The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Consumption and the Great Recession Mariacristina De Nardi, Eric French, and David Benson Introduction and summary The Great Recession of 2008–09 was characterized by the most severe year-over-year decline in consumption the United States had experienced since 1945. The consumption slump was both deep and long lived. It took almost 12 quarters for total real personal consumption expenditures (PCE) to go back to its level at the previous peak (2007:Q4). In this article, we document key facts about aggregate consumption and its subcomponents over time and look at the behavior of important determinants of consumption, such as consumers’ expectations about their future income and changes in consumers’ wealth positions related to house prices and stock valuations. Then, we use a simple permanent-income model to determine whether the observed drop in consumption can be explained by these observed drops in wealth and income expectations. We begin our data analysis by using macroeconomic data to study the behavior of consumption and its subcomponents. We then use microeconomic data from the Reuters/University of Michigan Surveys of Consumers1 to study nominal expected income growth and inflationary expectations. Our main findings from the macrodata are the following. First, the Great Recession marked the most severe and persistent decline in aggregate consumption since World War II. All subcomponents of consumption declined during this period. However, the large drop in services consumption stands out most, relative to previous recessions. Second, while the decline was historic, the trends in consumption and its subcomponents leading up the recession were not substantially different from past recessionary periods. Third, the recovery path of consumption following the Great Recession has been uncharacteristically weak. It took nearly three years for total consumption to return to its level just prior to the recession. In contrast, Federal Reserve Bank of Chicago the second-worst rebound observed in the data followed the 1974 recession and lasted just over one year. We find that this persistence is reflected most in the subcomponents of nondurables and especially in services. Our main findings from the analysis of the microdata are as follows. First, expected nominal income growth declined significantly during the Great Recession. This is the worst drop ever observed in these data, and this measure has not yet fully recovered to pre-recession levels. Second, the decline exists for all age groups, Mariacristina De Nardi is a senior economist and research advisor; Eric French is a senior economist and research advisor; and David Benson is an associate economist in the Economic Research Department of the Federal Reserve Bank of Chicago. The authors thank Richard Porter and an anonymous referee for helpful comments and Helen Koshy for editorial advice. © 2012 Federal Reserve Bank of Chicago Economic Perspectives is published by the Economic Research Department of the Federal Reserve Bank of Chicago. The views expressed are the authors’ and do not necessarily reflect the views of the Federal Reserve Bank of Chicago or the Federal Reserve System. Charles L. Evans, President ; Daniel G. Sullivan, Executive Vice President and Director of Research; Spencer Krane, Senior Vice President and Economic Advisor; David Marshall, Senior Vice President, financial markets group; Daniel Aaronson, Vice President, microeconomic policy research; Jonas D. M. Fisher, Vice President, macroeconomic policy research; Richard Heckinger,Vice President, markets team; Anna L. Paulson, Vice President, finance team; William A. Testa, Vice President, regional programs, Richard D. Porter, Vice President and Economics Editor; Helen Koshy and Han Y. Choi, Editors; Rita Molloy and Julia Baker, Production Editors; Sheila A. Mangler, Editorial Assistant. Economic Perspectives articles may be reproduced in whole or in part, provided the articles are not reproduced or distributed for commercial gain and provided the source is appropriately credited. Prior written permission must be obtained for any other reproduction, distribution, republication, or creation of derivative works of Economic Perspectives articles. To request permission, please contact Helen Koshy, senior editor, at 312-322-5830 or email Helen.Koshy@chi.frb.org. ISSN 0164-0682 1 education levels, and income quinfigure 1 tiles. Relative to previous recessions, Level of real personal consumption expenditures those with higher levels of income billions of 2005 dollars and education are more pessimistic coming out of this recession than 10,000 their poorer and less-educated counterparts. Third, expectations 8,000 for real income growth have also declined, and the decline in expected 6,000 real income growth is more severe when personal inflation expectations 4,000 are used instead of actual Consumer Price Index (CPI) inflation. Fourth, 2,000 expected income growth is a strong predictor of actual future income 0 1962 ’70 ’78 ’86 ’94 2002 ’10 growth. Since expected income growth is a very important determiNote: PCE is personal consumption expenditures. Source: Haver Analytics. nant of consumption decisions, the observed drop in expected income has the potential to explain at least part of the observed decline in figure 2 consumption. Nominal PCE to nominal GDP ratio during recessions since 1962 In the context of a simple permaPCE – GDP ratio nent-income model, we find that the 0.72 negative wealth effect (coming from decreased stock market valuations 0.70 and housing prices) and consumers’ 0.68 decreased income expectations were 0.66 big factors in determining the ob0.64 served consumption drop. In fact, 0.62 we find that in this model, the ob0.60 served drops in wealth and income 0.58 expectations can explain the observed 0.56 drop in consumption in its entirety, 0.54 depending on what we assume about 1962 ’68 ’74 ’80 ’86 ’92 ’98 2004 ’10 future income growth beyond the Notes: PCE is personal consumption expenditures; GDP is gross domestic time horizon covered by the Reuters/ product. Shaded areas indicate recession periods as defined by the National University of Michigan Surveys of Bureau of Economic Research. Source: Haver Analytics. Consumers data set. Reinhart and Rogoff (2009) have stressed the similarities between the current financial crisis and many earlier ones shows a flattening out of the consumption growth rate stretching across centuries, continents, and economies. in 2008–09. The fact that this pattern is clearly visible, These crises entailed large declines in real housing even over a period of almost 50 years, highlights the prices, equity collapses, and profound declines in outseverity and persistence of the Great Recession and put and employment. They emphasize the importance the very slow recovery that is following it. of balance sheet repair. We complement their research Figure 2 shows that consumption growth outpaced by emphasizing the role played by consumers’ income gross domestic product (GDP) growth through past expectations, as well as wealth effects. recessionary periods. The nominal PCE–GDP ratio has increased in each recession since 1962. In contrast, Macrodata: Total real PCE during the Great Recession, it increased more modestly. Figure 1 displays the level of real PCE from 1962 Since the latest recession, this ratio has either fallen to 2011:Q3. Even over this long horizon, the chart or stagnated. Thus, as a share of GDP, consumption 2 1Q/2012, Economic Perspectives has been hit harder than in previous figure 3 recessions. Normalized real PCE levels over recession periods Petev, Pistaferri, and Eksten (2011) document that while real per peak level = 1 capita consumption declined mono1.20 tonically until the middle of 2009, 1.15 real per capita disposable income 1.10 was relatively stable and its decline 1.05 was significantly smaller. This stability in per capita income is explained 1.00 entirely by a strong increase in gov0.95 ernment transfers to households, as 0.90 wages and financial income fell. The 0.85 increase in government transfers was partly due to higher take-up rates 0.80 −16 −14 −12 −10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16 for unemployment insurance and quarters since peak food stamps and partly due to the increased generosity of means-test1973:Q4 1981:Q3 2001:Q1 ed programs enacted by the federal 1980:Q1 1990:Q3 2007:Q4 government (such as extended unNote: PCE is personal consumption expenditures. employment benefits and increases Sources: Haver Analytics and authors’ calculations. in food stamps and emergency cash assistance). Given that these transfers are means tested, they primarily help figure 4 poorer households. Consistent with Real total quarterly PCE growth over 2008–09 this finding, we find that in the versus previous recessions since 1974 Reuters/University of Michigan Surveys of Consumers, the drop in quarterly growth (annual rate) income expectations for the next 6 12 months among poor households 4 was smaller than that among all other households. 2 Figure 3 compares the time path of real PCE over several recession0 ary time periods, where the level of −2 PCE is normalized to 1 at the business cycle peak (as defined by the −4 National Bureau of Economic Research, NBER) prior to each −6 −16 −14 −12 −10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16 recession. The NBER dates for quarters since peak the recessions’ peaks are 1973:Q4, 1980:Q1, 1981:Q3, 1990:Q3, Average of previous recessions 2001:Q1, and 2007:Q4. 2007:Q4 Figure 3 highlights that in the Note: PCE is personal consumption expenditures. 2008–09 recession, consumption Sources: Haver Analytics and authors’ calculations. dropped 3.4 percent from peak to trough (six quarters after the peak) and was slow to increase afterward. This pattern contrasts with every other recession since Figure 4 displays the time path of the real PCE 1974. During all previous recessionary periods, consumpgrowth rate for the 2008–09 recession around the NBER tion either fell only modestly or increased following peak and compares it with the average real PCE the peak. growth rates from all other recessions since 1971. This graph shows that the average real PCE growth Federal Reserve Bank of Chicago 3 rate around the 2008–09 recession was significantly lower than the corresponding average over the previous five recessions. Consumption has grown 4.1 percent in total over the past five years, or an average rate of 0.8 percent per year. This consumption growth rate contrasts sharply with its average rate since 1971 of 3.1 percent, adding up to about 15 percent growth over an average five-year period. Thus, consumption expenditures are about 15% – 4% = 11% below what they would have been had they grown at their historical averages from 2007:Q4 onward. All subcomponents of PCE fell during the Great Recession. Durables growth was somewhat weaker than in the previous five recessionary periods, both in terms of average growth rate and pattern of recovery. However, nondurables, and especially services, were the most depressed compared with previous recessions. Macrodata: Total real PCE services figure 5 Normalized real PCE services over several recessions peak level = 1 1.20 1.15 1.10 1.05 1.00 0.95 0.90 0.85 0.80 −16 −14 −12 −10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16 quarters since peak 1973:Q4 1981:Q3 2001:Q1 1980:Q1 1990:Q3 2007:Q4 Notes: PCE is personal consumption expenditures. For each recession, the level of PCE services is normalized to 1 at the business cycle peak (as defined by the National Bureau of Economic Research) prior to the recession. Sources: Haver Analytics and authors’ calculations. figure 6 Normalized real nondurables PCE over several recessions peak level = 1 1.20 Figure 5 highlights that the be1.15 havior of PCE services was starkly 1.10 different over the 2008–09 reces1.05 sion from all other recessions since 1.00 1974. In all other recessions, PCE services grew both before and after 0.95 the peak, while during the latest 0.90 recession, it stagnated starting two 0.85 quarters after the peak (four quar0.80 ters before the trough) and kept −16 −14 −12 −10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16 stagnating for four additional quarquarters since peak ters afterward. PCE services took 1973:Q4 1981:Q3 2001:Q1 until 2010:Q4 to return to peak levels. 