∗

Be careful when Doing Business

Bjørn Høyland †, Kalle Moene, and Fredrik Willumsen

1 Introduction
The World Bank report Doing Business is receiving a lot of attention from policy-makers around
the world. Recently, the director-general of the Malaysian Industrial Development Authority
was quoted saying that “Malaysia aims to move from the 24th to a top 10 position in the World
Bank’s ’Doing Business’ ranking list. We continue to ask ourselves what it will take to reach
the top 10, and are we willing to do what it takes to get there.” (Asia in Focus, Jan. 8 2007).
Similarly, Macedonia placed a one-page advert in the Economist’s annual forecasting report
“The World in 2008”, where the key message was that Macedonia had improved their position
in the World Bank’s Doing Business annual publication.
Doing Business evaluates countries’ business environment, and rank countries according to
their score on an index measuring the overall soundness of the business environment. As the
quotes above show, policy-makers have started not only to pay attention to this report, but even
to design policies aimed at doing better on the Doing Business ranking.
While it may be straight-forward to design a policy to move up on the ranking, it is not clear
that policy-initiatives that makes a country move up on the index make a real-life difference to
the underlying business environment of a country.
We warn against designing policies for the purpose of improving their relative position on the
ranking, as there are reasons for questioning the relationship between the ranking in the Doing
Business report and the underlying factor, the business environment. Instead, policy-makers
wishing to improve the business climate in their country should focus on those features that
make the largest real difference to companies considering opening or expanding their business
operations in the country. It is important to keep in mind the reasons for starting a business
may have as much to do with other factors than the business environment, for example closeness
to key input factors, e.g. raw materials, or to markets.
The way Doing Business evaluates the business environment is to pick indicators that may
be considered as signals of the underlying, but non-observable, conditions for doing business. In
this report we do not investigate whether Doing Business has chosen the best way to specify
these signals. We treat the set of possible signals as given, and investigate how the index
best can utilize this information to differentiate countries in a meaningful way. It should be
noted, however, that there may be reasons to question the set of signals and their more detailed
specifications.
Several commentators have for instance questioned whether the “employing workers indica-
tor”, in its present format, is capturing the employment climate best suited for doing business.
The indicator treats labor just as any other input, implying that extra (non-wage labor) costs
and (employment) rigidities are always bad for doing business. This view may be appropriate
in some cases, but certainly not in all. Some of the most efficient labor markets are in fact
in some way or the other regulated or influenced by organized interests, such as unions and
employers associations. The employing workers indicator might still focus only on the costs of
∗
This report has been commissioned by the Norwegian Ministry of Foreign Affairs, but the views expressed in
the report is solely the authors.
†
Corresponding author: [email protected]

1
such impacts and regulations, even in cases where they were designed solely in accordance with
business interests.
The relationship between pay and performance is an obvious example of this. Higher wages
and non-wage employment costs may promote greater employee effort and a stronger employee
commitment to the business enterprize. This may raise profits directly, and in turn reduce
labor turnover, which increase profits further by saving on other costs related to training and
social conditions in the workplace. All in all overall efficiency may go up. Similarly, protecting
employees against overwork and child labor may increase social stability and thus long run
business efficiency and work performance.
There may also be an element of mutual gift exchange between fair treatment and good
performance that is beneficial to doing business. In addition, several regulations and standard-
izations of wages induce a more stable workforce that may increase the efficiency in the use of
other factors of production, such as physical and human capital. It may increase the incentives
for training and education, which in turn lead to a better utilization of the existing business
opportunities.
Doing business with extraordinary high profits is not necessarily the same as doing business
efficiently. Some employment regulations are for instance meant to reduce firms’ local monopsony
power. Policies that raise the power of local workers may show up in higher wage costs in the
short run, while they raise overall long run performance by reducing the underutilization of the
local workforce that monopsony power entails.
We discuss the selection of indicators in section 21 . But both for the employing workers
indicator and others we do not try to redefine the indicators. As the different indexes use
different numbers of indicators, is is not clear why all indexes should be treated as equally
important for identifying viable business climates. The reason is twofold. First, some indicators
offer far more nuanced information than others. Second, some indicators measure aspects that
are far more important to business than others. The method applied in Doing Business is very
simple. This has some very real benefits. It is accessible to policy-makers. Both friends and
foes of the index can make up an informed opinion about the results. While we may accept
the claim by the authors of Doing Business that some more fancy, albeit standard statistical
scaling method do not change the results, the second section show that if the uncertainty in the
data is taken into account, it becomes difficult to tell most countries apart on the aggregated
ranking. A move of 20 or 30 places on the ranking may not reflect any real-life improvement of
the underlying business-environment of the country, it may simply be due to random noise. As
the “employing workers” sub-index has been criticized by ILO and some NGOs, we report the
effect of excluding this sub-index in 2.2.
In section 3 the economies are ranked using a statistical method that allows for the un-
certainty in the data to be accounted for. We model the uncertainty by treating each of the
indicators as an unbiased but noisy signal of the underlying business environment. By exploiting
the variation in the different indicators, we are able extract an estimate of this environment.
However, as the indicators contain different amounts of noise, and since we have a finite number
indicators, the underlying business environment will be estimated with some uncertainty. The
main advantage of this approach is that it allow us to evaluate to what extent countries placed
at different positions on the ranking, really can be said to offer businesses different conditions.
We then perform this ranking for the indicators used to rank economies in Doing Business. The
analysis is then repeated with the employing workers indicators excluded. Finally, all indicators
from the country tables are used to estimate a third ranking.
We evaluate which of the indicators that influence the ranking the most in section 4. This is
done for the indicators used to create the ranking in Doing Business as well as for all indicators.
1
We thank the World Bank team for giving access to their data and calculations. This made it straightforward
to reproduce the results and get a full overview of the coding decisions involved in this process, including which
indicators are included or excluded.

