demand.

Multisector income formation and distribution

A distributional model of the case study region was developed that has two components: an input-output table and a multisector income distribution matrix.’ The I-O table for the MNF economy was adapted from a basic table provided by the US Forest Service as part of its impact analysis for planning (IMPLAN) system.l’ At the two digit industry level, the table contains 59 sectors, which had a combined total gross output of $2.9 billion in 1982. Coal mining is the major sector, comprising about 28% of the region’s intermediate output. In contrast the forestry sector contributes only about 1% to the region’s intermediate production. However, its major forward linkage, the lumber and wood processing sector, is the third largest contributor, with about 7% of the region’s intermediate output.

The multisector income distribution matrix presents the income payment profile of each of the 59 sectors in the MNF disaggregated by 10 income classes. It was constructed by applying standard regionaliza- tion techniques to a larger matrix for West Virginia constructed from a combination of primary and secondary data. The matrix also includes a damage vector to account for the income effects of negative externalities associated with surface mining in the from of:

0 aesthetic damages; 0 water treatment costs incurred by municipalities; 0 recreation based damages; 0 flood damages owing to increased run off; and 0 damage to land and buildings from earth moving.”

It is worth noting that aesthetic damages dominate in the case of low and middle income groups but property income damages are comparable in magnitude to aesthetic damages for the upper income groups.

The simulation conducted in the case study assumes a doubling of final demand for surface mined coal on the public lands portion of the MNF. This policy will stimulate mining activity directly and economic activity elsewhere in the region through multiplier effects. It will also cause a decrease in recreational activity, generate water pollution etc, affecting the entire population of the area.12 Specifically, the $221 million increase in the final demand for surface mined coal leads to a $273 million increase in aggregate regional gross output and a $76.1 million increase in total regional income. This gain in income is significantly greater than the $25.2 million damage total.

The model’s solution can be viewed from another perspective to yield an income distribution impact matrix:

y = Iykjl (1) where y denotes a change in income, k represents income class and j represents the sector from which the change in income emanates. The income distribution impact matrix results of our simulation are presented in Ta’ble 1. The data represent the distributions of increments and decrements to income, which differ from the distribution of baseline income. The table shows that more than 50% of the total gross increase in income accrues from mining. Indirect impacts are felt most strongly in

RESOURCES POLICY December 1989 285


Table 1. Income distribution impact matrix for the Monogahela National Forest (in millions of 1982 dollars).

Sector

Agriculture Mining Construction Food Textiles Apparel Publishing Chemicals Paper Oil and coal products Wood products Primary metals Fabricated metals Machinery Electrical machinery Miscellaneous

manufacturing Other equipment Transport Communications Utilities Trade Bankina and insurance Business services Repair services Personal services Recreation Miscellaneous services Public administration Household activities

Total income increase

Mining damage vector

0.012 0.000 0.052 0.023 0.033 0.423 0.164 0.029 0.009 0.056 0.014 0.177 0.010 0.011

1.955

-3.197

income class I$) Under 5 ado- 5 000 9 999-

0.006 0.093 0.894 4.830 0.008 0.031 0.006 0.034 0.000 0.002 0.010 0.050 0.007 0.027 0.000 0.003 0.000 0.002 0.006 0.027 0.008 0.031 0.001 0.002 0.002 0.008 0.001 0.005 0.005 0.021

0.053 0.002 0.219 0.079 0.222 1.219 0.451 0.105 0.045 0.155 0.043 0.555 0.033 0.157

