Long before the Great Recession, Swedish policymakers had been acting as doctors to the masses. The country’s innovative social protection plan is called the Active Labor Market Program (ALMP). “Active” is the crucial word here. ALMPs are different from typical social safety nets for the unemployed more commonly found in countries such as the US, Spain, and the UK. Those “passive” programs usually provide cash benefits to the unemployed to replace their lost income (of course, the recipients had contributed every month to their unemployment insurance while they were working). While there is little doubt that unemployment checks help unemployed people to continue to support their families and make ends meet, the Swedes designed their programs to be “activating”—to help people get back into new jobs as quickly as possible.15
Since the 1960s, Sweden had been developing ALMP programs that provide workers with support, skills, and a plan to get back to work. While there is a lot of variation in how countries organize their ALMPs, Sweden’s were particularly well-developed and comprehensive to help keep workers active. In Sweden, when people lose their jobs, both they and their firms register at a government job center. So people participate by default. Within the next thirty days, the center creates an “individual action plan” with the person who has become unemployed. This individual has an interview with a job trainer every six weeks to see how the job search is going. The program also requires that the participant continue their job search (and it verifies their efforts) during their participation in the ALMP. In order for people to access cash benefits, they have to take part in a guided, step-by-step plan to get back into work.
Sweden’s ALMPs placed a much greater emphasis on activating workers than the US or Spain. The unemployed in Sweden were not just having their hands held, but were being actively reached out in order to help them stay economically active, and program managers worked with companies to get new jobs generated for their recently laid-off workers. That is not to say that unemployment offices in the US and Spain didn’t provide job search opportunities—they did, but their programs were far less active in their out-reach and aims than the Swedish ALMPs. One of Sanjay’s patients, for example, encountered a US version of an “ALMP”; he had to wait three hours to get a pamphlet that told him to prepare a resumé, take a shower, and wear a suit.
Before the Great Recession, ALMPs had played a critical role in preventing unemployment from causing depression in those countries that deployed them. In Finland, a randomized controlled trial in 2002 tested the effectiveness of the country’s ALMP program, Työhön (meaning “let’s go to work”). Researchers assigned 629 people who had lost their jobs to a job-training program with skilled caseworkers. A control group of 632 people received printed information about finding a job (the same printed information provided by the ALMP program) but did not receive the actual help of a Työhön trainer. The results were remarkably different between the two groups. Within three months, the researchers found, workers who were enrolled in the Työhön program experienced significantly fewer symptoms of depression and anxiety, with the greatest benefits seen in those previously at high risk of depression. The researchers found that two years later ALMP participants had significantly fewer depressive symptoms and higher self-esteem, were less likely to have given up hope on finding a job, and were more likely to have successfully returned to work than the control group.16
In practice, ALMPs provided mental health resilience to job loss in at least three ways. First, they helped people who lost their jobs find new ones as soon as possible—eliminating a key source of depression and anxiety. Indeed, studies of depression showed that symptoms of short-term depression often disappeared once an unemployed individual returned to work. Second, ALMPs helped reduce the mental health risks that accompanied job loss by providing formal social support through a trainer, as opposed to leaving people to cope with job loss on their own. Third, the data showed that these programs could even help people who were not unemployed but worried they might be; those at risk of losing their job knew they would get assistance to find a new one, and this appeared to prevent depressive symptoms among those at high risk for unemployment. If properly implemented, ALMPs were a classic win-win situation: improving the economy and preventing depression.17
Sweden had been learning from these data since the 1960s, and by the 1980s began operating some of the most highly sophisticated ALMPs in the world. One reason was that their politicians invested in well-resourced programs. Each year, Sweden spent a total of $580 per capita to help the unemployed get back into work. The US, UK, and Spain were spending less than half of this figure. Not only did Sweden invest more, but it invested proportionally more in active programs rather than passive paychecks. In the mid-1980s, Sweden spent about three-quarters of its funds to help the unemployed on active programs, whereas the US spent about one-third, the UK one-quarter, and Spain about one-tenth of their budgets on active programs. In 2005, the Organization for Economic Cooperation and Development published a comprehensive report comparing unemployment programs across European countries. It found that Sweden’s investment was paying off: its ALMPs had among the quickest response times in setting up interviews and individual action plans for the newly unemployed.18
These results all seemed well and good during normal economic times. But would these programs be enough to help protect Sweden from experiencing a large rise in suicides during an economic meltdown?
The protective effects of Sweden’s ALMPs were put to the test during its recession in the 1990s. In a scenario that resembles the current recession, Sweden’s housing market collapsed in 1991 and 1992, bringing nearly all of its 114 banks to near-closure. GDP fell by 12 percent. Ten percent of Swedish workers lost their jobs—a rise on par with unemployment spikes in many countries during the current recession.19
Remarkably, despite Sweden’s large spike in unemployment, suicide rates actually fell steadily during the period between the 1980s and 2000s, when the government invested, on average, about $360 per capita per year in active labor market programs (see Figure 7.3). There was no significant correlation between Sweden’s fluctuations in unemployment and its suicide rates.
