The Wider Impacts Sub-Objective TAG Unit

Transcription

1 TAG Unit DRAFT FOR CONSULTATION September 2009 Department or Transport Transport Analysis Guidance (TAG) This Unit is part o a amily which can be accessed at

2 Contents 1 The Wider Impacts Sub-obective Introduction WIs estimation method 1 2 Estimating the Agglomeration Impact Agglomeration Average generalised costs Eective densities 6 3 Estimating Output Change in Imperectly Competitive Markets 8 4 Estimating Economic Welare Impacts Arising rom Labour Market Changes Labour supply impact Move to more/less productive obs impact Economic impacts arising rom labour market changes 16 5 Wider Impacts Calculation: Final Steps Build-up and lagged productivity impacts Proiling over the appraisal period 18 6 Data Sources and Data Preparation Overview Data summary Economic data required or WIs Transport model data: overview Transport model data: demand Transport model data: generalised cost Geographical detail o data More advice on meeting data needs 28 7 Wider Impacts Sensitivity Estimates Introduction Sensitivity test with non-tempro data Sensitivity test where a LUTI model is available Sensitivity test or inter-city schemes Sensitivity test or reight trips 31 8 Dealing with Transport Data Deiciencies Overview Intra-zonal ourneys not modelled No recorded demand or generalised cost or OD pairs Insuicient segmentation o modes or purposes 33 9 Transport Data Checklist Presenting the Analysis Overview Inormation used or the decision on which Wider Impacts to appraise Central Wider Impacts estimates Wider Impacts sensitivity results 39

3 10.5 Methodology and urther uncertainties Further Inormation Reerences Document Provenance 45

4 1 THE WIDER IMPACTS SUB-OBJECTIVE 1.1 Introduction Wider Impacts TAG Unit 2.8 explains, in general terms, the orm o the appraisal required or the Wider Impacts and Regeneration Sub-Obectives TAG Unit 2.8 explains that, in the presence o imperect markets, wider impacts are not estimated as part o transport user beneits and must thereore be estimated separately. This TAG Unit provides detailed guidance on the estimation o these wider welare impacts, not measured elsewhere in the scheme appraisal This TAG Unit provides detailed guidance on how to appraise the ollowing Wider Impacts, as described in TAG Unit 2.8: agglomeration; increased or decreased output in imperectly competitive markets; labour market impacts: more/less people working and the move to more/less productive obs; TAG Unit 2.8 sets out the appraisal requirements or each o the Wider Impacts and a summary approach to estimating them. Thereore, TAG Unit 2.8 guidance should be reviewed beore the actual estimation o Wider Impacts applying the detailed methodology set out in this TAG Unit (3.5.14) This unit is structured in the ollowing way: guidance and equations or the estimation o Wider Impacts guidance on data inputs guidance on the appraisal o Wider Impacts tested as a sensitivity case; guidance on additional transport data assumptions; and guidance on presenting the analysis and the Wider Impacts assessment report Sotware is available or the appraisal o Wider Impacts. It is strongly recommended that this is used. The sotware automates the Wider Impacts estimation that is set out in this TAG unit. Please WiderImpacts@dt.gsi.gov.uk or inormation on how to access it. 1.2 WIs estimation method The ollowing chapters present a step-by-step guide to the calculation o Wider Impacts (WIs) TAG Unit 2.8 noted that the estimation o the impact o transport on employment location across areas and over time is challenging and requires use o a Land Use Transport Interaction (LUTI) model. For this reason, particular assumptions must be applied in the production o the central case WIs estimates. These are made clear in the methodology set out later in this TAG Unit. This Guidance Unit also sets out how these assumptions can be varied in the production o sensitivity estimates The particular assumptions that must apply in producing the central WIs estimate are: Employment location must be assumed to be ixed over the modelled period. Residential location must be taken to be ixed over the modelled period Detail on the modelling and assumptions that may be applied in the production o sensitivity estimates is given in chapter 7 o this TAG Unit. 1

5 1.2.5 For each WI, the estimation is described in general terms and the estimation process presented diagrammatically. The detailed equations to be used in estimating each WI ollow on rom the diagrammatic representation set out or each impact. 2

6 2 ESTIMATING THE AGGLOMERATION IMPACT 2.1 Agglomeration The agglomeration metric, known as eective density, provides a measure o the mass o economic activity across the modelled area. This measure relects the accessibility o irms and workers to each other, with the importance o one irm/worker to another declining over distance Since the level o agglomeration in a location is a unction o the proximity o businesses to one another and to workers, the relevant measure o agglomeration (eective density) is the generalised cost or business, commuting and reight travel. The irst step in estimating agglomeration impacts is thereore to calculate the average generalised cost or business, commuting and reight ourneys or each: origin/destination ourney pair scenario modelled year The level o agglomeration is estimated or the modelled base case without the scheme intervention. In this case generalised costs and trip numbers without the intervention are ed into the estimate o agglomeration. Agglomeration is also estimated or the alternative case, where the transport scheme has been implemented. In this case generalised costs and trip numbers with the intervention in place are ed into the estimate o agglomeration TAG Unit 2.8 explained that the level o productivity o irms and workers in an area is aected by the level o agglomeration in that area. In general, more agglomerated activity can lead to higher productivity through any or all o the ollowing: (i) greater business interaction, (ii) more eicient/eective labour market interaction, (iii) more eicient/eective input and output markets due to reduced reight costs Once eective density is estimated in the base and alternative case scenarios, the likely productivity response or the change in the level o agglomeration between the base and alternative case is estimated by applying an elasticity o productivity (with respect to eective density). Taking the relative changes in productivity by sector as a result o changes in agglomeration, the absolute changes in productivity are estimated according to the GDP and employment or the sectors in the areas being assessed. This gives an estimate o total output or each sector and each area. The resulting agglomeration impact is then summed across all origin areas and sectors to give the total agglomeration impact across the modelled area or each modelled year The diagram below sets out the agglomeration estimation in an illustrative way. The welare impacts associated with agglomeration (WI1) are estimated by applying the equations that ollow on rom the diagram: 1 Although reight is considered one source o agglomeration, it is not well known how changes in generalised costs aect changes in destination choice, time o day or mode. Thereore it is recommended that where changes in reight demand and cost aect the alternative estimate o agglomeration, these changes are only actored into WIs in a sensitivity case and not the central case WI estimates. See Chapter 7 below and TAG Unit 3.1.4, Freight Modelling, or urther advice. 3

