## Thursday, November 26, 2020

### What ails public procurement: an analysis of tender modifications in the pre-award process

by Shubho Roy and Anjali Sharma.

One source of low state capacity in India lies in the ability of the state to contract with private persons. The contracting process starts at procurement and runs till the final payment or dispute resolution. This process suffers from delays and disputes, to the point where managers in government feel uncertain about whether a given contract will work correctly, and private firms feel that doing business with the government is problematic. While difficulties in the last step (payment delays (link, link) and/or contract disputes (link, link) tend to loom large in the discussion, these are the final manifestations of weaknesses of the overall process of government contracting.

Research on government contracting is required in order to diagnose the sources of difficulty and design solutions. As an example, Lewis-Faupel et. al. (2016), analyse a database of 35,600 contracts awarded by the government under the Pradhan Mantri Gramin Sadak Yojana (PMGSY) to understand whether electronic procurement is associated with better procurement outcomes. They find some improvement in quality but not much impact on costs and delays. While state capacity in government contracting is an important problem, in the Indian context, there is a limited empirical literature on the subject. Due to difficulties with data, most studies rely either on small data sets (Goyal (2019)) or on the case study approach (Nag (2015)).

In this article we use a new dataset about the public procurement process in India to measure one aspect of the public procurement tender process: modifications made by the procuring entity to tender documents. We analyse the frequency and nature of these modifications and the possible impact these modifications might have on entities that participate in government tenders. We find that a large proportion of tenders that are published see modification and that government procurers make frequent modifications, especially in high value tenders. We offer some speculation on the causes and consequences of tender modifications.

### Difficulties of measurement of the public procurement process

The Indian state buys through various entities and at various levels. Each government department or entity is responsible for following the General Financial Rules (GFR) and purchasing the goods and services it needs to meet its operating requirements. Most government departments or entities do not have a centralised purchasing office. Purchasing is, often, distributed geographically and by value. Small value purchases (relative to the entity's budget) are usually carried out by lower rung offices, while larger value purchases are carried by regional or even the national headquarters. Similarly, purchases of goods or services may be carried out at the location where it is required. For example, a contract to maintain an agency's local office may be advertised only in the city where the local office is situated. The fact that purchasing takes place at many locations makes it harder to assemble datasets about it.

The first step in the process is tendering for bids. Advertisements for government procurement called Notice Inviting Tenders (called NITs in this article) are typically published in local newspapers where the government entity is interested in procuring goods or services. If the proposed procurement is above a certain threshold, it is advertised in newspapers with a national circulation. Any attempt to get data from NITs would call for scanning hundreds of newspapers, every day.

### A new opportunity for measurement

In 2011, a website called the Central Public Procurement Portal (CPPP) was established. From 2012 onwards, all central government organisations, including Central Public Sector Enterprises (CPSEs) and autonomous and statutory bodies under the Central government, are required to publish their tenders on the CPPP. Since then, in addition to publishing of tenders and awards information online a large part of the tendering process has also been automated using the portal. The mandate has evolved from e-publish to e-procure.

This website allows us to observe tenders from the date they are published till they are awarded. That is, from the time that the general public/interested bidders are informed that the government is interested in purchasing some goods or service or works, to the time that the government selects a supplier and awards the contract.

### The problem of tender modification

In this article, we focus on one measure of state capability in the procurement process: modifications that are made to the NIT during the procurement process. When the government issues the NIT, interested vendors respond by submitting a bid document to the government. The NIT contains details about: (1) the item being procured (including details such as procurement category, technical specifications, bill of quantity, estimated value of procurement, delivery specifications and period of work), (2) qualifications for eligible bidders (financial and technical eligibility conditions), and (3) critical dates for the tender process (publishing date, submission date, opening date). Any modification that is made to the NIT after it is first published is a source of substantial cost on all entities interested in bidding for the government procurement. For instance, if the government modifies the qualification requirements for the bidder after publishing the NIT, those who originally qualified for the NIT may not qualify after the change. The effort taken to develop a bid is then wasted. Once tender modifications are endemic, at every stage in the process, bidders build in expectations about future fluctuations through tender modifications, and this reshapes their decisions to try to sell to the government and the price at which engagement with the government could be profitable.

