Monday, January 13, 2014

Fifty Years Ago

Fifty years ago this past week, LBJ declared war on poverty.

Poor people lost.

Fig.1 The official government poverty rate: ominous signs.
Or maybe not...

TOF's first question is: what caused that precipitous drop in poverty rates prior to the War on Poverty. The short answer is that TOF hasn't a clue. Perhaps the wartime economy resulted in poorer households and things were recovering toward normal in the 50s and 60s. Or perhaps the transition from rural to urban was just then running to completion.  (Late Moderns may not realize the extent of rural poverty in the 1940s and earlier.)  Or the general prosperity inherent in being the only large economy on Earth that was not bombed flat in the 40s. Or else the data collection was only just getting started and the numbers are really incommensurable to later measurements.

TOF's second question is: WTF? The War on Poverty seems to have stopped the drop! Since 1967, official poverty rates have fluctuated between roughly 15% and 12% in what looks like a 25 year cycle. There is no secular trend in the metric.

TOF's third question is:


Are We Measuring Poverty Properly?

Faithful Reader may recall that a quality is defined by the means used to measure it. Count things differently, and you get a different count. What we sometimes discover is that we are not measuring what we think we are measuring. We are only measuring what we have been able to measure. Those who remember the Vietnam War may recollect, if they did not inhale, that "body count" was used to measure progress in that conflict.  That turned out not to be not the brightest idea the Best and Brightest ever had.

Poverty rates are the proportion of households that fall below the poverty threshold. "For example, a family with two adults and two children under 18 whose total income is $22,811 is not poor. If the same family makes $22,810, everyone in the household is poor (based on 2011 threshold data)."* Thus, there are two components to the measure:
  • setting the threshold ("specification limit") and 
  • measuring household resources (the number to be compared to the limit)

On the Threshold of Success

Caught by surprise by LBJ's announcement of a war on poverty, the government bureaucracy flailed about a bit to figure out how to measure it.  LBJ knew it when he saw it -- and we must not doubt his sincerity, because he had seen some pretty abject poverty in his time -- but how was the government to know if any success was being made in the war?  Body count would not do.

The USDA had developed an "economy food budget" a few years earlier, and SSA statistician Mollie Orshansky developed that into a "subsistence budget" for poor families by multiplying the USDA figure by three -- the assumption was that families spent one-third of their income on food. She adjusted the figure for different sizes and compositions of families.*
(*) adjusted the figure. See page 18 here.
In the Game of Check and Measure, data that already exists will always triumph over data that must be created. Orshansky's metric was updated for inflation and we were off and running.

 But...


Is the metric really appropriate? Do people still spend 33% of their income on food?  Older people have more medical expenses. Perhaps Medical-Out of Pocket expenditures (MOOP) should be included. Similar questions led PBS to worry that there are really more poor people than the metric identifies.

Are there poor people not flagged by the threshold?  Almost certainly yes.
Are there people flagged by the threshold who are not actually poor?  Again, almost certainly yes.

That's life in the wonderful world of metrology. Type I and Type II errors.  Reduce one and usually the other increases. Some folks would love to inflate the numbers in poverty from a variety of motives.  Others would like to deflate them, from contrary motives. Few are motivated by a love of statistics done well.

Resourceful Americans

Secondly, what resources should be counted against that threshold, however it has been defined?  Some data gleaned from news reports in 1993 and 2005 indicate the effect of the definition on the value returned for comparison to the threshold:

Fig. 2: Different ways of measuring household resources.
In 1993 nearly a quarter of Americans were considered as living in poverty if only their own income was considered. (The 2005 news story did not include this figure.)  However, when "cash transfers" (i.e., welfare check and similar payments) were counted as income, this dropped to 15%.  A further reduction to 12% was had by including the value of non-cash transfers like food stamps, housing assistance, and the like. Finally, if home equity is included, poverty rates dropped to just under 10%.

Line 2 is the Official Poverty Measure (OPM). That's the value plotted in Fig. 1. Line 3 is close to the Supplemental Poverty Measure (SPM) developed at the Census Bureau, but is not exactly the same. SPM was prepared in 2010, et seq.