0.074 0.086 0.088 0.095 0.036 0.016 0.006 0.007 0.473 0.003 0.004 0.004 0.004 0.002 0.002 0.001 0.001 0.023 0.322 0.379 0.391 0.418 0.164 0.089 0.038 0.036 2.108 0.142 0.209 0.213 0.252 0.132 0.073 0.035 0.039 1.197 0.292 0.370 0.352 0.500 0.202 0.188 0.080 0.163 2.402 1.415 1.811 I.619 1.887 0.825 0.472 0.253 0.275 10.189 0.962 1.223 1.144 1.415 0.757 0.705 0.461 0.552 7.834 0.168 0.171 0.174 0.198 0.109 0.081 0.048 0.052 1.135 0.068 0.085 0.087 0.100 0.044 0.022 0.007 0.007 0.474 0.145 0.148 0.127 0.120 0.050 0.027 0.016 0.018 0.862 0.049 0.054 0.048 0.054 0.029 0.021 0.013 0.014 0.339 0.693 0.758 0.697 0.696 0.371 0.259 0.184 0.200 4.590 0.057 0.063 0.073 0.083 0.039 0.027 0.016 0.018 0.419 0.164 0.153 0.053 0.109 0.040 0.094 0.040 0.141 0.940

11.609 13.682 12.124 13.530 5.945 3.728 1.861 3.185 76.123

-4.254 ~3.861 -3.074 -3.047 -1.175 PO.792 -0.328 -0.713 ~25.207

8.504

-4.725

15 OOO- 19 999-

20 008- 24 999-

25 009- 34 999-

35 oocr- 49 999-

10 OOO- 50 008- I4 999- 74 99%

0.083 0.013 6.639 1.543 0.045 0.011 0.044 0.020 0.002 0.000 0.069 0.012 0.036 0.013 0.004 0.001 0.002 0.001 0.037 0.009 0.040 0.008 0.004 0.001 0.011 0.003 0.007 0.003 0.032 0.014

75 008- 99 999-

Over 100 000

0.092 7.684 0.055 0.045 0.003 0.085 0.039 0.003 0.002 0.043 0.046 0.004 0.014 0.010 0.042

0.063 6.609 0.054 0.041 0.003 0.082 0.038 0.003 0.002 0.044 0.045 0.005 0.014 0.009 0.042

0.053 7.121 0.057 0.048 0.003 0.086 0.044 0.005 0.003 0.049 0.047 0.005 0.016 0.012 0.050

0.020 2.961 0.024 0.019 0.001 0.029 0.018 0.001 0.001 0.019 0.017 0.002 0.006 0.005 0.021

0.009 0.008 0.428 0.616 1.606 40.503 0.004 0.005 0.294 0.008 0.004 0.269 0.000 0.000 0.014 0.004 0.005 0.432 0.007 0.009 0.238 0.001 0.001 0.024 0.000 0.000 0.013 0.003 0.007 0.244 0.003 0.003 0.245 0.000 0.000 0.024 0.001 0.001 0.076 0.001 0.001 0.054 0.006 0.012 0.245

sectors that are closely linked to mining, such as trade, banking and insurance. The distribution of the total gross income increments is skewed modestly towards middle and upper income households. The last row in Table 1 includes losses associated with environmental damages from surface mining, which are more evenly distributed.

Positive measures of income distribution impacts

Given the requirements of the decision making context, the typical impact analysis as presented above needs to be extended. We now proceed to the formal specification and illustration of the use of three positive measures of income distribution impacts for this purpose.

The individual impact matrix

As would be the case for any resource development scenario, the simulation results indicate a differential impact across sectors as well as income groups. An individual’s relative gain will depend on his or her association with a given sector as an employee and possibly with several sectors as a dividend, interest, rent or royalty earner. Those associated with the focal point of development will stand to gain the most and gains will decline as linkages with the key sector become weaker. Nearly all individuals will suffer externality losses (eg diminished aesthetics and water pollution) and market transmitted losses will take place in those sectors displaced by the development effort (eg decreased recreational activity when land is shifted to mining).

