. 9
( 11)


275.00 299.99 527 2,359 2,978 2,630 2,636 29 20 10
300.00 324.99 847 3,296 2,976 2,475 808 10 4 9
325.00 349.99 1,089 3,113 2,589 1,898 184 7 14 22
350.00 374.99 1,578 2,616 1,676 796 41 19 30 48
375.00 399.99 2,618 1,734 639 252 15 38 54 86
400.00 424.99 3,224 784 240 183 83 95 78
425.00 449.99 3,025 321 149 55 100 79 37
450.00 474.99 2,389 118 28 21 33 9 17
475.00 499.99 1,380 30 28 27 15 24 26
500.00 524.99 559 27 20 20 25 22 16
525.00 549.99 140 20 15 11 14 22 16
550.00 574.99 75 9 13 12 25 19 21
575.00 599.99 29 11 8 15 17 8 4
600.00 624.99 4 12 18 12 2 4 33
625.00 649.99 16 9 9 35 56 26
650.00 674.99 15 21 14 26 10 19
675.00 699.99 10 23 29 31
700.00 724.99 3 3 27 12
725.00 749.99 3 4
750.00 774.99 1 9 37
775.00 799.99 34 34 6
800.00 824.99 11 4 14
825.00 849.99 14 24 13
850.00 874.99 19 23 25
875.00 899.99 20 17 12
900.00 924.99 12

Notes: 1 Negative sums indicate areas where woodland values exceed agricultural values.
Blank cells indicate that no 1 km cells fall into this category. There are 20,563 1 km cells.
262 Applied Environmental Economics

(1) of Table 9.2), while Plate 3d indicates the social value of conversion when the
broad de¬nition of woodland (timber, carbon storage and recreation, with the latter
again measured using our CV cross-study value) is used (column (7)).
The overall pattern of values shown in Table 9.2 is similar to that for sheep
farms, with expansion of the de¬nition of woodland bene¬ts increasing the value
of the latter. However, the pattern of farm-gate values illustrated in Plate 3c (where
woodland bene¬ts are de¬ned as arising just from timber and subsidies) is different
to its sheep farm equivalent (Plate 3a). Here we ¬nd that the optimal locations for
conversion to woodland (shown as negative sums) are clustered in upland rather
than lowland areas. This difference in itself is of interest and shows that in contrast
to the sheep sector, where it was the superiority of woodland in the lowland areas
which was the driving force behind the net bene¬t of conversions, here it is the
fall-off in milk farm values as we approach the most upland areas which allows
woodland to become viable “ but only at the extremes of topography. This difference
is repeated in our social value analysis of the wider de¬nition of woodland values
(Plate 3d) where, with the exception of peat soil regions, it is again the upland areas
which show more promise of net conversion bene¬ts (in contrast to the sheep farm
equivalent illustrated in Plate 3b).
The sheep and milk farm analyses differ not only in their relative pattern but also
in the absolute level of conversion values. Even when all possible woodland values
are considered, milk values almost always substantially exceed those generated by
woodlands. Given that we know there are very few milk farms in the extreme upland
areas of Wales this differential is probably even stronger than Table 9.2 indicates.
Furthermore, as the discount rate used here is not out of line with (and may even
be below) that likely to be used by milk farmers in everyday decision-making, any
increase in the discount rate due to risk aversion would only reinforce the result.
The social value assessment given here uses the government discount rate and so
results are valid as they stand.
In summary, conversions out of milk production and into woodland are generally
not justi¬ed by this study. We now extend our analysis to consider changes in the
species of tree used in conversions.

Conversion from agriculture to broadleaf woodland
Sheep farms
Table 9.3 presents results for conversions from sheep farming to broadleaf wood-
land, maintaining the discount rate at 6 per cent. It is useful to contrast these results
with the sheep to conifer conversion summarised in Table 9.1. In the latter, if we
consider only timber values, conversion generally (but not always) fails to generate
net bene¬ts when viewed from the farm gate, but almost always creates social gains.
However, the case for conversion is less clear in Table 9.3 where the slow growth
Cost-bene¬t analysis using GIS 263

rates associated with broadleaves mean that delayed timber bene¬ts are heavily dis-
counted; indeed, it is discounting which principally drives the differences between
Tables 9.1 and 9.3. Accepting such a discount rate means that in less than half of
the cells is conversion from sheep farming to broadleaved woodland justi¬ed upon
social grounds and in no case do farm-gate values support conversion.
Broadening the de¬nition of woodland bene¬ts to include carbon sequestration
does improve the farm-gate case for conversion, although in almost all cases the
value of sheep farming marginally outperforms that of woodland. However, social
values now generally support conversion except on areas of peat soil.
We now turn to consider recreation values. With respect to conversions to
conifers we have up to this point focused attention upon the lower-bound CV
measures. However, while evidence of a link between tree species and recreation
values is somewhat anecdotal (see Hanley and Ruffell, 1991, 1992), we feel that
the use of upper-bound measures has at least some justi¬cation with respect to
broadleaf woodlands. The use of such measures does signi¬cantly improve the
apparent viability of land use conversions, with virtually all cells producing net
social bene¬ts and most generating farm-gate gains from conversion. However,
because we expect strongly declining marginal recreation values for additional
woodlands in any given area (i.e. once a given locality has a recreational wood-
land then the marginal value of an additional woodland is relatively low) then we
cannot take the values given in columns (4) and (8) of Table 9.3 at face value.
This being so, it is of more interest to use this analysis to identify optimal lo-
cations for conversion rather than to look at total values. Plate 3e illustrates the
farm-gate value of conversion using our wider de¬nition of woodland bene¬ts (and
the upper-bound ITCM value of recreation), i.e. the net bene¬t map underpinning
column (4), while Plate 3f illustrates the social value equivalent of this analysis, i.e.
column (8).
It is clear from both Plate 3d and Plate 3f that, when our wider woodland bene¬ts
de¬nition is applied, the net bene¬ts of conversion from sheep rearing are highest
in areas of high population accessibility (enhancing recreation values) and decrease
as we move to more remote locations. The only areas where conversion is never
justi¬ed are on peat soils where large-scale carbon liberation occurs. This echoes,
in particular, the results shown in Plate 3b.
Analysis of the social values illustrated in Plate 3f indicates that the South Wales
valleys are an area of particular interest. In the highly populated valleys and around
the cities of Cardiff and Swansea there is a clear and very substantial net social
bene¬t from conversion out of sheep farming and into multipurpose broadleaf
woodland. This falls rapidly as we move away from such areas and into the sparsely
populated upland areas which run down through the centre of Wales or the more
inaccessible Pembroke and Lleyn peninsulas which characterise the west coast of
Table 9.3. Distribution of the net bene¬ts of retaining sheep farming in Wales as opposed to conversion to broadleaf
(beech) woodland:1 6% discount rate

Farm-gate values Social values
Lower limit Upper limit timber timber+ timber+carbon+ timber+carbon+ timber timber+ timber+carbon+ timber+carbon+
(£/ha/yr, (£/ha/yr, only carbon recreation (CVM) recreation (ITCM) only carbon recreation (CVM) recreation (ITCM)
1990) 1990) (1) (2) (3) (4) (5) (6) (7) (8)
’350.00 ’325.01
’325.00 ’300.01
’300.00 ’275.01
’275.00 ’250.01
25 312
’250.00 ’225.01
114 14 431
’225.00 ’200.01
126 177 923
’200.00 ’175.01
193 434 1,159
’175.00 ’150.01
294 1,259 3,345
’150.00 ’125.01
25 465 236 5,089 7,128
’125.00 ’100.01
223 993 5,775 8,588 5,160
’100.00 ’75.01
469 1,411 10,289 3,891 925
’75.00 ’50.01
1,517 3,916 3,166 3,074 401 190
’50.00 ’25.01
427 5,676 7,000 6,669 464 140 94
’25.00 ’0.01
0.00 24.99 6,345 8,991 4,538 8,822 232 81 3
25.00 49.99 1 10,816 2,500 608 1,392 4
50.00 74.99 3,400 1,703 294 172 317
75.00 99.99 8,894 295 211 125 197
100.00 124.99 6,810 269 74 17
125.00 149.99 872 101 91 77
150.00 174.99 173 118 3
175.00 199.99 214
200.00 224.99 159
225.00 249.99 40
250.00 274.99
275.00 299.99
300.00 324.99 34
325.00 349.99 165 380
350.00 374.99 418 305 65
375.00 399.99 57 19 10
400.00 424.99 51 14
425.00 449.99 3 174 343
450.00 474.99 393 282 88
475.00 499.99 86 33 7
500.00 524.99 7

Notes: 1 Negative sums indicate areas where woodland values exceed agricultural values.
Blank cells indicate that no 1 km cells fall into this category. There are 20,563 1 km cells.
266 Applied Environmental Economics

the country. The map also amply illustrates the ready interpretability of results
generated by the methodology developed in this research.

Milk farms
We now brie¬‚y consider the viability of transfers from milk farming to broadleaf
woodland. Results for this analysis are presented in Table 9.4. As before, while
the pattern of results obtained for milk farms is similar to that for sheep farms,
the absolute values are very different, with insigni¬cant levels of conversion being
justi¬ed under either the farm-gate or social value analysis. Given this, we do not
discuss these ¬ndings further here.

