Autonomy – not rules – may be the government’s best weapon in the fight against corruption

Katie Parry, Oriana Bandiera, Michael Best, Adnan Khan, Andrea Prat 13 May 2020

a

A

Governments buy a lot of stuff. In 2015, for example, general government procurement spending in OECD countries represented an average of 29% of total general government expenditures, or 12% of GDP (OECD 2016). The figure for developing countries is often at least as high; Pakistan, for example, spends around 20% of GDP on public procurement. 

As a result, weak procurement systems can substantially reduce the amount of money that governments have left to spend on other priorities. Ensuring value-for-money in everything the government purchases is a vital part of its mission to serve—and maintain the trust of—their citizens.

Optimum procurement policy: Strict rules or autonomy?

There are two main sides in the debate around optimum procurement policy. International organisations such as the OECD and World Bank tend to advocate for a policy of strict rules, combined with the development of strong monitoring and oversight institutions to ensure compliance. This discourages egregious misuse of government funds, but it also means that procurement offices (POs) must deal with a lot of red tape and may be discouraged from innovating—or even punished for doing so—even when there is the potential for savings.

Procurement is also an important function in the private sector. However, businesses tend to take the opposite approach – they generally grant considerable autonomy to POs, and trust them to use their discretion to seek out value for money. This philosophy is well summed up by Netflix’s five-word policy for expensing: “Act in Netflix’s best interest”. This approach calls for lighter monitoring during procurement and is often accompanied by pay-for-performance schemes that reward efficiency.

The economic literature has debated the trade-off between rules and discretions for some time. Recent work has indicated that the effect of greater discretion on procurement outcomes depends crucially on the quality of the procuring agency and its personnel (Mirenda et al. 2019, Baltrunaite et al. 2018), but this remains a question of active debate. In a paper released this year, however, researchers at the LSE and Columbia have found an innovative way to test the competing theories around procurement. They worked with the Government of Punjab to measure value-for-money around the purchase of generic goods1, and to assess the impact of giving POs either greater autonomy or financial incentives.

The price paid for generic goods varies considerably

Data collected over the course of the study show that there is considerable variation in the price paid by different public bodies for exactly the same good. Each point in Figure 1, for example, is a purchase of pens. The horizontal axis shows the price the public body actually paid for the pens and the vertical axis shows what the item is actually worth. As the orange line shows, POs paid anything between 3.5 and 115 rupees for pens that were worth 25.

Figure 1 The standard cost of—and amount paid for—pens

While the cost of each individual item is small—115 rupees is less than $1–the volumes are large. The purchase of generic items such as these accounted for over half of non-salary expenditures in many of the offices included in the study.

The savings from greater autonomy were considerable

The study tried two different approaches to increasing procurement efficiency. The spending of each PO is monitored by an Accountant General (AG), and the ‘autonomy’ treatment simplified or bypassed much of the pre-audit monitoring POs are subject to when procuring generic goods. As shown in Figure 2, a strict limit was put on the types of documents AGs could ask for as part of the spending approval process, and POs were given full discretion over how they spent a portion of their budget.

Figure 2 Procurement process summary

Panel A Status quo procurement process

Panel B Procurement process under autonomy treatment

Intuitively, this increase in PO autonomy is likely have to two effects that work in different directions: (i) removing red tape and increasing autonomy to achieve value-for-money should decrease prices paid for goods, while (ii) the of lower oversight might increase wastefulness in either an active way (i.e. corruption) or a passive way (i.e. laziness).

The study found that providing procurement officers with additional autonomy led to a reduction in the average unit prices paid by 8–9%, indicating that the removal of waste caused by complying with the monitoring activities more than offsets the loss of the benefits that that monitoring provides. Even in the relatively small group of offices included in this experiment, this virtually costless intervention led to savings that were enough to fund the annual operation of an additional five schools or to add 75 hospital beds.

Incentives were less effective, though still offered value for money

The second ‘incentives’ treatment introduced performance pay for procurement officers, rewarding them financially for achieving greater value-for-money. The study found that the average impact of these incentives hovered just above zero. It seems that where monitoring is onerous, POs are too tangled up in red tape to do anything to improve their performance.

Despite the low per-unit savings, however, the rate of return on this treatment was solid. The comparatively low cost of implementing the incentive scheme and the large volumes purchased meant that the government saved $1.45 for every $1 it gave out in bonuses.

Procurement in this context is highly demand-led and the items tracked in this study are highly generic, so neither treatment had any discernible effect on the variety, amount or composition of goods purchased, or the timing of procurement expenditure.

The effect of both treatments depended on the ‘type’ of monitor

The impact of these treatments varied considerably across participants. An important determinant of the success of the two treatments was the ‘type’ of AG assigned to monitor each PO.

The autonomy treatment worked best—delivering around 15% in savings—when the AG was extractive (as measured by the extent to which they delayed approvals2), while the incentives treatment worked best—delivering around 6% in savings—when the AG was effective. The fact that the autonomy treatment was more effective overall indicates that, in this context, the average AG is relatively extractive.

This is an important result; it implies that optimum allocation of authority between agents at different levels of a hierarchy will depend on their relative degrees of wastefulness. Other things being equal, the more wasteful the managers, the more savings can be realised by shifting autonomy to frontline personnel.

The appropriate response to inefficiency and corruption may sometimes be less monitoring, not more

More work is needed to understand whether this result holds in other contexts, and how these policies would affect the calibre of personnel attracted to procurement jobs in the long-term. It may be, for example, that greater autonomy might attract dishonest POs, though it might also improve the quality of supervision. Giving more autonomy to officers implies taking it away from the monitors, which may change the nature of AG applicants for the better.

Overall, this study indicates that we need to check our instinct to react to government inefficiency and corruption with increased monitoring. In many places, the elaborate system of pre-audit monitoring facing bureaucrats is probably doing more harm than good. Greater autonomy (and rigorous post-audit of purchases) may, counter-intuitively, be a key part of the solution.

Editors' note: This column was also pulished on VoxDev.org.

References

Bandiera, O, M C Best, A Q Khan and A Prat (2020), “The Allocation of Authority in Organizations: A Field Experiment with Bureaucrats”, Working Paper.

Baltrunaite, A, C Giorgiantonio, S Mocetti and T Orlando (2018), “Public procurement: Discretion may have its limitations”, VoxEU.org, 26 July.

Mirenda, L, S Mocetti and L Rizzica (2019), “The boss on board: Mafia infiltration, firm performance and local economic growth”, VoxEU.org, 26 October.

OECD (2016), “General government procurement as a percentage of GDP (2015)”.

Endnotes

1 Procurement officers were trained to enter detailed data on the attributes of the items they were purchasing into a web-based platform and were audited to physically verify that the data were accurate.

2 Again, the authors are agnostic as to whether these delays are caused by active obstruction (i.e. corruption) or passive obstruction (i.e. laziness)

a

A

Topics:  Competition policy Politics and economics

Tags:  procurement, Corruption, Punjab, inefficiency, autonomy, monitoring

Writer and editor; Adjunct Lecturer in Public Policy, Harvard Kennedy School

Professor of Economics and Director of STICERD, LSE

Assistant Professor of Economics at Columbia University

Professor in Practice, School of Public Policy and STICERD, London School of Economics

Richard Paul Richman Professor of Business and Professor of Economics, Columbia University; and Co-Director of CEPR's Industrial Organization programme

Events

CEPR Policy Research