I. Introduction

Kurt V. Krueger (2011) examined personal consumption rates for single persons. His comprehensive analysis covered several facets on the topic. He reviewed previous analyses of personal consumption rates by single persons from the Patton-Nelson series (Robert T. Patton and David M. Nelson, 1991; Michael R. Ruble, Robert T. Patton and David M. Nelson, 2009) and the U.S. Department of Justice (2002) as a part of the September 11th Victim Compensation Fund. Krueger then defined five alternative approaches that represent personal consumption as interpreted across various legal jurisdictions. He developed a taxonomy that allows grouping single-persons’ expenditures into components flexible enough to compute the various approaches that he defined. Finally, Krueger used 2005 to 2009 Consumer Expenditure Survey (CEX) microdata to estimate personal consumption rates for each of his approaches.

In this paper we update Krueger’s analysis using 2015 to 2019 CEX microdata and expand on his analysis as it relates to accumulation of estate. The CEX has evolved from 2005-2009 to 2015-2019 with the aggregation of some previously separate items, the addition of certain items and removal of others, and changes in the method used to estimate taxes. The sources of some expenditure data within the CEX have changed as well. While inflation was moderate over this period, it did vary across commodity groups. These factors, and more, potentially changed the mix of goods that people consume.

The original single-person consumption study used a narrow measure of accumulation of estate. We revisit this area of consumption, explaining why a somewhat broader approach may be appropriate and re-estimate accumulations as a percentage of income.

In Krueger’s (2011) study of single-person personal consumption, CEX microdata expenditures were aggregated into three categories: personal expenses, sharable expenses, and support to others. The first step in our analysis acquires the 2005-2009 CEX microdata and uses Krueger’s taxonomy to reproduce his results for single persons living alone. We are able to do this to a high degree of confidence. We then use that taxonomy to estimate personal consumption rates for single persons living alone using 2015-2019 CEX microdata. The 2005-2009 study looked at wage earners between the ages of 18 and 65. This 2015-2019 study expands that wage earner age range to age 70 to account for later exits from the labor market. Tables of personal consumption rates are developed that mirror Krueger’s. Both similarities and differences between results from the earlier and later data are discussed. Regression analysis develops continuous measures of personal consumption rates as functions of income.

The next section of this paper describes changes in the CEX from 2005-2009 to 2015-2019. It also discusses changes in the CPI across expenditure categories and how this may affect consumption.

Section III of this paper analyzes the 2015-2019 CEX microdata. Expenditures levels or personal consumption rates are developed for the various consumption approaches. The 2015-2019 results are compared to Krueger’s 2005-2009 findings. In many cases, the results are quite different. In addition, inflation adjusted 2005-2009 expenditures typically differ from 2015-2019 expenditures, suggesting why it is useful to re-estimate single-person expenditures using more recent data.

This section also delves more deeply into estate accumulation rates. We expand on Krueger’s initial definition by adding retirement contributions back into accumulation rates. The penultimate section describes regression analysis of selected rates as a function of income. The results will allow forensic experts to estimate consumption rates for any given income level. The presentation of regression results is followed by a conclusion.

II. Evolving CEX Data and Methods

Several changes occurred to the CEX between 2005-2009 and 2015-2019. These changes mainly reflect evolutions in technology, aggregations of formerly separate items, sources for expenditure data, and the estimation of taxes. The Bureau of Labor Statistics (BLS), 2005-2009, conducts the CEX, recording individuals’ expenditures on 700-plus items identified by universal classification codes (UCCs). As technology has evolved, some UCCs were dropped from the CEX and others were added. Examples of excluded UCCs include “pager service,” “personal digital assistants,” “telephone answering devices” and “rental of video cassettes, tapes, discs and films.” Newly added UCCs include “applications, games, ringtones for handheld devices,” “video game software” and “video game hardware and accessories.”

For some UCCs, the BLS combined them from separate items in the 2005-2009 surveys to an aggregate UCC in the 2015-2019 surveys. Five separate UCCs were combined into the aggregate “dinnerware, glassware and serving pieces.” Some separate clothing items were combined into aggregate broader UCCs. Formerly separate payments for cars or trucks were combined into “car/truck” payments. There are other examples as well.