Conversions between milk and sheep farming
In the above analyses we have calculated both farm-gate and social net bene¬t
sums for conversion from sheep farming to woodland and from milk farming to
woodland. However, these results also allow us to consider the net bene¬ts of
potential conversions between the two farming types (assuming a hypothetical
lifting of the market entry restrictions currently imposed by milk quotas) and to
ask whether this is more likely than a move into woodland. For simplicity in the
following discussion we will refer to the net bene¬ts of conversion to conifer
woodland although the analysis could also be repeated for broadleaves, producing
roughly similar results.
Considering those farm-gate values which farmers might actually have received
during our study period (i.e. ignoring non-timber woodland bene¬ts), then we have
shown that in lowland areas sheep farming generally, but only marginally, outper-
forms woodland, with some conversions being viable where poorer soils predom-
inate (for example, the north-west of Wales as illustrated in Plate 3a). However,
the farm-gate value of dairying (Plate 3c) always and very substantially outper-
forms that of woodland in such lowland areas and consequently exceeds the value
of sheep farming by a similar extent. Moving to consider upland areas, the farm-
gate value of sheep farming always exceeds that of woodland, this excess being
in places over £100/ha. The picture for milk to woodland conversions in upland
areas is more mixed. While in areas of less extreme elevation milk values still ex-
ceed those of woodland by over £200/ha, in the highest areas the situation changes
rapidly as dairy values fall rapidly, and the net bene¬ts of retaining milk production
drop below £100/ha. Thus, in the most mountainous areas, conversion to woodland
becomes pro¬table. Therefore, we can see that our model predicts that the farm-
gate value of sheep farming exceeds that of both woodland and milk production
in these upland areas. Such a prediction is borne out by actual farming practice in
these regions (see Chapter 8).
Cost-bene¬t analysis using GIS 267

Turning to consider social values, it is perhaps most valid to de¬ne woodland
value using the full range of bene¬ts considered in this study. Using this measure
we can see that woodland substantially outperforms sheep farming (Plate 3b) but is
itself consistently outperformed by dairying (Plate 3d) in lowland areas. Therefore,
in a scenario of full agricultural liberalisation and with farmers being paid for
positive externalities we would expect no conversions from dairying but complete
conversion from sheep farming, primarily into milk (if all policy restrictions had
been lifted) with woodland as a possible second choice.10 However, such a result
ignores the impacts upon milk price of such a supply expansion and given the very
strong likelihood of entry restrictions remaining upon the milk market we believe
that this does not invalidate analysis of the social bene¬ts of potential conversions
from sheep farming to woodland in lowland areas.
In the uplands the social value of woodland exceeds that of sheep farming in all
but peat soil regions, with net bene¬ts of conversion generally in the range of £100
to £200 per hectare. For dairy farming the picture is again less clear with about the
same area converting as not. In the former, the net bene¬ts of conversion generally
range up to about £100/ha with only a few areas exceeding this. Consequently,
assuming no entry barriers or requirement for risk premiums, we would expect all
sheep farms to convert, with approximately the same number turning to woodland
as to milk farming. Given the improbable nature of such assumptions we do not
foresee movement from sheep to milk production, so this implies that all conversion
would be towards woodland. The one exception throughout is the peat soil regions,
where afforestation is never justi¬ed on social grounds.

Results for the 6 per cent discount rate: summary
Looking back across the full range of analyses conducted using the 6 per cent
discount rate we can see that an economic case can be made for conversion from
sheep farming, particularly in lowland areas with high population accessibility, but
that under the subsidy schemes available in our study period such conversion was
not ¬nancially attractive to the farmer. (The intervening years have changed little
here, with subsidies still not being available for most of the non-market bene¬ts of
woodland.) Considering the choice of species, conifer woodlands generally seem to
be a more viable option for conversion than broadleaves. However, in the following
chapter we discuss omissions from this analysis (e.g. acidi¬cation impacts and
biodiversity effects) which are generally favourable to broadleaf trees and militate
against certain coniferous species. Given this, it is interesting to note that our
analysis of broadleaf values indicates that, using the wider de¬nition of bene¬ts,
conversions from sheep farming usually generate net bene¬ts.
10 As noted in Chapters 1 and 5, this is only a partial CBA; we were not able to consider all possible opportunity
costs “ a characteristic failing of many practical CBA applications.
Table 9.4. Distribution of the net bene¬ts of retaining milk farming in Wales as opposed to conversion to broadleaf
(beech) woodland:1 6% discount rate

Farm-gate values Social values
Lower limit Upper limit timber timber+ timber+carbon+ timber+carbon+ timber timber+ timber+carbon+ timber+carbon+
(£/ha/yr, (£/ha/yr, only carbon recreation (CVM) recreation (ITCM) only carbon recreation (CVM) recreation (ITCM)
1990) 1990) (1) (2) (3) (4) (5) (6) (7) (8)
’225.00 ’200.01
3 17 20
’200.00 ’175.01
17 15 26
’175.00 ’150.01
14 35 100 25
’150.00 ’125.01
11 49 100 118 21 74
’125.00 ’100.01
20 81 156 188 31 44 148
’100.00 ’75.01
19 187 146 186 61 181 203
’75.00 ’50.01
79 136 270 295 11 192 288 371
’50.00 ’25.01
174 283 335 326 27 309 440 520
’25.00 ’0.01
0.00 24.99 158 455 253 179 118 608 415 358
25.00 49.99 293 146 175 117 304 268 224 419
50.00 74.99 489 151 78 129 518 191 254 452
75.00 99.99 346 103 160 149 648 247 255 649
100.00 124.99 154 130 148 178 216 233 418 769
125.00 149.99 100 153 133 203 233 250 628 923
150.00 174.99 148 142 138 249 255 376 889 1,675
175.00 199.99 140 125 156 364 230 838 1,987 2,995
200.00 224.99 150 140 289 404 340 1,638 3,120 3,558
225.00 249.99 157 203 296 458 465 2,934 4,164 3,341
250.00 274.99 128 237 385 601 1,283 4,699 3,956 2,465
275.00 299.99 183 329 702 1,274 1,910 4,236 2,349 959
300.00 324.99 255 758 1,079 1,402 3,843 2,386 342 152
325.00 349.99 239 964 1,350 2,026 4,449 473 96 17
350.00 374.99 446 1,172 2,142 2,353 3,908 103 3 1
375.00 399.99 931 1,947 2,492 2,472 1,471 1 4 25
400.00 424.99 1,011 2,671 2,800 2,238 297 18 33 49
425.00 449.99 1,483 2,903 2,583 2,250 37 32 96 119
450.00 474.99 2,286 2,708 2,157 1,234 131 128 83
475.00 499.99 2,740 2,213 1,257 583 95 18 6
500.00 524.99 2,564 1,231 344 121 7 21 18
525.00 549.99 2,568 424 144 104 17 3 13
550.00 574.99 1,807 216 35 9 11 22 34
575.00 599.99 1,003 80 4 7 25 26 7
600.00 624.99 328 8 11 13 38 24 21
625.00 649.99 116 9 14 24 1
650.00 674.99 36 21 20 9 2 4 31
675.00 699.99 1 16 8 24 24 41 19
700.00 724.99 24 22 24 5 8
725.00 749.99 25 38 56
750.00 774.99 38 26
775.00 799.99 2 2 4 1
800.00 824.99 2 24 28
825.00 849.99 25 18 18
850.00 874.99 21 6
875.00 899.99 10
900.00 924.99 25 25 28
925.00 949.99 13 25 26
950.00 974.99 25 14
975.00 999.99 1

Notes: 1 Negative sums indicate areas where woodland values exceed agricultural values.
Blank cells indicate that no 1 km cells fall into this category. There are 20, 563 1 km cells.
Table 9.5. Distribution of the net bene¬ts of retaining sheep farming in Wales as opposed to conversion to conifer
(Sitka spruce) woodland:1 3% discount rate

Farm-gate values Social values
Lower limit Upper limit timber timber+ timber+carbon+ timber+carbon+ timber timber+ timber+carbon+ timber+carbon+
(£/ha/yr, (£/ha/yr, only carbon recreation (CVM) recreation (ITCM) only carbon recreation (CVM) recreation (ITCM)
1990) 1990) (1) (2) (3) (4) (5) (6) (7) (8)
’575.00 ’550.01
’550.00 ’525.01
’525.00 ’500.01
9 199
’500.00 ’475.01
37 217
’475.00 ’450.01
125 61 378
’450.00 ’425.01
116 155 803
’425.00 ’400.01
169 321 1,170
’400.00 ’375.01
234 992 1,823
’375.00 ’350.01
93 478 37 2,725 4,216
’350.00 ’325.01
200 912 2,963 5,814 5,056
’325.00 ’300.01
359 1,233 6,962 4,959 2,954
’300.00 ’275.01
1,263 2,170 3 5,092 2,653 1,612
’275.00 ’250.01
246 3,435 4,326 839 2,865 1,475 934
’250.00 ’225.01
3,998 5,464 4,565 7,486 1,518 601 288
’225.00 ’200.01
6,549 5,304 2,721 6,505 412 217 126
’200.00 ’175.01
18 4,452 2,455 1,532 3,570 156 58 37
’175.00 ’150.01
2,024 2,568 1,487 907 1,689 36 21 21
’150.00 ’125.01
7,549 1,499 676 361 352 17 15 8
’125.00 ’100.01
5,610 554 238 113 82 13 6 2
’100.00 ’75.01
3,032 141 60 36 20 2 1
’75.00 ’50.01
1,671 34 18 20 14 1 16
’50.00 ’25.01
526 16 15 8 3 14 18 10
’25.00 ’0.01
0.00 24.99 98 14 6 1 7 26 70
25.00 49.99 17 2 1 4 51 109 143
50.00 74.99 15 1 10 13 169 199 194
75.00 99.99 3 16 8 10 207 122 51
100.00 124.99 5 29 74 36 13 5
125.00 149.99 53 106 146 5 2
150.00 174.99 167 196 181
175.00 199.99 194 124 56
200.00 224.99 48 13 6
225.00 249.99 6 3
250.00 274.99
275.00 299.99
300.00 324.99

Notes: 1 Negative sums indicate areas where woodland values exceed agricultural values.
Blank cells indicate that no 1 km cells fall into this category. There are 20,563 1 km cells.
272 Applied Environmental Economics

Our analysis of milk farms suggests that, in general, there is not a strong economic
case for conversions from this sector to either conifer or broadleaf woodland. One
further interesting difference here is the result that if any such conversions were jus-
ti¬ed these would be in upland (but non-peat) areas. This seems mainly attributable
to a rapid fall-off in milk values as we reach the upland extremes of the Welsh
environment. However, we again remind ourselves of the relative scarcity of milk
farms in such environments.
Having analysed the effects of changing tree species we now consider the effect of
changing discount rates. Given our discussions in Chapter 5 and above, any increase
in rates seems unrealistic (and will almost inevitably rule out any possibility of
conversion), so a reduction seems more interesting.

