XIII. 2004 LERA BEST DISSERTATION COMPETITION

Three Essays on Labor Policy

Christopher Riddell
Queen's University

Chapter 1

Union Certification Success under Voting versus Card-Check Procedures: Evidence from British Columbia, 1978¿1998
      The first chapter of the dissertation examines the impact of mandatory election laws on certification success using data on private sector certification attempts from British Columbia over the 1978 to 1998 period. The paper exploits a unique natural experiment. Prior to 1984, British Columbia had the traditional Canadian card-check system, in which the union can be certi- fied without an election if it signs up a sufficient proportion of the proposed bargaining unit.1 In 1984, a mandatory election system was introduced in which a secret-ballot vote was required if the minimum threshold was signed up. The election system was repealed in 1993 and replaced with the original card-check procedure. Moreover, by exploiting the timing of various legislative changes that occurred over the sample period, the effect of mandatory voting can be isolated from other legislative changes.2 The paper then examines the role of management opposition across the two different union recognition procedures.3
     In general, mandatory election laws are believed to lead to lower certifi- cation success rates for two reasons: (1) management has more opportunity to oppose the bid in a voting system; and (2) peer-pressure from pro-union colleagues and union organizers may artificially inflate the "true" level of support in a card-check system. With respect to management opposition, a change in the union recognition law may have two effects: (1) the effectiveness4 of management opposition may change; and (2) the incidence5 of management opposition may change.
     Figure 1 shows success rates for the private and "public" sectors, with withdrawn certifications and raids excluded.6 The pattern in Figure 1 is striking. Success rates for private sector organizing drives fell by nearly 20 percentage points following the introduction of mandatory elections in May 1984. Moreover, in 1993, success rates appear to have recovered to their prevoting levels and remained as such through the remainder of the sample period. It is interesting that public sector success rates did not systematically decline.7
     Econometric analysis that exploits the timing of various changes in legislation suggests that mandatory elections alone can account for the entire decline in private sector success rates. The other major issue addressed in the paper is the role of management opposition. I estimate the impact of unfair labor practices on certification success over the 1987¿1998 period, which covers one card-check regime and one mandatory election regime.8 In addition, I use a decomposition procedure to provide some evidence as to how much of the fall in success rates associated with the voting regime can be attributed to a change in the incidence of management opposition and how much can be attributed to a change in the effectiveness of management opposition. Overall, the central results are as follows: (1) management opposition is, at least, twice as effective under an election system; (2) there is no difference in the incidence of management opposition across the different union recognition regimes; and (3) management opposition can account for 25 percent of the decline in success rates.9

Chapter 2

Welfare Checks and Health Outcomes
     The second chapter examines the impact of the distribution of welfare checks on the health outcomes of drug users. The relationship between social assistance and substance abuse has been a topic of much debate in the United States, where, in 1996, Public Law 104-121 terminated Supplemental Security Income and Social Security Disability Income for recipients with a primary diagnosis related to substance abuse. The rationale for the law was a belief that such recipients were spending the money on drugs and alcohol such that welfare was effectively "aiding and abetting addiction" (Cohen 1994). Canadian law rarely makes special provisions for substance abuse.10
     A few studies in the public health literature have shown evidence of socalled check effects„that is, a relationship between the day welfare checks are released and subsequent increases in outcomes such as hospital admissions,

drug-related fatalities, and 9-1-1 calls (see Riddell and Riddell 2005). In addition, economists have examined the relationship between identifiable, preannounced income changes and consumption. One such study directly tests for a consumption-related check effect among welfare recipients and finds that the probability and amount of daily expenditures increases markedly on the day of check arrival (Stephens 2003).
     This paper contributes to the check effect literature by conducting a unique test using hospital admissions data on injection drug users (IDUs) from the Canadian city of Vancouver.11 The paper addresses the question of whether drug users are more likely to leave the hospital against medical advice (AMA)„thereby interrupting their treatment„on the day that welfare checks are released. Check day is linked to drug consumption by examining drug overdose admissions. In particular, I test whether individuals induced to leave AMA by "Welfare Wednesday" are more likely to be readmitted with a drug overdose.12
     From a theoretical perspective, the possible relationship between welfare checks and drug consumption has various components. One component is the pure income effect story. On the basis of Cohen's findings (1994), it is clear that the 1997 change in U.S. law was predicated on the notion that terminating SSI/DI benefits may lead to reduced drug consumption. To date, I am unaware of any study that has convincingly examined the income effect side of the substance use-welfare debate. Indeed, I am unable to test directly for income effects in this paper, because I do not observe an individual's benefit level. In this paper, I address the question of whether the distribution of welfare payments affects the distribution of drug consumption over time. On the basis of the addiction literature, there are two key reasons why I anticipate a spike in drug consumption„or "binging"„in the days following welfare day: (1) the neurophysiology of intravenous cocaine use and, to a lesser extent, intravenous heroin use;13 and (2) environmental cues.14
     Figure 2 shows the distribution of discharges (AMA discharges relative to planned discharges, which together equal total discharges) over the sample period. A clear spike is observed on the day welfare checks are released. The results from econometric analysis confirm that individuals induced to leave AMA by Welfare Wednesday are readmitted to the hospital much sooner than other patients and are more likely to be readmitted with a drug overdose.

