GDP per Household and Average Household Income
Paul Krugman notes that average household income has grown markedly less than GDP per household since 1973.
He discusses the two most important factors. The cost of employee benefits (health insurance) have increased more than wages and are not counted in personal income, and household income is top coded — the very highest incomes are recorded just as more than x ($250,000 last I checked) so the increased income of the very rich does not appear in average household income.
I consider four other factors but don’t believe or don’t have much to say about three (all after the jump). The one that interests me, causes me to invade Rebecca’s turf and may have caused me to demonstrate my ignorance of NIPA terminology is GDP vs Net DP, which means net of depreciation of capital. GDP-NDP isn’t income and, therefore isn’t any household’s income.
I tried to Fred and got this graph
I think that (GDP-NDP)/GDP has increased about 2.5% of GDP since 1973 so about 2.8% of NDP. Caveat Lector, I might be making a fool of myself.
The idea is that information equipment (computers) has become a larger and larger fraction of the capital stock and that it loses value very fast due to technological obselescence.
Three other possible explanations are discussed after the jump
1. Undistributed profits including the increase in assets of nonprofit corporations (like most hospitals). Part of GDP but anyone’s personal income. Doesn’t do the trick — hasn’t grown much.
2. Under-reported income. Capital income seems to be under-reported. I would guess mostly honest mistakes but maybe a bit of tax evasion times don’t trust that the BLS doesn’t talk to the IRS (it doesn’t squeel dear reader — answer honestly) . The share of capital has grown. I think one could look at reported capital income compared to NIPA capital income minus reinvested profits, but I won’t try that right now.
3. GNP vs GDP we have gone way into debt. National income should be decreasing compared to GDP. But it isn’t (point of an informal debate Krugman had at NBER in 1988 with Summers when Krugman said “zero isn’t an especially important number.” Brad DeLong has blogged about the debate.
Robert –
I assume you’re refering to this post.
http://krugman.blogs.nytimes.com/2011/02/02/prices-and-plutocrats/
You have refrained from mentioning the obvious: Wealth has been massively redistributed from you and me to the already wealthy.
PK has a follow up.
http://krugman.blogs.nytimes.com/2011/02/03/economic-growth-and-household-income/
Notable quote:
A lot of the difference goes away when you focus on mean rather than median income; this tells you that the gap between economic growth and median incomes has a lot to do with rising inequality. The remaining gap probably reflects two things: additional increases in inequality not captured by the Census data thanks to top-coding, and a growth in benefits that aren’t counted as cash income.
Not that much of a mystery, then — but it remains striking how little of growth has trickled down to the typical family.
Here’s a bonus.
http://jazzbumpa.blogspot.com/2011/02/real-disposible-income-per-capita.html
Cheers!
JzB
Krugman’s chart shows that GDP growth is concentrated well “above” the lower 50% of the population.
Mean income going up faster than median indicates that the rich are getting richer and the poor are not getting poorer, but are getting relatively less, quickly.
What would one say if the median were higher than the mean?
What say you a “simple” thought exercise on the mean being lower than the median on Krugman’s graph about change in GDP per household?
Would that indicate lower income share of GDP growth was higher?
Would that make a better society?
Why? Why not?
I am not at all addresses the difference between mean and median, since I compare mean household income to GDP per household. Your comment is off topic.
Krugman said, absolutely correctly, that mean household income has increased more than mean household income because income inequality has increased. He did not assert that mean/median must always be a reasonable index of inequality. However when one compares actual real world income distributions (including USA 1973 vs USA 2007) a *large* increase in mean/median always corresponds to a large increase in all indices of inequality which anyone considers reasonable and, indeed, to Lorenz dominance. Note the key (weasel word) “large.” If it is ambiguous whether distribution A or B is less equal (neither Lorenz dominates the other) thenmdifferent indices can give different rankings with, for example, A more equal than B by Mean/median and entropy and B more equal than A by Gini) However, when all indices agree, it is reasonable to say that all show increased inequality and that all have increased because inequality has increased.
I am not saying that mean/median is a good index of inequality. It doesn’t satisfy minimum conditions which generally are required to call a number an index of inequality.
