Motivating Bureaucrats: Autonomy vs. Performance Pay for Public Procurement in Pakistan
- Civil servants
Public procurement systems are essential in facilitating government investments in public services, but high rates of corruption and leakage may lower the efficiency of government procurement. The way that decision-making power is distributed between procurement officers and their supervisors may be an important determinant of the overall level of corruption. Researchers partnered with several agencies of the Government of Punjab, Pakistan, to evaluate the impact of giving procurement officers more autonomy, compared to financial rewards, on public procurement performance. Shifting authority away from supervisors and towards procurement officers improved performance, particularly when supervisors were more inefficient or corrupt. In contrast, financial rewards for officers did not improve performance when their supervisors were inefficient or corrupt.
Public procurement systems are essential in facilitating government investments in infrastructure, health, education, and other public services. High rates of corruption and leakage may lower the efficiency of government procurement, leading to spending money on overpriced or low-quality inputs. This, in turn, could undermine government efforts to improve citizens’ living standards.
The way that decision-making power is distributed between procurement officers and their supervisors may be an important determinant of the overall level of corruption. Whether to give procurement officers more autonomy, or instead impose strict rules and intense monitoring, has been a subject of intense debate. For example, procurement officers’ decisions on procurement purchases may be vetoed or delayed by their supervisors, or supervisors might mark-up the prices of procurement purchases due to lack of effort or to extract bribes. In such cases, giving more autonomy to procurement officers might improve outcomes by reducing payment delays and allowing officers to buy from a wider range of vendors at lower prices. On the other hand, increased autonomy may also make it easier for procurement officers to embezzle funds and limits supervisors’ discretion in identifying and combating rule-breaking.
Giving officers a financial reward for good performance might also improve the efficiency of public procurement, yet such rewards might be ineffective if supervisors continue to delay purchases and mark-up prices. Can increased autonomy or financial rewards for procurement officers improve procurement performance and reduce leakages?
Context of the evaluation
This study took place in 26 out of 36 administrative districts in the province of Punjab, Pakistan, covering 80 percent of the province population and its largest districts. Researchers worked with different agencies in the Government of Punjab, including the Punjab Procurement Regulatory Authority (PPRA) and the Punjab Information Technology Board (PITB). They also worked with the four largest departments in the government– the departments of Higher Education, Health, Agriculture and Communication and Works. The offices within these departments and districts that were chosen for this study had procurement budgets of at least Rs. 250,000 (US$2,500) in the 2012-2013 fiscal year.
Each office of the Government of Punjab has one employee who is designated as the Procurement Officer (PO), and has the legal authority to make small and medium-sized public procurement purchases. Typical purchases include printer paper, newspaper, and pens. Procurement officers purchase these inputs on behalf of government offices, whose budgets are allocated by the finance department. Before making payments to vendors, the POs are required to submit their purchases for approval by an independent agency of the federal government known as the Accountant General’s office (AG). The AG has offices in each of the districts of the province, monitoring the purchase of offices in that district. AGs have the ability to delay approvals of purchases by demanding more thorough documentation for each purchase.
Details of the intervention
Researchers partnered with several agencies of the Government of Punjab, Pakistan, to evaluate the impact of giving procurement officers either greater autonomy or pay-for-performance rewards on public procurement performance. Six hundred procurement officers were randomly assigned to four groups:
- Autonomy: One group of officers received more leeway in decision-making over expenses through two policy changes: (1) a larger cash-in-hand budget of Rs. 100,000 (US$1,000 at the time) to make cash payments to suppliers without having to seek pre-approval from the AG, and (2) a checklist of the documents that AGs could lawfully request from POs in order to approve a purchase, which restricted the AGs authority and discretion over purchase approvals.
- Pay- for- performance: Another group received financial rewards for good procurement performance based on the quality of goods purchased relative to the price paid. The bonuses ranged from zero to two times a PO’s monthly salary. A special committee was tasked with evaluating performance using a new digital system described below and ranking procurement officers. POs assigned to this group received performance pay based on these rankings twice a year.
- Autonomy & pay-for-performance: The third group received both higher levels of purchasing discretion and performance pay.
- A comparison group, with no changes to autonomy and no performance pay
Researchers used a new digital platform to measure procurement performance. At the beginning of the study in 2014, they partnered with the Punjab IT Board to create a web-based platform—the Punjab Online Procurement System (POPS)—where procurement officers entered details of every procurement purchase. Through this platform, researchers collected detailed data on a set of generic goods and compared the prices different procurement officers paid for goods of similar quality, to produce a measure of “value for money.” POs with the best performance were those that payed the lowest price for items of the same quality.
To determine whether the effects of increased autonomy and performance pay were different under the supervision of different types of AGs, researchers used the share of transactions approved at the end of the fiscal year as a measure of supervisors’ inefficiency or corruption. “Good” supervisors were those who did not delay the procurement process, while “bad” supervisors waited to approve more than 40 percent of transactions until the end of the fiscal year.
See author Oriana Bandeira explain the intervention and results:
Results and policy lessons
Increasing the autonomy of procurement officers improved procurement performance: input prices lowered without a change in their quantity, quality, or attributes. Pay-for-performance had no effect on performance on average.
Increasing the autonomy of procurement officers reduced delays. Results suggest that the autonomy intervention improves performance by removing supervisors’ ability to hold up approvals. Extremely long delays caused by supervisors were 15 percent less likely when procurement officers were given more autonomy, and this effect was driven by offices with “bad” supervisors.
Increasing the autonomy of procurement officers reduced procurement prices, leading to a considerable level of savings. By causing fewer delays, the autonomy intervention reduced unit prices paid by 8–9 percent, (either on its own or in combination with performance pay) on average, indicating that giving bureaucrats greater autonomy did not increase leakage and instead allowed officers to spend taxpayer money more effectively.
Offering performance incentives was ineffective in reducing procurement prices. On average, the financial rewards had no impact on prices.
The effectiveness of each intervention varied based on the type of supervisor. Researchers found that increasing autonomy had no effect when the supervisor approved transactions quickly over the year (“good” supervisors), but it reduced prices up to 15 percent when POs had a “bad” supervisor, who held up more than 40 percent of transactions until the end of the fiscal year. The effect of performance pay had the opposite pattern: performance pay improved POs’ performance (i.e. reduced prices by 6 percent) when the supervisor was “good,” but had no effect when the supervisor was “bad.”
Overall, these results suggest that increased monitoring is not always the most effective response to bureaucratic corruption, and that greater autonomy for bureaucrats may actually improve performance in some instances. Giving autonomy to a public servant can be desirable when it means taking it away from a corrupt or inefficient supervisor. Performance pay, on the other hand, might be more desirable when supervisors are less corrupt and inefficient.
Importantly, increasing autonomy can also lead to large savings for governments: the researchers estimate that the savings from the autonomy intervention across the relatively small group of offices in the study were sufficient to fund the operations of 5 schools or to add 75 hospital beds.