Taxi industry inquiry

The inquiry’s modelling predicts around 100 new taxis on this set of assumptions. The inquiry has also conducted some sensitivity analysis on these results and has found that the likely entry is particularly sensitive to:

The initial payment to licence holders (assignment value); and

The size of the increase in driver remuneration that is assumed to flow through into increased driver payments.

On the initial payments, the inquiry considers that the evidence suggests that the appropriate payment to use is between $27,000 and $30,000.

The driver remuneration assumption largely depends on the extent to which owner/drivers are prepared to accept lower remuneration than that implied by the 55/45 split. If, for example, the owner engages a driver for one third or one half of the time, then the increase in cost caused by higher driver payments will only be 5 per cent or 3.33 per cent. This could cause more significant entry to occur.

The inquiry concludes from this modelling that actual entry as a result of the reforms proposed is likely to be between zero and 400 new taxis in the short term if demand, fares and costs are held constant.

In the Draft Report, the inquiry noted that restrictive licensing is very likely to cause significant losses in community welfare. The source of these losses is that prices for taxi services reflect not only the resource costs of producing taxi services – the costs of the vehicle and the driver’s time – but also payments to licence owners. The existence of high licence values suggests that there are some potential welfare gains for the community if these profits could be removed and lower prices charged. Although licence owners will lose from this re-arrangement, in general, consumers will gain more from lower prices than licence holders will lose.

Economists and review authorities have often attempted to measure the size of the welfare loss and/or potential welfare gain caused by restrictive licensing. This involves estimating the shape of demand and supply curves for taxi services and comparing the actual price and quantity combination under restrictive entry with an alternative combination estimating a different price and quantity combination.²⁶¹ At this lower price, there is no new entry by taxis because all the expected profits from entering have been removed.

A weakness with these estimates is that they assume that prices can merely be reduced to a competitive price (by regulation) with no effect on demand. But evidence from actual taxi markets suggests that the reality is not this simple. Supply and demand are likely to be interdependent: more supply increases vacancy rates, which lessens waiting times and creates more demand. The strength of these effects can be very important. Lowering prices, which would increase demand, also reduces the supply available to other potential consumers and can lengthen waiting times. This causes the demand curve for taxi services to shift inwards (demand is lower at each price level), meaning that less will be consumed at the prevailing price.²⁶²

As discussed further in the Draft Report, it is possible that the best approach to measuring welfare costs would be to consider a mix of both fare reductions and improvements in availability. These considerations were illustrated graphically in Figure 10.3 of the Draft Report.

These diagrams, replicated in Figure 9, illustrated the potential benefits for consumers from two different policy changes, starting from a position of fares at (f₀) and quantity of trips at Q₀: (a) reducing fares by removing assignment fees from fares and (b) keeping fares at current levels, but allowing free entry so that no above-normal profit would be earned from operating a taxi.

Which approach is likely to ‘better’ measure the costs of the restrictions depends on a number of market features. In an environment where licences have been strictly controlled, the benefits from lowering fares without increasing the number of taxis may be limited, because the increase in waiting times that results also causes large consumer losses. In these circumstances, it would be better to increase availability. Conversely, if the problem is not one of availability but high prices built into fares, then the losses could be better stemmed by focusing on fare reductions.

In these three figures, the initial situation is shown first and then the potential costs of restrictions (or gains from reform) are illustrated in two different ways. The first assumes that fares are reduced, assuming that waiting time can be held constant (fares fall from f₀ to f₁). The gains to the community are represented by the shaded triangle (the transfers between licence owners and consumers are not a source of welfare gain or loss). The second diagram illustrates a situation where fares remain at their present level, but sufficient new entry is allowed to reduce the profits from operating a taxi to zero (costs increase from c₀ to f₀). This reduces waiting times and increases the demand for taxi trips. The two policies can be compared with the ‘base case’ of no change.

In the example shown, drawn with a linear or straight line demand curve, the gains to the community from the increased number of trips and the value of those trips will significantly outweigh those from lowering fares without allowing new entry. This occurs even though costs to taxi operators under the greater availability scenario have increased compared to the base case, and ‘A’ appears as a welfare loss.

Figure 9 Identifying the welfare costs of entry restrictions - two alternative methodologies

As noted, the simplest modelling calculation of welfare loss comes from eliminating the estimated rents and assessing the value of the increased supply to consumers against the losses to licence owners. A calculation using a demand elasticity of -1.02, consistent with the inquiry’s survey evidence, and assuming no waiting time effects, suggests these costs (or ‘deadweight losses’) are in the order of $11 million per year. This is based on a linear demand curve, a fall in price of 19 per cent, an increase in quantity of 19 per cent and a calculation of (change in quantity)*(change in fares)*(0.5).

The inquiry’s further modelling of the different scenarios also takes into account waiting time effects and also some economies of scale effects. Waiting time effects, as described in Figure 9, result in shifts in the demand curve. Cost or scale effects are also relevant as the low level of fleet utilisation suggests that increases in passenger kilometres could be achieved without proportionate increases in costs (that is, there are some fixed costs that will not increase as quantity increases), meaning this is an additional source of efficiency gain.

The inquiry’s modelling suggests the following:

The restrictive entry policies impose significant costs that could be ameliorated by either reducing fares or allowing more entry

As the own-price elasticity of demand is currently estimated to be relatively high, the gains from reducing fares tend to outweigh those from greater availability, even to the point where allowing open entry but not reducing fares would result in a welfare loss (A would be less than B, in Figure 9)

The best outcome would combine fare reductions with more entry.

