The Strategic Failure

There is a distinction that organizes much of the discussion on productive development and that, nevertheless, is almost never made explicit. We call industrial policy the set of measures a State deliberately directs at the productive apparatus: tariffs, promotional regimes, subsidized credit, public procurement, tax incentives. We call strategy something else: the way those measures are coordinated with each other and with decisions taken outside the industrial perimeter — exchange rate, science policy, infrastructure, technical education — so as to push the productive structure in a chosen direction. These are independent planes. A country can have well-designed instruments and no direction; it can also have a clear direction and fail at execution. We argue here that Argentina built, over nearly half a century, a dense and costly industrial-policy apparatus, and that the central problem of that experience sits on the second plane: the absence of a selection criterion that would hierarchically order the instruments toward a specialization trajectory. The consequence did not end in 1976 or in 1991. If the argument is correct, it should still be observable today in the composition and behavior of the local capital market, and that is the quantitative hypothesis we close with.

Two distinct grounds for intervention

It is worth clearing up a persistent misunderstanding first. Neoclassical theory does offer grounds for intervention: it does so in the form of market failures — imperfect competition, public goods, externalities — and it admits, since the late eighties, a symmetric literature on government failure, which warns that intervention may fail to solve the problem or make it worse, whether because the official pursues their own objectives or because they lack the necessary information. The underlying disagreement, then, is not whether to intervene or not. It is something else: for the structural-change approach, rooted in structuralism and neo-Schumpeterian thought, the reason to intervene is not that the market deviates from an optimum but that its normal functioning consolidates the position a country occupies in the international division of labor. That an economy inserts itself by supplying primary goods is not a failure; it is the expected outcome of its endowment and its history. The question is whether that position is the desired one.

Prebisch formulated the problem with a precision that is sometimes lost in the popularized versions. The difficulty does not lie in producing primary goods instead of industrial goods, but in specializing in activities where rising efficiency is not accompanied by rising income. A country can adopt the most advanced available technology, raise its productivity, and impoverish itself, if it produces goods that are replicable at any latitude, prone to overproduction, under a competitive market structure. Structural change means, in that key, moving toward activities where technical progress is appropriable as rent. Defined that way, the concept detaches from the manufacturing sector as an accounting category: contemporary industrial policy should ask which activities — industrial or not — allow capturing innovation rent, rather than taking for granted that the answer is manufacturing industry.

History backs this second approach more than the first. Chang (2013) documents the gap between discourse and practice in today’s developed countries: nineteenth-century British free trade was a consequence, not a cause, of four centuries of deliberate industrial promotion. Kicking away the ladder names exactly that move — using the instrument and then preaching against it. The observation is empirical and requires no commitment to any theory of value: the countries that made development leaps in a few decades used selective industrial promotion.

Selectivity, sequencing, reciprocity

What distinguishes the successful experiences from the unsuccessful ones is not the presence or absence of protectionism, nor of foreign investment, nor of import substitution. All three elements were present in both East Asia and Latin America. Fajnzylber (1983, ch. 2) shows that South Korea and Taiwan applied import substitution intensively, with prohibitions, quotas, and tariffs calibrated to each stage’s need for foreign currency. The difference lies in three attributes of the design. It was selective, in that it prioritized identified sectors. It was sequential, because the prioritized sectors changed with the five-year plans and openness was administered gradually. And it was reciprocal: Hikino and Amsden (1995) show that the subsidy was tied to verifiable performance standards, typically export targets, with an effective penalty for non-compliance. The State did not distribute resources; it exchanged them for measurable results.

Amsden and Hikino further contribute the conceptual framework that, in my view, best organizes the problem. Late-industrializing countries did not compete with their own technology but as learners: they borrowed and improved already-created technology, aiming deliberately at intermediate-technology industries, where international technology supply and growing global demand both exist. That learning paradigm contains a paradox that is rarely underlined: the faster an economy converges toward the frontier, the sooner the opportunities to grow by borrowing are exhausted. The leap to the innovation paradigm — generating one’s own capabilities — is not guaranteed by the prior success of learning, and is in fact where several Asian trajectories hit their ceiling.

Bell (1984) adds the warning that keeps all of this from being read as a recipe. The maturation of an infant industry is not reached passively: it demands explicit technological effort, and that effort does not guarantee the capability. Worse still, the infant industry must converge against a moving target, because the productivity of mature foreign firms also grows; reaching international costs at a given moment is not enough if the slopes equalize afterward, because then the initiation costs are never recovered. Bell further notes that the available evidence on infant industries in developing countries is scarce and of uneven quality. This is a real evidence gap and it is worth stating plainly: much of the debate on protection rests on case studies, not on a comparable statistical body.

Argentina: instruments without hierarchy

The Argentine case is not one of absence of policy. Katz and Kosacoff (1989) describe a dense sequence of instruments since 1930: exchange controls, prior permits, tariffs, the Banco de Crédito Industrial, the IAPI, the national-interest industries regime, the 1958 foreign investment law, BANADE in 1970, provincial promotional regimes. The decisive turn takes place in the late fifties, when deepening toward complex sectors is resolved through foreign direct investment in a protected market, without export conditionality or technology-transfer requirements. Fajnzylber calls this frivolous protectionism, as opposed to protectionism for learning: smaller-scale reproduction, with different equipment, of the productive structure of the country of origin, without local innovation, in a captive market. And his diagnosis is precise in assigning responsibility: the problem was not the conduct of the multinationals, which maximized profits as one would expect, but the regulatory omission of the domestic agents who should have induced the right conduct.

