Here’s an interesting paper from Pat Kline and Enrico Moretti on local economic development, agglomeration, and the Big Push.
We study the long run effects of one of the most ambitious place based economic development policies in U.S. history: the Tennessee Valley Authority (TVA). Using a rich panel dataset of counties, we conduct an evaluation of the dynamic effects of the TVA on local economies in the seventy years following the program’s inception. We find that the TVA led to short run gains in agricul- tural employment that were eventually reversed, while impacts on manufacturing employment continued to intensify well after the program’s subsidies had lapsed – a pattern consistent with the presence of agglomeration economies in the manufacturing sector. Economists have long cautioned that the local gains created by place based policies may be offset by losses elsewhere, yielding ambiguous effects on the U.S. as a whole. We develop a novel approach to assessing the aggregate consequences of place based policies. Our findings suggest that the TVA boosted national manufacturing productivity by roughly 0.3% and that the dollar value of these productivity gains substantially exceeded the program’s costs.
Here’s how they conclude:
This paper makes two primary contributions. Our substantive contribution is to estimate the local and aggregate effects of one of the largest place based policies in U.S. history. To our knowledge, we are the first to empirically quantify the long run social costs and benefits of a place based policy. A second contribution is methodological: we have developed a tractable empirical framework for evaluating the aggregate welfare effects of placed based policies, with the potential to be applied to many other settings.
Our empirical findings are policy relevant. The evaluation designs of Section 3 provide strong evidence that the TVA sped the industrialization of the Tennessee Valley and provided lasting benefits to the region in the form of high paying manufacturing jobs. Notably, the impact on manufacturing employment persisted well beyond the lapsing of the regional subsidies, suggesting the presence of powerful agglomeration economies. By contrast, the agricultural sector, which is unlikely to exhibit substantial agglomeration forces, retracted dramatically once subsidies terminated.
Our analysis in Section 6 suggests the TVA substantially raised the productivity of the U.S. manufacturing sector by roughly 0.3% between 1940 and 1960. We estimate that the stream of benefits associated with this increase exceeded the program’s costs, though this conclusion rests on several unverifiable assumptions regarding the functioning of labor markets.
Most of the national impact of the TVA on worker welfare is accounted for by the direct effects of the program’s vast investments in public infrastructure. Our finding of a roughly constant agglomeration elasticity suggests the program’s indirect effects were minimal. A noteworthy implication is that although agglomeration economies represents an important market failure at the local level, this failure does not provide a rationale for federal intervention in the spatial distribution of manufacturing activity.
We caution, however, that our findings do not necessarily apply to all contexts, as the strength and shape of agglomeration economies may well vary across industries, periods and levels of aggregation. Our results are specific to the manufacturing sector and a period of U.S. history when manufacturing employment was expanding and earnings were relatively high. An important task for future work is to assess whether similar qualitative results hold for modern development efforts, such as those centered on building high tech clusters.