In their review of the empirical literature on obesity, Nettle et al. choose to limit their analysis to evidence derived from food insecurity surveys. Theory and evidence from behavioural ecology, as the authors rightly note, suggest that human obesity should be responsive to cues that predict – or would have predicted, in human evolutionary history – prospective food shortages. But the epidemiological literature on food insecurity relies, for the most part, on qualitative survey data (collected retrospectively) about perceived food scarcity. Measurement error alone could explain the observed gender differences (might men be less willing than women to report an inability to reliably feed their families?), confounding variables are numerous and often unobservable, and even enthusiastic proponents of what the authors refer to as the insurance hypothesis (myself included) should be surprised that a correlation with obesity is so reliably observed.
Had the authors expanded their purview to include household-level economic insecurity – that is, uncertainty with respect to financial well-being – they would have found a wealth of evidence supporting the link to obesity (for reviews, see Smith [Reference Smith2009] and Wisman & Capehart [Reference Wisman and Capehart2010]). This literature, which also draws on theory and evidence from behavioural ecology, typically employs econometric methods aimed at identifying causal relationships between economic insecurity and obesity. Smith et al. (Reference Smith, Stoddard and Barnes2009), for example, show that working-age men in the United States gained more weight over a 12-year period when they faced a higher risk of job loss (regional unemployment rates were used to instrument for risk of job loss – effectively “selecting” those individuals who face said risk through no fault of their own), while Barnes et al. (Reference Barnes, Smith and Yoder2013) show that cohabitation with other adults can be protective against weight gain – provided those other adults have paying jobs.
One of the most compelling confirmations of this hypothesis appears to have been accidental: Ferrie et al. (Reference Ferrie, Shipley, Marmot, Stansfeld and Smith1995; Reference Ferrie, Shipley, Marmot, Stansfeld and Smith1998) report on a natural experiment that occurred during the long-running Whitehall II Study of British civil servants. In the early 1990s one of Whitehall II's 20 civil service departments was threatened with privatisation (and subsequently, was indeed privatised). The timing of the Whitehall II survey dates was such that data were collected both before and after discussions of privatisation began, and then again after privatisation occurred. Although the authors clearly had no specific hypothesis about employment security and obesity in mind (Whitehall II being designed to test more general hypotheses about workplace stress and health outcomes), they nevertheless report strong evidence: Of the many health measures reported, the most significant differences (for both men and women) among those working in the threatened department were increased body mass index and increased likelihood of excessive sleep. These differences were noted both before and after privatisation actually occurred, and were independent of actual changes in employment status.
The timing and magnitude of the obesity epidemic has varied dramatically from country to country, but the largest increases have generally been observed in countries, such as the United States, the United Kingdom, Iceland, New Zealand, and Australia, that have aggressively pursued neoliberal economic policies (Ljungvall Reference Ljungvall2013; Egger et al. Reference Egger, Swinburn and Islam2012; Smith Reference Smith, Offer, Pechey and Ulijaszek2012a; Vogli et al. Reference Vogli, Kouvonen, Elovainio and Marmot2014). These policies have typically included expansion of international trade, privatisation of public services, softening of regulatory restraints on industry, monetary policy emphasising price stability rather than full employment, and a weakening of labour protections and other aspects of the social safety net. Although these policies may have been well intended (typically being enacted with the stated purpose of expanding opportunities for mutually beneficial exchange), they also, arguably, have had the collective effect of increasing the burden of financial risk faced by households. In the United States there is growing evidence that both earnings volatility (Gottschalk & Moffitt Reference Gottschalk and Moffitt2009) and economic insecurity more generally (Hacker et al. Reference Hacker, Huber, Nichols, Rehm, Schlesinger, Valletta and Craig2014) have increased over the period that obesity rates have risen (roughly, 1980–present), and that obesity rates have risen most dramatically among demographic groups that have seen the largest increases in economic insecurity (Smith et al. Reference Smith, Stillman and Craig2016).
Even if it is true that economic liberalisation has helped cause the global obesity epidemic, there is an obvious alternative mechanism to be considered. One might expect that liberalisation, in addition perhaps to making life more risky for the individual citizen, also tends to reduce regulatory constraints on the food industry. This could result in increased obesity rates via either lower food prices or industry marketing practices (e.g., promotion, product formulation). In the most direct test of these competing hypotheses (insecurity vs. “fast food”) of which I am aware, Offer et al. (Reference Offer, Pechey and Ulijaszek2010) found, in a panel of developed countries, that insecurity appears to play a far larger role. I myself have offered a related conjecture, suggesting that the similarities in the physiological effects of modern fast food and what anthropologists call “famine foods” are not accidental. Rather, one might say that industry is simply serving up what humans have evolved to crave during periods of economic insecurity (Smith Reference Smith2012b).
As a trained economist (and erstwhile biologist), I have always harboured a natural sympathy for the notion that the market mechanism is an elegant way of allocating societal resources. Give people incentives, the argument goes, and they will perform! But years of studying the role of insecurity in the etiology of obesity have given me pause. In the workplace, every opportunity to succeed necessarily comes paired with a threat of failure. This leads me to another core tenet of economics: Every benefit must be weighed against its cost. It seems that some of my profession's most cherished policy prescriptions may have come at the cost of a global obesity epidemic.
