ippc 6 presentation

17
Firm Participation and Competition in Public Procurement Jani Saastamoinen* 6 th International Public Procurement Conference, 14th – 16th August 2014, Dublin, Ireland *University of Eastern Finland

Upload: dr-paul-davis

Post on 12-Apr-2017

178 views

Category:

Business


0 download

TRANSCRIPT

Page 1: Ippc 6 presentation

Firm Participation and Competition in

Public Procurement Jani Saastamoinen*

6th International Public Procurement Conference, 14th – 16th August 2014, Dublin, Ireland

*University of Eastern Finland

Page 2: Ippc 6 presentation

Outline

1) Background

2) Research Problem

3) Data and Methods

4) Results

5) Conclusion

27.8.2014 Saastamoinen 2

Page 3: Ippc 6 presentation

27.8.2014 Saastamoinen 3

1) BACKGROUND

– Contracting authority (CA) mandated to exploit competitive environment in the markets when carrying out public procurement

• Efficient allocation of resources

• Attracting bidders crucial to obtain the efficient outcome

– Winner’s Curse

• Aggressive bidding leads to negative or lower than expected profit for the winning bidder

• Less likely in competitive markets (Wilson 1977; Thaler 1992)

Page 4: Ippc 6 presentation

• Contract options may influence the level of competition in a tender request

– Open vs. restricted tendering

– Competitive dialogue

– Electronic auctions

– Dynamic procurement

– Aggregated vs. lots

– Contract size

– Framework agreements

– Pure price competition vs. Economically most advategeous tender (EMAT)

27.8.2014 Saastamoinen 4

Page 5: Ippc 6 presentation

• Usually public prorcurement is carried out by first-price sealed bid common value reverse auctions

• In theory, the winning bid corresponds to the true value of the auctioned object (Wilson 1977)

• Competition intensifies with the number of bidders (Milgrom and Weber 1982)

• Biased estimation may lead to the Winner’s Curse (Milgrom 1989)

• Potential consequences for procurer: higher costs, lower quality, order delays, legal expenses, externalities of business failure (Bilginsoy 2000)

• Attempts to avoid the Winner’s curse may lead to cautious bidding resulting in a higher costs (Hong and Shum 2002)

• Empirically studied with bid dispersion which measures the difference between the winning bid and the second highest bid

• Substantial bid dispersion may be an indicator of the Winner’s Curse (Dyer and Kagel 1996)

• Bid dispersion decreases with the number of bidders (Dougherty and Lorenz 1976; Saidi and Marsden 1993; Dyer and Kagel 1996; Bilginsoy 2000)

27.8.2014 Saastamoinen 5

Page 6: Ippc 6 presentation

2) MOTIVATION AND RESEARCH PROBLEM

•Reseach is scarce on the impact of contracting options on the number of bidders and competition

•Provides a wider perspective to competition in public procurement

•Research questions:

– How different contracting options affect the number of bidders?

– How the number of bidders impacts competition?

27.8.2014 Saastamoinen 6

Page 7: Ippc 6 presentation

3) DATA AND METHODS

•Regional CA from Eastern Finland

– Annual purchasing volume € 150 million

– e-procurement only

•363 procurement contracts (including lots) awarded in 2011

•Contract values usually above the national or the EU threshold

•Statistical analysis carried out with OLS, logit and tobit regressions

27.8.2014 Saastamoinen 7

Page 8: Ippc 6 presentation

4) RESULTS

Descriptive statistics (categorical variables)

27.8.2014 Saastamoinen 8

Variable Frequency Relative Frequency (%)

Contract value above the EU threshold (eu_d) 126

[156]

47.4

[43.0]

Contract value above the national threshold (nat_d) 102

[153]

38.3

[42.1]

Lots contract (lots_d) 200

[278]

75.2

[76.6]

Service contract (serv_d) 106

[127]

39.8

[35.0]

Supplies contract (sup_d) 158

[231]

59.4

[63.6]

Restrictive tendering (res_d) 40

[53]

15.0

[14.6]

Framework agreement (frame_d) 172

[221]

64.7

[60.9]

Single bid tender (onebid_d) -

[96]

-

[26.4]

Contract awarded to multiple bidders (multi_d) 144

[183]

54.1

[50.4]

Notes: Figures in parentheses include the tender calls where only a single bid was submitted.

