Partnership Models: Analysis of Options

Transcription

1 Final Report 26 th March 2013 Partnership Models: Analysis of Options Prepared for NZTA

2 Authorship Tim Denne & Stephen Hoskins (09) Covec Ltd, All rights reserved. Disclaimer Although every effort has been made to ensure the accuracy of the material and the integrity of the analysis presented herein, Covec Ltd accepts no liability for any actions taken on the basis of its contents.

3 Contents Executive Summary i 1 Introduction Background Current Contracting Approaches (Pre PTOM) Partnership Contracts under PTOM 2 2 NZTA Description Incentives for Patronage Increase Incentives for Efficient Fares Impacts on Subsidy 7 3 NZTA Description Incentives for Patronage Increase Incentives for Efficient Fares Impacts on Subsidy 9 4 AT Model Description Incentives for Patronage Increase Incentives for Efficient Fares Impacts on Subsidy 11 5 Greater Wellington (GW) Description Incentives for Patronage Increase Incentives for Efficient Fares Impacts on Subsidy 13 6 CPI Model Description Incentives for Patronage Increase Incentives for Efficient Fares Impacts on Subsidy 16

4 7 Patronage Payment (PP) Description Incentives for Patronage Increase Incentives for Efficient Fares Impacts on Subsidy 18 8 Comparison of Results Impact of Revenue Change Incentives to Increase Patronage Adjusting Revenues for Fare Effects Incentives for Efficient Fares Subsidy Impacts Summary of Effects 23 9 Sensitivity Analysis Factors Analysed Commerciality Ratio (CR) Fares Costs Combinations Summary and Conclusions Model Form and Approach Incentives for Patronage Increase Efficient Fares Subsidy Levels Key Decision Choices 30

5 Executive Summary Background The Public Transport Operating Model (PTOM) contributes to the government s goal of growing patronage with less reliance on public subsidy. Consistent with this overarching goal, a number of contract models are being considered that would ensure both operators and councils had incentives to increase patronage. These partnership contracts would share the potential upside rewards (or gains) and the downside risks (or pain) between the funders of public transport (PT) and the operators. In this report we describe a number of options being considered and analyse their effects, particularly on patronage and subsidy level. Partnership Model Options The different models that have been proposed are listed in the table below. In general they are ensuring that both parties gain when patronage increases and lose when patronage falls. They do this by placing some revenue at risk. The different models pursue one of two basic philosophies. They either: Seek to share cost risk with the operator in addition to revenue risk NZTA1, NZTA2 and AT; or They share revenue increases relating to patronage only on the assumption that this is the only element that the operator can influence. Table ES1 Model Description Summaries Model NZTA1 NZTA2 AT GW CPI Patronage Payment (PP) Description Operator receives a share of the difference between revenue increase (since contract start) and indexation payment. This may be positive (revenue increase is more than indexation) or negative (revenue increase is less than indexation) Works the same as NZTA1 when the operator would receive a positive payment, but the downside risk is limited. If revenue decreases, operator risk is limited to share of the reduction in revenue; If revenue increases less than indexation, there is zero payment to the operator. Operator receives a share equal to the revenue change less the difference between gross and net indexation payments. Operator receives a share of the revenue increase that is not a result of price changes (ie patronage-related revenue) Operator receives a share of any increase in revenue above that resulting from fares increasing at the rate of inflation Operator receives a payment equal to patronage increase in absolute terms (numbers of passengers) times a payment per passenger. This can be set so that the patronage payment is equal to the average increase in revenue per additional passenger. All models set aside some revenue to be shared between the council and the operator at a pre-agreed sharing ratio. When revenues fall and there is none to share, the operator may be required to transfer money to the council. All introduce an incentive for operators to increase revenue via increased patronage but they all have limits to the amount shared. i

6 Apart from the PP model, all of the models have the same basic formula: Sharing amount = ΔR - x Where: ΔR = change in revenue since the base year; and x = some fixed element that constrains the quantity shared The value for x in the different models is shown below (Table ES2). The NZTA2 model has some constraints on this whereby it does not share anything across some outcomes. The PP model has a different form; most simply it is expressed as: Shared amount = ΔPax PP Where: ΔPax = change in passenger numbers PP = patronage payment (ie an amount paid per passenger) It can include a threshold also, eg in which the PP is not paid until the change in passenger numbers exceeds some threshold. It can be defined in a way that exactly corresponds to one of the other models through defining the value of the threshold and the PP. Table ES2 x-value in the different models Model x-value Formula NZTA1 Full indexation ΔR - I NZTA2 Full indexation (or zero) ΔR I (or ΔR or 0) AT Gross minus net indexation ΔR I g + I n GW Revenue change that results from fare change ΔR - (R 0 ΔF n (1 + ΔP f)) - (R 0 ΔP f) CPI Revenue change that results if fare change = CPI ΔR (R 0 CPI) Notes: ΔR = change in revenue; I = indexation; Ig = gross indexation payment; In = net indexation payment; R0 = revenue in start year; ΔFn = nominal change in fare levels; ΔPf = change in patronage as a result of change in fares; CPI = consumer price index We illustrate the basic approach used in the partnership models in Figure ES1. The upward sloping solid line represents the share of revenue that goes to the operator under a pure revenue share (Sharing amount = ΔR) in which the operator receives a fixed percentage of the change in revenue. Here we show sharing at a ratio of 50% so that, if revenue increases by $4,000, the operator obtains $2,000. The introduction of an x-value shifts the line to the right. So, if x = $2,000, the line shifts to the new dotted line; it crosses the x-axis when revenue change = $2,000 and at a revenue increase of $4,000, the operator receives $1,000 (50% of $4,000 - $2,000). ii

