Childhood obesity in the UK, is fast food exposure a factor?

It has often been suggested that the availability of fast food is a cause of increasing obesity prevalence.  As it is difficult to retrieve a causal effect of fast food on BMI when fast food is ubiquitous, we study an era when it is being introduced and some locations do not have exposure to it – or get it later - then by virtue of random or near random draw on geographical location we can observe the treatment effect of fast food independent of the process whereby people choose to live. Following this logic, our study exploits the beginning of the inception of fast food entry in Great Britain. This provides a natural experiment in terms of the dramatic geographic variation in the intensity of treatment of fast food exposure. In the years of 1977-1978 one particular fast food company experienced a sharp supply shock in the number of its outlets which provides us with greater confidence in the near random exposure to fast food.

We have collected data on the time and location of the openings of all fast food outlets in the UK between 1968 – 1986. Combining the fast food data with data on clinically measured BMI of the respondents from the 16 year follow-up of the 1970 British Cohort Survey (BCS), we estimate the effect of the proximity from one’s house and the durations since the openings of all fast food outlets established until 1986, on their BMI in 1986 and changes in BMI. Throughout all regressions we fail to find any evidence of fast food access having a positive effect on BMI. Moreover, we attempt to measure exposure in terms of duration and distance simultaneously. We generate an intensity measure by aggregating the time since the opening of the closest fast food outlets, divided by their respective distances from one’s house. Regressing adolescent BMI on our intensity measure along with testing several alterations of its functional form does not yield any support of a positive relationship of BMI and the exposure to fast food.

To assess whether fast food treatment was indeed as if randomly assigned, we have investigated in the determinants of fast food locations throughout the time period of our study using multiple datasets. We do not find a consistent pattern in observed area level determinants of fast food location as is commonly observed today. Neither do we find that the early wave of fast food establishments targeted obesity prone households or particularly unhealthy areas. Our results are robust when restricting the analysis to estimating the effect of distance to the outlets of the fast food company that experienced a sharp and unexpected supply shock. We also make use of an individual’s exogenous proximity to a fast food distribution centre as a predictor of fast food proximity in an IV- analysis.