The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
confidential How Do Retail Prices React to Minimum Wage Increases? James M. MacDonald and Daniel Aaronson Working Papers Series Research Department Federal Reserve Bank of Chicago December 2000(WP-00-20} FEDERAL RESERVE BANK OF CHICAGO confidential confidential How Do Retail Prices React to Minimum Wage Increases? James M.MacDonald Daniel Aaronson* December 2000 *Economic Research Service, U.S. Department of Agriculture, and Federal Reserve Bank of Chicago, respectively. Work was performed under a memorandum of understanding between the Economic Research Service and the Bureau of Labor Statistics(BLS), which permitted access to the confidential BLS data used in this paper. We thank Bill Cook and Scott Pinkerton of BLS for their advice and help, and thank Chin Lee and Gerald Schluter for comments on earlier drafts. The views expressed herein are not necessarily those of the U.S. Department of Agriculture, the Bureau of Labor Statistics, the Federal Reserve Bank of Chicago or the Federal Reserve System. confidential confidential How Do Retail Prices React to Minimum Wage Increases? A textbook consequence of an industry-wide cost shock is that it will be passed on to consumers through an increase in prices. The minimum wage offers a compelling natural experiment of such a cost shock, particularly among industries that employ lowwage labor. We assess the effect of recent minimum wage increases on restaurant prices, using specific item prices collected by the Bureau of Labor Statistics(BLS). We find that price responses follow textbook expectations in several dimensions. First, restaurant prices rise, by amounts that are broadly consistent with the modest costs imposed by minimum wage increases. Second, prices respond rathez quickly, within asix-month window around the wage incxease. Third, price increases are greater among fastfood outlets and in low-wage locations, where minimum wage increases would be expected to have greater effects on costs. But other elements of the price response are more complicated. A restaurant does not raise all of its prices by announts reflecting the costs of minimum wage increases {from 0.3 to 1.8 percent, depending on outlet type and location). Rather, it raises fewer prices(up to 25 percent of its items), but by 3 to 6 percent,on average. That response suggests that there maybe some item-specific costs to changing price, or that demand elasticities vary across items. Furthermore, we end that items at certain prices (such as fastfood items with prices ending in 99 cents) are less likely to be raised in the face of a minimum wage increase, and that outlets with recent price reviews are less likely to respond to minimum wage changes. confidential confidential Ttus pattern of restaurant price responses relates closely to notions of price stickiness. As has been known since Keynes,the dynamics of a price change can be complicated; in particular, prices z~nay not react instantaneously to cost changes. Figure 1 exemplifies the issue; it plots monthly changes in the Food Away from Horne component of the Consumer Price Index(CPn from 1995 to 1997, along with Producer Price Indexes for two key restaurant inputs: pork and ground beef. While input prices are quite volatile, the Away from Home CPI moves slowly and ~x~ethodiECally. Figure 1 suggests that prices may respond very slowly to cost changes, and the literature suggests conditions under which prices may not react at all. We present further evidence of sluggish price changes in our analysis below and relate those findings to other firm level results on nominal price rigidity. We then contrast the usual co-movements of prices and costs with the rapid and full response to a laxge, national, well-identified cost shock like the minimum wage. Recent Analyses of Minimum Wage Effects on Prices Recently, Card and Krueger(1995), and Aaronson (2001)analyzed the impact of the 1980s and i990s minimum wage legislation on restaurant prices.l Cazd and Krueger(1995) used their own surveys of fast food outlets in New Jersey and Pennsylvania, and reported that prices increased in New Jersey outlets, but not in Pennsylvania's, after an increase in New Jersey's minimum wage. But within New Jersey,they did not find larger increases in areas that were more likely to be affected by the changes. They did, however,find the expected incidence at the national level when they looked at 1990 and 1991 Federal minimum wage changes. Using the Food Away from Home CPI indexes for 271arge metropolitan areas, Card and Krueger found that prices rose more, between 1989 and 1992, confidential confidential in those cities with higher proportions oflow-wage workers in 1989. But the before-andafter nature of their comparisons meant that they could not evaluate issues oftiming, and their limited data generated improbable estimates of magnitude.2 Figure 2 updates Aaronson's(2001)findings, based on the CPI Food Away from Home index for 1978-973 The graph reports regression-adjusted price changes in the four months surrounding a minimum wage increase. The solid line captures the mean predicted change, while the dashed lines include the 95%confidence interval. Prices rise by statistically significant amounts in periods around minimum wage increases, and the increases approximate the expected cost effects o£ the minimum wage changes. The striking feature of the graph concerns timing: most of the pz~ce response occurs within a month or two of the minimum wage change. Despite the fact that minimum wage laws are often passed well before the enactment date, there is only a sma11 price increase in arxticipation of the "event," and no additional price reaction at long lags. Price Data Our dataset is the distinctive feature of our study. We use prices sampled over a three year period (January, 1995 through December, 1997) to construct the Food Away from Home component of the CPI. BLS Feld personnel collected prices for nearly 7,500 food items at over 1,000 different outlets. Outlets were drawn from 88 Primary Sampling Units(PSUs), which in turn included 76 Metropolitan Statistical Areas and 12 other areas representing the urban non-metro United States. PSUs were assigned to one of three reporting cycles; outlets in the eve largest were surveyed each month, while others were surveyed in two bimonthly cycles of odd and even numbered months. Because most prices were collected bimonthly, we compaze price changes over two month periods.a 4 confidential confidential We make use of a BLS "type of business" code assigned to each outlet. In "limited service"(LS)outlets meals are served fox on or off prer~ses consum~t~on, patrons typically pay at the register before they eat, and patrons do not typically place orc}ers while seated at a table, booth, or counter. In "full service"(FS} outlets, food is sold primarily for on premises consumption, waiter/waitress service is provided, orders are taken while patrons are seated at a table, booth or counter, and patrons typically pay after they eat.'~Ve group all other outlet types, such as schools, department stores, convenience stores, gas stations, ox vending machines,into an "other" category. Full service outlets account fox about half of atl price quotes, while LS outlets account for about 29%.S IS outlets employ higher proportions of teenage and unskilled workers; thus their prices should be more sensitive to minimum wage increases. Enumerators price multiple items at most outlets—about half of the sample's outlets have seven price quotes, and about a third have eight. Once an outlet is selected, specific items are selected for pricing with probability proportional to sales. During our 1995-97 period, an "item" most commonly was a meal, as BLS aimed to price complete meals as typically purchased at an outlet (for example, a meal item at an LS outlet might consist of a hamburger,french fries, and a soft drinl~). Our dataset codes items broadly, as breakfast, lunch, dinner, or snacks (corresponding to BLS "entry level item" codes}. Because we did not obtain exact meal descriptions for our dataset, we can't compute meaningful measures of average prices across outlets. But BLS strives to price identical items over time, and codes in our database describe temporal item substitutions due to discontinuances and alterations. Our ana.rysis focuses on price changes for identical items; we do not use price comparisons where one quote reflects BLS item substitution6 confidential confidential Aaronson {2001) and Card and Krueger(1995} used published BLS price indexes for 271arge metro areas, whereas we use actual prices in $8 urban and metro areas. Consequently, we can assess more observations over a widen zar~ge of labor market conditions. We can also identify prices in LS outlets, whereas published price indexes cover all outlets. But the database has some clear limitations. Because BLS introduced a