Chief executives and their departments are accountable for supplying the outputs specified in the Estimates, purchase agreements, and other authoritative documents. Departments supply many thousands of outputs to the government each year, far too many to use individual outputs as the basis for appropriating funds or, in my view, maintaining accountability. It would not be in the spirit of the New Zealand reforms to target only a few key outputs, as some countries have done.
The solution prescribed by the Public Finance Act is for appropriations to be made by output classes. An output class is defined by the act as any "grouping of similar outputs." Much effort has gone into defining and refining the output classes, and while some adjustments still are made each year, the structure has stabilised. Pursuant to 1992 amendments to the Public Finance Act, considerable effort has been given to define non-departmental output classes, and they now generally are specified in the same form as departmental output classes. Most of the output classes represent the activities conducted by departments and other entities in serving government. They describe the work performed for government, not the outputs produced. Many departments set aside an output class for policy advice and another for major administrative responsibilities such as managing contracts.
The output classes are categories of convenience; they are intermediate categories, more specified than Votes but more aggregated than discrete outputs. The Estimates and several other authoritative documents publish detailed output data within the various output classes. The output classes are important because appropriations are made to them, and departments manage and account for financial resources in terms of these classes.
Efforts to make the output classes informative categories has led to a proliferation of appropriation heads. The Logan Review (Review of the State Sector Reforms; 1991) reported that the shift to output class appropriations had increased the number of separate appropriations from 56 in 1988/89 to 774 just three years later. Logan recommended that the number of separate appropriations be reduced but that the Estimates and annual reports continue to publish detailed output information. But judging from the output class data presented in Appendix III, no progress has been made in this direction. The Estimates now have almost the same number of appropriations as they had when the Logan Review was conducted.
Three factors account for the large number of appropriations. One is the need to distinguish between departmental and nondepartmental outputs; another is the practice of itemising the various capital contributions, grants, and other payments; and the third is the effort to ensure that funds are spent on particular activities (such as managing contracts) and are not pooled with other administrative expenses. Regardless of the reasons, however, the fact remains that many of the output classes hold very small amounts of money. Moreover, there is a truly extraordinary disproportion between the largest and smallest output classes, even in the same Vote. Although each output class must meet a materiality test, the test does not seem to be strictly applied.
The large number of small output classes erodes managerial flexibility and complicates the task of allocating overhead and other indirect costs among the various classes. It also fosters the same kind of compliance mentality and budgetary gamesmanship that flourished under input budgeting. When managers are restrained (arbitrarily, in their view) from spending funds as they deem appropriate, they learn how to outwit the controls.
The output classes provide the decision and information structure for the accountability system. The same output class structure is used in the Estimates, Departmental Forecast Reports (DFRs), annual reports, and other documents. Some complications arise because the amounts recorded for appropriations in the Estimates are inclusive of GST, while those in documents generated by the departments themselves exclude GST, in accordance with generally accepted accounting practice. It is not difficult, however, to reconcile the various documents or to trace an output class from the Estimates, through its DFR, to the annual report. The format is accessible, transparent, and easy to use. One does have to navigate through thousands of pages to assemble the many pieces of the accountability puzzle, but it is an effort rewarded by the ease with which performance outturns can be compared to targets.
Output Specification in the Estimates.
Every department specifies output targets in at least three documents: the Estimates, the annual purchase agreement, and DFRs. Some departments also specify their outputs in corporate and strategic plans and in internal documents. As noted above, the various documents use the same output class structure.
Almost five hundred output classes are described in the Estimates submitted to Parliament each year. Specification in the Estimates on an output class basis proved to be a difficult task. Unlike inputs, which are standardised across departments, most outputs are unique to the department producing them. The necessary steps included identifying the outputs, organising them into output classes, developing qualitative and quantitative indicators of performance, and succinctly describing them in a manner that conveys useful information without overloading the Estimates. These steps entailed considerable negotiation between each department, the Minister, and central agencies, along with sometimes contentious wrangling over the best way of describing what the department was producing. It was also necessary to develop information on outputs supplied by third parties. The format of the Estimates has been changed several times during the 1990s, but with the introduction of the DFRs (beginning with the 1995/96 financial year) Treasury officials believe that a stable format has been attained which conveys a great deal of useful information. They anticipate only marginal adjustments in the future, no major overhaul, subject to accommodating legislative amendments due to changes in the preferences of the principal users - Members of Parliament.
It takes but a few minutes to learn how to work through the Estimates, because the same schedules and format are used for every Vote. As a result, a disproportionate amount of space is devoted to relatively minuscule amounts of money. The $4.8 million budgeted for Sports, Fitness, and Leisure Vote in 1996/97 takes up eleven pages in the Estimates; the $355 million budgeted for Corrections has sixteen pages. In general, the amount of output information presented in the Estimates increases with the amount of money allocated to the Vote.
I noted in chapter 3 that less information is provided on the output classes supplied by Crown entities than on departmental output classes. Nondepartmental entities are the end-users of two-thirds of the money appropriated for output classes. One should expect, therefore, that sufficient information is provided on nondepartmental outputs to permit monitoring of the use of these funds.