7 Operator revenue share Figure ES1 Revenue Sharing $5,000 $4,000 Pure revenue share (Shared amount = ΔR) $3,000 $2,000 $1,000 line shifts to the right by 'x' Shared amount = ΔR - x $0 -$10,000 -$8,000 -$6,000 -$4,000 -$2,000 $0 $2,000 $4,000 $6,000 $8,000 $10,000 -$1,000 Revenue increase -$2,000 -$3,000 -$4,000 -$5,000 Analysis Given the objectives relating to patronage and subsidy, we assess: incentives on both operators and councils to increase patronage; incentives on councils to increase fares while recognising the negative impact on patronage; potential impact on subsidy levels. Impact of Revenue Change All the models apart from the PP model are based on sharing the change in revenue and take the same basic form, as noted above. We illustrate the effects in Figure ES2 using a number of basic assumptions. 1 It shows how a change in revenue is shared with the operator under the different models compared with pure revenue share. 2 The sharing amount is the change in revenue minus some amount that sets a threshold. The threshold pushes out the level of revenue change before the operator starts to receive a positive share of revenue. This is most pronounced in the NZTA1 model for which there is no positive flow to the operator until revenues increase by more than 7.5% under our base assumptions. NZTA2 follows the pure revenue share line when revenue change is below zero and the NZTA1 model above the point at which the operator receives positive revenue. The CPI model follows the GW model exactly because we have set fare increase equal to CPI. 1 Start AGC = $1 million; start CR = 40%; gross indexation = $30,000; CPI = 2%; fare increase = 2% 2 The Pure Revenue Share would have a formula of: sharing amount = ΔR iii

8 Operator Revenue Share Figure ES2 Impacts of Revenue Change on Operator Revenue $30,000 Pure revenue share NZTA1 $20,000 NZTA2 AT GW $10,000 CPI $0-8% -6% -4% -2% 0% 2% 4% 6% 8% 10% -$10,000 Revenue Change (% of Start Year) -$20,000 -$30,000 -$40,000 An important thing to note is that the angle of the line is the same in all models. The operator obtains a share of any incremental increase in revenue and the amount is always the same, and is also the same as they would receive with a pure revenue share. This is because all of the models have the basic formula: a variable amount based on revenue change, less a fixed amount. The fixed amount determines the position along the x-axis; the variable amount determines the slope. In some ways this makes the models all exactly equivalent. To the extent that the fixed element of the equation cannot be influenced by the operator, it will be estimated at the beginning of the contract and included in the initial bid price. All that matters is then that the operator has an incentive to increase patronage. However, if the factors determining x are not predictable, the specification of the formula matters. The first three models (NZTA1, NZTA2 and AT) use indexation as the threshold. This shares costs with the operator in addition to revenue. It is equivalent to an approach in which, in order to be able to benefit from sharing revenue change, the operator must accept less generous indexation. The GW and CPI models both set a threshold equal to the revenue change expected from the change in fare level. The GW model calculates the revenue associated with the fare change; the CPI model assumes that fares change with CPI. Bot models isolate the impact of patronage change from fare change (if there was zero CPI and no change in fares, both models would be the same as the pure revenue share). iv

9 Incentives to Increase Patronage All the models provide incentives to increase patronage, with the possible exception of NZTA2 which, under a number of likely outcomes, will provide no patronage incentives. In the other models any increase in patronage will result in an increase in revenue and this marginal increase in revenue will be shared using the agreed percentage. Apart from PP and NZTA2, the marginal incentive is exactly the same across the models, as shown in Figure ES2 above: an additional passenger will result in the same level of net additional revenue for the operator. The PP model can be specified to provide the same marginal incentive. Adjusting Revenues for Fare Effects Two of the models, GW and CPI, adjust the revenue changes to take account of price (fare) effects. These isolate the impacts of revenue that are associated with patronage and share only these with the operator. The GW model does this more explicitly and is more complex. The CPI model operates on the assumption that fares will be increased equal to inflation and does not share the revenue increases that result from that adjustment. The PP model effectively isolates patronage-related revenue by measuring patronage directly. The other models do not isolate revenue that results from patronage change from revenue that results from fare adjustments. Incentives for Efficient Fares The models differ with the extent to which they provide incentives for efficient fares. The NZTA models and the AT model result in the council sharing any revenue increases, whether they are patronage or fare-related, and whether or not the fare increase is equal to, less than or more than inflation. With an estimated price elasticity of demand of -0.4 there is always a revenue gain from increased fare levels: the revenue gain will be greater than any negative impact on patronage. As the council gains a share from revenue increases it benefits from all fare increases. The GW model identifies the fare-related revenue and allocates this fully to the council. In doing so it provides an increased incentive to a council wishing to minimise subsidy levels to increase fares at more than the rate of inflation. It would be best introduced with a policy to limit fare increases to the rate of inflation. The CPI model assumes that fares will increase with inflation and all revenues up to this point will accrue to the council; however beyond this the council has the same incentive to increase fares as under the NZTA and AT models it benefits by sharing a proportion of any revenue increase. The PP model has some incentive to maximise fares if the council wishes to minimise subsidies because the council retains all revenues that are not associated with increases in patronage. v

