Thursday, February 26, 2015

Women in the Economy









Women in the Economy







By: Emily Kaput




 
“In a world in search of growth, women will find it, if they face a level playing field.” 
–Ms. Christine Lagarde, February 2015[i]

Today, 90% of countries in the world have legal regulations that inhibit female participation in the labor force. These regulations include laws that restrict female participation in specific professions, their ability to obtain a loan, and needing consent from their father or spouse to perform paid labor. An obstacle is the right to property (inherited or bought). Restricting female participation in the economy is popularly believed to decrease country’s ability to realize its full economic potential- this sentiment echoes in the halls of the United Nations, the International Monetary Fund, development banks around the world and by governments themselves.





This brings me to my research question: Do higher levels of women’s participation in the labor force inherently yield higher productivity? If not, what conditions, specific to the female population, contribute to higher productivity in economies?


The variable I chose to represent growth/size of an economy is Gross Domestic Product Per Capita.[ii] GDP would not be a properly scaled response variable as I am examining individual level contributions (female population) to growth in an economy.

The graph below displays the bivariate relationship between Female Labor Force Participation Rate and GDP Per Capita. My expectations: with higher levels of female labor force participation, GDP Per Capita would increase. This is not true in this bivariate representation.  There is a LOT going on in this graph. Certainly, where there is mid-range labor force participation by females, there is the highest GDP Per Capita. However, at the tail ends of this distribution GDP Per Capita is almost perfectly mirrored. Because this bivariate relationship does not tell me as much as I would have liked about female participation in the economy, I introduce control variables. These are: GDP attributed to Trade, GDP attributed to Agriculture, and Women’s Economic Rights.








I felt that measuring a country’s “openness” to trade was an important factor to include in this analysis.[iii] It is a popular notion that countries who actively participate in global trade are more likely to be liberaliz(ed/ing) societies.[iv] In addition to this, it is important to measure openness of an economy because international trade balance is an element of GDP. On the other hand, agrarian societies have historically been associated with traditional, conservative gender norms, and since the era of Western industrialization agrarian societies have been popularly known to have lower GDP.[v] Since GDP does not measure domestic activity (household productivity) these agrarian societies have a particular disadvantage when comparing their GDP with other less-agrarian countries.[vi]  *See appendix for graph of this bivariate relationship- congruent with expectations. 
Below, I have displayed the bivariate relationship between Women’s Economic Rights and GDP Per Capita. Note: with each score increase, GDP Per Capita increases.[vii]







In Table 1, I regress GDP Per Capita on these control variables and my key independent variable. The first two columns are my conditional models and the last column is my final, complete model with all variables.
Model 1: There was no significant relationship between Female Labor Force Participation Rate and GDP Per Capita. *The estimate displayed is likely a bad representation of the relationship, thank god.

Model 2: Guaranteed economic rights (only those countries who scored 3) have a positive effect on GDP Per Capita, as does increasing amounts of trade as a percent of GDP. The expected negative relationship between GDP Per Capita and agriculturally dominant societies is evident; there is a statistically significant decrease in GDP Per Capita per one percent increase in GDP attributed to agriculture.

Model 3: Here, where GDP is regressed on all variables, a clearer story is told. Female Labor Force Participation Rate has no relationship with GDP Per Capita- completely throwing out my initial assumption of a positive conditional relationship. Where women have guaranteed economic rights, however, there is a significant positive relationship. *Recall the exponential increase of GDP Per Capita in the bar graph displaying Women’s Economic Rights and GDP Per Capita. With every percent increase in Trade, GDP Per Capita increases by $66.57. And, unsurprisingly, GDP Per Capita decreases by $499.92 per one percent increase of agriculture’s share in GDP. Noting the R-Squared value: This model can only explain approximately 57% of the variation in this model. Meaning that there are other variables not included in this model, which would help to explain variance in GDP Per Capita around the world.

What does it mean, that female labor force participation has no effect on GDP Per Capita? Simple: other factors are necessary to include when exploring female contributions to growth in economies.  For instance, Women’s Economic Rights is a great place to start when explaining women’s role in the economy. In societies where women hare more economic rights, GDP Per Capita is higher than where there are little to no economic rights for women. Regardless of the Female Labor Force Participation Rate, economies seem to be more productive and citizens are ‘better off” when women are guaranteed economic rights.

