Inadequacies in costing systems raise an important question: on what basis should the government decide how much to spend on operating its departments? The old practice was to base these decisions solely on the cost of inputs, but the New Zealand reforms had good cause for rejecting this method. If, as explained earlier, the price paid by the government equals the cost of inputs, managers have incentives to be inefficient budget maximisers by consuming more inputs. In shifting to output budgeting, some reformers made the heroic assumption that the government could set the price it pays without basing its decision on the cost of inputs. This assumption may be valid where the provision of outputs is contestable, and budget decisions can therefore be made by paying the lowest price bid by a reliable supplier. When outputs are contestable, the government has no more need to know the structure or cost of inputs than a private buyer has in a market exchange. Even when it purchases the contested outputs from its own departments, government can be a demanding buyer and get the best price consistent with its specification of quantity and quality.
The original Public Finance Act (PFA) 1989 distinguished three modes of appropriation: inputs, cost, and price. Mode A (inputs) appropriations were "for the acquisition of goods and services of a non-capital nature relating to a specified class of outputs or a programme"; Mode B (cost) appropriations are "for the costs to be incurred in the supply of a specified class of outputs"; Mode C (price) appropriations are "for acquiring a specified class of outputs required by the Crown."
Mode A was a transitional form of appropriation; it was discontinued by the end of the 1990/91 financial year. The PFA contemplated that as departments improved their accounting systems and as an increasing proportion of outputs became contestable, many output classes would be shifted to Mode C appropriations. This hasn't happened. No output classes are appropriated on a Mode C basis. The only types of output class appropriation now in use are Mode B and Mode B net.
Most of the outputs currently purchased by government are not contested. Perhaps more should be, but the plain fact is that they are not. In these cases, the purchaser (whether it is the government or a buyer internally contracting for outputs in a vertically integrated firm) needs to know how the supplier has priced the inputs. The time-honoured (but reform-dishonoured) practice has been to calculate the cost of authorised inputs. In practice, the Treasury and departments resort to this practice whenever they lack sufficiently robust costing systems to establish a price on a basis that is independent of input costs. In reviewing departmental bids, Treasury Vote analysts explicitly consider the cost of operations. This method of budget preparation offers some protection against getting the price unduly low. There is nothing wrong in these circumstances with Vote analysts' examining the amounts estimated to be spent on key inputs such as personnel, training, information systems, and so on. Only in this way can government satisfy itself that the amounts to be spent are approximately right.
More consideration is given to line items in preparing and reviewing budgets than is commonly thought to be the case. Certain outputs, such as policy advice, are budgeted in input terms, and managers indicated in interviews that their departmental budgets often are examined by Treasury Vote analysts in these terms.
However, using input costs as building blocks for compiling the budget is not an optimal practice. Although there is no inconsistency in giving managers broad spending discretion while still using inputs as cost elements for budget decisions, this practice is undesirable because it opens the door to reestablishing central budget control of the details of expenditure and because it offers little incentive to improve efficiency in the provision of public services. Moreover, focusing on inputs diverts attention from outputs which should be the all-important driver of budget decisions.
As necessary as it may be that Treasury and departments take account of input costs, there is some risk that this approach may undermine the system of output budgets on which the new accountability regime is predicated. Care must be taken, therefore, that analysis of input costs not mean a return to line item or input budgeting. I can suggest three ways of guarding against this relapse. First, Treasury and departments can develop input cost models without delving into all the items of expenditure. Second, Treasury should seek to expand the use of benchmark prices for public services. Finally, Treasury should invest in cost accounting and analysis systems that give it a stronger assurance that the price paid is right and that make it feasible to pilot test variable budgeting in selected output classes.