complete outlet and item resampling in January 1998, we only use data through December 1997, and cannot look for long tags in response to the September, 1997 Federal minimum wage increase. And because our dataset contains no specific item descriptions, we cannot tie price changes to item-specific measures of input price changes (such as ground beef or chicken price indexes). Characteristics of the data We report descriptive statistics to introduce this distinctive data, to provide support for some of our modeling decisions, and to compare restaurant pricing behavior to evidence of nominal price rigidity reported in other firm-level data sets. Frequency of price changes Prior studies of nominal pricing rigidities often fnd that prices remain fixed for long periods(Caziton 1986, Cecchetti 1986, Kashyap 1995). That pattern is likely to hold in restaurants, where firms often review prices at regular time intervals, and change item prices only at those intervals (Hershey,2000). If restaurants change prices only at designated review times, then we will need to account for the possibility of Lead and lagged responses to minimum wage increases. confidential confidential Consider figure 3, which shows the proportion of prices in our dataset that remained unchanged in each bimonthly period. Full service meals show remazkable stability—on average, $7.4 percent of prices remained unchanged in any two month period, and that share shows little vaziation over time. Aside from the sharp changes immediately following minimum wage increases in 1996 and 1997, LS rxxeal prices also appear to be quite stable; across all outlet types and bimonthly comparison periods, 86.6 percent of prices remained unchanged. Another way to look at the incidence of price changes is to consider a price's typical life span. Duration descriptions are limited by the short sample period (with resultant left and right censoring of duration measures), but we can still offer some useful information. We looked at LS items whose prices increased just after the minimum wage increase in October 1996. Ten months later, 56% of those new prices remained unchanged. The pattern was quite similar among items that didn't change price just after the minimum wage increase: 49°70 of those prices remained unchanged 10 months later. The duration measures are consistent with figure 3, and the findings together suggest that the half-life of an LS restaurant price in this low-inflation period was about 10 months~ The distribution of price changes During months without a minimum wage increase, the mean bimonthly saznplewide price change was 0.37 percent, and varied little (from 0.36 to 0.38) across outlet types. During months with a minimum wage hike, the mean price change more than doubled in limited service outlets, rising to 0.88 percent, but rose only slightly (to 0.45} in ail other outlets. We return to quantifying this minimum wage effect later in the paper. confidential confidential Since most prices are unchanged in bimonthly comparisons(figure 3), samplewide price increases must combine zero increases for mQSt items with substantial mean price increases among Ehose items whose prices change.$ Table 1 reports on the size distribution of price changes. The top panel summarizes the results for linnited service restaurants, while the bottom panel covers full service. Columuns report data separately for price increases and decreases, and for months with and without minimum wage changes. Several features are noteworthy. First, there are both "large" and "small" changes in price but increases are clearly skewed to the right. Among limited service outlets, roughly one-quarter of all price increases are less than 2 percent and an additiona130 percent are between 2 and 4 percent. A tail of larger increases raises the mean price increase, conditional on an increase, to 5.3 percent when minimum wages aren't increasing and 4.8 percent when they aze. Similar magnitudes obtain among full service outlets: just over half of all increases are less than 4 percent, while mean increases are just under 5 percent9 One-ninth of price increases exceed IO percent, with a somewhat smaller incidence during minimum wage increases. At first glance, these wide distributions seem in contrast to some(S,s} models that predict firms will move when a threshold between optimal and actual prices is crossed. However,it is possible that the cost of changing prices varSes cross-sectionally ox should be modeled as a random variable as in Caballero and Engel(1999), where the cost of changing prices is sometimes low,leading to small price adjustments, and sometimes high, leading to large price adjustments. Consistent with cross-sectional vaziation in (S,$) pricing, the mean size of price increases is remarkably stable over our short time frame. Figure 4 reports mean confidential confidential bimonthly price increases between March 1995 and December 1997, among quotes increasing price. Aside from a sharp spike among limited service outlets in the summer of 1997 related to large increases in bacon usage and pork belly prices, the mean bimonthly price increase stays within a range of4 to 6 percent. The standard deviation of bimonthly mean price increases is 0.5 percent for full service outlets and 1.3 percent for limited service outlets, with the bacon episode accounting for half the difference in standard deviations. Furthermore, there's little change in the distribution of LS price increases when minimum wages are increasing, and Iater analyses find no minimurx~ wage effect on the mean size of price increases.j0 Lastly, although price cuts are rare, they tend to be larger in magnitude than price increases (table 1}. Over one quarter of price cuts exceed 10 percent, and cuts frequently exceed 20 percent. Although small price cuts occur, less than half are below 4 percent Because mean price increases vary little from month to month, the evidence suggests that firms may respond to industry-wide cost shocks by adjusting the likelihood of price increases rather than their size. Moreover, because many price cuts reflect sales, and hence are of larger size but limited duration, we may need to explicitly account for prior price changes in our later modeling of rrunimum wage effects. Across- and within-store synchronization Az~ important feature of modeling the inflation process is the mechanism by which prices move together. The sticky price literature generally assumes that price changes are staggered across stores but synchronized within-stores. Lath and Tsiddon (1992,1996} confirm this pattern using micro data from the food component of the Israeli CPI. 9 confidential confidential Kashyap(2995)notes that catalog price changes are not synchronized with common price shocks,such as changes in the money supply, suggesting prices are staggered acrossstores. Although we do not provide any formal tests, same simple descriptive statistics suggest that staggering probably occurs both across- and within-stores in the U.S. restaurant industry. Figure 3, which shows the monthly proportion of quotes that do not change price far the previous two months, provides evidence of across-store staggering. If stores pez~ectly synchronized their price changes and price durations over 10 months, we would erect to see a series of eight Os followed by a 1. This would imply a standard deviation of 33 percent, approximately 7 to 16 times the size of the standard deviations that we see in the data. But instead, figure 3 shows remarkable stability in the share of no-change quotes. The standard deviation of this proportion is 4.8 percent for limited service outlets (2.8 percent without the minimum wage nnonths included) and 2.0 percent for full servit;e outlets. For this pattern to occur, and given the price durations noted above, price changes must be staggered, not synchronized, across oatlets. We looked at within–store staggering by looking at price changes among those outlets with many quotes--7 or 8 price comparisons in a bimonthly period. Since most sample outlets have 7 or 8 quotes, we still had a lot of data, 8,809 outlet-mornths and b4,750 bimonthly price comparisons. Table 2sorts the data by outlet type, number of quotes, and time (whether or not a minimum wage is increasing}, and for each category reports the distribution of within-outlet price increases. If perfect synchronization occurs within stares, observations should bunch at zero and the maximum number of price increases(7 ox 8). Price changes do bunch at zero—most outlets change no sample prices l0 confidential confidential in most months, even when minimum wages increased. But when outlets do increase prices, they cleazly stagger price increases across items—there's no cleaz bunching at the maximum number of price increases. Table 2 also provides some suggestive evidence of how prices change in response to minimum wage increases. First, note there is little obvious change among FS outlets, but a sharp change among LS outlets: far fewer LS outlets keep all prices stable during rrunimum wage changes, ana far more LS outlets incxease many prices. Price points The final noteworthy element is the existence of psychological price points (Kashyap 1995). Almost 12 pexcent of LS item Prices end in 99 cents. Over 30 pexcent end in 9(about three times random chance). Only Z percent of full service restaurant prices end in 99 cents, but 8 percent end in 95 cents and the three most common price endings--95,00, and 25 cents-- encompass 20 percent of all observations. Price changes also cluster: charges of 5, 10, 20, and 30 cents account for half of all observed LS price changes(10 cents alone accounts for 26 percent), while those four, plus 25 cents,5Q cents, and a dollar, account far half of all FS price changes. Nominal price points suggest a discontinuous threshold whereby firms will lose marginal revenue if prices are raised.i~ If price points aze important, we would expect to see quote durations that are Ionger and price changes that are higher when starting from these thresholds. Furthermore, critical masses for price changes suggest that outlets may put off price changes until costs rise to a price changing point; if true, then price increases will lag cost increases. 