7 2.1.7 The agglomeration impacts ( WI1) are estimated or each modelled year using the ollowing equation: Where WI WI1 k, i d = d A, k, i B, k, i WI k ρ 1 GDPW k = WI1i B, k, i E B, k, i (2.1a), 1 (2.1) i, k k, 1 i are the sectoral agglomeration impacts or each area i and sector k. They will vary depending on the orecast year. i k WI1 is estimated or each origin area i, where i is the Local Authority District (LAD). Where the modelled transport zones are smaller than LAD areas, it will be necessary to aggregate average generalised costs or each zone to LAD level, in order to estimate A, k, B, k, i d i d, or LAD area i. is the industrial sector, with the sectoral groups deined in chapter 6, section 6.3. Sectoral employment and GDP data may thereore need to be aggregated to this level rom a more detailed level o data. is the orecast year in question. A, k, B, k, i d i, d are the eective densities o origin area i sector k in the alternative case (A) and the base case (B) respectively, to be calculated. This will vary depending on the orecast year. 4

8 k ρ is the elasticity o productivity with respect to eective density or sector k. This will not vary with orecast year. groups deined in chapter 6, section 6.3. k ρ is or the sectoral GDPW, B k E, i B, k i is the GDP per worker o Local Authority District area i sector k in the base case (B). This will vary depending on the orecast year., is total employment in sector k, origin area i in the base scenario (B). This will vary depending on the orecast year. WI1 are the total agglomeration impacts or all sectors k and areas i, to be calculated or a speciic orecast year. See chapter 5, section 5.2 below or inormation on discounting The inal agglomeration estimatewi1 may be positive or negative, depending on the impact o the transport scheme on generalised cost and on employment distribution across the area. 2.2 Average generalised costs Average generalised cost needs to be estimated or the base and alternative A, k, B, k, scenarios, to eed into the estimation o eective densities d i, d. This is done making use o parameters or linking average generalised costs to eective densities or each economic sector k. S, m, g i The average generalised cost o travel or each mode,,, is calculated by averaging over travel purposes, p, and weighting by the number o corresponding trips in every orecast year as indicated in equation (2.2) below. The ollowing speciication should be used in estimating average generalised cost, where the data allows. i g S, m, p, S, m, p, gi, Ti, S, m, p i, = S, m, p, Ti, (2.2) p Where: S, m, g i, are the average generalised costs o travel (weighted average by ourney purpose), between area i and area, or each mode m in the scenario S. This will vary depending on the orecast year, to the extent that costs vary in the modelling o transport (TEE) user impacts. S m represents the scenario; indicating the calculations should be done or both the alternative (A) and the base (B) case. is transport mode: private and public transport. This will not vary depending on the orecast year. 5

9 S, m, p, g i p, is the generalised cost o trips rom transport zone i to transport zone, scenario S, mode m and purpose p. It needs to be aggregated to LAD level. This will vary depending on the orecast year, to the extent that costs vary in the modelling o transport (TEE) user impacts. S, m, p, T i is purpose o travel. It includes business, commuting and in the sensitivity case reight trips. This will not vary depending on the orecast year., is the number o trips rom transport zone i to transport zone in the scenario S by mode m and purpose p. It needs to be aggregated to LAD level. This will vary depending on the orecast year, to the extent that the variable varies in the modelling o transport (TEE) user impacts As mentioned in TAG Unit , Introduction to Model Structures or Public Transport Schemes, time-o-day choice modelling should be considered i the scheme includes changing the dierences between peak and o-peak ares. In that case, equations 2.2 and 4.3 below will have a t subscript to incorporate the time dimension Where transport improvements lead to counter-intuitive changes in average generalised costs, the reasons or this need to be investigated. Provided this is not the result o an error, the use o base scenario trip weights in equation 2.2 combined with 2.3 may resolve the problem Average generalised costs are estimated or each transport model zone and aggregated to give an average generalised cost or each Local Authority District (LAD). The same aggregation process is ollowed or trip numbers. 2.3 Eective densities S, k, d i The equation or eective density,, is a measure o accessibility o zone i to obs in all the destination areas. It depends on the employment level or all sectors in the destination areas as well as the average generalised costs decayed by the alpha parameter or each sector. The unctional orm o eective density is thereore 2 : d S, k, i = E S, S, m ( g ), m, i, k α (2.3) Where: S E, is total employment or all k sectors in area in the scenario S - in the central WIs analysis where land-uses are held ixed, employment will be the same in the alternative (A) and the base (B) case scenarios. This will vary depending on the orecast year. S, m, g i, is the average generalised cost o travel rom area i to area in the scenario S or mode m computed in equation 2.2. This will vary 2 Further inormation on agglomeration and eective density can be ound on Graham (2005), (2006a) (2006b) and (2009) papers. 6