Note: TOF hopes that no one is so foolish as to look at two points, like 1993 and 2005, and declare triumph. One could as easily pick two other years and declare that poverty has increased.  It all depends on who is president in the second year chosen.

Not included in these metrics are equivalent transfers from private charitable organizations.  (How do we count the price differential at thrift shops?  Soup kitchens? Food banks?  Catholic Charities? Habitat for Humanity?)  Also not included is unreported income, which  studies of spending have noted as substantial. This not only includes various illegal activities, such as the sale of controlled substances (which involve only a small proportion), but also ordinary work "under the table."* 
*under the table. Poor people are not stupid. If they work more than a certain number of hours, their assistance will be cut and they will wind up with less income than if they did not work at all.  A relative of TOF once found herself in college working a part-time job with three such women. The owner took some advantage of this by occasionally stiffing them on their paychecks, knowing they could not report her. (Though to be fair, it was a very marginal sole proprietorship and she may simply have had insufficient revenues to meet the payroll.) The TOFling was not on welfare and could threaten to turn her in, which put an end to things, at least for the time.

In the meantime, the idea of "poor" has changed. While there are some who are objectively poor, most are only relatively poor. Forty-three percent of poor people own their home. (Recall the home equity effect in Fig. 2.)  One-third have a wide-screen or plasma TV; and so on. Now, when the cost of living is high, you can make a lot of money and have a lot of Stuff and still live on the edge; but still... TOF has seen people along Beach Road in Chennai living in grass huts with dirt floors, so there are degrees to this sort of thing.

Most poor people in the US do not go hungry.  The USDA survey of household food security reported for 2009 that
  • 82% of poor adults reported never being hungry at any time in the prior year due to lack
    of money for food
  • 96% of poor parents stated that their children were never hungry at any time during the year because they could not afford food.
  • 83% of poor families reported having enough food to eat.
But to put it another way,
  • 18% of poor adults reported being hungry at some time in the prior year due to lack of money for food
  • 4% of poor parents stated that their children had been hungry at some time during the year because they could not afford food.
  • 17% of poor families reported not having enough food to eat.

(*) Comparing the first and second items is interesting. More adults reported going hungry themselves than that their children had gone hungry. IOW, they fed their children first. This ought not surprise anyone, although it may.
The 2012 figures are comparable to these. "Reported being hungry" is bureaucratized to "very low food security" at the USDA site.

But now the question arises: should OPM be the Official Poverty Measure?
  • What expenditures should be considered when establishing the threshold?
  • What resources should be considered when tallying the income to be compared to that threshold?
How you answer these questions will pretty much determine what poverty rate you will get.  Because there is no such thing as poverty rates: there is only the result of applying a particular definition. The matter of measurement is not trivial. Not only decisions on what factors to include in the metric, but where one would go to secure the data on those factors. Some sampling series by the Census Bureau will harvest the needed estimates; but in other cases, the data has never been collected at all.

The Supplemental Poverty Measure


A recent paper by Wimer, et al. attempts to dig out from under this by modifying the Supplemental Poverty Measure (SPM) and retrofitting it to the period before 2010.  But, disinclined to calculate a new threshold for every year, Wimer "anchored" the threshold to 2012 and used the inflation index to create a single "constant dollar" threshold for past decades. They called this the Anchored SPM. This was reported uncritically by PBS.