11 confidential confidential Restaurant price data are generally consistent with other studies that report evidence of nominal price rigidity. First, price durations are long and price changes are rare. Second, the size of price changes varies cross-sectionally, with both small and large price increases common,but the cross-section distributions are stable over our three year period. Third, price staggering appears to occur both within- and across-stores. Finally, the use of certain price points is common and, as we show later, results in long quote durations and larger price increases. We next look more closely at how restaurants might respond to increases in minimum wages. Measuring the Effects of Minimum Wage Changes President Clinton signed a bill raising the minizx~um wage on August 20, 1996, after a debate that .lasted through the previous six months. The increase took place in two stages: an October 1, 1996 increase from $4.25 an hour to $4.75 an hour(11.8%), and a second increase l 1. rrxonths later, on September 1, 1997, to $5.15 (8.4%}. Price responses to the change should display temporal and geographic variations. Consider the timing of the 1996-97 Federat increase. When the law was passed (August 20, 1996) businesses knew minimum wages would be increased in 6 weeks {October 1, 1996) and again in a year(September 1, 1997). An outlet that reviewed prices quarterly might raise prices just before a minimum wage rise, or it might wait until the next review several months later.i2 Our azaalysis will consider lead and lag temporal effects. Changes in the Federal minimum affect specific geographic areas in different ways, because some states maintain higher minimums, and because prevailing market wages vary widely. States with no minimum wage standards and those with minimums I2 confidential confidential pegged to the Federal level experience the full change in the Federal minimum, while states with minimums above the new Fedezal level remain unaffected by the policy changes. Some State minimums fall between old and new Federal minimums; these states experience increases that are less in percentage terms than the first group of states. Finally, some State minimums changed during 1995-97, and businesses in those states faced minimum wage increases at times that differed from other states. We captare varying geographic effects by defining the measure MW,t:the percentage increase in the effective minimum wage in state i during month t.13 Because of State policies, percentage increases in the Federal minimum wage exceed sample mean values of MW;t, which is 1Q.7'% in October, 1996, compared to the 11.8°Io Federal increase, and 7.0% in September, 1997, compared to the 8.4% Federal increase. Market wages can vary widely across and within states. In some areas, prevailing low-skill wages may exceed minimum wages, and increases in the minimum wage should have little effect on market wages for low-skilled labor, and consequently little effect on costs and prices. Tn other regions, minimum wages may exceed market wages, and changes in the minimum wage will have stronger effects on obsexved wages,costs and prices. We expect increases in effective miniznuxn wages to have greater impacts on costs and hence an prices in low-wage areas. Our database includes unique outlet codes and precise outlet locations (addresses, zip codes, and telephone numbers)so we can link items to outlets and we caz~ link outlet records to related geographic information. We use data from the Current Population Survey(CPS)to summarize hourly wage distributions in the outlet's Metropolitan Statistical Area(MSA).14 We use observations for all of 1996 for each MSA,and 13 confidential confidential calculate the distribution of MSA hourly wages. We use the 20`h percentile howly wage (WAGE2Q)as our measure of low-skill wages in a metro area. Model and Basic Empirical Results Our basic statistical model can be summarized as follows: 1} In (Pk~,t/Pk~.t-z)=f(PPI, IPRICE, MEALTYPE,MW). Pk~,t is the price of the item k at outlet j in month t. The dependent variable is therefore the percentage change in price over a bimonthly period. Table 3 details the explanatory variables. PPI, a vector of two--month percentage changes in the Producer Price Index for Processed Foods, measures input price shocks faced by sample outlets, and specific measures include contemporaneous aswell asone- and two-period lags. The vector IPRICE captures some dynamics of changes in the item's price over time, reflecting pricing strategies as well as seasonal price patterns in restaurants. Some sale prices fall by substantial amounts in our sample(table 1), and are likely to increase substantially some time after the promotion. Similarly,items with price increases frequently decline again at some later period, because the original increase may have reflected seasonal cost increases or because rivals didn't match an increase. In order to capture these dynamics, we enter a variable, IPUP,equal to the percentage increase in the item's price in a previous period, ox zero if the price did not increase. Similarly, IPDOWN is the percentage decrease in price in a pxevious period, or zero if there was no decrease. We enter one-, two- and three-period lags on the item price variables.'s 24 confidential confidential The BIS uses ELI {entry level item)codes in item selection, and Food Away from Home codes identify dinner,lunch,snack, and breakfast/brunch items. We dropped snack observations(a small part of the database}, since they were disproportionately offered through unusual outlet types,like vending machines or gas stations. The model includes dummy variables fox lunch and breakfast meals, with dinner as the base. Finally, MW is the change {increase)in the effective minimum wage in period t at outlet j. We also use lead and lag measures of MW to assess price effects one period before and one period after periods including minimum wage increases. We found no evidence of longer leads or lags, either with these two month periods or with analyses of those prices at outlets that are surveyed monthly. We present the basic analysis of price effects in Table 3, using ali outlets. BLS generally collected several prices at an outlet in 1995-97(the mode is 7); because quotes from the same outlet are unlikely to be statistically independent, standazd error calculations in all analyses account for quote clustering, using Huber-White robust estimation techniques. Adjustment matters here, as robust standard errors are generally two to three times larger than OIS standard errors. Item price effects are important and statistically significant in table 3; prior price cuts lead to current period price increases, and past price increases indicate current price cuts (responses aze dampened, however; full reversion to a previous price would imply coefficient values, in absolute terms, of 1, while these fall well below I and usually below 0.1). PPI effects are small--food is 20 percent of outlet costs, and outlets purchase many different combinations offood products--while unreported meal type effects were small and not statistically significant. 