The government often requires potential bidders to submit 'earnest money deposits' or EMD with their bids, which is a financial guarantee. Potential bidders need to take steps to tie up the funds required for this EMD. This typically involves an explicit cost, such as getting a line of credit from a bank, or an opportunity cost, such as keeping this money aside and not using it for any other purpose. If the period of the tender process is then extended by the government procurer, the period for which this cost has to be borne by the potential bidder also gets extended. In large value tenders, this cost can be substantial. This cost has to borne by all potential bidders. For instance, if six firms bid for a civil works tender where the EMD requirement is Rs.1 million, the aggregate cost to the economy is the cost of keeping Rs.6 million of capital aside till one supplier is selected. The selected supplier can internalise this cost in its bid. But for the five that are not selected, this cost does not have the offsetting benefit. It will influence the supplier's overall business, not just its current or future government engagement.

Similarly, modifications in the original technical specification of the goods and services in the NIT, may require interested bidders to substantially change their bid documents or even drop out of the process after incurring the costs of putting together the bid documents. If the date of the delivery of goods or services is modified, the ability of the supplier to manage its supply chain is hampered. Even when a firm feels confident that it will win a contract, unpredictable delays in tendering hamper efficiency in the production and supply process.

We find three mechanisms through which the NIT can be modified: (1) cancellation, (2) re-tendering, and (3) corrigenda. An outright cancellation is where the government procurer retracts its decision to purchase some goods or service. A re-tender is where a published NIT is withdrawn and replaced with a new NIT. This fresh NIT usually has substantially different provisions and requirements for the bidders. Corrigenda are issued when the government procurer wishes to make changes to the NIT that it deems as not so substantial as to warrant a re-tender, and instead amends certain terms of the NIT. Each of these three mechanisms of modifying NITs introduce uncertainty and costs for all potential bidders in the public procurement process.

These three mechanisms also make the tender process uncertain and costly for the government procurer. Procurers have to spend time and resources to prepare NIT documents, and frequent document changes add to this cost. For instance, if a tender is published and then cancelled, the effort and the resources go to waste. Similarly, frequent corrigenda also require time and resources to manage and monitor the change management process. If modifications to the NIT create disputes and complaints about probity from potential bidders, additional costs of investigation and enforcement alongside possible litigation arise. If frequent modifications to the NIT create an outcome of too few or no bidders, there can even be an extreme outcome of a failed procurement.

Given the uncertainty and costs that modifications to NIT can impose on both the government procurers and the bidders, we think of these three mechanisms for making changes as sources of errors in the tender lifecycle. Our dataset makes possible the measurement of the `modification rate' which is a metric of the friction in the procurement process.

### A few entities account for most of the procurement

In November, 2020 there were more than 26,000 government procuring organisations registered on the CPPP. An average of 0.225 million tenders were published on the portal in FY 19 and FY 20 each. These tenders had an estimated value of Rs.8 trillion in each of these two years. We find that 11 procuring organisations within the Union government accounted for around 50% of the NITs by count and more than 80% of the NITs by value in both FY 18-19 and FY 19-10. Table 1 gives the details of the NITs published by these entities.

Table 1: The largest organisations by NIT count and value across FY 19 and FY 20

Organisation Share in count (%) Share in value (%) Average tender size (Rs. million)

Airports Authority of India 2.0 1.8 32
Border Roads Organisation 2.1 0.7 12
Defence Research and Development Organisation 3.3 0.2 3
Delhi Metro Rail Corporation 0.2 1.2 188
E-in-C Branch of Military Engineer Services 24.8 2.8 4
Engineers India Ltd 0.5 31.4 2,163
Food Corporation of India 2.1 14.6 241
IHQ of MoD (Army)-(OSCC) 14.9 1.3 3
Ministry of Road Transport and Highways 1.1 8.5 263
National Highway and Infrastructure Development Corporation 0.1 2.0 893
National Highway Authority of India 0.9 19.6 795

Total 51.9 84.2 57

### The modification rate is high

For each of these entities, we study all three mechanisms for making modifications to NITs: cancellation, re-tendering and corrigenda. For cancellation and re-tendering, we examine the data from the CPPP analytics dashboard over the two year period, i.e. FY 2018-19 and FY 2019-20.

For corrigenda, we find that there is no aggregate data available and it has to be hand collected at the level of each NIT. We focus on collecting corrigenda data for NITs published by the organisations in Table 1. The tenders for which we collect data are those that are currently not active. A tender that is not active is one which has either been awarded or cancelled. We focus on NITs published in the period of January to March, 2020. We pick this period for our analysis as it is before to the start of the Covid lockdown and allows us to study the corrigenda pattern in normal times. In this period, we record which NITs have no corrigenda and which NITs have at least one or more corrigenda. This analyse a set of 11,714 tenders across the 11 procuring organisations in Table 1 for the presence or absence of corrigenda.

