Printer Friendly

Workers should want to pay more for Social Security.


Although economic theory cannot resolve all the controversies concerning Social Security, simple supply and demand analysis can helpfully change the terms of the debate concerned with how the Social Security tax is divided between employers and workers. Defenders of the existing system typically argue that workers pay only half the tax (6.2%) because, as required by law, the other half is paid by employers. Opponents of the existing system typically argue that workers pay all of the Social Security tax because their compensation is reduced by employers to offset their portion of the tax.

For example, former House Majority Leader Dick Armey (2005) states that "Social Security alone takes 12.4 percent of workers' incomes. That's more than most workers pay in income taxes." In a reply to Armey, Marilyn Scheiner (2005) states that Armey should have said that "Social Security alone takes 6.2 percent of workers' income. The employer matches the employee's contribution for a total of 12.4 percent going into the Social Security coffers." (1) The first thing economic analysis tells us about this debate is that both sides are wrong on the facts, barring unlikely conditions. The second and more interesting insight is that each side of the debate is making a claim that supports the position of the other side.

Any student in an economic principles course should know that the distribution of a payroll tax burden depends on the relative elasticities of the labor demand and supply curves. If the labor supply curve is very inelastic relative to the labor demand curve, then most of the burden of a payroll tax will fall on the workers. On the other hand, if the labor demand curve is very inelastic relative to the labor supply curve, most of the burden will fall on the employer. So the debate over who bears most of the cost of the Social Security tax has to consider the relative elasticities of labor supply and demand. What the Social Security legislation states about the distribution of the tax burden is completely irrelevant. (2) This may seem to suggest that the opponents (supporters) of Social Security should argue that labor supply is less (more) elastic than labor demand, because this implies that less (most) of the tax burden is borne by the employer.

But why should the desirability of Social Security depend on the distribution of the tax burden? Supposedly there are benefits from Social Security, so its desirability should depend on how those benefits compare to the tax and how the net benefits are distributed. Even though labor supply and demand elasticities are relevant to the distribution of gains or losses from Social Security, we shall see that the desirability of Social Security (as measured by its net value and its benefits to workers) is completely independent of those elasticities. Once we take into consideration the benefits of Social Security, we find that when advocates and opponents argue over the incidence of the tax, they have their arguments completely backward. The strongest argument in favor of Social Security is that workers are paying more than its cost. On the other hand, the strongest argument against Social Security is that the burden of financing it falls mostly on employers. If employers are paying most of the cost of Social Security, then, for the sake of workers, it should be scrapped immediately.


Our argument is based on straightforward implications of labor supply and demand curves shifts in response to the costs and benefits of Social Security. (3) We begin in Figure 1 with labor supply curve S and labor demand curve D, assumed relevant without (before) Social Security. The vertical axis represents the monetary wage and the horizontal axis represents the quantity of labor. The equilibrium wage and labor quantity are given by [W.sub.B] and [L.sub.B], respectively.


We next introduce Social Security and consider the shifts in the demand and supply curves under the assumption that the employer has the responsibility to pay directly the entire tax. As indicated, this assumption is incorrect but has no effect on the actual incidence of the tax. (4) The relative elasticities of the curves do affect the incidence of the tax but as we shall see, not the efficiency of the Social Security program. As we have constructed the curves, the labor supply curve is more elastic than the labor demand curve. (5) This construction allows us to better highlight our main point, which is that the larger the proportion of the Social Security tax that falls on the employer, the stronger the case against Social Security. It will also allow us to show directly that supporters of the Social Security program should not want the labor demand curve to be inelastic relative to the labor supply curve, assuming they want workers to benefit as much as possible from Social Security and really believe the program creates positive net benefits. (6)

We consider four cases with different assumptions on the present value of the expected retirement payments workers are earning with the current tax payments. In case 1, the present value of the expected retirement payments is 0. In case 2, the present value of the expected retirement payments equals one-half the total taxes being paid. In case 3, the present value of the expected retirement payments equals the total taxes paid. Finally, in case 4, the present value of the expected retirement payments equals 150% of the total taxes paid. (7)