10 Subsidy Impacts The subsidy impacts are not straightforward to estimate. In simple terms the models can be examined in terms of how revenues are distributed, with anything distributed to operators being equivalent to subsidy. However, this ignores two effects: the possible incentive effects for operators to increase patronage and revenue, such that there is a larger amount to share; and the impacts on initial bid price. For example, expectations of increased revenue for operators following the introduction of the sharing model would be expected to lead to the operators bidding for a lower level of subsidy at the start of the contract. The expected impacts on initial bid price are further complicated by the effects of uncertainty. Greater revenue uncertainty for operators is expected to result in greater requirement for subsidy to compensate for risk. Uncertainty is greatest where the revenue that the operator might earn is most affected by factors that are beyond its control. The models all share a proportion of revenue and thus provide up-side and down-side risk, but some models limit this to patronage-related revenue, whereas others include all revenue and thus expose operators to price risk. In addition, three of the models relate the payments to operators to indexation (cost change). It is likely that the level of uncontrollable risk that the operator is exposed to will have the greatest impact on relative subsidy levels between the sharing models. Those that adjust the payments for indexation introduce most uncertainty, particularly the NZTA1 model. If costs rise significantly, the operator will face a net cost unless there is a substantial increase in revenues. And these effects are most pronounced for contracts with low commerciality ratios. NZTA2 is more certain because there are a number of circumstances in which there will be no sharing. The AT model has uncertainty associated with the effects of indexation, but this is much less than under NZTA1. There are also uncertainties associated with the fare level decisions of councils. These are reduced by the models that take account of fare changes (GW and CPI) and are removed in the PP model that simply rewards patronage. Summary of Effects Table ES3 summarises the effects under the different models. All share an amount that is some proportion of the change in revenue. The PP model does not, but it is equivalent to doing so. All include a threshold or adjust the revenues such that they are not pure revenue share models. The GW and CPI models adjust revenue to focus on patronage revenue only. The PP model focuses directly on patronage, and it can be specified to include a threshold. vi

11 All models provide a marginal incentive to increase patronage, although the NZTA2 model provides no incentive across a range of outcomes. Three models (NZTA1, NZTA2 and AT) share costs in addition to revenue and the other three models isolate revenue from patronage. None of the models provide incentives to change fares efficiently, that we define as increasing at the rate of inflation. This is because, if the price elasticity of demand is less than -1 as NZTA assumes) the council will always gain from a fare increase. Under the GW and PP models there is a particular incentive to raise fare levels at the expense of patronage. Subsidy levels are most likely to be differentiated by the differences in certainty for operators. Those that adjust the amount paid to operators with changes in costs introduce most uncertainty, particularly NZTA1. The PP model has greatest certainty as it limits payment explicitly to patronage. The GW and CPI models attempt to isolate patronage revenue, but do so imperfectly. Table ES3 Summary of Models Factor NZTA1 NZTA2 AT GW CPI PP Sharing amount based on change in revenue Threshold/adjustment Indexation Indexation Marginal incentive to increase patronage Indexation difference Fare-related effects Real fare change Can be included Shares costs Isolates patronage Efficient incentive to change fares Certainty of operator outcome Sensitivity Analysis Commerciality Ratio (CR) Low CR (and low revenues) means there is a need for a higher percentage increase in patronage (and revenue) to overcome fixed costs. This means that the NZTA1 model, in particular, requires a very high growth in patronage to provide sufficient revenue to repay a share of the indexation payment. In contrast, a high CR means there is a need for a smaller percentage increase (although the same absolute increase) in passenger numbers to generate sufficient revenue. Fares If fares do not move there is a requirement for more revenue to be earned from patronage increases to exceed the thresholds that limit payments to operators. The GW vii

12 model takes account of this and adjusts the shared amount for fare changes so that an operator is not penalised. Costs If costs fall, the amounts paid to operators under the GW and CPI models are unaffected, but under the NZTA and AT models, greater amounts are shared with operators. The NZTA1 model is affected most significantly. Summary and Conclusions The different models pursue one of two basic philosophies. They either: o Seek to share cost risk with the operator in addition to revenue risk NZTA1, NZTA2 and AT; or o They seek to isolate revenue increases relating to patronage only as the only element that the operator can influence. All the models (apart from PP) have the same basic form of: Sharing Amount (SA) = ΔR x The NZTA2 model varies and across some circumstances, can be specified as SA = Δr or SA = 0 Because x is a fixed amount, and ΔR varies with patronage, all (apart from NZTA2 and PP) provide the same incentives for patronage increase (and PP can be adjusted to achieve this also). ΔR also varies with fare levels. The GW and PP models (and to a lesser extent the CPI model or only alongside fares policies) do not provide operators with benefits from fare-related revenue changes. Because demand is estimated to be inelastic (patronage levels will fall less than price increases), increasing fares will always raise more. Councils may have an incentive to increase fares above inflation to obtain more revenue. The GW model isolates revenue from fare increases, but does so in a way that incentivises more fare increases (the council retains all fare-related revenue). Guidelines/rules on fare increases would reduce this uncertainty. The specification of x determines how much revenue is retained by the council before the operator receives a share of revenue (and thus potentially how low the subsidy is), but the expectation of future x means that it will affect initial bid prices also. Thus with perfect information all models would result in the same level of subsidy. Because there is not perfect information, what affects subsidy levels will be the extent to which the models increase or decrease certainty for the operator. More uncertainty will require more subsidy to compensate for risk. Uncertainty is greatest when the level of x varies with costs (particularly under the NZTA1 viii

