Last week I had post summarizing results from a number of other posts in which I compared how various series moved by presidential administration. For example, I looked at how the economy, national debt, dollar valuation, and so forth grew under various administrations. But I also looked at some social issues, such as abortions, unwed motherhood, etc. In almost every series, Democratic Presidents did better than Republican Presidents.
Anyway, I got some comments and some correspondence from people who felt that Presidents can’t possibly be affecting some of these social issues. As more than one person asked – can one really believe that someone is less likely to get an abortion because Clinton is President than they are if GHW or GW is President? How can this possibly be? In other words, people feel that if the numbers who Democrats are doing better on social issues which drive Republican voters, it must be purely a coincidence.
My answer is – yes, it can be, and there are two transmission mechanisms I can think of that make it so. (To be honest, to me both of these transmission mechanisms seem fairly obvious – it took me a bit of time to realize there was any dispute about this.) The first such transmission mechanism is itself social. Democrats are more likely to favor sex education and funding for contraceptives than Republicans, who tend to favor abstinence only education. There have been enough studies out there showing that abstinence only education simply doesn’t work, so I won’t rehash them.
But there is a second transmission mechanism that is particularly interesting for me –I believe that for the most part, if abortion is an issue of choice, economic considerations must enter into the mind making a decision. In particular – the person making the decision will consider her income. (We’ve already seen that the economy grows faster under Democrats than under Republicans, and the disparity becomes pretty big if one takes into account growth in the national debt. Put another way… Democratic Presidents seem to have policies that lead to faster economic growth.)
Notice what I wrote: her income. Because the rate of abortions has been decreasing over time, and real incomes have been rising, in general we expect a negative correlation between real incomes and abortions just because of the direction the two series are moving. But this negative correlation may well be just a coincidence. We need a way to determine if it is just a coincidence or not.
Which leads us to this… if income is correlated with abortions, then female income should be more strongly correlated with abortions than male income. This is because while a male may be involved in the decision about whether to have an abortion, except in very rare instances a female will be involved in the decision.
Now, how do we measure abortions? Data on abortions is available from the Census. It provides the number of abortions, the rate of abortions per thousand women (age 15 to 44), and the ratio per thousand live births. (Note – this data is more complete than the series I used in my earlier post – it goes back to 1975!) Of these three measures, the one that is most strictly related to a choice of any sort is the last one. After all, the total number of abortions and the number of abortions per thousand women will rise and fall depending on whether more women are age 22 or 44. On the other hand, a woman that is capable of carrying a child to term is also capable of having an abortion, and vice versa in most instances.
Thus, we can compare the correlation of real income of males to the ratio of abortions per thousand live births to the correlation of real income of females to the ratio of abortions per thousand live births. Regardless of the direction of that correlation, we expect the absolute value of that correlation to be higher for females.
Note that the census table also provides data for abortions by Whites and abortions by Blacks. Thus, we can test for all groups, Whites only, and Blacks only. (Data on real median income comes from real median income for males and females of all ethnic groups, real median income for White males and females, and real median income for Black males and females is also available from the census.
Finally, because real incomes have tended to rise, and because abortions have tended to fall since the early 1980s one might want to also consider in effect “detrend” the data. (Note – this is not true for Black people – abortion rates per thousand live births among Black women have tended to bounce around in a narrow range, never really decreasing, throughout the entire period. Its outside the scope of this post to speculate on why.) A fairly standard way (at least for econometricians) to make the data “stationary” is simply to take either the first difference of the series. I’ve always thought looking at the percentage change in the series is a bit more intuitive, so I’ve done that instead. (I hope readers with a strong background in statistics will forgive my lack of precision in this paragraph.)
The tables below do all of this. Tables 1, 2, and 3 look at the correlation between real income and the ratio of abortions per thousand women. Each table shows this correlation for the real income of both men and women, and the tables show this information for all ethnic groups, White people, and Black people.
Tables 4, 5, and 6 are the same, but they look at the percentage change in each series. What do we see? Well, in every single case – every single one, the correlation is higher for women’s income than for men’s income. (Granted, on the very last table the two figures are close, but even then the result holds.)
From this, it seems very likely that the median income of women is either affecting the rate of abortions (or at least being moved by the same factors as those that affect the rate of abortions). Now, I would assume at most a fraction of a percent of Americans would disagree with the idea that ideally there would be fewer abortions, whether they feel abortions should be made illegal or, as Bill Clinton put it, “safe, legal, and rare.” My reading of the data – if one wants to reduce the likelihood of a woman ending her pregnancy with an abortion (except among Black women), one should try to increase the real median income, and probably, in particular, the real median income of females.
As always, drop me a line if you want my spreadsheet.
It occurs to me I need a post looking at how real income behaved across administrations. Hopefully, it will get written and posted in the next few days.