15 confidential confidential We multiply all reported minimum wage coefficients and standard errors by 10,to save space (results should therefore be read as the effects of 10 percent, rather than 1 percent, minimum wage increases). In equation 1 of table 3, the minimum wage effect is positive and highly significant(t=5.09}, suggesting that outlet prices rise by 0.33 percent, on average,for contemporaneous 10 percent increases in the minimum wage. Equation {2) adds lead and lag effects for the minimum wage change (MW;,t_~ measures price responses to prior period minimum wage increases while MW;,~+~ tests for price responses to next period minimum wage increases). Each effect is positive; the one period lag effect is highly significant(t=3.18}, while the one period lead is significantly greater than zera at the 90%confidence level(t=1.72}. Lead and lag effects raise the estimate on the contemporaneous effect, and the full effect is more than double that in equation 1,16 ~'he Incidence of Minimum Wage Price Effects Table 4 examines minimum wage effects in more detail, with models that account for differences in outlet types and in prevailing area wages. The table reports minimum wage effects, but the models retain all other explanatory variables in table 3.'~ Equations 1-3 explore outlet type effects, running regressions for full (equation 1) and limited service outlets (equations 2 and 3). Full service minimum wage effects are statistically significant but quite small in a specification that mirrors table 3's equation 2. In contrast, the effects in limited service outlets are much larger(equations 2 and 3}. The coefficient an MW;t in equation {2}, positive and highly significant(t=6.44), suggests that LS prices rise by 0.8 percent in response to contemporaneous 14 percent increases in minimum wages (four times the FS effect). Equation (3)lead and lag effects are positive, 16 confidential confidential large, and statistically significant(t=2.25 for the lead effect and t=2.56 for the lag effect).18 Again,inclusion of lead and lag effects raises the contemporaneous estimate. Equations 4-6 add an interaction term between MW,~ and WAGE20,the hourly wage at the 20~' percentile of an area's wage distribution. High values of WAGE20 should indicate high wage areas, which maybe less strongly affected by changes in the minimum wage. In the all-outlet sample (equation 4)the main MW coefficient remains positive and highly significant, while the interaction Term is negative, fairly large, and statistically significant(t=2.07). Minimum wage increases have larger effects an pxices in low wage areas. Among full service outlets (equation 5), the coefficients on MW and the interaction term are of the expected sign but are only marginally significant. Equation (6)adds the interaction term to the LS sample. As in the all-outlet sample,IS wage effects are negative, substantive, and statistically significant(t=2.09). Price effects are larger in low wage areas. Finally, equation(7)adds interaction terms between WAGE20 and the lead and lag MW effects. Inclusion of those measures has no effect on the contemporaneous effects. The Zagged effects have the expected sign (positive on MW;,~_~ and negative on the interaction) but are not quite significant. The lead effects lose all power,suggesting that there is no interaction and that inclusion of it simply created multicollinearity with the main effect. How Large are Minimum Wage Price Effects? So far, we've found that restaurant prices increased when minimum wages rose, that increases were larger among limited service outlets and in low-wage areas, and that the response occurred in a six month window around the wage increase. The patterns 17 confidential confidential suggest that minimum wages affect outlet costs in different ways, and that price changes vary in response to those different cost effects. We have no information on factor shares for the outlets in our sample, and no data that would a11ow for estimation of factorsubstitution possibilities. As a result, we can't directly estimate the effects of minimum wages on outlet costs, and therefore cannot directly identify the extent to which costs are passed through in the form of higher prices. Instead, we rely on Lee and O'Roark's(1999) calculations of likely pass-through under various assumptions about factor shazes, substitution possibilities, and firm behavior. They used 1992 Current Population Survey wage and employment data, and U.S. Input-Output tables, to calculate the direct and indirect(through purchases)shares of low wage tabor in "eating and drinking places"{SIC 581}. Assuming full pass-through of costs to prices, no substitution, and no spillovers of nninimum wage increases to other wages, a 10 percent minimum wage increase would lead to a 0.74 percent price increase among eating and drinking places in Lee and O'Roark's model (table 5). Eating and drinking places include bars that serve no food, and our transactionbased sample includes supermarkets that serve food eaten on premises, schools, gas stations, and the like. But there's still a wide overlap between the Food Away from Home all-outlet sample and SIC 581, and there's a strikingly close correspondence between the Lee-O'Roark estimate of the full pass-through price increase and our estimated price increase for the all-outlet sample~.74 vs. 0.73 in table 5. We base our calculation on the surri of the lead, lag and contemporaneous coefficients, the most comparable approach.19 Further comparison can be found in Aaronson (2000) who used aggregated 18 confidential confidential 1978-95 BLS indexes to estimate along-run price response of 0.72, quite close to our 0.73 for disaggregated 1995-97 BLS data. Table 5 also reports price effects by outlet type and by prevailing wage level. In each case, we report short tun (current period)estimates as well as longer run estimates that add lead and lag effects. LS price effects are more than twice as large as the a11-outlet estimate—a 1.56 percent price increase in response to a 10 percent minimum wage increase, white FS price effects are much lower,0.40 percent if we exclude the Lead effect (negative and not significant). In turn, LS effects are larger still in low wage areas-1.83 percent when wages of 5.50 an hour set the area 20`~ percentile. How Do Oatlets Raise Prices? We've seen that prices at LS outlets rise by nearly 1.6 percent in response to a 10 percent inczease in minimum wages. Restaurants can arrange that increase in a variety of ways. They could raise all prices by 1.6 percent, or they could raise fewer prices by greater amounts. If they choose the latter, they've got to decide which item prices to increase. We pursue that question in this section. We begin by assessing the incidence of price increases, rather than the average price response to minimum wage change. Table 6 reports minimum wage effects for the LS sample, based on logit estimates of the model used throughout, but with the dependent variable now a dummy variable set equal to one for price increases. The contemporaneous minimum wage effect is positive and highly significant (t=3.b7} in equation 1, while the lagged effect is positive and marginally signifcant (t=1.76) and the lead effect is small and not significant. The wage interaction is 14 confidential confidential statistically significant(t=2.29)and works in the expected direction, with a smaller incidence of price increases,following a minimum wage increase,in high wage areas. In equation (2), we investigate the linkage between discrete restaurant price reviews and the formation of leads and lags in price responses. Most outlets change no sampled prices in most months: that decision may reflect the underlying econorxucs, but it may also reflect periodic rather than continuous price reviews. Tf price reviews are periodic, then outlets that have recently changed prices may not arespond to a minimum wage increase until another price review. We define a variable P_REV,and set it equal to one if the outlet changed price for any sampled item in the previous period (that is, prices were reviewed in the previous period). We then interacted P_REV with each of our MW terms. The results are striking. Minimum wage changes are less likely to lead to a price increase if the outlet had reviewed prices in the period prior to the increase. In the lagged period,firms are considerably mare likely to raise price if the outlet raised price just after the minimum wage changes.20 Equations 3-6 explore the role of pricing points. We define a variable PP99 and set it equal to one if the LS item price at the beginning of tie period ended in 99 cents. The coefficient on PP99 in equation (3)is negative and quite significant—firms are considerably less likely to raise a price ending in 99 cents. Equation {4)interacts PP99 with our minimum wage variable, and the results indicates that the logit coefficient on PP99 changes very little when minimum wages increase. We add an interaction with the actual value of the beginning period price in equation (5); the marginally significant price suggests that restaurants are particularly unlikely to raise prices on low priced 94 cent 2Q confidential confidential items (priced at 0.99 or 1.99). The last equation combines the PP99 and P_REV measures and fnds that results are robust to the inclusion of all.21 Because the coefficients in logit models can't be directly