Results for the 3 per cent discount rate
A 3 per cent discount rate is worth considering for two reasons: ¬rst, it more closely
approximates what we believe to be the rate used by sheep farmers for everyday
decision-making; secondly, it is closer to the social discount rates recently proposed
by many commentators and currently being considered by H. M. Treasury. The 3
per cent rate thus has applicability to the sheep farm-gate results and to both the
sheep and milk farm social value analyses.

Conversion from agriculture to conifer woodland
Sheep farms
Table 9.5 reports results from the analysis of conversions from sheep to conifer using
a 3 per cent discount rate. Considering column (1) we can see that lowering the
discount rate to 3 per cent makes conversion from sheep into woodland bene¬cial
for almost all farmers even when we only consider timber values and the availability
of grants and subsidies during our study period; Plate 3g presents the corresponding
map of values. Given that this scenario represents the available returns to farmers,
why does such a rate of conversion not occur? The answer, as before, is most
likely to be related to a risk premium. A farmer™s risk premium can either be
modelled as a higher required discount rate (as discussed earlier in this chapter)
or, at existing discount rates, as a requirement that unfamiliar goods, such as those
provided by forests, provide a substantially higher income than does conventional
production (our discussion of Table 9.2 is relevant here). As before, net savings
on subsidies mean that social values of conversion under this scenario (column
(5)) are substantially above farm-gate values; indeed, using this analysis, all Welsh
sheep farms should be converted to woodland. Given that we are here ignoring all
non-timber bene¬ts, this is a powerful result.
Cost-bene¬t analysis using GIS 273

For both farm-gate and social value analyses the addition of carbon sequestration
values again produces a bimodal distribution, with the majority of cells now more
strongly bene¬ting from conversion to woodland (columns (2) and (5)). The further
addition of recreation values reinforces this result.

Milk farms
Table 9.6 repeats the above analysis but now considers conversion from milk farms.
Given the discussion presented in Chapter 5, the 3 per cent discount rate is not espe-
cially relevant to farm-gate analyses of milk farm conversions. However, that same
chapter shows that such a rate is, arguably, relevant to social values (although it is
currently being considered by the UK government for such purposes). Examining
the social values block we can see that it is only when carbon sequestration values
are included that signi¬cant conversions are justi¬ed. Here about 18 per cent of
cells generate net social bene¬ts from conversion, a proportion which rises sub-
stantially when lower-bound (most appropriate for conifers) CV-based recreation
values are added, although substitution effects mean that this has to be a signi¬cant
overstatement of conversion viability. Examination of the maps underpinning these
results con¬rmed that it is high population, high accessibility, lowland areas which
generate the largest net bene¬ts from conversion.

Conversion from agriculture to broadleaf woodland
Sheep farms
As before we now hold the discount rate constant (at 3 per cent) and consider the
impact of conversions to our representative broadleaf tree species, beech. Table 9.7
shows results for sheep farms. Considering ¬rst the farm-gate values, the contrast
between our 3 per cent discount rate analyses of conversions from sheep to conifers
as opposed to broadleaves is very marked. Whereas present timber values and
related grants were suf¬cient to generate net farm-gate bene¬ts from conversion in
the former instance (Table 9.5, column (1)), for the latter such conversion fails to
pass the cost-bene¬t test (Table 9.7, column (1)). Given that grants for broadleaf
trees exceed those for conifers, this result seems to be due to the longer rotations,
and hence delay to felling bene¬ts, typical of broadleaves.
Addition of carbon sequestration bene¬ts makes conversion of just over 10 per
cent of cells apparently pro¬table from a farm-gate perspective (column (2) of
Table 9.7). However, the likelihood of farmers requiring a risk premium means
that in reality we would not expect conversions to occur until recreation bene¬ts
are also paid. Even if, as argued previously, higher rate recreation values can be
justi¬ed for broadleaf woodlands, then such a premium means that relatively high
increases in subsidies would be required to generate attractive levels of farm-gate
Table 9.6. Distribution of the net bene¬ts of retaining milk farming in Wales as opposed to conversion to conifer
(Sitka spruce) woodland:1 3% discount rate

Farm-gate values Social values
Lower limit Upper limit timber timber+ timber+carbon+ timber+carbon+ timber timber+ timber+carbon+ timber+carbon+
(£/ha/yr, (£/ha/yr, only carbon recreation (CVM) recreation (ITCM) only carbon recreation (CVM) recreation (ITCM)
1990) 1990) (1) (2) (3) (4) (5) (6) (7) (8)
’375.00 ’350.01
2 10 38
’350.00 ’325.01
14 37 52 5
’325.00 ’300.01
36 50 68 2 12 33
’300.00 ’275.01
2 55 68 95 15 23 117
’275.00 ’250.01
17 70 91 144 27 83 118
’250.00 ’225.01
34 96 176 180 83 134 236
’225.00 ’200.01
63 193 192 178 10 142 218 452
’200.00 ’175.01
78 185 188 197 36 243 285 464
’175.00 ’150.01
132 204 206 230 90 285 286 722
’150.00 ’125.01
209 226 258 296 165 325 475 890
’125.00 ’100.01
234 278 252 253 311 395 614 1,085
’100.00 ’75.01
303 219 182 255 357 398 1,003 1,393
’75.00 ’50.01
324 180 179 274 451 513 1,592 2,423
’50.00 ’25.01
309 164 175 427 450 1,330 2,975 3,503
’25.00 ’0.01
0.00 24.99 191 154 260 527 358 2,934 4,134 3,924
25.00 49.99 180 198 408 608 481 4,981 4,419 2,846
50.00 74.99 165 252 498 853 1,129 4,462 2,250 1,029
75.00 99.99 159 502 937 1,388 2,023 2,350 905 426
100.00 124.99 191 881 1,261 1,862 4,699 932 385 271
125.00 149.99 245 1,031 1,835 2,445 5,083 447 241 182
150.00 174.99 416 1,677 2,803 2,572 3,050 246 165 136
175.00 199.99 788 2,922 3,104 2,851 1,167 139 109 78
200.00 224.99 1,030 3,386 3,071 2,253 435 85 45 31
225.00 249.99 1,276 3,123 2,098 1,389 189 29 28 27
250.00 274.99 2,548 2,164 1,218 471 57 26 18 21
275.00 299.99 3,220 1,156 450 243 15 19 23 15
300.00 324.99 3,316 616 198 126 7 19 14 19
325.00 349.99 2,440 246 107 72 24 29 23
350.00 374.99 1,501 104 58 40 16 10 16
375.00 399.99 746 51 29 24 19 22 28
400.00 424.99 254 32 34 31 26 26 19
425.00 449.99 125 29 15 4 22 25 22
450.00 474.99 31 6 4 2 20 15 9
475.00 499.99 22 4 9
500.00 524.99 13 3 10 22
525.00 549.99 1 34 45 35
550.00 574.99 23 6
575.00 599.99 2 10 13
600.00 624.99 15 20 25
625.00 649.99 22 11 12
650.00 674.99 12 9
675.00 699.99

Notes: 1 Negative sums indicate areas where woodland values exceed agricultural values.
Blank cells indicate that no 1 km cells fall into this category. There are 20,563 1 km cells.
Table 9.7. Distribution of the net bene¬ts of retaining sheep farming in Wales as opposed to conversion to broadleaf
(beech) woodland:1 3% discount rate