Chapter 3

The Long-Run Effect of Unemployment Insurance: A Study of Maine and New Brunswick, 1940¿1991
     The final chapter examines the long-run impact of unemployment insurance. Existing studies of the labor market effects of income support programs

focus mainly on short-term responses to policy changes.15 It is not surprising that some critics have argued that such studies may dramatically understate the long-term effects of these programs. Over a longer time horizon, individuals have the ability to make larger adjustments in their behavior and adjust on a wider number of margins, in some cases effectively building a "lifestyle" around the programs in question.
     This paper attempts to measure the long-term effects of large-scale changes in income support policy by taking advantage of a natural experiment. This experiment results from the fact that a national border„between New Brunswick and Maine„divides a region with a relatively homogenous population and resource endowment but that experienced dramatically different changes in income support policy. By 1940, Maine had a modest unemployment insurance (UI) system but New Brunswick had none. After that, New Brunswick's system grew much more rapidly, to the point where New Brunswick's system in 1980¿1990 was astonishingly generous to workers with very short work histories. Using the five decennial censuses spanning the period 1940¿1991, the paper exploits the comparison between these two jurisdictions over time to estimate the long-term effects of UI program parameters on labor market outcomes.16

     The focus in this paper is on the incidence of part-year work among workers and has two main components. The first component uses grouped cell-level census data for the entire 1940¿1991 period. A unit of observation here is an industry-gender-region-year cell; for each such cell I calculate the fraction of persons working part-year (defined here as less than forty weeks), and analyze its responsiveness to the relative income of part- versus full-year workers in that cell, as determined by prevailing wage rates and UI program rules. The second component estimates the entire distribution of annual work weeks using census microdata from the three decades, 1970¿1991, for which it is available in both regions.

     The results show that men in Maine and New Brunswick had a similar, high incidence of part-year work in 1940; however, whereas men's part-year work significantly declined over the next five decades in Maine, it did not in New Brunswick. This raises the possibility that, for men, the Canadian UI system may have preserved a "way of life" that has become largely extinct in Maine. Among women, part-year work was substantially less common in New Brunswick than Maine in 1940; by 1990 this pattern had been dramatically reversed as a result of growth in New Brunswick and decline in Maine. For women, Canadian UI may thus have created a new pattern of part-year labor force participation that would not otherwise have existed.

     The empirical findings provide strong support for the hypothesis that UI has played a key role in the divergence in part-year work between the two regions. For instance, the results indicate that, for men, the 1972 reforms alone can account for nearly 60 percent of the change in part-year work that occurred over the 1960¿1990 period. It is interesting to note that, although a substantial share of the divergence in part-year work between these two regions is attributable to Canada's 1972 UI reforms, the analysis shows that Canada's introduction of generous "seasonal benefits" in the 1950s also plays an important role. Because earlier studies have not always accounted for these benefits (which are hard to characterize), such studies overestimate the size of Canada's policy change in 1972 thereby underestimating the effects of changes in UI program parameters at that time.

Notes

1. The proportion required to avoid an election varies by province, but generally has ranged between 50 percent and 55 percent. Although it is possible for a certification bid to require a vote in a card-check province, the reality is that few organizing drives go to an election. Historically, all Canadian provinces, except Nova Scotia since 1977, have had "card-check" procedures, but this has changed with an increasing number of provinces going the election route over the past decade. In addition, over the 1984¿1998 period, union density fell by about 7 percentage points in Canada„similar to the 8 percentage point decline in the United States during the same years (Riddell and Riddell 2004). Many observers believe that the recent decline of both unions and the card-check system is no coincidence.