The first approach would be to develop cost models that subsume the various inputs required to produce outputs. For example, Treasury and departments might develop input cost models that estimate the total cost attributable to each full-time equivalent employee or square metre of office space. The cost model would encompass numerous line items (travel, wages and benefits, training, information services, etc.). With a well-developed input cost model, the government can obtain reasonable assurance that training needs are being met, maintenance is accorded proper priority, and information technology is advancing. Ideally, these cost factors would be apportioned among activities by means of an activity-based cost system or some other accepted cost allocation method. Government would then be in a position to intelligently decide on the amount to be spent for each activity without controlling each of the inputs. Second, Treasury should actively develop and apply benchmark pricing in public expenditure decisions. Benchmark prices have two advantages: they generate pressure for efficiency and they provide confidence that if operations are well managed, the price will be right. Benchmarking would be facilitated by increasing competitive tendering where appropriate for public services. Finally Treasury should break new ground in the pricing of outputs. Budgeting on the basis of inputs should be regarded as, at best, an interim measure to be used until departments acquire competence to budget on the basis of output prices.
Moving from input to output prices as envisaged in Mode C appropriations would require major improvements in cost accounting, allocation, and analysis. The accounting and financial management innovations were introduced to comply with external reporting requirements; the next frontier in New Zealand financial management will be to devise systems and practices for internal management needs. This will be a tougher challenge, not only because cost accounting systems are as yet undeveloped, but also because these systems will have to be tailored to the needs of each department. While certain principles and practices can be standardized, no single size fits the conditions faced by all departments. A few departments already have made laudable progress, and some good practices were identified in the Coopers & Lybrand report. This is not the place to discuss the various approaches and technical issues in cost accounting and analysis, but it would be appropriate to set out some of the steps that might be taken. Significantly, these steps generally correspond to the pioneering work underway in Treasury and Customs in pricing outputs.
(1) An early step should be the desegregation of outputs into standard, measurable units to which cost data would be attributed. Although output classes may be suitable categories for compiling the Estimates and maintaining accountability, they are too broad and variegated for analysing costs. Arguably, the main categories should be the activities which drive costs. To be useful, these have to be more finely decomposed than is currently the practice in constructing the output classes.
(2) For outputs that can be broken down into standard units, such as processing superannuation claims or clearing international passengers arriving in New Zealand, benchmark prices based on cost analysis, comparison with similar operations, or best practices should be more widely developed. These prices would not be directly linked to input costs.
(3) For outputs that cannot be standardised, strong effort should be made to establish benchmark prices based on analysis on the cost of efficiently providing the service. Doing so would require full cost attribution and close examination of work processes to assess the impact of changes in the process on cost.
(4) Once price is set independently of cost, managers have renewed interest to be efficient. However, there is risk that a budget based on output prices may not cover input costs. In the private sector, money-losing producers go out of business, but this option is not open to most government departments. It is essential, therefore, that periodic analysis be undertaken to measure price against cost and, when indicated, necessary adjustments should be made in the volume of and/or payment for services.
(5) The purchase agreement should be negotiated with the same benchmark or output prices as are applied in the budget. This would more closely integrate the purchase agreement into the decisional process, rather than just having it adapted to fit the department's budget constraint.
(6) With appropriate cost accounting and benchmark prices, it should be feasible to compile a true performance-based budget in which the volume of outputs is directly linked to the volume of financial resources. Ideally, this type of budget would indicate the increments in outputs that would be obtained by providing increments in resources.
(7) The final step would entail a great leap forward in budget practice - a move from fixed to variable budgets. Doing so would require reliable apportionment between fixed and variable costs, with the amount appropriated varying with changes in the volume of outputs. The appropriation would be for a fixed amount but the department would be permitted to spend an additional amount - up to a predetermined limit - if outputs exceeded the specified level. Mode C provides a basis for this type of appropriation. Inasmuch as variable budgets would break with generations of budgetary tradition, it would be appropriate to move in this direction cautiously and only to the extent that all of the following conditions were satisfied:
(a) the department has a robust and reliable cost accounting system;
(b) the outputs are demand driven, so that the department is paid more only because more is required of it, not because it is adept at drumming up more business;
(c) unit output prices have been agreed with Treasury; and
(d) the department demonstrates that it can cope with variable budgets when outputs fall below target. Variable budgeting should not be a method for allowing only upward adjustment.
These steps entail an ambitious programme of financial management improvement. The basic message is that financial data should become means of managing the cost of operations, not merely means of complying with external reporting requirements.