1980:Q1 1990:Q3 2007:Q4 Regarding the main services Notes: PCE is personal consumption expenditures. For each recession, the level of nondurables PCE is normalized to 1 at the business cycle peak (as defined by the subcomponents, Petev, Pistaferri, National Bureau of Economic Research) prior to the recession. and Eksten (2011) document that Sources: Haver Analytics and authors’ calculations. spending on health services increased, held stable for housing and utilities, but declined substantially for services related Macrodata: Total real nondurables PCE to transportation, food, and recreation. In sum, the most We can see from figure 6 that the rise in PCE adjustable services dropped, while those components nondurables was similar to that experienced in most that the consumer has little discretion to adjust did not. other recessions before the peak, but its recovery path 4 1Q/2012, Economic Perspectives in the latest recession was among the worst. Petev, Pistaferri, and Eksten (2011) document an unusual decline in spending on food, an important indicator of consumer well-being, which raises concerns about the extent and depth of the strains on households during the latest recession. An interesting new paper by Aguiar, Hurst, and Karabarbounis (2011), however, documents that during the most recent recession, a significant fraction of foregone market work hours went to home production (based on diary information)—35 percent, including childcare. This is an important channel that could produce more goods (such as food) and services (such as childcare) at a lower cost. More research is needed to determine if home production could completely explain the observed decline in food spending. figure 7 Normalized real durables PCE over several recessions peak level = 1 1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8 0.7 0.6 −16 −14 −12 −10 −8 −6 −4 −2 0 2 4 quarters since peak 8 10 12 14 16 1973:Q4 1981:Q3 2001:Q1 1980:Q1 1990:Q3 2007:Q4 Notes: PCE is personal consumption expenditures. For each recession, the level of durables PCE is normalized to 1 at the business cycle peak (as defined by the National Bureau of Economic Research) prior to the recession. Sources: Haver Analytics and authors’ calculations. Macrodata: Total real PCE durables Figure 7 displays a large drop for durables over the most recent recession. Five to six quarters after the peak, this recession actually displayed the largest drop in durables, compared with the previous five recessions. In addition, the pace of recovery in durables was slow—it took 12 quarters for durables to regain the previous peak level. Petev, Pistaferri, and Eksten (2011) document that the bulk of the decline in real per capita spending is attributable to purchases of cars (a 25 percent decline by the end of 2008) and partly of furniture (a 9 percent decline). To summarize, our main findings from the macrodata are as follows. First, the Great Recession marked the most severe and persistent decline in aggregate consumption since World War II. All subcomponents of consumption declined during this period. However, we find that the significant drop in consumed services stands out most, compared with previous recessions. Second, while the decline was historic, the time path of consumption and its subcomponents leading up the recession was not substantially different from past recessionary periods. Third, the recovery path of consumption following the Great Recession has been uncharacteristically weak. It took nearly three years for total consumption to return to its level just prior Federal Reserve Bank of Chicago 6 to the recession. In contrast, the second-worst rebound observed in the data followed the 1974 recession and was just over one year. We find that this persistence is reflected most in the subcomponents of nondurables and especially in services consumption. Microdata: Expected income This section uses consumer expectations for future income from the Reuters/University of Michigan Surveys of Consumers, both in nominal and real terms, to see whether shocks to expected future income are contributing to the consumption dip that we have experienced. The survey asks two questions to identify the magnitude and sign of income changes. 1. “During the next 12 months, do you expect your income to be higher or lower than during the past year?” 2. “By about what percent do you expect your income to (increase/decrease) during the next 12 months?” The resulting index of expected income growth ranges widely across individuals, but on average, the estimates tend to accord with what we might have anticipated ex ante. The historical mean is +5.5 percent, split between +4.8 percent during recessions and +5.6 percent during expansions. While the realized 5 measure is much more variable, figure 8 shows that expected nominal disposable income tracks realized income quite well. The survey also asks about expected changes in the price level over the next 12 months. Historically, this survey estimate has been very similar to realized CPI inflation. We construct expected real income growth by subtracting each individual’s inflation expectations from his expected nominal income growth. We construct time series from the microdata. For each month of the survey, we take cross-sectional means within each demographic group and then aggregate to quarterly frequency to minimize noise. The data begin in 1978 and go through the first half of 2011, though some series only go back to 1990. Thus, we typically have five recessionary periods to examine. Microdata: Nominal income growth expectations figure 8 Realized and expected nominal disposable income income growth 14 12 10 8 6 4 2 0 −2 −4 −6 1978 ’82 ’86 ’90 ’94 ’98 2002 ’06 ’10 Nominal disposal income growth Expected nominal income growth Sources: Haver Analytics, Reuters/University of Michigan Surveys of Consumers, and authors’ calculations. figure 9 Average expected nominal income growth rates around recessionary periods expected nominal income growth 12 Except for the Great Recession and the 1980 recession, income ex10 pectations show a downward trend 8 for up to four quarters around the NBER peak, but then stabilize and 6 actually rise by the end of our four4 year window (see figure 9). For both the 1980 and most recent re2 cession, we observe larger and more 0 prolonged dips before and after the −2 NBER business cycle peak. Besides −16 −14 −12 −10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16 the abnormal drop, both in terms quarters since peak of size and duration, the recovery periods also stand out for their Great Recession 2007:Q4 1990:Q3 1980:Q1 length and sluggishness. Even well 2001:Q1 1981:Q3 after ten quarters from the peak, exSources: Reuters/University of Michigan Surveys of Consumers and authors’ calculations. pected nominal income growth was still well below its pre-recessionary level. It should be noted that the most recent recession is the only one during which nominal Figure 10 shows that since the late 1970s, nomiincome expectations reached negative growth rates. nal income growth expectations have not varied deIn all of the previous recessions that we study, even mographically until the most recent recession. The when nominal income growth rates went down, they prime-aged individuals (30–59) experienced the largest stayed well above 4 percent. Of course, inflation has drop in expected nominal income growth during the been lower during the most recent recession. We disGreat Recession and have now only partly recovered, cuss real income patterns in the next section. 6 1Q/2012, Economic Perspectives figure 10 Expected nominal income growth by age group expected nominal income growth 20 expected nominal income growth 20 15 15 10 10 5 5 0 0 −5 −16 −12 −8 −4 0 4 8 12 16 −5 −16 −12 −8 quarters since peak 1981:Q3 −4 0 4 8 12 16 12 16 quarters since peak 1990:Q3 expected nominal income growth 20 expected nominal income growth 20 15 15 10 10 5 5 0 0 −5 −16 −12 −8 −4 0 4 8 12 16 −12 −8 −4 0 4 8 quarters since peak 2007:Q4−Great Recession quarters since peak 2001:Q1 Age 18−29 −5 −16 Age 30−59 Age 60−69 Age 70+ Sources: Reuters/University of Michigan Surveys of Consumers and authors’ calculations. ten quarters after the peak. For younger consumers, expectations dropped well before the peak—five quarters ahead—but then stabilized after the peak. In past recessionary periods, nominal income expectations of the elderly population had hovered around or just above zero. However, these expectations have been markedly negative since the NBER peak in 2007:Q4. Focusing on this population, Christelis, Georgarakos, and Jappelli (2011) use the 2009 Internet Survey of Health and Retirement Study (HRS) to look at the effects of three different shocks—the drop in house prices, the decline in the stock market, and the increase in unemployment—on households’ expenditures during the Great Recession. This data set refers to the population aged 50 years and older. The HRS Internet Survey contains detailed measures of both housing wealth losses (between summer 2006 and summer 2009) and losses in various financial assets (between October 2008 and mid-2009). It also contains measures of consumption growth and qualitative indicators of consumption changes, allowing the researchers to estimate the effect of the losses on adjustments in consumption expenditure. Their main finding is that losses on housing and financial wealth, together with the income loss from becoming unemployed, led households to reduce their Federal Reserve Bank of Chicago spending. The estimated elasticity of consumption to financial wealth implies a marginal propensity to consume with respect to financial wealth equal to 3 percentage points. The decline in house prices also had an important impact on consumption: The estimated elasticity implies that the marginal propensity to consume out of housing wealth is 1 percentage point. Put differently, these estimates suggest that every dollar of financial wealth lost reduces consumption three cents per year and every dollar of housing wealth lost reduces consumption one cent per year. Additionally, households in which at least one of the two adult members (or the single head) became unemployed in 2008 and early 2009 reduced consumption by 10 percent in 2009. See Hurd and Rohwedder (2010a, 2010b) and the citations therein for more estimates on the responsiveness of consumption to asset and income shocks. Figure 11 shows that all income levels adjusted their expected income growth downward during the most recent recession. In past recessions, these adjustments were smaller. In the most recent recession, the first quintile (the poorest) dropped their income growth expectations the least. By the end of 2010, all income levels had roughly converged to the same post-peak level and their expectations are now much closer together. This result is consistent with Petev, Pistaferri, 7 figure 11 Expected nominal income growth by income quintile expected nominal income growth expected nominal income growth 15 15 10 10 5 5 0 0 −5 −16 −12 −8 −4 0 4 8 12 16 −5 −16 −12 quarters since peak 1981:Q3 −8 −4 0 expected nominal income growth expected nominal income growth 15 15 10 10 5 5 0 0 −5 −16 4 8 12 16 quarters since peak 1990:Q3 −5 −12 −8 −4 0 4 8 12 quarters since peak 2001:Q1 1st quintile 16 2nd quintile −12 −8 −4 0 4 8 12 16 quarters since peak 2007:Q4−Great Recession 3rd quintile 4th quintile 5th quintile Sources: Reuters/University of Michigan Surveys of Consumers and authors’ calculations. and Eksten’s findings. First, they find that increased government transfers propped up income among the poorest households during the Great Recession. Second, using the Michigan Index of Consumer Sentiment (constructed using a subset of questions from the Reuters/University of Michigan Surveys of Consumers), they document that high-income individuals became more pessimistic than other groups during the Great Recession.2 Finally, using the Bureau of Labor Statistics’ Consumer Expenditure Survey (CEX), they find that respondents in the top decile of the wealth distribution are the ones who decreased spending during the Great Recession (–5.4 percent). This finding holds for the subcategories of nondurables and services. This drop in consumption might be due to the large negative wealth effect experienced by these households due to declining house prices and stock market valuations. Figure 12 shows that in previous recessions, income expectations across education groups were rather flat over the cycle. In the most recent recession, everyone reduced their expected income growth. Microdata: Real income growth expectations Nominal income growth during the Great Recession was low, but realized inflation was also low. To study the behavior of real income expectations, we measure 8 inflation in two ways. First, we use actual CPI inflation over the 12-month period covered by the survey question, which assumes that consumers have perfect foresight over the next year concerning inflation. Second, we use the answer to the survey question about the individual’s expectation about growth in prices over the next 12 months. Using these two measures, we construct individual-level expected real income growth and then aggregate up to population-quarter means. The two inflation series have diverged in the past, but after the late 1970s the differences are minor. At the start of the Great Recession, however, a large gap opened up, making for the largest discrepancy we have observed between these two data series. The swing in 2008:Q2 is +6 percent in expected inflation, compared with –1 percent actual CPI inflation. The two measures have since become much closer (see figure 13). The gap in these two measures, of course, affects measured real income growth expectations as we document next. In figure 14, there is no clear cyclical pattern prior to the Great Recession in real income expectations. Before the most recent recession, real income growth was rather flat; it dropped into negative territory several quarters before the peak; and it then went up to about 4 percent four quarters after the peak. From then on, however, it had a large drop, reaching –3 percent five 1Q/2012, Economic Perspectives figure 12 Expected nominal income growth by educational level expected nominal income growth expected nominal income growth 15 15 10 10 5 5 0 0 −5 −15 −12 −8 −4 0 4 8 12 quarters since peak 1981:Q3 −5 −15 16 −12 −8 −4 0 expected nominal income growth expected nominal income growth 15 15 10 10 5 5 0 0 −5 −15 −12 −8 −4 0 4 4 8 12 16 12 16 quarters since peak 1990:Q3 8 12 −5 −15 16 −12 −8 −4 0 4 8 quarters since peak 2007:Q4−Great Recession quarters since peak 2001:Q1 High school dropouts High school graduates Some college College + Sources: Reuters/University of Michigan Surveys of Consumers and authors’ calculations. figure 13 Time series of 12 months forward inflation since 1978 (CPI versus personal inflation expectations for the Reuters/University of Michigan Surveys of Consumers) year-over-year inflation 16 14 12 10 8 6 4 2 0 −2 −4 −6 1978 ’80 ’82 ’84 ’86 ’88 ’90 ’92 ’94 ’96 ’98 2000 ’02 ’04 ’06 ’08 ’10 CPI inflation Consumer expected inflation Note: CPI is Consumer Price Index. Sources: Haver Analytics, Reuters/University of Michigan Surveys of Consumers, and authors’ calculations. Federal Reserve Bank of Chicago 9 quarters after the peak. In summary, figure 14 real income growth expectations Expected real income growth, deflated by CPI inflation deflated by CPI showed a deterioraexpected nominal income growth tion and lower average growth dur6 ing the latest recession than during previous recessions. 