measuring corruption in infrastructure: evidence from transition and developing countries

This article was downloaded by: [University of Chicago Library]On: 07 October 2014, At: 17:51Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,UK

The Journal of DevelopmentStudiesPublication details, including instructions for authorsand subscription information:http://www.tandfonline.com/loi/fjds20

Measuring Corruption inInfrastructure: Evidence fromTransition and DevelopingCountriesCharles Kenny aa World Bank , Washington, DC, USAPublished online: 20 Feb 2009.

To cite this article: Charles Kenny (2009) Measuring Corruption in Infrastructure:Evidence from Transition and Developing Countries, The Journal of DevelopmentStudies, 45:3, 314-332

To link to this article: http://dx.doi.org/10.1080/00220380802265066

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all theinformation (the “Content”) contained in the publications on our platform.However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, orsuitability for any purpose of the Content. Any opinions and views expressedin this publication are the opinions and views of the authors, and are not theviews of or endorsed by Taylor & Francis. The accuracy of the Content shouldnot be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions,claims, proceedings, demands, costs, expenses, damages, and other liabilitieswhatsoever or howsoever caused arising directly or indirectly in connectionwith, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes.Any substantial or systematic reproduction, redistribution, reselling, loan, sub-

http://www.tandfonline.com/loi/fjds20

http://dx.doi.org/10.1080/00220380802265066

licensing, systematic supply, or distribution in any form to anyone is expresslyforbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

Measuring Corruption in Infrastructure:Evidence from Transition and DevelopingCountries

CHARLES KENNYWorld Bank, Washington DC, USA

Final version received August 2007

ABSTRACT This paper examines what we can say about the extent and impact of corruption ininfrastructure using existing evidence. There is evidence that most perceptions measures appear tobe very weak proxies for the actual extent of corruption in the infrastructure sector, largely (butinaccurately) measuring petty rather than grand corruption. Survey evidence is more reliable, butlimited as a tool for differentiating countries in terms of access to infrastructure finance orappropriate policy models. The paper suggests that a focus on bribe payments as the indicator ofthe costs of corruption in infrastructure may be misplaced.

I. Introduction

Estimates regarding the cost of corruption in infrastructure suggest that, forexample, 5– 20 per cent of construction costs are being lost to bribe payments, and asmuch as 20–30 per cent of electricity is being stolen by consumers in collusion withstaff (Gulati and Rao, 2006). If five per cent of investment and maintenance costs ininfrastructure are lost to corruption, the financial burden alone may add up to aboutUS$18 billion a year in developing countries.1

Attempts to measure corruption have proliferated over the past 15 years – notleast with the launch of Transparency International’s Corruption Perceptions Index(CPI) and the rollout of enterprise and consumer surveys which include questions onthe extent of informal payments for licenses, government services and contracts.Corruption-related metrics have been used to determine aid allocations (for the USMillennium Challenge Account, for example) and guide approaches to reform bothat the general and sector level. Such exercises are of utility only if available measuresreasonably reflect an underlying reality of corruption at the sector level. This paperexamines that question.

Correspondence Address: Charles Kenny, Senior Economist, World Bank, 1818 H Street NW, Washington

DC, 20433 USA. Email: [email protected]

Journal of Development Studies,Vol. 45, No. 3, 314–332, March 2009

ISSN 0022-0388 Print/1743-9140 Online/09/030314-19 ª 2009 Taylor & Francis

DOI: 10.1080/00220380802265066

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

The paper discusses both petty corruption (here taken to include ‘speed payments’and other small bribes to obtain everyday services) and grand corruption (includingpayments to secure government contracts or major licenses, change regulations orinfluence the shape of laws). It looks at different approaches to estimating the extentof corruption (expert perceptions, surveys, indirect techniques) and reports on theresults of such studies.

Looking in particular at results from a large survey of corruption in transitioneconomies, the paper suggests that survey evidence is subject to considerableuncertainty. Perceptions measures appear to perform even worse in measuring theextent or impact of corruption on sector outcomes. While there is comparativelylimited evidence for developing countries, it appears likely that the broad lessonsfrom transition economies apply more widely. The paper discusses evidence for therelative costs of corruption impacts and suggests that a focus on bribe payments asthe indicator of the costs of corruption in infrastructure may be misplaced. Thepaper concludes with a discussion of implications for policy towards and analysis ofcorruption in transition and developing countries.

II. Making Estimates of the Extent of Corruption in Infrastructure

Table 1 lays out various ways of measuring the extent and/or impact of corruptionthrough surveys and other approaches. One direct way to examine the extent ofcorruption in a country or sector is to look at cases where it has been revealed as partof a criminal investigation. Of course, such a technique is open to a number ofserious biases – for example, there will be more cases where the justice system isefficient, less corrupt and focusing on the prosecution of corruption. There will alsobe more cases when corrupt activities themselves are less sophisticated and easier todetect. As a result, other approaches are usually preferred if attempting to makecross-country or cross-sector evaluations of corruption, most commonly involvingperceptions, surveys and indirect measures.

