Data Inspired Insights

Tag: data (Page 4 of 4)

Women in the Workplace – Understanding the Data

Cross Posted from OpenDataKosovo.org:

Continuing our series on Gender Inequality and Corruption in Kosovo (see Part I and Part II), in Part III and the next few parts, we are going to take a detailed look at the problems women face in the labour market in Kosovo.

To do this, we will be using information from several sources, including data on participation rates, by gender, from the Gender Statistics database at the World Bank, and a range of labour market statistics from various Kosovo Labour Force Surveys, released by the Kosovo Agency of Statistics.

High Level Concepts

Before diving into the statistics, let’s first visualize and explain some of the high level concepts in labour market statistics.

Chart 1 – Population Breakdown 2014

WAC_3_1

At the highest level, the section of the population that is relevant when looking at labour market statistics is people who are of working age and are able to work. In Kosovo, this population includes all people aged 15 to 64 and is known as the ‘working age population’.

Labour Force and Inactive Populations

At the next level, the working age population can be broken down into two main subgroups – those that are considered in the labour force (i.e. ‘participating’) and those that are ‘inactive’. It is important to note that someone who is ‘inactive’ is not the same as someone who is ‘unemployed’. In Kosovo, to be considered ‘actively looking for work’ (and therefore be classified in the labour force) the following criteria must be met. The person must be:

  • currently available for work, that is, available for paid employment or self- employment within two weeks; and
  • seeking work, that is, have taken specific steps in the previous four weeks to seek paid employment or self-employment.

If either of the above criteria is not met, the person is classified as inactive.

Calculating the Participation Rate

Once the population is classified as either in the labour force or inactive, it is possible to calculate the participation rate, one of the key labour market statistics. The participation rate measures the labour force population (people employed and/or actively looking for work) as a percentage of the working age population.

WAC_E_3_1

In Kosovo, the participation rates in 2014 were as follows:

  • Male Participation Rate (2014): 61.8%
  • Female Participation Rate (2014): 21.4%
  • Overall Participation Rate (2014): 41.6%

Unlike the unemployment rate, described below, the participation rate tends to provide more stable and reliable data than the unemployment rate, as it is not affected by short-term fluctuations and the business cycle.

Employed vs. Unemployed

Analyzing the population further, the ‘labour force’ can be subdivided into two populations – those that are employed and those that are unemployed. In most cases it is obvious whether someone is employed or not, but in some situations it may not be so clear (e.g. when a person is working for the family business in an unpaid capacity). To handle these scenarios, the agency tasked with compiling the labour market statistics in each country typically has a specific definition (or definitions) of what qualifies as employment. In Kosovo, to be classified as ‘employed’ a person must meet the following high-level criteria:

“People who during the reference week performed some work for wage or salary, or profit or family gain, in cash or in kind or were temporarily absent from their jobs.”

In addition, the Kosovo Agency of Statistics includes some more detailed criteria in their methodology that clarifies when work done on family owned farms classifies as employment. This will become important later.

Calculating the Unemployment Rate

Having separated the employed from the unemployed, it is now possible to calculate the unemployment rate. To do this, we divide the number of unemployed people by the total number of people in the labour force.

WAC_E_3_2

In Kosovo, the unemployment rates in 2014 were as follows:

  • Male Unemployment Rate (2014): 33.1%
  • Female Unemployment Rate (2014): 41.6%
  • Overall Unemployment Rate (2014): 35.3%

The unemployment rate is useful as a more immediate indicator of conditions in the economy. The obvious information is provides is an indicator of how many people without a job are currently looking for employment. But, in addition, it also provides information about how much spare capacity an economy has, the risk that inflation may pick up, whether structural issues are keeping people out of work and so on.

Chart 1 – Participation and Unemployment Rates by Gender 2014

What is Next?

In the next article, we will take a look at how the participation rate (for both males and females) in Kosovo compares across the region and internationally. In the meantime, please feel free to play around with the interactive visualization below, which shows the entire working age population of Kosovo broken down into its various subgroups.

Click on the chart below to interact with the data!

sunburst_pic

Sunburst chart created by Festina Ismali

 

 

Corruption in Kosovo: A Comparative Analysis

Cross posted from OpenDataKosovo.org:

Previously in Part I of this series, we looked at corruption in Kosovo from the perspective of Kosovo civil servants, as documented in a United Nations Development Programme (UNDP) report entitled Gender Equality Related Corruption Risks and Vulnerabilities in Civil Service in Kosovo[1].

In Part II we are now going to look at global corruption perception statistics compiled by Transparency International to consider how Kosovo compares internationally.

An International Comparison of Corruption

Transparency International is an organization that works to reduce corruption[2] through increasing the transparency of Governments around the world. Arguably Transparency International’s most well known contribution is the Corruption Perceptions Index (CPI), an index measuring “the perceived levels of public sector corruption worldwide”. In 2014[3] the CPI was calculated by aggregating 12 indices and data sources collected from 11 different independent institutions specializing in governance and business climate analysis over the past 24 months. The 2014 CPI covered 175 countries, including Kosovo.

In addition to the CPI, Transparency International does its own survey and data collection in the form of the Global Corruption Barometer (GCB survey). The GCB survey focuses on the public’s opinion of corruption within their own country, and in 2013 (the latest edition of the GCB available at the time of writing) collected the opinions of over 114,000 people across 107 countries – including Kosovo.

So what did these two reports show?

Results

In the CPI, Kosovo performs poorly, placing 110th out of 175 countries with a score of 33 out of 100 (unchanged from 2013). To give some perspective, Kosovo finished equal 110th with 4 other countries – Albania, Ecuador, Ethiopia, and Malawi. This placed it behind Argentina (107th), Mexico (103rd), China (100th), India (85th) and Greece (69th), countries that are often associated with high levels of corruption. Finally, this was the lowest ranking for any country in the Balkans region (tied with Albania).

Chart 1 – GCB Survey Q6 – Perceptions of Corruption by Institution for 6 Countries

WAC_2_1

The GCB survey, however, shows that the people in Kosovo have a different perception of corruption in several areas to that reported in the CPI. Based on the responses to question 6[4] (see Chart 1) and question 7[5] (see Chart 2) of the GCB survey, people in Kosovo are somewhat more optimistic about the levels of corruption in their country than the low rating on the CPI might indicate. Kosovo scores well in several areas:

  • Only 16% of people reported having paid a bribe in the last 12 months. This placed Kosovo 35th out of the 95 countries that provided a response to question 7.
  • 46% of Kosovars generally believe their public institutions to be corrupt or extremely corrupt. This sounds high but actually puts Kosovo ahead of the US (47%) and only slightly behind Germany (40%). The results for certain institutions were even better:
    • The Military is believed to be corrupt or extremely corrupt by only 8% of those interviewed – only four countries had a lower percentage than Kosovo on this part of question 6.
    • NGOs and Religious bodies were also seen as uncorrupt by large majorities.
    • 44% of people believed public officials and civil servants were corrupt, placing Kosovo ahead of Germany, France and the US, among others.

Chart 2 – GCB Survey Q7 – Reports of Bribes Paid by Institution for 6 Countries

WAC_2_2

But not all the results were positive. Questions 1[6], 4[7] and 5[8] in the GCB survey in particular highlight a more pessimistic outlook:

  • In response to question 1, 66% of Kosovars stated that they believed corruption had increased over the past 2 years, while only 8% believed it had decreased.
  • In response to question 4, 74% of Kosovars stated they believed their Government is run by large entities largely or entirely for their own benefit.
  • In response to question 5, only 11% of Kosovars surveyed believed the actions of their Government in the fight against corruption are effective.

What does all this mean? Why does Kosovo perform so poorly on the CPI, and on some GCB survey questions, but on other questions the perceived level of corruption of people in Kosovo is comparable to some developed nations?