Much effort has been expended by departments in developing detailed output information. Most adhere to Treasury guidelines for specifying outputs in terms of quantity, quality, timeliness, and cost. Inevitably, however, the quality of the information is uneven, but almost all departments have made enormous progress in specifying outputs. A comparison of the 1995/96 Estimates with those tabled several years earlier indicates significant improvement in the coverage and usefulness of the information. Moreover, the recent Estimates give evidence of considerable care and effort in identifying the outputs to be produced in the next year. The information appears to be fresh and up-to-date; it is not simply a repeat of the previous year's entries.
Nevertheless, there is ample room for improving the output class information in the Estimates. I believe that the usefulness of the information would be enhanced by using a tabular format (where appropriate) rather than or in addition to narrative descriptions, by comparing the outputs specified for the next year with those budgeted or produced for the current or past year, by presenting some of the information by output subclasses, and by linking the volume and quality of the outputs with the amounts to be appropriated. Part A2 in the Estimates highlights some of the initiatives or changes to be made in the next year, but I believe Parliament and the interested public would benefit from additional information.
Without trend or comparative data, it is hard to interpret or analyse the output information. If all that is wanted is specification of outputs, so that actual performance can be compared to targets, the present arrangement is satisfactory. But if one wants to come to an informed judgment on the amount of money that should be spent, then information on the previous financial year(s) would be useful. Moreover, efforts should be made to provide cost data on major output subclasses, so that one can analyse what the government intends to buy with the appropriated funds.
The lack of trend data on outputs in the Estimates has been a contributing factor in the substantial informational demands made by Parliament each year. The Finance and Expenditure Committee (FEC) issues a "standard estimates questionnaire" each year, and select committees examining individual Votes address "supplementary questions" to particular departments. The FEC has revised the standard questionnaire for 1996/97 to focus on Ministerial responsibilities rather than on departmental operations, thereby reinforcing the concept that the Estimates reflect the Minister's perspective. The new approach reflects each Minister's accountability to Parliament for appropriation requests. In previous years, the standard questionnaire requested each department to report on its plans to improve operational efficiency, financial management, and other aspects of the department's management. It also instructed departments to explain year-to-year changes in appropriations due to (among other things) changes in the volume of activity and outputs. In contrast to the Estimates, which provide only one year's output data, the standard questionnaire requested trend data on each department's output classes. The new standard questionnaire asks Ministers to discuss critical issues faced by the Government in preparing the Estimates, and significant changes in appropriation requests, including new policies and significant increases and decreases in the amounts appropriated. It also requests Ministers to indicate the key mechanisms used to evaluate the impact of outputs in achieving the stated outcomes.
The supplementary questions generally inquire concerning functions or services to be introduced or curtailed during the year. They also request a considerable amount of input information (for example, on travel, consultancy and personnel expenses), indicating that despite the reforms, the Parliament has not lost interest in this type of information.
I do not believe that inserting cost and trend data on outputs into the Estimates would make this document too unwieldy. Considerable economy in space can be achieved by rearranging some of the information currently provided in the Estimates. The key issue is not the number of pages, but whether the Estimates should enable Parliament and other users to make informed judgments on the amounts to be appropriated, or whether they should also provide performance targets against which actual results can be compared. I believe the Estimates can efficiently serve both roles.
Departmental Forecast Reports (DFRs).
These reports are the most recent additions to the accountability regime. They were first prepared at about the time interviews for this study were conducted with chief executives and senior managers. Undoubtedly, the attitudes expressed in these interviews were coloured by the initial experiences that departments had in complying with this new requirement. Many managers saw the DFRs as make work that duplicated information readily available in other documents and as an indication of insensitivity to the burdens placed on departments. They particularly resented the government's failure to compensate them for the additional work required to prepare these reports. Small departments felt the pinch the most.
Against these complaints, the reasons that led to introduction of the DFRs merit mention. These reports replaced the departmental budgets previously included in the annual Estimates. They enable the government to sharpen the distinction between the resources appropriated to Ministers as purchasers and those provided departments as producers. They give each department an opportunity to describe what it plans to do and spend, in its own words and numbers, though these have to be consistent with the Estimates. It was anticipated that the DFRs would lessen Parliament's demand for information via the standard and supplementary questionnaires. Finally, the DFRs would provide benchmarks for enforcing accountability; the annual reports compare outturns to the results forecast in the DFRs.
The DFRs add up to more than 1,500 pages. Although the size varies greatly among departments, all are required to include the provisions prescribed by the Public Finance Act. Section 34A of the Act provides that each department should forecast its financial position at the beginning and end of the year; its revenues and expenses for the year; projected cash flows; the performance of each output class; and financial performance. Each DFR also contains statements of responsibility, significant assumptions and accounting policies, and certain other information. Treasury has provided an indicative format which is followed by most departments.
These guidelines urge that the output classes be specified "from a departmental/provider perspective . . . that differs from the form and purpose of departmental output-class specification in the Estimates." The Treasury guidelines also stress the link of the DFRs and annual reports. The DFR's statements of objectives must be included in the annual reports as a reference point for auditing the statement of service performance.