Farm-gate values Social values
Lower limit Upper limit timber timber+ timber+carbon+ timber+carbon+ timber timber+ timber+carbon+ timber+carbon+
(£/ha/yr, (£/ha/yr, only carbon recreation (CVM) recreation (ITCM) only carbon recreation (CVM) recreation (ITCM)
1990) 1990) (1) (2) (3) (4) (5) (6) (7) (8)
’400.00 ’375.01
’375.00 ’350.01
’350.00 ’325.01
’325.00 ’300.01
’300.00 ’275.01
25 308
’275.00 ’250.01
61 1 364
’250.00 ’225.01
158 102 709
’225.00 ’200.01
141 334 1,047
’200.00 ’175.01
260 740 1,962
’175.00 ’150.01
19 397 24 2,826 5,691
’150.00 ’125.01
165 795 2,492 7,718 7,189
’125.00 ’100.01
350 1,222 9,828 7,432 2,089
’100.00 ’75.01
897 2,181 362 6,691 619 287
’75.00 ’50.01
42 3,284 6,196 6,512 740 205 92
’50.00 ’25.01
2,967 8,853 6,668 10,498 231 97 17
’25.00 ’0.01
0.00 24.99 11,859 5,629 1,533 2,565 68
25.00 49.99 567 4,288 462 182 405
50.00 74.99 7,413 398 175 156 220
75.00 99.99 10,549 259 140 11 1
100.00 124.99 1,336 92 81 88
125.00 149.99 231 169 19 68
150.00 174.99 257 12 194 331
175.00 199.99 135 395 279 90
200.00 224.99 75 82 16
225.00 249.99 1 71
250.00 274.99 16 198 319
275.00 299.99 378 270 97
300.00 324.99 95 20 2

Notes: 1 Negative sums indicate areas where woodland values exceed agricultural values.
Blank cells indicate that no 1 km cells fall into this category. There are 20,563 1 km cells.
Table 9.8. Distribution of the net bene¬ts of retaining milk farming in Wales as opposed to conversion to broadleaf
(beech) woodland:1 3% discount rate

Farm-gate values Social values
Lower limit Upper limit timber timber+ timber+carbon+ timber+carbon+ timber timber+ timber+carbon+ timber+carbon+
(£/ha/yr, (£/ha/yr, only carbon recreation (CVM) recreation (ITCM) only carbon recreation (CVM) recreation (ITCM)
1990) 1990) (1) (2) (3) (4) (5) (6) (7) (8)
’300.00 ’275.01
’275.00 ’250.01
’250.00 ’225.01
11 15
’225.00 ’200.01
11 18 21
’200.00 ’175.01
20 18 61 9
’175.00 ’150.01
3 22 63 114 11 44
’150.00 ’125.01
20 69 113 136 14 27 106
’125.00 ’100.01
11 124 147 179 32 106 158
’100.00 ’75.01
55 156 192 217 3 111 203 278
’75.00 ’50.01
91 214 266 306 28 235 344 413
’50.00 ’25.01
204 273 379 266 64 371 433 460
’25.00 ’0.01
0.00 24.99 174 395 175 174 216 528 359 471
25.00 49.99 345 143 155 130 362 238 231 436
50.00 74.99 543 118 92 159 710 255 258 488
75.00 99.99 202 121 191 144 357 228 343 809
100.00 124.99 151 159 128 205 213 230 552 822
125.00 149.99 113 146 126 201 253 309 754 1,375
150.00 174.99 153 129 150 349 248 566 1,457 2,428
175.00 199.99 145 148 259 412 292 1,396 2,619 3,388
200.00 224.99 142 205 320 532 401 2,128 4,117 3,564
225.00 249.99 145 236 360 561 926 4,374 4,160 3,295
250.00 274.99 161 325 643 959 1,649 4,986 3,258 1,436
275.00 299.99 209 687 981 1,506 3,204 3,272 954 322
300.00 324.99 270 962 1,262 1,789 4,722 916 130 44
325.00 349.99 355 1,014 1,843 2,212 4,259 156 56 28
350.00 374.99 796 1,601 2,392 2,546 2,189 29 8 20
375.00 399.99 949 2,495 2,847 2,267 377 13 20 27
400.00 424.99 1,214 2,947 2,521 2,360 89 26 25 9
425.00 449.99 2,031 2,786 2,455 1,520 1 25 24 19
450.00 474.99 2,686 2,310 1,576 817 11 3
475.00 499.99 2,836 1,680 543 150 2 4 40
500.00 524.99 2,710 614 117 95 27 43 7
525.00 549.99 2,055 209 91 35 21 3 10
550.00 574.99 1,155 117 13 3 20 52 54
575.00 599.99 399 12 2 1 44 12
600.00 624.99 175 3 3 3
625.00 649.99 62 2 24 36
650.00 674.99 3 26 18 10
675.00 699.99 20 5 0
700.00 724.99 0 1 11
725.00 749.99 20 30 27
750.00 774.99 18 33 26
775.00 799.99 26

Notes: 1 Negative sums indicate areas where woodland values exceed agricultural values.
Blank cells indicate that no 1 km cells fall into this category. There are 20,563 1 km cells.
280 Applied Environmental Economics

Turning to consider social values, and remembering that sustainability criteria
may justify use of the 3 per cent rate here, we can see that even if we only consider
timber bene¬ts a large majority (84 per cent) of cells would pass a cost-bene¬t
test of conversion. Addition of carbon bene¬ts indicates that almost the only cells
that would not pass such a test are those located on peat soils. Further addition of
recreation bene¬ts merely reinforces this result.

Milk farms
Table 9.8 summarises results for a conversion from milk production to broadleaf
woodland under a 3 per cent discount rate. Consideration of the farm-gate values
detailed here has to be tempered by the knowledge that a 3 per cent rate is lower than
that we would expect milk farmers to use for everyday decision-making (and that
a risk-weighted rate would be even higher than this). Even so, Table 9.8 indicates
that the long delays associated with broadleaves mean that farm-gate values do not
justify anything but the most minor conversions even when all bene¬ts are paid.
The situation with social values is very similar, with little conversion out of milk
being justi¬ed.

Other discount rates
Given the above discussions and comparisons with observed rates of conversion, it
seems likely that farmers are attaching signi¬cant risk premiums to any decision to
convert to woodland, an observation made elsewhere regarding other non-standard
production (Cobb, 1993). This can be modelled either as a required surplus of
net bene¬ts or as an in¬‚ated discount rate. Given this, consideration of further
reductions in discount rate does not appear to be justi¬ed.11

CBA summary and the present situation
CBA summary
Inspecting the analyses presented in this chapter we feel that the link between our
value estimates calculated at a 6 per cent discount rate, the wider case for using such
a rate and the rates of conversion observed in reality is compelling. Furthermore,
the fact that this is also the UK government™s current discount rate for socially
bene¬cial projects makes the analyses reported in Tables 9.1 to 9.4 of particular
Considering results for a 6 per cent discount rate and taking conifer woodlands
¬rst, we found (Table 9.1) that for sheep farmers the level of grants and subsidies
11 Analyses of lower discount rate scenarios were undertaken. These merely extended the trends observed when
we moved from a 6 per cent to a 3 per cent discount rate.
Cost-bene¬t analysis using GIS 281

paid during our study period was insuf¬cient to justify conversion, a situation which
seems unlikely to have changed up to the present day. However, increasing these
transfers in line with the wider de¬nition of external woodland bene¬ts would
substantially shift the balance of farm-gate values in favour of conversion. Further-
more, our analysis suggests that relatively modest increases in woodland subsidy
could result in signi¬cant rates of uptake among Welsh sheep farmers. Interest-
ingly, our analysis of social values shows that these are already strongly in favour
of conversion and that the increase in subsidies outlined above could generate very
substantial net social bene¬ts. However, turning to consider milk farms, Table 9.2
suggests that neither farm-gate nor social values justify substantial transfers out of
this sector and into conifer woodlands.
When we consider potential conversions to broadleaf woodlands, Table 9.3 shows
that, relative to conifers (Table 9.1), the longer rotation periods mean that the 6
per cent discount rate militates heavily against conversion from sheep farming,
although this is still generally justi¬ed if all the non-market bene¬ts of woodland
are appraised or we shift from farm-gate to social value assessments. However,
Table 9.4 shows that with a 6 per cent discount rate conversions from milk farming
to broadleaf woodland are not generally justi¬ed.
Shifting to a 3 per cent discount rate considerably increases the bene¬ts of wood-
land and so strengthens the case for conversion from sheep farming. However, while
such a rate may theoretically be justi¬ed for the calculation of social net bene¬ts,
it is not in line with present government policy and does not seem to re¬‚ect sheep
farmers™ attitudes towards this type of conversion. Furthermore, this switch does
not fundamentally alter the position with regard to farm-gate values on milk farms
although some positive net social bene¬ts may be derived from conversion if a wide
de¬nition of woodland bene¬ts is employed. Such a low discount rate may not be
valid for assessment of farm-gate values on milk farms.
Clearly, if conversions are to occur, then both farm-gate and social valuations
indicate that these will be most readily derived from the sheep farm sector. In
reality, decision-makers are likely to be faced with only limited resources to effect
such conversions. In such situations our methodology is particularly suited to the
identi¬cation of optimal sites for conversion onto which subsidies can be targeted.
Plate 3f provides a useful illustration of this capacity, showing how we can target
sites according to the net social bene¬ts created by conversion.
Our results also reveal an interesting dichotomy between economic analysis and
policy practice. We have shown that highly populated, readily accessible, lowland
areas provide the optimal location for conversions out of agriculture and into wood-
land. Such sites combine high rates of tree growth with high recreational demand.
However, it is only in recent years, with the advent of the Community Woodland
Scheme and similar schemes, that policy has begun to recognise the strength of this
282 Applied Environmental Economics

argument.12 The legacy of virtually all preceding policies has been a concentra-
tion of woodlands in upland areas, inaccessible to the majority of the population.
Figure 9.1 illustrates the present locations of Forestry Commission conifer wood-
lands in Wales (superimposed upon an elevation map). Comparison with our maps
of optimal conversion areas reveals the disparity between those areas and the actual
locations of the current woodland stock. The overall message of our analysis is
clear: extended economic analysis of both the internal and external net bene¬ts of
conversion shows considerable justi¬cation for bringing forestry down the hill.