2
The results show that some of the indicators excluded from the ranking in Doing Business are
in fact useful when it comes to distinguishing the economies, which perhaps is the main purpose
of the report.
In section 5 we compare the business environment for economies at different levels of de-
velopment. We limit the discussion to three relatively coherent groups of countries, the OECD
countries, countries in Latin America and Caribbean and the countries in Sub-Saharan Africa.
The results show that the included indicators does a much better job at distinguishing among
OECD and Sub-Saharan African countries than among countries in Latin America and the
Caribbean.
Finally, we summarize the key concerns regarding the use of Doing Business as a policy tool
in section 6. First, many of the indicators presented in the main text and in the country tables
do not form the basis for neither the sub-rankings nor the overall rankings. Furthermore, the
coding decisions involved in the creation of the rankings would have benefited from being more
clearly spelled out so that independent replications could be facilitated. Second, the rankings
hide the weak discriminating powers of the indicators. If we accept that the indicators collected
by the Doing Business team do capture the underlying business environment, the indicators fail
to conclusively distinguish between a very large proportion of the economies. This renders the
firm conclusions presented in Doing Business regarding how particular economies are improving
vis-a-vis other economies questionable. Using our method the underlying data is simply not able
to clearly distinguish the quality of the business environment of an economy ranked at number
75 from one ranked at number 40 with any reasonable degree of certainty. Third, the effect of
the employing workers index on the overall ranking is marginal. Fourth, the set of indicators
collected is much more suited to distinguish between some sub-sets of economies than others.

2 Selecting the indicators

While the main body of the text and the country tables in Doing Business presents a wide
range of indicators across ten different areas, the final ranking does not exploit the information
contained in all these indicators. Unfortunately, it is not always clear from the report which
indicators are included in the final ranking, and what criteria are used for including or excluding
an indicator in the calculation of the ranking. One reason may be that data for some of the
indicators are only available for 2007. As we will argue below, however, more indicators would
give more reliable estimates.
It may be misleading that Doing Business gives the impression that a wide range of indicators
are used to generate each particular sub-indicator, when in fact the rankings sometimes include
only one, often fairly crude, measure. Consider, for example, the sub-index measuring the ease
of closing of a business. Both the publicly available dataset, and the country tables in Doing
Business, report three indicators; time, cost, and recovery rate. However, the rankings, both
the sub-index and the overall ranking, are not based on these three indicators. They are based
solely on the recovery rate. This information is presented in the data notes only. As the section
closing a business also includes discussions of both time and costs in addition to the recovery
rate, a reader not paying attention to the data notes may believe that the sub-index is made on
the basis of all three of these indicators. This may lead some policy-makers to reach the wrong
conclusions. We would hence recommend that the World Bank team made it clear, in the main
text and in the country tables which indicators that formed the basis for the rankings, and the
reasons for including or excluding the indicators in the rankings. This brings us to an related
issue, the variation in the number of indicators.

3
2.1 Variation in the number of indicators
The number of indicators included in each of the ten indexes varies from one (for credit, in-
vestments, and closing a business) to six (for trading). Most indexes (taxes, register, license,
and enforcing contracts) have three indicators, while both starting a business and employing
workers have four indicators. Under the condition that the indicators under each of the indexes
measure the same underlying concept, it is the case that the precision increases with the number
of indicators. One would thus expect that an index with many indicators is capable of measuring
the underlying concept better than an index with few indicators. By calculating the percentile
in each of the indexes and then averaging them, Doing Business treats all indexes as equally
informative as well as equally important as measures of a good business environment. In this
report we will take the assumption that the indicators measure one underlying dimension, the
“true” business environment, as given. Our method, however, unlike the method used in Do-
ing Business, estimates the influence of the different indicators, rather than assuming that the
indicators used in each of the ten sub-indexes together are responsible for 10% of the overall
business environment.

