A howlingly weird finding

Something which I absolutely do not understand from Dani Rodrik:

The main message of the scatter plot can be summarized easily. Labor has moved from tradable sectors (agriculture and manufacturing) where labor productivity growth has been rapid to non-tradables, especially wholesale and retail trade, where productivity growth has been negative.

This is the explanation for Latin America\’s dismal growth in relation to E Asia\’s.

But here\’s what I don\’t understand at all. Supermarkets have been expanding at a rapid rate across Latin America.

From perhaps 10% of sales in 1990 to 50-60 % in the mid 2000\’s. And India and Vietnam, which have, according to Rodrik, much higher labour productivity growth, have had much lower penetrations of supermarkets (2-20% over the same period).

Now, OK, supermarkets are not the only components of the wholesale and retail trades. But they\’re a useful indicator of them at least.

And yes, there\’s all sorts of things people do complain about supermarkets: like that they put small shops out of business. But that (and it\’s one of the things we think we know about modern retail, isn\’t it?) means that we would expect rising labour productivity in that sector.

Supermarkets are, at least we think they are and certainly see they are in developed economies, (vide nef\’s calculation that £250,000 of spending in local shops creates 5 jobs while the same spending in a supermarket creates 1), users of less labour: the same thing as saying they have higher labour productivity.

So, to the extent that we can use supermarkets as being representative of wholesale and retail trades, we seem to be saying that a technology (and yes, supermarkets are a technology) which we\’re pretty certain increases labour productivity is in fact reducing labour productivity.

Which is, umm, weird.

Now Rodrik is an economist and as we all know I\’m not. So I\’m entirely buggered if I know what\’s going on here.

A couple of ideas occur:

1) The amount of labour previously being used is mismeasured. Given that much would have been in the informal sector, as it becomes formalised we might see the apparent amount of labour being used rise but that is due to our earlier guesses of how much was in fact being used originally.

2) That 1) is nonsense but that the output of the new system is being mismeasured. As an example, to take those UK figures. We might say that £250,000 of sales is the same output, whether delivered by small shops or a supermarket. But is it actually? If the effect on the individual shopper is different then not necessarily. For it might take 45 minutes to do the shopping in a supermarket but 3 hours (entirely made up numbers) in a series of small shops. And if we say that £250k of sales is the same by either method, well, no, it\’s not, is it? We\’ve saved the consumer 2 hours 15 minutes so the consumer experience (umm, the consumer surplus?) of the supermarket is much greater than that of the small shops. Meaning that we\’re not correctly measuring the output of the supermarket labour, because we\’re just measuring it by sales and not the actual product of that labour: ie, the time saved by consumers.

3) Is linked to 2). That this new technology of supermarkets allows consumers access to a much wider range of goods. Chicken, beef, carrots, onions, are all always available while the diet from the small shops was, say, beans and rice only. Again, we\’re underestimating the value to the consumer and thus undercounting the productivity of the retail labour.

4) Supermarkets really do have lower labour productivity than Mom and Pop stores and wholesalers.

5) I dunno….anyone who does have a clue want to weigh in here? For it seems entirely bizarre to conclude that achieving the same result (the distribution of food and goods to the population) through a technology which we\’re pretty sure uses less labour will lead to a reduction in labour productivity.

4 thoughts on “A howlingly weird finding”

  1. Is it because as you move from small family type businesses to larger corporate ones, you get more accurate reporting of amount of labour used? Mom & Pop probably work much more than a 40 hour week each running their own store, plus they get the kids to mind the till when they go to the wholesaler, or do some paperwork. Whereas a big comapny has to employ someone on fixed hours to do all the work, which shows up in the employment figures.

    I know in my line of business (farming) most self employed farmers work well over the normal hours, their wives similarly, and family members are cheap/free labour as well. I’ve filled in employment census forms too, and the information they get given is pure guesswork really. So the quality of the stats for small businesses is definitely iffy, even in the UK. In S America even more so I suspect.

  2. could it have something to do with capital intensity? If Rodrik is not talking about labour productivity but instead total factor productivity, then it might be easier to understand.

    to illustrate, so say:

    mom+pop L=10, K=1, TFP=10:


    supermarket L=5, K=5, TFP=5:


  3. cancel that! he is talking about labor productivity.

    hmm. well capital intensity still makes a difference to that …. but I’m as puzzled as you are

  4. Rodrik answers:
    In response to Tim Worstall’s puzzlement, the modern supermarkets he has in mind make up a tiny proportion of employment in “wholesale & retail trade.” The bulk of employment in this sector is informal, which is why productivity is low and has plummeted further as the ranks of informality have swelled.

Leave a Reply

Your email address will not be published. Required fields are marked *