The present situation
Finally, we can consider the extent to which the ¬ndings presented in this chapter
need to be modi¬ed by events which have occurred since our 1990 study period.
First, let us consider the timber, carbon-¬xing and recreation values which dominate
our analysis of woodland.
The analysis of timber values presented in Chapter 5 considered a variety of
studies examining possible trends within real timber prices. Arguments can be put
forward in favour of both increases and decreases in real prices. However, the weight
of long-term analysis currently suggests that neither viewpoint can be adequately
established and that an assumption of constant real prices is less prone to error than
either of the alternatives. Such a view is reinforced by recent government policy
papers describing an expanding and vibrant forest estate and industry. This target
can only be achieved, in the absence of new planting by the Forestry Commission,
by maintaining the real value of woodland grants and subsidies, which form a
substantial portion of the discounted income received by forest-owners. Similarly,
as noted in Chapter 7, the increasingly pessimistic predictions of the IPCC and other
experts regarding the apparent acceleration in climate change suggests that carbon
sequestration values will be at least non-declining and arguably may increase in
real terms over time. Such assumptions seem well founded given the recent US
exit from the Kyoto Climate Change Convention on reductions in greenhouse gas
Considering the real value of open-access woodland recreation, in our conclu-
sions to Chapter 3 we presented results suggesting that such values were non-
declining and may even be rising slowly over time. Models of economic devel-
opment suggest that such results are to be expected as economic growth leads to
increasing demand for leisure activities. Although studies of changing work patterns
can challenge the assumption that growth necessarily leads to increased voluntary
12 Interestingly it may well be a non-governmental organisation, the Woodland Trust, which plays a signi¬cant
role in future forest development, funded in part by a grant of over £6 million from the Millennium Fund
(Smith, 1996).
Cost-bene¬t analysis using GIS 283

Average elevation (metres)
Forestry Commission
≥ 350 Sitka spruce

0 10 20 30 40 50 km

Figure 9.1. Location of Forestry Commission sub-compartments of Sitka spruce in Wales
(superimposed upon elevation).

leisure time, increased af¬‚uence should raise the unit value of recreation services
especially for environmental quality goods such as woodland recreation. Again,
assumptions of non-declining values appear justi¬ed.
Overall, therefore, our assumptions of constant real values for the timber, carbon
sequestration and recreational bene¬ts of woodland seem reasonable and may even
284 Applied Environmental Economics

turn out to be conservative. Note, however, that from a farm-gate perspective only
timber and related subsidies provide direct income streams to the prospective farm
forester and therefore the gulf between the market and social value of woodland
seems set to persist for the foreseeable future.
Turning to consider the opportunity cost of agriculture, the 1990s have been,
with the exception of a few good years just before the middle of the decade, a torrid
and depressing period for farming both across the UK and within Wales. Although
real agricultural prices have fallen signi¬cantly over the 1990s this trend does not
represent the full extent of impacts upon farm incomes in Wales. Reductions in
real subsidy levels have compounded price falls such that incomes have more than
halved in all major sectors over the decade. The magnitude of these losses is so
large that the next decade will almost certainly see a continuation of the reduction
in farm numbers seen over recent years. Those that survive may well bene¬t from
policy measures intended to address the current problem. However, as recognised
by the National Assembly for Wales, a return to general levels of pro¬tability based
upon traditional agriculture seems a distant prospect in Wales.
Taking these trends together we can see that from a farm-gate perspective the
attractiveness of forestry as an alternative to conventional agriculture does seem
to have improved over the course of the 1990s, making our ¬ndings appear as
conservative estimates of the ef¬ciency gains of conversions. However, this does
not mean that this change will be suf¬cient to induce large-scale change in the
near future. For the reasons explained in this chapter, farmers may be risk-averse
with respect to changing activities and, while woodland may have improved some-
what in its ¬nancial viability, this may not be suf¬cient to overcome the perceived
security offered by traditional agriculture (although this seems the security of a
familiar poverty). However, what the trends of the 1990s do clearly suggest is that
the superiority of multipurpose woodland over certain sectors of Welsh agriculture,
when viewed from a social values perspective, is likely to have grown over the
course of the decade. This means that the economic CBA case is stronger than ever
for restructuring transfer payments to re¬‚ect the non-market values of woodland
and so facilitate land use conversions out of the most inef¬cient areas of Welsh
farming. Given this, our social value ¬ndings can justi¬ably be described as con-
servative estimates of the current value of transferring land out of farming and into
multipurpose woodland.
Conclusions and future directions

This research draws upon a series of interrelated studies designed to provide an
improved cost-bene¬t analysis of a proposed conversion of land use out of con-
ventional agriculture and into woodland. The analysis covers a number of diverse
questions and is necessarily complex. Consequently a number of conclusions can
be drawn. To simplify this process, we ¬rst review the achievements of this research
before considering, in the subsequent section, the problems of the study and ongoing
work. This is followed by our concluding comments.

Summary of research
As reviewed in the opening chapter of this volume, woodland produces a variety of
market-priced and non-market bene¬ts and costs. The ¬rst phase of this research
was concerned with monetary valuation of one of the principal non-market bene¬ts,
woodland recreation. Given the open-access nature of this good, which produces no
internal return to the land-owner but is of signi¬cant social value, we were forced
to rely upon non-market valuation methods. Chapter 2 reviewed these methods,
highlighting the theoretical appropriateness of both the contingent valuation (CV)
and travel cost (TC) techniques. The chapter also provided a theoretical analysis of
the values elicited by these methods.
Chapter 3 opened with an appraisal of UK applications of these methods to the
valuation of woodland recreation. This review raised a number of interesting issues;
for example, studies failed to identify any signi¬cant link between recreational
values and tree species. We also highlighted a number of problems with prior
studies in terms of their methodology, data analysis and reporting. In an effort to
identify values which could be transferred to woodlands in our study area, cross-
study analyses of both TC and CV estimates were conducted. These yielded separate
and signi¬cantly different valuation measures for subsequent consideration.

286 Applied Environmental Economics

Concerns regarding prior applications were in part the motivation behind our
own studies of recreation value, also presented in Chapter 3. Here we investigated
a number of study design issues, analysing the impact which differing approaches
had upon resultant value estimates. While our initial study was somewhat crude,
we feel that subsequent studies provided some idea of the potential impact of
design effects upon recreation value estimates. More speci¬cally we found that CV
estimates varied signi¬cantly with issues such as question ordering, the inclusion
or exclusion of questions regarding recreational budgets, choice of willingness to
pay format, payment vehicle and respondent type. While much of this variation can
be interpreted in line with economic theory, this does raise the complex question
of which value is the most appropriate for practical purposes. Our research into
the TC method found that its valuations were also subject to variation according to
the methodology employed. In particular we assessed the impacts of measurement
effects, choice of unit values and estimation technique. Variations in estimates
were found to be just as wide, or even wider, for the TC as for the CV approach.
However, the chapter also presents the ¬rst of a series of GIS-based analyses which
dominate the latter part of this volume. Here GIS techniques were used to improve
the measurement of key variables underpinning the TC method so as to produce
more accurate estimates of recreational values.
Chapter 4 opened by considering the equally important question of how many
people will visit a speci¬ed woodland site. Data from our ¬eld studies were used
to estimate a visit demand function which, although theoretically simple, exhibited
some methodological sophistication and proved reasonably reliable in predicting
visits when assessed against a subsample of sites for which actual arrivals were
known. Combining this with the various recreational visit values estimated pre-
viously, we obtained a range of woodland recreation bene¬t values. These varied
according to the valuation method used and methodological assumptions employed.
From these we identi¬ed a preferred upper- and lower-bound estimate of recreation
value for use in subsequent analyses.
The next three chapters switched the focus of analysis to consider tree growth and
its related bene¬ts. Throughout this we considered two species of tree: a represen-
tative conifer (Sitka spruce) and a typical broadleaf (beech). Chapter 5 assessed the
costs and bene¬ts of planting these species, producing estimates of net present value
and its annuity equivalent. This necessitated a study of the appropriate discount rates
for the various decision-makers under consideration (farmers and policy-makers).
The chapter also provided market and shadow price assessments to facilitate in-
vestigation of the value of woodland both to the farmer and to society. This dual
assessment was a feature of all subsequent chapters.
Chapter 6 presented GIS-based models of timber yield. Our methodology allowed
us to use the Forestry Commission™s sub-compartment database, thus permitting
Conclusions and future directions 287

a very substantial increase in sample size compared to previous studies. The GIS
also allowed us to incorporate data taken from the Soil Survey and Land Research
Centre™s (SSLRC) LandIS database detailing the environmental characteristics of
a site. The high quality and extent of these data facilitated the estimation of yield
models which were more robust than those previously reported in the literature.
Information from Chapter 5 allowed us to convert these yield estimates into maps
of timber value for both our conifer and broadleaf species.
The yield model also provided the basis for our analysis of carbon sequestration
in Chapter 7. Forestry Commission models of carbon storage in timber and carbon
liberation from its products were combined with information concerning soil carbon
¬‚ux to produce assessments of the net impact of planting trees upon the carbon cycle.
A review of the literature on valuing carbon storage was used to provide a monetary
evaluation of the results from this model which, as before, involved analyses for
both of our selected tree species.
Chapter 8 shifted attention from woodland to agriculture. The GIS-based models
of agricultural value presented utilise farm-level rather than parish or other aggre-
gated data. This methodology permitted the inclusion of the environmental char-
acteristics of individual farms as explanatory variables in the value functions. A
cluster analysis was used to identify homogeneous sectors within the farm database
and separate modelling exercises were conducted for the two principal sectors “
sheep and milk production. Finally a shadow pricing exercise provided comparisons
with estimated levels of farm-gate income.
All the preceding analyses were synthesised in Chapter 9 which provided a cost-
bene¬t appraisal of converting land out of the two agricultural sectors considered
and into either of the woodland types considered. Net bene¬ts were calculated from
both farm-gate and social perspectives. Comparison of predicted values with the
actual very low numbers of conversions led us to conclude that sheep farms were
using a risk-weighted discount rate of about 6 per cent. While this rate meant that
the level of woodland grants and subsidies made conversion unattractive from the
farmers™ perspective, our analysis showed that conversion from sheep farming to
conifer woodland would generate substantial net social bene¬ts which would jus-
tify the relatively modest increase in grants and subsidies necessary to induce such
conversion. The scope for conversion from sheep production to broadleaf woodland
was reduced by the long rotations of such tree species although some conversion
was still justi¬ed (see the discussion of this issue in the following section). A par-
ticularly important ¬nding was that the optimal location for conversion out of sheep
farming was not, as in general planting practice, in remote upland areas but, rather,
near heavily populated, high accessibility, lowland locations. However, when we
turned to consider milk farms we found little economic justi¬cation for conversion
to either conifer or broadleaf woodland.
288 Applied Environmental Economics

