Epistemic Status: somewhat confident: I may have made coding mistakes. R code is here in Markdown (with plots), or here in Github if you feel like checking.

Introduction: 

In their 2022 article, Matthew Killingsworth and Daniel Kahneman looked to reconcile the results from two of their papers. Kahneman (2010) had reported that above a certain income level ($75,000 USD), extra income had no association with increases in individual happiness. Killingsworth (2021)  suggested that it did. 

Kahneman and Killingsworth (henceforth KK) claimed they had resolved this conflict. They hypothesized correctly that:  

1. There is an unhappy minority, whose unhappiness diminishes with rising income up to a threshold, then shows no further progress 
2. In the happier majority, happiness continues to rise with income even in the high range of incomes 

1) refers to Kahnemann’s leveling off finding; 2) refers to Kllingsworth's continued log-linear finding.

(More info on this discussion can be found in Spencer Greenberg’s thoroughly enjoyable blog post. Spencer goes into the correlation/causation debate which I don't go into here.)

Here, I reproduce these findings from KK (2022) using a different dataset – the 2012 Health Survey for England. This dataset is smaller (n= 7,179, rather than 33,391), and uses a different collection technique. The NHS’s wellbeing variable was collated from 14 individual responses to a self-questionnaire, whilst the KK variable uses experience-sampling: i.e. pinging participants at random points in the day. KK regard experience sampling as the gold-standard, and whilst the NHS data had around 100,000 data points on wellbeing, the KK data had over 1.7 million experience-sampling reports.

So, you might be thinking: why bother to replicate the paper with worse NHS data? The first answer is a mixture of convenient access through my uni and personal curiosity. 

Though, I also think this analysis had some useful insights for others too. The KK paper looked solely at hedonic wellbeing – asking participants, “how do you feel right now?”. Whereas, the wellbeing scale here (WEMWBS) looks to capture hedonic and eudaimonic aspects, including questions about immediate emotional experience and life satisfaction. (More about the variable is here.^[1]) Given this, taken together with other minor differences – e.g. surveys versus experience pings; UK adults, rather than US employees –I felt it would be somewhat surprising if the KK findings replicated on the NHS data. 

… And, they did. 

Summary of Findings

Overall, I found similar results to Killingsworth and Kahnemann (2022) in 4 notable respects:

1. There is a linear relationship between log income and median happiness, although this relationship is small 
2. For a large proportion of the population (i.e. at p=50, 70, 85), this log-linear relationship continued above a fairly high income threshold (£50,000 in 2012), and: 
3. For a small minority (p=85) the slope of the log-linear graph  increased slightly (although not significantly)
4. For a large minority of the population (p = 5, 10, … 35), extra income had no association with increased levels of happiness. 

This constitutes a small update towards the KK findings, and thus against the view that there is a ‘happiness ceiling’ which the happiest in society have already reached.

Results

Before going into the results, I want to clarify two terms here: 

1. Relative Leveling Off: the slope of the happiness against (log) income plot decreases after a certain income threshold. 
2. Absolute Leveling Off: the slope of the happiness against (log) income plot is indistinguishable from 0 after a certain income threshold.  

The 2022 paper uses an income threshold of $100,000. Accounting for historic exchange rates and inflation, I use an income threshold of £50,000. This is roughly equal to $110,000 today. ^[2]

KK use ‘quantile regression’. This is like OLS, but estimates a certain percentile (not necessarily a quantile) of a response variable rather than the mean.

First, I investigate leveling off at the median, then at various percentiles. 

Median Regressions

Figure 1 gives the results from KK (2022), and my results (below). Both estimate median wellbeing in the population, which is standardised.

Figure 1

The overall shape I found look fairly similar. Both slopes – before and after £50k – are small. The 2022 paper found that the difference in median happiness for households with income $15,000 and $250,000 was about five points on a 100-point scale. I found a corresponding difference of around 4.85 on a 14–70 point-scale for £15,000 and £250,000.  