2. Three legislative changes were made to the British Columbia Labour Code over the sample period of October 1978 to May 1998 (see Riddell 2004 for a detailed analysis). The first, when mandatory elections were introduced, was in 1984. Unlike many legislative packages passed in Canada that included mandatory elections, the 1984 amendment made very few changes to the Labour Code. Second, in 1987, sweeping changes were made to the code when the Industrial Relations Reform Act was introduced. Finally, in 1993, mandatory elections were repealed. Many of the "voting era" amendments, however, were maintained.

3. The only previous paper that has isolated the effect of mandatory elections is Johnson (2002), who finds that elections reduce certification success by 9 percentage points. For the United States, Easton and Kriesky (2001) find some evidence that success rates are higher in cases decided by card-check relative to elections

4. A voting regime may increase the effectiveness of employer tactics for two reasons. First, the amount of time that employers have to influence the organizing drive is unambiguously greater in the voting system. In the Canadian card-check system union organizers collect signatures from members of the proposed bargaining unit, and then make an application. In British Columbia, it is rare for a union in one of the card-check regimes to fail to gather 55 percent support„sufficient for certification without a vote. The employer is contacted when the BC Labour Relations Board (BCLRB) conducts its investigation of the petition requirements. Thus, under card-check many organizing drives are virtually complete before the employer is aware of the drive. Under voting, unions still gather cards, but then a secret ballot vote is held. This extra step in the certification process gives the employer greater opportunity to launch a campaign. Second, management opposition is likely to be more effective in a voting regime because of the secret ballot vote itself. Under card-checks, if the employer coerces employees into refusing to sign cards, union organizers and pro-union colleagues can counteract the coercion tactics. In a secret ballot vote, the opportunity to counteract employer threats is likely diminished.

5. If the effectiveness of management opposition is greater in a voting regime„and employers know this„a cost-benefit calculation implies that employers will adopt such tactics more frequently given that the chances of defeating the bid are greater while the costs remain the same. As well, there is simply more time available to the employer under elections. Thus, those employers in a card-check system who did not realize organizing activity was occurring until after the cards had been collected would have an opportunity to oppose the union in a voting system.

6. A large majority of the social services certifications are public sector, but a few from health services are private sector and, unfortunately, cannot be isolated.

7. This finding is suggestive of an explanation based on management opposition, because employer resistance to certification is likely to be negligible in the public sector due to concerns over public image and the lack of profit maximization objectives, management opposition (particularly egregious management opposition) is likely to be negligible in the public sector.

8. Another issue is the potential endogeneity of ULPs (i.e., if the employer believes success is contestable and can be influenced, they are more likely to pursue an anti-union campaign compared to situations where success is very likely or very unlikely). Instrumental variable regressions are estimated to account for this potential endogeneity. Overall, the findings are similar. The results imply that standard ULP estimates are biased toward zero.

9. This result may understate the effect of election laws and management opposition on union organizing in a broader context. First, the number of certification attempts fell by around 50 percent following the 1984 law. Second, no evidence is provided on first contract success. Both of these issues are being addressed in future research.

10. Income assistance laws are under provincial jurisdiction in Canada. Our reading of the provincial laws for the five largest provinces is that only Ontario (in a legislative change made in 1997) has a strong restriction on collecting disability support if dependent on (or addicted to) alcohol or drugs (see subsections 2 and 3 of section 5, part 1 of the Ontario Disability Support Program Act). The Ontario law, however, is not as restrictive as the U.S. law because an individual with an addiction can remain eligible for benefits if they have a "substantial physical or mental impairment" (see subsection 3), which would generally include HIV/AIDS and various psychiatric conditions, both of which are common among the injection drug users in our data.

11. The drug use situation in Vancouver is dire. In 1993 drug overdoses from heroin became the leading cause of death among men 30¿49 years of age (see Riddell and Riddell 2005 for complete discussion and references). Through the 1990s, Vancouver has consistently had the highest overall levels of overdose deaths in Canada with more than 300 in 1998 alone and more than 2,000 between 1991 and 1998. According to a statement by the British Columbia general coroner, Vancouver's overdose death rate is "very likely" the highest in North America. The incidence rate of HIV infection among Vancouver's injection drug users is conservatively estimated to be 25 percent, with some articles in the media quoting a rate as high as 50 percent. Even with the conservative estimates, Vancouver is widely believed to have one of the highest levels of HIV infection among injection drug users in North America and similar to cities such as New York and Bangkok.

