Author Archives: jdingel

Shift-share designs before Bartik (1991)

The phrase “Bartik (1991)” has become synonymous with the shift-share research designs employed by many economists to investigate a wide range of economic outcomes. As Baum-Snow and Ferreira (2015) describe, “one of the commonest uses of IV estimation in the urban and regional economics literature is to isolate sources of exogenous variation in local labor demand. The commonest instruments for doing so are attributed to Bartik (1991) and Blanchard and Katz (1992).”

The recent literature on the shift-share research design usually starts with Tim Bartik’s 1991 book, Who Benefits from State and Local Economic Development Policies?. Excluding citations of Roy (1951) and Jones (1971), Bartik (1991) is the oldest work cited in Adao, Kolesar, Morales (QJE 2019). The first sentence of Borusyak, Hull, and Jaravel’s abstract says “Many studies use shift-share (or “Bartik”) instruments, which average a set of shocks with exposure share weights.”

But shift-share analysis is much older. A quick search on Google Books turns up a bunch of titles from the 1970s and 1980s like “The Shift-share Technique of Economic Analysis: An Annotated Bibliography” and “Dynamic Shift‐Share Analysis“.

Why the focus on Bartik (1991)? Goldsmith-Pinkham, Sorkin, and Swift, whose paper’s title is “Bartik Instruments: What, When, Why and How”, provide some explanation:

The intellectual history of the Bartik instrument is complicated. The earliest use of a shift-share type decomposition we have found is Perloff (1957, Table 6), which shows that industrial structure predicts the level of income. Freeman (1980) is one of the earliest uses of a shift-share decomposition interpreted as an instrument: it uses the change in industry composition (rather than differential growth rates of industries) as an instrument for labor demand. What is distinctive about Bartik (1991) is that the book not only treats it as an instrument, but also, in the appendix, explicitly discusses the logic in terms of the national component of the growth rates.

I wonder what Tim Bartik would make of that last sentence. His 1991 book is freely available as a PDF from the Upjohn Institute. Here is his description of the instrumental variable in Appendix 4.2:

In this book, only one type of labor demand shifter is used to form instrumental variables2: the share effect from a shift-share analysis of each metropolitan area and year-to-year employment change.3 A shift-share analysis decomposes MSA growth into three components: a national growth component, which calculates what growth would have occurred if all industries in the MSA had grown at the all-industry national average; a share component, which calculates what extra growth would have occurred if each industry in the MSA had grown at that industry’s national average; and a shift component, which calculates the extra growth that occurs because industries grow at different rates locally than they do nationally…

The instrumental variables defined by equations (17) and (18) will differ across MSAs and time due to differences in the national economic performance during the time period of the export industries in which that MSA specializes. The national growth of an industry is a rough proxy for the change in national demand for its products. Thus, these instruments measure changes in national demand for the MSA’s export industries…

Back in Chapter 7, Bartik writes:

The Bradbury, Downs, and Small approach to measuring demand-induced growth is similar to the approach used in this book. Specifically, they used the growth in demand for each metropolitan area’s export industries to predict overall growth for the metropolitan area. That is, they used the share component of a shift-share analysis to predict overall growth.

Hence, endnote 3 of Appendix 4.2 on page 282:

This type of demand shock instrument was previously used in the Bradbury, Downs and Small (1982) book; I discovered their use of this instrument after I had already come up with my approach. Thus, I can only claim the originality of ignorance for my use of this type of instrument.

Tim once tweeted:

Researchers interested in “Bartik instrument” (which is not a name I coined!) might want to look at appendix 4.2, which explains WHY this is a good instrument for local labor demand. I sometimes sense that people cite my book’s instrument without having read this appendix.

Update (10am CT): In response to my query, Tim has posted a tweetstorm describing Bradbury, Downs, and Small (1982).

The rapid rise of spatial economics among JMCs

Two years ago, my list of trade candidates also included a dozen candidates in spatial economics. Last year I listed 20 candidates. There are 45 spatial-economics JMCs in this year’s list. That looks like a rapid rise.

Of course, measurement problems abound. My view of “spatial economics” may have broadened during the last two years, in which case the listings would tell you more about me than about the candidates. That would be hard to quantify. But, to focus on one label within the broader spatial economics nexus, I’m pretty sure that I’m seeing more candidates explicitly list “urban economics” as one of their fields than in years prior.

