Some impressions from the CAED (Comparative Analysis of Enterprise Data) conference in Budapest. One of the best papers I've seen was The effect of industrial policy on corporate performance… by Chiara Criscuolo et al from LSE's Centre for Economic Performance.
ink to the paper here.
Here is a brief synopsis. Many countries (Latvia is no exception) spend lots of taxpayers’ money on industrial policies, which are supposed to raise productivity and employment. The basic concern however, is that the recipients use the money to finance activities they would have undertaken anyway. This would imply that taxpayers’ money is wasted. In spite of this concern (and ubiquity of industrial policies), there are surprisingly few proper econometric evaluations of such policies. Of course, I am not talking about naïve surveys where recipients are asked whether the program was useful. The problem is in constructing what economists call counterfactuals for firms that receive assistance. In other words, we don’t know what they [firms] would have done had they not received assistants. For example, suppose the government has a program that provides subsidies to firms that increase employment. Then a firm that plans to hire more workers (regardless of the subsidy) has an incentive to apply for the subsidy. A naïve study would find that subsidies are correlated with increases in employment. Yet in this particular case, clearly, a subsidy has no ‘effect’ on employment.
How to ‘identify’ the true effect (if there is any)? Economists use subtle identification strategies called instrumental variables. Intuitively, the idea is to find a factor (instrument) that influenced firm’s decision to participate in a program, but not its decision to change its employment. The ‘instrument’ could then be used to ‘isolate’ the true causal effect. Unfortunately, such ‘instruments’ are hard to find. Chiara Criscuolo and her co-authors look at the UK government’s program that gives grants to firms for investment in economically depressed areas. Being in the EU, however, the UK government cannot designate any region as a depressed one, it needs permission from the European Commission. The commission, in turn, designates depressed regions for Europe at large, so that the UK government has little to say about this. This provides an ‘instrument’, because, for some regions, whether a firm’s receives a subsidy or not is decided by the EU, and not by the firm.
What are the findings? Does industrial policy matter? The study finds large positive effects on employment and investment for subsidized firms. Another score for industrial policy, as opposed to free markets? Not so fast. The study does not find any effect on aggregate productivity, that is, on the efficiency of firms. Moreover, Criscuolo and her co-authors find that most of the employment increase is for large incumbent firms, the proportion of employment in new entrants actually falls. It therefore appears that industrial policy in the UK reduced aggregate productivity by hampering reallocation from more productive to less productive firms. How does this translate from ‘econ-speak’? Well, in a market economy more efficient firms can produce with lower costs and, therefore, can provide goods and services at lower prices. If the market is competitive, more efficient firms survive and less efficient firms perish. The resources released by less efficient firms (eg labor) are then absorbed by more efficient firms. Thus, firms constantly search for new ways to be more efficient and, in the process, provide better products at lower prices. When the government pumps money into less efficient firms, it delays the inevitable at the expense of more efficient firms and consumers.
Incidentally, in 2007-2013, Latvian government will spend 25 million lats to support businesses in depressed regions.
In Latvian here