The extent of corruption is most frequently estimated through ‘corruptionperception’ indices. Assessments, including elements of Transparency International’sCorruption Perceptions Index (TI CPI) and the Economist Intelligence Unitrankings, ask ‘experts’ including senior corporate officials to rank their perceptionsof the level of corruption in various countries. Such studies rarely lead to directdollar estimates of the extent of bribery or economic impacts, especially at the sectorlevel, but they can be used as an independent variable in regression analysis toprovide evidence of correlations between highly perceived levels of corruption andpoor development outcomes. Across countries, for example, highly perceived generallevels of corruption are sometimes associated with lower spending on proxies foroperations and maintenance. Related to this, general perceptions of corruption havebeen associated with lower quality infrastructure in some studies, such as a lowerpercentage of roads in good condition and more frequent power outages (Tanzi andDavoodi, 1998). Perceptions measures designed to capture the extent of grandcorruption appear reasonably stable over time, and they correlate with a number ofobjective indicators that we might expect them to correlate with (broad measures ofdevelopment and factors such as the extent of regulation, for example). Perceptionsof corruption in a given country are also broadly correlated across different surveys

Measuring Corruption in Infrastructure 315

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

Table 1. Measuring corruption

Perceptions indicators Strengths:. Based on expert evaluation of the

extent of corruption, aggregatedfrom numerous sources

. Examples: World Bank InstituteControl of Corruption,Transparency InternationalCorruption Perceptions Index

. Available over time for a large number ofcountries (CPI available since 1995, covers 163countries)

. Scores reasonably stable across time, correlatedwith objective measures of the quality ofinstitutions

. Measures perceptions of decision-makers –regardless of underlying accuracy, perceptionsmay drive decisions on (for example) investment

Weaknesses:. Perceptions may be weakly correlated with

reality, may suffer from significant biases. Highlycorrelated with perceptions of broadergovernance

. General indicators (not sector specific)

Surveys Strengths:. Based on interview responses of

those involved in corrupttransactions

. Examples: World Bank InstituteBusiness Environment andEnterprise Performance Survey,Bangalore Citizen Report Cards

. Improved accuracy based on answers frompersonal experience of corruption

. Can provide detailed evidence on levels and typesof corrupt payments in different sectors or typesof interaction with the government

Weaknesses:. Accuracy and potential extent compromised by

need for anonymity, unwillingness to discussillegal transactions, limited individual knowledge

. Extent of survey evidence considerably morelimited than perceptions data

. Measures payments rather than impacts ofcorruption

Judicial system reports Strengths:. Based on number and type of

convictions for corruption. Measures actual cases of corruption, provides

significant detail

Weaknesses:. Unlikely to be suitable for cross-country (or

cross-jurisdiction) comparison. Bias will beintroduced by institutional environment,competence and integrity of judicial system, andcompetence of the corrupt

Indirect and outcome indicators Strengths:. Objective indicators covering fi-

nancial flows, sector outcomes. Examples: public expenditure

tracking surveys, audits, perfor-mance indicators (rollout, price,quality, losses)

. (In some cases) widely available

. (Often) covers development outcomes rather thanintermediate indicators

Weaknesses:. Will capture impact of issues connected with

governance and sector environment other thancorruption

. Can be expensive/project specific (for example,audits)

316 C. Kenny

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

(even when these survey noticeably different groups). Nonetheless, there are reasonsto believe that broad aggregate perceptions indices are inaccurate, particularly intheir use as proxies for corruption in infrastructure.

It is important to emphasise that these indices measure perceptions of corruption,not corruption itself. Across countries, the fact that the World Bank Institute (WBI)measure for control of corruption has a correlation coefficient of 0.95 with both WBIgovernment effectiveness and rule of law measures, suggests that it may be verydifficult to tease out concerns regarding corruption with broader concerns aboutgovernance in general using perceptions indices (Thomas, 2006). This is not to arguethat when respondents are asked to think about corruption, they in fact only thinkabout general levels of crisis or weak governance – there is good evidence that this isnot the case (Kaufmann et al., 2006). It is only to point out that general concernsabut a country are likely to play a significant role in responses to a question which isbased on perceptions of an opaque, amorphous subject.2 Argentina’s TI CPI rankdropped precipitously from 5.2 in 1995 to 2.8 in 2002, for example (Seligson, 2005).This may reflect declining governance, or perceptions may have been altered by thefinancial crisis, which began to unfold in 1999.

It is not clear that corruption perceptions indices are good ‘leading indicators’ ofcorruption. In Peru, tapes of the head of the National Intelligence Service bribinglegislators and others precipitated a significant drop in the country’s CorruptionPerceptions Index, from 4.4 in 2000 to 3.5 in 2004, but this came after the tapes werereleased. There was no significant change in the index prior to the release of thetapes. The 1998 and 1999 rankings were 4.5, placing Peru ahead of the CzechRepublic and South Korea, for example, even as over 1,600 Peruvians were receivingbribes from the intelligence service. It is worth noting that the CPI continueddropping even as national polls suggested that the percentage of transactions whichinvolved paying a bribe fell from 6.4 to 4.5 over 2002–2004 (Ausland and Tolmos,2005). This evidence that the CPI acts as a ‘lagging indicator’ even during perhapsthe largest corruption scandal in recent history is a concern, if the index is to be usedto guide policy and investment decisions.3

Where we have estimates of the signal to noise ratio in corruption perceptions, theresults are not reassuring. Olken (2006) finds that villager perceptions of corruptionin village road projects in Indonesia are correlated with objective measures ofcorruption estimated from expenditure tracking and physical audits of the roads(‘estimated corruption’) but the correlation accounts for little of the variation inperceptions. Personal characteristics were significantly more correlated withcorruption perceptions than were levels of objectively estimated corruption. Beliefsabout corruption in Indonesia were strongly correlated across different levels ofgovernment and better-educated and male respondents were much more likely toreport corruption. Very significant as a determinant of perceived corruption at theproject level was ethnic heterogeneity. This bias was particularly important becausedirectly estimated corruption was significantly negatively related to measures ofheterogeneity. In other words, even at the level of the village project, perceptions ofgrand corruption appear to be a weak guide to actual levels of corruption andsubject to systemic biases.