10 m depending on the orecast year, to the extent that costs vary in the modelling o transport (TEE) user impacts. is transport mode: private and public. This will not vary depending on the orecast year. k α is a decay parameter or each group o sectors k as deined in Chapter 6, section 6.3 below. The decay parameter will not vary depending on the orecast year Note that the summation over LAD areas includes = i (the intra-areas) When reporting the welare estimates ( WI1 ) or the central case, eective density (i.e. equation 2.3) must be estimated with ixed employment in the base and alternative case. Where land uses are varied (i.e. where employment is allowed to vary in the alternative case) thewi1 estimate should only be reported as a sensitivity test. Further advice on reporting the outputs o the WIs analysis is given in chapter 10 o this TAG Unit For inter-city schemes, a sensitivity agglomeration estimate must be produced. This is to relect the rationale that the strength o agglomeration productivity impacts diminishes with distance (with a higher alpha value representing stronger distance decay). This sensitivity estimate is explained urther in chapter 7, section

11 3 ESTIMATING OUTPUT CHANGE IN IMPERFECTLY COMPETITIVE MARKETS TAG Unit 2.8 explained that the impact estimated as Output change in Imperectly Competitive Markets is the welare impact that results rom increased or decreased output being valued more highly by consumers than the cost o producing this output. This is computed using a simpliied approach The welare impact rom Output change in Imperectly Competitive Markets is estimated as a ixed proportion o total user beneits to business ourneys. The output o this step is the estimated welare change or each year o the scheme The diagram below sets out the output change in imperectly competitive markets impact estimation in an illustrative way. The welare impact o increased or decreased output in imperectly competitive markets is calculated by applying the equations that ollow Given that the welare impact rom Output change in Imperectly Competitive Markets is a proportion o the beneits to business ourneys and the up-rate actor ι is estimated to be 0.1, we have that, 3 = (3.1a) WI ιbub Where WI3 ι BUB WI3 = 0. 1BUB (3.1) are the impacts o increased or decreased output in imperectly competitive markets, to be calculated 3. WI3 will vary depending on the orecast year. is the imperect competition up-rate actor, currently estimated to be equal to 10% (0.1). The up-rate actor will not vary depending on the orecast year. are total user impacts to business ourneys (time, money costs, reliability gains/losses etc). BUB will vary depending on the orecast year to the extent that the modelled transport (TEE) user impacts vary by year. 3 Until there is urther evidence to suggest otherwise, the impacts rom increased competition, WI2, are assumed to be negligible. This is because in a country like the UK, the transport system is already well developed and it is unlikely that transport would be a binding constraint on competition. 8

12 3.1.5 These impacts are calculated or the whole o the modelled area rom the business user beneits in the Transport Economic Eiciency (TEE) analysis. The impact would be positive where business user beneits are positive overall, and negative where business user beneits associated with the transport scheme are negative BUB may include business and reight user beneits, i the intervention being appraised leads to changes in the cost o reight ourneys. See Chapter 7, section 7.5 or more inormation. 4 It is more likely that the transport scheme will result in an increase in output, as the transport scheme would try to provide an optimal level o transport provision that drives marginal costs down. However, in some cases where schemes increase travel costs, the marginal cost o transport could go up as a result o the intervention. 9

13 4 ESTIMATING ECONOMIC WELFARE IMPACTS ARISING FROM LABOUR MARKET CHANGES 4.1 Labour supply impact Transport costs are likely to aect the overall costs and beneits to an individual rom working. In deciding whether to work, an individual will weigh travel costs against the wage rate o the ob travelled to. As described in TAG Unit 2.8, the labour supply impact is essentially computed by looking at how the estimated change in transport costs aects the incentives or an individual to work, thereore aecting the overall level o labour supplied, the additional value added to the economy and the resulting tax revenue to the government The present calculation is concerned with estimating the additional value added to the economy as a result o transport improvements aecting the number o people attracted into work. The calculation is done three parts: i. calculating how commuting costs change as a result o the scheme and how this will aect the beneit an individual obtains rom working; ii. calculating how the change in the beneit rom working will impact on the overall amount o labour supplied; and iii. calculating the additional national output produced by the new labour supplied The irst step is to estimate the total commuting costs and travel time savings or workers commuting rom one origin (home) zone, and rom this to ind the change in annual commuting costs (including travel time, but expressed in money terms) or each worker rom the base to the alternative case. The output o this step is a set o estimated changes in annual costs o commuting per worker by zone and year. This change in annual commuting costs can be considered as a change in (perceived) annual return rom working. As such, it is divided by the earnings o the workers living in the zone to give the perceived relative change in net earnings. It is assumed that all savings are passed on to workers, and thereore that the irms do not appropriate any o them in the orm o proits In the second step, the return to work (labour supply) elasticity with respect to wages in absolute value is used to convert the perceived relative change in net earnings to a relative change in labour supply. Note that we need to apply a negative sign at this step to correct or the act that we are using the elasticity to changes in generalised cost, which have the opposite eect on labour supply to changes in wages The relative change in labour participation is weighted by the number o workers living in the origin zone and working in all the zones In the third step, the change in labour participation is multiplied by the median 5 wage o the marginal worker (who is less productive than the average worker) and the number o workers commuting rom the origin (home) zone The three step process is repeated and added or all origin (home) zones. The inal output o this section is the total output change in money terms rom the increased or decreased labour supply, or each year TAG Unit 2.8 noted that only part o the labour supply impact is not already measured implicitly within TEE appraisal. The inal step o the estimation thereore requires 5 Using median rather than average wage has the advantage o eliminating the potential small number o outliers with very high wages that would otherwise skew the average wage. 6 The additional proit or irms as a result o taking on the extra workers, (the tax take on marginal proit), is already included in the TEE appraisal. 10