Problems, progress and plans
Prior to presenting our ¬nal conclusions it is essential that we draw the reader™s
attention to several problems and omissions in this research and highlight, in miti-
gation, certain ongoing work addressing some if not all of these problems.
This was a relatively ambitious project covering a wide range of analyses all of
which have scope for improvement. One such area is the need for further consider-
ation of the impact of statistical error in a multimodel system. In particular, while
actual versus predicted tests were conducted on recreational demand and timber
yield estimates, to date such a validation analysis has not been performed for our
agricultural models.
A number of issues arise from our analysis of recreation values. One point, which
is more of a ¬nding than a criticism, is that our CV and TC studies have raised
signi¬cant concerns over the impact of study design, implementation and data
analysis upon resultant valuation estimates. While this is an interesting research
¬nding it does raise questions regarding the use of such values in our subsequent
cost-bene¬t analysis. We have attempted to address these issues by using upper- and
lower-bound estimates in this analysis but feel that this is a less than ideal solution.
In summary, more research into the understanding and control of design effects
is necessary.
Another valuation issue concerns the limitations of the recreation bene¬ts transfer
analysis presented here. While the GIS-based de¬nition of the variables used is
reasonably sophisticated, encompassing factors such as population distribution and
accessibility, other factors such as site characteristics were omitted. However, in
mitigation, our most recent work (overviewed in Chapter 4) shows that these omitted
factors do not radically alter the relative distribution recreation values away from
that predicted by the simpler models used in this analysis. This suggests that our
overall conclusions are not in error here.
A further issue is that, like most studies, the present analysis becomes dated even
while it is under construction. This is particularly true of our agricultural model
which relies upon data from the early 1990s. In Chapter 8 we reviewed the inter-
vening period from then to the present day, noting that the latter half of the 1990s
saw substantial falls in Welsh agricultural incomes. Although, as noted in Chapter 5,
timber prices have also fallen during this period the overall effect seems likely to
have been either neutral or shifting marginally in favour of timber. Such moves imply
that our predictions of the economic potential for land use change out of agriculture
and into multipurpose woodland can be defended as conservative estimates of the
present-day position.
A ¬nal issue we would highlight is that, while our study attempts to signi¬cantly
extend the analysis of costs and bene¬ts, we have omitted certain items. Of these
Conclusions and future directions 289

the more important omissions include sporting revenues (which in some locations
may be highly signi¬cant; see McGilvray and Perman, 1991), livestock shelter, and
externalities such as biodiversity and habitat value (Jenkins, 1984, 1986; Good,
1987; Good et al., 1991; Peterken, 1993; Garrod and Willis, 1994; Woodhouse et al.,
2000; Cowling and Heijnis, 2001), acidi¬cation impacts and landscape amenity
effects (Campbell and Fairley, 1991; Dillman and Bergstrom, 1991; Lavers and
Haines-Young, 1993; Fleischer and Tsur, 2000). Some have argued that values
associated with the bene¬t streams issues, such as biodiversity and habitat values,
may be better incorporated into decision-making by attempting to harmonise CBA
with non-economic appraisal systems such as multicriteria analysis (MCA) and
some commentators have attempted to bring these approaches together (Turner
et al., 2000). We have not attempted such a harmonisation of appraisal approaches,
partly because of time constraints, but principally because of the present lack of a
consistent theoretical framework for such analyses.
Many of the concerns raised above are already the subject of ongoing research.
Considering those externalities which are omitted from our analysis to date, one
area of ongoing work is the assessment of landscape amenity. Funding from various
authorities1 has supported the development of a GIS-based hedonic pricing (HP)
model of such values. The viewshed calculation capabilities of a GIS (which allow
the analyst to measure the extent and type of view observed from any given point
taking into account the natural terrain and man-made visual intrusions and obsta-
cles) make it the ideal tool for compiling map databases of an area, thus obviating
the need to rely on the crude distance-based measures typical of most HP models of
landscape amenity. This work is now well advanced (see Lake et al., 1998, 2000a,b;
Bateman et al., 2001a) and seems promising. A related development has been the
increasing scope for creating realistic 3D visualisations of landscapes from GIS
databases. Our initial research (Lovett et al., 2001; Appleton et al., 2002) leads us to
believe that such techniques would be highly appropriate for enhancing contingent
valuation, conjoint analysis and other expressed preference valuation techniques
so that they might be more readily applied to the valuation of future and planned
A further area of ongoing research examines the biodiversity and habitat values
of woodland, and the implications for these values of implementing the optimal pol-
icy changes implied by the present study. This work combines our various datasets
with those from the British Trust for Ornithology (BTO) to use certain bird species
as ¬‚ags for the wider biodiversity implications of policy change. This research is
still under development but initial results (Bateman et al., 1997c; Woodhouse et al.,
2000; Dolman et al., 2001) and other papers using GIS techniques (Gurnell et al.,
1 Including the Economic and Social Research Council (ESRC), Commission for the European Community
(CEC), Ordnance Survey and the Scottish Executive.
290 Applied Environmental Economics

1996; Swetnam et al., 1998) suggest that this will provide a powerful tool for
identifying the wider effects of the decision on which tree species to use in conver-
sion schemes. Our ¬ndings con¬rm the expected superiority of broadleaves over
conifers as providers of desirable biodiversity outcomes, a factor which has the po-
tential to reverse the apparent economic superiority of softwoods over hardwoods
observed in Chapter 9. As discussed in Chapter 8, an important complicating factor
here is that recent decreases in agricultural incomes have been accompanied by
an increase in stocking densities and consequent overgrazing and ecological dam-
age across many areas of Wales. However, as noted previously, such trends will
only serve to enhance the net bene¬ts of conversion from conventional agriculture
into multipurpose forestry, thus tending to make the results presented here appear
somewhat conservative.
One area in which we have to date achieved little more than a review of the litera-
ture (Bateman, 1992) is the incorporation of the acidi¬cation effects of woodlands,
particularly those composed of conifers. Here, while some evidence is contradic-
tory, the general consensus is that conifers can cause acidi¬cation damage to wa-
tersheds. There is considerable scope for addressing this issue. First, the literature
is extensive, particularly with reference to Wales (see, for example, the numerous
papers contained in Edwards et al., 1990). Second, there is a burgeoning literature
concerning the valuation of acidi¬cation impacts.2 Finally, a number of previous
studies have shown that a GIS provides the ideal tool for catchment analysis (see, for
example, Adams et al., 1995). This should make the future analysis of acidi¬cation
impacts reasonably tractable.

As discussed earlier this research has addressed a number of objectives. However,
we choose to emphasise two general points as its principal features, one method-
ological, the other empirical.

Principal methodological feature
The principal methodological achievement of this research is, we believe, the im-
proved incorporation of spatial and environmental variables into a variety of eco-
nomic models through the medium of GIS. This enhances the researcher™s ability
to model spatial complexity within a variety of economic analyses (Lovett and
Bateman, 2001).
2 This includes two large ongoing studies, one led by Alan Krupnick at Resources For the Future (RFF) in
Washington, D.C., the other conducted by the authors and others at CSERGE, UEA, as part of the CEC
EMERGE project.
Conclusions and future directions 291

A number of examples of this methodology are presented here. For example, the
GIS is employed to incorporate road infrastructure characteristics and the distri-
bution of population in our model of woodland recreation demand. The software
is also used to manipulate and integrate environmental data into our analysis of
agricultural values. Similarly, the GIS provides an ideal medium for combining
a variety of diverse data which had not previously been linked, such as the inte-
gration of SSLRC LandIS and Forestry Commission sub-compartment databases
in our analysis of timber yields. A further feature of this methodology is that the
resultant maps provide easily interpretable results which can readily be used by
decision-makers to analyse the impact of policy changes, and they also provide
information on the most appropriate sites for targeting policy initiatives.
The ¬‚exibility and analytical power of a GIS makes it, we feel, the ideal tool
for incorporating and analysing the spatial complexity which is such an important
part of the real world but is often so conspicuously absent from many economic