Table 2 gives a summary of the modification rate for the three mechanisms. It shows that across the 11 procuring organisations, potential bidders face substantial uncertainty due to tender modifications. Since the corrigenda data is from a period different from the tender cancellation and re-tendering data, these three modification rates cannot be added up to arrive at the aggregate modification rate. However, it is clear that a large proportion of NITs see changes, and that issuing corrigenda is the most frequent mechanism through which NITs are changed. Corrigenda are issued for nearly 20% of the NITs that get published in our analysis period.

We may expect that organisations that do more tendering would develop greater organisational capability to do this well. We might expect that difficulties of contracting in the past would have triggered off feedback loops into organisation design that address these difficulties. However, across these 11 entities, we see the reverse relationship. This raises concerns about the extent to which difficulties faced by these organisation have generated feedback loops into modified organisation design and reduced tendering inefficiencies.

Table 2: Sources of modifications in NITs (in %)

Organisation Re-tender Cancellation Corrigendum
(FY 19 & FY 20) (FY 19 & FY 20) (Jan-Mar 2020)

Airports Authority of India 4.4 13.0 26.6
Border Roads Organisation 2.8 8.0 7.5
Defence Research and Development Organisation 4.4 7.0 18.3
Delhi Metro Rail Corporation 0.1 25.3 56.3
E-in-C Branch of Military Engineer Services 16.6 5.7 17.2
Engineers India Ltd 5.7 4.3 43.0
Food Corporation of India 1.4 7.3 35.6
IHQ of MoD (Army)-(OSCC) 6.0 10.4 10.5
Ministry of Road Transport and Highways 7.8 15.3 43.1
National Highway and Infrastructure Development Corporation 1.7 28.2 78.7
National Highway Authority of India 0.5 16.0 19.5

Total 10.5 8.1 19.5

### Corrigenda are frequent

We turn to the nature and the frequency of the corrigenda being issued. For this, we identify a sample of active tenders with corrigenda for the entities listed in Table 1. Our analysis in Table 2 relied on tenders that were not active. However, on the CPPP website, corrigenda details are only available for tenders that are currently active. An active tender is one for which the tender lifecycle is yet to be completed. In this analysis, we identify 104 tenders. These tenders had 536 corrigenda issued against them when the sample was collected. Since, these are active tenders, more corrigenda may have been added to them subsequently.

While a cancellation terminates the procurement process and the re-tender starts it afresh, a corrigendum keeps the process running. The government procurer may make incremental changes to terms of the NIT with a corrigendum. For example, the government may add details to the technical specifications on the item being procured. Or it may change the date by which bids have to be submitted. Or it might demand additional documents from bidders as proof of eligibility. While the changes introduced by a single corrigendum may be small, over time, with many corrigenda, these changes may become substantial. So while corrigenda do not terminate the process, they introduce significant uncertainty in the procurement process.

Our analysis of 536 corrigenda across 104 tenders shows that changes by corrigenda are common. Table 3 presents some of the features of these corrigenda. We observe:

• On an aggregate basis, around five or more corrigenda are issued for every tender. This number varies across organisations, with entities like NHAI issuing as many as 30 corrigenda per tender.

• Around 60% of these corrigenda are issued to change the bidding dates. The remaining 40% are to make changes such as alterations to the technical specifications, the bill of quantity or for issuing more details or clarifications than were given in the original NIT document.

• There is a systematic under-estimation by procuring organisations of the time it will take to get bids. Across the 11 procuring organisations, the original estimate of the time it will take to receive bids was 16 days. However, through frequent issuance of corrigenda this period was extended to 81 days. And since these are active tenders, the bid period may increase even further.

• Some entities, like NHAI and NHIDC not just have more tenders with corrigenda but also more corrigenda per tender.

