There is a some alternative measures. For instance, Generalized Entropy Indexes, Atkinson Index, Piesch index, Kakwani index. Each measure can exhibit differnt properties. So, for a detailed explanation, see, for example: Measuring Inequality - Frank A. Cowell

Depend on you want to measure.., they have different properties

For that you can study

Inequality: Methods and Tools by Julie A. Litchfield March 1999

Text for World Bank’s Web Site on Inequality, Poverty, and Socio-economic Performance: http://www.worldbank.org/poverty/inequal/index.htm

Gini is the EASIEST way to measure inequality. It is easy to teach, it is easy to understand how the math works.

What does it measure?

It's a mathematical index. It's a representation. A formula with a simple numerical output.

I suggest you review the website of the PEP Research Network.

They have extensively explored many approaches to defining poverty and inequality. Many of their papers include advanced and/or novel measures of inequality. You can search their website by theme or region, and most papers in recent years will provide baselines and complete methodological descriptions for these methods.

I would not say that one measure of inequality is "superior" to another. They all have costs and benefits. If you want to reach the most vulnerable, I suggest searching them out directly by supporting on-the-ground NGOs, but in terms of data and research applications, the depth/intensity of poverty at a regional level, for example, will in a very quick snapshot tell you where to expect inequalities which are having pronounced impacts on a population in a given area.

How do you define inequality?

Often, the implicit mathematical assumption is inequality = Gini or inequality = (whatever the best or most suitable index/indices was for the particular study given the available information).

For example, inequality of opportunity is probably much more relevant for people who are concerned about revolutions, relative to actual outcomes which can also be explained by a lot of other reasons (younger people earn less, some years some people are unemployed a lot, perhaps there are two cultural groups with different approaches to the economy (Malaysia, perhaps) ... ).

I have seen people use lots of different measures of inequality.

You can show the charts, discuss possibly the representations of the underlying data. Directly represent the underlying data, and even do so with 5 or 10 dimensional modelling of poverty and inequality.

So here this one was higher, there that one was lower. This time this measure accentuated more heavily the vulnerable population, this time that measure covered up inequality at higher levels of aggregation. (I guess it's always good to probe the data for these kinds of things if you want to be careful in not overstating the case for some policy options, for example.)

So I put the question back ... what's the goal?

Which inequality? Why?

And I would find myself going to capacities and functionings and figuring ... well why bother so much about inequality. Instead, we can identify a set of basic things we think all people should have access to and guarantee it flat out. Easier said and done in Scandinavia perhaps than the Central African Republic.

Anyways, I could ramble on about this for a while. I think in recent years there has been very much of a trend away from unidimensional poverty, and so inequality measures are behind while "inequality in what?" precedes the second question somewhat. What's the point in improving on or complementing the Gini when we do not yet know the form of the target to which the assessment tool will be applied? In the meantime, Gini is quick and easy, and usually not all that different from other measures. Still, other tools help to evaluate and/or demonstrate relative robustness in different ways, and are thus highly desirable as a complement to the basic toolkit, imo. The tools are there. With the obvious warning of a need to understand their limitations, they are often just a few clicks away.

So I reckon Gini's best for most purposes, but researchers will want to make regular use of many other methods in trying to understand a particular dataset.

Oh yeah, I think it would be interesting to have a binding constraint on Gini, or some other measure of inequality. For example, once you reach 0.40, the only way for the rich to get richer is for them to find a way to motivate and/or otherwise get others more interested in getting richer.

Worst case scenario, if they are truly so greedy, and depending on the dynamics of the economy, then they may prefer to just pay the couch potatoes to live a lazy and blissfully ignorant life with expanded ability to pay for cable and popcorn, rather than all the complication of trying to figure out how to get them to work harder for economic goods, the sales of which the wealthy can benefit from.

It's tempting to think we'd need the perfect tool. But people don't usually support things they don't understand. Gini is easy. 0.40 is high enough, no?

I think a minimum guaranteed income together with a Gini cap would be super cool. Nothing too generous, but enough to make sure that everyone knows that society has their back while they figure out how to make the best they can of things.