The BLS uses both interview and diary surveys to collect individuals’ expenditures on UCCs. Expenditures on items that are purchased infrequently—cars or airfare, for example—are collected in an interview survey. Expenditures on items that are purchased frequently and more easily forgotten—food items, for example—are collected in the diary survey. Between the 2005-2009 and 2015-2019 surveys, the sources for some items switched from the interview survey to the diary survey or vice-versa.

The 2005-2009 CEX did not specifically track income taxes, only income tax withholdings. Krueger (2011) used the Internal Revenue Service’s 2008 Statistics of Income (2010) for single-person filers to compute federal taxes. He then used the ratio of state to federal withholdings to estimate state income taxes. Beginning in 2013, the CEX has estimated federal and state income taxes based on the consumer unit’s income and characteristics, using the NBER TaxSim program.¹ The tax estimates incorporate variables that are not present in the public-use data. CEX tax estimates are contained in the interview survey. We use an expectation-maximization algorithm with bootstrapping to impute estimated taxes from the diary survey based on the year of the survey, individual's age, before tax income, and income composition. (For a description of the imputation process, see James Honaker, Gary King and Matthew Blackwell, 2011.) The BLS imputes missing data five times, then uses the mean of the five imputations as an estimate of tax payments. We follow the BLS by also using the mean of five imputations to develop taxes and after-tax income in the diary survey.

Inflation may be one factor playing a role in changing consumption patterns between 2005-2009 and 2015-2019. Between 2009 and 2019 the CPI for all items rose by 19.2%. Increases in component parts, however, grew at varying rates. Over that decade, energy costs rose 11.5% while medical care costs rose 32.7%. Prices for food items rose 18.5% while prices for IT commodities actually fell by 47%.² These relative price changes may have altered consumption patterns between the two time periods. Certainly, other factors, such as changes in technology and tastes have influenced consumption as well.

Table 1 compares aggregate CEX data from 2009 to 2019.³ The first set of two columns show 2009-dollar expenditures in CEX summary categories and the associated share those expenditures make up of 2009 income. The second set of two columns show 2019-dollar expenditures in CEX summary categories and the associated share those expenditures make up of 2019 income. The last column shows the percentage point change in shares from 2009 to 2019. Expenditure shares on housing fell by 2.1 percentage points; expenditure shares on healthcare rose 1.3 percentage points. These are figures aggregated across all households. As only single single-person households are considered in this analysis, and they are differentiated by sex, income, wage-earner status and tax considerations, larger divergences between the two decades begin to appear. Identifying those differences is the goal of this paper. Furthermore, to the extent that expenditure rates have changed, forensic experts may want to use results based on the newer data as they reflect the evolution in purchasing patterns over time.

Table 1CEX Expenditures and Relative Income Shares, All Households, 2009 and 2019

View Fullscreen

III. 2015-2019 Expenditures: Measure of Single-Person Personal Consumption

1. Maintenance

Following Krueger, personal maintenance expenses are measured by the expenses of single persons at a minimal income level. Maintenance expenses are the sum of personal and sharable expenditures made by single persons with incomes between $15,000 and $19,000. Note that the poverty level for single individuals went from $10,830 in 2009 to $12,490 in 2019.⁴ Both are below the $15,000-$19,000 level suggested by Krueger. Not considering income taxes, Table 2 (comparable to Krueger’s Table 7) shows that for wage earners, economic losses to potential survivors exist for all earnings levels over $25,507 for males and $26,038 for females. For non-wage earnings, those figures are $26,968 for males and $28,429 for females. When taxes are considered, those figures increase slightly.⁵

Table 2Personal and Shareable Expenses, 2019$

View Fullscreen

Comparing the 2005-2009 results to the 2015-2019 results, nominal maintenance expenditure levels increased in all categories. This is expected given inflation over the convening decade. As a comparison, the overall CPI is used to adjust 2005-2009 maintenance expenditures to 2015-2019 levels and compared to actual 2015-2019 expenditures. For wage-earning females, inflation-adjusted 2005-2009 maintenance expenditures are about 1% above the actual 2015-2019 expenditure levels, either considering or not considering taxes. For all others, inflation-adjusted 2005-2009 maintenance expenditures are from 6.5% to 15.1% below current expenditure levels. These results suggest why it is useful to re-estimate single-person expenditures using more recent data rather than using the older findings and simply adjusting for inflation.