Turning to survey techniques, these can question victims directly about the extentand level of corruption they face. At the level of petty corruption, Davis (2004) used


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

a survey approach to estimate that the average speed payment or bribe made to getconnected to piped water in India works out at $2.64 per legal customer (see alsoSeligson, 2005). There are other firm and customer surveys that have includedquestions on the extent of petty corruption in infrastructure service provision.Approximately 20 World Bank Business Environment Surveys have askedinfrastructure-related questions regarding the need to pay ‘gifts’ in order to getwater, electricity or phone connection, for example.4

With grand corruption, we have no cross-country comparable data of the extent ofbribery in infrastructure firms. This reflects in part the concentrated nature ofinfrastructure provision which adds complexity to ensuring the anonymity of surveyresponses. We do have data on corruption in construction contracting in somecountries (reported in the next section) but even this is limited. The approach ofsurveying firms which are likely to be directly involved in grand corruption appearslikely to produce more accurate measures than perceptions indices. The averagerespondent is likely to have personal experience of the transactions that can becomecorrupted. Nonetheless, survey answers can involve some uncertainty. Hendersonand Kuncoro (2006) suggest that differences in survey design and technique accountfor the difference in estimates of corrupt payments between 10.5 per cent of costsfound in their survey of Indonesian firms and 3 per cent of profits found by theIndonesian Annual Survey of Medium and Large Enterprises. The problem ofaccuracy may be magnified when questions are less specific, or asked about ‘firmslike yours,’ or levels of corruption in general.Beyond perceptions and survey measures, indirect estimates can use measures of

losses as a proxy for the extent of corruption, potentially capturing the impact ofboth petty and grand corruption. In Andhra Pradesh, transmission and distributionlosses were reduced from 38 to 26 per cent during 1999–2003, in large part throughtheft control and the regularisation of 2.25 million unauthorised connections. Thisstrongly suggests that corruption was significantly linked to losses in this case. InBangladesh and Orissa, in India, around 55 per cent of generated power is paid for,the rest is lost to technical and commercial losses. Of this, perhaps 15–18 per cent isaccounted for by ‘true technical’ losses, suggesting leakage due to illegal connectionsor under-billing accounts for as much as 30 per cent of generated power (Gulati andRao, 2006).5

A related approach looks at levels of outputs compared to inputs in construction.Regarding grand corruption at the local level in infrastructure, Olken (2006) usedmeasures of reported physical inputs and costs, surveyed labour inputs and costs andphysical audits of outputs to determine that about 24 per cent of expenditures in anIndonesian road construction project were ‘lost.’ Canning and Fay (1996) reportvariations in the cost of construction of a kilometre of similar road that vary by asmuchas 5 to 10 times.Much of this will be due to differences from factors including location,some will also be due to less efficient, more corrupt procurement practices.6 Publicexpenditure tracking surveys (which track the flow of resources through layers ofgovernment bureaucracy) are a potential approach tomeasuring themisappropriationof government funds if combined with unit-cost and quality of service data on finaloutputs. They have primarily been used in social sectors to date. These surveys havefound significant leakage of funds, such as between 30–76 per cent of non-wage fundsfor primary education in African countries (Reinikka and Svensson, 2006).

318 C. Kenny

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

There are a number of different ways to attempt to measure infrastructurecorruption and its impact, although existing evidence is fragmentary and open towide margins of error. As the following sections discuss, evidence to date suggeststhat these measures may not be able to provide any strong degree of accuracyregarding the extent and impact of corruption.

III. Evidence from BEEPS

The Business Environment and Enterprise Performance Survey (BEEPS) is thelargest and most detailed cross-country survey which included a wide range ofquestions regarding both petty and grand corruption. It covers over 4000 firms in 22transition countries and was conducted for the first time in 1999–2000 (details of thesurvey can be found in Hellman et al., 2000). The survey examines a wide range ofinteractions between firms and the state.7 The assumption of the survey is that manyof the interviewed firms will be directly involved in corruption, although (for obviousreasons) survey questions tend to revolve around the corrupt activities of ‘a typicalfirm in your industry’ rather than asking directly about corrupt activities of therespondent firm.

In 1999, the median firm in BEEPS reported spending one to two per cent of itsrevenues on unofficial payments to public officials.8 At the aggregate level, acrossthe region, the average firm suggested that it divided up its illicit payment budgetas follows: 28 per cent to deal with licenses, health and fire inspections, 18 per centon tax-related issues, 15 per cent on securing government contracts, 12 per cent fordealing with customs, 11 per cent to facilitate connections to utilities and two percent to influence the design of legislation or regulation.9 Such results suggest thatpetty bribery for infrastructure connections is somewhat of an issue in the regionand we will see that there is evidence to suggest construction industries areparticularly susceptible to corruption in licensing, taxation and obtaininggovernment contracts, in turn suggesting that infrastructure investment may bean area of concern.

Significant variation within countries in the reported level of petty infrastructurecorruption is suggested by the 1999 BEEPS survey. The survey asks respondents howoften firms like theirs have to bribe to get connected to public services such astelephone and electricity connections, with answers ranging from between ‘always’(given a value of one) and ‘never’ (given a value of six). Figure 1 displays the averageand standard deviation of answers to this question for the countries, compared to theaverage across all countries. As can be seen, only in the case of Estonia does thestandard deviation not overlap with the cross-country average.

There are notable country differences (the percentage of firms answering ‘never’varies from between 31% in Ukraine and 92% in Estonia). Nonetheless, thevariation in answers within countries is considerably larger than the variation acrosscountries, to the extent that the great majority of average country responses areunlikely to be statistically significantly different from each other.10 This may wellreflect different interpretation of the intermediate categories (mostly, frequently,sometimes, seldom), but it is also likely to reflect the fact that different types of firms,in different parts of a country, face different risk of infrastructure corruptionvictimisation.


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

Figure

1.Averageandstandard

deviationacross

countriesoffrequency

ofbribes

forconnectionto

publicservices

(1¼Always,6¼Never).Notes:

X-axiscrosses

ataveragevalueforansw

er,chart

displaysaverageandstandard

deviationforeach

country.