4 Figure 15 shows that perceived real income growth based on con2 sumers’ inflation expectactions paints a much more pessimistic picture of 0 consumers’ purchasing power during −2 the Great Recession. Consumers’ perceived real income growth dipped −4 in and out of negative territory well −6 before the recession started, and −16 −14 −12 −10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16 sustained a large drop starting four quarters since peak quarters before the peak. That drop brought expectations from almost Great Recession 2007:Q4 1990:Q3 1980:Q1 +2 percent to a –4 percent growth 2001: Q1 1981:Q3 rate three quarters after the peak. Note: CPI is Consumer Price Index. It took two more quarters for exSources: Haver Analytics, Reuters/University of Michigan Surveys of Consumers, pectations to go back up to a and authors’ calculations. –2 percent growth rate, and they have remained stagnant ever since. The recession window in figure 15 figure 15 ends in 2011:Q4, with expected real Expected real income growth, income growth of –2.5 percent. In using consumers’ inflation expectations 2011, the series has recorded values of –3.1 percent, –3.7 percent, and expected nominal income growth –2.9 percent for the first three quar6 ters of the year, respectively. 4 Our main findings from the analysis of the microdata are as fol2 lows. First, expected nominal income growth declined significantly dur0 ing the Great Recession. It is the −2 worst drop ever observed in these data, and this measure has still not −4 recovered to pre-recession levels. Second, the decline exists for all −6 age groups, education levels, and −16 −14 −12 −10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16 income quintiles. Relative to previquarters since peak ous recessions, those with higher Great Recession 2007:Q4 1990:Q3 1980:Q1 levels of income and education 2001: Q1 1981:Q3 have been more pessimistic this Sources: Reuters/University of Michigan Surveys of Consumers and authors’ calculations. time than their poorer and lesseducated counterparts. Third, expectations for real income growth Do the Michigan microdata have also declined, and the decline in expected real have predictive power? income growth is more severe when we look at personal inflation expectations instead of actual CPI Below we show that the Reuters/University of inflation. Michigan Surveys of Consumers have remarkable forecasting power for both future disposable income 10 1Q/2012, Economic Perspectives Table 1 Regression results Dependent variable Lagged Lagged Forecasted income Michigan consumption annual growth income growth growth, variable expectations variable Q3/Q3 R-squared Annual income growth 1 year forward –0.35 (0.10) 0.80 — 0.61* (0.17) 0.29 Annual income growth 2 years forward 0.06 (0.08) 0.36 — 1.24** (0.17) 0.08 Annual income growth 3 years forward –0.34 (0.13) 0.42 — 2.16*** (0.20) 0.08 Annual consumption growth — 1 year forward 0.71 (0.23) 0.08 0.05* (0.13) 0.37 Annual consumption growth — 2 years forward 0.77 (0.23) –0.25 0.13** (0.16) 0.18 Annual consumption growth — 3 years forward 0.58 (0.27) –0.49 1.15*** (0.19) 0.11 Annual consumption growth 1 year forward –0.20 (0.14) 0.75 (0.21) 0.18 0.39* (0.14) 0.39 Annual consumption growth 2 years forward 0.10 (0.14) 0.76 (0.23) –0.31 –0.07** (0.19) 0.17 Annual consumption growth 3 years forward –0.09 (0.16) 0.59 (0.27) –0.44 (0.21) 0.11 1.36*** Notes: Regressions are run with data from 1978:Q1 to 2011:Q2. Newey-West standard errors in parentheses. Average annual income and consumption growth are 2.78 and 2.91, respectively. Using data up to 2011:Q3, forecast of growth between: *2011:Q3 and 2012:Q3; **2012:Q3 and 2013:Q3; ***2013:Q3 and 2014:Q3. Sources: Authors’ calculations based on data from Haver Analytics and Reuters/University of Michigan Surveys of Consumers. and consumption growth.3 We estimate the regression for disposable income (Yt ) in period t first: ((Yt+k+4 – Yt+k )/Yt+k ) = α0 + α1 ((Yt – Yt–4)/Yt–4) + α2 gMt + εt+k , where α0, α1, α2 are parameters to be estimated, and α1 and α2 are reported in table 1. The variable ((Yt+k+4 – Yt+k )/Yt+k ) is next year’s annual income growth k quarters from now, so k is 0 when forecasting income growth over the next year and 4 when forecasting income growth over the subsequent year. ((Yt – Yt–4 )/Yt–4 ) is income growth over the past year, and gMt is expected real income growth from the Michigan surveys, where we deflate using expected inflation from the survey. As can be seen in table 1, lagged income growth has a negative coefficient, and expected income growth has a positive coefficient. The coefficient on expected income growth in the next year is 0.8, indicating that a 1 percent decline in expected income growth reduces next year’s income growth 0.8 percent, taking into account the previous year’s income growth. The righthand column shows that predicted income growth over Federal Reserve Bank of Chicago the next year (2011:Q3 to 2012:Q3), using lagged income growth and expected income growth, is 0.6 percent, well below its average of 2.8 percent over the 1978–2011 sample period. Income growth between 2012:Q3 and 2013:Q3 is also forecasted to be low. Expected income growth also turns out to be a good predictor of consumption growth. Table 1 presents regressions using future consumption growth as the left-hand-side variable and lagged consumption growth and the Michigan expectations variable as the righthand-side variables. Using these estimates, the consumption forecast for 2011:Q3 to 2012:Q3 calls for a meager growth rate of 0.1 percent. In short, the low expected income growth in the expectations data of the Reuters/University of Michigan Surveys of Consumers suggests that the U.S. will experience low growth in both income and consumption over the next two years. Obviously, there are many things not included in this specification, so the estimates should only be taken as suggestive. However, the results are fairly robust to changes in model specification and to the addition of a few other variables, such as the unemployment rate. 11 Quantifying the effects of the drops in wealth and income expectations Data from the Federal Reserve Board of Governors’ flow of funds accounts show that in 2008, American households experienced a loss of $13.6 trillion in wealth, with most of the loss concentrated in stock market wealth. While stock market wealth has partially recovered since then, housing wealth has continued to decline. The resulting wealth loss, combined with lower expected income growth, has the potential to explain the extent to which consumers cut back consumption during the Great Recession. Now, we quantify the effects of these declines by first calibrating a simple model of consumption that matches the observed level of consumption in 2007:Q4 and that implies empirically plausible marginal propensities to consume (MPCs) out of both assets and permanent income. Then, we show the model’s predicted consumption in 2011:Q2 under different expectations for income and asset values. We find that for reasonable parameter values, the decline in asset values can explain one-third of the gap between actual and potential consumption, while declines in permanent income expectations can easily explain the rest. That is, the weak growth in consumption that we have experienced in the past few years can be explained by the combination of realized wealth losses on equity shares and housing and a more subdued outlook for future income growth. where ∞ 4) Yt = ∑ (1 / (1 + r )) τ −t Yτ τ =t is the present value of discounted future labor income. We compute Yt by assuming that consumers observe income up to 2011:Q2 and that from that point on, income expectations for the next year are those measured in the most recent Reuters/University of Michigan Surveys of Consumers, but they revert to long-run income growth afterward. Mathematically, we can write this as Yt+k = (1 + gM )kYt, k ≤ 4 Yt+k = (1 + g)Yt+k−1, k > 4, where Yt is disposable income, gM is the perceived real income growth for the next year in the 2010:Q4 Reuters/University of Michigan Surveys of Consumers (the most recent release of this variable suggests even more pessimism on consumers’ part than in 2010:Q4), while g is the average growth rate of income over the past 40 years. Putting these equations together yields ∞ 5) Yt = ∑ τ =t (1 / (1 + r )) τ −t Yτ = Yt (1 + (1 + g M ) / (1 + r ) + ((1 + g M ) / (1 + r )) 2 + ((1 + g M ) / (1 + r ))3 + ((1 + g M ) / (1 + r )) 4 Model × [1 + (1 + g ) / (1 + r ) + ((1 + g ) / (1 + r )) 2 + ...] We define Ct as consumption expenditures at time t (where time is measured in quarters). Households maximize 1) ∞ ∑ β ln(C ), t = t0 t t subject to the following asset accumulation equation, 2) At +1 = (1 + r ) At + Yt − Ct , lim βt t →∞ A C t =0 t given At 0 and given income expectations, and r denotes the interest rate earned on assets (At ). To avoid the additional complication of dealing with uncertainty, we make the simplifying assumption that individuals are certain of future income. However, we allow them to revise their perceived income process if they make a mistake. The solution to the consumer’s optimization problem is: 3) 12 Ct = (1 − β)(Yt + At ) , = Yt (1 + (1 + g M ) / (1 + r ) + ((1 + g M ) / (1 + r )) 2 + ((1 + g M ) / (1 + r ))3 + ((1 + g M ) / (1 + r )) 4 (1+ r ) (r − g ) ). We call the income process above income process 1. Then, to show the importance of low expected income growth, we consider a more pessimistic scenario, which we call income process 2, in which rather than reverting back to a long-run expected growth after four quarters, pessimism about income growth persists forever. In this case, 6) ∞ Yt = ∑ τ =t (1 / (1 + r )) τ −t Yτ = Yt ( ( r(1−+grM) ) ). Figure 16 reports four different lines for the time path of real disposable income since the beginning of 2007. The black line shows a counterfactual disposable income level—the level that would have existed had 1Q/2012, Economic Perspectives figure 16 Disposable income and assumed income processes billions of 2005 dollars 12,000 11,500 11,000 10,500 10,000 9,500 2006 ’07 ’08 ’09 ’10 ’11 Disposable income Counterfactual disposable income Income process 1 Income process 2 ’12 Sources: Haver Analytics and authors’ calculations. it continued to grow at its historical average rate of 3.2 percent from 2007:Q4 onward. The blue line shows realized disposable income up to 2011:Q2. The grey dotted line begins with realized disposable income in 2011:Q2. It then tacks on the expected level of disposable income using expectations data from the Reuters/ University of Michigan Surveys of Consumers for all periods thereafter. This corresponds to income process 2. The blue dotted line begins in 2012:Q2, assuming that income grows according to the Reuters/University of Michigan Surveys of Consumers between 2011:Q2 and 2012:Q4 and then at its historical rate afterwards. It corresponds to income process 1. Calibration The three key moments we wish to match are the marginal propensity to consume (MPC) out of assets, the MPC out of permanent income, and the level of consumption in 2007:Q4. Most estimates of the MPC out of assets are in the range 0.01–0.05, and most estimates of the MPC out of permanent income are between 0.5 and 1. We assume the MPC out of assets is 0.03 per year. We use per capita income growth for the individual’s decision problem. Thus, we set g = .032 – .014 = .018 (average disposable income growth over the 1967:Q4 to 2007:Q4 period less population growth of those Federal Reserve Bank of Chicago aged 16 and older over the same period). We then pick r and β to match the MPC out of assets and the level of consumption in 2007:Q4. Thus, we match ∂Ct = (1 − β) = .03 ∂At C2007:Q 4 = (1 − β)[Y2007:Q 4 1+ r r−g + A2007:Q 4 ], where C2007:Q4 = $9,312.6 billion (at an annualized rate), Y2007:Q4 = $9,944 billion (annualized), and A2007:Q4 = $69,139 billion. The unit of time in this analysis is a quarter. So, we convert annual growth rates to quarterly ones, using the formula (1 + g)(1/4) – 1 when taking the quarterly growth rate for g. For dollar amounts, we divide by 4. After converting everything to quarterly rates, we use the above two equations to solve for β and r. Table 2 presents all variables at quarterly and annualized rates. At annualized rates, β = 0.97 and r = 0.060.This gives a quarterly MPC out of permanent income equal to ∂Ct = (1 − β)[(1 + r ) / (r − g )] = .730, ∂Yt 13 Table 2 Table 3 Model parameters Results Annual Quarterly (dollars in billions) Realized consumption level 2011:Q2 Predicted consumption level 2011:Q2, given information in 2007:Q4 Consumption loss 9,379 Income process 1 Income process 1 and lower short-term interest rate Income process 2 1,206 10,472 Exogenously set 1,093 gM – 0.016 – 0.0040 Consumption loss due to asset value decline Population growth 0.014 0.0035 Asset value decline 9,746 g 0.018 0.0045 Predicted consumption decline due to MPC out of assets 0.030 0.0074 asset price decline 289 9,944 2,486 Y2007:Q4 Consumption loss, given disposable 9,313 2,328 C 2007:Q4 income decline 69,139 69,139 A 2007:Q4 Income process 1 917 Income process 1 and lower short-term Endogenously determined interest rate 710 β 0.970 0.993 Income process 2 4,038 r 0.060 0.015 Consumption loss given both asset Implied MPC out of income 0.730 and income declines Note: MPC is marginal propensity to consume. Sources: Authors’ calculations based on data from Haver Analytics and the Reuters/University of Michigan Surveys of Consumers. which is about in the middle of the normal range estimates in the literature for the MPC. Over the past 40 years, annual population growth for those aged 16 and older is 1.4 percent, which we define as p. We assume this rate of population growth continues in the future. Income growth in the individual’s decision problem is in per capita terms. We then account for aggregate growth at the end by adjusting up disposable income by 1.4 percent at an annual rate. Results Table 3 explains our key findings. All quarterly numbers in this section are annualized; that is, they are the quarterly flows multiplied by 4. Consumption expenditures in 2011:Q2 were $9,379 billion. Had they grown at average rates from 2007:Q4 onward, they would have been at $10,472 billion in 2011:Q2, which is 10 percent higher than they are today. This difference of $1,093 billion, line 3 of the table, is the shortfall we seek to explain with the model. Figure 17 depicts this shortfall graphically. Lines 4 and 5 in table 3 trace out the effects of the decline in asset prices. Net worth fell $9,746 billion in real terms over this period. Given a quarterly MPC of 0.0074 out of assets, we predict a ($9,746 billion) × (0.0074) × 4 = $289 billion fall in consumption, at an annualized rate. The following lines in the table predict the consumption fall due to various permanent income scenarios. To perform this computation, we first put ourselves in 14 999 4,328 Note: All amounts in billions of dollars. Sources: Authors’ calculations and data from Haver Analytics. 