Case 1

In this case the only effect on the demand and supply curves comes from the tax burden, because there are no offsetting benefits from the value of Social Security payments. With the direct payments of the employer covering the entire tax, the demand curve for labor shifts down by the total tax per worker--the marginal value of workers is now worth less to the employer at every employment level by an amount equal to the Social Security tax paid for each worker. The marginal cost of working is unaffected by the direct tax and so the labor supply curve doesn't shift. In Figure 1 the new demand curve is shown as [D.sub.1] with the supply curve remaining at S, and the new equilibrium wage and labor quantity are shown as [W.sub.1] and [L.sub.1], respectively. As is easily seen, most of the cost falls on the employer in this case. The direct tax per worker is given by the distance ac, which equals the vertical drop in the demand curve. The employer is able to offset this tax by an amount equal to be in Figure 1 with a lower wage. So the cost to the employer is ab per worker and the cost to each worker is bc. If having the employer pay most of the tax is seen as an argument in favor of Social Security, at least from the workers' perspective, then this case would be very favorable to workers. But obviously workers lose in this case (getting nothing in return for a sacrifice of bc in wages), though they are not losing as much as their employer.

Case 2

In this case, the labor demand curve remains at [D.sub.1] (as it does in all the remaining cases), being affected only by the direct Social Security tax the employer pays per worker. However, the labor supply curve shifts down to [S.sub.2] from S in Figure 1, with the vertical distance between the two supply curves given by the present value of the Social Security payments the workers are gaining, which is assumed equal to one-half the per worker tax. In this case, the introduction of Social Security reduces the equilibrium wage from [W.sub.B] to [W.sub.2] and the equilibrium labor quantity from [L.sub.B] to [L.sub.2]. The total tax per employee paid by the employer is given by the distance dg (the same as distance ac). But the employer's payment is now offset by a larger reduction in the wage, given by the distance eg, than in case 1. This wage reduction is slightly larger (by an amount given by ef in Figure 1) than the present value of the retirement payments.

Workers are better off in this case, because their loss declined to ef from bc. And they are still doing better than the employer, who is losing an amount equal to de per worker (the per worker tax, dg, minus the reduction in wage, eg). But it is the employer who has benefited the most from the improvement in the return on Social Security. Instead of losing almost all of the tax per worker, he or she is now losing a little less than one-half of it. Based on the argument that the larger the tax burden Social Security imposes on the employer relative to workers, the stronger the case for it, one would erroneously conclude that Social Security is a better program in case 1 than in case 2.

Case 3

In this case, the present value of the retirement payments being earned for workers equals the current tax payments being paid by the employer. This means that the labor supply curve, [S.sub.3], lies below S in Figure 1 by exactly the same vertical distance that the labor demand curve lies below D in each of our four cases. Each worker is willing to work for a money wage that is lower by the same amount the employer is paying in per worker Social Security taxes, which results in an equilibrium wage of [W.sub.3] and an unchanged equilibrium labor quantity of [L.sub.B]. In this case, Social Security has no effect on the well-being of either the employer or the workers. The wage declines by the full amount of the employer's tax payment, so the incidence of the tax on the employer is 0, and workers end up paying the entire tax with the lower wage, just as some opponents of Social Security claim. But of the cases we have considered so far, this is the one in which the Social Security program is the most efficient and the best for the workers and for the employer.

Yet if Social Security is seen as more desirable the higher the ratio of the employer's burden to the workers' burden, then this is the worst case so far because this ratio is lower here than in any previous case. The employer has gone from losing almost all his or her direct tax payment (in case 1), to losing a little less than half of it (in case 2), to losing nothing--a large gain. In contrast, the workers have gone from a small loss (in case 1), to an even smaller loss (case 2) to no loss at all--a small gain. It is also worth noting that when a tax comes with a benefit of equal value, the net benefits are the same (0) for both employer and workers regardless of the relative elasticities of the labor supply and demand curves. But as will be explained in section III, the greater improvement experienced by employers relative to their workers, is the result of differences in the relative elasticities of labor demand and supply.