If I’m right that the supply of spatial economists is rising, then one immediately wonders if the demand side will keep pace. I haven’t looked at JOE postings, but I doubt that ads mentioning “urban economics” are growing at the same rate as candidates listing it as a field.

Last month, in response to a Beatrice Cherrier query about why urban economics’ “boundaries & identity are so difficult to pin down,” Jed Kolko noted that “urban economists typically align strongly to another field — trade, labor, PF, finance (esp the real estate types), macro.” That fluidity has advantages and disadvantages. It certainly makes it challenging to compile a list of relevant job-market candidates. But my very short time series of arbitrarily collated candidates suggests growth in the supply of young spatial economists.

Spatial economics JMPs (2019-2020)

Here’s a list of job-market candidates whose job-market papers fall within spatial economics, as defined by me when glancing at a webpage for a few seconds. Illinois has six candidates! I’m sure I missed folks, so please add them in the comments.

The annual list of trade candidates is a distinct post.

Of the 45 candidates I’ve initially listed, 18 used Google Sites, 12 registered a custom domain, 3 used GitHub, and 12 used school-provided webspace.

Here’s a cloud of the words that appear in these papers’ titles:

Trade JMPs (2019-2020)

It’s November again. Time flies, and there’s a new cohort of job-market candidates. Time really flies: I started this series a decade ago! Many members of that November 2010 cohort now have tenure or will soon.

As usual, I’ve gathered a list of trade-related job-market papers. There is no clear market leader: the most candidates from one school by my count is three (Berkeley, Maryland, UCLA). If I’ve missed someone, please contribute to the list in the comments.

A separate post lists candidates in spatial economics, broadly defined.

Of the 31 candidates I’ve initially listed, 14 registered a custom domain, 9 used Google Sites, 2 used GitHub, and only 6 use school-provided webspace.

Here’s a cloud of the words that appear in these papers’ titles:

Why your research project needs build automation

Software build tools automate compiling source code into executable binaries. (For example, if you’ve installed Linux packages, you’ve likely used Make.)

Like software packages, research projects are large collections of code that are executed in sequence to produce output. Your research code has a first step (download raw data) and a last step (generate paper PDF). Its input-output structure is a directed graph (dependency graph).

The simplest build approach for a Stata user is a “master” do file. If a project involves A through Z, this master file executes A, B, …, Y, and Z in order. But the “run everything” approach is inefficient: if you edit Y, you only need to run Y and Z; you don’t need to run A through X again. Software build tools automate these processes for you. They can be applied to all of your research code.

Build tools use a dependency graph and information about file changes (e.g., timestamps) to produce output using (all and only) necessary steps. Build automation is valuable for any non-trivial research project. Build automation can be particularly valuable for big data. If you need to process data for 100 cities, you shouldn’t manually track which cities are up-to-date and which need to run the latest code. Define the dependencies and let the build tool track everything.

Make is an old, widely used build tool. It should be available on every Linux box by default (e.g., it’s available inside the Census RDCs). For Mac users, Make is included in OS X’s developer tools. I use Make. There are other build tools. Gentzkow and Shapiro use SCons (a Python-based tool). If all of your code is Stata, you could try the project package written by Robert Picard, though I haven’t tried it myself.

A Makefile consists of a dependency graph and a recipe for each graph node. Define dependencies by writing a target before the colon and that target’s prerequisites after the colon. The next line gives the recipe that translates those inputs into output. Make can execute any recipe you can write on the command line.

I have written much more about Make and Makefiles in Section A.3 of my project template. Here are four introductions to Make, listed in the order that I suggest reading them:

What’s an “iceberg commuting cost”?

In the recent quantitative spatial economics literature, the phrase “iceberg commuting cost” appears somewhat often. The phrase primarily appears in papers coauthored by Stephen Redding (ARSW 2015, RR 2017, MRR 2018, HRS 2018), but it’s also been adopted by other authors (Fratto 2018, Gaigne et al 2018, Matt Turner’s lecture notes). However, none of these papers explicitly explains the meaning of the phrase. Why are we calling these commuting costs “iceberg”?