2.2 Effect of selecting only a subset of the indicators

Now, compare the overall ranking produced in Doing Business with the ranking it would have
produced if one of the sub-indexes, for example employing workers, were excluded. It is of
substantial interest to see the effect of not including this set of indicators, as it has been ac-
cused being ideologically biased. We see that Norway would have ranked as number 7 rather
than number 11 if the employing workers index was excluded. A western European country
that benefits from the inclusion of the employing workers index is Switzerland. It would have
fallen from number 16 to number 20 if the employing workers index was excluded. However, the
economies that benefit the most from the inclusion of the employing workers index are countries
like Afghanistan (+11), Haiti (+15), Bhutan (+12), Uganda (+14), Micronesia (+16), Nige-
ria (+14), Marshall Islands (+21), Palau (+24), Brunei (+23), Kiribati (+16), Samoa (+11),
Maldives (+19), Tonga (+12) and Fiji (+12). Including the employing workers index moves
Georgia from number 28 to number 18. The main losers from the inclusion of the employing
workers index are Equatorial Guinea (-12), Sao Tome and Principe (-15), Sierra Leone (-11),
Ecuador (-15), Indonesia (-11), Nepal (-12), Paraguay (-23), Croatia (-14), Ghana (-22), Serbia
(-12), Montenegro (-11), Pakistan (-14), Macedonia, FYR (-12), Panama (-12), Peru (-11), and
Slovenia (-12).
As stated there may be reasons to question whether the employing workers indicator in its
present format is capturing the employment climate best suited for doing business. Our main
point, however, is not this, but the finding that excluding the indicator does not change much
the Doing Business Index, in contrast to what some commentators may believe.

3 Uncertainty in the measurement

Most data, both surveys and government statistics contain some measurement error. Here we
take the view that all the indicators are measured perfectly, but that they provide a noisy
signal about the underlying “true” business environment. The statistical technique we are using
extract this signal by estimating the optimal weighting scheme of the indicators. We hence take
the Doing Business assumption that the business environment in a country can be characterized
by one number as given, and ask how we can utilize the information contained in the data set
to extract this number in an optimal way. Since we are dealing with noisy signals, any analysis
based on such data should take this uncertainty into account when concluding from the data. It
is important that the reader is made aware of this uncertainty, to be able to make an informed

4
decision. Here we use a standard statistical approach for accounting for uncertainty.2 In this
section, we first replicate the result from Doing Business while accounting for the inherent
uncertainty in the data.

3.1 The uncertainty of the estimates

Figure 1 plots the economies as ranked by our method. Remember, we use the same data and the
same set of indicators as Doing Business. There are several aspects worth commenting. First,
the information used in Doing Business do not distinguish well between the economies that are
not placed near the end-points of the scale. While it clearly distinguishes the best economies
from the worst, it does not do a particular good job in distinguishing between the economies
that are somewhat in between. So, while the mean rankings are similar to the results presented
in Doing Business, our estimates suggest that there are considerable uncertainty around these
estimates. One should hence be careful when inferring from the mean ranking (or the ranking
in Doing Business) to the underlying business environment. Four groups seem to be present in
the figure. 1) An handful of countries that are clearly much worse for business than all other
countries. 2) A group of about 25 economies that do better then the worst off group, but worse
than most countries. 3) A large group of more than 100 countries, amongst which it is almost
impossible to distinguish. 4) A small group of countries that clearly provides a better business
environment than most economies in the third group.

3.2 Employing workers

Figure 2 shows the ranking of the economies when the indicators from the “employing workers”
sub-index are removed. The main message from this plot is that not much is lost by removing
these indicators. The ranking is not clearly influenced by this sub-index. Several Western
European countries, however, move up on the ranking. While some countries in the middle of
the ranking jump or fall many places when these indicators are excluded, the differences between
these countries and the countries they surpass or are overtaken by, are minor in the estimated
business environment.

3.3 Using all of of the information

Figure 3 uses all information provided by Doing Business to produce the same summarizing
plot. The difference between the least favorable business climates is smaller when all available
information is used. However, we are better able to distinguish among the most favorable
economies when using all information. The figure does provide a more gradual change in the
estimated business climate than the previous figure. when using all information, it is easier to
see that there is a group of some 30-40 economies that are struggling at the lower end of the scale
and a group of about 25-30 countries that seem to provide a more friendly business environment
than the large majority of economies.

4 What matters for our ranking?

This section investigates which of the indicators that matter the most in determining the rank
of the economies, i.e. the positions in Figures 1-3. We simply investigate which of the indicators
distinguish best between the economies. The implicit assumption in Doing Business is that all
of the ten aspects are equally important. Our approach allow us to evaluate which indicators
that matter for differentiating between the countries.
This ranking is not over which aspects are the most important for doing business. It is over
which aspects that matter for distinguishing between the countries. Figure 4 shows to what
2
We account for the uncertainty using Bayesian methods.

