Shopping Externalities and Self Fulfilling Unemployment Fluctuations*

More documents

Recommendations

Info

Table 1: Calibration targetsLabor Market TargetsMonthly transition rate, UE 0.433Monthly transition rate, EU 0.024Elasticity UE wrt tightness 0.650Product Market TargetsExpenditures of U relative to E 0.85Shopping time of U relative to E 1.25St dev log prices 0.15Max-min ratio 1.80Price paid by U relative to E 0.98Other TargetsPro…t margin 0.05Real annual interest rate 0.0355.2 Data Sources and Target ValuesTable 1 summarizes the value of the empirical targets used to calibrate the parameters of themodel. We construct empirical measures of the workers’transition rates between employmentand unemployment following the methodology developed by Shimer (2005). We …nd that,over the period 1987-2007, the average UE rate was 43 percent per month and the averageEU rate was 2.4 percent per month. We also …nd that the elasticity of the UE rate withrespect to the vacancy-to-unemployment rate was approximately 25 percent. However, asdiscussed in Petrongolo and Pissarides (2000) and Menzio and Shi (2011), this is a biasedestimate of the elasticity of the UE rate with respect to labor market tightness because bothemployed and unemployed workers search for jobs. Hence, we choose to target an elasticityof the UE rate with respect to theta of 65 percent, which is the value estimated by Menzioand Shi (2011) after accounting for search on the job. In Section 5.5, we consider alternativecalibrations in which the targeted elasticity varies between 45 and 70 percent.We choose the target for the di¤erence in expenditures between employed and unemployedworkers by looking at available estimates of the decline in expenditures experiencedby households transiting from employment to unemployment. Bentolila and Ichino (2008)use the PSID to estimate the e¤ect of moving from employment into unemployment on ahousehold’s food expenditures. They …nd that a year of unemployment leads a 19 percentdecline in food expenditures. Restricting attention to households who move into unemploymentbecause of either business closures or mass layo¤s, Stephens (2001) …nds that a year ofunemployment leads to a 14 percent decline in food expenditures. Stephens (2004) obtainssimilar …ndings using data from the Health and Retirement Survey (HRS) in addition to34
the PSID. 12 Based on these estimates, we choose a baseline target of 15 percent for thedi¤erence in expenditures between employed and unemployed workers. However, in Section5.5, we allow this target to vary between 10 and 40 percent.We use cross-sectional data from the American Time Use Survey (ATUS) to choose thetarget for the di¤erence in the amount of time devoted to shopping by employed and unemployedworkers. We restrict attention to individuals aged 22-55. We …nd that employedindividuals spend between 24 and 33 percent less time shopping than non-employed individuals,and between 13 and 20 percent less than unemployed individuals. 13 Krueger and Mueller(2010) also measure di¤erences in shopping time between workers in di¤erent employmentstates. They …nd a di¤erence in shopping time between employed and unemployed individualsof 29 percent in the US, 67 percent in Canada and Western Europe and 56 percent inEastern Europe. On the basis of these …ndings, we choose a baseline target of 25 percent forthe di¤erence in shopping time between employed and unemployed workers. In Section 5.5,we allow this target to vary between 10 and 50 percent.We use the Kielts-Nielsen Consumer Panel Data (KNCDP) to measure price dispersion,as well as the di¤erence in price paid by employed and non-employed households. 14restrict attention to individuals aged 22-55. We consider di¤erent de…nitions of a product.In our narrowest de…nition, we group together all goods that have di¤erence barcodes, but areidentical along all other dimensions (i.e. brand, size, color, shape, packaging, etc. . . ). In ourbaseline de…nition, we group together all goods that are identical, except for their barcodeand brand. In our broadest de…nition, we group together all goods that are identical, exceptfor their barcode, brand and size. We de…ne a market as Scantrack Market Area, which isthe notion of market used by Nielsen. We de…ne a time period as a quarter. Then, for eachtriple of product, market and time period, we measure the distribution of transaction pricesand we compute the standard deviation of log prices and various percentile ranges. For ourbaseline de…nition of a product, we …nd that the median standard deviation of log prices isapproximately 25 percent (see Table 2 in Kaplan and Menzio 2013), which is similar to the12 The elasticity of food expenditures with respect to income is likely to be low compared to other expenditurescategories, such as luxury goods or semi-durable goods. Therefore, the estimated e¤ect of moving intounemployment on food expenditures is likely to be low compared with the e¤ect on overall expenditures.13 The estimation results vary depending on the de…nition of shopping time. We consider a broad de…nitionof shopping time which includes time spent purchasing all goods and services plus related travel time, and anarrow de…nition of shopping time which includes time spent purchasing consumer goods and groceries plusrelated travel time. All details are available upon request.14 The KNCPD is a panel dataset covering approximately 50,000 households over the period 2004-2009.Respondents use in-home UPC scanning devices to record information (price, quantity, outlet, etc. . . ) abouttheir purchases of grocery and non-grocery household items, which account for roughly 30 percent of totalexpenditures. This data is similar, although much broader in scope to that used by Aguiar and Hurst (2007)in their analysis of the shopping behavior of retired households. The reader can …nd more details in Kaplanand Menzio (2013).35We
Page 2: 1 IntroductionWe propose a novel th
Page 5 and 6: claims). However, they certainly in
Page 7 and 8: the evidence brought forward that
Page 9 and 10: function with boundary conditions (
Page 11 and 12: are given byS t (p) = ((u t )) u t(
Page 13 and 14: The …rm and the worker bargain ov
Page 15 and 16: the paper, which is mainly devoted
Page 17 and 18: espect to u as (1 +S 0 (u) =u )(1
Page 19 and 20: a seller meets a buyer is constant.
Page 21 and 22: The eigenvalues, the determinant an
Page 23 and 24: to denote the set of J’s such tha
Page 25: JJů = 0ů = 0E1E1J1sJ1sE2J3sE2J3sE
Page 28 and 29: understand that regime switches can
Page 30 and 31: De…nition 3. An i-j Regime Switch
Page 32 and 33: 5 Sentiments and the Great Recessio
Page 36 and 37: …ndings in Sorensen (2000) and Mo
Page 38 and 39: An increase in unemployment of 1 pe
Page 40 and 41: 11.51110.5109.5Mean posted priceMea
Page 42 and 43: with the pessimistic regime. The de
Page 44 and 45: etween employed and unemployed work
Page 46 and 47: Table 4: Robustness: steady-state u
Page 48 and 49: vacancy rate because they are assoc
Page 50 and 51: [22] Heller, W. 1986. “Coordinati
Page 52 and 53: AppendixA Proof of Lemma 1Consider
Page 54 and 55: Claim 5. The only F (p) that satis
Page 56 and 57: where the parameters 0 and 1 are
Page 58: After substituting (C5) into (C6),

Shopping Externalities and Self Fulfilling Unemployment Fluctuations*

Create successful ePaper yourself

Delete template?

Save as template?