Like Sweden, Spain experienced large increases in unemployment from its recessions of the 1980s and 1990s. But Spain operated a relatively poorly resourced unemployment program, investing only about $90 capita per year, and focusing this money only on cash benefits. For men in Spain, trends in unemployment correlated strongly with suicide rates, as shown in Figure 7.4.20
FIGURE 7.3 Active Labor Market Programs, Unemployment, and Suicide in Swedish Men, 1980–200521 (solid line: unemployment; broken line: suicide)
We wanted to be as sure as possible that ALMPs were the determining factor in reducing the risk of suicide during economic recession. So we examined suicide rates and unemployment programs across all of the European countries using over two decades of data, comparing ALMPs to all the other main types of social protection programs, including healthcare services, family support such as childcare support, housing subsidies, old-age pensions, and passive unemployment benefits to see which, if any, could prevent an increase in suicide rates during recessions. Healthcare spending, for example, did not significantly reduce the risk of suicides from unemployment. This made sense: if the main risks of depression came from factors like unemployment, the formal healthcare system would probably not have much power to neutralize these risk factors. We also found that cash benefits didn’t reduce the risk. In test after test, we found that the ALMPs had the greatest and most significant preventive effects on suicide when compared to other social protection programs.
We estimated that, if done properly, ALMPs could neutralize the suicide risk of a recession. In our findings, published in the peer-reviewed medical journal The Lancet, we estimated that for a $100 investment per capita, on average ALMPs appeared to lower the risk of unemployment-related suicide from 1.2 percent to 0.4 percent.
FIGURE 7.4 Active Labor Market Programs, Unemployment, a
nd Suicide in Spanish Men, 1980–200522 (solid line: unemployment; broken line: suicide)
The data provided a striking example of how social protection programs could save lives. When countries invested more than about $200 per capita in ALMPs, the correlation of unemployment with suicides appeared to completely vanish. This was precisely why unemployment spikes had no correlation with increased suicides in Sweden, Finland, and Iceland, but unemployment was strongly correlated to suicide in Spain, the US, Greece, Italy, and Russia.
The lessons we learned from our research on ALMPs provided answers to several puzzles. First, ALMPs statistically explained why becoming unemployed was so much more dangerous for people in Eastern rather than Western Europe. At the time when the Soviet Union disintegrated, Finland, a major trading partner in Western Europe, suddenly lost one-third of its economy as the sales made to Soviet factories evaporated. Finland also had a drinking culture similar to the Soviet Union, and its unemployment had soared during the recession. Yet there was little or no discernible effect of the economic crash on suicides in the country. By contrast, the surge in unemployment in Russia, Kazakhstan, and the Baltic states corresponded to a devastating mortality crisis. This dramatic difference could be explained by the fact that Western European countries like Finland tended to invest considerably more in labor market protections (about $150 per capita) than did Eastern European countries ($37 per capita).23
When we presented our findings at research conferences, our colleagues in Russia and Poland suggested that unlike Sweden or Finland, their countries simply couldn’t afford to invest in ALMPs. But we found that, if executed well (i.e., in the Swedish style), ALMPs would essentially pay for themselves by boosting employment and reducing the burden on social welfare. A detailed analysis of Danish ALMPs, for example, concluded that the economic benefits of these programs far exceeded their costs, because the programs increased workers’ productivity and reduced reliance on welfare support. In Denmark, the programs generated a net savings of 279,000 Danish kroner (about $47,000) per worker over eleven years. Another study in 2010 performed a systematic review of 199 ALMPs studied in 97 research experiments. It found a similar consistent pattern of results as the Danish program: that ALMPs helped people return to work and, by keeping people economically active, reduced pressure on public welfare systems by increasing the economy’s labor supply—a main engine of economic growth.
With all of this evidence accumulating in favor of ALMPs, we were eager to translate these data into practice. After we published our research in 2009 about the benefits of ALMPs, we were invited to the British House of Commons and the Swedish Parliament to present our data and recommendations.24
The responses were remarkable—that is, remarkably dissimilar—in the two countries. When presented with the data that unemployment led to a rise in suicides, and that ALMPs could help mitigate the risks, the Swedish members of Parliament were unsurprised. One member asked: “Why are you telling us what we already know?” But when we presented the same data in the UK, in July 2009, to the House of Commons, the reaction was that the government was “already doing all it could to reduce unemployment.”
When the Conservative government came into power in 2010, the UK response became even worse. In 2012, the British Medical Journal published our paper showing that UK suicides had risen by more than 1,000 between 2007 and 2010 above pre-existing trends, corresponding to the continued rise in unemployment. Reporters soon contacted the UK Department of Health for a response. Its spokesman told the In de pen dent newspaper: “Losing a loved one [to suicide] can be devastating and we want to make sure that we are doing all we can to prevent suicide by giving people the right support when they need it most. We will shortly be publishing our new suicide prevention strategy, which brings together expertise across healthcare, criminal justice and transport to maintain or even decrease the current rates of suicide.” This sounded encouraging. But then the Health Department spokesman continued: “However, suicide rates in England have been at a historical low and remain unchanged since 2005. The department uses three-year rolling averages for monitoring purposes, in order to avoid focusing unnecessarily on fluctuations instead of the underlying trend.”25
By now, this tactic should sound familiar: averaging-out deaths is the same technique The Economist used to cover up death rates in Russia. When using rolling averages, any large jump in death rates can seem like a smooth bump in the road instead of a shocking spike (indeed, the Department appeared to have chosen the three-year period specifically for this end, instead of some other date range like five years). The Department’s comments were criticized by several university professors and statisticians, after which the statement quickly disappeared from their Internet webpage.
