Uncertainty treatment in input-output analysis
ifDakU
ifDakU
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
multiplier matrices), and f<strong>in</strong>ally statistically compares the difference between the last<br />
and the appropriate observed transactions flows. From his rather extensive<br />
experiments, <strong>in</strong>volv<strong>in</strong>g sample sizes n = 10, 20, 30, 50, 100, Social Account<strong>in</strong>g<br />
Matrices (SAMs) of Botswana and Korea, and different comb<strong>in</strong>ations of exogenous and<br />
endogenous number of <strong>in</strong>stitutions of the US SAM, Roland-Holst (1989) concludes that<br />
disaggregated multiplier estimates are unbiased, imply<strong>in</strong>g that „transactions tables, if<br />
they are observed with well-behaved (<strong>in</strong> this case normal) measurement error, will yield<br />
multiplier estimates centered on their true distribution‟ (p.720). Additionally, the author<br />
f<strong>in</strong>ds that if the Monte Carlo sample is large enough, the disaggregated multiplier<br />
estimates can be significantly more stable than their transactions matrix counterparts.<br />
In a similar study, Dietzenbacher (2006) suggested that Roland-Holst‟s (1989) sample<br />
was too small. Runn<strong>in</strong>g analogous Monte Carlo simulations, assum<strong>in</strong>g the transactions<br />
are unbiased and <strong>in</strong>dependently, normally distributed and us<strong>in</strong>g 30/34-sector IO tables<br />
of ten OECD countries and 128-sector IO table for the Netherlands, Dietzenbacher<br />
(2006) concludes that <strong>in</strong> practice the 'multiplier estimates are positively biased but the<br />
biases are negligibly small. The empirical relevance of this conclusion is that <strong>in</strong> carry<strong>in</strong>g<br />
out the typical <strong>in</strong>ter<strong>in</strong>dustry calculations we may proceed as if the multipliers are<br />
unbiased' (p.775).<br />
Rueda-Cantuche, Dietzenbacher, Fernández and Amores (RDFA, 2013) take the prior<br />
<strong>analysis</strong> one step further by impos<strong>in</strong>g stochastics on supply and use tables (SUTs)<br />
rather than on IO tables. The underly<strong>in</strong>g reason is that <strong>in</strong>termediate transactions,<br />
though historically obta<strong>in</strong>ed from published IO tables, nowadays are constructed from<br />
SUTs us<strong>in</strong>g certa<strong>in</strong> technology or sales structure assumptions. Without go<strong>in</strong>g <strong>in</strong>to all<br />
the details, 21 RDFA (2013) carry out Monte Carlo simulations us<strong>in</strong>g 2006 SUTs of<br />
Spa<strong>in</strong>, Italy, the Netherlands, Germany and F<strong>in</strong>land, and confirm the f<strong>in</strong>d<strong>in</strong>gs of<br />
Roland-Holst (1989) and Dietzenbacher (2006) for randomized <strong>in</strong>termediate flows that<br />
multipliers‟ biases do exist but are 'negligibly small'.<br />
Uncerta<strong>in</strong>ties <strong>in</strong> all components of the IO model with<strong>in</strong> a multi-region IO (MRIO) sett<strong>in</strong>g<br />
are exam<strong>in</strong>ed <strong>in</strong> a case study on the UK‟s carbon footpr<strong>in</strong>t by Lenzen, Wood and<br />
Wiedmann (2010). They first perturb the elements of the UK-MRIO table, gross<br />
<strong>output</strong>s, and sectoral CO 2 <strong>in</strong>tensities accord<strong>in</strong>g to their respective relative standard<br />
21 The procedure of SUTs randomization is more complicated than IO flows randomization. For example,<br />
the new issues that need to be dealt with <strong>in</strong>clude the follow<strong>in</strong>g: (a) the randomized SUTs need to be made<br />
consistent with each other, and (b) one has to make a choice of a technology or sales structure<br />
assumption <strong>in</strong> order to construct IO table from SUTs.<br />
35