Deleting the sharable component of expenditures from Table 2, Table 3 (comparable to Krueger’s Table 8) shows just personal consumption expenditures for single persons. Arguably, this could be used as a measure of personal maintenance for a married person. As in Table 2, nominal expenditures have increased by 2015-2019. Inflation adjusted 2005-2009 expenditures are higher than actual 2015-2019 expenditures for wage earners of both sexes and generally lower for non-wage earners. This again suggests why it is useful to re-estimate single-person expenditures using more recent data.

Table 3Personal Expenses, 2019$

View Fullscreen

2. Standard of Living

Under the standard-of-living method, personal consumption rates vary by income level and are measured as the aggregate of personal and sharable expenses. In Table 4 (comparable to Krueger’s Table 9), personal consumption rates under a standard-of-living criterion are shown for income ranges that mirror Krueger’s income ranges. When not considering income taxes, combined personal and sharable expense rates in 2015-2019 are significantly higher in all categories than in 2005-2009. This goes for males and females in all income brackets, whether they are wage earners or non-earners. When considering income taxes, combined personal and sharable expense rates in 2015-2019 are significantly higher in most categories than in 2005-2009. The exceptions are for wage earners in some higher income brackets.

Table 4Personal and Shareable Expenses, Percentage of Total Income

View Fullscreen

By 2015-2019, for wage-earning males, when not considering taxes, combined personal and sharable consumption rates exceed income until the $30,000-$35,000 range. Above that income range, support is available to survivors. For all other males, combined personal and sharable consumption rates exceed income until the $40,000 to $60,000 earnings range. For wage-earning females, personal consumption rates exceed income between $35,000 and $50,000 in income, depending on consideration of taxes. For non-wage-earning females, rates exceed income until beyond $60,000 in income. Beyond those income ranges, potential losses exist for survivors.

It is interesting to compare real results across time periods. To do this, a vector of 2019 incomes is developed, ranging from $20,000 to $150,000 in $2,500 increments. This vector is used to acquire combined personal and shareable consumption rates at each income level from Table 4. This is done separately for the four wage earner categories. As the next step, the vector of 2019 incomes is deflated to 2009 values using the CPI. Then, Krueger’s Table 9 is used to estimate combined personal and sharable consumption rates at each deflated income level. This is the approach a forensic expert might use if newer personal consumption rate data were not available. Finally, the results using the deflated 2009 incomes combined with Krueger’s 2009 consumption rates are compared to the consumption rates using the 2019 incomes combined with the newer consumption rates in Table 4.

The comparative combined personal and sharable consumption rates are shown in Figures 1 through 4. The solid line in each figure shows the consumption rate using 2019 earnings deflated to 2009 dollars combined with the consumption rates in Krueger’s Table 9. The dashed line in each figure is the personal consumption rate using 2019 earnings and the resulting consumption rates in Table 4. Figure 1 shows that when not considering taxes, for wage-earning males, the 2019 consumption rates are frequently higher than the deflated 2009 estimated rates, especially at the highest incomes. Figures 2 through 4 show a mix of results for all others: deflated 2009 rates are generally higher and sometimes lower than 2019 rates. At one end of the divergence, the 2019 estimate is 19.6 percentage points higher than the deflated 2009 estimate; at the other end of the divergence, the 2019 estimate is 8.9 percentage points lower than the deflated 2009 estimate. This again shows the difference between re-estimating single-person expenditures using more recent data versus using the older findings and simply deflating earnings for overall inflation.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

In Table 5 (comparable to Krueger’s Table 10) sharable expenses are removed from Table 4, allowing the decedent the ability to share those expenses with survivors. When sharable expenses are available to survivors, economic losses exist across all income levels. Compared to the 2005-2009 results, when not considering taxes, personal expense rates in 2015-2019 are higher for both wage-earner and non-wage-earner single persons in all income ranges, with the divergence varying across income groupings. When taxes are considered, personal expense rates are higher in 2015-2019 than in 2005-2009 at lower income levels. For all but non-wage-earning females, the reverse is true at higher income levels.