interpreted, table 7 reports the predicted probabilities of a price increase under different conditions. We set the base probability of a price increase to that predicted on the basis of the intercept(no changes in PPI or IPRICE measures over the previous several periods). The base probability of a price increase is 10.3 percent for items that aren't price points, and 1Q percent minimum wage increases have large contemporaneous effects, raising probabilities to 23 and 33 percent in high and low wage areas, respectively. Probabilities decline noticeably if the outlet had just had a prior period price review,to 16 and 24 percent, respectively. Lagged effects are of interest; if the outlet changed no prices in the immediate wake of a minimum wage increase, the probability of a lagged increase is quite small, 12.9 percent. But for outlets that responded immediately with price increases on some items then there's a substantial lagged possibility(21 percent)of a pxice increase. Price point of€ects are quite large. The probability that an item's price will be increased if price ends in 99 cents falls by 35 percent, to 6.7 percent, in the base case. While probabilities rise after a minimum wage incxease, price point items remain substantially more stable than other items. Table 8 considers the second aspect of price construction—how much are item prices increased, when outlets decide to raise price? The sample is restricted to LS items with price increases and the dependent variable is the log difference in prices. Equations (1)and (2)support the cruder inferences drawn on figure 4 and table 2: minimum wage effects have virtually no effects on the size of price increases, as the coefficients are small 21 confidential confidential and not nearly significant. As table 2 suggests, firms respond to minimum wage increases by raising more prices, not by rraising prices more. In equation (3), the coefficient on PP99 is positive and significant; when firms raise prices on price point items, they xaase them by more, although the coefficient is not that large (suggesting that PP99 item prices rise by 7.2 percent, on average). When we interact PP99 with its actual beginning price, we find that typical percentage price increases are larger among low-priced items (about 8.9% for 99 cent items, and 8.3 far %a $1.99 items, versus 5% for $G.99 items), when they are increased. Finally, we can see that P_REV plays a role, too; price increases are somewhat smaller for those items with prior month price reviews.22 Discussion Food prices respond quickly to minimum wage increases; most of the observed response(about b0%)occurs in the two-month period immediately after a minimum wage increase, while the rest occurs in the periods two months after or two months immediately preceding the wage increase. The timing is close to that observed by Aaronson using aggregated BLS data over the 1978-95 period; he found that about 56% of the price increase occurred in the two months immediately after the increase. Actual price changes could occur even more quickly than our estimations suggest,if they occur eazly in the lead and lag periods (footnote 18). Since other evidence indicates that restaurant prices respond only slowly to cost changes, the minunum wage response suggests that prices will respond quickly to common and permanent industry cost shocks. 22 confidential confidential Prices rose by amounts that were predicted on the basis of full pass-through of costs. While the finding does not prove that casts are fully passed through(wage spillovers, along with less than full pass-through, would give the same result), it does provide greater confBence for the results. Moreover,the incidence of increases accords with what we know of the low-wage labor market,Prices rose far more in LS outlets than in other restaurants, reflecting greater reliance on low wage labor in the former, and prices rase more in low wage areas than in high wage areas. Our paper's title implies an interest in the size, speed, and incidence of price increases following minimum wage increases, but it also connotes an interest in the mechanics of how restaurant increase prices. Our xesults imply that prices in limited service restaurants zise by 1.56 percent, on average,in response to a 10 percent minimum wage increase(and by 0.73 percent in the a11-outlet sample}. But outlets don't construct that response by raising all prices 1.56 percent; instead they increase price by about S percent, on average, on a much smaller set o€items. That pattern suggests one source of stickiness, item-specx~c casts of changing prices. We found two other specific sources, a marked reluctance to change price on items with prices at a psychological pricing point (ending in 99 cents), and a temporally discrete approach to price review. Despite evidence of stickiness, the rather small cost increases associated with minimum wage increases did generate a quick aggregate price response. As a widely publicized and permanent industry-wide cost shock, we suspect that minimum wage changes do not generate the sort of coordination failures that other cost or demand changes produce. 23 confidential confidential References Aaronson, Daniel,2Q01, "Price Pass-Through and the M~iniznum Wage," The Reviex~ of Economics and Statistics, forthcoming. Card, David and A]an Krueger, 1995,Myth and Measurement: The New Economics of the Manimum Wage,Princeton, NJ: Princeton University Press. Carlton, Dennis,"The Rigidity of Prices," American Econorrcic Review (June, 1986): 637658. Cecchetti, Stephen, 1986,"The Frequency of Price Adjustment: A Study of the Newsstand Prices of Magazines," The Journal ofEcoraorrcetrics, pp. 255-274. Hershey, Robert D., Jr.,"The Cost of Not Living On a $5.15 Minimuxr~", The New York Times, September 19, 2000: p. C1. Kashyap, Anil, 1995, "Sticky Prices: New Evidence Frorn Retail Catalogs," The Quarterly Journal ofEconomics, pp. 245-274. Lach,Saul and Daniel Tsiddon,"The Behavior of Prices and Inflation: An Empirical Analysis of Disaggregated Price Data," J'ournad ofPolitical Economy (April, 1992}: 349389. Lach,Saul and Daniel Tsiddon, "Staggering and Synchronization in Price-Setting: Evidence from Multi-Product Firms," American Economic Review, (December, 1.996): 1175-1196. Lee, Chinkook and 4'Roark, Brian,"The Impact of Minimum Wage Increases on Food and Kindred Products Prices: An Analysis of Price Pass-Through," Economic Research Service, US Department of Agriculture. Technical Bulletin No. 1877. July, 1999. Rubin, Aiissa J.,"Populaz Minimum Wage Hike Gets Solid Senate Approval," Congressional Quarterly, July 13, 1996, pp. 1964-65. U.S. Department of Labor, Bureau of Labor Statistics, ~landbook ofMethods, Washington: U.S. Government Printing Office, 1997. Weisman, Jonathan, "Republican Defectors Help Propel Minimum Wage Bill to Passage," Congressional Quarterly, May 25, 1946, pp. 1461-54. 24 confidential confidential Table 1: Bsmonthly Pxice Changes, by Outlet Type and Minimum Wage Change Month w/ Min Wage ? Type of Price Change Limited Service Outlets % of Monthly Quotes Mean Price Change Incidence of Price Changes 0-2% 2-4% 4-6°/a 6-8% 8-10% 10-20% >20% Full Service Outlets of Monthly Quotes Mean Price Change Incidence of Price Changes 0-2% 2-4% 4-6% 6-8% 8-10% 10-24% >20% (1) No Increase 11.4% 5.31% 25.5 3U.6 Ib.8 9.4 6.3 7.7 3.8 10.8% 4.76% 32.7 25.1 15.3 1.~.1 6.2 9.2 1.6 (2 Yes Increase 3) No Decrease 4 Yes Decrease 22.4% 2.7% 8.36% 4.83% --Share of Quotes(%)-29.3 22.8 18.3 ~ 34.1 i I.8 18.9 7.0 8.0 6.7 4.b I6.2 7.9 12.8 I.5 2.6% 8.I9% 11.2% 1.7% 7.45% 4.88% --Share of Quotes(%)-25.3 26.2 19.8 27.6 12.2 18.$ 11.3 10.7 7.4 7.1 17.1 8.4 1.2 6.8 1.7% 9.29% 24.5 15.3 15.3 11.2 6.t 18.4 9.2 21.4 143 8.9 9.8 12.5 21.4 11.6 Notes: A month with a minimum wage increase is one immediately (within two months) following any state or federal change that effectively raised the minimum.wage in a state. 25 confidential confidential Table 2: Incidence of Price Increases Among Outlets with Many Quotes Price Quotes at Outlet Minimum Wage T ? (Z) 7 No (21 7 Yes t3) 8 No (4) 8 Yes Limited Service Outlets Outlet-Months Bimonthly Price Pairs Number of Price Increases 0 1 2 3-5 6 7 8 1852 287 1070 162 12,964 2,009 8,568 1,296 -Share of Outlet-Months(%}77.1 60.3 67,5 59.3 7.6 6.6 12,1 14.2 3.3 5.6 5.7 4.9 6.7 12.5 9.7 11.1 1.2 4.5 2.1 1.2 4.2 10.5 1.7 5.6 1.1 3.7 Full Service Outlets Outlet-Months Bimonthly Price Pairs Number of Price Increases 0 1 2 3-5 6 7 8 3086 505 1,578 259 21,602 3,535 12,624 2,152 -Share of Outlet-Months(%)80.3 S1.2 81.Q ?8.1 5.7 5.3 5.6 5.9 2.4 3.0 2.7 2.2 6.0 5.8 4.6 6.fi 1.8 0.8 1.1 1.9 3.8 4.0 2.0 1.9 3.0 3.3 Notes: Months with increasing minimum wages are October and November, 1996, and Septennber and October, 1997. 