5
 Original data

Hong Kong, China Singapore ●

●
Denmark ●
United States ●
Norway ●
Estonia ●
Sweden ●
Finland ●
Netherlands ●
New Zealand ●
Germany ●
Canada ●
Ireland ●
Austria ●
Luxembourg ●
Australia ●
Japan ●
Panama ●
United Kingdom Belgium ●
●
Israel ●
Switzerland ●
Korea ●
Iceland ●
Spain ●
Latvia ●
France ●
Lithuania ●
Malaysia Taiwan, China ●
●
Romania ●
United Arab Emirates ●
Georgia ●
Mauritius ●
Dominican Republic ●
Thailand ●
Portugal ●
Serbia ●
Turkey ●
St. Kitts and Nevis ●
Czech Republic ●
Puerto Rico ●
Hungary ●
Saudi Arabia ●
Italy Egypt ●
●
Antigua and Barbuda ●
St. Vincent and the Grenadines ●
Chile ●
Tunisia ●
Bosnia and Herzegovina Dominica ●
●
Morocco ●
Macedonia, FYR Philippines ●
●
El Salvador ●
Trinidad and Tobago Croatia ●
●
Slovenia ●
St. Lucia ●
Tonga ●
Grenada ●
Mexico ●
China ●
Sri Lanka ●
Jordan ●
Kuwait ●
Poland ●
Cape Verde ●
India ●
Bulgaria Seychelles ●
●
Liberia ●
Maldives ●
Guatemala ●
Pakistan ●
Argentina ●
Solomon Islands ●
Djibouti Costa Rica ●
●
Jamaica ●
Kiribati ●
Brunei ●
Albania ●
Montenegro ●
Vietnam ●
Greece ●
Colombia ●
Slovakia ●
Brazil ●
Algeria ●
Indonesia ●
Honduras ●
Ghana ●
Uruguay Oman ●
●
Gambia ●
Marshall Islands ●
Fiji Peru ●
●
Armenia ●
Belize ●
Papua New Guinea Syria ●
●
Togo Suriname ●
●
Senegal Samoa ●
●
Tanzania ●
Namibia ●
Timor−Leste ●
Comoros ●
Bangladesh Micronesia ●
●
São Tomé and Principe ●
Gabon ●
Belarus ●
Lebanon ●
Vanuatu ●
Guinea−Bissau ●
Guyana South Africa ●
●
Swaziland ●
West Bank and Gaza ●
Palau ●
Bolivia ●
Cameroon ●
Yemen ●
Paraguay ●
Guinea ●
Nicaragua Moldova ●
●
Mozambique Ecuador ●
●
Kenya ●
Iran ●
Nigeria Ukraine ●
●
Sierra Leone ●
Benin ●
Côte d'Ivoire ●
Bhutan ●
Madagascar ●
Botswana ●
Nepal Equatorial Guinea ●
●
Russia ●
Mauritania ●
Uganda ●
Cambodia ●
Ethiopia Lesotho ●
●
Sudan ●
Lao PDR ●
Malawi ●
Haiti ●
Burkina Faso ●
Mali ●
Azerbaijan Mongolia ●
●
Venezuela ●
Zambia ●
Congo, Dem. Rep. Congo, Rep. ●
●
Angola ●
Zimbabwe ●
Rwanda ●
Burundi ● 95% Credibility Interval
Niger Eritrea ●
●
Kyrgyz Republic Afghanistan ● 50% Central Tendency
●
Central African Republic ●
Kazakhstan ●
Tajikistan ●
Uzbekistan ●
Iraq Chad ●
●

Figure 1: The figure shows the ranking of the economies. The black circle indicates the median value,
the thick line gives the central tendency of their location, the thin line covers the 95 percent confidence
interval.

6
 Without EWI

Hong Kong, China Singapore ●

●
Denmark ●
Estonia ●
Norway ●
Sweden ●
United States ●
Finland ●
Germany Netherlands ●
●
New Zealand ●
Ireland ●
Austria ●
Canada ●
Luxembourg ●
Panama ●
Belgium Japan ●
●
Switzerland ●
Australia ●
Korea ●
Israel ●
Spain United Kingdom ●
●
Iceland ●
France ●
Latvia ●
Lithuania ●
Taiwan, China ●
Romania ●
Georgia Malaysia ●
●
Dominican Republic ●
United Arab Emirates ●
Portugal Mauritius ●
●
Serbia ●
Thailand ●
Turkey ●
St. Kitts and Nevis ●
Hungary Czech Republic ●
●
Egypt Puerto Rico ●
●
Saudi Arabia ●
Morocco ●
Italy ●
St. Vincent and the Grenadines ●
Tunisia ●
Macedonia, FYR Bosnia and Herzegovina ●
●
Chile ●
Antigua and Barbuda ●
Dominica ●
Philippines ●
El Salvador ●
Slovenia ●
Croatia ●
Trinidad and Tobago ●
Mexico ●
Tonga ●
St. Lucia ●
Poland ●
Jordan ●
Grenada ●
Sri Lanka ●
Cape Verde Seychelles ●
●
China ●
India ●
Bulgaria Argentina ●
●
Kuwait ●
Liberia ●
Guatemala ●
Pakistan ●
Djibouti Costa Rica ●
●
Montenegro Solomon Islands ●
●
Jamaica ●
Albania ●
Maldives ●
Kiribati ●
Greece ●
Brazil ●
Vietnam ●
Brunei ●
Colombia ●
Indonesia ●
Algeria ●
Slovakia ●
Honduras ●
Ghana ●
Oman ●
Gambia ●
Uruguay ●
Peru ●
Syria Fiji ●
●
Belize ●
Armenia ●
Papua New Guinea Marshall Islands ●
●
Senegal ●
Tanzania ●
Togo Suriname ●
●
Timor−Leste ●
Comoros ●
São Tomé and Principe ●
Namibia ●
Bangladesh Samoa ●
●
Gabon ●
Micronesia ●
Guinea−Bissau ●
Belarus ●
Vanuatu ●
Lebanon ●
South Africa ●
Bolivia ●
Guyana Swaziland ●
●
Paraguay West Bank and Gaza ●
●
Palau ●
Cameroon ●
Yemen ●
Moldova ●
Nicaragua Guinea ●
●
Mozambique Ecuador ●
●
Kenya ●
Iran ●
Nigeria Ukraine ●
●
Sierra Leone ●
Benin ●
Madagascar Côte d'Ivoire ●
●
Equatorial Guinea ●
Bhutan ●
Nepal Botswana ●
●
Russia ●
Mauritania ●
Uganda ●
Cambodia ●
Ethiopia Lesotho ●
●
Sudan ●
Lao PDR ●
Malawi ●
Haiti ●
Burkina Faso ●
Mali ●
Mongolia ●
Venezuela ●
Azerbaijan ●
Zambia ●
Congo, Dem. Rep. Congo, Rep. ●
●
Angola ●
Zimbabwe ●
Burundi ●
Rwanda ● 95% Credibility Interval
Niger Eritrea ●
●
Afghanistan Kyrgyz Republic ● 50% Central Tendency
●
Central African Republic ●
Kazakhstan ●
Tajikistan ●
Uzbekistan ●
Iraq Chad ●
●

