Beyond Intermediates: The Role of Consumption and Commuting in the Construction of Local Input-Output Tables

It is a well-established fact in the literature on simulating Input-Output tables that mechanical methods for estimating intermediate trade lead to biased results where cross-hauling is underestimated and Type-I multipliers are overstated. Repeated findings to this effect have led to a primary emphasis on advocating the accurate estimation of intermediate trade flows. This paper reviews previous research and argues for a qualification of the consensus view: When simulating IO tables, construction approaches need to consider spill-over effects driven by wage and consumption flows. In particular, for the case of metropolitan economies, wage and consumption flows are important if accurate Type-II multipliers are to be obtained. This is demonstrated by constructing an interregional Input-Output table, which captures interdependencies between a city and its commuter belt, nested within the wider regional economy. In addition to identifying interdependencies caused by interregional intermediate purchases, data on subregional household incomes and commuter flows are used to identify interdependencies from wage payments and household consumption. The construction of the table is varied around a range of assumptions on intermediate trade and household consumption to capture the sensitivity of multipliers.

Development Accounting with Intermediate Goods

Do intermediate goods help explain relative and aggregate productivity differences across countries? Three observations suggest they do: (i) intermediates are relatively expensive in poor countries; (ii) goods industries demand intermediates more intensively than service industries; (iii) goods industries are more prominent intermediate suppliers in poor countries. I build a standard multi-sector growth model accommodating these features to show that inefficient intermediate production strongly depresses aggregate labor productivity and increases the price ratio of final goods to services. Applying the model to data, low and high income countries in fact reveal similar relative efficiency levels between goods and services despite clear differences in relative sectoral labor productivity. Moreover, the main empirical exercise suggests that poorer countries are substantially less efficient at producing intermediate relative to final goods and services. Closing the cross-country efficiency gap in intermediate input production would strongly narrow the aggregate labor productivity difference across countries as well as turn final goods in poorer countries relatively cheap compared to services.