14 isolation o the additional welare impact to TEE appraisal beneits. We will deal with how this is done in section 4.3 o this TAG Unit The diagram below sets out the labour supply impact in an illustrative way The welare impact o labour supply change, i.e. the impact on GDP rom more/less people working (component GP1 ), is estimated by applying the ollowing equation or each orecast year, which combines the three steps mentioned above and the summation over zones 7. GP1 = A, c, ( ) B, c, S, G i, Gi, Wi, LS S, ( ) ( i (1 τ 1) y Wi, S, ε m Wi, ) (4.1) Where GP1 is the impact on GDP rom more/less people working, to be calculated. GP1 will vary depending on the orecast year and it is at the transport model zone level. LS ε is the elasticity o labour supply with respect to eective wages (net o taxes and other transport costs). This will not vary depending on the orecast year. S W, i, is the number o workers living in transport model zone i and working in transport model zone or scenario S 8. For the central case, i.e. where a LUTI model is not being used, the number o workers or all scenarios will simply be the number o workers living in zone i and working in zone in the in the alternative case (A), as taken rom the transport model. I a LUTI model is available a sensitivity test may be undertaken 7 Note that we have simpliied the weighting by the number o workers that appears in both the numerator and denominator o the change in labour supply raction (second term o the equation ater the labour supply elasticity). 8 W (rather than total employment E) is used to ensure data on workers travelling rom home in zone i to a ob in zone is used. 11

15 A, c, where a LUTI model can be used to estimate workers living in zone i and working in zone in the alternative case (A). Where a LUTI model is used the estimates should be reported as a sensitivity test. The number o workers will vary with orecast year. B, c, i, G i,, G are the round-trip commuting average generalised costs o travel between zone i and zone in the alternative case (A) and the base case (B) respectively, calculated as shown in equation (4.3) below; all the modelled zones should be considered in the calculation, including the intra-zonal pairs (i=). These will vary depending on the orecast year, to the extent that they vary in the modelling o transport (TEE) user impacts. c is only commuting purpose. It does not include business or reight. This will not vary depending on the orecast year. y m τ 1 η are gross mean workplace-based earnings in zone ; must be or the m same time period as and it will vary depending on the orecast year to relect wage growth. is the median gross wage o marginal worker entering the labour market in zone. This will vary depending on the orecast year. must be or the same time period (e.g. per week or per year) as. y m is the average tax rate on earnings required to convert the gross y earnings into net earnings, with which the change in commuting costs can appropriately be compared. It is currently estimated to be equal to 30%. This will not vary depending on the orecast year. is the parameter that captures the lower productivity (compared to average) o workers on the margin o the labour orce. Currently it is calculated as the raction o 0.69 o the average wage, but it could take a dierent unctional orm or vary by zone. This will not vary depending on the orecast year. m = ηy (4.2a) m = 0. 69y (4.2) Given equation 4.2a, equation 4.1 simpliies to: GP1 ( ( ) = ε LS η S, A, c, B, c, W i Gi G ),, i, 1 τ (4.1a) 1 i S c G,, i, The round-trip commuting generalised costs o travel in the scenario S are the generalised costs o travel by mode m or the commuting purpose c weighted by the number o corresponding trips in the orecast year : 12

16 S, m, c, S, m, c, ( g g ) B, m, c, i, +, i Ti, S, c, m Gi, = (4.3) B, m, c, T m i, Where S c G,, i, is the round trip commuting generalised cost or scenario S. These will vary depending on the orecast year and they will be in money terms. S, m, c, g i, is the average generalised cost o travel rom zone i to zone in scenario S by mode m or the commuting purpose c in the orecast year. The measure o cost must be or the same period as the wage B, m, c, T i m y terms and (see below). It will be in money terms and travel time will be converted by multiplying by the money value o commuting time. Average generalised cost will vary depending on the orecast year, to the extent that the costs vary in the modelling o transport (TEE) user impacts., is the number o commuting trips rom zone i to zone in the scenario (B) by mode m. This will vary depending on the orecast year, to the extent that it varies in the modelling o transport (TEE) user impacts When the transport scheme involves a time-o-travel dimension, equation 4.3 would need to be modiied to take that into account. The principle applied is the same, only that the weighting would need to take into account that the number o trips vary by time as well and that will aect each o the average generalised costs It is required that the generalised costs o travel and the wage terms and y should all be annual values. They must be consistent with one another, and i they are not annual values, the resulting value o arrive at an annual value. GP1 m will need to be scaled to In theory changes in residential location, modelled using a LUTI model, may aect the level o labour participation. Where a transport scheme aects the residential location o those not participating in the labour market there is the potential or the transport scheme to result in a change in labour participation, through its impact on average generalised cost o travel to obs. Where a LUTI model is used to model residential relocation, eeding this into the estimated labour supply impact, the results must be reported as sensitivity estimates and not in the central case. Central case estimates must assume that residential location is ixed. Further advice on sensitivity testing is given in chapter Move to more/less productive obs impact As well as aecting incentives to supply labour, transport costs are likely to aect the overall costs and beneits to an individual rom working in dierent locations and the beneits to business o operating and employing people in dierent locations. 9 Another possible segmentation that could occur is by socio-economic group. It will enter the assessment through equations 2.2 and 4.3, where another dimension would be introduced to obtain average generalised costs and round-trip commuting costs. See chapter 9 below or urther data issues. 13