Principal empirical feature
This research presents a cost-bene¬t analysis of the agriculture/forestry trade-off
in one large area of the UK. The results of this analysis have, we feel, impor-
tant consequences for future policy. Accepting the caveats set out above, we feel
that the research has highlighted the potential for generating substantial net social
bene¬ts by converting some sheep farms to multipurpose woodland. Furthermore,
the identi¬cation of optimum conversion sites, facilitated by the methodological
advances discussed above, indicates that planting policy to date has been diametri-
cally opposed to that which is required to maximise economic bene¬ts in that it has
been concentrated in remote upland areas rather than accessible lowland locations.
However, our analysis has also shown that levels of woodland grant and subsidy are
insuf¬cient to induce conversion (a result which re¬‚ects real-world observations).
Nevertheless, our results indicate that only modest increases in these grants and sub-
sidies would be necessary to create the ¬nancial incentive for land use conversion
and thereby release the economic net bene¬ts arising from such change.
In essence, our analysis has highlighted the marked difference between the market
appraisal of the status quo and its social value. By including externalities in our
analysis we have shown that the situation is one of poorly targeted government
intervention leading to market failure, a situation which can readily be remedied by
linking transfer payments to the total economic value of goods rather than to their
market price.
Finally, while we recognise that the research presented in this volume is not
fully comprehensive with respect to the full complexities of land use change, we do
292 Applied Environmental Economics

believe that it represents a signi¬cant improvement on the current state of decision
analysis. Furthermore, we feel that the methodology developed here is readily
amenable to extension and that future research may develop this into a practical
decision support system of considerable assistance to policy- and decision-makers
as well as being of interest to academics and users of the land alike.

Achen, C.H. (1982), Interpreting and Using Regression, Quantitative Applications in the
Social Sciences No. 29, Sage, London.
Adams, R., Dunn, S.M., Lunn, R., Mackay, R. and O™Callaghan, J.R. (1995), Assessing
the performance of the NELUP hydrological models for river basin planning, Journal
of Environmental Planning and Management, 38(1): 53“76.
Adger, W.N. and Brown, K. (1994), Land Use and the Causes of Global Warming, John
Wiley and Sons, Chichester.
Adger, W.N., Brown, K., Shiel, R.S. and Whitby, M.C. (1992), Carbon dynamics of land
use in Great Britain, Journal of Environmental Management, 36: 117“33.
Adger, W.N., Pettenella, D. and Whitby, M. (eds.) (1997), Climate Change Mitigation and
European Land Use Policy, CAB International, Wallingford, Oxon.
Adger, W.N. and Subak, S. (1996), Estimating above-ground carbon ¬‚uxes from UK
agricultural land, Geographical Journal, 162(2): 191“204.
Adler, M.D. and Posner, E.A. (1999), Rethinking cost-bene¬t analysis, Yale Law Journal,
109(2): 165“249.
Aitken, M., Anderson, D., Francis, B. and Hinde, J. (1989), Statistical Modelling in
GLIM, Clarendon Press, Oxford.
Ajzen, I. and Fishbein, M. (1977), Attitude“behaviour relations: a theoretical analysis and
review of empirical research, Psychological Bulletin 84(5): 888“918.
Akabua, K.M., Adamowicz, W.L. and Boxall, P.C. (2000), Spatial non-timber valuation
decision support systems, Forestry Chronicle, 76(2): 319“27.
Alberini, A. and Carson, R.T. (2001), Yea-sayers, nay-sayers or just plain confused?
Mixture of populations in contingent valuation survey responses. Paper presented at
the Eleventh Annual Conference of the European Association of Environmental and
Resource Economists (EAERE), University of Southampton, 28“30 June 2001.
Allanson, P., Savage, D. and White, B. (1992), Areal interpolation of parish agricultural
census data, in Whitby, M.C. (ed.), Land Use Change: The Causes and
Consequences, ITE Symposium No. 27, HMSO, London.
Alston, J.E. (1986), An analysis of growth of US farmland prices 1963“82, American
Journal of Agricultural Economics, 68: 1“9.
Altman, D.G. and Gardner, M.J. (1989), Calculating con¬dence intervals for regression
and correlation, in Gardner, M.J. and Altman, D.G. (eds.), Statistics with Con¬dence:
Con¬dence Intervals and Statistical Guidelines, British Medical Journal, London,
pp. 34“49.

294 References

Anderson, K. and Tyers, R. (1991), Global Effects of Liberalizing Trade in Farm
Products, Thames Essay No. 55, Harvester Wheatsheaf, Hemel Hempstead.
Appleton, K.J., Lovett, A.A., S¨ nnenberg, G. and Dockerty, T.D. (2002), Rural landscape
visualisation from GIS databases: a comparison of approaches, options and
problems, Computers, Environment and Urban Systems, 26: 141“62.
Arrow, K.J. Parikh, J. and Pillet, G. (with contributions from Babu, P.G., Beltratti, A.,
Chichilnisky, G., Fankhauser, S., Faucheux, S., Froger, G., Gassmann, F., Haites,
E.F., Hediger, W., Hourcade, J-Ch., Morgan, M.G., Parikh, H., Pearce, D.W., Peck,
S., Richels, R., Schubert, R., Suarez, C. and Wellenmann, U.) (1996), Decision
Making Framework to Address Climate Change, report by IPCC Working Group III
Writing Team II, Intergovernmental Panel on Climate Change, Geneva, Switzerland.
Arrow, K.J, Solow, R., Portney, P.R., Leamer, E.E., Radner, R. and Schuman, E.H. (1993),
Report of the NOAA Panel on Contingent Valuation, Report to the General Counsel
of the US National Oceanic and Atmospheric Administration, Resources for the
Future, Washington, D.C.
Askew, G.P., Payton, R.W. and Shiel, R.S. (1985), Upland soils and land clearance in
Britain during the second millennium BC, in Spratt, D. and Burgess, C. (eds.),
Upland Settlement in Britain: The Second Millennium BC and After, British
Archaeological Report: British Series, vol. 143: 5“33.
Aspinall, R. (1991), GIS and forestry, Mapping Awareness, 5(4): 12“14.
Atkinson, S.E., Crocker, T.D. and Shogren, J.F. (1992), Bayesian exchangeability, bene¬t
transfer, and research ef¬ciency, Water Resources Research, 28(3): 715“22.
Australian Bureau of Agricultural Economics (ABAE) (1985), Agricultural Policies in the
European Community: Their Origins, Nature and Effects on Production and Trade,
Government Publications Service, Canberra.
Ayres, R. and Walter, J. (1991), The greenhouse effect: damages, costs and abatement,
Environmental and Resource Economics, 1: 237“70.
Azar, C. (1998), Are optimal CO2 emissions really optimal? Four critical issues for
economists in the greenhouse, Environmental and Resource Economics, 11(3“4):
Backlund, K., Kristr¨ m, B., L¨ fgren, K.-G. and Samakovlis, E. (1995), Global warming
o o
and dynamic cost-bene¬t analysis under uncertainty: an economic analysis of forest
carbon sequestration, mimeo, Department of Economics, University of Ume˚ , a
Baldock, D. and Conder, D. (eds.) (1987), Removing Land from Agriculture. Council for
the Protection of Rural England and Institute for European Environmental Policy,
Baldwin, B. (1996), Spruce trees provoke turf war, Horticultural Information Service,
Department of Horticulture, University of Saskatchewan.
Balkan, E. and Kahn, R. (1988), The value of changes in deer hunting quality: a
travel-cost approach, Applied Economics, 20: 533“9.
Baron, J. and Maxwell, N.P. (1996), Cost of public goods affects willingness to pay for
them, Journal of Behavioural Decision Making, 9(3): 173“83.
Barrow, P., Hinsley, A.P. and Price, C. (1986), The effect of afforestation on
hydroelectricity generation, Land Use Policy, 3 (April):141“51.
Bateman, I.J. (1991a), Recent developments in the evaluation of non-timber forest
products: the extended CBA method, Quarterly Journal of Forestry, 85(2):
(1991b), Placing money values on the unpriced bene¬ts of forestry, Quarterly Journal
of Forestry, 85(3): 152“65.
References 295