2007:Q4 and predict Y as of 2011:Q2, had income grown steadily at its long-run historical average. Second, we calculate Y , given realized income in 2007:Q4 and the two income processes that we described previously. To be clear, taking into account population growth rates, we calculate Y2011:Q 2 , given the information set from 2007:Q4, as Y2011:Q2 = Y2007:Q4 r1+− rg ((1+ p ) (1+ g ))14, where the term in the exponent (14) is the number of quarters between 2007:Q4 and 2011:Q2. Once we calculate the loss in Y under different income and interest rate scenarios, we use the model to calculate the resulting consumption loss. The consumption loss associated with income process 1 is $0.917 trillion, which is reasonably close to the observed consumption loss. This computation is sensitive to the time path of the interest rate as well. The baseline calibration yields a yearly interest rate of 6 percent. In the lower short-term interest rate scenario, we assume that over the first year the yearly interest rate is 3 percent and then reverts back to 6 percent. In this case, income is less heavily discounted; hence its present value is higher and the implied consumption drop is smaller, $710 billion rather than $917 billion. Unsurprisingly, the very pessimistic income expectation scenario considered in income process 2 generates a huge consumption loss of $4.038 trillion, which is almost four times larger than the consumption shortfall we wish to explain. 1Q/2012, Economic Perspectives figure 17 Real PCE with and without the Great Recession billions of 2005 dollars 11,300 11,000 10,700 10,400 10,100 9,800 9,500 9,200 8,900 2006 ’07 ’08 ’09 PCE ’10 ’11 ’12 Counterfactual PCE Note: PCE is personal consumption expenditures. Sources: Haver Analytics and authors’ calculations. Because our model predicts that consumption is linear in resources (assets and the present value of future income), we can add up the losses from assets and income. Note that the predicted consumption decline given the asset fall plus the predicted decline given income process 1 of $1.206 trillion lines up almost exactly with what actually occurred. Conclusion This article documents key facts about aggregate consumption and its subcomponents and looks at the behavior of important determinants of consumption over the cycle, such as consumers’ expectations about their future income and changes in consumers’ wealth positions due to changes in house prices and stock valuations. We performed a simple computation to determine whether the observed drop in consumption can be explained by the observed drops in wealth and income expectations. Federal Reserve Bank of Chicago In the context of a simple permanent income model, we find that the negative wealth effect (coming from decreased stock market valuations and house prices) and decreased consumer income expectations were crucial factors in determining the observed consumption drop. In fact, we find that in this model, the observed drops in wealth and income expectations can explain the observed drop in consumption in its entirety, depending on what is assumed about future income growth beyond the time horizon covered by the Reuters/University of Michigan Surveys of Consumers data set. 15 NOTES 1 This survey also collects the data that form the well-known Michigan Consumer Confidence Index. The survey is published monthly by the University of Michigan and Thomson Reuters. consequence. The median decline in wealth was 15% in Shapiro’s data, and those who lost at least 10% of their net worth had almost twice the mean wealth and 3.5 times the median wealth of the sample. As a possible explanation for the pessimism of the wealthy, Shapiro (2010) finds that these households were exposed more to the stock market and experienced larger declines in wealth as a 3 2 See Souleles (2004), Ludvigson (2004), and Barsky and Sims (2009) for more on the predictive power of the Michigan surveys. references Aguiar, Mark A., Erik Hurst, and Loukas Karabarbounis, 2011, “Time use during recessions,” National Bureau of Economic Research, working paper, No. 17259, July. Barsky, Robert B., and Eric R. Sims, 2009 “Information, animal spirits, and the meaning of innovations in consumer confidence,” National Bureau of Economic Research, working paper, No. 15049, June. Christelis, Dimitris, Dimitris Georgarakos, and Tullio Jappelli, 2011 “Wealth shocks, unemployment shocks and consumption in the wake of the Great Recession,” University of Naples, Italy, Centre for Studies in Economics and Finance, working paper, No. 279, October. Hurd, Michael D., and Susann Rohwedder, 2010a, “Effects of the financial crisis and Great Recession on American households,” National Bureau of Economic Research, working paper, No. 16407, September. __________, 2010b, “The effects of the economic crisis on the older population,” University of Michigan, Michigan Retirement Research Center, working paper, No. WP 2010-231, March. 16 Ludvigson, Sydney C., 2004, “Consumer confidence and consumer spending,” Journal of Economic Perspectives, Vol. 18, No. 2, Spring, pp. 29–50. Petev, Ivaylo, Luigi Pistaferri, and Itay Saporta Eksten, 2011, “Consumption and the Great Recession: An analysis of trends, perceptions, and distributional effects,” Stanford University, mimeo, August. Reinhart, Carmen M., and Kenneth S. Rogoff, 2009, This Time Is Different: Eight Centuries of Financial Folly, Princeton, NJ: Princeton University Press. Shapiro, Matthew D., 2010, “The effects of the financial crisis on the well-being of older Americans: Evidence from the cognitive economic study,” University of Michigan, Michigan Retirement Research Center, working paper, No. WP 2010-228, September. Souleles, Nicholas S., 2004, “Expectations, heterogeneous forecast errors, and consumption: Micro evidence from the Michigan Consumer Sentiment Surveys,” Journal of Money, Credit and Banking, Vol. 36, No. 1, February. 1Q/2012, Economic Perspectives Medicaid and the elderly Mariacristina De Nardi, Eric French, John Bailey Jones, and Angshuman Gooptu Introduction and summary Expenditures on medical care by Medicaid and Medicare, America’s two main public health insurance programs, are large and growing rapidly. Although Medicare is the main provider of medical care for the elderly and disabled, it does not cover all medical costs. In particular, it covers only a limited amount of long-term care expenses (for example, nursing home expenses). The principal public provider of long-term care is Medicaid, a means-tested program for the impoverished. Medicaid now assists 70 percent of nursing home residents1 and helps the elderly poor pay for other medical services as well. In 2009, Medicaid spent over $75 billion on 5.3 million elderly beneficiaries.2 An important feature of Medicaid is that it provides insurance against catastrophic medical expenses by providing a minimum floor of consumption for households. Although Medicaid is available only to “poor” households, middle-income households with high medical expenses usually qualify for assistance also. Given the ongoing growth in medical expenditures, Medicaid coverage in old age is thus becoming as much of a program for the middle class as for the poor (Brown and Finkelstein, 2008). Another important feature of Medicaid is that it is asset and income tested; in contrast, almost all seniors qualify for Medicare. This implies that Medicaid affects households’ saving decisions, not only by reducing the level and risk of their medical expenses, but also by encouraging them to consume their wealth and income more quickly in order to qualify for aid (Hubbard, Skinner, and Zeldes, 1995). Although Medicaid covers poor people of all ages, this article focuses on Medicaid’s coverage for the elderly. Many recent proposals for reforming Medicaid could have significant effects on the financial burdens of the elderly, on the medical expense risk that they face, and on their saving decisions. Moreover, Medicaid is Federal Reserve Bank of Chicago a large and growing component of the federal budget. The share of total federal, state, and local government expenditures absorbed by Medicaid rose from less than 2 percent in 1970 to almost 7 percent in 2009,3 and it is expected to increase even more in the future. Controlling the cost of Medicaid is an important component in correcting the federal government’s longterm fiscal imbalance. Mariacristina De Nardi is a senior economist and research advisor; Eric French is a senior economist and research advisor; and Angshuman Gooptu is an associate economist in the Economic Research Department of the Federal Reserve Bank of Chicago. John Bailey Jones is an associate professor of economics at the University at Albany, State University of New York, and a consultant to the Federal Reserve Bank of Chicago. The authors thank Daisy Chen, John Klemm, and representatives of Medicaid offices in Florida, Alabama, Indiana, Wisconsin, and Ohio, who helped verify the facts in this paper, and a referee and Richard Porter for comments. © 2012 Federal Reserve Bank of Chicago Economic Perspectives is published by the Economic Research Department of the Federal Reserve Bank of Chicago. The views expressed are the authors’ and do not necessarily reflect the views of the Federal Reserve Bank of Chicago or the Federal Reserve System. Charles L. Evans, President ; Daniel G. Sullivan, Executive Vice President and Director of Research; Spencer Krane, Senior Vice President and Economic Advisor; David Marshall, Senior Vice President, financial markets group; Daniel Aaronson, Vice President, microeconomic policy research; Jonas D. M. Fisher, Vice President, macroeconomic policy research; Richard Heckinger,Vice President, markets team; Anna L. Paulson, Vice President, finance team; William A. Testa, Vice President, regional programs, Richard D. Porter, Vice President and Economics Editor; Helen Koshy and Han Y. Choi, Editors; Rita Molloy and Julia Baker, Production Editors; Sheila A. Mangler, Editorial Assistant. Economic Perspectives articles may be reproduced in whole or in part, provided the articles are not reproduced or distributed for commercial gain and provided the source is appropriately credited. Prior written permission must be obtained for any other reproduction, distribution, republication, or creation of derivative works of Economic Perspectives articles. To request permission, please contact Helen Koshy, senior editor, at 312-322-5830 or email Helen.Koshy@chi.frb.org. ISSN 0164-0682 17 In this article, we describe the Medicaid rules for the elderly and discuss their economic implications. We focus on the rules for single (that is, never married, divorced, or widowed) individuals to avoid the additional complications involved in considering couples. The main difference between singles and couples is that the income and asset limits for Medicaid eligibility are higher for couples. Medicaid is administered jointly by the federal and state governments, but each state has significant flexibility on the details of implementation; hence, there is large variation across states in income and asset eligibility and in coverage. This variation may well provide elderly people in different states with different saving incentives, and it might even encourage them to move from one state to another. We focus on finding the features common to all states, and identifying the most salient state-level differences. 