An individual miner may have a good idea of his new employment prospects but other members of the community may be uncertain about


13Note that for expository purposes we include only the residents of the lo-county MNF economy in the base of our distributional indexes. Given the fact that a Nation- al Forest is in the public domain, the negative impacts on actual and potential visitors could legitimately be taken into account as well. At the same time, those that benefit from increased coal extraction (eg electric utility customers who reap a cost saving from a relatively low priced fuel) could also be counted. Such an analysis specifically, as well as the delineation of the appropriate population, in general, is however beyond the scope of this study.


their future, since they are impacted through indirect economic effects and externalities. The information contained in Equation (1) above can

be rearranged, however, to provide more insight to citizens of the area comprising the MNF. First, the average gain or loss per person in each cell of Table 1 is calculated to reveal the range of individual impacts. The range is then split into appropriate income change categories, r, which become the column dimension of a new matrix, replacing the sector designation, j. This rearrangement of the data in Table 1 aggregates all individuals with the same level of gain or loss, regardless of the sector from which their gains or losses originated. Moreover, the focus is shifted to impacted individuals rather than total income flows.

This is the basis for the individual impact matrix presented in Table 2, which is defined as a probability distribution of potential impacts for each member of a given socioeconomic group. The numbers not in parentheses in the body of the table represent the number of individual households, ~1, in each income class, k, which receive gains or losses in each income change class, r. This matrix is standardized by dividing each cell entry in a given row by its corresponding row sum (see the results in parentheses in Table 2). Our individual impact matrix I, is thus

bkr> I _

bk)

There are a total of 65 837 households in the lo-county MNF region.13 The entries in parentheses in Table 2 show that a household originally in the lowest income group, for example, has a 90% chance of loss between $499 and 0, a 2% chance of a gain between 0 and $499 and an 8% chance of a gain between $500 and $999. The groups with the highest probability of a loss are those in the low and middle income groups since environmental damage tends to be spread more evenly than gains in personal income. For a lower income household this means that if it incurs the same damage as a higher income household, it is more likely to suffer a net loss because its gross personal income gain is likely to be smaller than that of the higher income household.

Table 2. Individual impact matrix.’

income class of recipient ($)

Under 5 000

5 000 to 9 999

10000t0 14999

15000t019999

20000 to 24999

25OOOto 34999

35000to 49999

50000to 74999

75000to 99999

Over100000

Total

Income change class ($) Under -999 to -499 -1000 -500 to 0

11206

(0.9) 9654

(0.66)

6018

(0.51) 4441

(0.45)

3624

(0.5) 3168 (0.48)

213 709

(0.1) (0.32)

(0.0689 16 43 22

(0.06) (0.17) (0.09)

(0.:; (0.0:) 102 514 38902

(0.01) (0.59)

0 to 499

238 (0.02)

3241

(0.22)

2333

(0.2) 1861

(0.19) 1194

(0.16)

1044

(0.16)

196 (0.09)

(0.:;

7 (0.03)

2

(0.01) 10135

(0.15)

500 to 999

960 (0.08)

313 (0.02)

1923

(0.16)

883

(0.09)

292

(0.04) 211

(0.03)

306

(0.14) 142

(0.18)

(0.Z)

(0.0; 5050

(0.08)

1 000 to 2499

177

(0.01)

218 (0.02)

1375

(0.14) 1182

(0.16)

1088

(0.17)

150 (0.07)

(0.045:

(0.:;

(0.:; 4287

(0.07)

2 500 to 4 999

1189 (0.08)

1364

(0.12)

(0.:;

(0.0616, 212

(0.03)

312

(0.14)

139

(0.18)

(0.::

(0.036,

3413 (0.05)

5 000 to 9 999

1149

(0.12)

901

(0.12) 826

(0.13)

(0.:;)

(0.:;

(0.:;

(OAT) 3005

(0.05)

10 000 to Over 24999 25000

276 (0.13)

(O.lY)

(0.0289

(0.024 (0.E)

389 40 (0.01)

Total

12 404

14 574

11 856

9 775

7 259

6 549

2 183

770

257

210

65837

(1.0)

a = less than one-half of 1% (0.005).



