Proponents of this story like to cite academic research — some of it from Democratic-leaning economists — that seemingly confirms the idea that unemployment insurance causes unemployment. They’re not equally fond of pointing out that this research is two or more decades old, has not stood the test of time, and is irrelevant in any case given our current economic situation.I'm not sure what research the "proponents of this story" or Krugman are paying attention to, but the mainstream theory of search and unemployment - our standard tool for understanding how the labor market works - is both old and new, has certainly stood the test of time, and it tells us that more generous unemployment insurance (UI) increases the unemployment rate.
In standard search models, there are several channels through which UI affects the unemployment rate. In early one-sided search models, e.g. McCall (1970) an unemployed worker searches until he or she finds a job that is sufficiently attractive. The unemployed worker has a reservation wage strategy - offers at or above the reservation wage are accepted and those falling below it are rejected. More generous UI makes the opportunity cost of searching lower, and the unemployed worker increases his or her reservation wage. Why not be more picky if searching is less painful? This tends to increase the average duration of unemployment for any unemployed worker, and therefore to increase the unemployment rate.
Economists later developed two-sided search models. The big innovators in this research area were Dale Mortensen and Chris Pissarides. In a two-sided search model there are firms looking for workers, and would-be workers looking for firms, and it takes time for them to match. In this model, the effect of UI works somewhat differently from what happens in the one-sided search model. In a Mortensen-Pissarides (MP) model, most or all of the action is isolated in the behavior of firms. Unemployed workers simply search no matter what, but firms face a choice as to whether or not to bear a cost of searching for workers (posting a vacancy). If UI is more generous, this makes the bargaining position of workers better when negotiating with firms, because the opportunity cost of rejecting a job offer is lower (in this sense the effect has some similarity to what happens with one-sided search). This has the effect of increasing the wage the worker receives, which is bad for firms. As a result, fewer firms post vacancies in the labor market, the job finding rate of unemployed workers drops, and the unemployment rate rises.
Another strand of economic research studies the optimal design of UI systems. Like any form of insurance, UI is doing something useful - it provides insurance - but there are incentive effects to worry about. Insurance companies that provide auto insurance and fire insurance understand that there is a moral hazard problem - the better the insurance the less care the insured takes to prevent a loss. And it is difficult to observe the insured's effort in preventing losses. Similarly, in the case of UI, the search effort of an unemployed worker will affect the would-be worker's chances of finding a job, and that search effort is difficult to observe. More generous UI, as above, reduces the opportunity cost of searching for work, and one of the adjustments the would-be worker makes is to put less effort into search. This increases the average duration of unemployment and the unemployment rate. Work in this area (e.g. Hopenhayn and Nicolini's work) shows that, if the UI system is designed to account for moral hazard, that UI benefits should last for a long time after a worker becomes unemployed, and the benefit level should decline during a worker's spell of unemployment.
These predictions - which come from theory - are broadly consistent with empirical evidence. For example, if we look at a cross section of countries, what we observe is that countries with more generous UI systems than in the US (Western Europe, UK, Canada) tend to have higher unemployment rates.
Search models have been very widely used, and they are currently ubiquitous. They're used by labor economists and macroeconomists (New Keynesians too). Further, Diamond, Mortensen, and Pissarides received the Nobel prize in economics for their work on search theory. Quoting from the Nobel committee's announcment:
This year's three Laureates have formulated a theoretical framework for search markets. Peter Diamond has analyzed the foundations of search markets. Dale Mortensen and Christopher Pissarides have expanded the theory and have applied it to the labor market. The Laureates' models help us understand the ways in which unemployment, job vacancies, and wages are affected by regulation and economic policy. This may refer to benefit levels in unemployment insurance or rules in regard to hiring and firing. One conclusion is that more generous unemployment benefits give rise to higher unemployment and longer search times.Note the last sentence.
UI benefits were extended for the long-term unemployed in the US during the last recession, and those extended benefits are still in place. Krugman wants to argue that cutting off those benefits would be bad policy. There may be good reasons why Krugman's conclusion is correct. For example, the standard UI system in the US whereby benefits are cut off after 6 months may be suboptimal. As well, it may be the case that the effect of extended benefits on the unemployment rate is not quantitatively important. But it only weakens Krugman's case to make statements that are false.
Further, in the quote I started with, Krugman seems to want to argue that there is something about our current situation that is so different that everything we know gets turned on its head. That's not so obviously false, but I think it's incorrect.