However, this does not imply causality. According to a report given by the World Bank,[viii] increases in women’s economic rights tend to be the result of economic growth. Middle-and-low-income countries have the same level of restraints on women’s economic rights, and several papers explore this relationship between growth in an economy and its effect on women’s economic rights. A post about this relationship would be a great follow-up for this post.

Note from the author: My examination of women and growth in economies around the world succeeded only in raising more questions, which is preferable (in my opinion). In fact, I am relieved! The relationship between women and the economy had better NOT be explained by so few variables!



[i] Lagarde, Christine. "Fair Play—Equal Laws for Equal Working Opportunity for Women." Web log post. IMFdirect. International Monetary Fund, 23 Feb. 2015. Web. <http://blog-imfdirect.imf.org/2015/02/23/fair-play-equal-laws-for-equal-working-opportunity-for-women/>.
[ii] United Nations, United Nations Statistics Division, National Accounts Estimates of Main Aggregates (2010), Per Capita GDP- USD.  GDP Per Capita is negatively skewed, meaning that there are several observations that are abnormal (distribution is NOT normal). These outliers are: Monaco, Luxembourg and Norway (GDP Per Capita higher than the mean, $70,000.00). I chose not to “fix” this variable because there were only a handful of outliers, and the “fix” only reflected the data and created a positively skewed distribution. On the whole, I believe GDP Per Capita data to be an adequate representation of global GDP Per Capita.
[iii] Alexander C. Chandra, Lucky A. Lontoh, And Ani Margawati. "Beyond Barriers: The Gender Implications of Trade Liberalization in Southeast Asia." Trade Knowledge Network (2010): n. pag. International Institute for Sustainable Development. Web.
[iv] World Bank, World Development Indicators, Trade, Value Added (% of GDP) (2010).
[v]World Bank, World Development Indicators,  Agriculture, value added (% of GDP) (2010).
[vi] "News." Economic Development = Equal Rights for Women? World Bank, 24 Sept. 2013. Web. 17 Mar. 2015. <http://www.worldbank.org/en/news/feature/2013/09/24/Economic-Development-Equal-Rights-for-Women>.
[vii] Teorell, Jan, Stefan Dahlberg, Sören Holmberg, Bo Rothstein, Felix Hartmann & Richard Svensson. 2015. The Quality of Government Standard Dataset, version Jan15. University of Gothenburg: The Quality of Government Institute, http://www.qog.pol.gu.se. ciri_wecon, 2010, Panel Data.
[viii] Hallward-Driemeier, Mary; Hasan, Tazeen; Rusu, Anca Bogdana. 2013. Women's legal rights over 50 years : progress, stagnation or regression ?. Policy Research working paper ; no. WPS 6616. Washington, DC : World Bank Group. http://documents.worldbank.org/curated/en/2013/09/18287629/womens-legal-rights-over-50-years-progress-stagnation-or-regression

















6 comments:

  1. funny sub-sub title!

    Tables should perhaps be scattered throughout the text instead of being all up front.

    Research Question is clear.

    "initial assumptions" should be initial "hypotheses". You don't want to assume your results before testing them.

    Beyond the obvious, why is this an important question? What are the policy implications of such a study? What is driving the desire to study this?

    You start discussing the regression results before discussing research design. Why are the variables that you are including in the model included? What are the different models testing for? What is the goal of each of them?

    Table 3 is a bit unclear. Did this model stop using the control variables that were included in the previous table? How were Women's Economic Rights divided into four different groups? You seem to use Political Rights and Economic Rights interchangably...are these the same? How were variables standardized? Which countries were examined?

