Do unions really benefit workers? Think again. Assar Lindbeck & Dennis Snower. (2002). The Insider-Outsider Theory: A Survey

http://anon-ftp.iza.org/dp534.pdf
We are typically told unions represent labour. They always demand their seats in all kinds of committees on behalf of workers. They say, “we are fighting for all the workers.” However, are they, really? The insider-outsider theory shows something different. This theory was proposed first to explain why developed economies have natural unemployment. Even in a boom, unemployment rate never goes below 5% in Canada. Then economists typically say “we are in full employment” despite those poor five percent of unemployed labour forces. Why do we have them?
The principle observation of the insider-outsider theory is that it takes additional costs to replace existing employees by new workers (Lindbeck & Snower, 2002, pp. 1-2). The proposers of the theory name these costs ‘labour turnover costs (LTCs)’. These not only include the costs to recruit, hire, and train new employees but also encompass the costs to dismiss the existing employees. If an employee is very skilful and fired, then this worker may work for another competitor to the ex-employer and her business will suffer from losing some market share; To prevent the fired worker from working for a competitor, the employer may need to pay considerable amount of severance pay; if employees are unionized, especially, a dismissal of employees may lead to a lawsuit or stoppage of production led by the union, and possibly cost more than continuing to employ this worker. Therefore, LTCs give insiders market power. According to Lindbeck and Snower:
“[T]hese labour turnover costs (LTCs) divide workers into three groups: (i) insiders, whose positions are protected by these LTCs, (ii) outsiders, who have no imminent prospect of such protection (e.g., the unemployed, workers in the informal sectors, and inactive individuals), and (iii) entrants, who hold jobs that may lead to insider status.” (p. 2)
In this respect, roughly, union members and skilled workers are insiders; on the other hand, non-union members and unskilled workers belong to groups of outsiders or entrants (pp. 2-3). Union members, as insiders, can enjoy the benefits from union bargaining, such as job security, health insurance, signing bonus, stock options, etc.; on the other hand, non-union members, as outsiders or entrants, are excluded from these benefits. The benefits that insiders enjoy are not only from the expenses of employer but also from (directly or indirectly) imposing “less favourable” conditions towards outsiders and entrants. For example, union contracts typically include “last in, first out” provision; entrants would typically receive lower wages than insiders even if they were actually more productive; outsiders are not allowed to work at insiders’ positions even though they are willing to receive lower wages than insiders; outsiders may get stuck in “jobs with relatively low wages and/or low expected tenure” (p. 4). Since insiders’ market power stems from these LTCs, insiders are motivated to raise them because “[t]he smaller are the firm’s labour turnover costs (ceteris paribus), the more profitable it is for the firm to stop bargaining with its insiders and start bargaining with the outsiders instead” (p. 8). If LTCs are low enough, the labour market is contestable enough for employers to consider outsiders.
Lindbeck and Snower divides LTCs “into two categories: “production-related” costs, which must be expended in order to make outsiders productive within a firm, and “rent-related” costs, which are the outcome of insiders’ rent-seeking activities” (Lindbeck & Snower, 2002, p. 3). The insiders enjoy higher wages and more favourable benefits (union wages premium), because of their bargaining power, than non-members in the same industry. Fang and Verma (2002, Union wage premium, Persepectives on Labour and Income, 3 (9), p. 20) estimated this premium in Canada, as of 1999, is 7.7% on the top of average non-union member wage rate. Economically, this premium is a type of rent. If an employer attempts to cut wages and benefits of unionized employees, then she will easily face labour disputes, such as strikes or picketing, or, at best, collective bargaining, which incurs costs, as well.
Do unions benefit workers? Of course, yes. However, only a part of labour. Unions mainly (very often, only) benefit their members; non-members are usually ignored, or sometimes face hostility from the insiders when their interest is eroded by outsiders.