Case 4

In our final case we assume that Social Security yields positive returns, yielding a present value in retirement payments equal to 150% of the taxes paid. As before, the demand curve shifts from D to [D.sub.1]. The labor supply curve shifts down by one and a half times the demand curve's downward shift, or from S to [S.sub.4]. The equilibrium wage is shown in Figure 1 by [W.sub.4], and the equilibrium labor quantity is given by [L.sub.4]. (8) As is easily seen, both the workers and the employer are now both benefiting from a Social Security program yielding a positive return for the first time. But the big winner is the employer, who pays a direct tax given by distance jk but reduces the wage by an amount given by the distance ik--to [W.sub.4] from [W.sub.B]. Therefore, the employer realizes a total gain per worker given by the distance ij. The benefit to the workers is much smaller, with each gaining a retirement benefit with a present value given by the distance hk but having their salary reduced by and amount given by ik, for a net benefit of hi. (9)

This is the only case considered in which Social Security is good for workers, but it is also the case in which the gain the employer receives relative to that which the workers receive is the greatest. Not only is the employer not bearing any burden from paying the Social Security tax, he or she is making a profit from the program that is significantly larger than the net benefits received by the workers. The employer shifts a cost to the workers in lower wages almost 50% more than the Social Security taxes he or she pays. This leaves the workers with very little of the 50% return Social Security provides.

If all Armey and Scheiner knew about the Social Security program was that workers were paying in lower wages almost 50% more than the tax, Armey would see this as reinforcing his criticism of the program and Scheiner would be appalled with the program's unfairness and probably call for legislation to increase the tax on employers. If Armey and Scheiner had subjected the effects of Social Security to a little supply and demand analysis, they would have concluded that the wage reduction in excess of taxes is a clear indication that the benefits workers realize from Social Security are greater than its cost to them.


The inelasticity of the labor demand curve relative to the labor supply curve turns out to be a double-edged sword. With only a tax, if the employer's (demander's) demand curve is inelastic relative to the workers' (suppliers') supply curve, then the employer shoulders most of the tax burden, as shown in case 1. But when the tax comes with a benefit paid to the workers, as the value of the benefit increases, the inelasticity of the demand curve starts favoring the employer. Downward shifts in the supply curve in response to benefit increases cause a greater decrease in the wage paid by the employer, the more inelastic the labor demand curve. This allows the employer to capture most of the benefit, even though it supposedly goes entirely to the workers. Just as the incidence of the tax is independent of who pays it directly, the incidence of the benefit financed by the tax is independent of who receives it directly.

On the other hand, it is easily shown that if the labor demand curve had been constructed to be far more elastic than the labor supply curve, then case 1 would have found the workers paying almost all of the Social Security tax in the form of a lower wage. But given these relative elasticities, as the net value of Social Security increased, the wage would decline very little, allowing the workers to capture most of the gain. When the net value of Social Security is 0 (case 3), then the welfare of both the employer and the workers are left unaffected by Social Security. Recall, in this case the division of the tax and the benefits is independent of the relative elasticities of the labor demand and supply curves. In case 4, however, where Social Security yields a positive return, workers would capture most of that return, with the decline in wages being just slightly greater than the direct tax, which, without loss of generality, is being paid entirely by the employer in our analysis.

This explains our statement near the beginning of section II that those who believe that there are net benefits from Social Security and want workers to realize most of those benefits should not want the labor demand curve to be inelastic relative to the labor supply curve. The more elastic the demand for labor relative to the elasticity of the supply of labor, the greater the proportion of any net benefits resulting from Social Security that goes to workers.