320 C. Kenny

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

Turning to grand corruption, amongst the firms surveyed by BEEPS were 376construction companies. The BEEPS data provides evidence on what constructionfirms believe is typical pay-off as a percentage of the contract value made for securinga government contract in their industry. From this we can estimate that the averageperceived pay-off for a government construction contract in the region is aroundseven per cent of the contract value – 76 per cent of firms suggest bribes greater thanzero but less than ten per cent (although we will see this number is uncertain).11 Itappears that construction firms in the sample think it is more common to pay bribesin their industry than firms in other industries, that firms like theirs spend a largerpercentage of revenues on bribes, and they bribe more frequently to get licenses, dealwith taxes and get contracts. These results are highly statistically significant, againsuggesting the utility of the survey approach (Kenny, 2007).

It is worth noting that the BEEPS data suggests that there is a weak link betweencross-industry estimates of corruption and estimates of corruption given by the subsetof construction industries at the national level (see Table 2). Regressions, usingnational averages across 26 countries, suggest that there is no statistically significantcorrelation in average responses between construction firms and all firms in answer toquestions regarding the propensity to bribe in general, the level of corrupt payments asa percentage of revenues or the size of payments used to secure government contracts.The BEEPS survey was not designed to provide a large dataset for exploringconstruction alone so it is perhaps unsurprising that, partially as a result, evidenceregarding the variation in corruption across countries in the sector is not strong.

One of the BEEPS questions is ‘when firms in your industry do business with thegovernment, how much of the contract value would they typically offer in additional

Table 2. BEEPS: correlation between average country answer all firms and construction firmsfor three corruption variables

Constant

Averageacross

industries

It is common in my line of business to have to paysome irregular ‘additional payments’ to get thingsdone (1¼ always, 6¼ never)

Coefficients 2.44 0.36P-value 0.08 0.27Adj R2 0.01

N 26

On average, what percentage of revenues do firmslike yours typically pay per annum in unofficialpayments to public officials? (1¼ 0%, 7� 25%)


N 25

When firms in your industry do business with thegovernment, how much of the contract valuewould they typically offer in additional orunofficial payments to secure the contract?(1¼ 0%, 6� 20%)


N 24

Note: Dependent variable is the average country answer for construction firms, independentvariable is answer for all firms, N indicates number of observations.


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

or unofficial payments to secure the contract?’ Answers, expressed as a range ofpercentages of the contract value, are on a six point scale from zero to abovetwenty per cent. Taking construction firms in the 1999–2000 BEEPS dataset whichanswered this question, we can see how much the answers tell us about sector-levelcorruption using country dummies. All but nine firms out of 143 report some levelof bribery. How much of the variation in estimates of industry corruption by firmsin a particular industry across countries is explained by which country therespondent is in? If construction firms were perfectly informed about the typicallevel of corrupt payments to government in their industry in their country and theyunderstood and answered the question in the same way, we would expect 100 percent of the variation in answers to be explained by the country of residence of therespondent. In fact, around 14 per cent of the variation can be explained, and noneof the country dummies is statistically significant (Table 3, see Figure 2 for cross-country variation compared to standard deviations).12 This performance is similarif we limit the analysis to private firms or to private firms which deal with thegovernment.A similar finding applies to the answer to the question ‘how often do firms like

yours nowadays need to make extra, unofficial payments to government officials togain government contracts?’ – which is available for a considerably larger number offirms (Table 4). The variation explained by country dummies, which one would hopeto be very high, is in fact very low (with an R-squared of 0.1). The results apply evenif we remove those firms that say companies never pay such bribes, suggesting that

Table 3. Construction firms in BEEPS: reported corruption, significance of country dummies

All firms Private firms

Private firmswhich trade withGovernment

Coefficients P-value Coefficients P-value Coefficients P-value

Intercept 2.91 0.00 3.00 0.00 3.00 0.00Azerbaijan 0.71 0.08 0.33 0.51Czech Republic 0.09 0.83 0.00 1.00 0.43 0.61Estonia 70.51 0.19 70.69 0.09 70.67 0.41Poland 70.64 0.13 70.73 0.09 70.88 0.29Russia 70.48 0.15 70.52 0.15 70.52 0.49Slovakia 71.00 0.29Turkey 70.11 0.77 70.26 0.49 70.13 0.87Ukraine 70.24 0.53 70.21 0.60 70.10 0.90Uzbekistan 70.69 0.08 70.70 0.11 70.67 0.41

(Slovakia excluded) (Slovakia excluded) (Azerbaijan excluded)Adj R2 0.09 Adj R2 0.02 Adj R2 0.02

Observations 143 Observations 125 Observations 87

Note: Construction firms, sub-sample of countries with more than 10 construction companies.Question is: ‘When firms in your industry do business with the government, how much of thecontract value would they typically offer in additional or unofficial payments to secure thecontract?’ Answers coded from 1¼ 0% to 6� 20% (Don’t know/Don’t do business withgovernment excluded).

322 C. Kenny

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

reticence about reporting such behaviour does not lie behind the result. Overall, thissuggests that we can say little with statistical confidence about which countries inthe 1999 BEEPS dataset have more corrupt construction industries than average,

Figure 2. Average and standard deviation across countries of construction company contractbribes of government contracts. Notes: X-axis crosses at average value for answer, chart

displays average and standard deviation for each country.