Figure 2: The figure shows the ranking of the economies. The black circle indicates the median value,
the thick line gives the central tendency of their location, the thin line covers the 95 percent confidence
interval.

7
 All available data

Hong Kong, China Singapore ●

●
Denmark ●
United States ●
New Zealand ●
Sweden ●
Norway ●
Finland ●
Estonia ●
Canada ●
Netherlands ●
Ireland ●
Japan Germany ●
●
Belgium United Kingdom ●
●
Australia ●
Austria ●
Israel ●
Iceland ●
Korea ●
Switzerland ●
Luxembourg ●
Panama ●
Spain ●
Latvia ●
Malaysia France ●
●
Lithuania ●
Taiwan, China Mauritius ●
●
Georgia Romania ●
●
Portugal Thailand ●
●
United Arab Emirates ●
Dominican Republic Serbia ●
●
Turkey Puerto Rico ●
●
Czech Republic Hungary ●
●
Italy Saudi Arabia ●
●
Egypt St. Kitts and Nevis ●
●
Antigua and Barbuda ●
Chile ●
Tunisia ●
St. Lucia ●
Bosnia and Herzegovina ●
Slovenia ●
Tonga ●
Macedonia, FYR China ●
●
Poland ●
Sri Lanka ●
Mexico ●
El Salvador ●
Kuwait ●
Bulgaria St. Vincent and the Grenadines ●
●
Trinidad and Tobago Dominica ●
●
Croatia ●
Morocco ●
Pakistan ●
Jordan ●
Jamaica ●
Grenada ●
Philippines Solomon Islands ●
●
Costa Rica ●
Slovakia ●
Montenegro ●
Maldives ●
Brunei ●
India ●
Colombia ●
Guatemala ●
Seychelles Argentina ●
●
Ghana ●
Vietnam ●
Cape Verde Greece ●
●
Indonesia ●
Kiribati ●
Algeria Brazil ●
●
Djibouti Oman ●
●
Honduras ●
Peru ●
Fiji Uruguay ●
●
Liberia ●
Albania ●
Papua New Guinea ●
Armenia ●
Marshall Islands ●
Belize ●
Gambia ●
Samoa ●
Syria ●
Namibia ●
Bangladesh Tanzania ●
●
Togo South Africa ●
●
Lebanon ●
Suriname ●
Senegal ●
Micronesia ●
Guyana Vanuatu ●
●
Belarus ●
Gabon ●
Nicaragua Timor−Leste ●
●
São Tomé and Principe Palau ●
●
Paraguay ●
West Bank and Gaza ●
Moldova ●
Comoros ●
Swaziland ●
Yemen ●
Bolivia ●
Cameroon ●
Guinea−Bissau ●
Guinea ●
Mozambique Kenya ●
●
Nigeria Ecuador ●
●
Iran ●
Ukraine ●
Benin ●
Botswana ●
Nepal Côte d'Ivoire ●
●
Madagascar ●
Russia ●
Sierra Leone ●
Bhutan ●
Uganda Equatorial Guinea ●
●
Mauritania ●
Lesotho ●
Ethiopia ●
Cambodia ●
Sudan ●
Malawi ●
Lao PDR ●
Haiti ●
Mongolia ●
Burkina Faso ●
Zambia ●
Mali ●
Azerbaijan ●
Venezuela ●
Angola Congo, Rep. ●
●
Congo, Dem. Rep. Zimbabwe ●
●
Niger ●
Eritrea ● 95% Credibility Interval
Kyrgyz Republic Rwanda ●
●
Afghanistan Burundi ● 50% Central Tendency
●
Central African Republic ●
Kazakhstan ●
Tajikistan ●
Uzbekistan ●
Iraq Chad ●
●

Figure 3: The figure shows the ranking of the economies. The black circle indicates the median value,
the thick line gives the central tendency of their location, the thin line covers the 95 percent confidence
interval.

