Briefing Paper January 2006

New Estimates Confirm that Three-quarters of
a Million Pennsylvania Workers Would See Higher Pay With a $7.15 Per Hour Minimum Wage

Technical Appendix

As well as updating KRC’s own estimates of the impact of a higher minimum wage, this Briefing Paper seeks to clear up confusion that exists among lawmakers, members of the media, and the public because of the many different numbers that have been cited in the minimum wage debate.

Some of this confusion may result from the politics surrounding the issue. For example, opponents of a higher minimum wage sometimes inject into the discussion very low numbers that may leave lay audiences thinking the minimum wage is not very important. One example is the Commonwealth Foundation (CF) statement that “less than 1 percent of all American workers earned the minimum wage” in 2004 (see “Minimum Wage Facts,” online at www.commonwealthfoundation.org). While factually accurate, citing the number of workers who earn exactly the current minimum wage says nothing about the impact of an increase in the minimum wage of $1 or $2. The National Federation of Independent Business (NFIB) and Chamber of Business and Industry are also fond of citing the small number of workers who earn at or below the current minimum wage of $5.15. When the NFIB and Chamber do this, they do not always share with their audience that large numbers of workers earn in the wage ranges above $5.15 that would be affected by a minimum wage increase to $7.15.

Even if some confusion is deliberately created, however, when it comes to the simplest estimate of the effect of a minimum wage increase to $7.15 — i.e. the number of workers earning $5.15 to $7.14 — there is less incompatibility between published estimates than widely believed. The remainder of this Technical Appendix looks in detail at the estimates made by the Keystone Research Center (KRC), CF, and the Rendell Administration.

KRC’s Methodology

The primary source for all of the published estimates of the number of workers affected by a minimum wage increase is the Current Population Survey (CPS), a monthly household survey conducted by the Census Bureau and analyzed regularly by the Bureau of Labor Statistics (BLS). The CPS is the main source relied on by economists to evaluate wage trends at the national and state levels.

In previous estimates of the number of workers at the state and county level who would benefit from a minimum wage increase, KRC used the 2004 CPS, at that time the most recent full year of CPS data available. With the 2004 CPS, we estimated the number of Pennsylvania workers who would benefit directly from a minimum wage increase as the sum of two groups.

1. The number of workers reporting hourly earnings between $5.15 and $7.14 — 453,000.

2. The number of workers reporting weekly earnings whose “imputed” hourly earnings (weekly earnings divided by usual hours worked) were between $5.15 and $7.14 — 57,000.

Thus a total of 510,000 workers — 53,000 plus 57,000 — were estimated to benefit directly.

We also calculated from the 2004 CPS the number of workers earning $7.15 to $8.14 — 350,000. Even though employers paying in this range are not legally required to give their workers a raise, research shows these employers likely would give their workers a “ripple effect” increase to keep them above the new minimum wage. This increase helps these employers attract and retain quality workers. That a ripple effect exists and that workers above $7.15 would enjoy some increase have been acknowledged by opponents of a higher minimum wage. According to The Patriot-News, “opponents [of a minimum wage increase] say the proposed hike wouldn’t simply raise wages for. . . workers who . . . make less than $7.15. It could also force businesses that pay $8 to $9 an hour to adjust their pay scales.”

To sum up, KRC previously estimated the total number of workers directly and indirectly affected by a minimum wage hike to $7.15 as 510,000 plus 350,000 — 860,000 workers in all.

The new estimates in this paper of the number of workers who would benefit from a minimum wage increase differ from our previous ones for two reasons.

1. To the extent possible, we rely on data for 2005.

• BLS recently released the 2005 number of Pennsylvania workers reporting hourly earnings between $7.15 and $8.14 — 423,000, down from 453,000 in 2004.
• The number of workers reporting weekly earnings but whose imputed hourly earnings are $5.15 to $7.14 is not yet available. Earlier years’ data, however, reveal that the size of this group compared to that of workers directly reporting hourly earnings of $5.15 to $7.14 is stable over time: the group with imputed earnings from $5.15 to $7.14 is a little over 12 percent of the group reporting hourly earnings in this range. We used this fact to estimate the imputed hourly group in 2005 as 52,000, down from 57,000 in 2004.
• We continue to use 2004 data as the initial data source for the number of workers earning $7.15 to $8.14. (BLS has not yet published data for this wage range for 2005.)