65 and older, SSI recipients account for 40 percent of all beneficiaries and 27 percent of all Medicaid expenditures. “Dual eligibles” represent 29 percent of all beneficiaries and 9 percent of all Medicaid expenditures and are the second-largest group of Medicaid beneficiaries. “Medically needy” individuals represent 10 percent of all beneficiaries and 23 percent of all expenditures. “Others,” a category largely made up of those with catastrophic medical expenses who are not technically “medically needy,” represent 29 percent of all beneficiaries and 41 percent of all expenses. Although the Center for Medicare and Medicaid Services technically refers to “others” as categorically needy, a large share of this group are what we will refer to as medically needy, because their circumstances (catastrophic medical expenses) are more like those of the strictly medically needy than those of the other categorically needy groups. Overview of the Medicaid program The categorically needy: SSI beneficiaries Medicaid and Medicare were created by the Social Security Act Amendments of 1965. Although the program was initially intended to cover the population on welfare (for example, recipients of Aid to Families with Dependent Children, AFDC, or Supplemental Security Income, SSI ), over time legislation has expanded coverage to non-welfare recipients overwhelmed by their medical costs. Box 1 provides a chronology of important Medicaid-related legislation for the elderly. Two key themes emerge from box 1. First, Medicaid has increased the number of services provided over time. Second, Medicaid has attempted to limit the abuse of the system by using increasingly stringent and comprehensive asset tests to determine eligibility. For our purposes, it is useful to divide elderly Medicaid recipients into three groups: 1) the categorically needy, whose low income and assets qualify them for Medicaid. This group includes those who qualify for SSI, as well as “dual eligibles,” whose Medicare deductibles and co-pays are covered by Medicaid; 2) the institutionalized medically needy, who qualify for Medicaid because their financial resources do not cover their nursing home expenses; and 3) the noninstitutionalized medically needy, who qualify for Medicaid because their financial resources cannot cover catastrophic noninstitutional medical expenses. Each group faces a different set of asset and income tests. Figure 1 presents data on Medicaid enrollment and expenditures. In 2008, Medicaid spent roughly $75 billion4 on 5.3 million beneficiaries aged 65 and older (data from the Center for Medicare and Medicaid Services). These data provide information on the number of people and expenditures in the different groups. Of those aged In most states, SSI recipients qualify for Medicaid as categorically needy recipients. Under the Social Security Act Amendments establishing SSI in 1972, states were mandated to provide elderly SSI recipients with Medicaid benefits. The law exempted states that in 1972 were using Medicaid eligibility criteria stricter than the newly enacted SSI criteria (Gruber, 2000). The 11 states that had the more restrictive rules for Medicaid are referred to as 209(b) states (Gardner and Gilleskie, 2009). SSI pays monthly benefits to people with limited incomes and wealth who are disabled, blind, or aged 65 years and older. There is a (maximum) monthly SSI benefit that is paid for by the federal government. States can supplement this benefit. Figure 2 plots the federally provided monthly SSI benefit from 1975 to 2010. Table 1 shows the state-level supplements for all states that have offered a supplement over the sample period. In contrast to the federal benefit, which in real terms has been constant, the state supplements have varied greatly over time as well as across states. To qualify for SSI, individuals must pass both an income test and an asset test. In non-209(b) states, the income test is based on the combined federal and state maximum monthly benefit. Individuals with no income receive this maximum monthly benefit if they pass the asset test. Otherwise, each individual’s “countable income” is deducted from the maximum to produce a net benefit. In most states, individuals receiving any benefit, no matter how small, are categorically eligible for Medicaid. This implies that the implicit marginal tax rate for the threshold dollar of countable income—the incremental dollar that pushes the individual over 18 1Q/2012, Economic Perspectives BOX 1 Medicaid time line Social Security Act Amendments of 1965 n Medicaid program enacted. n Medicare program for the elderly also started. Social Security Act Amendments of 1972 n Enacted Supplemental Security Income (SSI) program for elderly and disabled, replacing state-level programs that served the elderly and disabled. n Required states to extend Medicaid to SSI recipients or to elderly and disabled meeting that state’s 1972 requirements. Omnibus Reconciliation Act of 1981 n Section 1915(c) home- and community-based waiver program launched. This program allows people with serious health problems to obtain home-based care instead of nursing home care. Tax Equity and Fiscal Responsibility Act of 1982 n Allowed states to make institutionalized individuals pay for Medicaid services if they owned a home and did not plan to return to that home. Omnibus Reconciliation Act of 1986 n Allowed states to pay for Medicare premiums for Medicare beneficiaries with incomes below the poverty level (qualified Medicare beneficiaries, QMBs). Omnibus Reconciliation Act of 1990 n Allowed states to cover Medicare premiums for Medicare beneficiaries with incomes between 100 and 120 percent of the poverty level (specified low income beneficiaries, SLMBs). Omnibus Reconciliation Act of 1993 n Tightened prohibitions against transfer of assets in order to qualify for Medicaid nursing home coverage. Instituted a three-year look-back period. Required recovery of nursing home expenses from beneficiary estates. Deficit Reduction Act of 2005 n Increased cost sharing (for example, increased copayments for certain drugs) and reduced certain benefits. n Extended the look-back period for assessing transfers from three to five years. n Imposed an upper bound on the amount of home equity excluded from asset tests. Sources: For 1965–93, Kaiser Commission on Medicaid and the Uninsured (2002); for 2005, Kaiser Commission on Medicaid and the Uninsured (2006). the income threshold—is extremely high, because that last dollar of income eliminates the individual’s Medicaid coverage. The conversion of actual income into countable income depends on whether the income is earned or unearned. Earned income consists of financial or inkind income from wages, self-employment (net), and sheltered workshops5 Each dollar of earned income in excess of $65 counts as 50 cents of countable income. Unearned income includes Social Security benefits, worker or veteran compensation, annuities, rent, and interest from assets. Each dollar of unearned income Federal Reserve Bank of Chicago counts as one dollar of countable income. In addition, the first $20 of income, earned or unearned, is disregarded; the amount varies slightly across states. By way of example, in 2010 the maximum federal benefit for single, aged SSI recipients was $674. To qualify for SSI, an individual must have had less than $674 × 2 + $65 + $20 = $1,433 of earned income or $674 + $20 = $694 in unearned income. Finally, several types of income, most notably food stamps, are excluded from the income test.6 The income standards used by the 209(b) states do not have to follow this formula, although some do. 19 figure 1 Medicaid enrollment and expenditures by maintenance assistance status in 2008, age 65+ A. Enrollment B. Expenditures 21% 27% 40% 41% 29% 23% 10% 9% SSI recipients Dual eligibles Medically Needy Other Source: Centers of Medicare and Medicaid Services, Medicaid Statistical Information System (MSIS). as that last dollar of assets eliminates the individual’s SSI and Monthly federal SSI benefit for aged individuals Medicaid benefits. Such a penalty living independently, 1975–2010 provides a strong disincentive to saving and encourages people to monthly SSI benefit spend down their assets until they 800 fall below the threshold. 700 The asset threshold varies 600 across states, with a modal value of $2,000. It is also the case, however, 500 that many important categories of 400 wealth are exempt, including one’s 300 principal residence. Box 2 lists 200 assets that are excluded for elderly individuals. 100 Table 2 shows the current 0 income and asset thresholds for 1975 ’80 ’85 ’90 ’95 2000 ’05 ’10 each state. The 209(b) states appear Nominal benefit Real benefit (2010$) at the bottom of the table. The only common factor across 209(b) states Source: Data from U.S. Social Security Administration, available at www.ssa.gov/oact/ COLA/SSIamts.html, deflated using Consumer Price Index data from Haver Analytics. is that individuals have to apply for Medicaid separately from their SSI benefit application. Although some of the 209(b) states impose tighter The law only requires that the states impose criteria income or asset restrictions for Medicaid, SSI eligino stricter than those in effect in 1972 (House Ways bility implies Medicaid eligibility in most of these and Means Committee, 2004). states. The asset test is more straightforward. Individuals The categorically needy: Dual eligibles with assets at or below the state-specific threshold figure 2 qualify. Individuals with assets above the threshold do not qualify. This implies that the implicit marginal tax rate for the threshold dollar of assets is extremely high, 20 “Dual eligibles” are individuals who are enrolled in Medicaid and have Medicaid pay their Medicare premiums. Medicare covers basic health services, 1Q/2012, Economic Perspectives Table 1 State SSI supplements (in 2010 dollars) for aged individuals living independently (selected years, 1975–2009) State Alaska California Colorado Connecticut District of Columbia Hawaii Idaho Illinoisa Maine Massachusetts Michigan Minnesota Nebraska Nevada New Hampshire New Jersey New York Oklahoma Oregon Pennsylvania Rhode Island South Dakota Utah Vermont Washington Wisconsin Wyoming 1975 1980 1985 1990 1996 2002 2009 575 622 529 552 503 439 588 409 482 363 407 217 249 233 109 146 118 90 78 45 25 0 270 286 611 0 245 171 0 40 30 25 7 0 233 69 40 10 8 7 6 370 255 196 158 122 51 63 27 NA NA NA NA NA NA NA 41 26 20 17 14 12 233 450 363 261 215 175 156 233 49 64 55 50 19 17 233 126 90 71 125 113 98 233 271 199 140 63 17 10 233 223 124 73 60 50 44 37 49 122 55 45 38 33 41 97 61 63 52 43 38 233 247 167 124 144 120 105 95 109 209 122 107 75 64 45 69 32 4 3 3 2 2 81 85 65 53 38 33 233 126 111 109 107 89 78 233 0 40 30 25 21 18 15 0 26 20 10 0 0 233 117 109 107 105 65 72 246 146 114 77 47 35 32 47 284 265 203 172 117 102 85 0 53 41 33 14 12 25 Illinois supplements are determined on a case-by-case basis. Notes: Converted to 2010 dollars using Consumer Price Index data from Haver Analytics. NA indicates not applicable. Sources: For 1975–2002, U.S. House of Representatives, House Ways and Means Committee (2004); for 2009, Social Security Administration (2009b). a including physicians and hospital care, for the elderly. Medicare Part B, which covers outpatient services such as doctor visits, costs $96.40 per month. As a dual eligible, an aged individual can get Medicaid to cover Medicare premiums and services that Medicare does not cover. Depending on their income, dual eligibles can qualify as Qualified Medicare Beneficiaries (QMBs), Specified Low-Income Medicare Beneficiaries (SLMBs), or Qualified Individuals (QIs). QMBs are assisted with Medicare Part B premiums and co-payments. In most states, the QMB income limit is 100 percent of the federal poverty level ($903 for single elderly people), and the asset limit is $6,600. However, nine states (including New York) do not impose any asset limits, and a subset of these states also provide more generous income limits and disregard amounts. SLMBs are elderly individuals with income between 100 percent and 120 percent of the federal poverty level. SLMBs are assisted with premiums only. QIs are individuals with income between 120 percent and 135 percent of the poverty level who, depending on funding availability, Federal Reserve Bank of Chicago may receive assistance with Medicare Part B premiums (Kaiser Commission on Medicaid and the Uninsured, 2010a and 2010b). Table 3 shows the asset and income limits for QMBs, SLMBs, and QIs. The medically needy Individuals with income or assets above the categorically needy limits may nonetheless not have enough resources to cover their medical expenses. Under the medically needy provisions, Medicaid pays part of these expenses. The implementation of medically needy coverage, however, varies greatly across states and types of medical care. The types of care covered under these arrangements include institutional (long-term) care, as well as home- and community-based service (HCBS) care. As pointed out earlier, the term “medically needy” has both a loose and a strict definition. The loose definition we use refers to all programs for receiving Medicaid due to catastrophic medical expenses. However, in formal Medicaid language, the term “Medically Needy” 21 BOX 2 Assets excluded from the SSI asset test 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. The home you live in and the land it is on, regardless of value. Property that you use in trade (gas station, beauty parlor, etc.). Personal property used for work (tools, equipment, etc.). Household goods and personal effects. Wedding and engagement rings. Burial funds (up to $1,500). Term life insurance policies (regardless of face value) and whole life insurance policies (with face value up to $1,500). One vehicle (regardless of value). Retroactive SSI or social security benefits for up to nine months after you receive them (includes payments received in installments). Grants, scholarships, fellowships, or gifts set aside to pay educational expenses for up to nine months after you receive them. Some property may be partially excluded, such as the property used to produce goods or services needed for daily life, and nonbusiness property that produces income, such as rented land, real estate, or equipment. Source: Social Security Administration (2009a). refers to just one of several mechanisms for coping