Table 5Personal Expenses, Percentage of Total Income

View Fullscreen

3. Support to Survivors

The support to survivors standard reverses the above logic and measures losses to survivors as the decedent’s earnings multiplied by the percent of decedent’s earnings that would have been used to support others. Table 6 (comparable to Krueger’s Table 11) shows the ratios of support to others compared to total income (not considering income taxes) and combined support to others and sharable expenses relative to income. At all income levels, there is an economic loss to decedents. The 2015-2019 rates are near the 2005-2009 rates, sometimes higher and sometimes lower depending on the income grouping and earner versus non-earner status.

Table 6Support to Others and Support to Others plus Shareable Expenses, Percentage of Total Income

View Fullscreen

4. Estate Accumulations and Support to Survivors

Some states allow for the recovery of lost accumulations to an estate as a result of a wrongful death.⁶ Iowa State Bar Association (2020), Iowa Civil Jury Instructions, for example, allow for the recovery of “the present value of the additional amounts (decedent) would reasonably be expected to have accumulated as a result of [his] [her] own effort if [he] [she] had lived out the term of [his] [her] natural life.”⁷ We explore an alternative approach toward estimating accumulations to an estate than the approach taken by Krueger.

Krueger estimated the accumulations as “income less total CEX expenditures” (Krueger, 2011, p. 158). Krueger included what the CEX classifies as Pension and Social Security expenditures in CEX expenditures. Hence, Krueger is excluding retirement payroll deductions and Social Security contributions from estate accumulations. The CEX Pension and Social Security category includes the following UCCs: 1) government retirement deductions, 2) railroad retirement deductions, 3) private pension fund deductions, 4) amounts placed into individual retirement plans, and 5) Social Security contributions. Items 1 through 4 can be considered potential accumulations to an estate. Consequently, we adjust Krueger’s measure of estate accumulations by adding back in government retirement deductions, railroad retirement deductions, private pension fund deductions, and amounts placed into individual retirement plans. We continue to exclude Social Security contributions.

Table 7 (similar to Krueger’s Table 12) shows accumulation rates when retirement payroll deductions are considered part of estate accumulations. For wage-earning males, accumulations begin in the $35,000 to $45,000 income range depending on whether income taxes are considered. For wage-earning females, accumulations begin in the $35,000 to $40,000 range not considering taxes, but not until at least $60,000 when accounting for taxes. For non-wage earners, males start to accumulate in the $40,000 to $60,000 range when taxes are not considered and in the $60,000 to $100,000 range when taxes are considered. Only relatively high-income female non-wage earners accumulate, and this only occurs when taxes are not considered.

Comparing the 2015-2019 results to the 2005-2009 results, even with the inclusion of retirement payroll deductions, net accumulation rates are lower in 2015-2019 than in 2005-2009 at income levels below $70,000. This holds regardless of sex, earner/non-earner status or tax considerations. As income goes beyond $70,000, the comparative results are mixed. The newer accumulation rates are higher for wage earning females between $70,000-$79,999 not considering taxes, wage earning males considering taxes, earning over $70,000 and wage-earning females considering taxes from $60,000 to $80,000 and over $150,000.

How a forensic expert handles accumulations is an interesting question. Iowa’s legal decision, for example, provides little guidance on how to interpret its jury instruction. One approach would be to estimate accumulations over a decedent’s working life and then estimate dis-accumulations—while accounting for Social Security and other payment sources—in retirement. Using this approach, it would be meaningful to use the accumulation definition discussed above and represented in Table 7.

Table 7Net Estate Accumulations, Percentage of Total Income

View Fullscreen

In Table 8, the decedent’s sharable expenses and support to others are added to positive values of estate accumulation to provide an alternative measure of potential support to survivors as shown in Krueger’s Table 13. Not considering income taxes, for wage earners, the 2015-2019 rates are higher than the 2005-2009 rates at very low levels of earnings, and dip below the 2005-2009 rates as earnings increase. The same effect occurs for non-wage earners, but the dip occurs at somewhat higher income levels. When income taxes are considered, the 2015-2019 rates are always higher than the 2005-2009 rates.