The relative magnitudes of the net costs (in cases 1 and 2) and the net benefits (case 4) of Social Security going to the employer and the workers depend on the relative elasticities of the labor demand and supply curves. As we have constructed the demand and supply curves in Figures 1, the employer is the biggest loser when the Social Security program is all costs and no benefits, but he or she becomes better off relative to the workers as the net benefits from the program become positive. However, regardless of the relative elasticities of the labor demand and supply curves, neither the employer nor the workers realize any net benefit from Social Security until the present value of the retirement payments is greater than the present value of the tax payments, which occurs only in our case 4. No matter what the relative elasticities of labor supply and demand, instead of picking up any (much less half) of the cost of Social Security, if the system generates positive net benefits, the employer will actually make a profit from it. In the situation we have considered, the employer captures most of the net benefits, leaving the workers very little. In addition, the division of the net benefits from Social Security is unaffected by statutory division of the direct taxes between the employer and the workers or by the fact that all the retirement payments are sent directly to the workers.

The implication of the analysis is clear. Workers should hope that their employer is profiting from Social Security by reducing wages by more than the employer's direct tax. If workers are paying less in reduced wages than the direct tax, no matter how that direct tax is divided by statute, they are paying more for Social Security than they are receiving in return. Indeed, no matter what the relative labor demand and supply elasticities, the more the employer profits from the Social Security program by reducing wages, the more net value workers realize from the program. The last thing workers should want is a Social Security program that imposes positive net costs on their employer, as it does in cases 1 and 2. And the last thing Social Security advocates should be claiming is that employers pay half the tax because they do not lower workers' wages. If the Social Security program does not result in lower wages for workers, it is because the payoff they receive from the program is sufficiently negative to shift up the labor supply curve by enough to offset the negative wage effect caused by the downward shift in the labor demand curve.


Armey, D. "Letter to the Editor." Washington Post, May 31, 2005, p. A16.

Borjas, G.J. Labor Economics, 3rd ed. New York: McGraw-Hill Irwin, 2005.

Entin, S.J. "Tax Incidence, Tax Burden, and Tax Shifting: Who Really Pays the Tax?" Center for Data Analysis Report, #04-12, November 5, 2004; available online at

Hamermesh, D.S. Labor Demand. Princeton, NJ: Princeton University Press, 1993.

Prescott, E.C. "Why Do Americans Work So Much More than Europeans?" Federal Reserve Bank of Minneapolis Quarterly Review, 28(1), 2004, 2-13.

Scheiner, M. "Letter to the Editor." Washington Post, June 4, 2005, p. A16.

Steuerle, C.E., A. Carasso, and L. Cohen. "How Progressive Is Social Security When Old Age and Disability Insurance Are Treated as a Whole?" Straight Talk on Social Security and Retirement Policy, No. 38, Urban Institute, May 1, 2004; available online at

Summers, L.H. "Some Simple Economics of Mandated Benefits." American Economic Review, 79(2), 1989, 177-83.

Toner, Robin, and M. Connelly. "Poll Finds Broad Pessimism on Social Security Payments." New York Times, June 19, 2005, p. 18.

Triest, R. "The Effect of Income Taxation on Labor Supply in the United States." Journal of Human Resources, 25(3), 1990, 491-516.

Ziliak, J.P., and T.J. Kniesner. "Estimating Life Cycle Labor Supply Tax Effects." Journal of Political Economy, 107(2), 1999, 326-59.

(1.) Entin (2004) provides another example of arguing against Social Security by claiming that workers pay the entire tax. According to Entin, "The entire Social Security payroll tax on wages is remitted by employers to the Treasury, but according to statute, it supposedly is paid half by employees and half by employers (statutory obligation). Most economists would argue that, legislative language notwithstanding, the initial incidence and the ultimate economic burden of the entire tax is borne by workers."

(2.) For a straightforward analysis of the incidence of a payroll tax see Borjas (2005, pp. 170-74).

(3.) Our analysis is similar to that contained in Summers (1989). Summers is concerned with the relative efficiency of government mandating that employers provide certain benefits to workers compared to providing those benefits publicly and financing them through taxes.

(4.) The reader can easily establish this result by considering shifts in the demand and supply curves associated with workers and employers each paying directly half the tax or any other proportion summing to 1.