with unaffordable medical expenses. As a rule, we will use the lowercase term “medically needy” to refer to the loose definition, and the uppercase term “Medically Needy” to refer to the formal program. Figure 3 presents a diagram of how individuals may qualify for medically needy coverage under the various provisions. In addition to having different mechanics, the provisions impose different asset and income thresholds. For example, Medicaid imposes more generous asset limits for noninstitutional care. We discuss these provisions below. The institutionalized medically needy We begin by looking at provisions for institutional (that is, nursing home) care.7 If an institutionalized elderly individual’s monthly income is within 300 percent of the SSI limit, then she qualifies for Medicaid (Gruber, 2000) in 39 states, plus the District of Columbia, through the expanded nursing home provision. Virtually all of the person’s income will still be applied toward the cost of care, and the individual will get an allowance. If an institutionalized person’s income is greater than 300 percent of the SSI limit, but still insufficient to cover her medical expenses, she may qualify for Medicaid through one of two mechanisms. The first option is to use the formal Medically Needy provision, which can be used for any sort of medical expense, to cover institutional care. The individual will have a “spend- 22 down” period that lasts until her net income—income less medical expenses—falls below the Medically Needy threshold. After qualifying as medically needy, the person still has to direct most of her income to pay for her care. She can keep only a small amount as a personal allowance, while Medicaid uses the rest to keep the individual at the institution (Gruber, 2000). The second mechanism for receiving institutional care is to use a Qualified Income or Miller trust. Income deposited in these trusts is excluded from the Medicaid tests. The individual deposits enough income in a trust to fall below the 300 percent limit and qualify for expanded nursing home coverage. Once the individual passes away, the state receives any money remaining in the trust, up to the amount that Medicaid has paid on the individual’s behalf8 (Weschler, 2005). Of the 39 states offering enhanced nursing home coverage, 25 also offer Medically Needy coverage. The remaining 15 states are required by federal law to allow applicants to use Miller trusts. Four of the states that provide medically needy coverage permit Miller trusts as well (Stone, 2002). Of the 11 states not offering expanded nursing home coverage, nine offer Medically Needy coverage. The difference between these states and the states offering expanded nursing home coverage is that individuals in these states are not automatically eligible for Medicaid nursing home care if their incomes are below 300 percent of the SSI level. However, given 1Q/2012, Economic Perspectives Table 2 Income and asset limits (in $) for SSI Medicaid recipients, 2009 Maximum federal SSI and Medicaid SSI benefit plus Disregarded state supplement income State asset limitb, d Non-209(b) states Alabama 2,000 674 2,000 1,262 Alaskaa Arizona No limit 903 Arkansas 2,000 674 California 2,000 907 Colorado 2,000 699 Delaware 2,000 674 District of Columbia 4,000 907 Florida 5,000 674 Georgia 2,000 674 Idaho 2,000 701 Iowa 2,000 674 Kansas 2,000 674 Kentucky 2,000 674 Louisiana 2,000 674 Maine 2,000 907 Maryland 2,500 674 Massachusetts 2,000 907 Michigan 2,000 907 Mississippi 4,000 724 Montana 2,000 674 Nebraska 4,000 907 Nevada 2,000 711 New Jersey 4,000 907 New Mexico 2,000 674 New York 4,350 769 North Carolina 2,000 903 Oregon 4,000 676 Pennsylvania 2,000 907 Rhode Island 4,000 907 South Carolina 4,000 903 South Dakota 2,000 689 Tennessee 2,000 674 Texas 2,000 674 Utah 2,000 907 Vermont 2,000 920 Washington 2,000 721 West Virginia 2,000 674 Wisconsin 2,000 759 Wyoming 2,000 699 209(b) states SSI: 2,000 Connecticut Medicaid 1,600 845 2,000 1,044 Hawaiic Illinois 2,000 674 SSI: 2,000 Indiana Medicaid: 1,500 674 Minnesota 3,000 907 SSI: 2,000 Medicaid: 1,000 768 Missouric SSI: 2,000 Medicaid: 1,500 715 New Hampshirec North Dakota 3,000 674 Ohio SSI: 2,000 Medicaid: 1,500 674 Oklahoma 2,000 719 Virginia 2,000 722 Monthly (earned) income limit for SSI/Medicaid eligibility 20 1,433 20 2,609 20 1,891 20 1,433 230 2,109 20 1,483 20 1,433 20 1,899 20 1,433 20 1,433 20 1,487 20 1,433 20 1,433 20 1,433 20 1,433 75 1,954 20 1,433 20 1,899 20 1,899 50 1,563 20 1,433 20 1,899 20 1,507 20 1,899 20 1,433 20 1,623 20 1,891 20 1,437 20 1,899 20 1,899 20 1,891 20 1,463 20 1,433 20 1,433 20 1,899 20 1,925 20 1,527 20 1,433 20 1,603 20 1,483 278 2,033 20 2,173 25 1,438 20 1,433 20 1,899 20 1,621 13 20 1,508 1,433 20 1,433 20 1,523 20 1,529 Based on Alaska Public Assistance payments. Disabled individuals under the age of 65 face no asset limits. Individuals receiving reduced SSI benefits may not qualify for Medicaid. d In 209(b) states, SSI and Medicaid asset limits are sometimes different. Source: Kaiser Commission on Medicaid and the Uninsured (2010b). a b c Federal Reserve Bank of Chicago 23 Table 3 Income and asset limits (in $) for dual eligibles, 2010 State Monthly Monthly Monthly Income income limit, income limit, income limit, disregard Asset QMBs SLMBs QIs amount limit Non-209(b) states Alabama 903 1,083 1,219 Alaska 1,108 1,333 1,503 Arizona 903 1,083 1,219 Arkansas 903 1,083 1,219 California 903 1,083 1,219 Colorado 903 1,083 1,219 Delaware 903 1,083 1,219 District of Columbia 2,706 2,708 NA Florida 903 1,083 1,219 Georgia 903 1,083 1,219 Idaho 903 1,083 1,219 Iowa 903 1,083 1,219 Kansas 903 1,083 1,219 Kentucky 903 1,083 1,219 Louisiana 903 1,083 1,219 Maine 1,354 1,535 1,670 Maryland 902 1,083 1,218 Massachusetts 903 1,083 1,219 Michigan 903 1,083 1,219 Mississippi 903 1,083 1,219 Montana 903 1,083 1,219 Nebraska 903 1,083 1,219 Nevada 903 1,083 1,219 New Jersey 903 1,083 1,219 New Mexico 903 1,083 1,219 New York 903 1,083 1,219 North Carolina 903 1,083 1,219 Oregon 903 1,083 1,219 Pennsylvania 903 1,083 1,219 Rhode Island 903 1,083 1,219 South Carolina 903 1,083 1,219 South Dakota 903 1,083 1,219 Tennessee 903 1,083 1,219 Texas 903 1,083 1,219 Utah 903 1,083 1,219 Vermont 903 1,083 1,219 Washington 903 1,083 1,219 West Virginia 903 1,083 1,219 Wisconsin 903 1,083 1,219 Wyoming 903 1,083 1,219 20 20 20 20 20 20 20 QMB: 1,803; SLMB: 1,625; QI: NA 20 20 20 20 20 20 20 75 20 20 20 50 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 209(b) states Connecticut 1,779 1,960 2,092 Hawaii 1,039 1,246 1,402 Illinois 903 1,083 1,219 Indiana 903 1,083 1,219 Minnesota 903 1,083 1,219 Missouri 903 1,083 1,219 New Hampshire 903 1,083 1,219 North Dakota 903 1,083 1,219 Ohio 903 1,083 1,219 Oklahoma 903 1,083 1,219 Virginia 903 1,083 1,219 QMB: 876; SLMB: 877; QI: 873 20 25 20 20 20 13 20 20 20 20 No limit 6,600 No limit 6,600 6,600 6,600 No limit No limit 6,600 6,600 6,600 6,600 6,600 6,600 6,600 No limit 6,600 6,600 6,600 No limit 6,600 6,600 6,600 6,600 6,600 No limit 6,600 6,600 6,600 6,600 6,600 6,600 6,600 6,600 6,600 No limit 6,600 6,600 6,600 6,600 No limit 6,600 6,600 6,600 10,000 6,600 6,600 6,600 6,600 6,600 6,600 Notes: QMB indicates qualified Medicare beneficiaries; SLMB indicates specified low-income Medicare beneficiaries; and QI indicates qualified individuals. NA indicates not applicable. Source: Kaiser Commission on Medicaid and the Uninsured, 2010b. 24 1Q/2012, Economic Perspectives figure 3 Eligibility flowchart for non-SSI Medicaid beneficiaries KEY GROUPS: a) Institutionalized Medically Needy b) Dual eligibles Receiving long-term care? YES At a nursing home? NO NO YES State has expanded nursing home Medicaid program? HCBS: people who receive long-term care services at home Qualify for QMB based on income and assets? YES YES Do you fit your state’s financial eligibility criteria? (300% of SSI in most states, but <300% of SSI in a few) Income within 300% of SSI? YES YES NO MEDICAID ELIGIBLE MEDICAID PAYS FOR MEDICARE PREMIUMS AND CO-PAYS Does state have medically needy program? NO YES NO NO Does state have medically needy program? TRY SPENDING DOWN MEDICAID ELIGIBLE NO TRY NURSING HOME CARE YES NO TRY SPENDING DOWN PUT EXCESS INCOME IN TRUST YES MEDICAID COVERS RESIDUAL COST MEDICAID COVERS RESIDUAL COST Federal Reserve Bank of Chicago MEDICAID ELIGIBLE Qualify for SLMB based on income and assets? PUT EXCESS INCOME IN TRUST MEDICAID ELIGIBLE NO MEDICAID PAYS FOR PREMIUMS ONLY SPEND DOWN TO MEDICALLY NEEDY INCOME LIMIT AND SEE IF MEDICAID WILL COVER YOU 25 that most individuals in nursing homes incur medical expenses far greater than 300 percent of the SSI level, there is little practical difference in Medicaid eligibility across the different states. All individuals with incomes below 300 percent of the SSI level in either type of state will deplete all their resources and will be eligible for Medicaid nursing home care through the Medically Needy program. The remaining two states, Indiana and Missouri, lack both provisions. However, Indiana and Missouri are both 209(b) states. To reduce the hardships that SSI beneficiaries may face in 209(b) states, federal rules require these states to allow individuals to spend down to the states’ income and asset limits for Medicaid.9 The rules thus mandate that 209(b) states offer the equivalent of a Medically Needy program, even if the states do not formally offer the Medically Needy option (Carpenter, 2000). Four 209(b) states—Indiana, Missouri, Ohio, and Oklahoma—offer a spend-down provision in accordance with this mandate. With this provision in place, institutionalized individuals in every state have at least one way to qualify for Medicaid if they are destitute and institutionalized.10 Table 4 shows the provisions offered in each state and the associated income and asset limits. In most states, the Medically Needy income limits (income less medical expenses) are stricter than the income limits for the categorically needy. Medicaid’s ability to recover assets from the estate The asset limits presented in table 4 are similar to the asset limits for the categorically needy presented in table 2. There are two key distinctions between the two sets of asset tests, both relating to their treatment of housing. First, the Medicaid asset test for the categorically needy excludes the individual’s principal residence, whereas the Deficit Reduction Act of 2005 stipulates that the Medicaid asset test for the medically needy places limits on the amount of home equity that is excluded. Although there are limits on the amount of home equity that can be excluded, the second-to-last column of table 4 shows that these limits are quite generous.11 Second, and more importantly, houses owned by institutionalized individuals who do not plan to return to that house no longer serve as principal residences.12 Therefore, the home equity of that individual is no longer excluded from the asset test. More precisely, the U.S. Department of Health and Human Services (2005c, p. 2) states that an individual’s house is included in the asset test when he “has no living spouse or dependents and moves into a nursing home or other medical institution on a permanent basis without the intent to return, transfers the home for less than fair market value, or dies.” An essential part of the definition is “the intent to return” provision, designed to exempt individuals 26 whose stays at the institution are temporary. In most states, the intent to return is based on the beliefs of the institutionalized individual, with no reference to the individual’s underlying medical condition. Only the 209(b) states are allowed to use more objective criteria, such as a professional medical diagnosis or the duration of stay, to assess the likelihood that the individual might return to his home. A mechanism that is available to non-209(b) states is to restrict the institutionalized individual’s income allowance so much that the individual can no longer cover property taxes and maintenance costs, forcing her to sell her home. However, individuals may be able to resist such “squeezes” by using reverse mortgages to fund taxes and maintenance (U.S. Department of Health and Human Services, 2005c). Once an individual dies, his home ceases to be protected. The Omnibus Reconciliation Act of 1993 requires states to seek from beneficiary estates reimbursement for long-term care, both in-house and institutional, and services provided concurrently with long-term care. However, states cannot pursue homes occupied by the beneficiary’s spouse or dependents (U.S. Department of Health and Human Services, 2005d). Furthermore, because the state may be one of many claimants to the estate, and given the general complexity of estate law—which in a few states explicitly protects estates from Medicaid claims—Medicaid collects relatively little money from estates.13 In 2004, estate recoveries equaled 0.8 percent of Medicaid spending on nursing homes, with the most successful state, Oregon, recovering 5.8 percent of its nursing home expenditures (U.S. Department of Health and Human Services, 2005a). Table 5 provides information on asset recovery practices and outcomes. One device states use to enhance their recovery prospects is to place liens on their beneficiaries’ assets. The Tax Equity and Fiscal Responsibility Act (TERFA) of 1982 allows states to place liens on the homes of permanently institutionalized Medicaid beneficiaries. After the beneficiary dies, states may also place “postdeath” liens on her estate (U.S. Department of Health and Human Services, 2005b). TERFA liens can help states protect themselves from abuses of the “intent to return” provision. While the intent to return is generally based on the subjective opinion of the beneficiary himself, TERFA liens