If it wanted to help its people, the British government could learn much from Sweden’s experience. The UK would of course need to invest more in ALMPs and stop job losses from happening. But the Conservative government was doing precisely the opposite: austerity was creating an active labor-destroying program. The data revealed that the austerity program cut public-sector jobs in the most deprived regions of the country. Moreover, it was implementing policies that made it easier for the private sector to lay off people during the recession. As one unusually blunt 2010 report commissioned by the government explained, “some people will be dismissed simply because their employer doesn’t like them,” but argued that this is a “price worth paying” to boost the economy, though the logic of how mass unemployment would drive economic growth was left unexplained.26
The consequences of the UK’s real-world experiment with austerity soon became tragically apparent in its suicide data. As in the US, the Great Recession in the UK featured an initial spike in unemployment and job losses in 2007. As employment began to recover in 2009, suicides began to fall. But the following year, when the Conservative government came to power, the UK began a massive austerity program, which in 2012 alone cut 270,000 public-sector jobs. The UK then experienced a second wave of “austerity suicides” in 2012.27
It is said that those who don’t learn from history are doomed to repeat it. Our models had predicted a repeat of suicide trends in Spain and Sweden, and we were now seeing their contrasting histories on playback into the present. Sweden and Spain both experienced large recessions, but suicides again rose substantially in Spain and actually fell in Sweden.
The range of observed suicide trends across the UK, the US, Iceland, Greece, Italy, Spain, and Sweden revealed that a rise in suicides can be averted during recessions. More suicides could have been prevented if the right steps were taken both before and during the recession to help people return to work.28 These suicides are only the tip of the iceberg; for each suicide there are an estimated ten suicide attempts and between 100 and 1,000 new cases of depression.29
As the Swedish and Finnish experiences demonstrate, unemployment may put people’s mental health at risk, but suicides needn’t inevitably follow. Nor should we wait for our healthcare systems to pick up the pieces when people become depressed or suicidal. Antidepressants may help some people deal with the consequences of unemployment. But wouldn’t it be better to treat the cause of the problem rather than the symptoms? The Swedish invested proactively in programs that reached out to people who lost jobs and helped them develop an action plan for returning to work.
The sufferings of the White Widows of Italy and their husbands, “suicided” by austerity, must not be forgotten. The way forward is surprisingly clear in the data. The question is whether we will now take action.
8
A PLAGUE ON ALL YOUR HOUSES
The crows of Bakersfield started dying in May 2007.
Most Californians knew the town of Bakersfield as a hot little hellhole between San Francisco and Los Angeles—a pit stop for gas and cheap Indian food. It’s a town that seemed destined to be ignored.
But when the crows of Bakersfield started dying, people started talking. First came the reports of children finding dead ravens f
loating in backyard swimming pools. Then other species of birds started falling out of trees. One local news crew filmed a swarm of swallows collapsing like little comets from the sky.1
The dying birds were frightening enough, but panic erupted when humans began developing strange symptoms. Some people had tremors. Others experienced a condition that doctors call “myoclonus”—a spasmodic, involuntary contraction of muscles that looked like the shudder of the possessed. Most became confused, and a few were paralyzed.2
By the peak of that summer, several people had already presented at the Bakersfield Memorial Hospital with a constellation of these symptoms, which some thought might be polio, and others believed was surely the wrath of an angry God.
“Could it be over-heating?” wondered a doctor, attempting to calm nerves. That year had after all been exceptionally hot, and the winter the driest since 1988. The Kern River, typically rippling downstream from the Sierra Nevada mountains, had baked dry into a cracked muddy road. A truck driver from Arkansas had even collapsed and died of heat stroke at Bruce’s truck stop on the eastern edge of town.3
But after a few weeks, scientists from the California Encephalitis Project, a state laboratory supported by the Centers for Disease Control and Prevention (CDC), provided the Bakersfield doctors with a more rational explanation. Spinal fluid samples from the affected people tested positive for West Nile Virus.
West Nile Disease had appeared before in New York City and Texas, killing birds at first. Mosquitoes were the main vector for the disease, carrying the virus and transporting it from one victim to the next. Mosquitoes would become infected after biting the dead birds, carry the virus in their salivary glands, then transmit it to living birds, horses, and humans. In some people, West Nile infection led to a day of fever, body aches, and a rash; in others, particularly the elderly or those with weak immune systems, it caused a potentially fatal infection of the brain.
The Body Economic Page 15