2. Since the increase to $7.15 would only go into effect on January 1, 2007, we estimate how much change there will be in the size of our directly and indirectly affected groups by then. This change results from two factors.

a. Nominal wage changes. To the extent that employers adjust wages for inflation, or give workers increases over and above inflation, more workers will move above our $7.14 and $8.14 thresholds, reducing the number of beneficiaries from a minimum wage increase.

b. Changes in total employment. If total employment increases, the number of workers earning $5.15 to $8.14 can also be expected to increase.

Since both these factors have been operating over the past several years, we assume that their effect between now and January of 2007 will be the same as it was over the period 2001 to 2005. From 2001 to 2005, the number of workers earning between $5.15 and $7.14 declined 6.6 percent per year. We assume it will do the same in 2006 and that it will decline a further 3.3 percent by January 2007, bringing our total for the number of directly affected workers to 427,000 by January 1, 2007.

For workers earning $7.15 to $8.14, the annual rate of change from 2001 to 2004 (recall that we do not have data for this group for 2005) was -2.6 percent. Applying this percentage to 2005 and 2006 and applying -1.3 percent for the half year to January 1, 2007, the number of workers earning $7.15 to $8.15 is estimated to be 327,000 on January 1, 2007.

Table 1 summarizes KRC’s prior and updated estimates.

KRC’s and Other Estimates Compared

Tables 2 and 3 compare KRC’s methods and numbers with those of the Commonwealth Foundation, developed by economist David McPherson, and the Pennsylvania Department of Labor and Industry.10

In the range of $5.15 to $7.14, there are two primary reasons that CF, PDL&I, and KRC generate different estimates of the number of workers (Table 2).

1. Workers who report a weekly not an hourly wage are included in the CF and KRC sample but not in the PDL&I one. (Both CF and KRC impute the hourly wage for this group by dividing weekly wages by usual weekly hours.) In the KRC sample, the group with imputed hourly wages is only 12 percent of the much larger group of workers who report an hourly wage. Therefore, its inclusion or exclusion makes only a relatively small difference in estimates.

2. Second, the three groups use different methodologies to project changes in the number of workers in the $5.15 to $7.14 range from the time of data collection (2004 or 2005) to January 1, 2007. PDL&I does not, in fact, project these changes, relying instead on 2005 numbers. CF and KRC use different methods to project how much the numbers will change by January 1, 2007.

A further minor difference between these estimates exists in the time period of the data used. CF uses a sample from the second half of 2004 and the first half of 2005. KRC and PDL&I rely partly or completely on 2005 data.

Above the wage range ending at $7.14, there is another difference between these groups’ estimates: only KRC estimates the number of indirectly affected workers who will earn $7.15 to $8.14 on January 1, 2007.

Table 3, with the actual numbers, shows that the KRC and PDL&I estimates for directly affected workers are similar. The similar final number results because the differences in methodology between the two estimates offset one another.

• KRC’s inclusion of workers with imputed hourly earnings increases its estimate relative to PDL&I’s.
• KRC’s projection of the reduction from 2005 to January 1, 2007 in the number of workers earning $5.15 to $7.14 lowers its estimate relative to PDL&I’s.

Table 3 also shows that the CF number is much lower than either the KRC or PDL&I number. This difference is the result of CF’s projection methodology, which assumes 2.5 percent wage inflation from 2004/05 until January 1, 2007. There are two flaws with this methodology.

• First, CF does not take account of job growth that will increase the number of affected workers.
• Second, the CF method is plagued by the tendency of respondents to household surveys to report their wages in whole numbers (e.g., $6, $7, $8 and so on). As a result of this tendency, some 124,000 Pennsylvania workers are estimated to earn almost exactly $7 in CF’s original data (for the period 7/2004 to 6/2005). Inflating a $7 wage by 2.5 percent for each of two years brings these workers to $7.35 per hour by January 1, 2007, moving them out of the McPherson/CF directly affected range.11 In practice, however, these workers may actually earn anywhere from approximately $6.75 to $7.25 to start with, and some of them will still earn less than $7.15 by 2007. KRC’s alternative method of projecting the number of workers in the $5.15 to $7.14 range avoids difficulties due to spikes in the wage data at whole numbers.

While the flaws in McPherson’s method lead him to overestimate the decline in directly affected workers over time, he deserves credit for explicitly modeling the change in the number of such workers between the time when his data was collected and January 1, 2007. KRC’s original estimate (of the number of directly affected workers) was too high because it did not model this change. The new KRC estimate, which models this change in a way this is likely to prove more accurate than McPherson’s, is between the original KRC 510,000 and the CF 320,000, although slightly closer to the former.

This document is an on-line summary of a Keystone Research Center report. The entire report is available for download as a PDF file at the KRC Web site www.keystoneresearch.org © 2001 Keystone Research Center