Summary of Keister & Moller 2000 This review summarized wealth inequality in the form of net worth. Authors examined empirical evidence of wealth accumulation and distribution, presented estimates of trends in wealth inequality, explored a diverse collection of research that explained those trends, including treatments of aggregate influences and individual and household factors. Finally, they talked about intergenerational processes and wealth mobility. In early 1990s, US was the most unequal industrial society in the world. Top one percent of American wealth holders could own up to 40% of total net worth. Authors mainly looked at literature in net worth, not income, because they argued that income alone was inaccurate estimate for real financial status. In another word, income did not have high enough correlation with net worth. An example could be the Gini index for the U.S. was 0.52 if we looked at income but went up to 0.84 if we looked at net worth. Authors thought that by looking at net worth as wealth, we would get a more accurate and more severe number of inequality. Empirical data for examining wealth mainly came from survey data, estate tax data, and government s aggregate estimates of household wealth. But there were a variety of problems such as underestimate the wealth of rich and lack of long-term coverage of the same population. Researchers like Wolff, Greenwood, Keister and so on made their efforts to deal with data disadvantage. As for the trends in wealth inequality, if we asked one simple question that how much wealth were owned by top one percent of people in the U.S., data told us that the overall trend for this amount had been increasing from 1950s to 1990s. In 1950s, top one percent owned 35% total household sector wealth. It went down in 1970s but went up again in 1980s and 1990s. In 1995, to one percent owned nearly 38.5% total wealth. Two things were worth mentioning: among these top one percent, what the top one and half percent owned increased, according to data in 1983 and 1989; and that what bottom 80% owned declined a few percent during the same time. Factors of wealth inequality were explored in this review. On a macro level, market fluctuations had impact on individual portfolio behavior thus the value of combination of assets that families owned changed over time. Social pressure, solicitous insurance sales man, unstable financial market in this process could also make wealth accumulation different. On a micro level, family income, age, race and family structure could all have effects on savings and wealth. For example, education differences, structural barriers and discrimination, portfolio behavior and consumption, and family structure (gender, marriage, family size, divorce etc.) could be factors of race gap in wealth. To integrate impact of macro and micro factors, a few researchers looked at things such as how demographic characteristics influenced percentile position in wealth distribution, how surplus theory explained wealth inequality, and levels of aggregation. Our authors finally looked at intergenerational processes, mobility and inheritance that tried to relieve extreme wealth inequality. By using baby boomers for example, research showed that baby boomers who had higher incomes and more accumulated wealth, also had fewer children. Research also found weak but 1
statistically significant relation between parents income with children s income. But research on wealth mobility were still rare and that wealth research were limited by only looking at certain groups of people. More in-depth and broad research covering more demographics were needed. Summary of Killewald et al. 2017 This latest review on wealth inequality started with a few methodological concerns facing wealth research: how to address challenges to causal inference posed by wealth s cumulative nature and how to operationalize net worth given its highly skewed distribution. Authors then overviewed data sources available for wealth research; explored the trends in wealth levels and inequality and evaluated wealth s distinctiveness as an indicator of social stratification. Authors also reviewed recent empirical evidence on the effects of wealth on other social outcomes, as well as research on the determinants of wealth. To be short, our authors mainly looked at three themes: challenges of studying wealth, data that wealth researchers got, and effects and determinants of wealth inequality. For Part I, authors addressed concerns of studying wealth, especially on measurement and stacking specification error on net worth. They technically examined research and introduced alternative approach to measuring wealth more accurately: model wealth accumulation. They listed a variety of problems and explored studies that might give a clue to tackle those problems. For example, reverse causality concerns on wealth may be tackled by Panel methods estimating within-individual change; lack of long-term data might be tackled by marginal structural models. Regarding the fact that wealth distribution is highly right-skewed, they summarized different methods to deal with it, such as using median regression, log-transform net worth, using inverse hyperbolic sine (IHS) transformation and so on. Our authors also considered two conditions of wealth, one was wealth being the independent variable, the other was wealth being dependent variable. Authors gave different techniques regarding different roles. For Part II, authors summarized wealth data and patterns. They included an overview of wealth surveys including 12 U.S. national ones, 2 subnational ones; 13 surveys from 10 different countries and 4 comparative surveys. As for wealth trends, they looked at the data of 1989 