In response to my emailed question, the Ministry for Social Development told me that all records prior to 1996 are held in hard copy rather than electronically. In other words, nobody knows for how many years current beneficiaries have received benefits. Figuring it out would require manually going through each beneficiary's paper records, putting them into electronic form, and then running the averages. As it stands, the data does not exist in any usable form.
I wonder how the government can make any kind of informed policy decisions when it has no clue whether current beneficiaries are experiencing transitional problems or whether their current spell on benefit is just the latest in a series of spells. Surely optimal policy varies depending on this. The best I can find is a study by Moira Wilson a few years back that took a cohort of 250,000 granted a benefit in 1993 and followed up five years later. But again, that's just a two-period snapshot.
I've asked MSD if there's any data available on the cohort that became eligible for the DPB since 1996: their lifetime records will be available electronically. We'll see what comes back.
It would be awfully nice if Wellington could devote a few resources to getting the data in order. A few summer interns doing data entry could make a decent start. Some of the kinds of questions to which we cannot know the answer for lack of data:
- How many total years have current beneficiaries received benefits, in how many spells, with breaks of how long between spells and with what reasons for transitioning between types of benefit or on and off of benefit?
- Is there a blip upwards in the birth rate among women on the DPB as work entry requirements come close to binding? In other words, is the timing and number of births elastic to the duration of benefits? It would be a perverse result for the child-poverty campaigners if the DPB were encouraging more within-poverty births; we haven't sufficient data to tell.
- If we were to move to US-style lifetime caps, how many people would be subject to them? The optimal lifetime cap surely depends on the distribution of lifetime receipt. Length of current spell gives a lower bound estimate; what's a more reasonable estimate?
- What are the longer term trends in movement between the unemployment benefit and other benefits? If the recession drags on and we see folks flip from unemployment benefit to the main benefit, what's their expected duration on main benefit before rejoining the workforce?