13 model and to a lesser extent the AT model). There is also uncertainty associated with the councils changing of fares; as noted above these can be addressed through fares policies. If fare policies (that limit fare rises to CPI) are set alongside the introduction of a partnership model, uncertainties relating to fares are largely eliminated. This then provides a clear choice between models that either seek to isolate patronage related revenue increases (GW, CPI, PP) and those that seek to share some underlying cost risk (that cannot be eliminated) with the operator in addition to revenue risk. If isolating patronage increases is desired, the PP model does this most explicitly and simply. The GW model is somewhat complex and can deal with situations in which fare levels vary from CPI. The CPI model is based on an assumption that fare policy is adhered to. If sharing cost risk, the NZTA1 model shares all of the cost risk and the AT model shares a proportion of that risk. The NZTA2 models varies between sharing all cost risk (when revenue growth is high) or no cost risk when it is not. ix

14 1 Introduction 1.1 Background The Public Transport Operating Model (PTOM) contributes to the government s goal of growing patronage with less reliance on public subsidy. Consistent with this overarching goal, a number of contract models are being considered that would ensure both operators and councils had incentives to increase patronage. These partnership contracts would share the potential upside rewards (or gains) and the downside risks (or pain) between the funders of public transport (PT) and the operators. In this report we describe a number of options being considered and analyse their effects, particularly on patronage and subsidy level. In this introductory section we first characterise contracts without a sharing arrangement, setting out how the individual parties are affected by changes in revenues and costs. We then describe six options for a partnership approach and point out the effects of changes in the underlying factors that determine costs and revenues. 1.2 Current Contracting Approaches (Pre PTOM) Currently contracts differ with respect to: revenue distribution and indexation. Contracts can be set in gross or net terms with respect to revenue. Under gross contracts the operator is paid a fee that covers their full costs, with councils retaining the revenues. There is no incentive for the operator to increase patronage; any incentives would be from contractual performance measures. Under net contracts the operator retains the revenue from fares and is thus rewarded if patronage increases. A fee is paid to make up the difference between revenues and costs. Indexation payments are made to compensate operators for estimated increases in costs. These too may be in gross or net terms. In both cases they use an NZTA index which represents an estimate of changes in costs for a bundle of relevant inputs, 3 Gross indexation increases the fee by an amount equal to the previous year s total annual costs times the index. Net indexation increases the fee by an amount equal to the previous year s fee times the index. The amount will thus differ with the commerciality ratio. Gross indexation fully compensates for changes in costs and could be said to produce the same outcome as if the contract was opened for competitive bidding each year 3 The Procurement Manual states that contract prices must be adjusted to compensate for fluctuations in the price of inputs (eg wage rates, fuel prices) for any contract with a term of 12 months or more. This is undertaken on a quarterly basis using a standard index provided by the NZTA. There are separate indices for diesel bus and ferry contracts. 1

15 (assuming that the index is an accurate representation of actual cost increases). The difference between net and gross indexation is greatest at a high commerciality ratio, where it might also be argued that the operator is best able to take on revenue risk. Contracts can be gross-gross, net-net, gross-net (gross contract, net indexation) or netgross. In this report, and in the discussion of different partnership models, the intent is to get away from the distinction between gross and net contracts, ie those that fix which party obtains revenue. The partnership approach is attempting to identify what is the best way to distribute upside and downside risks relating to patronage and revenues. For the description of options below we use a neutral model in which there is no prior decision on which party receives the revenue, although in practice and for consistency, we have assumed that any excess revenue is retained by (or accrues to) the council. Indexation can still be gross or net, reflecting local preferences. 1.3 Partnership Contracts under PTOM Options In this section we examine the different options that have been proposed for sharing risks and rewards. The different approaches all introduce an incentive for operators to increase revenue via increased patronage but they differ with respect to the limits to this. The different models pursue one of two basic philosophies. They either (1) seek to share cost risk with the operator in addition to revenue risk NZTA1, NZTA2 and AT; or (2) they share revenue increases relating to patronage only, assuming that this is the only element that the operator can influence. The models are summarised in Table 1 Table 1 Model Description Summaries Model Description Threshold NZTA1 NZTA2 AT GW CPI Operator receives a share of the difference between revenue increase (ΔR) (since contract start) and indexation payment. This may be positive (ΔR > indexation) or negative (ΔR < indexation) Works the same as NZTA1 when the operator would receive a positive payment, but the downside risk is limited. If revenue decreases, operator risk is limited to share of the reduction in revenue; If revenue increases less than indexation, there is zero payment to the operator. Operator receives a share equal to the revenue change less the difference between gross and net indexation payments. Indexation payment Indexation payment Difference between gross & net indexation Operator receives a share of the revenue increase that is not a result Revenue that of price changes (ie patronage-related revenue) results from fare change Operator receives a share of any increase in revenue above that resulting from fares increasing at the rate of inflation Revenue that results if fare change = CPI Patronage Operator receives a payment equal to patronage increase in absolute Can be set at Payment (PP) terms (numbers of passengers) times a payment per passenger. any level 2