(5.) Estimates on the elasticity of the labor supply vary, but most are quite small. For example, Triest (1990) finds that men have a wage elasticity of supply close to 0, whereas the elasticity of labor supply for women falls into the range 0.1 to 0.2. Ziliak and Kniesner (1999) find supply elasticities for men ranging from 0.14 to 0.20, with the elasticity increasing with wealth. On the other hand, Prescott (2004, p. 11) estimates that the elasticity of labor supply is "nearly 3 when the fraction of [labor] time allocated to the market is in the neighborhood of the current U.S. level." Estimates on labor demand elasticities typically run a little higher (in absolute value) than labor supply elasticities. See Hamermesh (1993, table 3.2, pp. 82-86) for an extensive list of estimated labor demand elasticities.

(6.) In section III we discuss how the relative benefits to workers and employers from Social Security would change if we assumed that the labor demand elasticity is greater than the labor supply elasticity.

(7.) In cases 2, 3, and 4 we make the assumption that each worker earns the same amount (and less than the maximum income on which the Social Security tax is applied), and each expects to receive the same present value of retirement benefits. These assumptions simplify the analysis by keeping the marginal benefits and costs from Social Security the same for all workers. This simplification does not affect our general conclusion in most labor markets. It should be acknowledged, however, that Social Security provides a lump-sum benefit and imposes a lumpsum cost for those workers who earn more than the taxable maximum, and therefore our analysis does not apply for these workers or for labor markets dominated by them. With regard to expectations on retirement benefits, they are probably best represented as somewhere between cases 1 and 2, though probably closer to 2. Based on a New York Times/CBS News Poll, Toner and Connelly (2005) report that 94% of American adults believe that Social Security is in trouble (in crisis, 20%; serious trouble, 40%; and in some trouble, 34%). Only 3% believe Social Security is not in trouble, with another 3% having no opinion. The sampling error was plus or minus 4 percentage points. How accurate expectations are regarding Social Security benefits is subject to future policy changes that are difficult to predict. Based on plausible projections of the current system Steuerle et al. (2004) estimated that for men born between 1956 and 1964, the internal rate of return on the main Social Security program (Old Age and Survivors Insurance) would be 2.0%. For those born since 1964, the return will surely be lower, and likely negative for those just entering the workforce, but estimates on these returns are highly conjectural.

(8.) We have obviously distorted our diagrams by showing little of the monetary wage to better highlight that part of the wage influenced by Social Security taxes and benefits.

(9.) For convenience, we have been referencing the losses and gains from Social Security to each worker hired. This slightly understates the losses to the employer and to all workers in cases 1 and 2 and slightly understates the gains in case 4. In cases 1 and 2 the employer is slightly worse off than we have indicated because a few workers, on whom a small profit was being made before Social Security, are laid off. And the workers laid off are also slightly worse off because they were making a little more than their opportunity cost. In case 4, the gain to the employer is slightly understated because of the additional profit made on the extra workers hired, and those workers are slightly better off because they are making a little more than their opportunity cost.

Clark: Professor of Economics and Probasco Chair of Free Enterprise, University of Tennessee at Chattanooga, 313 Fletcher Hall, Dept. 6106, 615 McCallie Avenue, Chattanooga, TN 37403-2598. Phone 1-423-425-4118, Fax 1-423-425-5218, E-mail j-clark@utc. edu

Lee: Professor of Economics and Ramsey Chair of Private Enterprise, University of Georgia, Terry College of Business, Dept. of Economics, Brooks Hall 509, Athens, GA 30602. Phone 1-706-542-3970, Fax 1-706-542-3376, E-mail
COPYRIGHT 2006 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2006 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Clark, J.R.; Lee, Dwight R.
Publication:Economic Inquiry
Geographic Code:1USA
Date:Oct 1, 2006
Previous Article:One lump or two: unitary versus bifurcated measures of injury at the USITC.
Next Article:Reading, writing, and sex: the effect Of losing virginity on academic performance.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters