atively invisible effects are reckoned, die overall impact on the quality of padent care becomes obscure. In any case, raising the average quality of patient care does not mean that consumers are necessarily better off. Such quality comes al a cost. We will have less of other good things. After all, would we really be better off if the government banned the sale of economy cars?

Although it is often argued that freeing-up medical educauon would deluge the market with incompetent hacks, such arguments carry more force with others than they do with economists. Presumably patients would seek to identify those doctors who are well-trained and competent Since such information is highly valued, it would doubdess pay someone to provide it somehow or another. In other words, private orgaruzations would find a way of certifying competence. Hospital associations are one of several possible means of doing so. After all, we can correcdy judge the relative research achievements of professors by their University association. Would hospitals do a less capable job at identifying competent doctors? Alternatively, doctors might be privately certified by insurance organizations similar to Underwriters Laboratories. Such companies would carefully screen doctors because mistakes would be bom by them in the form of insurance losses. Whatever the method that would emerge, powerful incentives would exist to screen doctors efhciently. Inefficient methods rarely survive marketplace competition.

BILATERAL MONOPOLY

Bilateral monopoly is said to exist in a labor market when a monopolist labor seller, the union, faces a monopsonist labor buyer, thereby causing the labor market to be noncompetitive on both the demand and supply sides. In this case a collectively bargained wage may or may not create unemployment, depending upon the size of the wage demand.

Consider in Figure XV the formation of a union in a monopsonistic labor market in which the initial equilibrium is at Njn and Wm- Unionization makes it possible to raise both the wage and the employment level above these amounts. For example, we will show that a union by achieving the collectively bargained wage Wu, can enjoy employment of Nu • We start by deriring the marginal factor cost of labor, which is here composed of two parts. Notice from the marginal/average mles that at a union wage of Wu , the marginal factor cost of labor curve is now horizontal at Wu out to Nu. On this stretch, the employer can hire any amount of labor up to Nu at the constant

price of Wl, , i. ., at a constant average factor cost of W. To increase employment further, however, the employer mu.st hire labor at a wage given by the rising supply (AFCj) curve. Along this rising ponion, the MFCi curve rises above the supply curve. Hence, the MFC] curve has a discontinuit) at Ny ; it is horizontal up to N„ and then follows the original MFC] curve above Ny.

Faced with diis situation, how many workers should the monopsonist hire? As weve seen, the monopsonist will continue hiring workers until the marginal revenue product of labor equals the marginal factor cost, i.e., until MRPj = MFCj. In Figure XV the MRP] exceeds the MFC] out to Nu workers hired; hence the firm would want to hire Nu workers. It would not want to go beyond Nu however, because at higher levels of employment, the marginal factor cost of labor exceeds the marginal revenue product Hence the emplojTneni of Nu workers, where die discontinuity arises in the MFCj curve, is as close as the firm can get to equating marginal factor cost and marginal revenue product. Ni, is therefore the employment equilibrium and the wage rate is Wu- Four important points should be noted about this equilibrium. First, no unemployment occurs; at Wu the quan-1 tity supplied and demanded are identical at . Second, Wu exceeds the pure monopsony wage of W. Third, the equilibrium employment of Nu also exceeds ] the equilibrium employment under pure monopsony, Nm- Fourth, any collectively bargained wrage lying! below the competitive one of Wc, where demand andj supply cross, will likewise raise wages without decrease ing employment.

What is going on here intuitively? Rememt that die pure monopsonist restricts because hiring additional workers drives the margina factor cost of labor (MFCj) above the wage rate, Wj But so long as the union can present him with a labor supply (- ] ) curve at Wu, the MFC] curve identical to it. This means the firm does not bid up I price of labor in order to hire more workers. Hence, l incentive to restrict hiring vanishes smce he is no long bidding up the wage rate by increasing employment.

However, at any collecuvely bargauied wj rate above Wc, the quantity of labor supplied excee that demanded. Therefore, ifthe union achieves a lectively bargained wage above Wc, employment will ot less than the competitive level, Nc. Nonetheless Figure XV, the union could achieve a union Mage high as W* before employment would fall below d initial monopsony employment of Nm- Under the

circumstances, however, unemployment must exist (why?); whereas in the initial monopsony equilibrium at Wm and Nn, no unemployment occurs.

While we can say for sure that the wage will not fall below Wn, (why?), the upper limit depends, among other things, on the amount of unemployment the Union would be willing to tolerate to achieve higher wages. Where the wage will .fmally setde is said to depend upon the "bargaining power" of both sides, a fancy way of saying we dont precisely know.

THE POWER OF UNIONS OVER WAGES

Although there are some dramatic exceptions, the overall ability of tmions lo raise wages is surprisingly small. For example, while unionists, on the average make about 15% more than non-union workers in the same trade, almost all ofthe difference is caused by the higher productivity of union labor. Unionized workers simply embody more human capital, which employers often reinforce by donating more physical capital. Thus, the marginal productivity of union labor is high. Such high marginal productivity would lead to high wages, whether the labor force is unionized or not. Unionism per se raises wages by only a few percent.