Probably we will get lots of amazing art, music, etc. What's that worth?

It depends on what you are looking for. For example if you have different population groups, and you want know the between-group inequality you can use the Theil index. At the same time, is a useful measure if you don´t have information at the level of individuals. Here you can find more about this inequality indicator .

utip.gov.utexas.edu/papers/utip_14.pdf

Regards.

I prefer inequality ratio (the share of national income/consumption for the richest decile divided by the share of national income/consumption for the poorest decile). In addition, it is good to consider also these deciles separately in order to understand the whole picture of national economic inequality. For a study of inequality in 143 countries see my recent book (below).

Book Missing a Decent Living for Everyone: Success and Failure in...

Almost any of them. But Gini is easy to teach and use. It makes sense to use indicators which are relatively broadly used and understood. Is there a worse indicator of inequality than Gini? One simple to use equation will give you one short number which summarizes the situation. There's not much of any room for confusion or spin. But there are many other ways, possibly infinite.

I think simply showing income shares by quintile in a column, then repeating for different time periods in each column, or showing income shares (of GDP) of each quintile as a separate line on a graph can be good.

Or you can just report the raw income distribution on a graph, perhaps in log since some numbers will be exponentially larger.

Since it is very common to have low income in one year and higher income in a second year, you can compare mobility between quintiles year on year compared to five-year intervals to find out about entrenched poverty compared to a temporary downturn or upturn.

I think it might depend on data availability a lot of the time and the specific use in others. If you're looking for a snapshot for which you can easily perform cross country comparisons at many points in time, Gini can summarize many pictures very quickly, but if you're trying to investigate claims of economic determinants of dischord or which policies are effective at mitigating certain causes of inequality, then Gini might not help much, except that you could give the reader a quick snapshot. If you say "Gini is 0.25 and now I'm going to talk about how the revolution can be avoided," I would think you would be speaking of a revolution of the rich against some socialist government. It you say "Gini is 0.45 but I can't tell you if the level of inequality reflects the same people staying poor or that the situation of two groups has recently changed", then the only use of Gini is, again as a snapshot.

Consider the question "is GDP the best measure of welling?".

You could also compare Gini of indices of consumption which evaluate the main classes of consumption, which could yield very different results for income which might not be very accurate in representing economic power of a certain class, but might tell you a lot about what that economic inequality means in terms of their living conditions.

Anyways, if you check pep-net.org, you'll find lots of papers which use a whole bunch of different methods. Many of them report Gini, and a few others test sensitivity of Gini to different situations, but mostly it's just referred to in a first second or two of a methods section as a snapshot of the magnitude of inequality prior to moving on to more advanced and/or innovative approaches.

This may be useful: 

http://srmo.sagepub.com/view/assessing-inequality/SAGE.xml

 Inequality is a relative term that is deviation from standard/average level in different context in different fields of studies. It is better to construct suitable index of important causal and impact factors related to the variable in question.

I have a different take on this...

You don't specify what the nature of your problem is, so a different approach might work for your use case. I suggest Kendall's Tau

(Since Gini coefficient comes from economics, I'll describe my approach in terms of economics to make it more concrete.)

Suppose the different "incomes" are coming from specific "persons" and you want to compare an experiment where different "incomes" result from two experimental conditions and how it affects "persons' incomes".

If you want to compare these two results, you can use the Kendall-Tau similarity measure to compare the two lists. If they are exactly in the same order, then the Kendall-Tau is 1. If the lists are backwards from each other, then it's -1. If they are in random order, then it's 0.

Now, one thing that's a bit dissatisfactory with this approach, is that the ordering ignores the "income" differences between the two lists. An interesting alternative is the Weighted Kendall-Tau, which does take into account of the income differences as well as the ordering.

A problem with this approach, is you're comparing pairs of experiment results, so the resulting comparison is "richer": You end up with a table of N experimental results which is no longer a simple summary. Of course, this approach doesn't work if the people are "anonymous" or are from totally different populations (or sizes) so that you can't compare "persons" directly.

I hope this helps.