17 4.2.2 A LUTI model can be used to model changes in employment location between areas. Changes in the location o employment are likely to impact on the overall productivity o employment. The estimation o the move to more productive obs WI is thereore in two parts: modelling the impact on the transport scheme on the location o employment, and estimating the impact o the changes in employment location on productivity An index o productivity dierentials or Local Authority Districts (LADs) is used to estimate the productivity impact rom modelled employment. The index orms part o the WIs economic data set. For each LAD, the Move to More/Less Productive Jobs impact is estimated by multiplying the change in employment in the area resulting rom the transport intervention with the indexed GDP per worker in the area and summing across areas. The output o this step is the change in total output rom the Move to More/Less Productive Jobs eect, or each year A LUTI model can also be used to estimate changes in residential location that orm part o the modelled response to the transport scheme. These changes in residential location will eed into the estimates o commuter user impacts, and could thereore be expected to aect the level o employment in dierent zones in the alternative case TAG Unit 2.8 noted that only part o the move to more/less productive obs impact is not already measured implicitly within commuter time savings. The inal step o the estimation thereore requires isolation o the additional welare impact to appraisal, equivalent to the tax receipts rom the additional productivity. This issue will be dealt with in the ollowing section The move to more productive obs impact should be calculated only in cases where a LUTI model is being used to orecast the employment and residential relocation consequences o the scheme or policy being appraised. In the central case, GP3 should be assumed to be zero. In the sensitivity case, GP3 may be estimated as below. Further advice on reporting the outputs o the WIs analysis is given in chapter 10 o this TAG Unit The diagram below shows the estimation o the move to more/less productive obs impact in an illustrative way. 14

18 4.2.8 The calculation o the GDP eect o the relocation o obs to more/less productive areas, 10, is calculated as: GP3 Where GP3 N, A, B, ( E E ) GP3 = GDPW PI (4.4) i i is the move to more/less productive obs impact o the alternative case (A) compared with the base (B), to be calculated. This will vary depending on the orecast year. i i N GDPW, is the national average GDP per worker. This will vary depending on the orecast year. A, B, i E i, E are total employment in Local Authority District (LAD) i in the alternative case (A) and the base case (B). These will vary depending on the orecast year. Where modelled zones are smaller than LAD level it will be necessary to aggregate the data to LAD level. PI i is the index o productivity per worker in LAD area i. This will not vary depending on the orecast year, meaning that it is assumed there is no technical progress. 10 Until there is urther evidence to suggest otherwise, the impact rom people choosing to work longer hours, GP2, is assumed to be zero. 15

19 4.3 Economic impacts arising rom labour market changes The wider impacts rom labour market changes additional to Transport Economic Eiciency, TEE, appraisal, WI4, are estimated as the tax wedge rom the more/less people working impact ( GP1 ) and the tax wedge o the move to more/less productive obs impact ( GP3 ). The tax wedges relect income tax, national insurance contributions and corporation tax The diagram below sets out the labour market changes The corresponding equations are: 4 = τ 2GP1 + τ 3GP (4.5a) WI 3 WI4 = 0.4GP G P (4.5) Where WI4 GP1 τ 2 are the labour market changes additional to Transport Economic Eiciency (TEE) appraisal. WI4 will vary depending on the orecast year. is the impact on GDP rom more/less people working computed above. GP1 will vary depending on orecast year. is the tax take on increased labour supply parameter, currently estimated to be equal to 40%. The tax take will not vary depending on orecast year. 16

20 GP3 τ 3 is the move to more/less productive obs impact computed above. GP3 will vary depending on the orecast year. is the tax take on move to more productive obs parameter, currently estimated to be equal to 30%. The tax take will not vary depending on orecast year The inal labour market estimate may be positive or negative, depending on the impact o the transport scheme on generalised cost, employment and residential location across the area. 17