The phrase was imported from international economics, where the concept of “iceberg transport costs” is widely used. That idea is also explicitly defined. Alan Deardorff’s glossary says:

A cost of transporting a good that uses up some fraction of the good itself, rather than other resources. By analogy with floating an iceberg, costless except for the part of the iceberg that melts. Far from realistic, but a tractable way of modeling transport costs since it impacts no other market. Due to Samuelson (1954).

Two bits of trivia that aren’t very relevant to the rest of the post: these should be called “grain transport costs” because von Thunen introduced the idea with oxen-pulled grain carts more than a century before Samuelson (1954) and basic physics means there are actually economies of scale in shipping ice.

Why do we use the iceberg assumption? As Deardorff highlights, it lets us skip modeling the transportation sector. By assumption, the same production function that produces the good also produces its delivery to customers. For better or worse, that means that international or long-distance transactions don’t affect factor demands or transport prices by being international or long-distance per se (Matsuyama 2007). This is one way of keeping trade models simple. Per Gene Grossman: “few would consider the ‘iceberg’ formulation of shipping costs as anything more than a useful trick for models with constant demand elasticities.”

In urban economics, saying that commuting costs take the “iceberg” form means that the model abstracts from transportation infrastructure and the transport sector. Commuters “pay” commuting costs by suffering lower utility. There is no supplier of transportation services that earns any revenues. (Given that most US roads are unpriced, this isn’t much of an abstraction.) But, just as folding transportation services into the goods-producing firm’s production function has consequences for trade models, saying that commuting enters the utility function directly has consequences for the economic content of urban models.

Given that these models do not feature a labor-leisure tradeoff, there is an equivalence between utility costs and time costs. As described by Ahfeldt, Redding, Sturm, and Wolf (2015): “Although we model commuting costs in terms of utility, there is an isomorphic formulation in terms of a reduction in effective units of labor, because the iceberg commuting cost enters the indirect utility function (5) below multiplicatively.” If the cost of commuting is mostly about the opportunity cost of time, then this modeling device captures it reasonably well in a model with homogeneous workers.

If workers are heterogeneous in their hourly wages, then their opportunity costs of time differ. Higher-wage workers have higher opportunity costs of time. In the classic model of locational choice (see Kevin Murphy’s lecture), this causes higher-wage workers to be more willing to pay for residential locations that give them shorter commutes. In the typical quantitative spatial model, however, preferences are Cobb-Douglas over housing and a tradable good. As a result, even with heterogeneous agents, the utility-cost and time-cost formulations of commuting costs are equivalent.

But what if commuting costs are paid with money? In addition to more time on the road, driving a greater distance involves burning more fuel. (Actually, in these models, it typically involves burning more of the numeraire good.) This is not equivalent to the utility formulation, because the cost of a tank of gas is not a constant proportion of one’s income. Moreover, if the car itself costs money, then lower-wage workers might take the bus. The monetary costs of accessing different commuting technologies can have big consequences for urban form, as suggested by LeRoy and Sonstelie (1983), Glaeser, Kahn, and Rappaport’s “Why do the poor live in cities?” and Nick Tsivanidis’s paper on bus-rapid transit in Bogota. The iceberg formulation of commuting costs cannot tackle these issues.

Similarly, even though transportation infrastructure is surely more capital-intensive than much of the economy, we cannot speak to that issue when we parsimoniously model transport as simply coming out of people’s utility.

“Iceberg commuting cost” is a short, three-word phrase. I hope the 600+ words above suggest what it might mean.

Market-size effects, across places and over time

The dividing line between neoclassical trade models and the now-quite-dated “new trade theory” is economies of scale. Neoclassical models feature constant (or decreasing) returns. Free trade is efficient in such settings. With the introduction of increasing returns, Brander, Spencer, Krugman, Helpman, and others “open[ed] the possibility that government intervention in trade via import restrictions, export subsidies, and so on may under some circumstances be in the national interest after all” (Krugman 1987).