26 confidential confidential Table 3: Magnitude and Timing of Price Responses to Minimum Wage Increase Description Coefficients (s.e) C2) (~) 0.315 0.379 {0.032) (0.026) PPI~ %change in the Producer Price Index for Processed Foods, period t 0.050 (0.029) 0.070 (0.030) PPI~_~ %PPI change, period t-1 -O.Q70 (0.037} -0.040 (0.038) PPI~_Z % PPI change, t-2 U.061 (0.028) 0.046 (0.030) IPUPL_1 %increase in item price, one period Iag; zero if no increase -0.089 (0.013) -0.090 (0.013) IPDOWNt_~ %decrease in item price, one period lag; zero if no decrease 0.388 (0.445) 0.389 (0.045) IPUPt_Z %increase in item price, two period lag; zero if no increase -O.U70 (0.012) -0.070 (O.aI2) IPDOWN~_2 %decrease in item price, two period lag; zero if no decrease 0.096 (0.023) 0.096 (0.023} IPUPt_3 %increase in item price, three period lag; zero if no increase -0.03b (0.010) -0.036 (0.009} IPDOWNt_3 %decrease in item price, three period lag; zero if no decrease 0.060 (0.030) 0.060 (0.034) MW;~ %change in miuumum wage in state i, period t. 0.331 (0.065) 0.4Q6 (0.063) MW;,~_i %minimum wage change, state i, period t-1. 0.207 (0.006) MW;,t+~ %minimum wage change, state i, period t+1 0.115 (0.007) Variable Intercept 0.065 RZ 6$,887 N Note: equations also included categorical meal type variables 27 confidential 0.066 68,887 confidential Table 4: Incidence of Price Response to 10% Minimum Wage Increase Outlet Type Variables MW,t C1) FS C2) IS 0.191 (0.086) 0.845 (Q.125) C3) LS (4) All ~5) FS ~6~ LS (7} LS Coefficients and standard errors 0.937 1.449 1.048 (0.134) (0.531) (0.692) 2.b98 {0.883) 2.69fi (4.883) MW;,~_~ 0.212 (0.083) 0.305 (0.119} 0.200 (0.068) 0.227 (0.088) 0.263 (0.121) x.222 (Q.797) MW;,t+F -0.062 (Q.075) 0.319 (0.142} 0.124 {0.070} -O.Q41 (0.080) 0.296 (O.15S} -0.262 (0.810} -0.162 {O.Q78) -0.129 (0.099) -0.278 (0.133) -0.278 (0.133) MW;=*WAGE20 MW;,~.1*WAGE20 -0.147 (0.120) MW ;,t+~*WAGE20 0.086 (0.128) N R2 35,759 .017 21,064 .151 21,Q64 .152 61,716 .Ob8 32,822 .Q18 18,024 .1 G4 Notes: Each regression also includes other variables listed in table 1. FS refers to full service outlets, while LS refers to Iimited service outlets. MW;,t-1 and MW;>t+l refer to 1 period before and 1 period after minimum wage changes. All standard error estimates are adjusted for nonindependence within outlets. 2s confidential 28,024 .165 confidential Table 5: Compazing estimates of the price effect of a 10% minimum wage increase Authors Lee O'Roark Aaronson MacDonald &Aaronson & " " " Scope of estimate All outlets All outlets All outlets (lr, sr) Full Service (lr, sr) Limited Service (lr, sx) Limited Service, high wage (lr, sr) Limited Service,low wage {lx, sr) %Price Rise 0.74 0.72 0.73, Q.41 0.34, 0.19 1.56,0.94 1.31, Q.6S 1.83, 1.17 Notes: `Ir' refers to Long run response, while `sr' refers to short run (one period) response. Hi wage cities have WAGE20 equal to 7.37 an hour, while low wage cities have WAGE20 equal to 5.50 an hour(median of cities in top and bottom quartiles, respectively. 29 confidential confidential Table 6: Logit Model Coefficients: Probability of Price Increase, Limited Service Variables (1) 2.939 (0.800) Coefficients and standard errors (2) (3) (4) (5) 2.983 2.854 2.852 2.841 (Q.802) (0.799) (0.799) (0.798) {6) 2.896 (0.801} MW;,t.~ 0.3I3 {0.178} 0.220 (0.193) 0.308 {0.179) 0.308 (Q.179) 0.308 (0.179) 0.212 (.I94) MW;,C+I O.OSS {0.176) 0.224 (Q.206) 0.048 (0.176) 0.048 (0.176) O.Q49 (0.176} 0.214 (0.205) MW;t*WAGE20 -0.2$4 (0.124) -0.273 (0.125) -0.272 (0.125) -.270 (0.125) -0.269 (0.124) -0.260 (4.124) ~VIW;t MWIt*P_REV -Q.496 (0.278) -0.482 (4.279) MW;,=_~*P_REV 0.616 (0.313) 0.632 (0.301) MW;,;+i*P_REV -0.461 (0.304) -0.454 (Q.305) PP99 -0.485 (0.129) PP99*MW;~ -0.4b4 (Q.137) -x.726 (0.2Q4) -0.482 (0.130) -O.U98 ~a.2g~> PP99*P1 0.062 (0.03b) Notes: Standard errors are in parentheses. Estirz~ates drawn from logit estimation; ;full results reported in Appendvc. Dependent variable is dummy variable equal to one if price increased in period. MW+1 and MW-1 refer to 1 period before and 1 period after minimum wage change.P REV is a dummy variable equal to 1 if any outlet prices were changed in the previous period. PP99 is a dummy variable equal to 1 for items with beginning prices ending in .99. P1 is the beginning period price. WAGE20 is the 20~` percentile wage in the sampling area. 30 confidential confidential Table 7: Predicted probabilities that an item's price will increase Events Price Review Last Period? P1 ends in 99 cents? (I) No No (2) Yes No {4} Yes Yes (3} No Yes Predicted Probability of Price Increase(%) 6.7 6.7 10.3 10.3 Base Probability 10% Minimum Wage Increase In high wage area In low wage azea 23.4 33.1 15.4 23.5 15.9 23.5 1 Period After Increase 12.4 21.Q 8.0 I4.1 1 Period Befoxe Increase 12.4 8.2 8.0 5.2 ~ 10.5 15.9 Notes: Predicted probabilities from logit model of price increase. Full model reported in appendix table 1. Low wages are set to $S.SO an hour; high wages are set to $737. 31 confidential confidential Table 8: Minimum Wage Price Effects, Among Items that Raised Price, Limited Service Variables (~) 0.020 (0.338) Coefficients and standard errors ~2) (3) (4) 0.750 0.949 x.837 (1.966) (1.977) (1.9b5) (5) 0.646 {1.958) MW,,~_i -0.007 (0.431 -0.052 (0.494} -0.076 (0.495) -0.114 (0.493) -Q.148 (0.495) MW;,~+i 0.844 (0.589} 0.926 (0.696) 0.918 (Q.700) 0.876 (0.702} 0.940 (0.696) -0.095 {0.314) -0.123 {0.312) -0.107 (0.309) -0.0$4 (0.306) MW;~ MW;~*WAGE20 P REV -0.691 (0.303) PP99 1.712 (0.448} PP99*P1 R2 N 0.460 2841 0.476 2,316 0.483 2,316 4.083 (0.924) 4.195 {0,930) -0.592 (0.165) -0.620 (0.170) 0.487 2,316 0.489 2,316 Notes: All equations also include other variables listed in table 1. MW+1 and MW-~ refer to 1 period before and 1 period after minimum wage change. P_REV is a dummy variable equal to 1 if any outlet prices were changed in the previous period. PP99 is a dummy variable equal to .l for items with beginning prices ending in .99. P1 is the beginning period price. WAGE20 is the 20th percentile wage in the sampling area. 32 confidential confidential Figure 1: Comparing Changes in the Away Prom Home CPI and Meat PP3s and beet — — rk 15~ ,0 1 If; aO ~, 3~ 1~ , 1 ls .~ ~ -10 • r■ ~ ~ rl 1j ~' Jan-95 Apr-95 Jul-45 Oct-95 Jan-96 Apr-96 Jul-96 Oct-96 Jan-97 Apr-97 Jul-97 Oct-97 Figure 2: Response of Food Away from Home CPI to Minimum Wage Increase, 1978-97 0.035 0.030 0.025 ,~ 0.024 0 ~ 60 ~.~15 S Q.~~~ U ~ O.Ot?5 0.000 -0.005 -0.010 -0.015 t-4 t-3 t-2 t-1 t Month 33 confidential t+i t+2 t+3 t+4 confidential Figure 3: Percent of Items with Unchanged Price, Full and Limited Service Outlets too so ~ ~a r0 K W so ao Mar-9S Jun-95 Sep-45 Dec-95 Maz-96 !un-46 34 confidential Sep-96 Dec-96 Mar-97 Jun-97 Sep-97 Dec-97 confidential figure 4: Mean Price Increases, Fu21 and Lamited Service Outlets ' " LS FS 1000 8.00 8 6.00 ;c 4.00 2.00 Mnr-95 Jun-9S Sep-95 Dec-95 Mar-96 Jun-96 Sep-96 Dec-96 Maz-97 Jun-97 Sep-97 Aec-97 35 confidential confidential 'Price responses are critical for measuring the welfare implications of minimum wage legislation. 2 They estimated puce increases of 3.3% fora 10% increase in minimum wages, suggesting that low wage labor accounted for one third of restaurant costs, if cost increases were fully passed through to prices. But total payroll accounts for less than 30% of restaurant sales, according to Retail Census data. Low wage labor more likely accounts for one sixth of costs (Aaronson, 2000). 3 The calculations, presented in Aaronson (2000), are based on the published monthly and bimonthly Food Awa;+ from Home indexes for 27 U.S. cities. 4 To maintain consistency in our price analyses, we randomly assigned outlets in.the five largest PSUs to odd or even two month cycles. Counts are based on all itenns for which prices were obtained in October or November, 1996,just after the October 1, 1996 increase in the Federal minimum wage. A more complete description of the sample's construction, and particularly of outlet and item selection procedures, can be found in the BLS Handbook ofMethods. 5 BLS replaced an old ordering with these type of business codes in July, 1996, and began to report Food Away from Home price indexes for type of business groupings in January, 1998. Businesses surveyed early in our sample period were assigned these codes retroactively. 6 It's possible that firms could respond to a minimum wage increase by reducing quality, instead of raising price. We don't have measures of quality, but our dataset notes whether an item is the same as the item priced in the previous month,or whether BLS has had to 36 confidential confidential substitute another item. The data show no evidence of any increased incidence of item substitutions following minimum wage increases. ~ The uaeasures are consistent in the following sense: if 93.3 percent of prices were unchanged each month (implied by 86.6 percent bimonthly stability), and if the probability of change was independent of the length of time that the price had been unchanged, then 1 Q months later we would expect 50 percent of prices(1-0.933~°)to be unchanged. With the ten month survival rates so close to 50 percent, the probability of price change does appear to be independent of the length of time that it had been fixed. $ Price changes are also highly clustered near the mean, with excess kurtosis of62.E and 80.8 for price changes among LS and FS outlets, respectively. Distributions of increases and decreases are also quite peaked compared to normal distributions, with excess kurtosis of 14.2(LS)and 8.6(FS)for increases, and 1..6(LS)and 6.8(FS)for decreases. Kashyap(1995)also reports positive excess kurtosis in his sample. 