Friday, April 02, 2010

The Psychological Effects of Recession - Human Sciences Talk

Linked here is a talk I gave to the UCD Human Sciences symposium that took place in Newman House on Tuesday. Below is a summary of this. All of this is preliminary but, given that we are dealing with a real-time situation, I feel some degree of responsibility to present findings for discussion even if it may be  some more time before the results are "camera-ready".

(i) The talk begins with an overview of the extent of the Irish recession. Ireland has witnessed a fall off in GNP that is the largest in the last 100 years. Retail sales are one third less than peak. Unemployment rates spiraled and are now stabilising at over 13 per cent, with far higher rates particularly among young men.

(ii) The second part of the talk addresses how recent research and developments in behavioural economics differs from traditional understanding of recession. In particular, loss aversion and endowment effects imply asymmetric effects of economic fluctuations with psychological effects being steeper on the downside. Myopia implies that downturns may have greater psychological effects than would be estimated from models where individuals smoothly discount the future. Recent work on emotion and identity further highlights the extent to which the state of unemployment may be particularly aversive. Furthermore, there has been an increasing emphasis in economics in measuring well-being directly on the basis that consumption itself is incomplete as a marker of welfare, particularly in economically wealthy countries.

(iii) The third part of the talk examines evidence from the "Celtic Tiger" period in Ireland. Following an increase in well-being that occurred in the late 1980s and early 1990s largely due to a decline in unemployment and increased job security, psychological well-being levels remained mostly static during the Irish economic boom. Furthermore, the effect of income on well-being is small once one climbs above the bottom 15 per cent of the income distribution. This is robust to a wide range of models. The effect of unemployment is still enormous through the Celtic Tiger period, being a great impact on well-being than either divorce or the presence of a chronic illness. To date, no variable I have ever looked at can explain this effect. I worried somewhat before that the results on Ireland are out of line with the Stephenson/Wolfers work but if you look closely at their paper, you can see that Ireland is one of the few countries that provides poor evidence for their central point when looked at from the point of view of the time series (see Figure 16).

(iv) The fourth part of the talk examines the well-being of a sample of approximately 1,300 students that we tracked from 2009-2010. Approximately 300 of these had graduated, and approximately 80 of these are unemployed. These results are preliminary but the effect of unemployment on life satisfaction is once again enormous and completely robust to a wide range of contemporaneous and lagged controls including personality measures, parental income and several others.

(v) The fifth part of the talk presents preliminary findings from our IRCHSS-funded project on well-being. Once again, unemployment has a massive effect on well-being, dominating all other effects including home-value and income reductions by a large margin. A big limitation of our data so far is a sample size of approximately 500 which reduces our ability to look at interactions and a restriction to households with internet access, which raises representativeness concerns. But the effects are substantial and in the direction expected. We are working further now on examining the interaction of home value reductions and unemployment in determining well-being. Furthermore we are examining the extent to which people with different types of values and identities are differentially affected by economic changes. A further point that is interesting is that the effect of  income reductions on life satisfaction, while not in the ballpark of unemployment changes, look qualitatively higher than previous work in Ireland. Again, this results is on probation but I have a strong sense that concerns such as procedural unfairness, anger about bank bailouts and so on may be magnifying the well-being effect of income losses in Ireland. Furthermore, we need to experiment with the ordering of these questions. I worry that asking people about income reductions may prime them to give lower life satisfaction scores than they would otherwise give. We have not controlled for this so far in our research.

(vi) The sixth part of the talk examines the evolution of the suicide rate in Ireland. Ireland's suicide rate increased enormously during the economic boom, partly in keeping with a long-run increase in the suicide rate. It stabilised post 2002 and, while suicide researchers urge caution in interpreting recent suicide data due to potential for changes in death reports and so on, it looks likely that the suicide rate in Ireland began to decline post 2004. This decline continued in 2008 the first year of our recession in Ireland. Vital statistics provided by the CSO indicate increases in the number of suicides in 2009 relative to 2008. For example there were 111 suicides reported in the third quarter of 2008 as opposed to 126 in the third quarter of 2009. This is worth treating seriously and certainly merits investigation as to the composition and potential motivations for suicides in 2009. However, the idea of a stable relationship between macroeconomic fluctuations and suicides in Ireland looks fanciful with some of the highest suicide rates in the history of the country having taken place at times of high economic growth. Furthermore, as said above, most suicide researchers in Ireland caution against the use of the raw figure in the year of publication as they are subject to revision.

(vii) The final part of the talk argues that far greater emphasis be placed on the development of unemployment policies in Ireland. Furthermore, it must be widely acknowledged that our current policies are failing badly. The viewpoint that unemployment results from a preference for leisure looks at best a tautology from the viewpoint of the well-being data, and at worst a dangerously prejudiced misconception. Ongoing research here is gathering further data to examine the effect of unemployment and home value reductions at the micro level on well-being. We are also comparing the effects of recession when looked at from traditional life satisfaction and well-being measures to when we examine more "microscopic" measures such as the day reconstruction method. The effect of unemployment stress on job search and financial decision making is also being examined.


Michael Daly said...

Great to see there is data on students followed up into the labour market and unemployment. It would be interesting to see the psychological factors that modify adjustment to unemployment and also what distinguishes those who get jobs, beyond subject studied and academic ability.

On the psychological effects of recession, loss aversion provides an explanation for why many people are willing to work overtime/take cutbacks to avoid losses. Emotional attachments and endownment effects do predict enhanced disutility following loss. However, from affective forecasting & loss aversion studies we also expect that people will overestimate the emotional impact of the loss though when it occurs they will adapt well. Does this apply to losing one's job? From the discussion there's an argument that people may even err on the side of underestimating the hedonic impact of unemployment.

What about students? What are they losing, their gamble in investing time/effort/money in an education with a pay-off at the end? Disappointment captures some of the sentiment well generated by expectations prior to/during the gambling being at serious odds to the outcome they face now.

Michael Daly said...

(not implying that students spent their undergrad in the 78 club there!)

Liam Delaney said...

"Emotional attachments and endownment effects do predict enhanced disutility following loss. However, from affective forecasting & loss aversion studies we also expect that people will overestimate the emotional impact of the loss though when it occurs they will adapt well. Does this apply to losing one's job?"

This is dead right Michael. A student in my psychology class raised this exact point last year. The more I have thought about this, the more I think that people actually do predict pretty well the psychological effects of the main negative life effects. I really have never found any evidence in any data-set I have analysed that people adapt back to normal levels following receiving a chronic illness, getting divorced or being made redundant. If I had to put my hand on my heart and say what I think then it would be that most people experience these events as partly unanticipated shocks and that they cause major and prolonged psychological distress. In general, people are aware that such states cause major and prolonged psychological distress but may not insure themselves psychologically against them for reasons outlined by Easterlin and others (e.g. overfocus on consumer goods).

Liam Delaney said...

In terms of why recent graduates would experience unemployment as aversive I think we need to go beyond economic reasons. It simply is not the case that this is due to a reduction in income though if I am proved wrong by later data I will admit this. I think we need to look microscopically at the unemployment state from start to finish to figure out what is going on. I will be putting out some qualitative results on this soon.

David O'Donnell said...

Very useful work. Keep it up.