Every once in a while I toy with the thesis that someone ought to make a big deal about the fact that a lot of the standard statistical data about the United States that we track is of a kind of low quality. One noteworthy example is the poverty rate formula, which is basically nonsense. Apparently Barack Obama believes we should change it. This is a good idea on many levels, though one issue any time you change anything like this is that you don't want to lose the ability to track trends across time.
« More Or Less | Main | Maliki's Game-Changer »
Woo Data
19 Jul 2008 03:13 pm
Comments (17)
Every once in a while I toy with the thesis that someone ought to make a big deal about the fact that a lot of the standard statistical data about the United States that we track is of a kind of low quality. One noteworthy example is the poverty rate formula, which is basically nonsense. Apparently Barack Obama believes we should change it. This is a good idea on many levels, though one issue any time you change anything like this is that you don't want to lose the ability to track trends across time.
If the new poverty rate formula yields a lower poverty rate than the current one (that is, if, under the new formula, Americans who were previously considered to be living in poverty are no longer considered to be living in poverty), would you still consider it a "good idea?"
Income-based poverty measures are fundamentally flawed because income is such a poor measure of people's actual material conditions of life. If we're serious about measuring poverty, we need to look at consumption data rather than, or in addition to, income data.
"one issue any time you change anything like this is that you don't want to lose the ability to track trends across time"
Huh? I thought the reason you change how these statistics are calculated is because you want to avoid inconvenient comparisons with the past. How we calculate unemployment has changed many times. Everytime it has changed, the number has gone down. Politicians don't want accurate comparisons to the past, they want numbers that make them look good. So they change the calculations to achieve that effect. The rest of us might not like it, but we don't really matter, do we? And few of us even notice the changes, so there is no reason to stop doing it.
Re: If we're serious about measuring poverty, we need to look at consumption data rather than, or in addition to, income data.
Nonsense. Consumption doesn;t tell us much either, since it may bet hat consumption is being financed by borrowing. I would suggest income minus fixed obligations (e.g., an individuals's cost of housing) plus wealth minus debt is what we should look at.
I believe poverty formulas are fairly simple and based on the cost of food. The costs of housing and health care have changed far more quickly than food costs since the original definitions were established. Here's more on Mayor Bloomberg's efforts to change the definition of poverty in NYC with much more background on why it matters.
http://www.nytimes.com/2008/07/14/nyregion/14poverty.html
one issue any time you change anything like this is that you don't want to lose the ability to track trends across time.
If the formula is "nonsense," then it presumably hasn't been tracking any useful trends up to now, so I don't see how correcting it would raise this issue.
JonF,
Nonsense. Consumption doesn;t tell us much either, since it may bet hat consumption is being financed by borrowing.
You're the one who's talking nonsense. Most consumption is not financed by borrowing. And even where consumption is financed by borrowing, so what? How does that mean it's not a measure of standard of living?
I would suggest income minus fixed obligations (e.g., an individuals's cost of housing) plus wealth minus debt is what we should look at.
Why?
There is no poverty data to be lost. There is economic data and a poverty formula. You can change the formula and still have unchanged access to historic trends.
The original income threshold levels were based on food prices, but the ADJUSTMENTS have been based on overall price trends, not just on food prices. So the fact that other things have gone up more (or less) than food is already reflected in the updated thresholds.
So the main concern in evaluating trends over time is how well the CPI reflects actual changes in the cost of living for poor people.
I agree, as well, that consumption data would be useful. There are lots of people whose income may be temporarily low, but who have plenty of resources for the long term and shouldn't really be counted as part of the poverty problem.
The original poverty line was based on food, then adjusted to the percentage of a family's budget that was spent on food. This has since been adjusted to inflation - however, the proportions of a family's budget normally spent on food has shrank as housing, health care, education, and debt service become much larger expenditures than previous.
If you want to get really technical, many people have argued that the original food budgets used to represent the minimum amount of food were pitched at a bare above-starvation level, and represent a diet that is really lacking in nutritional standards (especially re: fruit, vegetables, vitamins) and really high in starch (especially in corn).