I found more evidence of relative leveling off after £50,000 – although the slopes are statistically indistinguishable at 5% (p = 0.055). However, the slope above £50,000 was  significantly greater than 0 (i.e. no absolute levelling off). 

Regressions at Various Percentiles

In Figure 2, above, I give results from Kahnemann and Killingsworth. 

They found no relative or absolute leveling off amongst the majority of the population. For the 30th, 50th, and 70th percentile the slopes were statistically indistinguishable before and after the income threshold, and were significantly greater than 0: i.e. their curves look just like the global log-linear curves found above. 

The 15th percentile saw absolute and relative leveling off – the slope decreased above £50,000 and was statistically indistinguishable from 0. 

Figure 2

• Note: t-values are given for the slope for individuals with income above the threshold, not for the interaction. This is unclear in the KK graph. 

In Figure 2, below, I give my results for the same percentiles. Overall, I found a similar shape to KK – that is, for incomes below £50,000, happier percentiles had lower slopes than less happy percentiles; for incomes above £50,000 happier percentiles generally had higher slopes.  

(KK found monotonic increases in the slopes across percentiles above the income threshold; I didn't.)

There was evidence of relative levelling off, with significant decreases in slope above £50,000 for the 5th, 10th …, 35th percentiles (see Table 1). 

I also found an increase in slope above the income threshold for 85th percentile, approaching significance (p = 0.13). However, this does not mean that the marginal utility of each extra £1 increases, contrary to common economic assumptions, since the x axis is in log units.

There was more evidence of absolute levelling off: none of the slopes for percentiles 5, 10, ..., 35 being significantly positive. In contrast, slopes for the 50th, 70th,and 85th percentiles were positive. This is indicative of a happy majority for which extra money is associated with extra income. 

Some of the slopes above £50,000 had 0 slopes and t-values, which seems slightly strange. I suspect this could be an issue with a lack of data – there were only 1332 respondents with income above £50,000.

Implications

To recapitulate, I found similar results to Killingsworth and Kahnemann (2022) in 4 notable respects:

1. There is a linear relationship between log income and median happiness, although this relationship is small 
2. For a large proportion of the population (i.e. at p=50, 70, 85), this log-linear relationship continued above a fairly high income threshold (£50,000 in 2012; >$100,000 today), and: 
3. For a small minority (p=85) the slope of the log-linear graph  increased slightly (although not significantly)
4. For a large minority of the population (p = 5, 10, … 35), extra income had no association with increased levels of happiness.  

This was notable because the NHS happiness data used a variable which looked to capture eudaimonic and hedonic aspects, and was collected using a different technique in the UK.

So it was somewhat surprising I got similar results. Particularly, since replications in psychology/economics are fairly rare. So, this led me to slightly update towards the Killingsworth and Kahnemann paper’s results. 

I think the most interesting implication of the KK paper comes from what we can draw from (3) and (4). As they write, 

The acceleration at the 85% percentile and the flattening at the 15th percentile are contrary to the trend that would be expected from floor or ceiling effects (e.g., a ceiling effect would predict a diminishing slope at the high end of the happiness distribution, not an acceleration).

That is, there isn’t a ceiling to happiness which the happiest in society are nearing – continued percentage increases in income from economic growth may lead to higher levels of happiness (from 3). Continued economic growth may also more inequality of happiness, given that for the unhappy minority (like 20-35% of the population), extra income has no effect above a certain level (from 4). 

Table 1: Happiness at Different Percentiles (above, KK; below, me)
 

1. ^{^}

   I have not looked into this variable very deeply – documentation for NHS data is here and the first paper to describe WEMWBS is here 

2. ^{^}

   Kahneman’s original paper from 2010 uses a threshold of $75,000. Accounting for inflation, in 2012, this was worth $80,264, or around £51k 2012 UK pounds (at historic exchange rates £1 = $0.631 in 2012. Call it £50k for aesthetics.