Turning now to the issue of pricing over time, time inconsistency arises when a regulator will have incentives ex post to reverse commitments it may have wanted to make ex ante.

For example, given that many of Telstra’s costs result from past investment that are now sunk, ex post the regulator may want to discount some of those costs, thereby reducing user charges. As Telstra’s obligations to provide service prevent it from ‘walking’, the temptation for a regulator to gain public standing and political legitimacy by being seen to be tough, and by forcing down prices, can be very strong. However, those gains come at a steep long-run cost: as the regulated firm comes to expect such time-inconsistent conduct by the regulator, it either refuses to put new assets at risk or demands a higher risk premium for doing so. As a result, prices must ultimately rise, the quality and range of service must suffer, or both.

Most regulators therefore try to avoid acting in ways that are time inconsistent. Indeed, the ACCC itself, in its approach to regulating electricity transmission, has stressed the risks of time inconsistency and sought to make credible commitments to avoid it.^[21] This has not, however, been its approach in telecommunications, as can be seen from a consideration of the manner in which the ACCC has determined depreciation and the PSTN capital charge.

The capital costs determined in a TSLRIC model are essentially a lump of costs, and those costs need to be spread over time. One approach to doing so is to ‘levelise’ the capital charge; that is, set it such that its value is equal in each year of the asset’s life.^[22] Adopting this approach, the annual capital charge would be calculated as:

(1)

where:

The issue of choosing between these approaches to determining the stream of costs first arose in the context of Telstra’s PSTN Undertaking for 1997/98. In assessing that Undertaking, the ACCC commissioned NERA to estimate the TSLRIC of PSTN access.

NERA’s view, which was clearly expressed in its report to the ACCC (NERA 1999), was that the depreciation profile used in the calculation of TSLRIC must reflect economic depreciation.^[23] If the depreciation profile that is actually used fails to mirror the economic depreciation profile, this will lead to a failure to recover the cost of investment over time.

NERA expressly rejected the use of an annuity approach to depreciation, stating that it is even less appropriate than straight-line depreciation because a constant annualised capital cost (depreciation plus cost of capital) means that depreciation increases each year; that is, it is actually back-loaded. While it is possible to tilt the annuity to allow for price and output declines, NERA argued that it requires a large tilt to achieve a declining depreciation profile over time.

Despite NERA’s findings, the ACCC requested NERA to calculate results based on an annuity function. For call conveyance costs, NERA found that using an annuity function had a significant effect on the results: charges for the Undertaking period were 20 per cent lower than the results obtained using proxy economic depreciation profiles (NERA 1999: 63).

In its Final Report on the Assessment of Telstra’s Undertaking (ACCC 1999: 60), the ACCC failed to rely on any of the NERA results that used economic depreciation profiles, instead relying solely on the annuity-based results, which were undertaken only as a sensitivity analysis by NERA and only at the suggestion of the ACCC (see NERA 1999: 63, footnotes 48 and 49).

The ACCC has, since that first Undertaking assessment, continued to rely on the tilted annuity for estimating the TSLRIC of both PSTN OTA and ULLS. In the case of ULLS, the effect of the tilted annuity formula is to steeply back-load the time profile of cost recovery. This is inconsistent with the profile of economic depreciation, as it fails to take into account other factors that impact on the value of the asset over time, such as wear and tear.^[24]

However, even putting the contrast with economic depreciation aside, the central problem is that the ACCC has never brought the deferred costs to account. Rather, at each assessment it has brought the counter back to zero, thus setting prices in a sequence of regulatory redeterminations ‘as if’ the deferred costs could be simply written off.

The consequences of this approach can be seen from Figure 1, which compares actual ULLS charges (displayed as a charge per ULLS line in nominal dollars on the Y-axis), as determined by the ACCC, with the charges that would have prevailed had the ACCC respected the time profile of cost recovery as implied by its initial cost modelling.

Sources: Calculated from ACCC (2002b), ACCC (2003), ACCC (2005b), ACCC (2007a).

Specifically, the line in Figure 1 provides an indication of what ULLS prices would have been had a time-consistent tilted annuity been applied to the Commission’s initial pricing decision on ULLS. The points traced out by that line are the sequence of charges required for the ‘fair bargain’ to be paid out. The line slopes up as the ACCC’s decision had deferred costs to future years, through the steep back-loading of the depreciation charge.

However, what the ACCC has done is not to set prices on the basis of that ‘fair bargain’. Rather, in part by constantly restarting the clock, the ACCC has reduced its estimate of the TSLRIC of ULLS in Band 2 areas from $35/service/month to just $13.90/service/month (see bars in Figure 1).

What is striking is not only the extent of the overall reduction (and hence of the departure from the pattern of prices implied by the tilted annuity) but also its pattern. In effect, as estimated by the ACCC, costs largely decrease, often sharply, but there are only exceptionally moderate rises. However, if depreciation in a forward-looking cost model is set on an actuarially fair basis, the changes in costs at re-estimation should follow a normal distribution (so long as cost shocks are independent), as it should be as likely that the initial estimate of depreciation (which reflects the anticipated change in asset values) will be an underestimate as it is that it will be an overestimate. This contrasts with the pattern of the successive ACCC estimates.^[25]

The author’s explanation of this outcome is that what the ACCC has done is to ignore the price path implied by its back-loaded depreciation profile, while successively reducing the estimate of the levelised cost. These reductions are equivalent to writing off the amount that (through the back-loading of the annuity) had implicitly been deferred to each period from previous periods.

The extent of the resulting cost-recovery shortfall is large in absolute terms. It can be quantified in terms of the loss that would be borne by the hypothetical, wholesale-only, access provider, operating a continuously fully optimised network. That amount, taken as a loss compounded from 2002 to the present at an interest rate of 10 per cent, approaches $22 billion.