21 5 WIDER IMPACTS CALCULATION: FINAL STEPS 5.1 Build-up and lagged productivity impacts The Wider Impacts would be expected to arise as a result o the generalised cost impacts o the scheme. There is no clear reason to expect any particular time delay in the realisation o these beneits, beyond any delay expected between the transport intervention being introduced and the realisation o generalised cost impacts. The Wider Impacts should thereore be estimated rom the same point, and over the same period, or which the transport scheme is expected to impact on demand and user travel costs. That is, no separate or Wider Impacts speciic ramp-up eect should be applied. General appraisal guidance on whether modelled transport impacts should be scaled up over time in order to allow people s behaviour to change is available in TAG Units (4.2) and (1.7.7). 5.2 Proiling over the appraisal period Typically, appraisals are undertaken by modelling one or more uture years. The impacts are then interpolated between a base or opening year and each o the modelled years. They are then extrapolated rom the last modelled year to the end o the appraisal period. Interpolation could be done by connecting the years by a straight line or using a ixed annual growth actor where appropriate As the magnitude o the WIs is driven by the magnitude o the user impacts, which themselves are extrapolated between modelled years, it is necessary to grow the WI estimates rom modelled year to modelled year. This should be done in accordance with growth in values o time. The relevant values o time to be applied to grow the impacts vary depending on the WI: Agglomeration impacts are driven by changes in business and commuter user impacts which are valued according to working time values. These WIs should thereore be grown rom modelled year to modelled year in line with a weighted average work plus commuter values o time. TAG Unit provides advice on assumed long term annual productivity growth and on the growth o the value o non-work time beneits. Increase or decrease in output in imperectly competitive markets is driven by changes in business user impacts which are valued according to working time values. This WI should thereore be grown rom modelled year to modelled year in line with work values o time. TAG Unit provides advice on assumed long term annual productivity growth. Labour market impacts are driven by commuting user beneits, valued according to commuting values o time. The WIs should thereore grow in line with commuting values o time. TAG Unit provides advice on the growth o the value o non-work time beneits No urther growth assumptions should be applied in extrapolating between modelled years General advice on extrapolation over an appraisal period is available WebTAG Unit on Cost Beneit Analysis Other points to note in producing the inal estimates are: The price base or the assessment, where in monetary terms, should be the same as used in other parts o appraisal. This is currently 2002 prices. 18

22 Finally, and again consistent with the main appraisal, the stream o annual Wider Impacts should be discounted to the irst year o beneits using HM Treasury Green Book discount actors and summed to give a Net Present Value. 19

23 6 DATA SOURCES AND DATA PREPARATION 6.1 Overview This section provides urther background on some o the data requirements or the estimation o Wider Impacts as set out in the previous chapter Beore applying the equations in chapters 2 to 4 o this TAG Unit, it is necessary to assimilate the data or the estimation o Wider Impacts. The data required alls into two groups: A) Economic data: this includes data on the productivity o labour, on employment numbers in an area, and on the likely productivity impacts that result rom changes in the level o agglomeration. This is needed to: Estimate the overall welare impact that results rom the impact o changes in accessibility on agglomeration and other Wider Impacts. Changes in accessibility aect the behaviour o irms and workers and in many cases their productivity. The economic data, in simple terms, translates the inormation on accessibility (across all irms and workers aected) into an estimate o the overall productivity and welare change associated with the modiied accessibility B) Transport model output data on the user impacts o a scheme i.e. generalised cost and travel demand inormation or the dierent users and dierent modes. This is needed to: Estimate how transport impacts on the accessibility o dierent users (e.g. commuters, business passengers and reight users). Changes in the cost o accessibility (i.e. accessing obs, markets and businesses) are key elements in determining a scheme s impact on agglomeration, labour supply, obs travelled to, and also the increased or decreased output produced in imperectly competitive markets The ollowing sections provide guidance on the data requirements or the assessment o Wider Impacts. 6.2 Data summary Tables 1 to 3 below provide a summary o the key data inputs to the Wider Impacts estimation as set out in chapters 2 to 4 o this TAG Unit. The tables indicate whether the data is part o the economic data set, should be sourced rom another place such as TEMPRO, or taken rom the modelled transport user impacts. Table 1: Agglomeration data Variable Name B i GDPW, Data Description Source Details, k GDP per worker in Local Authority District i sector k in the base case (B) varying by orecast year Economic Data Set i is origin area B is base case k is industrial sector is orecast year GDP per worker is in 2002 prices B k E, i, Total employment in the base case in sector k, Economic Data Set i is origin area 20

24 area i varying by orecast year B is base case k is industrial sector is orecast year S E, Total employment or all k sectors or scenario S area varying by orecast year Economic Data Set 11 LUTI and/or local orecasts or the sensitivity WIs estimate is destination area S is scenario: alternative (A) or base (B) case is orecast year k ρ Elasticity o productivity with respect to eective density 12 Economic Data Set ρ (rho) is the agglomeration elasticity k is industrial sector k α Distance decay parameter Economic Data Set α (alpha) is the distance decay parameter k is industrial sector S, m, g i, Average generalized cost o travel rom area i to area in the scenario S or mode m aggregated by purpose and varying by orecast year Transport Outputs Model i is origin area is destination area S is scenario: alternative (A) or base (B) case m is mode: private and public transport is orecast year Average generalised cost is in 2002 prices S, m, p, g i, Average generalised cost o travel rom zone i to zone in the scenario S or mode m and purpose p and varying by orecast year. It needs to be aggregated to the LAD level. Transport Outputs Model i is origin zone is destination zone S is scenario: alternative (A) or base (B) case m is mode: private and public transport p is purpose o travel including business, commuting and reight in the sensitivity case. is orecast year Average generalised cost is in 2002 prices 11 In the standard analysis where land-uses are held ixed, employment will be the same in the alternative case (A) and the base case (B). 12 The sector-weighted agglomeration elasticities should be taken as constant over the appraisal period. The exception is where robust orecast sectoral employment data is available and in these cases agglomeration elasticities may be re-weighted by sectoral mix or every orecast year. 21