14Note that the community impact index result differs slightly from that presented in an analogous simulation in op cif, Ref 9 because of subsequent refinements of the model. 15Note that the distributions of the gross gains and gross losses differ significantly. The distribution of the total direct and indirect impacts for the surface mining sector, as measured by the Gini coefficient, is 0.415 (as compared to its direct Gini of 0.6OOL The damaae vector Gini is 0.154, meaning that damages are spread much more evenly than income gains but in this case that also means that they are highly regressive. The net income change Gini of 0.544 is thus significantly greater than the gross change Gini of 0.415. The overall effect on the Gini coefficient for the entire regional economy is an increase from 0.420 to 0.426, or a 1.4% increase in inequality. Overall then, an increase in surface mining activity will have a slightly adverse effect on the existing skewness of income distribution in the MNF economy.

In effect, the fact that the I matrix gives an individual in a given group the range of individual impacts or the entire distribution should prove more valuable than the expected value to those individuals who have more complex attitudes toward risk (eg the Savage minimax regret criterion). These attitudes can be more formally incorporated into the anslysis with the third measure presented below.

The community impact index

The individual impact matrix contains a great deal of information and serves a useful purpose from the standpoint of the parties at interest in Forest Service policies. However, forest managers need a way of shifting through the matrix and arriving at a summary judgement. There are a number of alternatives. For one, there is the summation of individual net gains in dollar terms, though this is no more than a benefit-cost criterion. The summation of individual net utility changes is merely a weighted form of this criterion. Both statistics correspond to the dollar voting of well functioning markets.

An alternative is the one person one vote condition of most formal elections and many types of representation. Table 2 contains the elements of an individual voter approach to a summary statistic since each cell entry represents the number of households in a given income group and a net gain/loss category. We refer to our summary measure as the community impact index, which is defined as a tally of how the majority of the relevant poulation is qualitatively affected by a policy and of the size of the majority.

This measure is simply the aggregation of the I matrix into a scalar that represents the net number of gainers or losers. Let us define the number of people with net gains, G, as

G = XX+, kg (3)

where g denotes the subset of r corresponding to positive income change categories. The community impact index, C, can then be expressed succinctly as:

C= 5 05CIl II

(4)

In the simulation there are 26 319 households that stand to gain from the increased mining activity and 39 518 households that are net losers, so C = 0.400.i4 The result may thus cause a majority of residents of the MNF to oppose the mining policy despite the fact that the gross gains are three times the size of gross losses.

The community impact index of 0.40015 is somewhat surprising in light of the relative size of gains and losses. This means that a slight

majority of the households in the MNF economy suffer net losses as a result of the proposed policy. The dissection of the index in Table 2 reveals the reason for the apparent paradox of strongly positive aggregate net benefits to the region coupled with a majority of residents incurring net losses. Recall that environmental damages were rather evenly distributed, meaning that they represent a higher proportion of the income of lower income residents than of upper income residents. On the other hand, gains are skewed more favourably toward upper income residents. Thus, net losses of a rather small magnitude are prevalent among the many lower income and, to a lesser extent, middle


“See C. Russell, ‘Applications of the public choice theory: an introduction’, in C. Russell, ed, Collective Decision-Making, Johns Hopkins University Press, Balti- more, MD, 1979. 17Note that the results may be sensitive to two other sets of factors. The first is environmental damage. To examine the im- plications for our results of a misspecifica- tion of environmental losses, we simulated the effects of both a 50% decrease and a 50% increase in our initial estimates. For the lower bound case, the decrease in the size of damages relative to the size of the income gains (which remain unchanged) causes the distribution of net impacts to become less skewed than before., raising the Gini coefficient in the MNF from 0.420 to 0.423. On the other hand, there is a significant change in the community impact index, which is 0.547, meaning that a majority of the households in the MNF would receive net gains. The upper bound case results in a higher Gini coefficient (0.430) and a smaller community impact index value (0.355) than the baseline. The reason the two indexes move in the oppo- site direction is again that environmental damages are estimated to fall disprop- ortionately on lower income groups.