16 The NZTA models require operators to repay a share of the indexation payment; this functions as a threshold. Operators receive a positive share of revenues once revenue increases are greater than indexation; they must pay the council when revenue change is less than indexation. This is more difficult to overcome with low starting commerciality because the threshold is set in absolute terms. For example, if starting with a 25% commerciality ratio, for the revenue increase to be greater than indexation it would need to be four times as much in percentage terms (eg with a 2% per annum indexation payment, the operator would only receive a share of revenue if revenue increased by more than 8%). The Auckland Transport (AT) model shares risks and benefits (pain and gain) across both revenues and costs. It shares all changes in revenues between the operator and the council less the difference between the gross and net indexation amount. The Greater Wellington (GW) model is based on the fare adjustment formula included in the PTOM procurement manual. This formula is used to ensure that operators are not made better or worse off as a result of changes to fares. It does this by estimating the impact on revenues of changes in (real) fare levels; these revenues go to the council. This is adapted to being a partnership model by assuming the remainder of revenues are the result of changes in patronage levels and are shared with the operator. The CPI model is based on a simple assumption that revenues would be expected to remain constant in real terms (and thus to increase in nominal terms at a rate equal to CPI). This assumes that fares increase at the rate of inflation and there is no expected change in passenger numbers; any increase in patronage results in a real increase in revenues and this increase is shared between operators and councils. The Patronage Payment (PP) model is a straightforward payment for any increase in patronage using a payment per additional passenger. A threshold can be set, eg payments start above an x% increase in passengers. Negative payments can operate for reductions in passengers (or reductions below the threshold) Analytical Approach We analyse the models in more detail below. Given the objectives relating to patronage and subsidy, we assess: incentives on both operators and councils to increase patronage; incentives on councils to increase fares while recognising the negative impact on patronage; potential impact on subsidy levels. We would expect that the partnership model would: provide an incentive to both parties to increase patronage by both benefiting from each additional passenger and each losing from a reduction in passengers; take account of price changes that might affect patronage so that rewards are not distorted by inefficient pricing; 3

17 not incentivise inefficient pricing we comment below on the issue of price elasticities of demand; not lead to increasing levels of subsidy payment per additional passenger. Patronage Impact The patronage incentive can be analysed by examining the effect on the operator (and the council) of one more passenger, or one more dollar of revenue. Fares The issue of fare levels is complicated because of the limited information on price elasticity of demand. NZTA suggests that a constant elasticity of -0.4 is used, ie that every 1% increase in price will result in a 0.4% reduction in patronage. However, assuming an elasticity that is constant and less than 1 means that there is always an incentive to increase fares: the increase in revenue will be greater than the reduction in passengers in response to fare levels. Given an objective of reducing subsidy, increasing fares might be an obvious and favoured approach. In analysis we assume instead that efficient fares would increase at the level of inflation. This means that fare levels do not result in any negative impact on passenger numbers and any increase in patronage is the result of greater demand for public transport. Subsidy Impact The impact on subsidy levels is somewhat difficult to estimate. Simplistically, any amount paid to an operator in a partnership model is not paid to the council and therefore results in a need for more subsidy. However, the models are expected to provide incentives for operators to increase patronage (and thus revenue); and any increase in revenue in this way, that is shared by the council, will reduce subsidy. In addition, any expectations of future flows of money (to or from the operator) will affect the initial bid price which sets the baseline for all models. In theory, and given perfect information, subsidy levels will be the same under any model. The models will differ if: there are differences in how unexpected changes in revenues are distributed (unexpected changes in costs are addressed via indexation); there are differences in the certainty of future revenue flows under the individual models, with greater uncertainty expected to result in greater need for subsidy to compensate for risk. Operators will be able to predict some market uncertainties, but the council s approach to pricing may be unpredictable; thus uncertainty is increased in models where operator revenue is affected by price change. This means that there can be no definite estimate of the impacts on subsidy levels. To assist the analysis we provide the following pieces of information: additional (marginal) subsidy paid per additional passenger; uncertainty how vulnerable is operator revenue to price change? Below we describe a number of models that are currently being considered. In the next sections we describe and analyse the different models; we summarise the effects in Section 8 and undertake sensitivity analysis in Section 9. 4

18 2 NZTA Description The NZTA1 model shares revenue surpluses with the operator only after indexation costs have been covered. The amount shared is defined by the formula: Shared amount = ΔR - I Where: ΔR = change in revenue from the commencement of the contract I = indexation payment All other revenue is retained by the council. It is effectively a two-part sharing model with a fixed and a variable element. The fixed component is the share of the indexation payment. This is based on factors outside of the controls of the council and operator. The variable element is the change in revenue. To illustrate the approach, we show a number of possible cases in Table 2. Table 2 Sharing under NZTA1 Period Item Case A Case B Case C Case D Description Indexation Positive Positive Positive Negative Revenue change Positive Positive Negative Negative Start year Annual Gross Costs $1,000,000 $1,000,000 $1,000,000 $1,000,000 Revenues $400,000 $400,000 $400,000 $400,000 Commerciality Ratio 40% 40% 40% 40% Subsidy $600,000 $600,000 $600,000 $600,000 Year 1 Cost change 3.0% 3.0% 3.0% -1.0% Gross indexation payment $30,000 $30,000 $30,000 -$10,000 Net indexation payment $18,000 $18,000 $18,000 -$6,000 Patronage change 6.0% 2.0% -2.0% -2.0% Fare change 2.0% 2.0% 2.0% 2.0% Revenue change 8.12% 4.04% -0.04% -0.04% Revenue change $32,480 $16,160 -$160 -$160 Gross Share amount $2,480 -$13,840 -$30,160 $9,840 Indexation Revenue to council $31,240 $23,080 $14,920 -$5,080 Revenue to operator $1,240 -$6,920 -$15,080 $4,920 Net Share amount $14,480 -$1,840 -$18,160 $5,840 Indexation Revenue to council $25,240 $17,080 $8,920 -$3,080 Revenue to operator $7,240 -$920 -$9,080 $2,920 All cases start with annual gross costs of $1 million in the start year and revenues of $0.4 million requiring a subsidy of $0.6 million. Costs increase in year 1 by 3% (Cases A to C) or fall by 1% (Case D), resulting in indexation payments as shown. Patronage and fare changes are assumed and used to estimate the change in revenue in dollar and percentage terms. 5