The power of unions over wages is constrained by essendally two general kinds of forces. Recall diat the demand for labor is derived from the demand for the fmal products made by labor. Increases in wages, as weve seen, increase firms costs, thereby ultimately limiting the amount of fmal product sold. Raising wages therefore implies a penalty in lost output and hence less ultimate use of labor input. The loss of output therefore constrains the ability of unions to raise wages. Clearly, the greater the penalty in lost output.

the weaker the ability ofthe union to raise wages. The ability to substitute out of labor is the other counter-force to the power of unions over wages. Even if wage increases result in no loss of final product, unions will be constrained by the substitution of capital for labor in production. The easier the substitution, the weaker the union. Thus, the overall power of unions over wages will be strongest when:

(1) Product demand is inelasdc. As we just saw, the increase in wages will reduce the supply of the final product, raising prices and causing the quantity sold to fall. In Figure XV these effects are depicted by a reduction in supply from S to S so that the quantity sold falls from its initial value of Q*. Now consider the two demand curves Dj and D2 where, as we saw in Section FV, D2 is the more elastic curve. If the product demand curve is D2, the reduction in output.will be larger than if the demand curve is Di. In the former case, output falls all the way from Q* to Q2, whereas in the latter case, the drop is only from Q* to Qj. Thus, when the product demand is inelastic, the loss of output due to a unions wage demand will be reladvely small. As such, the reduction in the firms use of labor inputs will likewise be relatively small so that the union can raise wages at a relatively low cost in layoffs.

(2) Substitution of capital for labor is difficult. As we saw in Section rv the greater the substitutes forja good, the more elastic will be the demand for that good. Hence, in Figure XVII, the elastic labor demand curve Dj results from easy substitutability between capital and labor, whereas the inelastic labor demand curve of D2 characterizes relatively difficult substitution. Starting from the competitive equilibrium at W* and N*, consider the impact on employment

of raising the wage to tlie collectively bargained level Wu- If substitution is easy, so that were operadng along D], firms will massively subsdtute out of labor and into capital causing employment to drop to Nj. On the other hand, if substitution is difficult, so that were operating along D2, firms will cut back employment only to N2-

The airhne pilots and airtraffic controllers unions illustrate these principles. If airtraffic controllers raise their wages much, management would simply substitute into computers and mechanical tracking and controlling devices. As such, many jobs would be lost. At present, however, there are no adequate substitutes for pilots. Consequendy, the less elastic demand for pilots strengthens the unions ability to achieve high wages for its members. (3) Labor costs are a small percentage of total costs. If so, even a large wage increase will litde affect the cost curves of the firm. As such, the loss of jobs and output will be small. On the other hand, when an industry is labor intensive, an increase in wages will raise total costs dramatically. For example, labor costs are an extremely high percentage of the costs of coal mining, but a relatively small percentage of the cost of oil well drilling. As such, coal mine unions have drastically less power over the wages of members than do unions of oil workers.

Oftentimes, skilled workers in smalPcraft" unions enjoy particularly high wages because they make a relatively small contribution to their employers total cost. Our theory therefore predicts - and history confirms - that the membership of such unions would resist merging uito giants like the United Auto Workers Union. Such labor solidaritj would weaken

the craft unions bargaining posidon by making them share the fate of larger and weaker unions whose contribution to firms total costs is tremendous. (4) The supply of capital is inelastic. Firms faced with wage demands from unions attempt to subsdtute out of labor and into capital. Ifthe supply of capital is inelastic, the price of capital will be rapidly bid up as firms make the switch. As such, the amount of capital substituted for labor will be smaller than if the capital supply is elasdc.

THE DEMAND FOR OTHER FACTORS OF PRODUCTION

Although its clear that firms will substitute out of labor and into capital when the price of labor rises, weve yet to show how firms chose their long run mixture of capital and labor Clearly, if firms want to get the maximum possible output per total dollar spent, they will allocate their input purchases so that the last dollar spent on labor causes output to rise by exacdy the same amount as the last dollar spent on capital. For example, if the last dollar spent on labor and capital causes output to rise by 10 units and 4 units respectively, then the firm could have a greater output at the same cost by spending a dollar more on labor and a dollar less on capital. Spending a dollar less on capital will cause output to fall by 4 units, bul spending a dollar more on labor causes it to rise by about 10 units, so that the total output rises by 6 units (10 - 4). Making the same point a slighdy different way, the firm could have the same output at a lesser total cost by reallocating its input purchases in fav or of labor

In order to show how much output rises when the firm spends a dollar on capital, we need to introduce the marginal physical product of capital (MPPt)-This concept measures the change in output (Aq) or total physical product ( ) caused by using an extra unit of capital. For example, if output rises by 12 units when the firm uses another piece of capital, then the = marginal physical product of capital is 12 units of out-; put. Continuing with the example, if a dollars worth of capital is 1/4 units of capital, then spending anoth-i er dollar on capital will cause output to rise by 3 units j (12/4). Thus, to calculate the amount output rises where the firm spends another dollar on capital, vve; multiply the marginal physical product of capitals (MPPj.) by a dollars worth ofcapital. Now, how much -capital can we buy for a dollar? Clearly, that amount is $1/Pk where Pk is the pnce ofcapital. For example, if] the price of capital is $4 per unii of capital, then wCf can buy 1/4 units ofcapital for a dollar We can sum-