When your neighbour wins the lottery

I’m not sure if the format of the Dutch postcode lottery is common, but it certainly creates some interesting incentives. In this lottery, a random postcode is drawn from the 430,000 postcodes in the Netherlands, with each postcode having, on average, 19 households. Each person in that postcode who has purchased a ticket in the lottery receives €12,500 for each ticket that they hold (people can buy more than one ticket), and one ticket in the postcode is awarded a BMW.

If your postcode is drawn but you do not own a ticket, you know with certainty that you would have won if you had purchased one. You will also know that there were other winners in your postcode, and given the typically small size of a postcode, it is likely that you know some of those winners. With 30% of the Dutch participating in the lottery, there is a good chance that one of those winners is your neighbour.

It was with information from this lottery that Peter Kuhn and colleagues decided (ungated working paper here) to look at two of the more interesting questions in economics. First, how do people react to a wealth shock? Do they smooth consumption over their lifespan, or do they blow it all at once? Second, how do other people react to this change in relative wealth? If my neighbour wins the lottery, does it change my behaviour even though it does not affect my absolute well-being?

Kuhn and colleagues obtained their data by surveying people covering four groups - winning lottery participants, people who lived in a postcode that won but who had not purchased a lottery ticket, lottery participants in postcodes who had not won, and those who do not play the lottery from those unsuccessful postcodes. This combination allowed the researchers to compare behaviour of those with and without lottery tickets in winning postcodes, while controlling for differences in characteristics between those who play the lottery or not by examining differences in unsuccessful postcodes. The researchers asked questions about a range of factors, including household composition, demographic variables, labor supply, happiness, car ownership, income and lottery participation. Questions were asked about both current behaviour and behaviour a year earlier, which was intended to capture behaviour six months before and after the lottery result.

On the question of how the income shock affects the behaviour of the lottery winner, the results supported the idea that people smooth consumption over their lifecycle as winning the lottery had no effect on most household expenditure. What they did find, however, was an increase in car and durable purchasers by the winners, which supports the idea that shifting the timing of purchases of durables is one way in which people smooth consumption. It is also consistent with the idea that people have self-imposed borrowing constraints.

The result for which this study is more famous is the effect on the neighbours. Having a neighbour win the lottery increases the unsuccessful person’s probability of purchasing a car in the next six months by around 7 per cent, which the authors consider large relative to the size of the effects on the lottery winners themselves. This also reduces the average age of their car (no surprise given they are more likely to have purchased a new one). Interestingly, there was no effect on happiness for either the lottery winners or their neighbours, which is problematic for a relative income based model of happiness.

It’s not easy to interpret this result. That the increase in consumption by neighbours occurred only in relation to a visible item of consumption - a car that the neighbour will surely see - is suggestive of what the authors call “keeping up with the van den Bergs”, but the transmission path is unclear. Are they copying the purchase of a car, or responding to the known change in relative income? I sense that the dynamic between a neighbour who knows they missed out on a certain lottery win if they had bought a ticket and their triumphant neighbour would be much different to the case where that income shock was from another source.

I’d like to see two separate experiments, one involving knowledge that a neighbour won the lottery (but without the car or durable expenditure by the winner) and a second with only the signs of expenditure. My (speculative) suspicion is that the lottery knowledge would dominate. An experiment tracking neighbour responses to new car purchases might also be interesting, as if visible consumables are what the neighbour responds to, the windfall may not be strictly required.

As an end note, I’d heard about this paper before I got around to reading it, and I was surprised at the gap between what I had heard and the strength of the author’s claims. I’d seen this result waved around a lot as a sign that people respond to relative income changes, but there are some complicating factors that make it difficult to interpret the causative pathway. The happiness result also does not help the “relative income matters” case (nor possibly my speculative suspicion).