19 In Case A, the increase in revenue is greater than the increase in costs (and indexation payment). There is a small surplus that is shared, but most of the revenue is used to reimburse the council for the indexation payment. Under net indexation more is shared with the operator because there is less indexation to reimburse the council for. In Case B there is a smaller change in patronage such that revenues do not increase by as much as costs. As a result the council receives all of the revenue increase ($16,160) and the difference between this and the indexation payment ($30,000 - $16,160 = $13,840 or $1,840 under net indexation) is a loss shared between the operator and the council, ie the operator must pay back some of the indexation payment. Under Case C patronage and revenues fall and costs rise. The gap between indexation payment and revenue is even greater and the operator pays back approximately half of the indexation payment received. Case D has both costs and revenues falling, but the costs fall by more than the revenues. Here the indexation payment results in the operator paying the council, but because the sharing amount is positive (revenues are less negative than costs), the council pays the operator. 2.2 Incentives for Patronage Increase Once the model is introduced the operator benefits from any increase in revenue, be it patronage-related or price-related. The formula has a simple specification in which the shared amount increases by $1 for every additional $1 of revenue. The operator and the council will share this additional $1 using a pre-agreed percentage, so at a 50% sharing arrangement, each additional $1 of revenue will result n the operator receiving $0.50. Note, this is different from the potential outcome of the operator facing a net payment despite patronage increases (Case B). The negative payment is affected by the level of cost indexation and would shrink in size if patronage increased further. As noted above, there is a relatively high hurdle that needs to be overcome before a positive payment is received by an operator and it benefits from the partnership model (versus no partnership model). Restating the example given earlier, if starting with 25% commerciality ratio (CR), the operator would face a net penalty if the growth in patronage is not four times greater in percentage terms than the growth in costs (and two times at 50% CR and 1.5 times at 75% CR). This does not change the incentive for patronage increases (that is based on how the sharing amount changes with each additional dollar of revenue), but it does make the model unattractive to operators versus the status quo or means that initial bid prices for contracts will be increased to take account of this additional risk. Some patronage increases (or decreases) may be fare-related. For example, in theory patronage will increase if fares are not raised in nominal terms and thus drop in real 6

20 (inflation-adjusted) terms. Operators and councils are both rewarded under these circumstances. Thus NZTA1 is offering relatively efficient incentives for patronage increases by always rewarding these increases. 2.3 Incentives for Efficient Fares If fares increase, revenues increase and the council (and operator) gains from a share of the increased revenue. There is no limit to this incentive and, as noted above, given a price elasticity of demand of -0.4, fare increases will always be expected to raise more revenue. Thus there is an incentive to raise fare levels and no disincentive to limit this to inflation. 2.4 Impacts on Subsidy Although it is possible to estimate the effects of the model on total subsidy levels by calculating the level of payment to the operator relative to different starting positions (eg gross or net contracts), in practice these expected effects will be reflected back in initial bid prices such that the differences are not meaningful as an estimate of the final impacts on subsidy levels. To avoid influencing the choice of model we do not present these static effects on subsidy here. We limit the discussion of subsidy to impacts at the margin and uncertainty Marginal Subsidy The subsidy payment per additional passenger is equal to the revenue raised by the passenger times the sharing amount. So if the sharing ratio is 50%, for each additional passenger trip paying a $2.50 fare there is a subsidy of $1.25. This is regardless of the starting level of commerciality. However, the net impact on subsidy levels depends on what would have happened otherwise. If the additional passenger would have come even if the operator received no revenue, then the difference between with and without sharing is the payment of the $1.25 to the operator. However, if the passenger is additional because of the sharing model, then effectively there is a reduction in subsidy as a result; the council obtains $1.25 (and the $1.25 payment to the operator is not a cost to the council it would not have happened in the absence of the sharing model) Uncertainty As noted in Section 1.3, the main contributor to total subsidy levels will be uncertainty of income. The NZTA1 model introduces revenue uncertainty because of the relationship to indexation. Whereas under this model, at any time raising an additional dollar of revenue will always benefit an operator, the net benefit to the operator of any given level of patronage depends also on the level of indexation, and this is not certain. To explain: if revenue increases by $30,000, whether the operator benefits depends on whether or not costs have also increased by this amount. The aggregate uncertainty is greater than without the sharing model and is thus likely to increase required subsidy levels. 7