9 By comparison, Kashyap (1995} reports that more than 20 percent of the price increases in his study of catalog prices are less than 3 percent. His study looks at changes over six month intervals rather than the two month intervals that we report. 10 We applied Kolmogorov-Smirnov D-tests fox differences in the price distributions in table 1. We found no significant differences in price decreases in months with minimum wage increases compared to other months, and no significant differences in the distribution of increases among LS outlets. FS outlets with price increases do show a statistically significant shift, driven by the higher incidence of 0.2% increases in months without minimum wage increases. 37 confidential confidential "Price points maybe connected to coordination failures. In markets with a limited number of sellers, prices are strategic complements. A seller maybe reluctant to raise a price unless they're sure that other sellers will also raise price. That uncertainty may be particularly pronounced for items with price points, if consumers are particularly sensitive to changes in those prices. 1z Firms clearly had more foreknowledge of the 1997 increase. The 1996 increase could not have been predicted until sk~ortly before the House of Representatives vote on May 23, after a week of legislative maneuvezing that almost consigned the bill to defeat ;'Neisman, 1996; Rubin, 1996}. Even then, the final timing didn't become•clear until adoption of the conference report on August 2. Businesses therefore knew of the 1996 increase 2-4 months prior to implementation, while they knew of the 1997 increase, which was specified in the 1996 bill, 12-13 months before implementation. I3 boring 1995-97, 12 states had minimum wages above the Federal standazd. We summarize state-specific minimum wage changes with information from the Monthly Labor Review's annual (January issue} surveys of state labor legislation. Tf the effective minimum wage in a state increased by 10% on October I, then MW will equa10.10 for August-October price comparisons and for September-November price comparisons(and 0.0 for othex comparisons). 'a For the 12 non-metro urban areas, we use wage data for the non-metro parts of the outlet's state. CPS codes are unavailable for 9 MSAs,so sample sizes decline when area wage data are included in the analysis. is Adding lags reduces sample size. Three period lags(up to 6 months, with two month periods) were always statistically significant, while a fourth period was not. 38 confidential confidential 16 Tl~is pattern closely matches that found by Aaronson using aggregated data for 197$95. He found a sharp price spike at the minimum wage increase, statistically significant individual month effects at the one month lead and lag, and a full effect captured in a window of six months surrounding the minimum wage increase (three before and three afterj. Our two month contemporaneous,lead, and lag periods sum to six months. 17 Estimated minimum wage effects are quite robust to the inclusion or exclusion of the other explanatory variables, and the pattern of coefficients on the other explanatory 'variables changes little as the model changes to capture different minimum wage effects. _ As a final check, v`+e estimated models with fixed firm and PSU effects. Key minimum wage coefficients were unaffected by the inclusion of fixed effects, and the fixed effects themselves added almost nothing to the model's fit. 8 BLS enumerators collect prices in 3 roughly week-long collection periods during the first 22 days of a month. Outlets are visited during the same collection period each month, although not necessarily on the same day each month. Therefore, prices observed at "one period lags" after the October 1, X996 minimum wage increase include, at the near minirnam,comparisons made for the first week of December coanpared to the first week of October and, at the far rnaacimum,comparisons for the third week of January compared to the third week of November. One period lead effects include, at the near minimum,comparisons of the third week of September to the third week of July and, at the far ma~cimam, comparisons ofthe first week of August to the first week of June. In short, some "lead" and "lagged" price changes could have been made quite close to the date of the minimum wage change, as little as 10 days before oz after. 39 confidential confidential 19 While our findings are consistent with full pass-through, they are also consistent with less than full pass-through, coinciding with spillovers. 20 We also looked at outlet price increases two periods before a minimum wage increase, but adding another period provided no new information for the analysis. The results suggest the following sequence. Outlets that reviewed prices two periods before a minimum wage increase did not then review again in anticipation of the minimum wage change. Those that did raise price in anticipation of the nninimum wage increase(one period before} had only modest contemporaneous price increases, while those without anticipatory price increases before the minimum wage change raised pxices substantially. 2i We looked at other LS price points. About 3 percent of prices end in 98 cents, and over half end in the digits 8 or 9,but additional price point measures were not nearly significant, once we entered PP99. 22 Price point effects were also importazit in unreported FS outlet models. There, prices that ended in 25,50, or 95 cents were much Tess likely to be increased, and were changed by more when they were changed. 40 confidential confidential Working Paper Series A series of research studies on regional economic issues relating to the Seventh Federal Reserve District, and on financial and economic topics. FDiCIA After Five Years: A Review and Evaluation George J. Benston and George G. Kaufman wP-97-? Money,S[icky Wages, and the Great Depression Michael D. Bordo, Christopher J. Erceg and Charles L. Evans WP-97-2 Pzice Pass-Through and Minimum Wages Daniel Aaronson WP-97-3 Habit Persistence and Asset Returns in an Exchange Economy Michele Boldrin, Lawrence J. Christiano and Jonas D.M. Fisher WP-97-4 North-South Terms of Trade: An Empirical Investigation Michael A. Kouparitsas WP-97-5 Interactions Between the Seasonal and Business Cycles in Production and Inventories Steven G. Cecchetti, Anil K. Kashyap and David W. Wilcox WP-97-6 "Peso Problem" Explanations for Term Structure Anomalies Geen Bekaert, Robert J. Hodrick, and David A. Marshall WP-97-7 The Big Problem of Smail Change Thomas J. Sargent, Francois R. Velde WP-97-8 Bank Capital Standards for Market Risk: A Welfare Analysis David Marshall and Subu Venkataraman wP-97-9 Monetary Policy and the Term Structure of Nominal Interest Rates: Evidence and Theory Charles L. Evans and David A. Marshall wP-97-10 Employer Learning and Statistical Discrimination Joseph G. Altonji and Charles R. Pierret WP-97-11 A Model of Commodity Money, With Applications to Gresham`s Law and the Debasement Puzzle Francois R. Velde, Warren E. Weber and Randall Wright WP-97-12 The Evolution of SmaII Change Thomas J. Sargent and Francois R. Velde WP-97-13 The Role of Credit Market Competition on Lending Strategies and on Capital Accumulation Nicola Cetorelli WP-97-14 Algorithms for Solving Dynamic Models with Occasionally Binding Constraints Lawrence J. Christiano and Jonas Fisher wP-s7-15 confidential confidential Working Paper Series ~~onr~nu~a~ The Return from Community College Schooling for Displaced Workers Louis S. Jacobson, Robert J. LaLonde and Daniel G. Sullivan wP•9~-1s Modeling Money Lawrence J. Christicznv, Martin Eichenbaum and Charles I,. Evans WP-97-17 Monetary Policy Shocks: What Have'UVe Learned and to What End? wP-97-18 Lawrence J. Chrisfiano, Martin Eichenbaum and Charles ~ Evans Volunteer Labar Sorting Across Industries Lewis M. Segal, Ediza6eth Mouser and Burton A. Weisbrod vrn-9z-~s Would Freetrade Have Emerged in North America without NAFTA`t Michael A. Kouparitsas WP•97-2a The Roie ofthe Financial Services Industry in the Local Economy Douglas D. Evanoff, Philip R. Israilevich and Graham R. Schindler wP-972'1 The Trojan Horse or the Golden Fleece2 Small Business Investment Companies and Government Guarantees WP-97-22 Elijah Brewer III, Hesna Genay, William E. Jackson Ill and Paula R. 4l~orthington Temporary Services Employment Durations: Evidence from State UI Data Lewis M. Segal and Daniel G. Sullivan WP-9r-2s The Determinants of State Food Manufacturing Growth: 1982-92 Mike Singer WP-97-24 Requiem for a ZvIarket Maker: The Case of Drexel Burnham Lambert and Below-Investment-Grade Bonds Elijah Brewer III and William E. Jackson 111 WP-97-25 Plant Level Irreversible Investment and Equilibrium Business Cycles WP-ss-1 Marcelo Veracierto Search, Self-Insurance and Job-Security Provisions Fernando Alvarez and Marcelo Veracierto WP-98-2 Could Prometheus Be Bos~nd Again? A Contribution to the Convergence Controversy WF>-98-3 Nicola Cetorelli The Informational Advantage of Specialized Monitors: The Case of Bank Exameners Robert DeYoung, Mark J. Flannery, William W. La~tg and Sorin M. Sorescu wP-98-a Prospective Deficits and the Asian Currency Crisis WP-98-5 Craig Burnside, Martin Eichenhaum and Sergio Rebelo Stock Market and Investment Good Prices: Implications of Microeconomics Lawrence J. Christiano and Jonas D.1t~ Fisher WP-98.6 2 confidential confidential Working Paper Series ~~onrr.