    The following section is very troublesome:
    "Even more outrageous is the result seen in educating females...This is ridiculous. Based on this model, I do not think it is safe to say these variables are effectively regressed together. I would never, in a million years, tell a country to stop educating their female populations."
    The point of a statistical study is not to back up the assumptions/biases of the researcher at all costs. Though you are likely correct that the model is not an accurate estimate of reality, why is that the case? What is wrong with it? Personal opinions like 'atrocious,' 'outrageous,' and 'ridiculous' are highly subjective and do not lead the reader to trust the analysis - it makes it seem like this is an op-ed with an agenda.

    Graphs! (there aren't any...unless the blog is not displaying correctly)

    What does "well rounded and robust" mean when speaking about variables? I don't think that data quality is necessarily the biggest problem with determining the relationship here.

    "This is problematic when attempting to substantiate my initial claim"
    again, statistical research should be conducted to test a hypothesis not verify a claim. Just throwing out results because they don't agree with your assumptions is questionable science.

    I don't see linearity/distribution of the data addressed here. Are there any issues that may have skewed the results of the regression?

    What about causality? Is there a reason that this study doesn't identify causality? What is it?

    What is the conclusion? You state that the regression results are questionable but, taken at face value, what does this research conclude and how valid are those conclusions?

    I'm a bit confused by your last section about finding a better response variable than GDP. The research question is examining the effect of rights on GDP. Are you positing that this is not a worthwhile research question and that a new one should be formulated? You seem to imply that the research is not intended to examine the effect on GDP but rather some other more amorphous idea of "production." If this is the case then the research question should be changed. If not, then this section is not needed.

    ReplyDelete
    Replies
    1. Thanks!!! I appreciate the feedback- super helpful. :)

      Delete
  2. Research question is clear. Love the sub-title. However, I would like more explanation of why this is an important question. Undoubtedly it is, but if a strong economic argument could be made that female participation in the work force is good, then maybe more countries would work harder to involve women in the work force.

    The biggest issue with this study is the Dependent Variable. You picked GDP, which is a total measurement, but does not account for differences in populations. The better variable to pick here is GDP per capita, which would give a much more meaningful outcome, and might even give you the results you were looking for.

    There are no graphs. I think that summary and regression tables should be farther down in the paper, and not right at the beginning, as it makes it hard to constantly scroll back and forth between them.
    It’s not clear that linearity was examined, or that you transformed any of the variables. I think the confound variables are too similar to each other, and do not accurately control for factors that might affect GDP.

    There doesn’t seem to be any conclusion. Your findings, however illogical they may seem, seem to state that female involvement in the workforce causes drops in GDP. I think that if you use GDP per capita you may not come to that conclusion. However, you should definitely discuss the findings and what you think might have caused them.

    ReplyDelete
    Replies
    1. I didn't even think of GDP per capita. Thank you for that!!

      Delete
  3. Hi Emily, I enjoyed your blog post especially the idea of the topic! To start, I think the blog post as a whole will be more clear if you break out your sections and inject the tables amongst the text, I had a hard time scrolling up and down to see the tables while going through your findings. I would also have liked to see a more developed "introduction" with more background info on how you came to your hypothesis and your expected results. But I love the title and your personal commentary throughout, it really seems like a blog/opinion piece because your voice comes through often.

    With the exception of "women's economic rights" I did not get an understanding of the variables you included and why, which could be backed up with a short sentence about their relevance to GDP growth and/or women's education. I see them in the tables and they all seem to be relevant and important, though! I agree with Alex, and would like to see this regression using GDP per capita which may offer you a more realistic regression/comparison of female education in years and GDP per capita (both are individual unit data). I think you may see clearer results, maybe in the "right" direction!

    Your tables are nice, I would include a few lines to make them a little more streamlined. I would also recommend bring the (t-score) up to directly under the beta coefficient for clarity of which variable it belongs to. Right now it looks like the t-score goes with the variable below it, instead of above it. I would like to see a graph with the main relationship female education & GDP (per capita?) so that I could visualize the relationship.

    In your conclusion, you give a good critique of the GDP variable (can this be fixed?). I would like to see more conclusion about the findings: Why might this relationship be in the negative direction? What do the other variables and coefficients tell us? This is where you can put in some information for future research or policies!

    Great start and awesome title & picture (of course!)

    ReplyDelete
  4. This comment has been removed by the author.

    ReplyDelete