Page 9: Ippc 6 presentation

Descriptive statistics (continuous variables)

27.8.2014 Saastamoinen 9

Mean Median Standard

Deviation

Minimum Maximum Number of

Observations

n 4.30

[3.42]

3

[2]

4.13

[3.82]

2

[1]

42

[42]

266

[363]

N 21.24

[18.61]

12

[12]

30.59

[26.92]

2

[1]

175

[175]

266

[363]

B2 – B1 7170.67 332.73 35208.60 0 457450 266

ln(B2 – B1 + 1) 5.65 5.81 2.98 0 13.03 266

(B2 – B1)/B1 .43 .13 2.10 0 32.9 266

ln(B2) – ln(B1) .23 .13 .34 0 3.52 266

CV .24 .19 .23 0 1.61 266

Sd. 9697.77 611.59 40604.73 0 517791.2 266

B 68353.04

[52561.33]

5225.41

[3546.87]

421979.60

[351557.1]

.93

[.71]

6320431

[6062820]

266

[363]

p 80.47

[81.76]

90

[90]

22.98

[23.06]

40

[40]

100

[100]

225

[284]

Notes: The figure in parentheses includes also the tenders where only a single bid was received.

n: Number of bidders.

N: Number of interested bidders.

B2 – B1: The difference between the two lowest bids.

ln(B2 – B1+1): A logarithmic transformation of the difference between the two lowest bids.

(B2 – B1)/B1: A relative difference of the difference between the two lowest bids.

ln(B2)-ln(B1): A logarithmic transformation of the relative difference between the two lowest bids.

B : The average bid size (Euro).

p: The proportion of price in percentage points [0, 100] in the applied EMAT criterion.

Page 10: Ippc 6 presentation

Single bidder tenders (logit regression on a dummy dependent variable)

27.8.2014 Saastamoinen 10

Estimated

Coefficient

Marginal Effect

constant -.457

(.653)

-

eu_d .108

(.477)

.026

(.116)

nat_d -.391

(.425)

-.090

(.092)

sup_d 1.058***

(.302)

.228***

(.055)

frame_d -.055

(.310)

-.014

(.078)

lots_d .822***

(.301)

.164’**

(.047)

multi_d -1.400***

(.405)

-.357***

(.099)

p -.004

(.005)

-.001

(.001)

inv_N 2.447***

(.885)

.604***

(.225)

)ln(B -.109***

(.036)

-.027***

(.009)

Wald Chi2 43.96*** -

Pseudo-R2 .134 -

Number of Obs. 279 279

Notes: * p-value < 0.1; ** p-value < 0.05; *** p-value < 0.01.

Standard errors in parentheses. Robust standard errors used.

Page 11: Ippc 6 presentation

Tenders with several bidders (dependent variable = inv(n))

27.8.2014 Saastamoinen 11

Variable Estimate

(OLS)

Estimate

(Tobit)

constant .535***

(.127)

.599***

(.159)

lots_d .158***

(.045)

.194***

(.058)

sup_d .226***

(.040)

.251***

(.049)

frame_d -.033

(.057)

-.040

(.073)

eu_d .053

(.100)

.056

(.126)

multi_d -.278***

(.063)

-.325***

(.083)

res_d -.025

(.088)

-.071

(.112)

nat_d -.043

(.095)

-.066

(.126)

p -.001

(.001)

-.001

(.001)

inv_N .733***

(.148)

1.011***

(.254)

)ln(B -.027***

(.007)

-.031***

(.009)

F-statistic 21.20*** 12.76***

R2 .263 -

Pseudo-R2 - .231

Observations 279 279

Censored

observations

- 57

Average VIF 3.68 -

Notes: * p-value < 0.1; ** p-value < 0.05; *** p-value < 0.01.

Standard errors in parentheses, robust standard errors used.

Page 12: Ippc 6 presentation

Competition in tenders (dependent variable relative bid dispersion)

27.8.2014 Saastamoinen 12

A1 A2 A3

(OLS)

A4

(Tobit)

A5

(OLS)

A6

(Tobit)

constant .117**

(.052)

.100*

(.053)

.139

(.150)

.135

(.151)

.105

(.189)

.089

(.189)

lots_d .026

(.062)

.022

(.063)

.077

(.076)

.070

(.076)

sup_d -.033

(.052)

-.033

(.052)

-.008

(.066)

-.012

(.067)

frame_d -.099

(.079)

-.090

(.080)

-.130

(.089)

-.120

(.090)

eu_d .013

(.117)