A closer reading of the paper reveals that, as in so many of these cases, they are not reporting actual data. Aggregate and summarized data were used. Averages were applied. And an average can conceal as much as it reveals. Some of the numbers they wanted literally did not exist in the past, and they had to be recreated by estimations, extrapolations, and extrications from their rear pants pockets. Other "data" were created from models, and the one thing we know about models is that they are wrong. Some were just big hairy-arsed guesses (BHAG).*  The details are in their paper.
(*) BHAG. TOF does not wish to imply that the choices were whimsical or without foundation; only that they were choices, and different choices would have led to different results.
But friends, that's about as good as it gets in this line of work.  It's not as if social sciences were, like, you know, real science.  Here are the results, which start only in 1967, with the commencement of the War on Poverty.
Fig. 3: Index of poverty rates using Anchored SPM metric.
It plots index numbers (1967 = 100), and TOF hates index numbers with an implacable malice, despite economists feeling the love.  Actual poverty rates are shown in their appendix. The Anchored-SPM poverty rates run higher than the official poverty rates. (TOF was startled by this since resources were increased, but the threshold was also raised, at least for earlier years.)

If Luca Brasi sleeps with the fishes, and Lt. John Dunbar dances with the wolves, the poverty rate writhes with the economy. Overall, the time series mimics that of the OPM, but with a decreasing secular trend. While the OPM shows the overall poverty rates to be unchanged between1967 and 2011: 14% vs. 15%, resp., the Anchored-SPM shows a substantial drop: from 29% to 16%.  IOW, really really different (and high) back in the day; very nearly the same today. 

Absent any poverty program at all, the authors claim, the poverty rate would have remained roughly constant at 27%-29%.  

Now, the purpose of the paper was to show how wonderfully effective government poverty programs have been, so of course their conclusion was that government poverty programs have been wonderfully effective. Its social science.* This does not mean that they are wrong, but their choices in defining the "Anchored SPM" may have been influenced by their goals. Neither the OPM nor the "unanchored" SPM lead to the same conclusions. 
(*) social science. "Social science is rarely dispassionate, and social scientists are frequently caught up in the politics which their work necessarily involves." -- Daniel Patrick Moynihan 
HOWEVER, recall the tabulation above for 1993 (Fig. 2). This appeared in the Wall Street Journal, which can be suspected of motives, but not likely the same motives as liberal college professors: and they reported 
  • 23.4% unassisted poverty, 
  • 15.1% OPM, and 
  • 12.1% (not quite) SPM. 
So the broad-spectrum picture is likely in the ballpark even if the precise numbers are less than... precise. Roughly a quarter of Americans are in poverty and always have been for half a century* but about half of them are being raised above the threshold by assistance of various sorts. 
(*) have been so for half a century. Is this surprising? Serial correlation, dudes. A poor household in year k is likely to be a poor household in year k+1. This really messes up any correlation coefficient you may have thought of calculating.

"The Kids Are Alright."

The most singular feature of Fig.3 is the steady decline in childhood poverty. Let's pop a champagne cork on that one, folks.  Had they been counting actual children, rather than models, group means, estimates, and extrapolations, this would be at least somewhat encouraging. And it may well be.

The authors state:
By 2012, estimates that did not count the resources from non-cash transfers and the tax system would find child poverty at 30% and deep child poverty at 17%, rather than 19% and 5%.
Translation: people are poorer if you do not count all their income.  TOF will write that one down.

So why the decline in the Anchored-SPM and not in the OPM or SPM?

Crossing the Threshold of Hope

Basically, the new metric raises the poverty threshold for early years more than for later years. The thing that makes TOF's head itch is the computation of that anchored poverty threshold of $25,000 (constant dollars) for all years. Even allowing for the use of deflated dollars, this seems guaranteed to make poverty rates fall more markedly.  People in 1967 would have had a harder time crossing the Anchored-SPM threshold than crossing the OPM or SPM thresholds, while for recent years the thresholds are more nearly the same, and so the rates will be nearly the same.
Fig. 4: These are thresholds not poverty rates. To be "poor"
your resources must fall below these lines. Note that it's easier to
fall below the Anchored SPM line, esp. back in earlier time frames.

Inflation was not the only difference in wages and prices between then and now. Fitted sheets were $2.79 in 1965 (a figure TOF happens to have handy, go figure). Inflation would make that $20.64 today. But today, depending on where you shop, you could buy a complete sheet set -- fitted, and flat, and pillow case -- for that amount. A loaf of white bread cost about .25 in 1967 and corruption of the currency would bring that to $1.74 today, yet today's average price is $2.36. So fitted sheets are up less than inflation, while bread loaves are up more.