If they want to change the formula but not lose historical trends, all they have to do is count both; call one the Historical Poverty Measure and the other the Adjusted Poverty Measure and report both. Moreover, you can probably adjust backwards in time to construct historical series for the APM without too much difficulty.
This is a good idea on many levels, though one issue any time you change anything like this is that you don't want to lose the ability to track trends across time.
Well, they're not going to just find all the previously recorded stats and have a book-burning outside the library of every major university just because they adopted a new formula to measure something like poverty.
Tomorrow's experts, whatever formula they use, are going to be able to make comparisons to the past based on plugging whatever basic statistics they use into the formula for past years, just like they will for future ones. Presumably all of the mundane individual figures we take to plug into a formulaic measure like the povery measure you're referring (or ones that are as good as replacements for them) to are going to be ones that are of such interest to a broad range of disciplines and research topics that we will have been collecting data on them for a while.
I agree, as well, that consumption data would be useful. There are lots of people whose income may be temporarily low, but who have plenty of resources for the long term and shouldn't really be counted as part of the poverty problem.
Yes. People such as retirees drawing down their savings, and medical students temporarily living off loans who will making hundreds of thousands of dollars in a few years.
Liberals tend to oppose consumption-based measures of living standards because they show that "poor" people are actually much better off than income data alone implies.
The poverty-line calculation in the US is very flawed in the respect that it has a standard measure of poverty for the entire United States, regardless of cost of living in a given area. Thus, the same measure of poverty is applied to those living in a high cost of living area such as New York City as those applied to a low cost of living area such as backwoods Mississippi. In comparison, Canada uses a poverty calculation that assesses the cost of living in different metropolitan areas and establishes different guidelines that reflect that. The Canadian system is obviously much better at establishing some kind of consistent poverty statistic.
Also, I don't see why a new poverty statistic couldn't be developed while maintaining the current poverty statistic as a means of tracking changes over time. Sure, it might be more confusing to have two poverty lines, but it seems to be a sensible compromise.
The poverty-line calculation in the US is very flawed in the respect that it has a standard measure of poverty for the entire United States, regardless of cost of living in a given area. Thus, the same measure of poverty is applied to those living in a high cost of living area such as New York City as those applied to a low cost of living area such as backwoods Mississippi. In comparison, Canada uses a poverty calculation that assesses the cost of living in different metropolitan areas and establishes different guidelines that reflect that. The Canadian system is obviously much better at establishing some kind of consistent poverty statistic.
Also, I don't see why a new poverty statistic couldn't be developed while maintaining the current poverty statistic as a means of tracking changes over time. Sure, it might be more confusing to have two poverty lines, but it seems to be a sensible compromise.
Re: Most consumption is not financed by borrowing.
I agree. However, if consumption is not financed by borrowing thenm it must be financed by either income or by a drawdown of assets (AKA wealth). The only other possibility is theft, and you will agree that that is even less likely a source of consumption than borrowing?
Re: Why?
My formula is is basically similar to how we assess a business' economic viability. Debt matters. So do assets. So do fiuxed costs.
Of course it's crazy to set the poverty line based on a 1960s food budget (adjusted to inflation) while ignoring housing, health care, transportation, and everything else.
But an even more important reason to update the poverty measure is that family incomes are calculated solely on cash income -- ignoring non-cash benefits like housing vouchers, food stamps, child care subsidies, medicaid coverage, and the earned-income tax credit... each worth thousands of dollars per year.
That means we don't keep track of how much benefit these subsidies provide in reducing poverty. Which is ridiculous given that these kinds of non-cash subsidies -- many of them available only to the working poor -- are pretty much the only kind of anti-poverty program that get funded these days.
penis büyütücü
Comments closed August 02, 2008.
| 
Heads in the Sand
Um, a little more information please? Why is the poverty rate formula nonsense? What is the proposed change?
You allude to the fact that you know something about this topic, but neither the post or the link gives any substance at all.
I agree, however, with your thesis on statistical quality of government/social measures.
Posted by right | July 19, 2008 3:43 PM