The fact that “size matters” in new trade theory (size can influence the pattern of specialization because there are economies of scale) while it does not in neoclassical models became the basis for empirical investigations trying to distinguish these theories. Davis and Weinstein (2003) describe the idea behind this research strategy:

A fundamental divide may be identified between two classes of models. In the first class, unusually strong demand for a good, ceteris paribus, makes a country an importer of a good. An example would be a conventional two-sector neoclassical model with strictly downward sloping import demands. However, there is an alternative tradition within the trade literature which emphasizes an important interaction between demand conditions and production opportunities in which the production response to local demand conditions is so powerful that strong local demand for a product leads a country to export that product. When such conditions exist, the literature terms it a home market effect.

Stepping away from trade, there’s a very different economic context in which the role of market size is also crucial: the literature on innovation. The idea dates at least to Schmookler (1966) who memorably titled two of his chapters “The amount of invention is governed by the extent of the market.” It’s also key to endogenous growth theory. Acemoglu and Linn (2004) provided empirical evidence that market size influenced innovation in a particular sector:

This paper investigates the effect of (potential) market size on entry of new drugs and pharmaceutical innovation. Focusing on exogenous changes driven by US demographic trends, we find a large effect of potential market size on the entry of nongeneric drugs and new molecular entities… Our results show that there is an economically and statistically significant response of the entry of new drugs to market size. As the baby boom generation aged over the past 30 years… the data show a corresponding decrease in the rate of entry of new drugs in categories mostly demanded by the young and an increase for drugs mostly consumed by the middle-aged.

In “The Determinants of Quality Specialization“, I showed that high-income cities manufacture higher-priced, higher-quality goods in part because they are home to more high-income households who demand such products. Quantitatively, I found that the home-market effect plays at least as large a role as the factor-abundance mechanism in quality specialization across cities of different income levels.

What does this have to do with Acemoglu and Linn (2004)? I didn’t see much of a connection when I was writing my paper. Pharmaceuticals were just one of many industries in my data on US manufacturing plants, and pharmaceutical pills are probably less sensitive to trade costs than most goods. But I now see a closer relationship between looking for home-market effects in the cross section and looking for market-size effects in the time series.

The primary bridge is a recent QJE article by Costinot, Donaldson, Kyle and Williams. They used variation in disease burdens across countries as a source of variation in demand for drugs to look for home-market effects in international pharmaceuticals production. I’ve blogged about that paper before.

The latest connection is a paper by Xavier Jaravel called “The Unequal Gains from Product Innovations: Evidence from the U.S. Retail Sector”. His article investigates the time-series analog of my cross-sectional results on quality specialization. In recent decades, income growth has been concentrated at the top of the income distribution. Did the increase in the relative size of the affluent market benefit the affluent beyond the straightforward income gains? With economies of scale, increases in demand could induce supply-side responses that favor affluent-demanded goods. That’s the home-market-effect story for why high-income cities are net exporters of high-quality products: due to increasing returns, greater demand elicits a more-than-proportionate production response. Jaravel documents the time-series equivalent for national outcomes: “(1) the relative demand for products consumed by high-income households increased because of growth and rising inequality; (2) in response, firms introduced more new products catering to such households; (3) as a result, the prices of continuing products in these market segments fell due to increased competitive pressure.”

As a result, a two-by-two matrix neatly summarizes these contributions to the empirical literature on market-size effects:

Pharmaceuticals Vertically differentiated consumer goods
Time series Acemoglu & Linn (2004) Jaravel (forthcoming)
Cross section Costinot, Donaldson, Kyle, Williams (2019) Dingel (2017)

There’s an obvious relationship between the AL and CDKW papers, as explained by CDKW:

In their original article, Acemoglu and Linn (2004) exploit such demographic variation over time within the United States to estimate the impact of market size on innovation. Here, we employ the spatial analog of this strategy, drawing on cross-sectional variation in the demographic composition of different countries in a given year, to explore how exogenous variation in demand may shape the pattern of trade.

With the benefit of hindsight, some more subtle connections between the four cells of this two-by-two matrix seem pretty clear. For example, Jaravel’s adoption of the Acemoglu (2007) terminology for “weak bias” and “strong bias” in his footnote 3 mirrors the distinction between the weak and strong versions of the home-market effect introduced by Costinot, Donaldson, Kyle, and Williams (2019).

In summary, market-size effects seem to be important for understanding both innovation outcomes and the geographic pattern of specialization. We’ve found market-size effects in the time series and in the cross section, for both the pharmaceutical sector and vertically differentiated manufactured goods.