to 2013 collected by Pfeffer & Schoeni (2016) and found that top one percent of American rich owned 32% total wealth in 2001 and 36% in 2013. To demonstrate results sensitivity to different variable transformations, authors estimated and compared different wealth-income correlation across time. They concluded that the wealth-income correlation decreased over the past quarter century. In Part III, authors examined evidence on wealth consequences and determinants. Wealth as predictor could have impacts on a few social outcomes and income. For example, more parental wealth could lead to greater offspring education (educational and cognitive achievement, college enrollment etc.), better family labor market outcomes, and homeownership for children. Individual wealth could influence transitions to homeownership, self-employment, family structure, cultural signal of status and achievement, political power and so on. As outcome, wealth could be 2
influenced by a variety of factors including income, housing, risky assets, family structure and health. Authors also examined exogenous predictors of wealth in detail, including age, social origins, education, race, gender, macro-level context and policy. They concluded that wealth was an important dimension of social stratification and proposed future research attended to better decisions about appropriately operationalizing net worth; improvement on data availability; studying components of wealth that may illuminate the causal processes; establishing the causal role of endogenous processes (e.g., marriage, portfolio composition, self-employment, homeownership) needed to illuminate the pathways generating wealth disparities by race, gender, and social origins; more qualitative research to understand wealth generation and use; more research on group disparities of other racial/ethnic groups, nativity, and gender; and more comparative research that could reveal the macro-level determinants of wealth levels, intergenerational wealth mobility, and wealth inequality. Summary of Other Articles Piketty 2014 examined the history of the distribution of income and wealth. Building on prior publications over three centuries and over 20 countries, Piketty showed that wealth inequality followed a U-shaped trajectory across most developed countries since 1900, with the upswing occurring in the United States since about 1970 and in Europe since about 1980. Piketty argued that temporary decreased wealth inequality might not be a relief because it could be due to war-induced asset devaluation, high tax rates, and skills investments stimulating economic growth. He argued that rate of return to capital overtaking the economic growth rate also contributed to wealth inequality. In addition, his findings showed that developed countries generally experienced similar trends in wealth inequality through the twentieth century despite having different inequality levels. Saez & Zucman 2016 tried to estimate the distribution of wealth in the United States since 1913 by combining income tax returns with macroeconomic household balance sheets. They capitalized incomes from individual taxpayers and tested their method in three micro datasets: Survey of Consumer Finance, linked estate and income tax returns, and foundations tax records. According to their observation, wealth concentration was high in the beginning of the twentieth century, fell from 1929 to 1978, and increased since then. On the one hand, top 0.1% people owned 7% total wealth in 1978 and the number increased to 22% in 2012. As for characteristics of top wealth-holders, they were younger today than in the 1960s and earned a higher fraction of the economy s labor income. On the other hand, the bottom 90% wealth share first increased up to the mid-1980s and then steadily declined. Authors argued that the increase in wealth inequality in recent decades was due to the upsurge of top incomes combined with an increase in saving rate inequality. Wolff 2000 examined American wealth condition from 1983 to 2007. The household wealth inequality showed a sharp increase from 1983 to 1989 and little change from 1989 to 2007 in his research. The gains of 2001 to 2007 period were largely based on rising home prices financed by increasing mortgage debt. The author analyzed that the high level of mortgage indebtedness of the past three decades also 3
made the middle class vulnerable to the collapse of the housing market at the end of the decade of the 2000s. Regarding home value as a share or total assets from 2001 to 2007, there was a large increase and corresponding fall of the share of stocks, pension accounts, net equity in owner-occupied housing in total assets. As for racial disparity, the mean of African Americans was only 19% that of white families in total assets in 1983 and remained stable over the years. The mean wealth of Hispanic households was also very low compared to non-hispanic whites in 1983, a ratio of 0.16. As for age disparity, young household had a worse wealth position over the years 1983 to 2007, which exposed them to the joint collapse of the stock and housing markets at the end of the decade of 2000s. Most of our reading this week focused on race and wealth discussion. Oliver and Shapiro 1997 focused on racial inequality in America and specifically looked at black middle class and their economic well-being on SIPP data. They looked at qualitative interviews and examined major factors to wealth disparity including family structure, age, education, occupation, and income by studying cases. In general, they found that overall ratio black-to-white income ratio narrowed from 0.62 to 0.7 (and 0.76 for college graduates) by looking at data in 1967, 1964 and 1988. Regarding net