In passing I mention two things. First, mean/median often is used as an index of inequality. Second Krugman did not use it as an index of inequality.
He didn’t say we know inequality has increased *because* mean/median has increased. He said inequality has increased, because inequality has unambiguously increased. As far as I know, no one who has any basic knowledge of the data denies that the US income distribution has become less equal since 1987 (those who are determined to argue say that income inequality is a bad measure of inequality). Your attempt at a point has no relevance to any actual debate about the change in US inequality since 1973 which is of any interest to anyone who has basic knowledge of the relevant data.
OK back to mean/median. It is possible for distribution A to be undeniably more equal than (Lorenz dominate) distribution B but for A to have a higher median/mean. However this does not occur with actual real world income distributions. This is a stylized fact about income distributions (which is well known to anyone who knows much of anything about income distributions). It is not a claim about all distributions.
Krugman asserted two tgings which are definitely true : US income has increased a lot and that unambiguously inequality increasing change in the distribution has caused the growth of the mean to be greater than the growth of the median. There is no trace of any questionable claim about mean and median household income in Krugman’s post. None. Arguing with that part of the post is like arguing whether the earth is flat.
Ooops I left out the link to Krugman. It was to the follow up. I will now add the link.
Since I was discussing mean household income vs GDP per ousehold and didn’t mention median household income, increased inequality is relevant only because of top coding. In the post (to which I should have linked sorry) I am talking about wy the third bar is higher than the second, not why the third is higher than the first. In theory this should have nothing to do with income distribution (in practice it does have to do with income distribution because of top coding).
Also people like me have become richer due to the shifts in the US income distribution. Not as much as the super rich, but more than the average household. I haven’t personally, because I moved to Italy. Anyway I am of the class of relatively rich whose income has increased compared to mean income (and increased more compared to median income).
So I managed to confuse myself.
That’s always a success.
JzB
Ooops I left out the link to Krugman. It was to the follow up. I will now add the link.
Since I was discussing mean household income vs GDP per ousehold and didn’t mention median household income, increased inequality is relevant only because of top coding. In the post (to which I should have linked sorry) I am talking about wy the third bar is higher than the second, not why the third is higher than the first. In theory this should have nothing to do with income distribution (in practice it does have to do with income distribution because of top coding).
Also people like me have become richer due to the shifts in the US income distribution. Not as much as the super rich, but more than the average household. I haven’t personally, because I moved to Italy. Anyway I am of the class of relatively rich whose income has increased compared to mean income (and increased more compared to median income).
I’d like someone to do these numbers again: adjusting for the change in household size over the decades. It’s quite substantial, the shrinking of that size.
I don’t know whether or how it would change the numbers, but I sure think it would be interesting to find out.
The ratio Kuznets (Capital in the American Economy, 1961) concentrates on quite heavily is capital consumption/gross fixed investment. What percent of investment is replacement of consumed capital?
Showed that it had been rising since 1870
Also cap gains aren’t included in personal income, so I assume also not in household income.
There are good accounting reasons for this, but it does give an incredibly skewed view of what most of us think of as “income.”
Cap gains are also not included in GDP.
“I am not saying that mean/median is a good index of inequality. It doesn’t satisfy minimum conditions which generally are required to call a number an index of inequality.”
In particular, when you’re talking households or families, if divorce is and becomes more common among the poor than the rich (which it is), then that affects the median/mean significantly even though it’s arguably not a very good metric for inequality. In an extreme example, if you have two married couples, everyone making $40,000 each, then if one couple divorces mean household income drops from $80k to $53k, but median household income drops from $80k to $40k. Neither the drop in household income nor the median mean ratio is all that meaningful in such an example.
That is, however, irrelevant to a post that compares mean to mean.
Haven’t seen anyone mentioning the fly in the ointment here which is “Personal Income.” If you have a household of two people making 50k apiece your household income is 100k. But if both people get a raise to 70k a year and move out on their own, there has now been a “drop” in household income from 100k to 70k, but an increase in Personal Income of 50k to 70k. These two things should ALWAYS be talked about TOGETHER. Just like I see “Outsourcing” talked about incessantly, but the word “INSOURCING” almost never occurs in these artices I read about job loss in the US.