8
 Trading_TimeImp ● Trading_TimeImp ●

Trading_TimeExp ● Trading_TimeExp ●
Trading_costImp ●
Trading_costImp ●
Trading_CostExp ●
Trading_CostExp ● Trading_DocsImp ●

Trading_DocsImp ●
Trading_DocsExp ●
Closing_Recovery ●
Trading_DocsExp ● Credit_Sum ●
Credit_Sum ● InvestProtect_InvestProtect ●

Closing_Recovery ●
Closing_Time ●
Start_Procedures ●
InvestProtect_InvestProtect ● Closing_Cost ●
Licence_Time ● Licence_Time ●
InvestProtect_DirLiability ●
Start_Procedures ●
Start_Cost ●
Taxes_Payments ● Enforcing_Cost ●

Start_Cost ● Taxes_Payments ●

Enforcing_Cost ●
InvestProtect_Shareholder ●
Enforcing_Procedures ●
Enforcing_Procedures ● InvestProtect_Disclosure ●

Taxes_Total ● Taxes_Other ●
Employ_Rigidity ●
Employ_RigidHours ●
Taxes_Total ●
Licens_Procedures ● Employ_RigidHours ●

Licence_Cost ●
Employ_DifficultFire ●
Start_time ●
Start_time ● Licence_Cost ●
Employ_DifficultFire ● Register_Cost ●

Register_Cost ●
Licens_Procedures ●
Employ_Hire ●
Employ_Hire ● Register_Procedure ●
Register_Procedure ● Start_Capital ●

Start_Capital ●
Employ_CostFire ●
Taxes_Time ●
Employ_CostFire ● Taxes_ProfitTax ●

Taxes_Time ● Register_Time ●
Enforcing_Time ●
Register_Time ●
Employ_NonwageCost ●
Enforcing_Time ● Taxes_LaborTax ●

−1.0 −0.5 0.0 0.5 1.0 −2 −1 0 1

Figure 4: The figure shows which indicators that matter for the ranking of an economy, only Doing
Business indicators.

extent the indicators matter for the positions of the economies. We see that indicators related
to trading, access to credit and investor protection when closing down a business matter the
most for how countries are ranked. Indicators from the other groups of indicators matter less.
In fact, neither non-wage cost of employment nor labor tax seem to matter at all for how an
economy is performing on our ranking. This indicates that there are several relevant dimensions
to the business environment. But as Doing Business restricts attention to only one dimension,
we do the same. As we believe that one of the main points of the report is to distinguish the
business environments, it would by useful if the indicators where able to distinguish between
economies along the whole scale of the underlying business environment.

5 Rankings within groups

In this section we focus on three subsets of countries, those in the OECD, in Latin-America, and
in Sub-Saharan Africa. This allows us to evaluate to what extent the indicators differentiate
between relatively “similar” countries. In other words, can the indicators help businesses make
informed decisions about which countries to choose from a relatively similar group of countries.
We start by investigating the OECD countries in figure 5. The figure shows that Greece, Italy,
the Czech Republic, Spain and Portugal clearly provide a worse business environment than the
other OECD countries, assuming that the underlying data is able to capture this. There is
hardly any difference in the business environment in Denmark, Canada, the US, Ireland, and
Australia. The business environment is quite similar in the large majority of OECD countries.
The data are not conclusive on whether it is better conditions for business in the UK than in
Germany. Figure 6 compares the Latin American countries. With a few minor exceptions, there
is hardly any differences between the economies. While it is clear that Haiti and Venezuela offer
a less friendly business environment than all other economies in the region, the data do not help
use to distinguish the business environment in Guatemala from that in Mexico.
In figure 7, we focus on countries in Sub-Saharan Africa. We see that the the data do a

9
 OECD

New Zealand ●
Denmark ●
Canada ●
Ireland ●
Australia ●
United States ●
Sweden ●
United Kingdom ●
Finland ●
Norway ●
Iceland ●
Belgium ●
Switzerland ●
Netherlands ●
France ●
Japan ●
Luxembourg ●
Austria ●
Germany ●
Portugal ●
Korea ●
Spain ●
Czech Republic ●
Italy ●
Greece ●
95% Credibility Interval
50% Central Tendency
Figure 5: The figure shows the ranking of the economies, when only using OECD data. The black circle
indicates the median value, the thick line gives the central tendency of their location, the thin line covers
the 95 percent confidence interval.

Latin America & Caribbean

Panama ●
St. Kitts and Nevis ●
St. Vincent and the Grenadines ●
Dominican Republic ●
Dominica ●
St. Lucia ●
Chile ●
Grenada ●
Mexico ●
Costa Rica ●
Argentina ●
El Salvador ●
Jamaica ●
Honduras ●
Brazil ●
Belize ●
Colombia ●
Peru ●
Guatemala ●
Uruguay ●
Ecuador ●
Guyana ●
Suriname ●
Paraguay ●
Bolivia ●
Nicaragua ●
Venezuela ●
Haiti ● 95% Credibility Interval
50% Central Tendency

Figure 6: The figure shows the ranking of the economies when the model is estimated using only Latin
America and the Caribbean. The black circle indicates the median value, the think line gives the central
tendency of their location, while the thin line cover the ranks that the economy had more than 5% of
being located between.