25 S, m, p, T i, Number o trips rom zone i to zone or mode m and purpose p and varying by orecast year. It needs to be aggregated to the LAD level. Transport Outputs Model i is origin zone is destination zone S is scenario: alternative (A) or base (B) case m is mode: private and public transport p is purpose o travel including business, commuting and reight in the sensitivity case. is orecast year Table 2: Output in imperectly competitive markets data Variable Name BUB Data Description Source Details Total user impacts to business ourneys ( business user beneits ) varying by orecast year BUB should be extracted rom the cost beneit analysis (using or example TUBA or COBA). Total user impacts to business - time, money and reliability gains/losses. is orecast year BUB is in 2002 prices Where available, total user impacts to reight ourneys (FUB reight user beneits ) may also be included ι Imperect competition parameter Economic Data Set Where included, FUB will be used only in the sensitivity case. ι (iota) is the imperect competition parameter Currently, ι = 0. 1 Table 3: Labour market impacts data Variable Name Data Description Source Details LS ε Elasticity o labour supply with respect to net return rom working Economic Data Set LS ε is elasticity o labour supply S W, i, Number o workers living in zone i and working in Transport Model W is number o workers commuting 22

26 zone varying by orecast year 13 i is origin zone is destination zone S is scenario: alternative (A) or base (B) case is orecast year y Mean gross workplacebased earnings in zone varying by orecast year Economic Data Set is destination zone is orecast year Average earnings is in 2002 prices m Median wage o marginal worker entering the labour market in zone varying by orecast year Derived rom the Economic Data Set is destination zone is orecast year τ Average tax rate on 1 earnings Economic Data Set τ 1 (tau 1) is the average tax rate on earnings required to convert gross earnings y into net earnings. Currently, τ 1 = 0. 3 η Productivity parameter that captures the lower productivity o new entrants to the labour orce Economic Data Set η (eta) is the lower productivity o new entrants parameter. It is currently constant across zones: η = 0.69 A, c. B, c,,, Round-trip commuting i, G i G generalised costs o travel between zone i and zone varying by orecast year Transport Model G is average generalised round-trip cost i is origin zone is destination zone (A) is alternative case (B) is base case c is commuting purpose is orecast year Average generalised round-trip cost is in 2002 prices 13 I a LUTI model is being used to orecast relocation o population and/or obs, the alternative case values should be used and reported only as a sensitivity test. For the central case, and when a LUTI model is not being used, the numbers o workers will be the base case numbers as used in the transport model. 23

27 S, m, c, g i, Average generalised cost o commuting travel rom zone i to zone varying by orecast year Transport Outputs Model i is origin zone is destination zone S is scenario: alternative (A) or base (B) case m is mode: private and public transport c is commuting purpose is orecast year Average generalised cost is in 2002 prices B, m, c, T i, Number o commuting trips rom zone i to zone varying by orecast year Transport Model T is number o trips i is origin zone is destination zone (B) is base case m is mode: private and public transport c is commuting purpose is orecast year N GDPW, Average national GDP per worker varying by orecast year Economic Data Set N is national is orecast year GDP per worker is in 2002 prices A, E i, E B, i Total employment in Local Authority District (LAD) i varying by orecast year Economic Data Set E is total employment in LAD i is origin area (A) is alternative case (B) is base case is orecast year PI i Index o productivity per worker in LAD area i Economic Data Set PI is productivity index i is origin area τ Tax take on Labour 2 Supply parameter Economic Data Set τ 2 (tau 2) is the tax wedge o labour supply parameter Currently, τ 2 = 0. 4 τ 3 Tax take on Move to More/Less Productive Jobs parameter Economic Data Set τ 3 (tau 3) is the tax wedge o move to more productive obs parameter Currently, τ 3 =

28 6.3 Economic data required or WIs There is a speciied core economic data set or the analysis which should be used to ensure consistency o estimates across schemes. A provisional data set ( Wider Impacts Economic Dataset.xls ) is available in WebTAG or use alongside this guidance. Further inormation on the WIs sotware used to estimate Wider Impacts can be sought by ing DT at WiderImpacts@dt.gsi.gov.uk The economic data set provides data or the most recent year or which data is available on a consistent basis. The data is based in 2002 prices to be consistent with the transport model outputs and webtag guidance more generally The WIs appraisal requires workplace-based employment data or each o the years being modelled. In the central case this data should be taken rom TEMPRO, see TAG Unit 3.1.5, Data Sources. Any other sources o employment data should be applied and reported only as a sensitivity test to the main assessment For the purpose o the central case Wider Impacts analysis it is assumed that the transport scheme does not aect the overall location o employment or residents. For each modelled zone the assumption is that land use is ixed in the central case. In the sensitivity tests, local employment orecasts and/or modelled changes in employment rom a LUTI model can be incorporated. Where this is done it must be made clear what employment and residential location assumptions have been applied and where these dier rom those in TEMPRO Where the methodology includes the use o sectoral data this should be used or our sector groups to be consistent with the data groups used in the production o k k agglomeration elasticities and distance decay parameters : Manuacturing Construction Consumer Services ρ Producer Services Table 4 below provides the necessary sectoral aggregation inormation rom UK SIC(92) 2 digit classiication. α Table 4: Sectoral aggregation SIC(92) 2 digits Description 15 Food Manuacturing 17 Textile Manuacturing 18 Apparel Manuacturing 19 Leather Manuacturing 20 Wood Manuacturing 21 Paper Manuacturing 22 Publishing Manuacturing 24 Chemical Manuacturing 25 Plastic Manuacturing 26 Mineral Manuacturing 27 Basic Metals Manuacturing 28 Fabricated Metals Manuacturing Sector Group 25