(1992), The United Kingdom, in Wibe, S. and Jones, T. (eds.), Forests: Market and
Intervention Failures, Earthscan, London, pp. 10“57.
(1993), Valuation of the environment, methods and techniques: revealed preference
methods, in Turner R.K. (ed.), Sustainable Economics and Management: Principles
and Practice, Belhaven Press, London, pp. 192“265.
(1995), Environmental and economic appraisal, in O™Riordan, T. (ed.), Environmental
Science for Environmental Management, Longman, Harlow, pp. 45“65.
(1996), An economic comparison of forest recreation, timber and carbon ¬xing values
with agriculture in Wales: a geographical information systems approach, Ph.D.
thesis, Department of Economics, University of Nottingham.
(1999), Environmental impact assessment, cost-bene¬t analysis and the valuation of
environmental impacts, in Petts, J. (ed.), Handbook of Environmental Impact
Assessment, vol. I: Environmental Impact Assessment: Process, Methods and
Potential, Blackwell Science, Oxford, pp. 93“120.
Bateman, I.J., Brainard, J.S., Garrod, G.D. and Lovett, A.A. (1999a), The impact of
journey origin speci¬cation and other measurement assumptions upon individual
travel cost estimates of consumer surplus: a geographical information systems
analysis, Regional Environmental Change, 1(1): 24“30.
Bateman, I.J., Brainard, J.S. and Lovett, A.A. (1995b), Modelling woodland recreation
demand using geographical information systems: a bene¬t transfers study, CSERGE
Global Environmental Change Working Paper 95“06, Centre for Social and
Economic Research on the Global Environment, University of East Anglia and
University College London.
(1998), Transferring multivariate bene¬t functions using geographical information
systems, Nota di Lavoro 84.98, Fondazione Eni Enrico Mattei (FEEM),
Bateman, I.J., Carson, R.T., Day, B., Hanemann, W.M., Hanley, N., Hett, T., Jones-Lee,
M., Loomes, G., Mourato, S., Ozdemirogla, E., Pearce, D.W., Sugden, R. and
Swanson, J. (2002), Economic Valuation with Stated Preference Techniques: A
Manual, Edward Elgar, Cheltenham.
Bateman, I.J., Cole, M., Cooper, P., Georgiou, S., Hadley, D. and Poe, G.L. (2001c),
Visible choice sets and scope sensitivity: an experimental and ¬eld test of study
design effects upon contingent values, CSERGE Environmental Decision Making
Working Paper EDM01“01, Centre for Social and Economic Research on the Global
Environment, University of East Anglia.
Bateman, I.J., Day, B.H., Lake, I. and Lovett, A.A. (2001a), The Effect of Road Traf¬c on
Residential Property Values: A Literature Review and Hedonic Pricing Study, The
Stationery Of¬ce, Edinburgh and London.
Bateman, I.J., Diamand, E., Langford, I.H. and Jones, A.P. (1996b), Household
willingness to pay and farmers™ willingness to accept compensation for establishing a
recreational woodland, Journal of Environmental Planning and Management, 39(1):
Bateman, I.J., Dolman, P., Lovett, A.A. and Brainard, J.S. (1997c), Placing the
biodiversity consequences of land use change within a wider context: developing a
GIS/CBA methodology for assessing conversions from agriculture to farm forestry in
Wales, in O™Riordan, T. (ed.), Economics of Biological Resources and Biodiversity.
Proceedings of a DoE/CSERGE seminar, Department of Environment, London,
31 October 1996, published by the Centre for Social and Economic Research on the
Global Environment, University of East Anglia and University College London,
pp. 33“52.
296 References

Bateman, I.J., Ennew, C., Lovett, A.A. and Rayner, A.J. (1999d), Modelling and mapping
agricultural output values using farm speci¬c details and environmental databases,
Journal of Agricultural Economics, 50(3): 488“511.
Bateman, I.J., Garrod, G.D., Brainard, J.S. and Lovett, A.A. (1996a), Measurement,
valuation and estimation issues in the travel cost method: a geographical information
systems approach, Journal of Agricultural Economics, 47(2): 191“205.
Bateman, I.J., Jones, A.P., Nishikawa, N. and Brouwer, R. (2001d), Bene¬ts transfer in
theory and practice: a review, CSERGE Global Environmental Change Working
Paper GEC2000“25, Centre for Social and Economic Research on the Global
Environment, University of East Anglia and University College London.
Bateman, I.J. and Langford, I.H. (1997a), Budget constraint, temporal and ordering
effects in contingent valuation studies, Environment and Planning A, 29(7):
(1997b), Non-users™ willingness to pay for a National Park: an application and critique
of the contingent valuation method, Regional Studies, 31(6): 571“82.
Bateman, I.J., Langford, I.H., Jones, A.P. and Kerr, G.N. (2001b), Bound and path effects
in multiple-bound dichotomous choice contingent valuation, Resource and Energy
Economics, 23(3): 191“213.
Bateman, I.J., Langford, I.H., Munro, A., Starmer, C. and Sugden, R. (2000a), Estimating
the four Hicksian measures for a public good: a contingent valuation investigation,
Land Economics, 76(3): 355“73.
Bateman, I.J., Langford, I.H., Nishikawa, N. and Lake, I. (2000b), The Axford debate
revisited: a case study illustrating different approaches to the aggregation of bene¬ts
data, Journal of Environmental Planning and Management, 43(2): 291“302.
Bateman, I.J., Langford, I.H. and Rasbash, J. (1999b), Willingness-to-pay question format
effects in contingent valuation studies, in Bateman, I.J. and Willis, K.G. (eds.),
Valuing Environmental Preferences: Theory and Practice of the Contingent Valuation
Method in the US, EU, and Developing Countries, Oxford University Press,
pp. 511“39.
Bateman, I.J., Langford, I.H., Turner, R.K., Willis, K.G. and Garrod, G.D. (1995a),
Elicitation and Truncation Effects in Contingent Valuation Studies, Ecological
Economics, 12(2): 161“79.
Bateman, I.J., Langford, I.H., Willis, K.G., Turner, R.K. and Garrod, G.D. (1993b), The
impacts of changing willingness to pay question format in contingent valuation
studies: an analysis of open-ended, iterative bidding and dichotomous choice
formats, Global Environmental Change Working Paper 93“05, Centre for Social and
Economic Research on the Global Environment, University of East Anglia and
University College London.
Bateman, I.J. and Lovett, A.A. (1992), Modelling the potential impact of changing UK
agro-forestry subsidies using GIS: a bioeconomic approach, in Rideout, T.W. (ed.),
Geographical Information Systems and Urban and Rural Planning, The Planning
and Environmental Study Group of the Institute of British Geographers, Edinburgh,
pp. 100“15.
(1997), Modelling the yield and value of forestry over a large area: a GIS based model
of Sitka spruce and beech yield in Wales, Information Paper, Department of
Resource Management, Lincoln University, Canterbury, New Zealand.
(1998), Using geographical information systems (GIS) and large area databases to
predict yield class: a study of Sitka spruce in Wales, Forestry, 71(2): 147“68.
(2000a), Modelling and mapping timber values using geographical information
systems, Quarterly Journal of Forestry, 94(2): 127“38.
References 297

(2000b), Modelling and valuing carbon sequestration in softwood and hardwood trees,
timber products and forest soils, Journal of Environmental Management, 60(4):
Bateman, I.J., Lovett, A.A. and Brainard, J.S. (1999c), Developing a methodology for
bene¬t transfers using geographical information systems: modelling demand for
woodland recreation, Regional Studies, 33(3): 191“205.
Bateman, I.J. and Mellor, C. (1990), The UK timber market: an econometric model,
Oxford Agrarian Studies, 18(1): 53“61.
Bateman, I.J., Munro, A., Rhodes, B., Starmer, C. and Sugden, R. (1997a), Does
part-whole bias exist? An experimental investigation, Economic Journal, 107(441):
(1997b), A test of the theory of reference-dependent preferences, Quarterly Journal of
Economics, 112(2): 479“505.
Bateman, I.J. and Turner, R.K. (1993), Valuation of the environment, methods and
techniques: the contingent valuation method, in Turner, R.K. (ed.), Sustainable
Environmental Economics and Management: Principles and Practice, Belhaven
Press, London, pp. 120“91.
Bateman, I.J., Turner, R.K. and Bateman, S.D. (1993a), Extending cost bene¬t analysis of
UK highway proposals: environmental evaluation and equity, Project Appraisal,
8(4): 213“24.
Bateman, I.J. and Willis, K.G. (eds.) (1999), Valuing Environmental Preferences: Theory
and Practice of the Contingent Valuation Method in the US, EU, and Developing
Countries, Oxford University Press.
Bateman, I.J., Willis K.G. and Garrod, G.D. (1994), Consistency between contingent
valuation estimates: a comparison of two studies of UK National Parks, Regional
Studies, 28(5): 457“74.
Batterbee, R.W. (1984), Diatom analysis and the acidi¬cation of lakes, Philosophical
Transactions of the Royal Society of London: Series B, 305: 451“77.
Batty, M. and Longley, P.A. (eds.) (1996), Spatial Analysis: Modelling in a GIS
Environment, GeoInformation International, Cambridge.
Baumol, W.J. (1968), On the social rate of discount, American Economic Review, 58:
Beauchamp, T.L. and Bowie, N.E. (1988), Ethical Theory and Business, 3rd edn. Prentice
Hall, Englewood Cliffs, N.J.
Benson, J.F. and Willis, K.G. (1990), The aggregate value of non-priced recreation
bene¬ts of the Forestry Commission estate. Report to the Forestry Commission,
Dept. of Town and Country Planning, University of Newcastle upon Tyne.
(1992), Valuing informal recreation on the Forestry Commission estate, Forestry
Commission Bulletin 104, HMSO, London.
(1993), Implications of recreation demand for forest expansion in Great Britain,
Regional Studies, 27(1): 29“39.
Bergen, V., Behrndt, M. and P¬ster, G. (1992), Germany, in Wibe, S. and Jones, T. (eds.),
Forestry: Market and Intervention Failures, Earthscan, London, pp. 127“64.
Bergh, van den, J.C.J.M., Button, K.J., Nijkamp, P. and Pepping, G.C. (1997),
Meta-Analysis in Environmental Economics, Kluwer Academic Publishers,
Bergland, O., Magnussen, K. and Navrud, S. (1995), Bene¬t transfer: testing for accuracy
and reliability, Discussion Paper #D-03/1995, Agricultural University of Norway, …s.
Berman, M.D. and Kim, H.J. (1999), Endogenous on-site time in the recreation demand
model, Land Economics, 75(4): 603“19.
298 References

Berry, J.K. (1993), Cartographic modelling: the analytical capabilities of GIS, in
Goodchild, M.F., Parks, B.O. and Steyaert, L.T. (eds.), Environmental Modelling
with GIS, Oxford University Press, pp. 58“74.


. 9
( 11)