This is the problem of dynamic balance. When too many different items are thrown in a bucket, each responding to its own causal roots, the aggregate may behave differently than any of the particular items in the aggregate. One increases, another decreases and the aggregate shows no change.  So if poverty is not one thing but a combination of many things, the reason for the stability of the OPM for the past half-century may be that while some kinds of poverty increase, others decrease. 

Filled with deadly vapors!
Technology also matters. The price of an ordinary light bulb has recently jumped from $0.60 for 60W incandescent bulbs to $0.99 for 60W-equivalent compact fluorescent bulbs filled with mercury vapor. However, true costs are life cycle costs, and CFLs last longer than incandescents. But disposal costs are also part of life cycle costs, and disposal of mercury lamps is trickier. Externalities must also be accounted for, and we have the risks associated with the inevitable mercury-releasing breakage in homes or among children, who are more susceptible to neurotoxins. Since poor households must buy light bulbs, too, how are these factors to be accounted for among the fitted sheets and loaves of white bread.

It might could be that an aggregate adjustment on a price index may be less than useful in setting the threshold. A television in 2011 is not the same kind of thing as a television in 1967.  Nor is a telephone.  In 1967 you could not by a burner phone at any price. So, go figure.

It is clear that the 15% households that were poor in 2012 were likely richer than the 14% that were poor in 1967, not because they had more money, but because they could buy better things with that money. In 1970, only 36% of all US households had air conditioning; by the present time 80% of poor households have it. What was once a luxury is now within the reach of the poor.
For decades, the living conditions of the poor have steadily improved. Consumer items that were luxuries or significant purchases for the middle class a few decades ago have become commonplace in poor households, partially because of the normal downward price trend that follows introduction of a new product.
-- Understanding Poverty in the United States
Hence, poor people today are not poor in the same way or to the same extent as the poor of 1967. This is another aspect not captured by the OPM threshold, albeit one not mentioned in the PBS article or the Wimer paper, either.

Table Talk

Fig. 2 shows is that in 1993, official government programs effectively cut the percentage of poor people in half.  But by another measure, we have to wonder.  Does a person cease to be genuinely poor simply because he has become dependent on the largess of another? This is the old give-a-man-a-fish/teach-a-man-to-fish quandary. Certainly, the man must eat while he is learning to fish; but if he is already getting his fish with no effort, what is his incentive to learn?  

And what are the behavioral incentives on the fish-giver?  If so long as the man is dependent he will vote reliably for his padrone, then there are insidious incentives on the piscelargitor to neglect the angling lessons.  In change path visioning, this constitutes a 'rest point' or sub-optimum equilibrium state. There is just enough satisfaction on all sides to stay put and not push on to the next state.
Fig. 5: Change path and rest points in change management.
So the next question to ask is why the OPM poverty rates have been stagnant since the beginning of the War on Poverty (Fig. 1).  We need to ask the questions mentioned in Fig. 5.  What's working now?  What isn't?  What's keeping us in the status quo (besides a reluctance to rock the boat)?  But also: what happens if the status quo continues? What are the forces hindering progress? What are the forces assisting progress? (See: force field analysis)

Conclusion

  • The War on Poverty has been a success because without it poverty would be up around a quarter of the population instead of 12%-15%
  • The War on Poverty has not reduced the OPM for the past half-century; but...
  • Capitalism has brought more affordable goods within the reach of poor households, so that while there are as many (%) poor households as ever, they are materially better off than they once were.
  • The Anchored-SPM has shown a more substantial improvement, but does so largely by raising the threshold for earlier years compared to OPM and SPM.
  • All of this is treating the symptoms and not addressing the causes of poverty. People are not poor because they don't have enough money. That's not a cause, but a definition. It's like saying snow is white because it doesn't have enough color.  Applying a bandage to a wound may be necessary, but it is not sufficient.