Perceptions vs. Reality

One of the issues when looking at the results of the GCB survey is that the responses to most of these questions are subjective. What constitutes corruption or extreme corruption varies by country and culture based on what people are used to living with. What someone in South Asia or sub-Saharan Africa considers standard practice and harmless may be considered unbelievably corrupt by people in other parts of the world.

These different standards are really highlighted when we compare the percentage of people believing an institution is corrupt with the number of people reporting to have paid a bribe to that institution, using questions 6 and 7 of the GCB survey. There are four institutions that appear as options for both questions, allowing us to make a direct comparison:

  1. Education
  2. Judiciary
  3. Medical and Health, and
  4. Police

In the comparison (see Chart 3), we find numerous examples where the percentage of people that reported paying bribes was higher than the percentage of people who believed the institution was corrupt. The implication of this finding is that significant numbers of people in these countries believe that paying a bribe is not a sign of corruption.

Chart 3 – Comparison of Perceived Corruption with Bribes Paid

WAC_2_3

Kosovo and most developed nations were examples of the opposite case – they generally reported relatively high numbers of people who believed the four comparable institutions were corrupt, and relatively low percentages of people reporting bribes being paid. Bribery, of course, is not the only form of corruption, and this result could simply be an indicator that different forms of corruption are more prevalent in these countries. But it could also be an indicator that people in some countries are particularly cynical about the fidelity of their institutions.

To get a better sense of how concerned people really are about corruption, lets now take a look at some of the responses to other questions in the survey.

Is a Person’s Willingness to Take Action a Better Indicator?

One of the questions asked on the survey that could potentially reveal some further information was question 10 – “Are you willing to get involved in the fight against corruption?” Respondents were then provided with a range of activities, both active and passive, and were requested to indicate whether they would be willing to participate.

At a high level, the responses to this question appear to show an inverse correlation between the value of the CPI for a country and how willing people in that country were to do something active to fight corruption. In other words, the higher the percentage of people willing to do something active to fight corruption, the lower the CPI index for that country (i.e. a higher level of corruption).

Using a statistical model (such as regression), we can check whether this relationship is real and how strong it is. However to do this, we need to consider countries with regimes that punish dissent and crack down on protests and/or organizations that might try to combat corruption. In these countries, you would expect to have a low percentage of people willing to take action against corruption despite corruption being high.

To account for this, we need to have some sort of indicator of how worried people are about speaking out in their country. The best piece of information that we have from the GCB survey that can serve this purpose was the question asking if the respondent would be willing to report corruption.

Using these two pieces of information, we can try to test the following hypotheses:

  1. A high percentage of people willing to take action against corruption in a given country is indicative of a high level of corruption.
  2. A low percentage of people willing to take action against corruption in a given country, but a high level willing to report corruption is indicative of a low level of corruption.
  3. A low percentage of people willing to take action against corruption in a given country, and a low level willing to report corruption is indicative of a high level of corruption in a repressive regime.

Based on these hypotheses, we also expect that there would be no (or very few) cases where there is high percentage of people willing to take action against corruption and a low level of people willing to report corruption.

Building a Model

Using our two pieces of information described above, and with the assumption that the CPI is the most accurate indicator of the true level of corruption within a country[9], we can build a model to predict CPI for each country and test our hypotheses. The formula for this model will be as follows:

Where:

Yi = the actual value of CPI for country i

β0 = a constant

Xi1 = the percentage of people willing to do something active to fight corruption[10] in country i

β1 = a constant applied to Xi1

Xi2 = the percentage of people willing report an incidence of corruption in country i

β2 = a constant applied to Xi2

εi = the residual or error

Using ordinary least squares (OLS) and the data for the 101 countries for which the CPI and the two variables (X1 and X2) described above are provided, the results of the model is as follows:

β0 β1 β2
Coefficient 27.9735 -0.8417 0.8798
Standard Error 4.8427 0.0705 0.0738
R2 66.7%

The first thing to note is that the coefficients support the three hypotheses we mentioned above:

  1. A strongly negative coefficient β1 indicates that the larger the percentage of people willing to do something active to fight corruption, the lower the predicted CPI.
  2. A strongly positive coefficient β2 indicates that the larger the percentage of people willing to report corruption, the higher the predicted CPI.