The DFR is a free-standing document that is intended to be used independently of the Estimates, but should be consistent with them. A Treasury review of the first crop of DFRs concluded that some departments had merely repeated the output information published in the Estimates, but most had made some effort to describe the outputs from their perspective. My own sampling of the DFRs indicates substantial overlap with the Estimates, though the former tend to be more detailed. Like the Estimates, the DFRs concentrate on the year ahead and do not provide year-by-year trends on the mix, volume, or cost of outputs. Standing alone, the DFRs appear to be well designed and useful documents; read in tandem with the Estimates, one may question whether the additional work might be reduced by cutting down on duplication.
In reviewing the DFR output information in the light of the parallel information contained in the Estimates, I am not sure what is meant by the difference between the Minister's and the department's perspective. In terms of the outputs themselves, there should be no difference between the two. The department sells what the Minister buys. In fact, the two sides must agree on common outputs in the purchase agreement. It is of little consequence that the Estimates read, "This class of outputs involves the purchase . . .," while the DFR states, "This class of outputs involves the provision . . . ." Many DFRs provide more detail than is published in the Estimates, including information on the volume and cost of outputs, but there is no reason (other than space constraints) why some of this detail cannot be inserted in the Estimates.
There is one important difference between the Minister's purchase and the department's provider perspective. As purchaser, the Minister should be indifferent as to how the outputs are produced, but this must be a matter of substantial concern to the department. In late 1995, the chair of Parliament's Finance and Expenditure Committee urged departments to enhance ownership-type information in the DFR's, such as information on management developments and the capacity to produce. I am not sure, however, that the DFR is the appropriate vehicle for discussing the department's capacity to efficiently produce the outputs or broader ownership issues.
These agreements were not part of the original management design. They were introduced (beginning with the 1993/94 financial year) pursuant to a recommendation by the 1992 interdepartmental working party on output definition that chief executives specify the outputs to be supplied in agreements with Vote Ministers. The purchase agreements gained quick acceptance, and only three years after being introduced, they have secured a niche in New Zealand public management.
Although these agreements are welcomed, practice has not been standardised. As noted, some Ministers sign agreements after protracted negotiations aided by purchase advisers; others sign with little or no discussion. Some Ministers use the agreement as a vehicle for reviewing all the outputs supplied by the departments; others regard it as an opportunity to impress their priorities on the department. The most common pattern is one in which the chief executive drafts the agreements, most of the outputs specified in the signed document are those proposed by the chief executive, but the Minister inserts some matters that she/he is concerned about. Seen in this light, the key test is not whether Ministers influence the entire agreement - they rarely do - but whether they succeed in getting the department to accord priority to those outputs they care about. Most Ministers regard their influence on specific priorities as sufficient.
Purchase agreements are negotiated during the period that the budget is being prepared. Thus there are parallel discussions between the Minister and chief executive on the outputs to be purchased and between the Minister and the Minister of Finance and Treasury on budget expenditure decisions. There is inevitable spillover from one set of negotiations to the other, if only because the purchase agreement cannot be finalised until the Estimates have been settled. Early in the purchase agreement process there was consideration whether the parallel negotiations should be formally linked. Treasury met resistance from chief executives when it requested copies of their draft 1994/95 purchase agreements while the 1994 budget was being formulated. Treasury considered that even in draft form these documents would assist Vote analysts in reviewing departmental budgets; chief executives saw the request as an inappropriate intrusion by Treasury into their relationship with Ministers. Facing strong opposition, Treasury retreated, but it nevertheless remains an issue whether the purchase agreement should be wholly a bilateral matter or of broader interest to the government. Departments do provide purchase agreements, sometimes in draft when the agreement has not been finalised, to Parliamentary select committees for their Estimates examinations. I believe they should also be provided to Treasury. These agreements should enrich the analysis of whether departments have sufficient resources to produce projected outputs.
An important ongoing issue pertains to the level of detail in the agreements. Some consider the agreement a firm contract (subject to renegotiation during the year) on the services to be delivered; others view it more as an opportunity for periodic discussion between the Minister and chief executive on what the department is planning to do. The more the purchase agreement is perceived in contractual terms, the more likely that it will specify the outputs in great detail. In fact, some purchase agreements list just about every task the department expects to perform.
Although there may be considerable benefit in planning the work to be performed during the year and the tasks to be completed, I am sceptical about the value of trying to specify all work in advance. Departments have to be sufficiently supple and have sufficient slack to deal with the shocks and surprises that occur during the year. Overspecifying outputs runs the risk of generating the same "managing by the book" behaviour that characterised input budgeting. Particularly in policy advice and similar output classes, it makes little sense to specify every report or study to be undertaken; the list will almost certainly change as the year unfolds. Better, therefore, to emphasise those matters of priority to the Minister (as some purchase agreements do) and to leave some of the other details open ended. It is encouraging that an increasing number of purchase agreements describe the overall service to be provided to the Minister and do not itemise individual projects or studies.