10
very good job in identifying differences in the business environment. We see that Chad and the
Central African Republic offer a rather unfriendly business environment compared to all other
countries in the region. Mauritius and Liberia offer the best business environments in Sub-
Saharan Africa. We also see that the index is able to distinguish between two relatively large
groups of countries within this region. All in all, the underlying data collected by the Doing
Business team are not equally good at discriminating between countries in different regions.
The data do a better job in Sub-Saharan Africa than in the OECD area, and the worst job in
Latin America.

Sub−Saharan Africa

Mauritius ●
Liberia ●
Cape Verde ●
Ghana ●
Seychelles ●
São Tomé and Principe ●
Namibia ●
Togo ●
Senegal ●
Tanzania ●
Comoros ●
Gabon ●
South Africa ●
Guinea−Bissau ●
Cameroon ●
Swaziland ●
Guinea ●
Nigeria ●
Mozambique ●
Kenya ●
Sierra Leone ●
Benin ●
Equatorial Guinea ●
Madagascar ●
Côte d'Ivoire ●
Botswana ●
Mauritania ●
Lesotho ●
Uganda ●
Ethiopia ●
Sudan ●
Malawi ●
Mali ●
Burkina Faso ●
Zambia ●
Congo, Rep. ●
Congo, Dem. Rep. ●
Angola ●
Eritrea ●
Zimbabwe ●
Burundi ●
Niger ●
Rwanda ●
Central African Republic ●
Chad ● 95% Credibility Interval
50% Central Tendency

Figure 7: The figure shows the ranking of the economies when the model is estimated using only countries
in the Sub-Saharan Africa. The black circle indicates the median value, the thick line gives the central
tendency of their location, the thin line covers the 95 percent confidence interval.

6 Summary and Conclusion

Our report evaluates the methodology behind Doing Business. It pays particular attention to
two aspects:

• the uncertainty in the ability of the indicators to capture the underlying business climate,
and

• the effect of the controversial “employing workers” sub-index on the overall ranking.

By not taking uncertainty into account, we show that the rankings hide the weak discriminating
powers of the indicators. If we accept that the indicators collected by the Doing Business team
do capture the underlying business environment, the indicators fail to distinguish between a
large proportion of the economies. We also find that the effect of excluding the employing
workers index is marginal. Furthermore, we show that:

• several of the indicators presented in Doing Business are not used for rankings, and that

• coding-decisions taken before calculating the rankings are not transparent.

11
Several of the indicators presented and elaborated on in the main text and in the country tables
of Doing Business, neither form the basis for the sub-rankings nor the overall rankings they
present. In addition, the coding decisions involved in the creation of the rankings would have
benefited from being more clearly spelled out so that independent replication of the results could
more easily be done. For example, firing costs (weeks of wages), a sub-indicator in the employing
workers category, are truncated at 8 weeks, but there is no reference to this in the report.
The main message is that many countries may find it easier to change their ranking in Doing
Business than changing the underlying business environment. However, as the underlying data
distinguish rather poorly between a very large group of the economies, government and other
policy makers should refrain from letting themselves be seduced by their numerical ranking.

We recommend that the World Bank

• make it clear, in the main text and in the country tables, which indicators that form the
basis for the rankings, and the reasons for including or excluding indicators in the rankings,

• make the coding choices more transparent, by referring the alterations made to indicators
before using them in the final ranking in the main text,

• consider including the results of a standard factor analysis in the report, as the way the
index is built up is perfectly suited for this kind of analysis.

On the basis of our factor analysis, we find that

• it is very hard to distinguish a large group of countries, when taking estimation uncertainty
into account,

• it seems that indicators related to trade and credit are really important for distinguishing
the economies, while some of the other indicators have no power.

One might ask:

• is it possible to reform some of the indicators to better distinguish countries?