In Table 12, the inquiry provides the output from the base case and from three different modelling scenarios:

Reducing fares by the full amount of the existing economic rents, but fixing the quantity of taxis

Fixing fares, but allowing the quantity of taxis to vary

The inquiry shared some of its modelling, including a description of how the model works and some of its key outputs, with industry representatives TISV and the VTA after the release of the Draft Report. Each organisation engaged professional advisers who scrutinised the inquiry’s modelling results and produced their own modelling using the input data collected by inquiry.

During the course of the inquiry, industry groups made a number of submissions supporting approaches to licence release that are specifically related to measurable increases in demand and/or other criteria related to demand (such as increases in waiting times). The VTA supplied some preliminary modelling work it had commissioned from PwC, which provided some detail about its assessment of how the inquiry’s and the industry’s preferred approaches might work over periods of five and 10 years. The industry’s preferred approach is largely based on releasing licences depending upon changes in demand and possibly other performance triggers.

The inquiry considers that for a model to be able to predict behaviour and market outcomes, it must be capable of producing results that are consistent with rational behaviour. In this light, the inquiry suggests that modelling that predicts that taxi operators will consistently earn negative margins or margins below a ‘normal’ economic or commercial return are not likely to be useful for predictive purposes. Rather, it represents a weakness in the way the model is constructed.

Both modelling exercises provided to the inquiry by industry advisers were not consistent with this expectation of rational behaviour. Rather, assumptions were made that assumed a path of entry regardless of the underlying state of the industry. It was clear from reviewing both models that entry of as many taxis as modelled would produce a negative margin for operators (that is, not just below a normal commercial return, but a negative return). Why operators would continue to take licences up when they could not hope to make a normal return on their investment is not explained and the inquiry rejects this modelling as being a misrepresentation of the likely outcome.

The inquiry considers it to be fundamental that modelling of the inquiry’s proposals must produce outcomes consistent with operators making a sufficient return to be able to pay licence holders (including the Government where this is a new licence) $20,000 per year. The inquiry’s modelling starts with this fundamental feature and its model endogenously (that is, within the model) calculates the number of taxis that would be consistent with that level of profit. Therefore, it is relatively straightforward to measure the impact of the inquiry's preferred licensing approach over time. Changes to demand, fares and costs can also be made, with the number of taxis determined by payments to licence owners  such as $20,000. This modelling also incorporates the effect of changes in fares and in waiting times on demand (via a feedback loop) and in turn takes account of the impact of demand on the number of taxis supportable.

In contrast, modelling the likely effects of the proposals presented by industry is not so straightforward. They impose assumptions on growth in taxi licences based on estimated demand growth, rather than having clear relationships that can explain entry decisions by taxi operators. Therefore, the inquiry has considered two means by which the VTA (and other industry) proposals of releasing licences in accordance with observed changes in drivers of taxi demand might practically work:

By allowing the release of a suitable number of licences such that waiting times do not increase. This could be a reasonable approximation if waiting times are a relatively large component of the licence release index measure. Increases in demand that lengthen waiting times will then cause the release of more licences. (This is described as ‘Industry proposal 1’ in Table 14); or

By allowing the release of the number of licences that holds payments to licence holders at a capped level. That is, if it is assumed that assignment payments are a significant component of the licence release index measure, then increases in demand that feed through to higher assignment prices will feed through to licence increases, which will in turn reduce the assignment prices. The inquiry further assumes that ‘capping’ here means holding the current payment to licence holders of around $29,000 to $30,000, increased by 0.5 per cent per year less than fare growth. (This is described as ‘Industry proposal 2’ in Table 14).

The inquiry has run a number of simulations to determine the effects of these different policies. The assumptions run on the inquiry’s model are summarised as follows.

The inquiry’s modelling does not propose a reduction in fares, even though the new $20,000 licences are a reduction from the assignment value that is currently incorporated in the ESC’s fare model from 2008 (around $23,000) and well below current market assignment rates. This is because there is an offsetting cost increase for operators – higher driver payments.

Under the VTA approaches, the inquiry considers there is some immediate risk that fares will need to rise to incorporate a higher assignment value. This is because this higher assignment value must be factored into fares for an operator that assigns to just make a normal profit, and the current implied value is below the value last measured in 2008. Even if this is ignored, it will be necessary to factor in higher fare growth under the VTA approaches. This is because assignment costs (payments to licence holders) will continue to grow under these approaches. In contrast, under the inquiry’s approach, there will be no increase and this means that only 80 per cent of the cost base can rise over time. Payments to licence holders currently represent around 20 per cent of operator cost, although this will decline over time as other costs rise and assignment prices remain fixed.

The assumptions used in the modelling are outlined in Table 13. There are two assumptions of note. The first is the slower annual growth in fares under the inquiry’s approach, which reflects the inquiry’s capping of assignments (via the issue of new licences) at a fixed nominal value (i.e. zero growth). The second is the lower value for elasticity of demand than found in the inquiry’s survey work, as this base value of -1.02 is based on static comparisons of prices (i.e. assuming the price of all other goods and services remains constant). Over longer time periods, a lower value should be used, more consistent with changes in the real price of taxi fares.