may be established on the basis of objective criteria (U.S. Department of Health and Human Services, 2005b). Table 6 (p. 30) summarizes the criteria states use. TERFA liens also protect states if a beneficiary attempts to transfer the house to a third party (for example, a child) prior to applying for Medicaid. The 1Q/2012, Economic Perspectives Table 4 Income and asset limits (in $) for institutionalized medically needy Medicaid recipients, 2009 Asset State Coverage limit Income limit (income less Expanded medical nursing home expenses) coverage Non-209(b) states Alabama No NA NA Alaska No NA NA 360 Arizona Yes 5,000 b Arkansas Yes 2,000 108 California Yes 2,000 600 Colorado No NA NA Delaware No NA NA District of Columbia Yes 4,000 577 Florida Yes 5,000 180 Georgia Yes 2,000 317 Idaho No NA NA Iowa Yes 10,000 483 Kansas Yes 2,000 495 Kentucky Yes 2,000 217 Urban: 100; Louisiana Yes 2,000 Rural: 92 Maine Yes 2,000 903 Maryland Yes 2,500 350 Massachusetts Yes 2,000 9,035d Region 1: 341; Region 2: 341; Region 3: 350; Region 4: 375; Region 5: 391; Michigan Yes 2,000 Region 6: 408 Mississippi No NA NA Montana Yes 2,000 625 Nebraska Yes 4,000 392 Nevada No NA NA New Jersey Yes 4,000 367 New Mexico No NA NA New York Yes 2,000 767 North Carolina Yes 2,000 242 Oregon No NA NA Pennsylvania Yes 2,400 425 Rhode Island Yes 4,000 800 South Carolina No NA NA South Dakota No NA NA Tennessee Yes 2,000 241 Texas No NA NA Utah Yes 2,000 370 916 (991 for Vermont Yes 2,000 Chittenden) Washington Yes 2,000 674 West Virginia Yes 2,000 200 Wisconsin Yes 2,000 592 Wyoming No NA NA Federal Reserve Bank of Chicago Income allowed if Home institutionalized equity in 2003 limit Stateallowed Miller trust Yes Yes a Yes Yes No Yes Yes c No Yes Yes Yes Yes Yes Yes NA 500,000 Yes NA 500,000 Yes 76.65 500,000 Yes 40 500,000 Yes 35 750,000 No NA 500,000 Yes NA 500,000 Yes 70 750,000 No 35 500,000 Yes 30 500,000 No NA 750,000 Yes 30 500,000 Yes 30 500,000 No 40 500,000 No Yes Yes Yes No 38 500,000 40 750,000 40 500,000 60–65 750,000 Yes Yes Yes Yes Yes Yes Yes No No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 60 NA 40 50 NA 40 NA 50 30 NA 30 50 NA NA 30 NA 45 No No No No 500,000 No 500,000 Yes 500,000 No Disregardedc No 500,000 Yes 750,000 No 750,000 Yes 750,000 No 500,000 No 500,000 Yes 500,000 No 500,000 No 500,000 Yes 500,000 Yes 500,000 No 500,000 Yes 500,000 No 47.66 500,000 No 41.62 500,000 No NA 500,000 No 45 750,000 No NA 500,000 Yes 27 Table 4 (continued) Income and asset limits (in $) for institutionalized medically needy Medicaid recipients, 2009 Asset State Coverage limit Income limit (income less medical expenses) Expanded nursing home coverage Income allowed if Home institutionalized equity in 2003 limit 209(b) states Region A: 576; Connecticut Yes 1,600 Regions B and C: 476 Yes Hawaii Yes 2,000 469 No Illinois Yes 2,000 903 No NA No Indiana No e NA Minnesota Yes 3,000 677 No NA No Missouri No e NA New Hampshire Yes 2,500 591 Yes North Dakota Yes 3,000 750 No NA NA Yes Ohio No e NA NA Yes Oklahoma No e Group I: 281; Group II: 324; Yes 2,000 Group III: 421 Yes Virginiaf Stateallowed Miller trust 54 30 30 NA 69 NA 50 40 NA NA 750,000 No 750,000 No NA No 500,000 No 500,000 No 500,000 No 500,000 No 500,000 No 500,000 Yes 500,000 Yes 30 500,000 No NA indicates not applicable. Income limit frozen at $1,656. a Liquid asset limit—total assets, including housing, cannot exceed $100,000. b Income limit set at 250 percent, rather than 300 percent, of SSI limit. c Limit is $1,200 for those with professional care assistance. d State is required to offer a spend-down provision. e f The state of Virginia is split into three groups, each with a different Medically Needy income limit. Source: Kaiser Commission on Medicaid and the Uninsured (2010b); Miller trust information from Stone (2002). Deficit Reduction Act of 2005 extended Medicaid’s “look-back” period from the three years preceding application to five years. Transfers made during the lookback period are subject to Medicaid review. If the applicant is found to have made a net transfer, that is, sold some of his assets at prices below their fair market value, his eligibility will be delayed (ElderLawNet, Inc., 2011). The degree to which elderly individuals transfer their assets in order to become eligible for Medicaid has been the subject of several studies. These studies find that the elderly transfer little if any of their money to their heirs for the purpose of making themselves eligible for Medicaid. Thus, extending the look-back period past five years or more aggressive pursuit of transferred assets is unlikely to defray much of Medicaid’s expenses. Norton (1995) argues that elderly individuals are more likely to receive transfers in an attempt to avoid Medicaid. In contrast, Bassett (2007) finds that “the self-assessed probability of entering a nursing home is a significant determinant of making an asset transfer.” Bassett estimates that in 1993 there were about $1 billion “Medicaid-induced” asset transfers, equaling about 3 percent of total Medicaid expenditures. Many of the 28 people making the transfers, however, did not receive Medicaid long-term care benefits, implying a smaller final cost to Medicaid. Waidmann and Liu (2006) study asset transfers over the period 1995 to 2004. They conclude that “even the most aggressive pursuit of transferred assets would recover only about 1 percent of total Medicaid spending for long-term care.” Reviewing the literature, O’Brien (2005) concludes that the evidence “do[es] not support the claim that asset transfers are widespread or costly to Medicaid.” In summary, the evidence is mixed whether the elderly give or receive transfers to affect their Medicaid eligibility. However, there is a clear consensus that these transfers are small relative to the size of Medicaid transfers. The noninstitutionalized medically needy The structure of Medicaid coverage for noninstitutionalized medically needy individuals is similar to that for those in institutions. Individuals with specific needs, such as home health care, can qualify under provisions tailored to those needs. Individuals not qualifying under these limited provisions can qualify under the general medically needy provision, if their state offers it. 1Q/2012, Economic Perspectives Individuals needing long-term care can often substitute home-based care for care at a nursing home Share of Medicaid nursing home expenses or another institution. To promote the use of home-based collected from estates care, states can utilize 1915(c) home- and community Medicaid collections/ based service care (HCBS) waivers, which give them State nursing home costs (%) additional flexibility in how they provide these services (Carpenter, 2000). Services that can be offered under Alabama 0.8 an HCBS waiver range from traditional medical services, Alaska 0.0 a Arizona 10.4 such as dental care and skilled nursing services, to Arkansas 0.4 nonmedical services, such as case management and California 1.5 environment modification. Colorado 1.5 In most states, the income test used for 1915(c) Connecticut 0.8 Delaware 0.3 waivers is the same as the one used for expanded nursing District of Columbia 1.0 home coverage, namely 300 percent of the SSI limit. Florida 0.6 Other states (for example, California) impose more Georgia 0.0 stringent tests. As Table 4 shows, many states (includHawaii 0.9 Idaho 4.5 ing Arizona) allow the use of Miller trusts. As with Illinois 1.3 the expanded nursing home program, beneficiaries Indiana 1.8 are expected to direct their income toward the cost of Iowa 2.9 their expenses. The income allowances, however, Kansas 1.4 Kentucky 0.9 vary greatly across states (Walker and Accius, 2010). Louisiana 0.0 The asset limits for 1915(c) applicants are the ones Maine 2.5 for the categorically needy (Stone, 2002). Housing is Maryland 0.6 Massachusetts 2.0 excluded from the asset test, but the Omnibus ReconciliMichigan 0.0 ation Act of 1993 requires states to pursue estates to reMinnesota 2.8 cover the cost of long-term care. On the other hand, states Mississippi 0.1 do not have to pursue these costs if they decide it would Missouri 1.1 Montana 1.4 not be cost-effective (U.S. Department of Health and Nebraska 0.3 Human Services, 2005d). Given the limited success of Nevada 0.3 state cost recovery efforts in general, such efforts are New Hampshire 1.6 unlikely to play a large role in the case at hand. New Jersey 0.6 New Mexico 0.0 Some states limit access by requiring 1915(c) New York 0.5 beneficiaries to exhibit difficulties in performing at North Carolina 0.5 least three “activities of daily living” (bathing, dressing, North Dakota 1.2 grooming, and so on); functional eligibility for nursing Ohio 0.5 Oklahoma 0.3 homes requires only two. Most states impose limits Oregon 5.8 on how much they spend per year for home and comPennsylvania 0.1 munity-based service care. Furthermore, states are free Rhode Island 1.0 to choose how many applications to approve. They South Carolina 1.3 South Dakota 1.0 are also free to limit the number of waivers.14 Many Tennessee 0.9 states have more individuals in need of waivers than Texas 0.0 open “slots,” and thus operate waiting lists (Kaiser Utah 0.0 Vermont 0.4 Commission on Medicaid and the Uninsured, 2009). Virginia 0.1 Table 7 summarizes the 1915(c) HCBS waiver proWashington 1.8 grams offered by each state. West Virginia 0.1 In addition to utilizing 1915(c) waivers, states can Wisconsin 1.8 Wyoming 2.7 provide HBCS services under two other provisions: the federally mandated home health benefit provided Results for Arizona are not comparable to those for other states because of data issues arising from the extensive use of prepaid by all states; and the optional personal care benefit, managed care contracts. which in 2006 was provided by 31 states. In 2006, Sources: Probate data—Karp, Sabatino, and Wood (2005); policy range and collections data—U.S. Department of Health and Human the two programs incurred 34 percent of total HCBS Services (2005a). expenditures and assisted 61 percent of the HCBS beneficiaries. Most states screened applicants to these Table 5 a Federal Reserve Bank of Chicago 29 Table 6 Decision criteria for TERFA liens Number of Length months Intent Other of stay triggering to return Physician’s third-party State presumption presumption home declaration evaluation Other Alabama Yes 3 Yes Yes No No Arkansas Yes 4 Yes Yes No No California Yes No No No No No Connecticut Yes 6 Yes Yes Yes Yes Delaware Yes 24 Yes No No No Hawaii Yes 6 Yes Yes No No Idaho Yes Yes No No No No Illinois Yes 4 Yes No No No Indiana NR Yes Yes Yes Yes Yes Maryland Yes NR Yes Yes No Yes Massachusetts Yes 6 Yes Yes Yes No Minnesota Yes 6 Yes No No No Montana Yes Yes No No No Yes New Hampshire Yes No No No No Yes New York Yes No No No No No Oklahoma Yes 6 Yes Yes No No South Dakota Yes Yes No No No Yes West Virginia NR NR Yes No No Yes Wyoming NR No NR NR NR NR Notes: TERFA is the Tax Equity and Fiscal Responsibility Act of 1982. NR indicates no response. Source: Karp, Sabatino, and Wood (2005). programs with the income and asset tests for categorically needy recipients. There is variation in the financial eligibility limits states require to get this benefit. Some states keep it at the 300 percent level, but others restrict it further. Many states also provide a medically needy spend-down option (Kaiser Commission on Medicaid and the Uninsured, 2009). The noninstitutionalized medically needy: Other pathways For individuals unable to qualify under any of the preceding pathways, the Medically Needy provision provides an important “last chance” opportunity to qualify for Medicaid (Crowley, 2003). The income and asset levels for the noninstitutionalized Medically Needy applicants are the same as the ones for institutionalized individuals presented in table 4. Similarly, noninstitutionalized individuals with high incomes end up paying most if not all of their medical expenses before they receive aid. Because the income limits for the Medically Needy provision are usually stricter than the limits for the “income needy” (for example, the SSI recipients, dual eligibles, and certain HCBS beneficiaries), noninstitutionalized individuals also face a possible discontinuity in coverage. In consequence, the penalty to being 30 Medically Needy rather than income needy may be significant. By way of example, consider two individuals in Pennsylvania. Both individuals require health care costing $500 per month. The first individual has a monthly income of $900 per month, which in Pennsylvania allows him to qualify as categorically needy (table 2). This person pays nothing for medical care. The second individual has a monthly income of $1,100 and does not qualify as categorically needy. Deducting medical expenses leaves her with a net income of $600, which is above Pennsylvania’s Medically Needy net income limit (table 4). In short, receiving an additional $200 of income costs the second person $500 of Medicaid benefits. The quantitative importance of these discontinuities is of course an empirical matter, depending both on the formal provisions and their practical application by Medicaid administrators. Discussion In a number of recent studies, the joint effect of Medicaid and public assistance programs such as SSI is modeled as a consumption floor: If an individual is not able to cover her medical expenses and purchase a minimal amount of consumption, the government will cover the difference (Hubbard, Skinner, and Zeldes, 1995; Palumbo, 1999; De Nardi, French, and Jones, 1Q/2012, Economic Perspectives Table 7 Eligibility criteria for Medicaid 1915(c) HCBS waivers, 2008 Tougher Income limit Waiting Income limit for Waiting list functional Income for the aged ` list for the aged/disabled for the requirements;b