12
A Effect of removing “employing workers” from the ranking

Economy RANK Excluding.Workers Difference

1 Singapore 1 1 0
2 New Zealand 2 2 0
3 United States 3 4 1
4 Hong Kong, China 4 3 −1
5 Denmark 5 6 1
6 United Kingdom 6 5 −1
7 Canada 7 9 2
8 Ireland 8 8 0
9 Australia 9 11 2
10 Iceland 10 10 0
11 Norway 11 7 −4
12 Japan 12 14 2
13 Finland 13 12 −1
14 Sweden 14 13 −1
15 Thailand 15 17 2
16 Switzerland 16 20 4
17 Estonia 17 15 −2
18 Georgia 18 28 10
19 Belgium 19 22 3
20 Germany 20 16 −4
21 Netherlands 21 18 −3
22 Latvia 22 19 −3
23 Malaysia 23 25 2
24 Saudi Arabia 24 27 3
25 Austria 25 24 −1
26 Lithuania 26 21 −5
27 Mauritius 27 30 3
28 Puerto Rico 28 31 3
29 Israel 29 29 0
30 Korea 30 23 −7
31 France 31 26 −5
32 Slovakia 32 32 0
33 Chile 33 34 1
34 St. Lucia 34 39 5
35 South Africa 35 37 2
36 Fiji 36 48 12
37 Portugal 37 33 −4
38 Spain 38 35 −3
39 Armenia 39 42 3
40 Kuwait 40 45 5
41 Antigua and Barbuda 41 46 5
42 Luxembourg 42 36 −6
43 Namibia 43 50 7
44 Mexico 44 38 −6
45 Bulgaria 45 49 4
46 Hungary 46 44 −2
47 Tonga 47 59 12
48 Romania 48 40 −8
49 Oman 49 56 7
50 Taiwan, China 50 41 −9
51 Botswana 51 52 1
52 Mongolia 52 54 2
53 Italy 53 55 2
54 St. Vincent and the Grenadines 54 57 3
55 Slovenia 55 43 −12
56 Czech Republic 56 58 2
57 Turkey 57 51 −6
58 Peru 58 47 −11
59 Maldives 59 78 19
60 Belize 60 64 4
61 Samoa 61 72 11
62 Vanuatu 62 60 −2
63 Jamaica 63 67 4
64 St. Kitts and Nevis 64 73 9
65 Panama 65 53 −12
66 Colombia 66 61 −5
67 Trinidad and Tobago 67 76 9
68 United Arab Emirates 68 68 0
69 El Salvador 69 66 −3
70 Grenada 70 75 5
71 Kazakhstan 71 87 16
72 Kenya 72 71 −1
73 Kiribati 73 89 16
74 Poland 74 69 −5
75 Macedonia, FYR 75 63 −12
76 Pakistan 76 62 −14
77 Dominica 77 81 4
78 Brunei 78 101 23
79 Solomon Islands 79 85 6
80 Jordan 80 86 6
81 Montenegro 81 70 −11
82 Palau 82 106 24
83 China 83 78 −5
84 Papua New Guinea 84 94 10
85 Lebanon 85 90 5
86 Serbia 86 74 −12
87 Ghana 87 65 −22
88 Tunisia 88 79 −9
89 Marshall Islands 89 110 21
90 Seychelles 90 82 −8
91 Vietnam 91 88 −3
92 Moldova 92 84 −8
93 Nicaragua 93 92 −1
94 Kyrgyz Republic 94 93 −1

13
 Economy RANK Excluding.Workers Difference
95 Swaziland 95 103 8
96 Azerbaijan 96 97 1
97 Croatia 97 83 −14
98 Uruguay 98 98 0
99 Dominican Republic 99 95 −4
100 Greece 100 91 −9
101 Sri Lanka 101 96 −5
102 Ethiopia 102 105 3
103 Paraguay 103 80 −23
104 Guyana 104 108 4
105 Bosnia and Herzegovina 105 102 −3
106 Russia 106 107 1
107 Bangladesh 107 104 −3
108 Nigeria 108 122 14
109 Argentina 109 100 −9
110 Belarus 110 119 9
111 Nepal 111 99 −12
112 Micronesia 112 128 16
113 Yemen 113 116 3
114 Guatemala 114 111 −3
115 Costa Rica 115 115 0
116 Zambia 116 109 −7
117 West Bank and Gaza 117 114 −3
118 Uganda 118 132 14
119 Bhutan 119 131 12
120 India 120 121 1
121 Honduras 121 117 −4
122 Brazil 122 118 −4
123 Indonesia 123 112 −11
124 Lesotho 124 127 3
125 Algeria 125 123 −2
126 Egypt 126 124 −2
127 Malawi 127 126 −1
128 Ecuador 128 113 −15
129 Morocco 129 120 −9
130 Tanzania 130 125 −5
131 Gambia 131 137 6
132 Cape Verde 132 130 −2
133 Philippines 133 134 1
134 Mozambique 134 129 −5
135 Iran 135 133 −2
136 Albania 136 136 0
137 Syria 137 138 1
138 Uzbekistan 138 143 5
139 Ukraine 139 144 5
140 Bolivia 140 135 −5
141 Iraq 141 147 6
142 Suriname 142 152 10
143 Sudan 143 140 −3
144 Gabon 144 139 −5
145 Cambodia 145 145 0
146 Djibouti 146 146 0
147 Comoros 147 141 −6
148 Haiti 148 163 15
149 Madagascar 149 142 −7
150 Rwanda 150 150 0
151 Benin 151 151 0
152 Zimbabwe 152 155 3
153 Tajikistan 153 154 1
154 Cameroon 154 159 5
155 Côte d’Ivoire 155 162 7
156 Mauritania 156 161 5
157 Mali 157 165 8
158 Afghanistan 158 169 11
159 Togo 159 157 −2
160 Sierra Leone 160 149 −11
161 Senegal 161 156 −5
162 Burkina Faso 162 158 −4
163 São Tomé and Principe 163 148 −15
164 Lao PDR 164 167 3
165 Equatorial Guinea 165 153 −12
166 Guinea 166 168 2
167 Angola 167 160 −7
168 Timor-Leste 168 171 3
169 Niger 169 166 −3
170 Liberia 170 172 2
171 Eritrea 171 175 4
172 Venezuela 172 164 −8
173 Chad 173 174 1
174 Burundi 174 177 3
175 Congo, Rep. 175 173 −2
176 Guinea-Bissau 176 170 −6
177 Central African Republic 177 176 −1
178 Congo, Dem. Rep. 178 178 0
Table 1: The effect of excluding the Employing Workers Index on the overall ranking of
economies.

14