Second, the results may be sensitive to institutional considerations. The straightforward application of an extended I-O model to the problem at hand implicitly assumes that all new property related income is retained and respent within the MNF economy and that all new wage and salary income is distributed incrementally across the workforce ie additional man- power is obtained via overtime pay. (The increased coal surface mining yields 1 889 direct man year employment equivalents and 2 815 indirect and induced man year employment equivalents. There is sufficient slack to absorb this additional work time into either new jobs because of the 11% unemployment rate or overtime work because the average number of days actually worked by coal miners in the region ranged from 5560% of available work days.) These inherent assumptions can be related directly to features such as absentee property ownership and union policies. We tested the sensitivity of the results to them by varying the parameters within a more sophisticated extended I-O model (see op tit, Ref 9). The results for the ‘most likely’ case of a 50% capital income retention and 50-50 split of the wage bill between overtime work and new hires was a Gini coefficient of 0.384 and community impact index of 0.348. This means that a relaxation of these basic assumptions has only a marginal effect on our results.

Other case studies may find their results more sensitive to institutional factors than environmental ones. The main point is that the extended I-O model and positive income distribution measures are sufficiently general to incorporate many important real

continued on page 290


income group members, while large net gains are experienced by the vast majority of the much worse represented upper income groups.

Note that our example is likely to be representative of a large number of public policy issues related to natural resources (eg hazardous material disposal, offshore drilling, nuclear waste repositories). A positive level of net benefits may not be sufficient to receive public support in the presence of potentially ubiquitous environmental externalities.

To the extent that C measures the preferences of affected residents, the index will also be quite useful to policy makers where the size of the winning margin is an important guide to policy.” Note also that if the well being of one subgroup of the population is considered to be of relatively greater importance than that of the others, weights can be attached to the n values of the former. This differs from a dollar weighting of the benefit-cost criterion (eg higher weighting of low income individuals) and represents another way of incorporating equity into the ana1ysis.r’

The political articulation index

The previous index would correspond to an ideal of majority rule, under perfect information and standard assumptions about utility maximiza- tion. Though forums for public opinion and participation may be the main source of feedback to policy makers, input from some of the affected parties cannot be counted on where costs or perceived obstacles to participation outweigh the expected value of the benefits from influencing a policy outcome. The distribution of net benefits for

participants may differ from those of the relevant population, and thus the preferences expressed at a public hearing, for example, will not be representative of the community. Thus, incorporating the rate of participation into the previous index may provide policy makers with a valuable means of interpreting public feedback.

The political articulation index is defined as an indicator of the likely public response to a policy, which takes into account such factors as people’s intensity of preference, political influence, attitudes toward risk and transactions costs. Each of these factors can serve to weight individual impacts, with the overall weight in each case indicating an individual’s or group’s probability of active political participation.

For example, a person with $50 000 to gain from the new mining policy in the MNF would be more likely to attend and speak out at a public hearing than one who stands to gain only $5. Furthermore, if we assume asymmetry of the utility function stemming from risk aversion, a $500 gain would not give rise to the same fervour as a $500 loss. We would likewise expect a higher probability of participation for a $500 gain on the part of a poor person than for a wealthy person. Lower income individuals may feel they have less political influence than those in other groups and therefore that their vote carries less weight. They may also be stifled by transaction costs (travel expenses, hiring a babysitter etc). On the other hand, the relatively high probability of attendance by some individuals who stand to gain or to bear losses could be more than just the wielding of individual influence. For example, a very high proportion of the people who stand to gain from increased mining might be concentrated in a single, socioeconomically uniform neighbourhood or small community. This could result in a coalition that is buoyed by its strength in numbers and may translate into a high



continued from page 289 world features that will vary from setting to setting. “Note that public participation mandates do not actually require formal votes and do not require public officials be bound by such votes should they take place. The presentation in this paper should not be interpreted as an endorsement of such requirements nor imply that the policy maker’s decision be a binary function of gainers and losers alone. While we have not emphasized efficiency, equity and professional standards of natural resource management, we realize that these are likely to be arguments in the decision function along with public participation considerations.


probability of success in influencing the outcome, thereby overriding transaction cost obstacles for group members. All of these features can be captured in a weighting matrix, W.

