[e]

2008: Asset Prices Down, Wealth Inequality Up

(January 2009)

A recent paper by Parker and Vissing-Jorgensen (2009) argues that consumption and income inequality are both procyclical (at least in the US, since the early 1980s), and on the basis of this evidence they predict that the recession beginning in 2008 will lower inequality in both distributions. (See also an article in the Wall Street Journal by Robert Frank (2008) which makes the argument that both income and wealth inequality are falling.)

Their interpretation is that "high-income households currently bear an inordinately large share of aggregate fluctuations." This note argues that in fact wealth inequality has probably risen in the course of 2008 because of the leverage of the middle classes. (However, that conclusion is then reversed if we include an estimate of human capital.)

Data on Wealth Distribution

The Federal Reserve collects a "Survey of Consumer Finances" every 3 years. The most recent data available is for 2004. Below I show net worth broken down into four categories. The distinction between financial and nonfinancial assets is the survey's own, but I have divided financial assets into safe (transaction accounts, CDs, savings bonds, bonds, cash value life insurance) and risky (stocks, other managed assets, other), and I have treated the remaining categories as 50% safe and 50% risky (pooled investment funds, retirement accounts).

percentile of net worth	(1) safe financial assets	(2) risky financial assets	(3) nonfinancial assets	(4) debt	(5) net worth	(6) change	(7) % change	(8) income/net worth
less than 25	2	1	17	21	-1	-4	249%	-1705%
25-49.9	11	6	86	56	47	-20	-42%	90%
50-74.9	38	25	214	93	185	-53	-29%	33%
75-89.9	124	96	409	103	526	-120	-23%	17%
90-100	610	736	1981	214	3113	-691	-22%	8%

Table 1: assets by wealth quantiles, from the Federal Reserve Survey of Consumer Finances 2004, thousands of US dollars.

The results are divided into five net wealth quantiles, using cutoffs given by the SCF summary tables. Columns (1)-(4)in Table 1 show average values for each of four asset classes, measured in thousands of US dollars, and the fifth column shows net worth. The sixth column is calculated by multiplying nonfinancial assets by -20%, and financial assets by -40%, which are approximately the changes in house prices, and US stock prices, respectively, since summer 2006. Column (7), the most important, shows the predicted proportional change in net worth. This column shows a monotonic relationship: the wealthier groups have lost a smaller proportion of their net wealth. Note that the average net worth of the poorest quartile is negative in 2004, this net worth is predicted to have become even more negative, but it remains small as a proportion of their income (as shown in column (8)). This finding, that the middle classes have lost proportionately more of their wealth than the rich, casts some doubt on Parker and Vissing-Jorgensen's interpretation that "high-income households currently bear an inordinately large share of aggregate fluctuations," and a sentence from the Wall Street Journal, "If history is any guide, the upheaval already is shrinking inequality and could continue to narrow the wealth gap."

Qualifications

1. The conclusions are altered if we include human capital (the present value of future labour income) as a part of human capital. Taking present income as proportional to human capital, and multiplying it by 10 to approximate a present discounted value, we find much smaller decreases in net worth, and with the opposite ranking (ascending, the losses to total net wealth are: 1%, 4%, 7%, 9%, 12%). If the income distribution is procyclical then this reversal effect will be even stronger. Also, if the wealthier are typically older, then their PDV factor will be smaller, again contributing to a magnifying of the loss of the wealthy.
2. PVS compare the populations in different bands of consumption, whereas I have divided the population according to net wealth. I have also calculated the changes in net wealth according to band of income (defined in the SCF summary tables), and find the following results, showing that those on higher incomes also have lower proportionate losses of wealth:

   Percentiles of income	change in net wealth
   less than 20	-37%
   20-39.9	-32%
   40-59.9	-28%
   60-79.9	-23%
   80-89.9	-22%
   90-100	-21%

3. These figures could exaggerate the losses borne on housing if a significant share of households have "non-recourse" mortgages, such that the debt can be settled by returning the house, even if the value of the house is less than that of the debt.
4. There may be some capital loss in "safe" assets through higher default probabilities or inflation. However, inflation has been reasonable over 2008, and the Moody's index of AAA monthly bond yields shows no very large change in yield. See http://research.stlouisfed.org/fred2/series/AAA/119
5. There may be significant reversion to trend in house prices, which would make housing wealth difficult to account for properly.

References

Parker and Vissing-Jorgensen (2009) "Who Bears Aggregate Fluctuations? Estimates and Implications for Consumption Inequality"

Piketty & Saez (2003) "Income Inequality in the United States, 1913-1998", QJE

Frank, Robert, "Wealth Gap Is Focus Even as It Shrinks", WSJ October 27 2008