l WSql Viewpoint Note No. 150 August 1998 Why Performance Contracts for State-Owned Enterprises Haven't Worked Marn' Sh/vid' When privatization is not feasible or palatable, some developing country governments seek to improve the performance of state enterprises by negotiating performance contracts with their managers. Many of these contracts have been put in place with World Bank assistance. Research shows that they rarely work. This Note summarizes the rationale for performance contracts and explores the reasons why they haven't worked. It concludes that since a well-designed and -enforced performance contract can be as politically costly as a well-designed privatization, performance contracts are not likely to be successful in countries that lack the political will to privatize. \Xritten contracts betwveen governments and state put In pI.ce most gi v)ernuentse xv ere tr inl- to enterprises have been wvidelV used in 'ooricl run their- state enterprises Without any forul (of flank projects since the first 13ank operationll sup- perforirimce evaluation A,, (nc architect (4 porting performance contracts, a 19- opera- performanice contracts notecl this \va likoe play - tion in Senega. A Bank survev of cdevelopina ing foi thall without rules. scorcehoairds. or re.- couintries fountdi 505 suLIC Conttracts in thiirt\--t-xN cre.s Performance contraicts sceemied a logi-al countries, and another 10B.000 in China is (if >1 )lution to this problem. since similar c ontracts tune 199-i (NWorld Bank 199;). Although these had hbeen sucCesstul in the private settoir performance contracts go yv cliffer-ent nances- c-ouitrit-p/Ia;, memoraindului of understanding, ioi (in. including the propo)rents of perfor- signaling system-thev slare cominillon featurees ma nce contracts, minimilizecl the prohlemis All are negotiated. written agreements between governments would face in designing suchi governments and the mi-anagers of state enter- contracts, however. Much has been written about prises that specify targets that malnagernent the problcms that principals (in this case. gov- pledges to a"chieve in a given time framile and ernrilients) face because they cannot aiccurately clefine how performance will be measured at measure the effort expended bv their- agents the end of a specified peioci. (manamgers) or sort it out fromil othie-r factors af- fecting performance. These agency prcoblemis are The case for performance contracts CompoundLe1d in the public secto-. \w here politi- cians have manv points of view ancld burealu- Despite a global wave of privatizations, state crats halve mainy cliffer-ent agendas. tndlelr such enterprises still account foir about 10 percent circumnstalnces it is halcrd to judge performance of gross dcomestic proCdLuct (GDP) in develo(p- and to motivate managers and holdl themil ac- ing countries. These enterprises are often the counta,ible for results. MoreoverL unlike private largest and most valuable or problem.atic firms, owvners politicians may not benefit fromi better wvith monopolies in mining, petroleum. infra- performance. and so may try to make manaigers structure, and heal-v industry. For these firmis serve objectives that cinflict with efficiency, such performance contracts have often seemed to as rewarding political supporters wvith jobs or make good sense. Before the contracts were suLbsidies. WN The World Bank Group . Finance, Private Sector, and Infrastructure Network Uklu Why Performance Contracts for State-Owned Enterprises Haven't Worked Proponents of performance contracts argue that in hundreds of state enterprises in China-found they can be written in ways that clarify mul- no evidence that performance contracts had im- tiple objectives and make it easier to judge per- proved efficiency. The first study analyzed the formance. For example, a contract could apply effects of contracts in monopoly enterprises (in weights to the multiple objectives, spell out water, electricity, telecommunications, and oil the obligations for which managers will be held and gas) in Ghana, India, the Republic of Ko- accountable, and specify rewards (such as rea, Mexico, the Philippines, and Senegal. It bonuses) and penalties (such as demotion or found no pattern of improvement associated firing). Even where a government seeks to with the performance contracts in productivity maximize social or political objectives, a per- or profitability trends (figure 1). formance contract can improve efficiency by setting appropriate targets. For example, a state The second study used a much larger sample in enterprise required to retain redundant work- manufacturing but in only one country, China. ers could still achieve contractual targets aimed The results showed that the increasing use of at improving quality. And an overstaffed firm performance contracts in China could not stem could still improve labor productivity by mak- the fall in productivity among state enterprises ing better use of plant and equipment. The fact (figure 2). More important, the study found no that managers operate under such constraints robust, positive association between perfor- could be taken into account by judging perfor- mance contracts and productivity. And a com- mance against past trends. parison of a sample of state enterprises that had signed performance contracts with