Other kinds of W matrices may be useful, since the row categories may change depending on the group characteristics of interest to policy makers (eg age and ethnic or interest group). For any given group characteristic, a separate W matrix would be desirable for each argu- ment in the utility function of the decision maker (individual net gain, aggregate net gain and equity etc). Even with the valuable insight into several of the factors that might comprise W by Culhane and others, we acknowledge the difficulty in its empirical estimation with the current data available from public hearings. First, analysis of the distribution of policy impacts has been skimpy at best. Second, very little data exist on income, age, perceived size of gain or loss from a given policy and access to policy maker, ie factors that would affect preferences for the policy and participation. Subjective judgements by the policy maker may have to play a major role in the meantime.

The political articulation index, A, can be formally expressed as:

This index modifies the passive community impact measure to yield an active intensity of preference measure.

Interestingly, an unfavourable majority score on the community impact index, (4), can turn into a favourable majority on the political articulation index, (5), or vice versa. Such a reversal could possibly take place in our mining development example if those who stood to gain formed a strong coalition that significantly accentuated participation.

Before the development of recent distributional impact models, the Forest Service and other policy makers did not have the means to determine specifically how the majority of the parties at interest were affected and of necessity placed significant emphasis on preferences expressed at public hearings or in written testimony. Now policy makers have a choice. If forest managers are relatively sensitive to the actual welfare of individuals in a forest community, they will tend toward incorporating Equation (4) into their decision making function. To the extent that they are influenced by public action, they will gauge the outcome of a hearing or predict this outcome with Equation (5). Which measure is most appropriate is largely a normative question and need not be resolved here.”

Conclusion

Though economists are generally sceptical of public ownership or publicly administered resource allocation as substitutes for the market- place, these institutions are commonplace in most Western nations, especially in the case of natural resources such as forests, water, mineral rights and wilderness areas. A major unanswered criticism of these non-market allocation processes is whether they can be democratic or even responsive to consumer preferences in any significant way. Can unelected public officials make allocations on behalf of a constituency that the constituency itself would have chosen? This is an important concern for any nation founded on democratic ideals. Any answer in the affirmative must overcome several major obstacles. Government agen-


ties must be able to provide the data needed by citizens to render an informed judgement, the costs of obtaining accurate public feedback must not be prohibitive and the institutions for public input must not be

significantly biased. It is our contention that many of these problems can be overcome in a

participatory decision making setting with the aid of data on the distribution of economic impacts. In the past such impacts have often been associated with normative judgements, while more commonly used efficiency based assessments, such as cost-benefit analysis, are viewed as objective and straightforward. Yet, as our case study illus- trated, a positive net benefit result may clash with citizen preferences. Since a majority of the affected population incurred net losses, their support of the policy could be motivated only by a level of altruism that cannot be justified theoretically or empirically. In the context of public participation, overriding public sentiment places an aggregate efficiency objective in a clearly paramount position, which is a normative judgement in its own right. This is not to say that the popularity of a policy should be the only criterion, or that the trade off between efficiency and participatory democracy cannot be bridged by some redistribution. It is to say that the issues raised in this paper are complex and can never be resolved if not properly addressed.

We also contend that our analysis is applicable to an ever growing number of settings involving natural resources. There are an increasing number of opportunities for public participation. In addition, there appear to be more and more cases in which the concentrated gains from policy decisions are juxtaposed to environmental damage, which, although often only small in magnitude, is likely to be imposed on a large portion of the population.


assessing who gains and who loses from natural resource policy: distributional information and the...

Documents