allowedc States (% of SSI limit)a the aged (% of SSI limit)a aged/disabled cost limits (in $) Non-209(b) States Alabama 300, MT 7,094 Yes; yes UL Alaska 300, MT 0 No; yes 1,656 d Arizona NP Arkansas 300, MT 0 No; yes UL California 100 1,200 No; yes ≤2,022 Colorado 300, MT 0 No; no 2,022 Delaware 100, MT 0 250, MT 0 Yes; no 1,685 District of Columbia 300 0 No; yes 2,022 Florida 300, MT 0 300, MT 12,684 Yes; yes 674 Georgia 300, MT 763 Yes; no 674 Idaho 300, MT 0 No; no 674 e Iowa 300, MT 0 No; yes 2,022 Kansas 300 0 Yes; yes 727 Kentucky 300, MT 0 No; yes 694 Louisiana 300 8,433 No; yes 2,022 Maine 300 0 No; yes 1,128 Maryland 300 6,000 No, yes 2,022 Massachusetts 100 0 No; no 2,022 Michigan 300 3,404 No; no 2,022 Mississippi 300, MT 6,000 Yes; yes UL Montana 100 600 No; yes 625 Nebraska 100 0 No; yes 903 Nevada 300, MT 343 300, MT 0 No; no UL New Jersey 300 0 No; yes 2,022 New Mexico 300 5,000 No; no UL New York 300, MT 0 Yes; yes 787 North Carolina 100 6,000 No; yes 903 Oregon 300, MT 0 No; yes 1,822 Pennsylvania 300 0 No; yes 2,022 Rhode Island 300 0 300 99 No; no 923 South Carolina 300, MT 2,016 No; yes 2,022 South Dakota 300, MT 0 No; yes 694 Tennessee 300, MT 350 No; yes 1,348 Texas 300, MT 40,107 Yes; yes 2,022 Utah 300 0 Yes; no ≥903, ≤2,022 Vermont NP Washington 300 0 No; yes ≤2,022 West Virginia 300 0 No; yes 674 Wisconsin 300 13,296 No; no ≤2,022 Wyoming 300, MT 210 No; yes UL 209(b) states Connecticut Hawaii Illinois Indiana Minnesota Missouri New Hampshire North Dakota Ohio Oklahoma Virginia 300 0 100 100 100 0 100 0 100, MT 1,279 300 0 100 0 100 0 100 0 300, MT 1,224 300, MT 0 300 0 300 0 No; yes No; no No; no No; yes No, yes No; yes No; no No; no No; yes No; yes No; no 1,805 1,128 674 2,022 935 1,113 Varies 750 1,314 1,011 ≤2,022 MT indicates that the state allowed Miller trusts in 2009–10. Individual must exhibit difficulty performing three (rather than two) activities of daily living. c Cost allowance for 2009–10. These limits may be exceeded through the use of Miller trusts. d Offers a similar program. e Allowance is $1,128 for renters. Note: HCBS is home- and community-based service care; NP indicates not a participant; UL denotes unlimited with a Miller trust; ≤ means at most, but the income allowance depends on multiple factors. Source: Kaiser Commission on Medicaid and the Uninsured (2009); Miller trust information from Walker and Accius (2010). a b Federal Reserve Bank of Chicago 31 2010; French and Jones, 2011). Is this a reasonable approximation of the Medicaid system? Our review suggests that the effective consumption floor provided by Medicaid varies greatly by income and asset levels, as well as medical conditions. Individuals in nursing homes are given much smaller allowances, and are more likely to forfeit the value of their house, than noninstitutionalized individuals. This distinction has been recognized by Brown and Finkelstein (2008), among others. The extent to which institutionalized individuals must surrender their homes depends on a number of factors, including the interpretation of the intent to return, the willingness of the state to impose liens, and the effectiveness of estate recovery, all of which vary across states. We also find the potential for discontinuities in coverage. Medicaid recipients can be placed in two groups. The first group is the income needy, who receive benefits because they have low incomes. Income-needy individuals include those receiving expanded nursing home coverage, many recipients of HCBS services, and dual eligibles, as well as the categorically needy. The second group is the expenditure needy, who receive benefits because their medical expenses are large relative to their income. This group includes individuals utilizing Miller trusts, as well as the Medically Needy. In some cases, the net income (income less medical expenses) limits for the medically needy are stricter than the income limits for the income needy. This raises the possibility that the income needy receive more generous coverage. We believe that the scope for such unequal treatment is greatest for noninstitutionalized individuals. NOTES Figure is taken from the Kaiser Family Foundation (2010). 1 Figures are taken from the 2010 Medicaid Actuarial Report (Office of the Actuary, Centers for Medicare and Medicaid Services, 2010) for those who are “aged.” Data from the Medicaid Statistical Information System show that over 0.6 million disabled people are also aged 65 and older. 2 Figures are taken from the U.S. Bureau of Economic Analysis, 2011, tables 3.1 and 3.12. 3 Data from the Medicaid Statistical Information System (MSIS) cited in figure 1 show $68.3 billion, but these data do not include certain payments such as Medicare premiums paid for dual eligibles. For this reason, the MSIS data likely understate dual eligibles’ share of total expenditures. Also, the MSIS categories are slightly different from those in figure 1. However, virtually all “cash recipients” over age 65 are those receiving SSI and virtually all “poverty related” individuals over age 65 are dual eligibles. 4 5 Sheltered workshops are organizations that provide employment to people with disabilities (Sheltered Workshops. Inc, 2011). In addition to food stamps, the exempt categories include income that is set aside toward an approved plan for achieving self support (used by the blind and disabled to pay off educational or vocational costs), and certain types of assistance for home energy needs. 6 The remainder of this section utilizes overviews by Stone (2002), Walker and Accius (2010), and the Kaiser Commission on Medicaid and the Uninsured (2010). 7 Prior to the passage of the Omnibus Budget Reconciliation Act in 1993, it was acceptable to place extra income in a self-created discretionary fund to acquire Medicaid coverage. Since 1993, apart from limited trusts such as the Miller or Qualified Income trusts, most discretionary trust funds are treated as countable income or assets and may restrict people from obtaining Medicaid (see Goldfarb, 2005). 8 32 9 The mandate is in the 2000 House Bill 1111, Section 11.445, which specifies that an individual eligible for or receiving nursing home care must be given the opportunity to have those Medicaid dollars follow them to the community and to choose the personal care option in the community that best meets their needs (Niesz, 2002). This raises the possibility of a discontinuity in coverage. An individual whose income is $1 above the categorically needy limit may need to spend a considerable amount to qualify under the Medically Needy provision. However, in practice the discontinuity in coverage is unimportant in most cases because institutionalized Medicaid recipients must spend almost all of their income on their care. The median cost of nursing home care was $5,550 per month in 2010. Whether an individual’s income is slightly more or less than 300 percent of the SSI limit ($674 × 3 = $2,022), Medicaid will still provide a nursing home, but all of their income must be put toward the cost of the nursing home. 10 If a spouse or dependent resides in the house, the equity limits do not apply (ElderLawNet, Inc., 2011). 11 The inclusion of housing in the asset tests for institutionalized individuals applies to the categorically needy as well as the medically needy. Most categorically needy individuals, however, do not hold significant housing equity (U.S. Department of Health and Human Services, 2005c). 12 States do not have to pursue an estate if they determine pursuit would not be cost-effective. The definition of “cost-effective,” not surprisingly, varies across states (U.S. Department of Health and Human Services, 2005d). 13 For example, New Hampshire and Michigan limit 1915(c) waivers for the aged to those who are also disabled. Only two states, Arizona and Vermont, do not offer HCBS waivers, and Arizona offers a similar program. 14 1Q/2012, Economic Perspectives references Bassett, William F., 2007, “Medicaid’s nursing home coverage and asset transfers,” Public Finance Review, Vol. 35, No. 3, May, pp. 414–439. Gruber, Jonathan, 2000, “Medicaid,” National Bureau of Economic Research, working paper, No. 7829, August. Brown, Jeffrey R., and Amy Finkelstein, 2008, “The interaction of public and private insurance: Medicaid and the long-term care insurance market,” American Economic Review, Vol. 98, No. 3, June, pp. 1083–1102. Hubbard, R. Glenn, Jonathan Skinner, and Stephen P. Zeldes, 1995, “Precautionary saving and social insurance,” Journal of Political Economy, Vol. 103, No. 2, pp. 360–399. Carpenter, Letty, 2000, “Financial eligibility rules and options,” in Understanding Medicaid Home and Community Services: A Primer; Felicity Skidmore, (managing ed.), Washington, DC: U.S. Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation, Office of Disability, Aging, and Long-Term Care Policy; and George Washington University, Center for Health Policy Research, October, pp. 20–39. Kaiser Commission on Medicaid and the Uninsured, 2010a, “Dual eligibles: Medicaid’s role for low-income Medicare beneficiaries,” report, Washington, DC, December. Crowley, Jeff, 2003, “Medicaid medically needy programs: An important source of Medicaid coverage,” Kaiser Commission on Medicaid and the Uninsured, issue paper, No. 4096, January. De Nardi, Mariacristina, Eric French, and John Bailey Jones, 2010, “Why do the elderly save? The role of medical expenses,” Journal of Political Economy, Vol. 118, No. 1, February, pp. 39–75. ElderLawNet, Inc., 2011, “Medicaid rules,” ElderLawAnswers, June, available at www.elderlawanswers.com/Elder_Info/Elder_ Article.asp?id=2751#1. French, Eric, and John Bailey Jones, 2011, “The effects of health insurance and self-insurance on retirement behavior,” Econometrica, Vol. 79, No. 3, May, pp. 693–732. Gardner, Lara, and Donna B. Gilleskie, 2009, “The effects of state Medicaid policies on the dynamic savings patterns and Medicaid enrollment of the elderly,” University of North Carolina, working paper, February. Goldfarb, David, 2005, “Supplemental needs and their impact on Medicaid and SSI eligibility,” Goldfarb Abrandt Salzman & Kutzin LLP, report, July, available at www.seniorlaw.com/sntimpact.htm. Federal Reserve Bank of Chicago , 2010b, “Medicaid financial eligibility: Primary pathways for the elderly and people with disabilities,” report, No. 8048, Washington, DC, February. , 2009, “Medicaid home and community-based service programs: Data update,” issue paper, No. 772003, Washington, DC, November. , 2006, “Deficit Reduction Act of 2005: Implications for Medicaid,” report, No. 7465, Washington, DC, February. , 2002, The Medicaid Resource Book, July. Karp, Naomi, Charles P. Sabatino, and Erica F. Wood, 2005, “Medicaid estate recovery: A 2004 survey of state programs and practices,” AARP Public Policy Institute, paper, No. 2005-06, June. Niesz, Helga, 2002, “Money follows the person: State Medicaid legislation,” OLR Research Report, Office of Legislative Research, No. 2002-R-0837, November 12. Norton, Edward C., 1995, “Elderly assets, Medicaid policy, and spend-down in nursing homes,” Review of Income and Wealth, Vol. 41, No. 3, pp. 309–329. O’Brien, Ellen, 2005, “Medicaid’s coverage of nursing home costs: Asset shelter for the wealthy or essential safety net?,” Georgetown University, Long-Term Care Financing Project, issue brief, May. Palumbo, Michael G., 1999, “Uncertain medical expenses and precautionary saving near the end of the life cycle,” Review of Economic Studies, Vol. 66, No. 2, April, pp. 395–421. 33 Sheltered Workshops, Inc., 2011, Washington, MO, available at http://shelteredworkshopsinc.org/about.html. Social Security Administration, 2009a, Annual Report of the Supplemental Security Income Program, Baltimore, MD: Social Security Adminstration, May. , 2009b, State assistance programs for SSI recipients,” Social Security Online, January available at www.ssa.gov/policy/docs/progdesc/ssi_st_asst/ 2009/index.html. Stone, Julie Lynn, 2002, “Medicaid: Eligibility for the aged and disabled,” CRS Report for Congress, Congressional Research Service, No. RL31413, July 5. U.S. Bureau of Economic Analysis, 2011, National Income and Product Accounts of the United States, April, available at www.bea.gov/national/nipaweb/ index.asp. U.S. Department of Health and Human Services, Office of Assistant Secretary for Policy and Evaluation, 2005a, “Medicaid estate recovery collections,” Medicaid Eligibility for Long-Term Care Benefits series, policy brief, Washington, DC, No. 6, September. , 2005b, “Medicaid liens,” Medicaid Eligibility for Long-Term Care Benefits series, policy brief, Washington, DC, No. 4, April. 34 , 2005c, “Medicaid treatment of the home: Determining eligibility and repayment for long-term care,” Medicaid Eligibility for Long-Term Care Benefits series, policy brief, Washington, DC, No. 2, April. , 2005d, “Medicaid estate recovery,” Medicaid Eligibility for Long-Term Care Benefits series, policy brief, Washington, DC, No. 1, April. U.S. House of Representatives, House Ways and Means Committee, 2004, Green Book 2004: Background Material and Data on Programs within the Jurisdiction of the Committee on Ways and Means, March 2004, Washington, DC: U.S. Government Printing Office, March. Waidmann, Timothy, and Korbin Liu, 2006, “Asset transfer and nursing home use: Empirical evidence and policy significance,” Kaiser Commission on Medicaid and the Uninsured, issue paper, No. 7487, April. Walker, Lina, and Jean Accius, 2010, “Access to long-term services and supports: A 50-state survey of Medicaid financial eligibility,” Insight on the Issues, AARP Public Policy Institute, No. I-44, September. Weschler, Kathy, 2005, “A qualified income trust: Keep your Medicaid eligibility,” MDA/ALS Newsmagazine, Vol. 10, No. 3, April. 1Q/2012, Economic Perspectives