12. Welfare checks are distributed once a month in the province of British Columbia, usually on the last Wednesday.

13. The more rapid a drug's onset of action the greater the euphoria users experience and the more they "like" the drug. Cocaine and heroine„particularly when used intravenously„ are especially rapid in their delivery to the brain. Moreover, in the case of cocaine, the rate of removal from the brain is extremely fast. This is believed to be a key reason why individuals who become addicted to cocaine tend to move toward a "bingeand- crash" pattern of use, in addition to moving from intranasal and smoking administration toward intravenous use.

14. Research in both the psychology and neuroscience fields has found that consumption„ particularly drug consumption„may be linked to certain environmental cues (see Laibson 2001 for a review of these literatures). This notion implies that, once a culture of drug use has been established„such as in the Downtown Eastside of Vancouver„ it would be very difficult for a user to resist drug use once exposed to that culture, even if their intention was not to use drugs. Indeed, Welfare Wednesday itself may have become an environmental cue.

15. Of course, for a variety of reasons, the long-term effects of income support programs are much harder to measure than short-term effects. One reason is that large-scale changes in such programs tend only to be made at the national level. As a result, a "control group" is typically not available to indicate what might have happened in the absence of a policy change. Identification of long-term effects in nonexperimental studies is further hampered by the greater leeway for confounding variables such as macroeconomic conditions to change over longer periods„problems that can be avoided in short-term studies using, for example, a regression-discontinuity design. Experimental studies, which solve the control group problem via random assignment, tend to be very limited in duration and to focus on small variations in program parameters.

16. One paper that examines long-term effects of income assistance is Meyer and Rosenbaum (2001), who study the longer-run effects of reforms to the welfare system and the earned income tax credit, with women without children as a control group for those much more affected by these policy changes„women with children. A necessary feature of studies like Meyer and Rosenbaum's is the use of a control group that is drawn from the same economy as the "treatment" group. As a result, their estimates are likely contaminated by general equilibrium effects of policy changes. Further, the control group, of necessity, is demographically different from the treatment group (e.g., childless women versus mothers), and it is not clear that the maintained assumption of identical policy effects on both groups is plausible. The prospect of using the same demographic groups in similar neighboring regional economies over long periods of time, and in the presence of large differential changes in income support programs thus provides an important new perspective in estimating the effects of social policy changes.

References

Cohen, W. (1994) "Tax Dollars Aiding and Abetting Addiction: Social Security Disability and SSI Cash Benefits to Drug Addicts and Alcoholics." Report of the Senate Special Committee on Aging and Senate Subcommittee on Oversight of Government Management. Washington, D.C.: Government Printing Office.

Eaton, Adrienne E., and Jill Kiesky. 2001. "Union Organizing under Neutrality and Card Check Agreements." Industrial and Labor Relations Review, Vol. 55, no. 1, pp. 42¿59.

Johnson, Susan. 2002. "Voting or Card-Check: How the Union Recognition Procedure Affects Organizing Success." Economic Journal, Vol. 112, no. 4, pp. 344¿61.

Laibson, David. 2001. "A Cue-Theory of Consumption." Quarterly Journal of Economics, Vol. 116, no. 1, pp. 81¿119.

Meyer, Bruce, and D. Rosenbaum. 2001. "Welfare, the Earned Income Tax Credit, and the Labor Supply of Single Mothers." Quarterly Journal of Economics, Vol. 116, no. 3, pp. 1063¿114.

Riddell, Chris. 2004. "Union Certification Success under Voting versus Card-Check Procedures: Evidence from British Columbia, 1978¿1998." Industrial and Labor Relations Review, Vol. 57, no. 4, pp. 493¿517.

Riddell, Chris, and W. Craig Riddell. 2004. "Changing Patterns of Unionization: the North American Experience." In Thomas A. Kochan and Anil Verma, eds., Unions in the 21st Century: An International Perspective. London: Palgrave Macmillan.

Riddell, Chris, and Rosemarie Riddell. 2005. "Welfare Checks, Drug Consumption and Health." Journal of Human Resources, forthcoming.

Stephens, Melvin. 2003. "'3rd of Tha Month': Do Social Security Recipients Smooth Consumption between Checks.' American Economic Review, Vol. 93, no. 1, pp. 406¿22.

.