{"id":3636,"date":"2015-06-20T14:25:48","date_gmt":"2015-06-20T14:25:48","guid":{"rendered":"https:\/\/journal.c2er.org\/?p=3636"},"modified":"2018-04-30T21:27:19","modified_gmt":"2018-05-01T01:27:19","slug":"questioning-the-value-of-economic-multiplier-estimates","status":"publish","type":"post","link":"https:\/\/journal.c2er.org\/2015\/06\/questioning-the-value-of-economic-multiplier-estimates\/","title":{"rendered":"Questioning the Value of Economic Multiplier Estimates"},"content":{"rendered":"

Edward G. Keating, Irina Danescu, and Robert Murphy<\/h2>\n

RAND Corporation, Santa Monica, CA<\/h2>\n

Introduction<\/h3>\n

In the fall of 2014, the Australian Department of Defence asked the RAND Corporation to assess the economic consequences of shipbuilding.[1]<\/a>\u00a0 This tasking was part of a larger project on the Australian naval industrial base examining whether Australia should import ships from foreign manufacturers or build them in Australia. As part of RAND\u2019s inquiry, we conducted case studies of the economic impact of Newport News Shipbuilding (NNS) in Newport News, Virginia and Austal USA shipbuilding in Mobile, Alabama. Our insights are drawn from interviews with subject-matter experts (both employed by and unaffiliated with the shipbuilders), as well as an analysis of publicly available \u2013 e.g., government, corporate annual report \u2013 data concerning both firms.<\/p>\n

Concurrently, we reviewed the extensive literature on economic multipliers.\u00a0 Commonly used economic multiplier software packages include the Regional Input-Output Modeling System (RIMS) and Impact Analysis for Planning (IMPLAN). Though widely used, we were struck by the relative lack of value of economic multiplier estimates.\u00a0 Most centrally, estimated economic multipliers do not<\/em><\/strong> lend themselves to the seemingly desirable usage:\u00a0 Just because one has estimated a multiplier of, say, 2.0, it does not logically follow that a $100 million new project would have a $200 million net impact on its community.\u00a0 The net impact could be greater than $200 million or, perhaps more likely, less than $200 million.<\/p>\n

But, in fact, the actual impact is so context-dependent that we fundamentally question the value of economic multiplier estimates. Here we provide an overview of economic multipliers, how they work, and what they assume. We then discuss what we observed at NNS and Austal USA.\u00a0 Both shipbuilders, we find, have had favorable economic impacts on their regions; in particular, demonstrably improving the lives of their blue collar workers.\u00a0 However, we concluded that estimated multipliers do not appear to be especially informative as to shipbuilders\u2019 economic impact.<\/p>\n

Defining Economic Multipliers<\/h2>\n

There is an extensive literature[2]<\/a> on economic multipliers, but we were surprised by its lack of consensus or clarity. The economic impact of a given project varies with the economic conditions and specific context in which the shipbuilding spending would occur. An economic multiplier is a foundational concept in macroeconomics. Suppose that the government spends $100 buying a good or service from a company or individual. The party that receives the $100 might then be expected to spend the $100 (or at least a portion of it) on some other goods and services. The original $100 creates a cascade (i.e., multiples) of spending through the economy: it results in additional spending by shipbuilder workers at local restaurants, which then hire additional workers who rent additional housing, and so forth. The cascade of resultant spending is not going to be infinite, because recipients at each stage may save some of the money, will have to pay taxes on it, and may spend some of the money outside the measured region.<\/p>\n

If a leakage (saving, taxation, imports) rate can be estimated, the economic multiplier represents the sum of a convergent geometric series. For example, with a leakage rate of 50 percent, the original $100 in spending would result in a total regionwide spending increase of $200 ($100 plus $50 plus $25 plus $12.50 plus $6.25, and so on), or an economic multiplier of 2.0. Note that multiplier values in this formulation include the original level of spending. Multipliers are estimated using input-output modeling, which includes RIMS and IMPLAN previously mentioned. This approach establishes different categories of expenditures, and it is estimated where a dollar, if spent in a category, is subsequently spent across other categories. The calculations are essentially a bookkeeping-type exercise. A dollar sent to a recipient (e.g., a shipbuilder) is divided between categories of usage (e.g., a share to pay workers, a share to purchase materials from subcontractors, a share paid in dividends to shareholders, a share paid in taxes), and then each category, in turn, has its own allocation of consequent spending.<\/p>\n

Multipliers Do Not Necessarily Estimate Project Impact<\/strong><\/h2>\n

An example from the Second World War illustrates why we cannot simply multiply an estimated multiplier by a new project\u2019s size to estimate the new project\u2019s impact.\u00a0 The vast defense-related buildup that the United States\u2019 manufacturing base undertook came at the expense of the production of consumer products. Herman (2012, p. 153) notes: First came the auto industry, with a drastic cut by more than half. Then in October [1941] nonessential construction was ordered halted, to divert materials to defense plant construction. On October 21 manufacturers had to stop using copper in almost all civilian products, followed by sharp cuts in refrigerators, vacuum cleaners, metal office furniture, and similar durable goods. The total amount of economic activity generated by U.S. manufacturers did not increase by the magnitude of their government defense contracts. Rather, the net change was the value of those contracts less the value of the consumer products that would have been manufactured absent the government contracts.<\/p>\n

When we consider displacement or crowding out effects, we find that economic impacts can be considerably less than those estimated by RIMS or IMPLAN.\u00a0 Economic multiplier estimates implicitly assume that all new positions absorb previously unemployed workers rather than drawing workers away from other employers.\u00a0 If, ultimately, the new project simply displaces comparable projects, the economic impact would be zero. We have to ask, then, what activities are displaced by the government\u2019s shipbuilding contract. The higher the economic value of the displaced activity, the lower the net economic impact of the shipbuilding contract would be. On the other hand, to the extent that shipbuilding reduces unemployment or underemployment, a larger impact is implied. It is also possible that estimated economic multipliers understate long-run economic impact.<\/p>\n

Suppose, for instance, that the shipbuilding contract leads to skill development in a region that consequently attracts other industry. Then the economic implications would be larger than those estimated by an economic multiplier.\u00a0 The \u201cleakage arithmetic\u201d of economic multiplier calculations does not consider the possibility of favorable economic spillovers or synergies. \u00a0Such an effect would argue for use of a larger value as the applicable economic impact. We therefore have to consider some key issues in our case studies, including:<\/p>\n