Table 4. Construction firms in BEEPS: corruption frequency, significance of country dummies

Full sampleExcludingnever (6)

Dummy (never¼ 0,other¼ 1)

Coefficients P-value Coefficients P-value Coefficients P-value

Intercept 5.32 0.00 5.00 0.00 0.27 0.01Azerbaijan 71.15 0.01 71.75 0.23 0.39 0.00Estonia 70.47 0.31 71.38 0.35 0.21 0.12Georgia 70.07 0.88 71.00 0.50 0.10 0.46Hungary 0.63 0.20 70.22 0.13Kyrgyzstan 71.50 0.33Latvia 70.20 0.67 71.44 0.33 0.09 0.53Poland 70.23 0.60 71.73 0.24 0.06 0.64Russia 70.37 0.34 71.93 0.18 0.08 0.46Turkey 71.57 0.00 72.24 0.12 0.42 0.00Ukraine 70.52 0.22 72.69 0.07 0.05 0.68Uzbekistan 70.28 0.54 71.50 0.31 0.11 0.41

(Kyrgyzstan excluded) (Hungary excluded) (Kyrgyzstan excluded)Regression Statistics Regression Statistics Regression Statistics

Adjusted R2 0.07 Adjusted R2 0.03 Adjusted R2 0.07Observations 361 Observations 143 Observations 361

Note: Construction firms, sub-sample of countries with more than 20 construction companies.Question is: ‘How often do firms like yours nowadays need to make extra, unofficial paymentsto public officials to gain government contracts?’ Answers coded from 1¼Always to 6¼Never(Don’t know excluded).


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

based on questions asking construction firms themselves about how corrupt theirindustry is.The variation is large enough to show that construction in Azerbaijan (the worst

performer) is significantly more corrupt than the best countries (Uzbekistan, Polandand Estonia)13 and the survey results are useful for econometric analysis of the causesand consequences of corruption. However, the survey was not designed for, andwould not be suitable for, strongly differentiating levels of corruption within sectorsacross countries. Questions are not exact and are open to subjective interpretation;one cannot expect one person in a company to have perfect knowledge of companyrevenues, contract sizes and, in particular, the size of bribes paid. Furthermore, thereare many different types of construction firm, and they will frequently be workingwith different levels of government or different departments within those levels.However, if these factors are what account for the variability of responses, it suggeststhe danger of assuming one indicator can accurately gauge levels of corruption evenregarding one distinct activity at the level of the sector (let alone all activities at thelevel of the country). In turn, this suggests the danger of using even survey evidence asan ‘actionable indicator’ of levels of corruption at the sector level.

IV. Contrasting Perceptions and Survey Evidence

Despite these concerns, survey data is likely to be more reliable than perceptionsdata, and so it is worth comparing the two to measure the accuracy of generalperceptions scores as a proxy for corruption at the sector level. The results suggestthe need for caution in the use of such perceptions indicators, although this is on thebasis of small samples. Svensson (2005) notes that cross-country survey evidenceregarding incidence of bribes is not significantly correlated with expert perceptionsonce GDP per capita is taken into account. This result is confirmed by looking at ourBEEPs data and at the sector level.We have seen that survey evidence suggests that corruption in construction and

infrastructure is weakly correlated with general corruption levels. Given in additionthe weak relationship between surveyed general corruption levels and perceptions ofcorruption, it is perhaps unsurprising that there appears to be no link betweensurveyed measures of infrastructure corruption in BEEPS and general perceptionsmeasures. Regarding country averages for the question ‘how often firms like yoursneed to make extra, unofficial payments to gain government contracts?’ fromBEEPS, answers to this question are correlated only weakly with TransparencyInternational rankings for all firms, and not at all correlated with TI rankings whenusing averages from a subset of construction firms. There is a stronger relationshipbetween TI rankings and all-industry country averages regarding the need to pay forlicenses and permits (Table 5). This confirms Knack’s (2006) demonstration thatTransparency International corruption perceptions rankings correlate far morestrongly with petty corruption questions in the BEEPS data than with grandcorruption (the TI measure correlates better with payments for utility connectionsand licenses than with payments for government contracts, for example). Looking atpetty corruption in infrastructure in particular and general CPI measures, the (little)available data actually suggests no significant correlation between perceptions ofcorruption in getting an electricity connection and CPI scores.14

324 C. Kenny

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

Similar results hold more broadly when looking at a global dataset of businessenvironment surveys (Table 6). These surveys, which cover 51 developing countries,report on the size of bribes required to secure government contracts and thepercentage of managers who rank corruption as a significant impediment to doingbusiness (details of the survey are presented in Batra et al., 2003). Using countryaverage data, TI CPI scores and data on GDP per capita, the results suggest thatsurveyed corruption in contracting is related to the percentage of managerssuggesting that corruption is a serious issue. There is also a link between thepercentage of managers listing corruption as a significant problem and TI CPIscores, although the correlation is weaker than that between survey results and GDPper capita. Furthermore, the level of bribes paid in government contracting reportedby firms at the national level is insignificantly related to TI CPI scores.

V. It’s Not How Much You Divert, But How You Divert It That Matters

Even were data on the size and frequency of payments significantly improvedthrough an expanded effort to mount corruption surveys covering infrastructurein detail, a focus on such payments may underestimate and misplace theeconomic damage done by corruption in infrastructure projects. One source ofmis-estimation is to confuse the financial and economic costs of bribe paymentsthemselves. Payments are not a ‘deadweight loss,’ in that bribe recipients can anddo spend the money. More importantly, the major damage done by corruption isprobably not the narrow financial loss of bribe payments but the economic costin terms of skewed spending priorities, along with substandard construction andoperation.

Table 5. BEEPS and transparency international CPI

ConstantIndependentvariable Adjusted R2

DV: How often do firms like yours need to makeextra, unofficial payments to gain governmentcontracts? (1¼ always 6¼ never)

IV: Transparency International rankingSample: country averages (26 obs)

5.55* 70.005** 0.17

DV: How often do firms like yours need to makeextra, unofficial payments to get licences andpermits? (1¼ always 6¼ never)

IV: Transparency International rankingSample: country averages (26 obs)

5.45* 70.006* 0.32

DV: How often do firms like yours need to makeextra, unofficial payments to gain governmentcontracts? (1¼ always 6¼ never)

IV: Transparency International rankingSample: construction firm country averages(26 obs)

4.30 0.005 0.00

Notes: DV is dependent variable, IV is independent variable; *, **, ***indicate significant at 1,5, 10 per cent respectively.