Table 3: Frequency and nature of corrigenda

Avg. tender size (Rs. million) Avg. corri-genda /tender (No.) O/w Date change (%) Original bid period (Avg. in Days) Revised bid period (Avg. in Days)

Airports Authority of India 504.2 2.5 60 17 33
Border Roads Organisation 264.5 4 63 22 64
Defence Research and Development Organisation 9.7 1.6 63 18 26
Delhi Metro Rail Corporation 2,811.9 1.3 77 7 14
E-in-C Branch of Military Engineer Services 1.9 1.2 8 7 7
Engineers India Ltd NA 3.3 33 16 31
Food Corporation of India 1.6 2 - 16 19
IHQ of MoD (Army)-(OSCC) 2.0 3.8 89 16 49
Ministry of Road Transport and Highways 1.9 1 100 7 7
National Highway and Infrastructure Development Corporation 1,561.9 4.4 20 10 24
National Highway Authority of India 7,620.3 29.7 67 44 584

Total 1,848.2 5.2 60 16 81

### An illustration of changes through corrigenda for a complex project

Table 3 presents a subset of a larger set of parameters on which the original tender is modified through the corrigenda. We illustrate the extent to which original tenders can experience change using one example of a tender for a complex procurement: the NHAI Contract to upgrade a stretch of road at Panipat from two lanes to four lanes. The estimated value of the tender is Rs. 2.1 billion and the estimated period of work post award is 2 years. Potential bidders are required to submit an EMD of Rs. 21 million, 1% of the tender value.

Table 4 is a brief summary of the timeline of the NIT. As of the date of data extraction, the NIT has been amended 31 times, and has been active for 693 days as against the original bid period of 44 days.

Table 4: An example of frequent changes to a tender

Date NIT change event Time elapsed (in days)

10th Dec. 2018 Original NIT published on CPPP website. Original bid submission timeline: 23rd January, 2019. Bids to be opened on on 24th January, 2019. -
11th Dec. 2018 NHAI publishes first corrigendum. This adds a new set of information about the procurement. Additional documents added to the NIT by NHAI. 1
21st Jan. 2019 NHAI issues second corrigendum two days before the final date of submitting bids. This corrigendum extends the submission timeline from 23rd January to 12th February, 2019. 42
11th Feb. 2019 One day prior to the submission deadline, bid submission date extended by 7 more days to 19th February, 2019. 63
13th Feb. 2019 NHAI publishes three corrigenda on a single date. They are titled "Revised Tender Documents", "Financial Proposal", and "Letter". After nearly two months of publication of original NIT, NHAI makes substantial changes to the planned procurement. Interested bidders would have to go back to the drawing board and redo their bids. On 13th February, the deadline to submit the bids was 19th February, so the bidders would have just six days to absorb all the changes, make fresh documents and submit them. 65
14th Feb. 2019 The NHAI publishes another corrigendum. This one contained replies to queries raised by the bidders. The clarifications may have required the bidders to redo their bids or make substantial changes. 66
18th Feb. 2019 One day before the bid submission deadline, NHAI changes the deadline for the third time. Now, the final date for submitting bids would be the 25th of February, 2019. 70
22nd Feb. 2019 to 19th Oct. 2020
The NHAI goes on to publish 25 more corrigenda, each extending the deadline for submitting bids by about two weeks. The bid submission timeline is now 17th November, 2020. 679
2nd Nov. 2020
We extract corrigendum data for this tender. The tender is still active so more corrigenda can be added. 693

### Conclusion

In this article, we have used a novel dataset and discovered some new facts about the public procurement process.

While there is significant decentralisation in the procurement process across government entities, there is a possibility of improving public procurement by initiating reforms at a few entities that make up the bulk of procurement activity.

A large proportion of the tenders published are modified by the government procurer, either through a corrigendum or through re-tendering and tender cancellation. These tender modifications are taking place at a scale that increases uncertainty in the procurement process. There is a need for organisational reform within government organisations, so that better analysis is done before the first document is unveiled, so that the need for changes thereafter is minimum.

How the state contracts with private persons is one important element of the overall problem of state capacity. There is a need to build knowledge, and a literature, on this subject. This article constitutes one small element of that overall research program.

### References

Sean Lewis-Faupel, Yusuf Neggers, Benjamin A. Olken and Rohini Pande (2016), Can Electronic Procurement Improve Infrastructure Provision? Evidence from Public Works in India and Indonesia, American Economic Journal: Economic Policy, Vol. 8, No. 3.

Bodhibrata Nag (2015), Combating Corruption in Indian Public Procurement - Some Exploratory Case Studies, The Journal of Institute of Public Enterprise, Vol. 38, No. 1&2.

Yugank Goyal (2019), How Governments Promote Monopolies: Public Procurement in India, American Journal of Economics and Sociology, Vol. 78, No. 5.

Anjali Sharma is a researcher with the Finance Research Group. Shubho Roy is a doctoral candidate at the University of Chicago. The authors would like to thank Susan Thomas for discussions and useful inputs and Charmi Mehta and Sejal Gajjar for assistance with collecting the data.

LaTeX mathematics works. This means that if you want to say $10 you have to say \$10.