My 2017 Journal of Urban Economics paper (see 2015 NBER early version) presents new evidence on these claims. While the American Economic Review rejected this paper, I think we wrote a really good paper and I thank the JUE for recognizing this. MIT News just published a write up of this paper.
Markets allocate scarce resources. In the absence of political favoritism, capital will be allocated to the entity who values it the most because this entity will be willing to pay the highest interest rate to borrow the lender's capital. This logic predicts that in equilibrium that each potential borrower of capital will have equalized the "bang per buck borrowed". The anonymous markets allocate capital to equate the marginal valued added from having access to capital across firms.
Now suppose that the Powerful State plays favorites and allows certain firms to borrow at special rates such as an interest rate of 0% and to receive free land to build factories on. These entities will use more resources and this could lead to a serious misallocation if these favored entities are less productive or simply diminishing returns to capital exist. Read this great 2017 misallocation paper and read my USC colleague Joel David's work.
But, there can be growth benefits when government plays favorites. While I am a pure UChicago economist, consider the following economy. At the fringe of major cities in China, there is farmland growing low value added crops. Imagine if a powerful leader could cheaply payoff the farmers to vacate and that the cleared land is zoned for employment. Roads are built, infrastructure (sewers and electricity) are installed and then the following recruitment process occurs; anchor companies are recruited at cheap rents and other incentives are lured to open up production in the park. As in the suburban mall setting, these anchor tenants attract other firms who seek to coagglomerate with them and a productive new edge city forms. In our China paper (see above), we document that this is exactly what happens when parks feature few state owned enterprises and attract high human capital firms.
But, there can be growth benefits when government plays favorites. While I am a pure UChicago economist, consider the following economy. At the fringe of major cities in China, there is farmland growing low value added crops. Imagine if a powerful leader could cheaply payoff the farmers to vacate and that the cleared land is zoned for employment. Roads are built, infrastructure (sewers and electricity) are installed and then the following recruitment process occurs; anchor companies are recruited at cheap rents and other incentives are lured to open up production in the park. As in the suburban mall setting, these anchor tenants attract other firms who seek to coagglomerate with them and a productive new edge city forms. In our China paper (see above), we document that this is exactly what happens when parks feature few state owned enterprises and attract high human capital firms.
So, our paper documents both misallocation and the fact that the state can solve co-location problems. Why the second? Firms that seek to colocate can have trouble facing the land assembly problem. In big cities, there are few adjacent parcels where large firms can locate next to each other. This is possible at the fringe of a city. The urban economics literature (see Ellison and Glaeser and Rosenthal and Strange) has documented the benefits of firm agglomeration but there hasn't been enough research studying why agglomeration is costly. The answer is that in Major Cities in the U.S that the real estate capital stock is durable and tenants sign long term leases; these frictions limit the ability of firms to "co-agglomerate" and thus enjoy the benefits in terms of learning and lower transportation costs from being next to each other.
Since we study 110 expensive parks in China, we document heterogeneous treatment effects. Our paper is cool because we document heterogeneous costs of such parks (the opportunity cost of not using the land at the fringe for purely suburban housing) and heterogeneous benefits (that we parameterize as a function of standard urban agglomeration forces). Similar to Hsieh and Klenow , we find that the presence of State Owned Enterprises lowers the productivity of these parks. So, public capital is best used when it lures productive private firms who gain from co-agglomerating next to each other. We argue that due to the land assembly problem that these firms would not have been able to achieve this in the absence of the creation of the new park.