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

Imagine a road project that costs $1 million to build but generates $320,000 ineconomic returns each year after construction for 10 years. The project’s overalleconomic rate of return (ERR) is about 30 per cent (the average return for WorldBank transport projects completed in the recent past). If the project had sufferedfrom collusive bidding, and this had raised the price of construction by 20 per cent,to $1.2 million, the project’s ERR would drop to 26 per cent.15 This is a significantdecline but it still leaves the project at more than double the traditional ‘hurdle rate’of a ten per cent ERR.Imagine instead that the bidder agreed a contract price of $1 million but used

insufficient and substandard materials to build the road, spending only $800,000 onconstruction and pocketing the remaining $200,000. This reduces the road’s trafficcapacity so that yearly economic returns fall by a quarter. It also shortens the usefullife of the road to five years. This would reduce the overall ERR to 15 per cent.16 Thesame financial level of corruption has a considerably larger economic impact in thiscase, reducing the ERR by 15 percentage points rather than four.What if construction firms had used $200,000 to pay off legislators to divert money

from operations and maintenance funds for the construction of this road, andrecouped their expenditure through overbidding or poor construction? At this point,with the new road project sucking up resources from maintenance of existing roads,

Table 6. Investment climate questions on contract corruption: correlations

ConstantIndependentvariable Adjusted R2

DV: Value of gift expected to secure governmentcontract (% contract value)

IV: Transparency International CPISample: country averages (51 obs)

3.41* 70.30 0.00


IV: % managers ranking corruption as a majorconstraint to doing business

Sample: country averages (44 obs)

70.11 0.08* 0.34


IV: GDP per capita (log)Sample: country averages (50 obs)

6.65*** 70.50 0.01

DV: Transparency International CPIIV: GDP per capita (log)Sample: country averages (50 obs)

73.04* 0.74* 0.42

DV: % managers ranking corruption as a majorconstraint to doing business

IV: GDP per capita (log)Sample: country averages (43 obs)

131* 711.6* 0.25

DV: % managers ranking corruption as a majorconstraint to doing business

IV: Transparency International CPISample: country averages (48 obs)

58.9* 78.42* 0.17

Notes: DV is dependent variable, IV is independent variable; *, **, ***indicate significant at 1,5, 10 per cent respectively.

326 C. Kenny

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

reducing traffic capacities and road life across the network, economic rates ofreturn might turn negative. In short, the nature of corruption can have asignificant impact on the economic damage caused by that corruption, even if thefinancial losses are the same size. Bribery payments which are made to deliverquality projects and services at inflated prices may be far less damaging thancorrupt activities which impact the quality of delivery or the type of projectsundertaken.

The significant difference between the scale of payments and the developmentimpact of those payments may help to account for a further weak correlationbetween surveyed bribe payment levels and sector outcomes. The limiteddevelopment impact of petty corruption in infrastructure, for example, may helpto explain the apparently weak relationship between both perceptions and surveyedpetty corruption levels in infrastructure and infrastructure outcomes. Table 7 reportson correlations between answers to Business Environment Survey questions(the percentage of firms who say gifts are required for connections to infrastructure)and sector outcome indicators controlling for GDP per capita. There is apositive and significant link between the reported extent of petty corruption intelecommunications and the waiting list for a telephone mainline. Otherwise, therelationship between our indicators of infrastructure outcomes and measures ofpetty corruption are insignificant.17 It should be emphasised that all of these resultscover a very small number of observations (11–23 countries).

Again, general perceptions measures are not robustly correlated with infra-structure outcomes. For example, Estache et al. (2006) report that a general measure

Table 7. Investment climate questions on infrastructure corruption: correlations

ConstantLog GDPper capita Corruption

AdjustedR2 N

DV: log mobile phones/1,000IV: firms expected to give gifts to geta phone connection

73.16** 1.00* 70.00 0.72 21

DV: waiting list for telephonemainlines

IV: firms expected to give gifts to geta phone connection

71.52 0.18 0.02* 0.61 11

DV: % population with access toimproved water source

IV: firms expected to give gifts to geta water connection

77.32 11.4** 70.34 0.51 15

DV: % managers who rank electricityas a major constraint to doingbusiness

IV: firms expected to give gifts to getan electricity connection

123* 712.0** 0.32 0.77 13

DV: % of roads pavedIV: Value of gift expected to securegovernment contract

1.28 0.34** 70.16* 0.44 33

Notes: DV is dependent variable, IV is independent variable; *, **, ***indicate significant at 1,5, 10 per cent respectively; GDP per capita data for 2002 in PPP from WDI.


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

of perceived country-level corruption is associated with lower energy use. At thesame time, they found telecoms access positively associated with perceivedcorruption while measures of access to water were not correlated either way withperceived corruption.18

To sum up, existing measures of corruption available across countries and regionstend to be based on perceptions, which tend (if anything) to measure the extent of thepetty corruption that interviewees experience directly. Again, we have seen that mostindividuals and even firms appear only partially informed as to the extent of ‘grand’corruption involving officials and contractors. Construction and infrastructurecorruption varies considerably across and within countries with similar scores onoverall country-level measures of corruption. And finally, all of the usual measuresof corruption relate to the extent and scale of bribe payments. They do not provideinformation on the type of bribe payment, the level of theft of materials, theconstruction codes being ignored and so on. Without such measures we can say littleabout the extent of the most damaging forms of corruption in infrastructure acrossdifferent countries.