29 29 Machinery Other Manuacturing 30 Oice Machinery Manuacturing 31 Electrical Machinery Other Manuacturing 32 TV Communication Manuacturing 33 Optical Precision Manuacturing 34 Vehicles Manuacturing 35 Other Transport Manuacturing 36 Furniture and other manuacturing products not elsewhere classiied Manuacturing 45 Construction Construction 50 Motor Trade Consumer Services 51 Wholesale Consumer Services 52 Retail Consumer Services 55 Hotels Restaurants Consumer Services 60 Land Transport Consumer Services 61 Water Transport Consumer Services 63 Travel Support Consumer Services 64 Post Telecom Consumer Services 65 Financial Producer Services 66 Insurance Producer Services 67 Auxiliary Financial Producer Services 71 Machinery Renting Producer Services 72 Computer Services Producer Services 73 R&D Producer Services 74 Other Business Services Producer Services 6.4 Transport model data: overview TAG Unit 2.8 explains that, as the WIs appraisal builds on the modelled user beneits, it relies on the transport model having been well speciied and having good coverage across modes and geographical areas I transport model data or all relevant modes 14 are not incorporated into the assessment, then this is likely to result in errors in the estimation o Wider Impacts. This is because the omission o other relevant modes will lead to an incorrect estimation o the base case level o agglomeration and hence an incorrect estimation o the productivity impact resulting rom any changes in agglomeration caused by the transport scheme. In particular, the impact o the intervention on agglomeration will be exaggerated leading to an over-estimation o agglomeration beneits Multi-modal coverage by the transport model includes capturing data on existing demand and costs as well as data on changes in demand and costs. Both sets o data help determine the estimated ractional change in agglomeration. Hence data on demand and generalised cost are required or all lows, whether they are aected by the modelled intervention or not The need or the transport model to cover all modes that are utilised to a signiicant extent within the modelled area (whether the mode is aected by the transport 14 For the purpose o WIs analysis, relevant modes reers to all modes that are utilised in the modelled area in the base case as well as all modes that are aected by the intervention itsel. 26

30 scheme under consideration or not) may be a particular issue or rail where multimodal models are less common in scheme appraisals. However, it is expected that or the maority o rail schemes or which WIs are likely to be signiicant, a reliable multi-modal model will be available to the appraiser. Hence or most rail schemes that are likely to produce large WIs impacts (positive or negative), the multi-modal requirement o WIs analysis should be attainable through use o a robust multi-modal model. Speciic advice on integrating rail into multi-modal modelling or cases where a multi-modal model is not available (to model rail as well as other modes) is currently being produced. In the interim, i the appraiser believes that the WIs o their rail scheme will be signiicant but a multi-modal model is not available, he/she should seek urther advice rom DT Chapter 8 below discusses special cases where the data is missing or some modes. In most cases, a single mode transport model (i.e. private only) should not be used when estimating WIs In addition to the considerations above, general advice on selecting and speciying a transport model is available under the TAG sections: 3.1: Modelling, 3.10: Variable Demand Modelling, and 3.11: Speciication, Development and Use o Models or Maor Public Transport Schemes. 6.5 Transport model data: demand Demand data should be extracted rom the transport model or the ull set o Origin and Destination (OD) pairs and segmented by mode, ourney purpose and across time periods The OD matrices extracted need to be aggregated to match the level o aggregation or the economic data, normally to Local Authority District (LAD) level Other aggregation o demand data may also be needed as ollows: The generalised cost saving should be aggregated according to shares o dierent user groups (Commuting and Business/In-Work). The case o reight will be discussed in Section 7.5 o this TAG Unit. The data should be aggregated to give average demand or modelled year. This is done by a simple aggregation o OD pairs across time periods, modes and purposes, using appropriate annualisation actors (as described in the TUBA user manual 16 ). 6.6 Transport model data: generalised cost Generalised cost data should be extracted rom the transport model or the ull set o OD pairs. As noted above, it is necessary to include those users and modes that are not aected by the intervention as well as those that are The generalised cost data that is input to the WIs calculation should be valued using WebTAG appraisal values o time in 2002 prices The Wider Impacts assessment analyses the change in accessibility or dierent transport users, and the beneits that derive as a result o this change in accessibility beyond direct user beneits. To allow or this, the measure o the generalised cost change (resulting rom the scheme) needs to be as ull a measure as possible. This means it needs to capture time, travel cost, reliability and crowding beneits, where relevant. The guidance provided in TAG Unit must be ollowed when valuing time saving and operating cost beneits. In-drat guidance is available in TAG Unit 15 For commuting trips, there would generally be trips on both private and public modes, so a multimodal model would always be expected to be a minimum requirement or WI analysis. 16 TUBA User Manual, accessed at: Guidance.pd 27