21 3 NZTA Description NZTA 2 is a variant on NZTA 1 that limits the downside. It has the same threshold that applies on the up-side so that the operator does not receive a positive payment until there is a significant increase in revenues, but the downside costs are significantly limited. The formula used is: Shared amount = If ΔR < 0 = ΔR If ΔR 0 then If I > ΔR = 0 If I < ΔR = ΔR - I Where: ΔR = change in revenue from the commencement of the contract I = indexation payment When revenue falls, the shared loss is equal to the reduction in revenue and does not take account of changes in indexation payment. When revenue rises the operator shares the upside gain only if the revenue increase is greater than indexation, but if it is not, there is no penalty or reward. We use the same examples to illustrate the impacts in Table 3. The top part of the table is unchanged and is shaded. Table 3 Sharing under NZTA2 Period Item Case A Case B Case C Case D Description Indexation Positive Positive Positive Negative Revenue change Positive Positive Negative Negative Start year Annual Gross Costs $1,000,000 $1,000,000 $1,000,000 $1,000,000 Revenues $400,000 $400,000 $400,000 $400,000 Commerciality Ratio 40% 40% 40% 40% Subsidy $600,000 $600,000 $600,000 $600,000 Year 1 Cost change 3.0% 3.0% 3.0% -1.0% Gross indexation payment $30,000 $30,000 $30,000 -$10,000 Net indexation payment $18,000 $18,000 $18,000 -$6,000 Patronage change 6.0% 2.0% -2.0% -2.0% Fare change 2.0% 2.0% 2.0% 2.0% Revenue change 8.12% 4.04% -0.04% -0.04% Revenue change $32,480 $16,160 -$160 -$160 Gross Share amount $2,480 $0 -$160 -$160 Indexation Revenue to council $31,240 $16,160 -$80 -$80 Revenue to operator $1,240 $0 -$80 -$80 Net Share amount $14,480 $0 -$160 -$160 Indexation Revenue to council $25,240 $16,160 -$80 -$80 Revenue to operator $7,240 $0 -$80 -$80 8

22 The results are the same as in NZTA1 for Case A, but the other cases all have greatly reduced effects. Case B results in no sharing; the council retains all the additional revenue. In Case C the small revenue reduction is shared. In Case D, rather than a positive payment to the operator, there is a negative payment equal to half the revenue loss. 3.2 Incentives for Patronage Increase The limitation on downside risk means that the incentives for increasing patronage and revenue are limited also. Operators benefit from raising additional revenue only if: The revenue increase is greater (in dollar terms) than indexation payment; or Revenues are otherwise falling. In other cases, the council retains all of the revenue. Thus incentives for patronage increases are limited to periods of low inflation, or to high CR routes. 3.3 Incentives for Efficient Fares As with NZTA1 raising fares increases revenues, but in NZTA2 there can be an even greater incentive to do so because there are possibilities of no sharing options in which the council retains all revenue gains, although, this is no different from a gross contract with no sharing. 3.4 Impacts on Subsidy Marginal Subsidy As with the incentive payment, the marginal subsidy payment depends on the circumstances with respect to costs. If the revenue increase is greater than the cost increase then additional passengers result in additional subsidy equal to the revenue raised times the sharing ratio. However, as discussed above, this holds unless the additional passengers are directly attributable to the introduction of the partnership model, in which case there is a reduction in subsidy for each additional passenger: the council receives a share of the extra revenue. If revenues increase less than indexation there is no impact on subsidy levels. If revenues fall, subsidy levels fall also as some of the revenue fall is borne by the operator Uncertainty The uncertainties for operators are likely to be reduced compared with NZTA1 because in many circumstances zero sharing is likely to result, ie increasing revenue but not as much as increases in costs. Thus the model may have little impact on subsidy or incentives. 9

23 4 AT Model 4.1 Description Under the AT model the amount paid to operators is a share of the change in revenue and of the difference between the gross and net indexation payment. It was initially designed for net contracts, but here we apply it to a neutral contract. The formula used is: Shared amount = ΔR - Ig + In (where gross indexation has been paid) = ΔR + Ig In (where net indexation has been paid) Where: Ig = indexation payment that would apply under a gross contract In = indexation payment under a net contract ΔR = change in revenue The model is doing two things: sharing all revenue increases or decreases, and sharing the differences between indexation payment options in a way that results in more sharing of cost risk. As with NZTA1 and 2, it effectively has a fixed and a variable element: the fixed element is the difference between the indexation payments, which cannot be influenced by the council or operator; the variable element is the revenue change, which can. We show the effects using the same table below. Table 4 Sharing under AT Model Period Item Case A Case B Case C Case D Description Indexation Positive Positive Positive Negative Revenue change Positive Positive Negative Negative Start year Annual Gross Costs $1,000,000 $1,000,000 $1,000,000 $1,000,000 Revenues $400,000 $400,000 $400,000 $400,000 Commerciality Ratio 40% 40% 40% 40% Subsidy $600,000 $600,000 $600,000 $600,000 Year 1 Cost change 3.0% 3.0% 3.0% -1.0% Gross indexation payment $30,000 $30,000 $30,000 -$10,000 Net indexation payment $18,000 $18,000 $18,000 -$6,000 Patronage change 6.0% 2.0% -2.0% -2.0% Fare change 2.0% 2.0% 2.0% 2.0% Revenue change 8.12% 4.04% -0.04% -0.04% Revenue change $32,480 $16,160 -$160 -$160 Gross Share amount $20,480 $4,160 -$12,160 $3,840 Indexation Revenue to council $22,240 $14,080 $5,920 -$2,080 Revenue to operator $10,240 $2,080 -$6,080 $1,920 Net Share amount $44,480 $28,160 $11,840 -$4,160 Indexation Revenue to council $10,240 $2,080 -$6,080 $1,920 Revenue to operator $22,240 $14,080 $5,920 -$2,080 10