Azpiazu (1989) takes the analysis down to the level of instrument design, and there the result is even more uncomfortable. National and provincial promotional regimes coexisted with different enforcement authorities and overlapping benefits, which generated direct incentives to evade. Tax deferral without indexation, under inflationary conditions, diluted the investor’s fiscal burden and shifted financing onto the State. Exemption from import duties on capital goods, with no compensation to the local producer, penalized precisely the sector that Fajnzylber identified as strategic for its multiplier effect and for the kind of knowledge it mobilizes. The VAT rebate on sales subsidized domestic placement and operated, in practice, as an anti-export bias. Azpiazu himself describes the set as a juxtaposition of diverse benefits that indiscriminately promotes any capital formation: a good share of the projects were relocations of pre-existing plants, not capacity expansion.

The diagnosis repeats itself, with different vocabulary, in the period after 2001. Lavarello describes the accumulation of overlapping programs as geological layers: initiatives that pile up without dismantling the previous ones and without a hierarchical body to order them. Fiscal and financial resources directed at industry went from an average of 0.9% of GDP between 2004 and 2006 to 1.4% between 2010 and 2013, and four ministerial strategic plans coexisted — industrial, innovation, agro-industrial, federal planning — each coherent on its own and none articulated with the others. There was policy. What was missing was the body that decided what came first.

The friction the argument does not resolve

Two facts are inconvenient for this reading, and it is worth putting them on the table. The first: in 1955 Argentina had a net manufacturing value per capita of 145 dollars against Korea’s 8, and a ratio of net industrial to agricultural product of 1.32 against 0.20 (Hikino and Amsden, 1995, table 4). Argentina’s industrial starting point was overwhelmingly higher. In 1913 Argentine per-capita GDP was the highest among fifteen late-industrializing countries, at 1,770 1980-international dollars; by 1987 it had fallen to fifth place with 3,302, while Taiwan went from 453 to 4,744 and Korea from 610 to 4,143. What needs explaining is not an industrialization that never happened, but one that happened early and was not sustained. The second fact: the 1964-1974 phase grew without interruption at rates of up to 8% a year, with rising exports of manufactures of domestic origin and technology, under the same institutional architecture we are criticizing here. A thesis that attributes everything to a lack of strategic coordination does not comfortably explain that decade. It is likely that the explanation is, partly, macroeconomic and, partly, political: the cycle is cut short by distributive conflict and institutional interruption rather than by exhaustion of the accumulation model.

The hypothesis: reading structural change in asset prices

Here we move from the literature review to a conjecture of our own, not tested in this text. If the absence of structural change were a structural fact and not a narrative, it should have an observable signature in the capital market, because the composition of the export basket determines the composition of the foreign-currency flow and, through that channel, the sensitivity of local assets to the commodity price cycle.

The central hypothesis is that the factor loading of Argentine equity returns on a terms-of-trade factor stays stable or grows over the last three decades, while in regional peers with more diversified baskets that loading declines. The natural specification is a regression of monthly returns of the local index measured in dollars — with the non-trivial caveat of the implicit exchange rate during periods of exchange controls — against four factors: the Chicago soybean price, the dollar index, the sovereign spread, and a global emerging-markets factor. Estimating rolling 36-month betas over 1996-2025 and comparing the Argentine profile against Brazil, Mexico, and Chile, the structural argument predicts the absence of a declining trend in the Argentine commodity beta. A useful complement is crossing those betas with each country’s economic complexity index: if the mechanism is the one we posit, the slope between complexity and sensitivity to the commodity cycle should be negative and significant in the panel.

The identification problems are serious and I do not minimize them. The sample is short for inference on thirty-year processes. The Argentine panel is small, illiquid, and survivorship-biased, and its sectoral composition changed because of regulatory events unrelated to the productive structure. The complexity index is annual, built from trade data that carries its own measurement error, and is endogenous with respect to everything else. A stable beta is consistent with the hypothesis but also with alternative explanations, starting with the most obvious one: that sovereign risk dominates variance to such a degree that no real factor stands out. The contribution of the exercise would not be to prove the structuralist thesis but to put an observable price on it, and to turn into falsifiable a claim that today circulates mostly as narrative.

Explicitly outside the scope of this note is the evaluation of Argentine industrial policy after 2015, for which I did not work with primary sources, and the discussion of the concrete institutional design of a coordination agency, which is a political-economy problem and not one of development theory.

---

References

Azpiazu, D. (1989). La promoción a la inversión industrial en la Argentina. Efectos sobre la estructura industrial 1974-1987. CEPAL.

Bell, M. (1984). Assessing the performance of infant industries. Journal of Development Economics. [INCOMPLETE REFERENCE: volume and pages]

Chang, H-J. (2013). Kicking Away the Ladder: The Real History of Free Trade. [INCOMPLETE REFERENCE: edition consulted]

Fajnzylber, F. (1983). La industrialización trunca de América Latina. Nueva Imagen. Chs. 1, 2, and 3.

Hikino, T. and Amsden, A. H. (1995). La industrialización tardía en perspectiva histórica. Desarrollo Económico, 35(137), 3-34.

Katz, J. and Kosacoff, B. (1989). El proceso de industrialización en la Argentina: evolución, retroceso y prospectiva. CEAL. Chs. 3 and 4.

Kosacoff, B. and Ramos, A. (1997). Consideraciones económicas sobre la política industrial. CEPAL.

Lavarello, P. Políticas industriales en Argentina, 2003-2015. [INCOMPLETE REFERENCE: exact title, year, and publisher]

Seurot, F. (1986). La planificación en las economías socialistas. Chs. 2 and 3. [INCOMPLETE REFERENCE]