Table 8Shareable Expenses and Support to Survivors and Positive Net Estate Accumulations, Percentage of Total Income

View Fullscreen

IV. Regression Analysis

Table 9 shows regression results using the information in Tables 4 through 6. In each, the natural log of the relevant rate is regressed on a constant and the natural log of income. The regression results allow straightforward estimation of rates given any income. The first section of Table 9 contains regression results for combined personal and sharable expense rates (from Table 4). Before tax and after-tax results for wage-earning and non-wage-earning males and females have high R²s. All coefficients are statistically significant at the 99% confidence level and are of the expected sign: rates decrease as income increases. The same qualitative results are held in the second section of the table where just personal expense rates are considered (from Table 5). Regressions have high R²s, coefficients are statistically significant and rates decline in income.

Table 9Regression Results

View Fullscreen

Results in the third section of Table 9 are in contrast to the first two sets of results. This section relates to support to others. As suggested in the left-hand side of Table 6, for wage-earning single persons, support to other rates are fairly flat, ranging between 4.6% and 7.9% of income for males and 4.8% to 13.9% for females with no discernable pattern. For non-wage earners, absent males in the lowest income level, rates fall and then rise. This explains the poor logarithmic results. The last section in Table 9 contains results when support to others is combined with sharable expenses (the right-hand side of Table 6). Once again, regressions have high R²s, coefficients are statistically significant and rates decline in income. The regression results in Table 6 should be useful to forensic experts wishing to estimate relevant expenditure rates at any income level.

V. Conclusion

Using 2005-2009 CEX data, Krueger (2011) estimated various measures of personal consumption for single persons. Since that time, the CEX has undergone a number of changes, inflation rates have varied across broad Consumer Price Index components, technology has evolved and consumers’ tastes have changed. These factors, and more, have potentially altered consumers’ mixes of goods and services. We re-estimate Krueger’s measures using data from a decade later. The estimation of estate accumulation rates has expanded as well.

Personal maintenance expenses increased over the decade, typically, more so than overall inflation. Moving on to support to survivor measures, combined personal and sharable expense rates are almost always higher in 2015-2019 than in 2005-2009. Expenditure rates also differ from those acquired by deflating 2019 earnings to 2009 values and using the associated 2005-2009 rates. Losses exist for survivors once income climb above the $30,000 to $45,000 range.

When personal expenses are separated from sharable expenses so that sharable expenses are available to survivors, economic losses exist across all income levels. Personal expense rates are higher in 2015-2019 than in 2005-2009 for male and female wage earners and non-wage earners alike, when not considering income taxes. When taxes are considered, 2015-2019 rates are higher than 2005-2009 rates for lower income wage earners. The reverse is true of higher income wage earners. The 2015-2019 rates are almost always higher than 2005-2009 for non-wage earners of either sex.

The support to survivors standard measures losses to survivors as the decedent’s earnings multiplied by the percent of decedent’s earnings that would have been used to support others. At all income levels, there is an economic loss to decedents. The 2015-2019 rates are near the 2005-2009 rates, sometimes higher and sometimes lower depending on the income grouping and earner versus non-earner status.

The current work expands on Krueger’s measure of estate accumulation to include pension contributions. Even when estate accumulation rates are estimated in this broader manner, when not considering taxes, newer accumulation rates are almost always lower than Krueger’s accumulation rates. When considering taxes, for lower income households, those rates in 2015-2019 are lower than accumulation rates using the 2005-2009 CEX. For higher income households, 2015-2019 accumulation rates are generally higher than 2005-2009 rates for wage earners only.

Regression results allow for a more continuous estimation of single persons combined personal and sharable expense rates, personal consumption rates, support to others, and combined support to others and sharable expense rates. All but the support to others regressions have high R² and statistically significant coefficients. This should allow estimation of single-person consumption rates at most income levels.

Until now, forensic experts wanting to estimate personal consumption rates for single persons typically had to take current earnings, deflate them to 2009 dollars and apply Kreuger’s personal consumption rates. Our results indicate that doing so can produce results significantly different than using newer 2015-2019 personal consumption data. Experts may want to consider incorporating this newer data into their economic loss analyses.