worth, they found the least amount of inequality occurring among middle-income earners, but the income amount differed even within similar level household. As for the most inequality, they found it in white-collar occupations, and the number went as for every dollar owned by the white middle class, the black middle class owned fifteen cents. Authors presented finding from interviews as major factors to wealth inequality. After looking at monetary reward of increased education level, they found that more formal school raised income and wealth for whites and blacks and narrowed income inequality in the process. For example, by finishing high school, whites can increase household incomes by $5,774 over those not finishing high school; and black can increase income by $2,810. However, the gap was still there and at lower levels blacks gained less. In addition, their discussion on family structure and children revealed that most white families rearing children functioned on incomes well above the poverty line, while only those black families with one or two children could operate with budgets above the poverty line. Even small black families must bear incomes amounting to only 60% of those of their white counterparts as the average family raising three children needing $12,286 in income at disposal for black family and $30,151 in income for white family, for example. Keister 2000 did a study on the impact of racial differences in asset ownership on the distribution of household wealth. He used standard regression methods for the ownership of seven assets using 1983 to 1986 panel of the Survey of Consumer Finances (SCF) data. And He applied a simulation model to investigate the role that patterns of asset ownership had on the distribution of wealth, by looking at effects of racial differences in family wealth history, earnings, education, marital behavior, fertility, and others. Keister estimated how changes in historical patterns of portfolio behavior and educational attainment would have reduced inequality. To be specific, income had a clear positive and significant effect of the log-odds of ownership of each asset; whites were more likely than blacks to buy high-risk, high-return assets thus the 4
net worth of whites was likely to increase faster than that of blacks. Regarding race and wealth, race seemed to be indirect effects on wealth ownership and accumulation patterns, as even after completely removing the direct effects of race on asset ownership, the vast majority of wealth holders were white and wealth inequality remained extreme. The author suggested that while numerous factors may contribute to the creation of wealth inequality, much of the existing disparities could be alleviated by policies that encourage blacks to own assets that were likely to increase their net worth. Making high-risk asset ownership accessible and understandable to black families would possibly reduce the current dramatic disparities that are perpetuated by black-white differences in asset ownership. Scholz & Levine 2003 did a detailed review on race and wealth. They cast suspicion on trend data presented by Wolff 2000 which argued that wealth disparities continued to widen from 1989 to 1998 and favored findings by Kennickell 2000 that no statistically significant changes happened between 1989 and 1998 in the Gini coefficient. But for racial disparity, authors agreed that the facts about black-white wealth inequality were striking. They showed mean and median net worth for different age groups for white households and for black and Hispanic households, and found wide net worth disparities are apparent at every age. For instance, at ages 51 to 55, mean (median) net worth of white households was $467,747 ($156,550), while for black and Hispanic households it was $105,675 ($33,170). Authors then examined the extent to which various factors contributed to racial disparities in wealth accumulation and to observed wealth inequality. They suggested that: differences in income and demographic characteristics were the most important factors explaining black-white wealth differentials; factors such as transfers and family background (important), portfolio choices (overall modest role, but particularly due to differences in stock ownership) also had influence on racial difference on wealth; savings behavior was not strongly related to race and therefore did not contribute significantly to the black-white wealth gap etc. Our last reading, Altonji & Doraszelski 2005, focused on the role of permanent income and demographics in black and white differences in wealth. By using sibling fixed-effects models to control for intergenerational transfers and the effects of adverse history, authors argued that differences in savings behavior and/or rates of return played an important role in wealth inequality. To be specific, our authors looked at the combined effect of factors by estimating models for the growth of wealth as a function of income and demographic factors. They found that in the case of couples, income and demographic factors could explain 74% of the race gap in the growth of wealth when they used the growth model for whites but only 49 percent when they used the growth model for blacks. The discrepancy was even larger (84% vs 30%) when they studied growth in the ratio of wealth to permanent income. These findings mainly suggested that difference in savings behavior or rates of return on assets could play a key role in explaining the wealth gap. In addition to this, authors also found a higher self-employment rate and a stronger link between self-employment and wealth for whites than for blacks, which made an important contribution to the wealth gap between white and black couples. 5