General Insights

Aside from providing support for our hypotheses, the other thing this model reveals is the countries that are not very well explained by this model. Chart 4 shows the CPI predicted by the model as compared to the actual CPI value for 2014.

Chart 4 – Predicted CPI vs. Actual CPI by Country

WAC_2_4

At a high level, we can split the chart into two parts:

  1. Points below and to the right of the line reflect countries where the actual level of the CPI was lower than the predicted level
  2. Points above and to the left of the line reflect countries where the actual level of the CPI was higher than the predicted level.

Starting with the first group – countries that were more corrupt than the model predicted – these cases appear to fall into two categories:

  • Conflict Affected Countries – In these cases, of which Sudan is the most extreme example, there was typically a low percentage of people willing to do something active to fight corruption, and therefore the CPI was predicted to be significantly higher than it is in reality. This is likely to be due the citizenry of these countries facing more immediate problems. This pattern was seen across Sudan, Afghanistan, Iraq, Libya and South Sudan.
  • Other – In these cases, of which Russia was the best example, there was generally a high percentage of people willing to report corruption (86% for Russia) and a relatively low percentage of people willing to do something active to fight corruption (47% in Russia). As a result the model predicted a relatively high CPI. The explanation for this is not as clear as above, but the evidence would seem to suggest that the people in these countries are either not aware of the high level of corruption present in their country, or that they have a significantly different opinion as to what constitutes corruption.

Contrasting with the above cases, we can also see there are countries above and to the left of the line in Chart 4. This represents countries that were less corrupt than the model predicted. In these cases the responses to the two questions were indicative of a country with a higher level of corruption than actually existed. The following were two interesting cases:

  • Finland – the model was thrown off by a surprisingly low percentage of people willing to report corruption. Of the respondents from Finland, only 65% of people surveyed reported they would be willing to report corruption – a surprisingly low percentage for a country with a CPI value of 89. In fact, Finland and Japan were the only countries with a CPI above 60 that reported a percentage below 80% for this question.
  • The United States – neither of the data points used for the US in the model were hugely abnormal for countries in the same CPI range. 80% of people said they would be willing to report corruption (a little lower than you would expect) and 50% said they be willing to do something active to fight corruption (a little higher than you would expect). Both of these potentially show a slightly higher level of mistrust in government than other developed nations, something that does tie in with the politics of large parts of the US.

Unlike the above examples, Kosovo appeared fairly typical for the model. Let’s now take a deeper look into the results of the model for Kosovo.

Insights for Kosovo

For Kosovo, the model was able to fairly accurately predict the CPI using the two variables described. Kosovo has both a high percentage of people willing to do something active to fight corruption (80%) and a high percentage of people willing to report corruption (84%). As a result, the model predicted a high level of corruption in Kosovo, a CPI of 35, which was just below the actual CPI value of 33.

However, aside from proving the accuracy of the model in this case, these high values reveal important information about the people of Kosovo. It reveals Kosovars do believe corruption is an issue, and that they are willing to do something about it.

Summary

Overall, there are positives and negatives for Kosovo that can be taken from the Transparency International data. On the negative side, the CPI highlights that corruption is a significant issue in Kosovo. Even in a region with consistently low CPI scores (the best performer is Slovenia with a score of 58) Kosovo is a significant underperformer. The most disappointing aspect of this underperformance is that Kosovo has had the significant advantage of 15 years of assistance from various international agencies in setting up infrastructure for good governance.