24 4.2 Incentives for Patronage Increase All revenue increases are shared so, in the same way as in NZTA1, every $1 of additional revenue raised as a result of patronage increases is shared between the operator and the council. It does not take account of the effects of fare levels on patronage. 4.3 Incentives for Efficient Fares As with NZTA1 there is an incentive to increase fares because it increases revenues. There is no bound to this incentive. 4.4 Impacts on Subsidy Marginal Subsidy The marginal subsidy impact is the same as under NZTA1. This model has a fixed element, which is the difference between gross and net indexation and this is unchanged with changes in patronage and revenues. Each additional passenger (and quantity of fare revenue received) results in an additional subsidy payment at the same level as under NZTA1, ie $1.25 for every $2.50 of additional revenue Uncertainty Uncertainty levels are similar to those under NZTA1. An operator knows it will always gain from increasing patronage, but it final payment is dependent on factors beyond its control, ie the levels of indexation payment. 11

25 5 Greater Wellington (GW) 5.1 Description The Greater Wellington (GW) model shares revenue that is not related to changes in fares. It uses an approach based on a formula in Section of the Procurement Manual that sets out the basis for resetting contracts following a change in fare levels. The formula estimates the effects of price change on revenues by estimating: (1) the increased (or decreased) revenue resulting from current passengers paying the changed fare; (2) the change in the number of passengers (increase or decrease) in response to the change in fare level; and (3) the effects of the price change on any additional (or fewer) passengers that has resulted from other impacts on patronage levels. It is described for application to a net contract as a way to estimate an amount that would be clawed back from the operator to the council. However it could also be used with a neutral contract to define the amount that went to the council with the residual amount of revenue change being that attributed to patronage change. This patronagerelated revenue could then be shared between operator and council. The formula is: Sharing amount = ΔR - (R0 ΔFn (1 + ΔPf)) - (R0 ΔPf) Where: ΔR = change in revenue (as in other models) R0 = revenue in previous year ΔFn = change in fare level in nominal terms ΔPf = percentage change in passenger numbers as a result of change in fares = E ΔFr E = price elasticity of demand, assumed to be -0.4 based on NZTA advice ΔFr = change in fare level in real terms = ((1 + ΔFn)/(1 + CPI)) 1 CPI = consumer price index The formula has a number of elements: It starts with the total change in revenue (ΔR) before subtracting the effects associated with price change, which includes the following: o the change in revenue that is based on the current number of passengers paying more for their fares (R0 ΔFn), ie previous year s revenue times the change in fare level; o the change in the number of passengers affected (R0 ΔFn is multiplied by 1 + ΔPf). This adjusts the number of passengers that are assumed to pay the increased price to only those that continue to use PT following the change in fare level; o the change in total revenue as a result of the change in the number of passengers. 12

26 The model assumes that the remaining revenue change is as a result of increases in patronage. This is a slight over-estimate. Where patronage has increased because of actions taken by councils or operators to increase demand at the same time as a price change, the changes in revenues will include an element that is the changed price paid by these additional passengers; part of this is a price effect. However, calculating this adds a considerable level of complexity to the analysis which is unlikely to be worthwhile. The impacts are shown in Table 5, including some of the intermediate steps based on CPI of 2.5% and a price elasticity of demand of The other items are the same as under the previous models. The sharing amount takes no account of indexation and is the same under gross and net approaches. Table 5 Sharing under GW Model Period Item Case A Case B Case C Case D Assumptions Revenue change $32,480 $16,160 -$160 -$160 Real change in fares -0.5% -0.5% -0.5% -0.5% ΔP f 0.2% 0.2% 0.2% 0.2% Price element of fare change $8,796 $8,796 $8,796 $8,796 Gross Share amount $23,684 $7,364 -$8,956 -$8,956 Indexation Revenue to council $20,638 $12,478 $4,318 $4,318 Revenue to operator $11,842 $3,682 -$4,478 -$4,478 Net Share amount $23,684 $7,364 -$8,956 -$8,956 Indexation Revenue to council $20,638 $12,478 $4,318 $4,318 Revenue to operator $11,842 $3,682 -$4,478 -$4, Incentives for Patronage Increase Any patronage increase increases the shared amount and provides rewards to operators and the council. Likewise any reduction in patronage is penalised. The incentive is the same as under the NZTA1 and AT models. However, the GW model isolates revenue increases from changes in price and operators are not rewarded for such fare changes. Fare-change related revenue goes to the council. 5.3 Incentives for Efficient Fares The council receives revenue associated with fare changes whereas those associated with patronage increases are shared with the operator. If the council wished to maximise subsidy reduction as an objective it would have the incentive to increase fare levels. This model may operate best with strict guidelines on fare changes, ie limiting them to inflationary adjustments. 5.4 Impacts on Subsidy Marginal Subsidy The marginal subsidy is the same as in the other models. Each increase in patronage (that is not directly in response to the introduction of the sharing model) results in a subsidy equal to the revenue earned per passenger times the sharing amount, ie $