VI. What Does This Mean for Measurement of Corruption in Infrastructure?

Of course, measures of the most damaging forms of corruption will be some of thehardest to uncover. They are not part of everyday experience, so they are difficult topick up in general surveys. Even large-scale company surveys such as BEEPS do notyet tackle the non-financial elements of corruption. The response should not be to‘live with the measures we have,’ however, as these are likely to be poorly designed todetermine the extent of the problem and the success of remedial actions ininfrastructure. One approach would be to look at the most damaging consequencesof corruption, which are more easily measurable.Minimising the damage done by corruption involves countering the incentives to

build the wrong thing, and to build and then operate it badly. If this is our concern,we should focus attention on macro-sector issues such as overall budgeting andproject selection and on physical auditing of the status of physical capital. We shouldsee if budgets are adequate (and paid) for operations and maintenance, ifmaintenance is actually carried out in a way that preserves infrastructure quality,if the process for selecting projects picks those with high economic return and if thenew infrastructure is well constructed. If all of these conditions are met, we will knowthat the impact of corruption in infrastructure on overall development will becomparatively small. This suggests a focus on medium-term expenditure frame-works, public expenditure tracking surveys and physical audits as key corruptionmeasurement tools in infrastructure.An advantage of such an approach is that it may be easier to find good metrics

than in the case of sector- and type-specific direct corruption indicators. Rather thanrelying on perceptions or expert intuition, an approach that relies on inputs andoutputs can use objective indicators. We have good benchmarks for the cost ofmaintaining different classes of road, for example. Is the national budget puttingaside adequate resources to fully maintain the country’s road network based onthose benchmarks? We have many years of experience in project evaluation which

328 C. Kenny

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

should allow us to determine if project selection in a given sector appears to befollowing rational procedures or is driven by other concerns. It is a comparativelysimple engineering task to determine if a road or pipeline has been constructed andmaintained adequately or poorly through a physical audit.19

Such a measurement approach would involve errors of commission andomission. The ‘wrong’ project can be selected on the grounds of political interestand poor construction or maintenance can be the result of incompetence andinefficiency. Nonetheless, the project still remains the ‘wrong’ project and poorconstruction and maintenance remains a significant drag on developmentperformance whatever its cause. Furthermore, as suggested by the very highcorrelations between the perception-driven measures of corruption and governancethat we have, disentangling different types of governance failure will be a verydifficult, and quite possibly mistaken, exercise. As a result, the errors ofcommission are comparatively minor in terms of the indicators’ abilities tomeasure overall development impact. Regarding errors of omission, it may be thatwell-built projects involve corruption which raises the price of construction or thatconsumers accessing well-maintained networks still have to make speed paymentsor are illegally connected. This is, of course, a legitimate issue of concern.Measuring contract prices against benchmarks and business and customer surveyscan play an important role in uncovering and measuring such corruption.However, it is worth repeating that this type of corruption is likely to be lessdamaging in terms of development impact.20

At the same time, we should extend and expand survey work, increasing thefocus on petty corruption in infrastructure and grand corruption in construction.Accuracy will be improved if surveys are either limited to, or large enough to berepresentative for, a particular sub-sector, clients of a particular firm or suppliersto a particular ministry or department (indeed BEEPS have doubled in size sincethe 1999 survey used in this paper). With petty corruption, the role for increaseduse of consumer scorecards and other consumer survey methods is considerable.There may also be a role for survey/interview instruments aimed at internationalinfrastructure providers that might allow for anonymous data collection regardinggrand corruption in infrastructure provision itself. These survey results will bevery useful in cross-country statistical analysis of the success of interventions,where techniques can be utilised to take account of the considerable noise inunderlying data.

Even with more survey work, however, it is likely that if we want measures for‘progress’ in sector level anti-corruption efforts aimed at grand corruption ininfrastructure, in particular, we should again use more easily measured outputindicators which have been related to corruption variables rather than thevariables themselves (measures such as percentage of roads in good condition,transmission and distribution losses). Output measures may better capture thedevelopment impact of corruption in infrastructure and greater ease ofmeasurement for such indicators will allow for more accurate determination ofchange over time.

Regarding the use of corruption perceptions indicators in general and their use foranalysis and policymaking at the cross-sector level, the results here do suggest some


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

need for caution. The relative stability of perceptions measures over time and thestrong correlation between them suggests that they do capture some underlyingfeature of reality. What remains at issue is how closely that feature of reality isconnected with actual levels of corruption at either the sector or the macro level. Ininfrastructure, we have some reason to doubt too strong a connection. Theconnection may be stronger in other sectors but this is an issue worth furtherexamination before using perceptions as accurate measures of corruption in analysisor (even more) to guide policy decisions.

Acknowledgements

This paper does not necessarily reflect the views of the World Bank, its ExecutiveDirectors or of the countries that they represent. Thanks to Antonio Estache,Jonathan Halpern, Laszlo Lovei, Gregory Kisunko, Todd Moss, Oliver Morrissey,Tina Soreide, Richard Messick and an anonymous reviewer of the journal forcomments and, in particular, Jim Anderson both for detailed comments and forpointing out a significant error in an earlier draft. Remaining errors and opinions aremine.

Notes

1. Based on investment and maintenance estimates from Fay and Yepes (2003).

2. Aggregating perceptions scores from different sources does not necessarily improve accuracy.

Variation in sources for country scores within a given year and across years and lack of independence

between sources, both increase the magnitude of variation between scores required to declare a

‘statistically significant’ difference in perceived corruption – a variation which is already quite large. It

also makes comparing scores for a country over time problematic (Arndt and Oman, 2006; Knack,

2006).

3. Exacerbating this problem and helping to explain the continuing decline in Peru’s CPI rank as other

indicators suggested the corruption situation was improving, is that the CPI often incorporates data

that is two to three years old.

4. Accessed at: http://www.enterprisesurveys.org/

5. Similarly, Davis (2004) suggests that unaccounted for water accounts for 35 per cent of total flows in

India.

