Groucho Marxism

In a recently published article, the Canadian economist Blair Fix has argued that all empirical tests of the labour theory of value, henceforth LTV, are based on a pseudo-scientific approach he refers to as ‘metaphysical alchemy’. This is not the first time Fix has criticized the LTV. I dealt with his previous critique in another blog post, in which I argued that although Fix is correct in pointing out the flaws in other economists’ attempts to empirically validate the LTV, he is incorrect in arguing that this implies the LTV is fundamentally flawed as a concept. Here I will look at Fix’s recently published critique, which focuses on the work of the Pakistani Marxist economist Anwar Shaikh, who was the first to attempt to empirically validate the LTV.

The thrust of Fix’s argument is that economists of every persuasion misinterpret empirical data because they subconsciously impose their own ideological biases onto it. This results in them effectively seeing things in the data that aren’t really there, and this is what Fix refers to as ‘metaphysical alchemy’. Fix claims that Marxian economists such as Shaikh are guilty of this when they use empirical data to test the LTV. He also argues that national accounts tables used to test the LTV are particularly susceptible to this type of mis-analysis due to the many accounting identities present in the data. These identities create many correlations that can be easily misinterpreted as underlying causal relationships.

Fix’s definition of ‘metaphysical’ relates to concepts that cannot be directly observed. In the context of the LTV, he argues that ‘socially necessary labour’ is just such a concept, as it is unobservable and only gets ‘revealed’ through prices. The fundamental problem is that the LTV is a theory about commodities, but the data required to test it is not available at commodity level as it is kept private by firms. For this reason the LTV has only ever been tested at sector level. But a correlation between prices and labour time at sector level does not imply such a correlation holds at commodity level, as a sector-level correlation could just be a result of the number of commodities in each sector correlating with itself.

To make this a bit more rigorous: suppose our economy has m sectors, and for i = 1,2,…,m let q_i, p_i, and l_i be the quantity of commodities produced by sector i, average price of commodities produced by sector i, and average labour employed per commodity in sector i, respectively. Empirical tests of the LTV have demonstrated that P = (P_i) is correlated with L = (L_i), where P_i = p_iq_i is the total monetary value of output in sector i and L_i = l_iq_i is the total labour-time employed in sector i. But to test the LTV properly we need to show that p = (p_i) is correlated with l = (l_i). We cannot infer this from the fact that P is correlated with L, as both P and L are dependent on q = (q_i) and could be correlated simply for that reason.

In theory we can get around this problem by dividing P_i and L_i by q_i for each i. In practice, however, this q_i is not made publically available, so we cannot do this. It is difficult to see how we might get around this problem. However it is important to note that this is a problem with the data, not with the underlying theory. Fix seems to be arguing that the lack of available data with which to test the LTV somehow invalidates the theory itself; but that is obviously not the case. There are many historical examples of scientific theories that took years to be test empirically due to a lack of available data, but did not mean these theories were invalid. In fact Fix provides one such example himself.

As Fix notes in his introduction, the ancient Greeks proposed that matter is made up of fundamental ‘quanta’, but it was not until two millennia later that scientific advances allowed atoms to be observed, thereby proving them correct. The Greeks’ proposition was a metaphysical one, in Fix’s sense, as it related to a concept that could not be directly observed. But that does not mean that for 2,000 years the idea that matter is made up of quanta was fundamentally invalid; it just meant that it hadn’t been experimentally verified yet. This is the situation we currently find ourselves in with the LTV. At the present time, the LTV is best thought of as hypothesis that has yet to be tested empirically. The task in front of us is to figure out how to do that.