8 comments:

  1. Wait, I'm not sure what you're saying after reading this. Do you believe that while there are less poor people than before it's not REALLY a success in the sense that it treats the symptoms of poverty but is not in itself a cure? Like the "bread and circuses" of ancient Rome?

    ReplyDelete
  2. It's true that capitalism has brought consumer goods to poor households, but I don't think that bullet point does enough justice to the mechanisms. I don't think "capitalism" by itself can get you there. It's not like poor people exerted much demand on consumer markets in, say, the 1920's. To get there, you need the twin forces of overproduction and planned obsolescence. It's more complicated than this, of course, but much of the consumer goods owned by the poor is "waste" that is either obsolete or part of an excess that couldn't be absorbed by the market. The reason is obvious: you come up better when you get *some* money for the goods, or even when you give them out for free, when the cost of disposal is factored in. The fact that the poor get big screen TV's is pretty accidental to the capitalist system. (But less so to the logic of consumerism.)

    ReplyDelete
  3. Very interesting analysis. Pretty much confirms what I've always suspected - many of the official statistics we get from the government involve a lot of arbitrary definitions, number massaging, and in some cases pulling numbers straight out of the rectum. How do any of us know what the poverty "rate" truly is, or unemployment etc.? How does one arbitrarily split people into two categories named "poor" and "not poor" using the continuous variable of income? How do we account for the differences in cost of living? What about other resources such as education and culture that aren't captured in official statistics? As Thomas Sowell points out, income isn't static and "the poor" and "the rich" are often the same people, just at different points in their life. Someone who graduates from a top university and takes a menial job to make ends meet until they get their medical school acceptance letter is obviously in a different position than someone working the same job but with limited education and skills. Both would be measured as "poor" by official statistics. Trying to take the measures of millions of people, in vastly different circumstances is a hopeless task, and any official "rate" that comes out of such an analysis is likely devoid of meaning.

    ReplyDelete
    Replies
    1. And there are others who never get picked up in the statistics. For example, the social mobility data comes from studies of tax returns. If someone does not file a tax return because he's living under a bridge in Hoboken, he never shows up in any quintile, let alone an upwardly mobile one.

      That's why we should never get excited by a fluctuation of a few percentage points one way or the other, on any of these time series.

      Delete
  4. Your mention of the infamous body counts in Vietnam reminded me that I spent a lot of time as an operations analyst back then trying to figure out whether we were winning and by how much, if at all. We tried not only body counts buy weapons captured, supplies captured, numbers of VC who "rallied," and on and on. I recall one meeting my group had with one of the nation's top statisticians. His question was, what metric is the enemy using? We had no idea.

    ReplyDelete
    Replies
    1. Deming always said the most important things are unmeasured and unmeasurable. Makes me wonder what metric G. Washington used. Who was the statistician?


      I have not seen your byline in ANALOG in a while. You are that Joseph P. Martino, aren't you? "Not a Prison Make" and so on? I enjoyed your stories when I started reading the mag back in the 60s.

      Delete
    2. Yes, I'm that Joseph P. Martino. I find that I'm still writing John Campbell stories and trying to sell them to his successors, without success. The magazine has changed; my writing style hasn't. I'm currently working on a novel about the first manned expedition to Mars. I have a murder mystery out (THE JUSTICE COOPERATIVE, available from Amazon), and my most recent book is RESISTANCE TO TYRANNY, a primer on how to conduct an armed revolt, also available from Amazon. Back in the '90s I wrote a response to the Catholic Bishops War and Peace Pastoral, in which I applied Just War Doctrine to the use of nuclear weapons, something the good Bishops had notably failed to do (A FIGHTING CHANCE, from Ignatius Press). In short, I've been busy. Thanks for remembering.

      Delete
    3. Joseph, Mike, I can't believe that Joseph couldn't sell one of his novels to Baen, given the topics and themes mentioned. Joseph, have you tried with Toni W. there?

      Delete

In The Belly of the Whale - Now Available

    Dear Readers, Dad's final (? maybe?) work is now available at Amazon, B&N, and many other fine retailers. I compiled a list a fe...