In their review of the empirical literature on obesity, Nettle et al. choose to limit their analysis to evidence derived from food insecurity surveys. Theory and evidence from behavioural ecology, as the authors rightly note, suggest that human obesity should be responsive to cues that predict – or would have predicted, in human evolutionary history – prospective food shortages. But the epidemiological literature on food insecurity relies, for the most part, on qualitative survey data (collected retrospectively) about perceived food scarcity. Measurement error alone could explain the observed gender differences (might men be less willing than women to report an inability to reliably feed their families?), confounding variables are numerous and often unobservable, and even enthusiastic proponents of what the authors refer to as the insurance hypothesis (myself included) should be surprised that a correlation with obesity is so reliably observed.
Had the authors expanded their purview to include household-level economic insecurity – that is, uncertainty with respect to financial well-being – they would have found a wealth of evidence supporting the link to obesity (for reviews, see Smith [Reference Smith2009] and Wisman & Capehart [Reference Wisman and Capehart2010]). This literature, which also draws on theory and evidence from behavioural ecology, typically employs econometric methods aimed at identifying causal relationships between economic insecurity and obesity. Smith et al. (Reference Smith, Stoddard and Barnes2009), for example, show that working-age men in the United States gained more weight over a 12-year period when they faced a higher risk of job loss (regional unemployment rates were used to instrument for risk of job loss – effectively “selecting” those individuals who face said risk through no fault of their own), while Barnes et al. (Reference Barnes, Smith and Yoder2013) show that cohabitation with other adults can be protective against weight gain – provided those other adults have paying jobs.
One of the most compelling confirmations of this hypothesis appears to have been accidental: Ferrie et al. (Reference Ferrie, Shipley, Marmot, Stansfeld and Smith1995; Reference Ferrie, Shipley, Marmot, Stansfeld and Smith1998) report on a natural experiment that occurred during the long-running Whitehall II Study of British civil servants. In the early 1990s one of Whitehall II's 20 civil service departments was threatened with privatisation (and subsequently, was indeed privatised). The timing of the Whitehall II survey dates was such that data were collected both before and after discussions of privatisation began, and then again after privatisation occurred. Although the authors clearly had no specific hypothesis about employment security and obesity in mind (Whitehall II being designed to test more general hypotheses about workplace stress and health outcomes), they nevertheless report strong evidence: Of the many health measures reported, the most significant differences (for both men and women) among those working in the threatened department were increased body mass index and increased likelihood of excessive sleep. These differences were noted both before and after privatisation actually occurred, and were independent of actual changes in employment status.
The timing and magnitude of the obesity epidemic has varied dramatically from country to country, but the largest increases have generally been observed in countries, such as the United States, the United Kingdom, Iceland, New Zealand, and Australia, that have aggressively pursued neoliberal economic policies (Ljungvall Reference Ljungvall2013; Egger et al. Reference Egger, Swinburn and Islam2012; Smith Reference Smith, Offer, Pechey and Ulijaszek2012a; Vogli et al. Reference Vogli, Kouvonen, Elovainio and Marmot2014). These policies have typically included expansion of international trade, privatisation of public services, softening of regulatory restraints on industry, monetary policy emphasising price stability rather than full employment, and a weakening of labour protections and other aspects of the social safety net. Although these policies may have been well intended (typically being enacted with the stated purpose of expanding opportunities for mutually beneficial exchange), they also, arguably, have had the collective effect of increasing the burden of financial risk faced by households. In the United States there is growing evidence that both earnings volatility (Gottschalk & Moffitt Reference Gottschalk and Moffitt2009) and economic insecurity more generally (Hacker et al. Reference Hacker, Huber, Nichols, Rehm, Schlesinger, Valletta and Craig2014) have increased over the period that obesity rates have risen (roughly, 1980–present), and that obesity rates have risen most dramatically among demographic groups that have seen the largest increases in economic insecurity (Smith et al. Reference Smith, Stillman and Craig2016).
Even if it is true that economic liberalisation has helped cause the global obesity epidemic, there is an obvious alternative mechanism to be considered. One might expect that liberalisation, in addition perhaps to making life more risky for the individual citizen, also tends to reduce regulatory constraints on the food industry. This could result in increased obesity rates via either lower food prices or industry marketing practices (e.g., promotion, product formulation). In the most direct test of these competing hypotheses (insecurity vs. “fast food”) of which I am aware, Offer et al. (Reference Offer, Pechey and Ulijaszek2010) found, in a panel of developed countries, that insecurity appears to play a far larger role. I myself have offered a related conjecture, suggesting that the similarities in the physiological effects of modern fast food and what anthropologists call “famine foods” are not accidental. Rather, one might say that industry is simply serving up what humans have evolved to crave during periods of economic insecurity (Smith Reference Smith2012b).
As a trained economist (and erstwhile biologist), I have always harboured a natural sympathy for the notion that the market mechanism is an elegant way of allocating societal resources. Give people incentives, the argument goes, and they will perform! But years of studying the role of insecurity in the etiology of obesity have given me pause. In the workplace, every opportunity to succeed necessarily comes paired with a threat of failure. This leads me to another core tenet of economics: Every benefit must be weighed against its cost. It seems that some of my profession's most cherished policy prescriptions may have come at the cost of a global obesity epidemic.