.008

(.117)

.105

(.142)

.097

(.143)

multi_d .118

(.088)

.110

(.088)

.081

(.105)

.078

(.105)

res_d -.042

(.106)

-.046

(.107)

-.030

(.127)

-.039

(.127)

nat_d -.024

(.107)

-.018

(.107)

.053

(.131)

.057

(.131)

p - - -.001

(.001)

-.001

(.001)

inv(n) .348**

(.147)

.380**

(.150)

.414**

(.171)

.443**

(.172)

.423**

(.191)

.453**

(.192)

)ln(B -.004

(.009)

-.005

(.009)

-.003

(.189) -.003

(.010)

F-statistic 5.63** - 1.31 - 1.22 -

LR Chi2 - 6.38** - 11.94 - 11.87

R2 .021 - .045 - .055 -

Pseudo-R2 - .032 - .059 - .067

Observations 266 266 263 263 222 222

Censored

observations

8 - 8 - 8

Average VIF 1.00 - 3.43 - 3.66 -

Page 13: Ippc 6 presentation

27.8.2014 Saastamoinen 13

Competition in tenders (dependent variable absolute bid dispersion)

B1 B2 B3

(OLS)

B4

(Tobit)

B5

(OLS)

B6

(Tobit)

constant 5.733***

(.475)

5.643***

(.500)

-1.906

(.648)

-1.916

(.646)

-3.291

(1.074)

-3.334

(1.088)

lots_d .212

(.335)

.201

(.336)

.490

(.348)

.473

(.346)

sup_d -.148

(.292)

-.147

(.293)

-.069

(.334)

-.075

(.333)

frame_d -.186

(.399)

-.170

(.398)

-.471

(.466)

-.452

(.465)

eu_d .272

(.561)

.263

(.557)

.770

(.780)

.754

(.785)

multi_d -.185

(.352)

-.200

(.349)

-.114

(.390)

-.119

(.385)

res_d -.410

(.485)

-.419

(.477)

-.408

(.546)

-.427

(.541)

nat_d .417

(.502)

.430

(.496)

1.037

(650)

1.045

(.646)

p - - .009

(.009)

.009

(.009)

inv(n) -.242

(1.322)

-.076

(1.364) 2.545***

(.886)

2.606***

(.892)

2.143**

(1.051)

2.210**

(1.060)

)ln(B

.822***

(.034)

.820***

(.035)

.847***

(.040)

.845***

(.040)

F-statistic .03 .00 110.94*** 111.98*** 91.68*** 92.94***

LR Chi2 - - - - - -

R2 .000 - .675 - .667 -

Pseudo-R2 - .00 - .212 .204

Observations 266 266 263 263 222 222

Censored

observations

- 8 - 8 - 8

Average VIF 1.00 - 3.43 - 3.66 -

Page 14: Ippc 6 presentation

Graphical interpretation (relative bid dispersion)

27.8.2014 Saastamoinen 14

Bid dispersion Number of bidders

Contract size

Lots

Supplies

Multiple suppliers

Interested bidders

+

+

+

- - -

Page 15: Ippc 6 presentation

Summary of results

•Single bidder tenders

– A single bid is more likely in supplies and lots contracts

– When a contract is awarded to multiple suppliers and there is more general interest toward a request for tenders, receiving only one bid is less likely

•Tenders with several bidders

– Supplies and lots contracts attract less bidders

– Contracts awarded to multiple suppliers, larger contracts by value, and a general interest toward a request for tenders attact more bidders

•Competition

– Bid dispersion decreases when the number of bidders increases

• Intensifies competition

• Winner’s Curse less likely

27.8.2014 Esityksen nimi / Tekijä 15

Page 16: Ippc 6 presentation

5) CONCLUSION AND IMPLICATIONS

•This study showed that the choice of contracting options affects the number of bidders a request for tenders attracts

• In turn, the number of bidders is positively connected with the level of competition and potential Winner’s Curse in public procurement

• Interest toward a request for tenders translates into a higher number of submitted bids

– CAs should advertise upcoming invitations to tender could increase the number of received bids

•CAs should consider using multi-supplier contracts to instigate competition

•CAs should carefully consider when a contract should be divided in lots and an aggregated contract should be preferred in light of economic benefits arising from competition

27.8.2014 Saastamoinen 16

Page 17: Ippc 6 presentation

Thank you for your attention!

www.uef.fi