The magical 4 day work week

Limitations of using empirical evidence in determining policy

Every so often newspapers trot out the 4 day work week story. Spain just announced a pilot program where the government covers the expense of a pilot program. Spain is the second-least productive country in Europe, won’t this hurt productivity? No! According to a study, it’ll actually improve productivity, create jobs and raise wages. Simply put, Spain can’t afford not to do it! It’s all there, just read the study:

Empirical studies are increasingly used to justify policy decisions. They trump logical reasoning and questioning the results can get you branded as being anti-science. You don’t need an argument how a decreased work week will do all these goals, you just need it in the abstract of a published study.

I’m skeptical of using empirical evidence to justify policy. Too often it results in scientism, the promotion of science to form a narrative. When considering the validity of empirical evidence to argue a point, I try to ask the following questions:

Is there bundling of all good things?

In the study mentioned earlier, reduced hour work weeks would:

1. Created 560 thousand jobs (reducing unemployment by 2.6%)

2. Women would be primary beneficiary from new jobs

3. Wages increase 3.7%, labor share increase of 2.1%

4. Increase GDP 1.4%

These claims immediately drew my suspicions. These concepts aren’t necessarily tied together. It could be that productivity increases but wages don’t increase, as has happened across the industrialized world since the 1970s. Or maybe productivity would mean employers wouldn’t need to hire as many workers, or capital holders would seize the value of excess productivity. Also note that no negative side effects are mentioned in the abstract.

Or consider the vegan diet. Did you know a vegan diet:

1. Helps you lose weight

2. Helps you prevent cancer

3. Is how humans ate for most of our existence

4. Is the only way to save the world from environmental disaster

5. Is the only ethical way to eat

6. Grow the size of your son’s penis (?)

I’m sure some of these claims are true. But when bundled like that, I suspect someone is trying a little too hard to convince me. It could be that a vegan diet is beneficial for preventing cancer but not effective at weight loss. These claims don’t have to be tied together. Bundling all the benefits together creates a zealous undertone that undermines individual arguments.

This is the kitchen sink strategy to pushing policy. Most recent example is studies that show investment in green tech will create jobs, save money, drive innovation and save the environment. Be skeptical when there are no trade-offs and a wide range of benefits.

Would the opposite results be reported?

Did you know that having women on your board is good for your bottom line? It’s science:

Companies with women directors on their board also perform better than those without women by specific metrics. For example, when Fortune-500 companies were ranked by the number of women directors on their boards, those in the highest quartile in 2009 reported a 42 percent greater return on sales and a 53 percent higher return on equity than the rest

I don’t doubt this claim, but what if the opposite were true? Presumably since we need a study, we should be willing to accept the opposite result. Suppose for some reason, women on the board decreased returns. Would it be reported? Would it change anyone’s mind regarding policies that promote female board members?

It’s fine to promote a policy on egalitarian grounds. You can promote including women in the board room without resorting to empirical studies that show how great it will be for everyone. Unless you’re willing to accept the opposite conclusion, including empirical evidence amounts to “and also” bundling mentioned in the last section.

If the opposite were true, would you promote the reverse policy?

Suppose you accept empirical evidence as necessary when determining policy and you’re willing to accept the results, whatever they may be. Would you also be willing to promote opposite policy if unexpected results were found.

For instance, suppose the shorter work week decreased productivity, leading to a decrease in wages and and increase in unemployment. Would you be willing to promote policies that incentivized longer work weeks? Maybe bump up the number of hours required for full-time benefits or overtime pay?

If not, you shouldn’t be so keen on using empirical evidence to promote the policy you like.