6. The ongoing World Bank effort to build a database of road construction and rehabilitation costs

should help to provide benchmarks against which to estimate excess costs of construction in transport,

accessed at: http://www.worldbank.org/transport/roads/rd_tools/rocks_main.htm

7. Accessed at: http://info.worldbank.org/governance/beeps/

8. This amongst firms which reported a percentage and did not answer ‘don’t know’.

9. These are unweighted country average responses.

10. The average of within-country standard deviations is 1.2 compared to the standard deviation of

country averages which is 0.4.

11. This assumes mid-point values for the data ranges (2.5% for the 0–5% range, for example) and 30 per

cent for answers of ‘above 20 per cent’. The lowest possible estimate (assuming 0% for the 0–5%

range, and so on) is four per cent, the highest (assuming 5% for the 0–5 range and so on, and 100% for

the ‘above 20%’ answer) is 10 per cent.

12. The F-stat. is 1.43.

13. It is also worth noting that this situation may be considerably better with the larger samples of the

2005 dataset.

14. A 21 observation regression produced the following result: (% firms expected to give a gift to get an

electrical connection)¼ 22 71.56*(TI CPI). The CPI does not enter significantly at 10 per cent.

330 C. Kenny

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

http://www.enterprisesurveys.org/

http://www.worldbank.org/transport/roads/rd_tools/rocks_main.htm

http://info.worldbank.org/governance/beeps/

15. This (and subsequent calculations) view the corrupt payment as a transfer but accounts for a (high)

marginal cost of government funds lost to corruption of 1.50 (a 50% deadweight loss).

16. This is approximately the economic impact of poor road construction suggested by Olken (2004).

17. Although there are better dependent variables for water, non-technical losses would be a better

dependent variable for electricity (again, this is not available for as many countries).

18. These results, positive and negative alike, are open to all of the usual concerns with econometric

exercises regarding questions of causality and the stability of coefficients in the presence of

multicolinearity and omitted variables.

19. For example, in the Philippines, physical audits combined with a GIS system are being used to

determine if roads and bridges actually exist and what state they are in as part of a drive towards

improved transport governance. Furthermore, especially at the level of project selection and

measurement of infrastructure quality, there is no need to survey 100 per cent of proposed projects or

infrastructure stocks – a random representative sample would suffice to suggest if the sector is

performing well.

20. Of course, this will depend on the level of such payments – if they start doubling contract prices, for

example, they will become a serious issue but, luckily, also much easier to spot.

References

Arndt, C. and Oman, C. (2006) Uses and Abuses of Governance Indicators. OECD Development Centre

Studies Series (Paris: OECD).

Ausland, A. and Tolmos, A. (2005) Focus on corruption: how to secure the aims of decentralization in

Peru by improving good governance at the regional level. Mimeo, Kennedy School, Harvard

University, Boston.

Batra, G., Kaufmann, D. and Stone, A. (2003) Investment Climate Around the World: Voices of the Firms

from the World Business Environment Survey (Washington DC: World Bank).

Campos, N. and Giovannoni, F. (2006) Lobbying, corruption and political influence. IZA Discussion

Paper 2313, IZA, Bonn, Germany.

Canning, D. and Fay, M. (1996) Infrastructure and growth. Mimeo, World Bank, Washington, DC.

Davis, J. (2004) Corruption in public service delivery: experience from South Asia’s water and sanitation

sector. World Development, 32(1), pp. 53–71.

Estache, A., Goicoechea, A. and Trujillo, L. (2006) Utilities reform and corruption in developing

countries. Mimeo, World Bank, Washington DC.

Fay, M. and Yepes, T. (2003) Investing in infrastructure: what is needed from 2000 to 2010? World Bank

Policy Research Working Paper 3102, Washington DC.

Gulati, M. and Rao, M. (2006) Checking corruption in the electricity sector. Mimeo, World Bank,

Washington DC.

Hellman, J., Jones, G., Kaufmann, D. and Schankerman, M. (2000) Measuring governance, corruption,

and state capture: how firms and bureaucrats shape the business environment in transition economies.

World Bank Policy Research Working Paper 2312, Washington DC.

Henderson, J. and Kuncoro, A. (2006) Corruption in Indonesia. NBER Working Paper 10674,

Cambridge, MA.

Kaufmann, D., Kraay, A. and Mastruzzi, M. (2006) Measuring governance using cross-country

perceptions data, in: S. Rose-Ackerman (ed.) International Handbook on the Economics Corruption

(London: Edward Elgar).

Kenny, C. (2007) Corruption, construction and developing countries. World Bank Policy Research

Working Paper 4271, Washington DC.

Knack, S. (2006) Measuring corruption in Eastern Europe and Central Asia: a critique of the cross-

country indicators. World Bank Policy Research Working Paper 3968, Washington DC.

Olken, B. (2004) Monitoring corruption: evidence from a field experiment in Indonesia. NBER Working

Paper 11753, Cambridge MA.

Olken, B. (2006) Corruption perceptions vs. corruption reality. NBER Working Paper 12428, Cambridge

MA.

Reinikka, R. and Svensson, J. (2006) Using micro-surveys to measure and explain corruption. World

Development, 34(2), pp. 359–370.


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

Seligson, M. (2005) The measurement and impact of corruption victimization: survey evidence from Latin

America. World Development, 34(2), pp. 381–404.

Svensson, J. (2005) Eight questions about corruption. Journal of Economic Perspectives, 19(3), pp. 19–

42.

Tanzi, V. and Davoodi, H. (1998) Roads to nowhere: how corruption in public investment hurts growth.

IMF Economic Issues, No. 12.

Thomas, M. (2006) What do the worldwide governance indicators measure? Mimeo, Johns Hopkins SAIS,

Washington DC.

332 C. Kenny

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

7:51

07

Oct

ober

201

4

measuring corruption in infrastructure: evidence from transition and developing countries

Documents