a sample of The contrary evidence firms that had not found no significant differ- ence between the two groups. The logic of performance contracts is persua- sive, but the reality has been disappointing. Two Is it possible that performance contracts failed empirical studies-one analyzing the effect of to improve productivity because managers were such contracts on profitability and productivity told to maximize social benefits, such as increas- in twelve companies in six countries and the ing employment or developing backward re- other examining statistically the correlation be- gions? Although the studies did not measure tween performance contracts and productivity social benefits, the weights that contracts as- signed to productivity targets (two-thirds on FIGURE 1 PERFORMANCE CHANGES IN STATE ENTERPRISES average) and the stated goals of the parties to AFTER SIGNING PERFORMANCE CONTRACTS the contracts suggest that improving operating efficiency was the prime objective. Moreover, Enterprises most social and political goals imposed constant 12 costs on state enterprises during the period and 102Improved so should not have affected the trends being 10 ~ ~ ~ ~ ~ ~ ~ ~~~Imrvdmeasured. 8 Why not simply judge performance contracts on 6 whether the firms met the targets specified in the 4 contract? All the firms in the first sudy did achieve at least satisfactory ratings where some sort of 2 ,score was assigned, and all the contracts assign a a Cp d a high weight to economic goals. The problem is Return on assets Labor productivity Total factor productivity that many of the targets are soft or flawed mea- sures of economic performance. For example, Note: The data cover twelve state enterprises in Ghana, India, the Republic of Korea, Mexico, the 3 ecn ftesoefroeo h lcrct Philippines, and Senegal. 3 ecn ftesoefroeo h lcrct Source: Company data and World Dank estimates. companies (India's National Thermal Power Cor- FIGURE 2 CHINESE STATE ENTERPRISES UNDER PERFORMANCE CONTRACTS AND THEIR PRODUCTIVITY, 1980-89 Productivity Firms under performance contracts Index (1980 = 1.0) 700 2.75 poration) depended on the volumne of electricity it generated. The company achieved its target 600 M250 and received a score of excellent, yet its total 500 Productivity 2.25 factor productivity actually fell below precontract levels. Output went up, but inputs rose three times 400 2.00 faster. The target was flawed because it could be / 1.7 rmet by increasing inputs, even if efficiency de- 300 1.75 clined. The contracts have many such flaws, for 200 1.50 reasons explored below. 100 Firms 1.25 What is the problem? / 1.00 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 The theoiy of contracting suggests that to im- provTe performance, performance contracts must: Note:The data for productivity refer to the state enterprises with performance contracts each year. * Reduce the information advantage that man- agers enjoy over owners. * Motivate managers through rewards or pen- the second study, in China, found that the in- alties to achieve the contract's targets. centive (wage increases linked to profits) was * Convince managers that the government set too low to motivate improvements in most promises in the contract (such as to pay bo- of the firms and was aimed only at workers. nuses or impose penalties) are credible. Finally, governments' commitment to enforc- The performance contracts in the two studies ing the contracts and keeping their promises failed on all three counts. First, managers were was not credible. All the contracts lacked neu- able to use their information advantage to nego- tral, third-party enforcement mechanisms (the tiate targets that were either hard for outsiders state enterprises could not take the government to evaluate or easy for the firm to achieve. Per- to court, for example), and governments often formance is hard to evaluate, for example, when reneged on their promises. In Ghana, India, there are many targets (the contract for Korea's and Senegal, for example, the government did telecommunications company had forty) or not force public entities to pay their bills to when targets change frequently (a third of the the electricity companies. targets for Ghana's water company changed every year). Targets can also simply be soft; in There is evidence that a performance contract India, for example. negotiations dragged on so that overcomes the three contracting problems long that targets were set equal to ex post per- can improve efficiency. The study of China silllu- formance. The managers' information advantage lated what would have happened with a "good" was compounded by governments' failure to performance contract-one that addressed the give the bureaucrats responsible for negotiat- information, incentive, and commitment prob- ing the contracts and evaluating results the lems--and found that it would have had a sta- power, resources, and status they needed to face tistically significant and large positive effect, enterprise managers on a level playing field. boosting productivity growth rates by 10 per- Managers were thus able to negotiate targets cent. But only 2.2 percent of the firms in the that they could achieve withotut making addi- sample had "good" performance contracts. All tional efforts to improve productivity. the other performance