~uea~ Understanding the Effects of a Shock to Government Purchases Wendy Edelberg, Martin Eichenbaum and Jonas D. M. Fisher WP-98-7 A Model of Bimetallism Francois R. Velde, and Warren E. Weber WP-98-8 An Analysis of Women's Return-to-Work Decisions Following First Birth Lisa Barrow WP-98-9 The Quest for the Natural Rate: Evidence from a Measure of Labor Market Turbulence Ellen R. Rissman WP-98-10 School Finance Reform and School District Income Sorting Daniel Aaronson WP-98-11 Central Banks, Asset Bubbtes, and Financial Stability George G. Kaufman wP-sa-72 Bank Time Deposit Rates and Market Discipline in Poland: The Impact of State Ownership and Deposit Insurance Reform Thomas S. Mondschean and Timothy P. Opiela wP-98-13 Projected U.S. Demographics and Social Security Mariacristina De Nardi, Selahattin bnrohorog'Zu and Thomas J. Sargena WP-98-14 Dynamic Trade Liberalization Analysis: Steady State, Transitional and Inter-industry Effects Michael Kouparitsas WP-98-15 Can the Benefits Principle Be Applied to State-local Taxation of Business? William H. Oakland and William A. Testa WP-98-16 Geographic Concentration in U.S. Manufacturing: Evidence from the U.S. Auto Sapplier Industry Thomas H. Klier WP-98-17 Consumption-F3ased Modeling of Long-Horizon Returns Kent D. Daniel and David A. Marshall WP-98-18 Can VARs Describe Monetary Policy? Charles L. Evans and Kenneth N. Kuttner WP-96-19 Neighborhooc! Dynamics Daniel Aaronson WP-98-20 Inventories and output volatility Paula R. Worthington WP-98-2~ Lending to troubled thrifts: the case ofFHLBanks Lisa K Ashley and Elijah $rewer III w~-9s-z2 3 confidential confidential Working Paper Series ~~anr~nuea~ Wage Differentials for Temporary Services Work: Evidence from Administrative Data Lewis M. Segal and Daniel G. Sullivan wP-9&23 Organizational flexibility and Employment Dynamics at Young and Old Plants Jeffrey R. Campbetd and Jonas D. M. Fisher WP-98-24 Extracting Market Expectations from Option Prices: Case Studies in Japanese Option Mazkets Hisashi Nakamura and Shigenori Shiratsuka wp-99-1 Measurement Errors in Japanese Consumer Price Index Shigenori Shiratsuka WP-99-2 Taylor Rules in a Limited Participation Model Lawrence J. Christiana and Christopher J. Gust WP-99-3 Maximum Likelihood in the Frequency Domain: A Time to Build Example Lawrence J.Christiano and Robert J. Vigfusson 1M~-9g-a Unskilled Workers in an Economy with Skill-Biased Teciu►ology Shouyong Shi WP-99-5 Product Mix and Earnings Volatility at Commercial Banks: Evidence from a Degree o€Leverage Model Robert DeYoung and Karin P. Roland Wp-94.8 School Choice Through Relocation: Evidence from the Washington D.C. Area WP-s9-7 Lisa Barrow Banking Market Structure, Financial Dependence and Growth: International Evidence from Industry beta Nicola Cetaredli and Michele Gambera WP-99-8 Asset Price Fluctuation and Price Indices Shigenori Shiratsuka WP-99-9 Labor Market Policies in an Equilibrium Search Model Fernando Alvarez and Marcelo Veracierto WP-99-10 Hedging and Financial Fragility in Fixed Exchange Rate Regunes Craig Burnside, Martin Eichenbaum and Sergio Rebelo WP-89-1 ~ Banking and Currency Crises and Systemic Risk: A Taxononny and Review George G. Kaufman WP-99-12 Wealth Iaeyuality, Intergenerational Links and Estate Taxation Mariacristina De Nardi WP-99-i3 Habit Persistence, Asset Returns and the Business Cycle Michele Boldrin, Lawrence J. Chrisiiuno, and Jonas D.M Fisher WP-99-14 4 confidential confidential Working Paper Series ~~ont~ued~ Does Commodity Money Eliminate the Indeterminacy of Equilibria? Ruilin Zhou Wp-ss-~5 A Theory of Merchant Credit Card Acceptance Sujir Chakravorti and Ted To WP-99-~ s Who's Minding the Store? Motivating and Monitoring Hired Managers at Small, Closely Held Firms: The Case of Co~nercial Banks Robert DeYoung, Kenneth Spong and Richard J. Sullivan WP-99-1 Assessing the Effects of Fiscal Shocks Craig Burnside, Martin Eichenbaum and Jonas D.M. Fisher WP-9~1s Fiscal Shocks in an Efficiency Wage Model Craig Burnside, Martin Eichenbaum and Jonas D.M. Fisher WP-99-19 Thoughts on Financial Derivatives,Systematic Risk, and Central Banking: A Review of Sonne Recent Developments William G Hunter and David Marshall wP-s9-2o Testing the Stability ofImplied Probability Density Functions Robert R. Bliss and Nikotaos Paniginzogdou WP-99-21 Is There Evidence of the New Economy in the Data? Michael A. Kouparitsas wP-98-22 A Note on the Benefits of Homeownership Daniel Aaronson WP-ss-23 The Earned Income Credit and Durable Goals Purchases Lisa Barrow and Leslie McGranahan wP-ss-24 Globalization of Financial Institutions: Evidence from Cross-Border Banking Performance Allen N. Berger, Robert DeYoung, Hesna Genay and Gregory F. Udell WP-99-25 Intrinsic Bubbles: The Case ofStock Prices A Comment Lucy F. Ackert and William G Hunter WP-99-26 Deregulation and Efficiency: The Case of Private Korean Banks Jonathan Hao, William C. Hunter oral Won Keun Xang WP-99-2~ Measures of Program Performance and the Training Choices of Displaced Workers Louis Jacobson, Robert LaLonde and Daniel Sullivan wry-ss-28 The Value ofRelationships Between Small Firms and Their Lenders Paula R. WorthBngton WP-99-29 Worker Insecurity and Aggregate Wage Growth Daniel Aaronson and Daniel G. Sr~llivan WP-99-30 S confidential confidential Working Paper Series ~~anc~nuea~ Does The Japanese Stocic Market Price Bank Risk? Evidence from Financial Firm Failures Elijah Brewer Ill, Hesna Genay, William Curt Hunter and George G. Kaufman wp-99-31 Bank Competition and Regulatory Reform: The Case of the Italian Banking Industry Paolo Angelini and Nicola Cetorelli wP-98-32 Dynamic Monetary Equilibrium in aRandom-Matching Economy WP-00-7 Edward J. Green and Ruilin Zhou The Effects of Health, Wealth, and Wages on Labor Supply and Retirement Behavior WP-00-2 Eric French Market Discipline in the Governance of U.S. Bank Holding Companies: Monitoring vs. Influencing Robert R. Bliss and Mark J. Flannery Using Market Valuation to Assess the Importance and Efficiency of Public School Spending Lisa Barrow and Cecilia Elena Rouse Employment Flows, Capital Mobility, and Policy Analysis Marcelo Veracierto Does the Community Reinvestment Act Influence Lending? An Analysis of Changes in Bank Low-Income Mortgage Activity Drew Dahl, Douglas D. Evano,f~"and Michael F. Spivey WP-oo-3 WP-00.4 vUP-oo-5 WP-oo-s Subordinated Debt and Bank Capital Reform Douglas D. Evanoffand X.arry D. Watl WP-00-7 The Labor Supply Response To(Mismeasured But} Predictable Wage Changes Eric French wP-o0-8 For How Long Are Newty Chartered hanks Financially Fragite2 Robert DeYoung wP-oo-s Bank Capital Regulation With and Without State-Contingent Penalties David A. Marshall and Edward S. Prescott WP-00-10 Why Is Productivity Pracyclical? Why Do We Care? Susanto Basu and John Fernald WP-oo-11 Oligopoly Bantcing and Capital Accunnulation Nicola Cetorelli and Pietro F. Peretto wP-00-~2 Puzzles in the Chinese Stock Market WP-00-13 John Fernald and John H. Rogers The Effects of Geographic Expansion on Bank Efficiency Allen N. Berger and Robert DeYoung wP-00-14 D confidential confidential Working Paper Series ~~ont~nued~ Idiosyncratic Risk and Aggregate Employment Dynamics Jeffrey R. Campbell and Jonas D.M. Fisher WP-00-15 Post-Resolution Treatment of Depositors at Failed Banks: Implications for the Severity of Banking Crises, Systemic Risk, and Too-Big-To-Fail George G. Kaufman and Steven A. Seelig WP-0o-16 The Double Play: Simultaneous Speculative Attacks on Currency and Equity Markets Sujit Chakravorti and Subir Loll wP-oo-17 Capita3 Requirements and Competition in the Ban[cing Industry Peter J.G. Vlaar WP-ao-18 Financial-Intermediation Regime and Efficiency in aBoyd-Prescott Economy Yeong-Yuh Chiang and Edward J. Green WP-o~-t9 How Do Retail Prices React to Minimum Wage Increases? James M. MacDonald and Daniel Aaronson wp-o0-20 confidential