That said, there is a big positive that comes from the GCB survey data, and it is also potentially an important clue as to the best way forward for Kosovo and the international organizations involved in the region. That positive is that the people of Kosovo appear to be aware of the issues of corruption in their country, and more importantly, they are very willing to take an active role to fight it. Compared to Albania, a country with the same CPI as Kosovo, almost twice the percentage of survey respondents stated they were willing to do something active to fight corruption in Kosovo (80% vs. 44%), and significantly more people said they were willing to report corruption (84% vs. 51%).

What this suggests is that, if harnessed effectively, anti-corruption efforts in Kosovo could be very popular, and therefore powerful. But the right strategies have to be implemented and publicized to garner public support.

Somewhat unsurprisingly, we believe a key strategy has to be raising awareness of how data can be used to reduce corruption and bring about change. This can apply equally to data that is currently collected by government agencies but isn’t publically released, or new datasets that the public can assist in collecting. With the right data and right analysis, these datasets can help to improve governance in numerous ways including:

  • exposing systematic corruption
  • identifying gaps in anti‑corruption controls, and
  • better targeting of anti-corruption efforts.

Using this open data approach also helps reduce reliance on the bravery of individual whistleblowers. Although whistleblowers are often vital in helping to identify incidents and even patterns of corruption, the fact is that, even in developed nations, they will always risk retaliation and other subtler forms of retribution (reduced career prospects, being ostracized by their peers and generally being perceived as untrustworthy).

Overall, what the results of the Transparency International data shows us is that, with better coordination and targeting of anti-corruption efforts, there is the potential to actively involve large numbers of Kosovars. If that can be achieved and funneled into meaningful strategies, the future of Kosovo could be very bright indeed.

Have any suggestions for ways data could be used to fight corruption? Disagree completely? Feel free to leave your thoughts in the comments!

 

[1] Gender Equality Related Corruption Risks and Vulnerabilities in Civil Service Kosovo, United Nations Development Programme. November 2014. Gender Corruption final Eng.pdf

[2] Defined by Transparency International ‘… as “the abuse of entrusted power for private gain”. Corruption can be classified as grand, petty and political, depending on the amounts of money lost and the sector where it occurs.’

[3] The methodology for compiling the CPI is reviewed on a yearly basis with data sources added and removed as needed.

[4] “To what extent do you see the following categories in this country affected by corruption?” – responses of “corrupt” or “extremely corrupt” recorded as a positive response.

[5] “In your contact or contacts with the institutions have you or anyone living in your household paid a bribe in any form in the past 12 months?“

[6] “Over the past 2 years, how has the level of corruption in this country changed?”

[7] “To what extent is this country’s government run by a few big entities acting in their own best interests?”

[8] “How effective do you think your government’s actions are in the fight against corruption?”

[9] By their own admission, Transparency International’s CPI is not a perfect measure of corruption. Corruption by its nature is hidden and so there is no objective measure of the true level of corruption. However, the CPI is currently the most respected measure of corruption available and so we make the assumption that it is also the most accurate for the purposes of constructing this model.

[10] Taken as the average of the percentage of people who said they would take part in a peaceful protest and the percentage of people who said they would join an organization that works to reduce corruption as an active member

Women and Corruption Issues in Kosovo

For those that don’t know, over the past couple of months I have been spending time working with a tech startup/NGO here in Pristina called Open Data Kosovo. The main aim of the organization is to encourage and facilitate the release of data and other information by the government of Kosovo (and related bodies) in order to increase transparency and reduce corruption. So far they have been fantastically successful, getting both national and international media attention, which is all the more impressive when you consider they are only now coming to the end of their first year of existence.

One of the main things I have been working on since joining is putting together some analysis of the various datasets they have been publishing online to see what conclusions can be provided to the public that might help create a more informed discussion of the issues. The first piece has now been published on the Open Data Kosovo website and we are excited to see what kind of feedback we get. If you want to take a look, please click the link below:

More women in leadership would probably reduce corruption, but is there a more effective way? 

Newer posts »

© 2024 Brett Romero

Theme by Anders NorenUp ↑