contracts had either in- significant or negative effects on productivity. Second, the incentives provided under the con- tracts failed to motivate managers. The first study Why did so few performance contracts contain found that only two of the twelve contracts paid the provisions necessary for success? Perfor- a bonus or punished underachievenment. And mance contracting assumes that government Why Performance Contracts for State-Owned Enterprises Haven't Worked objectives can be maximized and performance guarantees for debts of state enterprises and improved by setting targets that take into ac- instructed banks to lend to them under the same count the constraints placed on managers. For criteria as for private enterprises. State enter- this to occur, politicians and bureaucrats must prises were required to pay a 10 percent re- state their objectives explicitly and agree to turn on assets as a dividend, and money losers weights that reflect their priorities, empower a were required to sell assets to pay their divi- supervisory body to translate these objectives dend. The government privatized almost all into monitorable targets negotiated with man- commercial and financial firms and most utili- agers, punish and reward managers on the basis ties, allowing it to concentrate its supervision of tieir perfoirmanice, and keep any proiimises orn relatively few firms (such as the water arid made in the contract. Few of these actions oc- sewerage companies). curred for the contracts studied. Do these fincdings mean that World Bank opera- Why would governments sign performance tions should not support or encourage perfor- contracts and then not try to make them work? mance contracts? Although the studies found few Some governments may have been motivated successful contracts, they did show that in those to pledge actions that were politically unreal- rare cases wvhere a performance contract is prop- istic because it enabled them to meet condi- erly written, it can improve efficiency. But they tions of a World Bank loan. Some governments also found that performance contracts can do may have underestimated the political costs of harm. If targets are set too low, managers might Viewpoint is an open forum intended to adhering to a performance contract, such as reduce their efforts to improve performance. And encourage dissemina- firing politically loyal but underperforming flawed targets can have perverse effects, as in tion of and debate mn managers, paying incentives that might raise a the case of India's electricitv companv. Since a best practices foran- manager's salary well above a minister's, shift- well-designed and carefully enforced perfor- panding the private ing funds from other purposes to pay electric- tnance contract can be as politically costly as a sector. The views pub- ity bills, or allowing overstaffed state enterprises well-designed privatization, performance con- lished are those of the authors and snould not to lay off workers. All governments seem to tracts are not likely to be successful in countries be attributed to the have underestimated the extent of their infor- that lack the political will to privatize, where World Bank or any of mation disadvantage relative to managers. thev mav be viewed as a soft alternative to its affiliated organiza- T tions. Nor do any of the privatization. The findings suggest that perfor- conclusions represent Improving enterprise performance mance contracts should be used only where gov- official policy of the ernments are politically prepared to make tough Executlve Directors Chile's successful experience in reforming its decisions and the contract is part of a broader or the countries they state enterprises points to actions that are key package of state enterprise reforms. represent. to improving efficiency: Chile increased com- To order additional petition by ending any legally mandated state References copies please call monopolies and barriers to entry, reducing 202-458-tlt Ior contact import tariffs to 10 percent across the board. Shirleyv Mary, and Lixin ColinxtI. 1997a. "Empirical Effects of Perfor- Suzanne Smitn, editor, r u m manceContracts:Evidencefromchina."WorldtBank,Developinent Room F6P-188, breakmg up Ronopolies m such secturs as elec- Research Group, Washington. D.C. The World Bank, tricity, and pushing state enterprises to con- . t9s7In.ofruiatioa, Incentives. and Conrnitinent: An Empirical 1818 H Street, NW, tract out competitive activities under strict rules Analysis of Contracts hert'een Government and State Enterprises." Or Internet address of competitive bidding. It placed state enter- Policv Resea'cl Working Paper 1769. World Bank, Development ssmith7@worldbank.org. prises under private commercial law, and mem- World Bank. 1995. Buoeaucrats in Business: TheEconosnics and Poli- The series is also bers of the boards of directors became liable dosofGovernnrentownesh NewYork: Oxford L'niversitv Press. available on-line (www.worldbank.org/ for their decisions. Private parties were named html/fpd/notes/ to boards, and boards were kept small (five Mamy Shirley, Research Manager, Development notelist.html). people) to reduce the political value of keep- Research Group @ Printed on recycled ing companies public. The government elimi- paper. nated all subsidies, transfers, and government